plasmid

1972, 36(3):361. Bacteriol. Rev.

R C Clowes plasmids.Molecular structure of bacterial

http://mmbr.asm.org/content/36/3/361.citation

Updated information and services can be found at:

These include:

CONTENT ALERTS more»(when new articles cite this article),

Receive: RSS Feeds, eTOCs, free email alerts

http://journals.asm.org/site/misc/reprints.xhtmlInformation about commercial reprint orders: http://journals.asm.org/site/subscriptions/To subscribe to to another ASM Journal go to:

on January 29, 2014 by guesthttp://m

mbr.asm

.org/D

ownloaded from

on January 29, 2014 by guest

http://mm

br.asm.org/

Dow

nloaded from

http://http://mmbr.asm.org/content/36/3/361.citation

http://mmbr.asm.org/cgi/alerts

http://mmbr.asm.org/cgi/alerts

http://journals.asm.org/site/misc/reprints.xhtml

http://journals.asm.org/site/subscriptions/

http://mmbr.asm.org/




BACTERIOLOGICAL REVIEWS, Sept. 1972, p. 361-405 Vol. 36, No. 3Copyright 0 1972 American Society for Microbiology Printed in USA.

Molecular Structure of Bacterial PlasmidsROYSTON C. CLOWES

Division of Biology, The University of Texas at Dallas, Dallas, Texas 75230

INTRODUCTION AND DEFINITIONS ...................................... 361SOME PROPERTIES OF BACTERIAL PLASMIDS .......................... 362Chromosomal Integration ............... .................................... 363Fertility Repression........................................................ 363Incompatibility ............................................................. 364Elimination ................................................................ 364Replication and Segregation (Membrane Site Model) ...... ................... 365

GENETICS AND LINKAGE OF PLASMID-BORNE GENES ................. 366MOLECULAR MODELS FOR PLASMIDS AND THEIR IMPLICATIONS FORTHE EVOLUTION OF R FACTORS.367

PHYSICAL COMPOSITION AND FORM .................................... 369METHODS OF ISOLATION ................................................. 371

Differences from Host DNA Base Ratio ................................... 371From Minicells ............................................................. 372Lysis with Nonionic Detergent (Brij 58) ......... ............................ 373Conjugal Transfer.......................................................... 373Isolation of CCC Plasmid Molecules.374

Cellulose nitrate adsorption ............ ................................... 374Alkaline sucrose sedimentation ........................................... 374Ethidium bromide-cesium chloride centrifugation ...... .................... 374

METHODS OF MOLECULAR WEIGHT DETERMINATION ................. 374Sedimentation Analysis ................. .................................... 374Loss of Supercoil Structure by Irradiation ........ ........................... 375Reassociation Kinetics...................................................... 376Electron Microscopy........................................................ 376

SIZE AND CONFIGURATION ............ .................................. 376F and F' Sex Factors.376Colicinogenic Factors........................................................ 379Drug-Resistance Factors ............... ................................... 380Monomolecular factors ............... .................................... 380Unstable R factors........................................................ 381Multimolecular R factors .............. ................................... 382

PHYSICAL INTERRELATIONSHIPS ........................................ 384MOLECULAR STRUCTURE ..... ............................................ 385Monomolecular Plasmids .......................... 385A Revised Map for R222 .................... 387Alternative Molecular Structures ofR Factors.387Speculations on R-Factor Evolution ....... ................ 389

ASPECTS OF PLASMID REPLICATION ............ ........................ 390Plasmid Copy Number-Relaxed and Stringent Regulation.390Mechanism of Relaxed Replication.391"Transition" in Drug Resistance ................ ............. 392DNA-Protein "Relaxation Complexes"......................................... 393Replication in Minicells ........... ............................ 394Transfer Replication .......... ............................. 394

CONCLUSIONS ........................................... 395LITERATURE CITED .......... ............................. 397

INTRODUCTION AND DEFINITIONS somal. There has been a natural tendency toattribute to many of these properties the same

During the past few years, an increasing genetic and molecular basis as has been de-number of bacterial properties have been derived from considerably more extensive studiesscribed which behave in such a way as to sug- of such related elements as the F sex factorgest that their genetic control is not chromo- (136) and the temperate bacteriophage X

361


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

(113). Since the application of the terminologydeveloped for F and X to these other geneticelements often leads to specific implications,definitions of terms are particularly important.

Thus, an extrachromosomal element is heredefined as a genetic element which is physi-cally separate from the chromosome of the celland is able to be perpetuated stably in thiscondition. The term therefore embraces suchelements as the P1 prophage but would ex-clude wild-type X prophage or the deoxyribo-nucleic acid (DNA) of abortive transductants.As presently used in bacteria, the term "ex-trachromosomal element" is synonymous withthe term "plasmid," which is further definedas an element which is nonessential for thegrowth of normal cells of the host species, sothat under most conditions it may be gainedor lost without lethal effect. Those plasmidswhich have been shown to be able to occupy achromosomal site are referred to as episomes.Thus, whereas a plasmid may or may not bean episome, all episomes are plasmids. Sincethe definition of "episome" implies a geneticrelationship between the plasmid and the hostcell, the same element may be an episome inone host but a plasmid in another. It has oftenbeen implied that plasmids are those episomeswhere a chromosomal location in one or an-other host has not so far been demonstrated. Itshould be borne in mind, however, that therewould appear to be no a priori reason that aplasmid must be able to integrate into, or havegenetic homology with, the chromosome of anyhost cell.

Plasmids may be conveniently classified intotwo major types, infectious and noninfectiousplasmids. Infectious (or self-transmissible)plasmids (e.g., F, ColIb) control the establish-ment in the host cell of a "donor state"(through such physical features as sex pili), toprovide a mechanism of conjugation that per-mits their transfer from the host cell into an-other cell, the recipient, which is thereby itselfconverted into a donor. Those infectious plas-mids that can also promote the transfer ofother genetic material, either chromosomal orplasmid, are known as sex factors. The ques-tion at the moment remains open whether ornot all infectious plasmids have this property.Noninfectious plasmids (e.g., ColE1) are notable to set up the donor state and require ei-ther that a sex-factor plasmid be present inthe same cell for their conjugal transfer or thattransfer be effected by a transducing phage.The plasmid nature of a noninfectious plasmidis thus more difficult to establish and is ofteninferred by determining that its transmission

is unlinked to any chromosomal marker. How-ever, conjugal transfer of both chromosomaland nonchromosomal material to the same re-cipient cell can occur during the course ofnormal inating conditions (see 42). This maybe mistaken for linkage of a plasmid withchromosomal markers (4) and can be distin-guished by observing conjugal transfer from anumber of Hfr donors with differing "origines"of transfer, in which case true linkage is unaf-fected whereas an extrachromosomal elementwill show a constant time of entry, by all Hfrstrains (42). Both types of plasmid can betransmitted efficiently in a generalized trans-ducing phage system (10, 12, 83, 105, 237).

It will not be the intention of this review toconsider in any detail many of the genetic andphysiological properties of bacterial plasmidswhich have been extensively dealt with in anumber of recent reviews and symposia (71,168, 178, 247, 252), nor will the physical as-pects of those plasmids with no clear geneticor physiological function (30, 54, 125, 154, 206)be examined, nor yet those that can be re-garded as phages (166, 214). In general, themain emphasis will be on the physical charac-teristics of those plasmids, with clearly definedproperties, that are not phages and, whereverpossible, the relationship of their genetic andmolecular properties. Other reviews focussingon the physical aspects of some of these plas-mids have recently appeared (63, 88, 111).

It now seems possible that many plasmidgenes do not have chromosomal counterparts.Thus, there is clearly a need for distinguishingbetween plasmid and chromosomal geneswhich, though phenotypically similar, may befunctionally quite distinct. In fact, the viewexpressed by Demerec et al. (66) that "loci onplasmids and episomes are not different inkind from loci on the chromosome" could bedisputed. Thus, the symbols used in the orig-inal publications for plasmid determinants willbe continued, in spite of the fact that fre-quently these do not conform to the standardtriletter lower case italics (66).

SOME PROPERTIES OF BACTERIALPLASMIDS

Most of the bacterial plasmids extensivelystudied have been those originating in strainsof Enterobacteriaceae, although recently anincreasing number have been characterizedfrom staphylococcal strains. In enteric bac-teria, the identification of the classical plasmidF has been followed in recent years by the dis-

362 CLOWES


mbr.asm

.org/D

ownloaded from



MOLECULAR STRUCTURE OF BACTERIAL PLASMIDS

covery of two other major groups of plasmids:the colicinogenic (Col) plasmids and the infec-tious drug-resistance (R) plasmids. Most ofthese plasmids also give rise to the donor statein their host cells with the result that the F sexfactor was for some time taken as a model forall plasmids. However, during the last fewyears it has become increasingly clear thatthese other groups of plasmids, althoughshowing many of the important characteristicsof the F factor, differ in other major respects.Some of these differences are outlined below.

Chromosomal Integration

A major feature of F is its ability to inte-grate within the chromosome of the host strainEscherichia coli K-12, by which property itwas characterized as an episome and, togetherwith X, formed the basis for the original defini-tion of that element by Jacob and Wollman(135). (It should be noted that wild-type Xundergoes lethal [vegetative] replication and isthus unsuitable for classification as a true bac-terial plasmid.) However, integration may infact be a non-essential property of plasmids,since a number of bacterial plasmids, for ex-ample, the colicinogenic factors ColIb orColEl, do not appear to integrate within thechromosome of at least one of their commonhosts, E. coli K-12. Transmission of ColIb orColEl from a series of Hfr K-12 donors is thusindependent of any chromosomal marker (42,171). Nevertheless, ColIb is an effective sexfactor in K-12 (41), and its ability to transferchromosomal markers has sometimes beenused to imply chromosomal integration. If thiswere true, and integration via normal recombi-nation were blocked by the presence of a recA(33) mutation in a donor strain, then we wouldexpect that chromosomal transfer would besimilarly blocked. We have shown that, al-though in the case of those transfer factorssuch as F, which are known to integrate, theuse of recA donors considerably reduces chro-mosomal transfer, nevertheless, a small levelof transfer (about 1 recombinant per 108 donorcells) remains (45). This level does not appearto be due to leakiness of the recA mutation,since there is a loss of polarity of markertransfer by F' factors from recA donors com-pared to transfer from recA+ donors (whereas,residual activity, if mediated via integrationby recombination, would be expected to main-tain the polarity of normal transfer). Moreimportantly, chromosomal transfer by ColIb isindependent of recA function and may there-fore be concluded to be independent of the

integration of ColIb into the chromosome (45),although the possibility of integration via amechanism related to the X integrase system(99), or some other mechanism (82), is noteliminated. More recent evidence, using E.coli dnaA mutants which are temperature-sensitive (ts) in the initiation of their chromo-somal DNA replication (117, 118), supports theidea that ColIb does not integrate in E. coli.Revertants of such ts mutants (but not of the"fast-stop" chain elongation DNA replicationmutants), able to grow at the normally nonper-missive temperature, are found at an increasedfrequency from ts strains carrying an F sexfactor. These revertants appear to arise as aresult of a mechanism (termed "integrativesuppression") resulting from integration of theF sex factor in the chromosome, so that a newinitiation site for chromosomal replication isprovided within the integrated sex factor (175).Such revertants cannot be isolated from thesame ts mutants harboring a ColIb factor (173),or indeed one of several R factors of the fi-type (173; T. Arai, personal communication),implying that these plasmids are not able tointegrate in E. coli K-12.

Since the time that an extrachromosomalnature was suggested for such potentially le-thal elements as the ColIb plasmid (43, 171), ithas also been established that certain tem-perate phages, such as P1, set up their pro-phage state without integration of the pro-phage genes within the chromosome (126). Itmay well be, therefore, that episomes are aspecial class of plasmids having an additionalproperty of being able to integrate within cer-tain host chromosomes, and it appears an openquestion at the moment whether chromosomalintegration has any important bearing on ei-ther the evolution or the stability of theplasmid-host relationship.

Fertility Repression

It has been shown that many sex factorsdo not generally express the properties ofthe donor state in all cells of a host strain(168, 171, 223), F again being an exceptionto the rule. The fertility properties of most sexfactors are generally repressed in the majorityof cells, and their fertility has been concludedto be regulated by an operator-repressor typesystem (168, 172). (Although a more recentanalysis of a number of transfer-defectivemutants has led to a more complex model, inessence it may be similar [1, 2, 80, 132, 246, 248,249].) Mutations to derepression of fertilityhave been identified and have been shown

VOL. 36, 1972 363


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

to behave genetically as of two major types,consistent with the idea that one type (whichis recessive) occurs in a regulator gene pro-ducing a diffusible repressor-like substance,and the other type are due to mutations atthe site of action of this repressor (the opera-tor) and are thus dominant (119, 167). F ap-pears to be defective in producing a repressor,since it is sensitive to the putative repressorsof some repressed factors in biplasmid hosts(168, 239). [However, among a series ofmutants of an F'iac factor selected as insensi-tive to this repression, two subgroups arefound, one dominant and another recessive,when present in a triplasmid cell containingthe repressing plasmid, parental F'lac and theF'lac mutant. From this it has been arguedthat repression is complex and requires at leasttwo molecules, one (which is plasmid-specific)produced by all plasmids including F, andanother, produced by self-repressed factors,which can also act on F (80, 246).lThe class of R factors able to repress F have

been termed fi+ (fi = fertility inhibition) andthose R factors which do not repress the fer-tility of F are termed fi- (239). Similar groupsexist within the colicin factors. Some, such asthe ColB factors, repress F (92, 109, 190),whereas others, such as ColIb, do not have thisproperty (171).As one manifestation of the donor state,

specific hairlike processes termed sex pili orsex fimbriae are elaborated on the surface ofthe host cell (23). Some workers have con-cluded that the genetic material is transferredfrom the donor to the recipient cell via thepilus (23), but even though pili have beenshown essential for the process of conjugation(see 56), this conclusion is an inference whichrests at best on indirect evidence (186). Infec-tious plasmids appear to determine one of atleast two distinct classes of pili. The factorstermed F-like synthesize F pili which are sim-ilar in their sites for adsorption of those samemale-specific phages (now more properlytermed F-specific phages, e.g., MS2, R17, f2,Qfl, 42, fd) which adsorb to pili synthesized bythe F factor. In contrast, I-like factors form Ipili which do not carry sites for adsorption ofthe F-specific phages, but carry sites for ad-sorption of another group of phages termedI-specific (since they were first described forthe ColIb factor) which do not in turn adsorbto F pili (168).

Until recently, the subdivision of F-like fac-tors appeared to be identical with that of fi+factors, and I-like factors appeared to be of the

fi- factor type. However, it is now clear thatmany fi- factors are not I-like (20, 58, 141, 233),some being reported to produce yet a thirdtype of pilus (141). (It has also been claimedthat, although most fi- factors isolated in Eng-land are I-like, the majority of fi- factorsarising in Japan are not I-like [58].) Moreover,certain I-like plasmids have been reported torepress the fertility of the F factor whenpresent in the same cell (they are fi+; see 101),whereas some self-repressed F-like plasmids(e.g., ColB3; Clowes et al., J. Gen. Microbiol.55, proc. iv, 1965) do not repress F-pili forma-tion (they are fi-). These variations probablyindicate that the interpretation of fertility reg-ulation is more complex than a simple re-pressor-operator interaction (80, 109, 249).

IncompatibilityUsually two isogenic plasmids cannot be

stably maintained in the same bacterial cell.Thus, an F'lac plasmid cannot be stably trans-ferred to an Hfr strain (69), and, if transferredto an (F'gal)+ strain, either the F'lac or theF'gal is perpetuated in individual cells, but notboth (65). This phenomenon, termed plasmidincompatibility, is distinguished from entryexclusion, a change on the surfaces of plasmid-containing cells which inhibits the transfer of arelated plasmid (76, 178). Incompatibility canalso occur between pairs of plasmids which arenot isogenic, e.g., ColV2-K94 and F (138, 160);F and ColV3-K30 (160); ColB2-K77 and 222/R(92, 109), and is usually a polarized process,one plasmid (the first of the pairs shownabove) excluding the other, irrespective ofwhich is the resident and which the superin-fecting plasmid (109, 160). It has been reportedthat pairs of fi- or of fi+ R factors are incom-patible, whereas an fi- and an fi+ plasmid canstably coexist in the same cell (194, 239). How-ever, more recent reports show that, althoughall fi+ R factors appear to be incompatible,several compatible groups have been foundamong fi- R factors (20, 58). Incompatibility isusually interpreted as indicating close interre-lationship, and it may indicate some commonevolutionary origin of those factors involved.Clearly, until the process of plasmid replica-tion and its regulation is more clearly under-stood, the basis for incompatibility is likely tobe obscure.

EliminationPlasmids may be lost spontaneously from

host strains, and this loss may be increased bycertain treatments (termed "curing"), in-

364 CLOWES


mbr.asm

.org/D

ownloaded from




cluding the use of intercalating dyes (acridines[46, 116, 236] or ethidium bromide [21]), ri-fampin (16, 35, 137), thymine starvation (46,160, 188a), crystal violet (203) or surface-activeagents (131, 208, 222). Different plasmids varyconsiderably in their ability to be "cured," andthis does not seem to be dependent upon suchproperties as sex-factor activity. For example,both ColEl and ColIb are completely refractoryto curing by acridine orange, which is mosteffective on F (46). Curing of F by acridineorange (AO) is restricted to the autonomousstate, integrated F being refractory (116).Curing of autonomous F by AO has beenshown to be due to the inhibition of F replica-tion (120).

Replication and Segregation (MembraneSite Model)

Many current ideas on plasmid replicationand segregation originate in a model first pro-posed by Jacob, Brenner, and Cuzin in 1963(133). Although it now seems likely that someaspects of this model may be oversimplified, itstill remains the basis of most ideas inter-preting replication, segregation, and otherimportant properties of plasmids. The hypoth-esis was proposed that those DNA moleculesthat are capable of replication (termed "repli-cons") are circular in structure and carry atleast two gene loci controlling their replica-tion; at one locus on the replicon is located aregulator gene which produces a diffusible sub-stance (initiator) acting upon the second locus,an operator of replication (replicator), topermit DNA replication to be initiated fromthat point; regulation of replication is thuspostulated to be under positive control. Frag-ments of DNA lacking these genes were pre-sumed incapable of replication and were sug-gested to be characterized by such fragmentsof bacterial DNA as exist in abortive transduc-tants. In the case of such elements as X, inwhich vegetative replication is unregulatedand finally leads to the lysis of the cell, theregulated (prophage) state is set up by integra-tion of the X genome within the chromosome ofthe host cell. By this means, replication of theX prophage genome is coordinated with thereplication of the host chromosome. Replica-tion of the F plasmid, when in the autonomous(F+) state, also appears to be regulated so thatthere is only one copy of the plasmid per chro-mosome, although in this state F is not inte-grated within the host chromosome but is,nevertheless, quite stable. To explain its regu-

lated replication and stable inheritance, it wasproposed that the F replicon was attached to acellular site to which the chromosomal rep-licon was also attached. This attachment site,suggested to be located on the cell membrane,would control and transmit signals leading toinitiation of replication of both chromosomeand all attached plasmids. During each cellgeneration, the membrane site with the at-tached plasmids would duplicate so that, atcell division, a copy of the membrane site withattached chromosome(s) and plasmid(s) wouldbe transmitted to each daughter cell, thus en-suring the stable inheritance of all cellularcharacters whether plasmid or chromosomallycontrolled. Incompatibility between isogenicplasmids was explained as due to a limitationto a single plasmid membrane site leading tocompetition for this site, so that only one oftwo isogenic plasmids could attach and bestably inherited. Incompatibility between ap-parently unrelated plasmids was suggested tobe due to the use of the same replication siteleading to similar competition, whereas unre-lated and compatible plasmids were presumedto have distinct membrane-attachment sites.This model therefore offered an attractive yetsimple explanation of the way that plasmidscould be stably inherited even though theywere present in as few copies as one per chro-mosome, in addition to suggesting a basis forthe phenomenon of incompatibility.

Several experiments support the idea of a''segregation unit" of chromosome and Ffactor. Cuzin and Jacob (57) investigated theeffect of a period of seven generations ofgrowth at an elevated temperature on anF'lac+ E. coli strain during which chromo-somal replication occurred, but F factor repli-cation was blocked by using a ts mutant Ffactor unable to grow at this temperature. Thenonreplicated F factor was found to be segre-gated into only those cells to which chromo-somal DNA which had been synthesized beforeF factor replication was inhibited was also seg-regated. More recent experiments have showna similar physical association during growthand cell division between an F' factor andchromosomal DNA following inhibition of F'replication by AO (120).However, several aspects of plasmid replica-

tion and incompatibility are difficult to ex-plain on the membrane-replicon model. Oneaspect that has been emphasized (189) is thatincompatibility of cells to a superinfecting F'element leads to the interpretation that F is

365VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

attached to its membrane-attachment site inHfr cells as it is in F+ or F' cells, yet in a

number of such Hfr strains, replication of thechromosome-F continuum is not controlled byF (31, 251). Another anomaly is that, althoughColV2 is incompatible with F and excludesautonomous F, it can stabilize in an Hfr strainwhere its replication and colicinogeny proper-

ties appear to be normal, and the transferproperties of colicin factor and chromosomeare unchanged (160). These and other objec-tions could be overcome if it were assumedthat regulation is effected through a repressor,

in an analogous way to which superinfecting Xreplication is controlled in X lysogens. Such a

system of negative control may be superim-posed on the positive control through an initi-ator. For example, Pritchard et al. (189) haveproposed that initiator is made constitutivelyby the host cell, but an inhibitor of this initi-ator is synthesized by replicon genes in a dis-continuous way, a burst being transcribed soon

after initiation of each replication cycle, inhib-iting initiation of a further cycle of replicationuntil the concentration of the inhibitor falls bydilution duegrowth.

to cell volume increase during

GENETICS AND LINKAGE OFPLASMID-BORNE GENES

In those cases where a number of geneticproperties of the same bacterial cell appear tobe extrachromosomal, as shown by their abilityto be transferred independently of chromo-somal markers both by conjugation and trans-duction and by their ability to be gained or

lost from the cell either spontaneously or after"curing," it becomes important to knowwhether these properties are carried on a singleplasmid structure or whether more than one

element is involved (106, 107, 170, 241). Thegenetic data of a number of plasmid-bornegenes are shown in Table 1, separated into twoclear-cut groups. To avoid the structural impli-cations arising from previously used termssuch as "associated" and "dissociated," whichhave been used in some instances (52, 111, 200)to refer to what will later be inferred to becointegration and recombination release of twoplasmid-gene groups, and in another instance(9) to designate what may well be a completelydifferent mechanism that leads to conjugalcotransfer of two plasmid-gene groups, I havedenoted these plasmid-gene groups as either"cointegrates" or "aggregates."

TABLE 1. Transfer and loss of plasmid genesa

Plasmid Conjugal cotransfer Coelimination or loss Cotransduction'

PLASMID CO-INTEGRATES

F'-Lac 100%(60/60)F+Lac+ 100%(40/40)F Lac -0%(0/60)F+Lac- (45) 0%(0/40)F-Lac- (46) ?0%(0/60)F Lac- 0%(0/40)F -Lac

ColV2 or ColV3 100%(151/151)Col-TF+(Col +TF I 0%(0/151)Col-TF- (45) 100%(20/20)Col -TF (160) ?

0%(0/151)Col'TF-F ColB ColV trp >90%(F ColB ColV Trp >90%(F ColB ColV Trp c.90%(F ColB ColV Trp.

cys Cys) (84) Cys) - (84) Cys)+ (84)222/R4 100%(RTF ,SUrSm,Cmr, 100%(RTF SusSm',Cm', 100%(RTF+,Sur',Smr,(Su,Sm,Cm,Tc)r TcO) (235) TcO) (236) Cmr,Tcr) (237)

A-T 100% A-T5 (10) ? 100%(22/22)A+T R (10)PII (pen-mer-ero) N.A. c.90% pen', mer', ero' (181) c.99% (4875/4880)pen',

ero' (181)

PLASMIDAGGREGATES

F,ColE1 100%F-El+ (43) <0.01%F-E1- (0/290) (46) ?(2%F+E1-;2%F-E1l+) 100%F -E1l (290/290)

F,ColE2 1%F+E2+99%FE2- (43) ? ?(0.1%F -E2-)

\,S 1%AXS R

99% \-SS (9) ? 0%(29/29)A S R (10)(0.1%AS R)

AA 4%1 AA96%A\As (9) ? 0%(16/16)A\AR (10)(0.4%-A R)

a TF represents transfer factor activity of colicin factor; RTF (or A) represents transfer activity of R factor; paren-theses in conjugal cotransfer of plasmid aggregates are results of interrupted matings.

hAll transductions using P1.

366 CLOWES


mbr.asm

.org/D

ownloaded from




In the group of plasmid genes termed cointe-grates, conjugal cotransfer and coelimination(either spontaneous or by curing) is usuallyseen, although partial transfer or partial loss isoccasionally observed. (In this group, cotrans-duction is also the rule rather than the excep-tion. It is only expected between pairs of genessituated near enough on the same DNA mole-cule to be incorporated in the same bacterio-phage particle, but when it does occur, it hasbeen taken as the most reliable genetic crite-rion for the location of two plasmid genes onthe same molecular structure. However, inthose cases where the DNA content of aplasmid is considerably less than the DNA ofthe transducing phage genome, it has beensuggested that the DNA of another physicallydistinct plasmid may be incorporated withinthe same transducing phage particle [102].This conclusion, if validated, would indicatethat cotransduction of plasmids cannot beconsidered as absolute evidence of linkage.)

In contrast, the genes shown in the group of"plasmid aggregates" are typically transferredindependently by conjugation, are also lost in-dependently, and cotransduction has not beenobserved. Within this group are included ex-amples of what would normally be regarded asindependent pairs of plasmids, such as F andColEl, which, although found independently innature, can be established in the laboratory inthe same host cell. In addition, less clearly de-fined and naturally occurring systems such asAS and AA of Anderson and Lewis (9; seep. 382) are included. In the plasmid aggregategroup, conjugal cotransfer can occur, even withhigh frequency. For example, after a normalmating procedure with F+ColE1+ donors, al-most 100% of the recipient cells acquire both Fand ColEl (43). Only if resort is made to inter-rupting the mating after short periods of con-tact (data shown in parentheses in Table 1)can cells be found to which either one or otherfactor has been independently transferred (43).Nevertheless, treatment of donor cells with AOgives rise to an almost complete elimination ofthe F factor without any observable curing ofthe ColEl factor (46), indicating physical inde-pendence of the two plasmids. Thus, during Ftransfer, the probability of independent ColEltransfer during the same period of mating isvery high. Conjugal cotransfer can thereforealso be seen to be an unreliable criterion ofgenetic cointegration, although this implica-tion has frequently been drawn (106, 169, 170,241) or, a less clearly defined structural asso-ciation has been inferred (5, 7, 9). However,just as chromosomal mobilization by ColIbmay occur without ColIb integration in the

chromosome (45), so transfer of ColEl by ColIb(43, 223) or by F may be brought about by amechanism which may not require direct phys-ical interaction between the two genetic ele-ments.The behavior of the plasmids F and ColE2

are more typical of plasmid aggregates. ColE2is not transferred to recipient cells at anythinglike the high frequency of ColEl, so that afterthe normal 2-hr period of conjugation, al-though about 99% of the recipients acquire theF sex factor, only about 1% in addition acquireColE2, and it is again necessary to resort tointerrupted mating in order to isolate thesmall number of cells (0.1%) that acquireColE2 alone (43). Some of the R-factor systemsinvestigated by Anderson and co-workers, forexample those involving the transfer factor Aand the drug-resistance markers S or A (9),show a close parallel with F+ColE2+ donors inconjugal transfer. From A+S+ donors, most ofthe recipients receive only the transfer factorA, a small number acquire in addition thedrug-resistance marker S, and only with inter-rupted mating is transfer of S in the absenceof A observed. AA behaves similarly to ASwith a slightly more efficient cotransfer of Aand A (9).

Circular genetic maps have been proposed toaccount for the data of some members of theplasmid group referred to as "plasmid cointe-grates." Watanabe (230) suggested a circularlinkage map for the R factor 222 (Fig. 1) as onestructure to satisfy the results of transductionstudies with phage P22. Similar maps havebeen proposed for the composite F-ColV-ColB-trp-cys plasmid by Fredericq (84) and for thePI penicillinase plasmid in Staphylococcusaureus by Novick (179; Fig. 2), both from theresults of deletion analyses. In each case, thegenetic data are consistent with the integra-tion of a group of genes within a single molec-ular structure. However, there has been somecontroversy as to whether the genetic nature ofthe plasmid aggregates, particularly some ofthe A-mediated drug resistance factors investi-gated by Anderson and co-workers (7, 233,238), can be similarly represented.

MOLECULAR MODELS FOR PLASMIDSAND THEIR IMPLICATIONS FOR THE

EVOLUTION OF R FACTORSIt has often been implicitly and explicitly

assumed that the origin and evolution of Rfactors is equivalent in many ways to that ofF' factors (63, 231, 233). A transfer factor -wasproposed to integrate in the bacterial chromo-some (as in the formation of an Hfr strain[209]) near to a chromosomal gene mutated to

367VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

SULFANILAMIDE

STREPTOMYCIN

CHLORAMPHENICOL TETRACYCLINE

RTF

FIG. 1. Circular genetic map of the 222/R factorproposed by Watanabe (230) on the basis of trans-duction data.

B

As

mer

FIG. 2. Proposed structure of the PI penicillinaseplasmid from Staphylococcus aureus (179). Thegenes pen, ero, bis, asa, asi, and mer control resist-ance to penicillin, erythromycin, bismuth, arsenate,arsenite, and mercury, respectively, and the seg genecontrols segregation. Mutants were mapped as

shown by genetic deletion analysis. The contourlength of the intact plasmid measured 9.4 Am, andthe deletion mutants having lost the BCD and ABCregions have contour lengths of 8.1 /um and 7.3 Am,respectively.

drug resistance. On release of the factor fromthe chromosome, this adjacent bacterial regionwould be incorporated as part of a newly cre-

ated R factor. In this way, a factor could buildup resistance consecutively to a number ofdrugs. Although the genetic evidence of theplasmid cointegrate group is not inconsistentwith this idea, there are several theoreticalconsiderations which speak against this "genepick-up" model. Firstly, drug-resistance genesmap at well-scattered sites around the E. coliand Salmonella chromosomes (207, 225). Incor-poration of one genetic segment would favorsubsequent crossing-over within this relativelyextensive region of homology and therefore belikely to reduce integration at a different partof the chromosome to permit pick-up of asecond segment. This model would be validonly if it were assumed that each of the drug-

resistance markers is derived from one of anumber of different host strains, none of whichhave extensive genetic homologies. Anotherconsideration which speaks against the "genepick-up" model is the fact that the modes ofaction of R-factor drug resistances are oftenfunctionally distinct from those of chromo-somal resistance; R-factor resistances ofteninvolve antibiotic-inactivating enzymes (62,63), whereas the corresponding chromosomalresistance comes about by a modification ofribosomal structure (245). More importantly,R-factor resistances function in a dominantfashion, indicating that in general the genesfor sensitivity and resistance may be presentsimultaneously (233). This would not be neces-sary had the drug resistance arisen de novo onthe chromosome of the haploid bacterium andfurthermore, in many instances, chromosomaldrug resistance (e.g., to streptomycin) is infact known to be recessive (153).One alternative to the "gene pick-up" model

would propose that extrachromosomal DNA iscommonly present in many bacterial species inthe form of various independently replicatingand autonomous replicons. Some of these rep-licons would take the form of transfer factorssuch as F. Others would take the form of non-infectious plasmids having colicin activity, asfor example, ColEl. Still others might carrygenes determining resistance to either anti-biotics or heavy metals, or would be capable ofmutation to this resistance. Since many non-infectious plasmids are conjugally transferredwhen coexisting in the same cell as one of anumber of infectious plasmids (9, 43, 187, 221),the presence of an independent transfer factorand an independent noninfectious drug-resist-ance factor in the same cell would be likely toendow the cell with the properties of an R fac-tor, i.e., infectious drug resistance (see Fig. 20).In some cases, genetic homologies, illegitimatepairing (82), or X-integrase-like activity mightpermit subsequent integration of these repli-cons within each other to form composite rep-licons with the structure of plasmid cointe-grates, but in the absence of cointegration theproperties of such strains would be expected tobe similar to those denoted for the grouptermed "plasmid aggregates." The model ofthe "plasmid aggregate" requires that the non-infectious plasmid should have the propertiesof a replicon, and it is not likely, therefore, tohave arisen by the segregation of a chromo-somal segment previously incorporated bygene pick-up. Thus, although the plasmidcointegrate model permits consideration of R-factor evolution via either chromosomal pick-

368 CLOWES


mbr.asm

.org/D

ownloaded from




up or by mutation of extrachromosomal ele-ments (followed by their mutual cointegration),the plasmid aggregate model would favor anextrachromosomal origin almost exclusively.

PHYSICAL COMPOSITION AND FORMThe probability that plasmids are composed

of DNA was first concluded from the ability ofF factors, R factors, and Col factors to be inac-tivated by the decay of radioactive 32P incor-porated in the host cell (151, 217, 243). How-ever, the first direct demonstration of thephysical nature of bacterial plasmids, layingthe basis for future quantitative studies, camefrom the work of Marmur and colleagues (164).In this work, an F'lac+ element originating ina strain of E. coli was transferred to a strain ofSerratia marcescens, in which the lactose-fer-menting activity associated with the F factorwas stably inherited. Isolation of the DNAfor example, ColEl. Still others might carryby analytical density-gradient ultracentrifu-gation in CsCl showed a major peak of DNAcharacteristic of the density of the S. marces-cens chromosomal DNA (1.718 g/cm3, corre-sponding to a base composition of 58% guanineplus cytosine [GCd), with a shoulder at a den-sity corresponding to that of E. coli DNA(1.709 g/cm3 = 50% Gd). The shoulder was notvisible in the parental strain of S. marcescensbefore transfer of the F factor. Later workdemonstrated a similar F'lac transfer tostrains of Proteus (in which the GC ratio andhence the density of the Proteus chromosomalDNA [1.699 g/cm3 = 38% GC] were furtherremoved from those of E. coli and of the F'factor), leading to the identification of the F'factor DNA as a separated satellite peak (77).The satellite peak was no longer present if theF' factor were cured by AO (see Fig. 3). Workwith the F' sex factor was later extended todemonstrate the presence of similar satellitebands in DNA of Proteus strains to which oneof several R factors had been transferred (74,201). In some cases, a complex R-factor bandwas seen with peaks at two different densities.In these experiments prior to 1967, the tech-nique used to isolate DNA from the donorstrain led to its breakage into fragments with amaximum size about 10 million daltons (162).The double-peaked R satellite band, therefore,might have arisen either as a result of the in-dependent transfer of two plasmids of distinctand different base composition or, alterna-tively, by the fragmentation of a largerplasmid greater than 10 x 106 daltons and of aheterogeneous base composition. Since F-fac-

tor DNA and X DNA are known to be com-prised of segments with different DNA baseratios (164, 219) as distinct from the morehomogeneous base-ratio distributions amongfragments of virulent phage and bacterial DNAspecies (163), no clear-cut distinction couldbe made between these two ideas.The various configurations taken up by DNA

molecules, which has formed the basis formuch of the structural studies of bacterialplasmids, were previously well investigatedand characterized by Sinsheimer and co-workers (27, 79) from studies with the double-stranded replicative form of the single-stranded phage, OX174, and by Vinograd andco-workers (229) in their experiments withpolyoma virus. They have shown (Fig. 4) thatDNA may take the form of (a) a linear, double-stranded (duplex) molecule which can close

(a) HOSTR DNA

DNAREFERENCE R-DNAi

Z W~~~~~~~~~~~~~~~~~~~~~~~~

O R++D ~ ~ ~ ~~~~~~~~~~~~~~~~~~~~ICl)

>~~~~~~~~~~~~~~

DENSITY ( g cm 3)FIG. 3. Diagrammatic representation of density

tracings of ultraviolet adsorption photographs afteranalytical, density-gradient profiles of DNA fromProteus mirabilis, together with a reference DNA ofBacillus subtilis phage SPOI, of density 1.742 glcm'.a, DNA from Proteus host at density 1.699 g/cm3; b,DNA from the same host strain harboring an F'factor; c, DNA from the host strain after the F'factor has been cured by acridine orange.

VOL. 36, 1972 369


mbr.asm

.org/D

ownloaded from



CLOWES

NEUTRAL

( C )

(b)

(a)

BACTERIOL. REV.

4L K A LINE

COVALENTLY-CLOSED CIRCULARDUPLEX(1.42)

1t

OPEN-CIRCULAR CIRCULARDUPLEX SINGLE-STRANDED

(1.14)4 _ _ _ _ _ _

coacooxcuLINEARDUPLEX

LINEARSINGLE - STRANDED

(3 to 4)

I ,,

( 1.28 )

(1|.14 )(1.00)

FIG. 4. Diagrammatic representation of the different configurations of DNA molecules. The relative sedi-mentation coefficients of the different DNA molecules of polyoma virus in neutral and alkaline sucrose gra-dients are shown under the structures (193).

upon itself to form a cyclic (circular) duplexmolecule either (b) by joining one of the DNAstrands, leaving the other with a break, or (c)through joining of both strands. Since bothstrands of this latter structure are covalentlybonded, it is referred to as a covalently closedcircular (CCC) duplex molecule. When isolatedextracellularly, this molecule takes the form ofa "supercoil," the DNA helix being twistedaround itself. The circular form (b), with onlyone of its DNA strands joined, is termed some-what confusingly an "open-circle" duplex be-cause its conformation is open rather thansupertwisted, a "nicked circle" because one ofthe two strands of the DNA duplex is brokenor "nicked," or a "relaxed" circle because it isnot supercoiled. The sedimentation values ofthese molecular forms of polyoma virus were

carefully measured by Vinograd and colleaguesin neutral solutions of either salt or sucrose togive the relative values shown in Fig. 4. Theratio of the sedimentation coefficients of theopen circular to linear duplex (1.14) and that ofthe covalently closed to the open circular form(1.25) can be used in identifying each specificpeak.Under alkaline conditions, which lead to

breakage of the hydrogen bonds betweenthe two strands of a DNA duplex, the strandsof the linear or open circular duplexes are sep-

arated, but the strands of the supercoil re-

main interlocked. Thus, when centrifuged un-

der alkaline conditions, each structure forms a

compact random coil with relative sedimenta-tion coefficients as shown in Fig. 4, the co-

valently closed circular (CCC) form being a

far more rapidly sedimenting molecule.The sedimentation coefficients of similar

forms of other molecules may be calculatedfrom their molecular weights by use of for-mulas derived by the empirical fit of a numberof experimental values.That of the linear duplex is derived from

So 20,W = 2.8 + 0.00834 M0 479 (Svedberg) (87);

the open circular duplex varies as

So20,W = 2.7 + 0.01759 M0 445 (Svedberg) (124);

and the covalently closed duplex varies as

S°210 = 7.44 + 0.00243 M0 58 (Svedberg) (124).

Using these three formulas, the ratios of thesedimentation coefficients can be seen to varywith molecular weight (Fig. 5).

Vinograd and his co-workers also investi-gated the effect of the dye ethidium bromideon these structures (13, 192). Ethidium bro-mide is known to intercalate between adjacentbase pairs of DNA, and in so doing the helixpitch is changed and the DNA duplex is ex-

tended, thus lowering its density. Intercalationof ethidium bromide occurs to the same extent

370


mbr.asm

.org/D

ownloaded from




*1.7

0 1.6 -

o SLINC0 '.5-

C

E 1.4 Sc

C,)~~~~~~~~SC0

0

SLIN

2 3 4 56 7 89 20 30 40 50 6070 8090

10 100Megadaltons ( Mdals)

FIG. 5. Ratios of sedimentation coefficients of configurational forms of duplex DNA as a function of molec-ular weight. The values S,,,, SOc, and SLIN, respectively, represent the S20 w values in Svedberg units forthe covalently closed circular, open (relaxed or nicked) circular, and linear duplex DNA molecules calculatedfrom the empirical formulae. Sccc = 7.44 + 0.00243 M0 58 (124); Soc = 2.7 + 0.01759 M0"445 (124); SLIN = 2.8+ 0.00834M479 (87).

in the open circular duplex or in the linearduplex, and this results in the same decreasein density (0.125 g/cm3) of both structures. Incontrast, intercalation of ethidium bromideinto CCC DNA is limited because of the re-striction of the rotation of the two strandsabout each other, due to the absence of a freeend of rotation. Thus, in the presence of ethi-dium bromide, the extension of CCC DNA isless than that of the other two duplex forms,and its density is thus lowered to a lesserextent (0.085 g/cm3). Hence in ethidium bro-mide, the CCC form is denser (0.125 - 0.085= 0.04 g/cm3) and can be separated from theother two double-stranded forms even in theabsence of any differences in base ratio.

METHODS OF ISOLATION

Differences from Host DNA Base Ratio

Host strains of Proteus, with a chromosomalGC content of 38% (density 1.698 g/cm3), havebeen frequently used for the isolation ofplasmid DNA species, which have GC contentsof between 45 and 55%, corresponding to den-

sities from 1.704 to 1.718 g/cm3 (see Tables 4and 6). Thus, plasmid DNA can be readily re-solved from Proteus DNA and seen as a sepa-rated satellite peak or band after analyticaldensity-gradient ultracentrifugation in cesiumchloride solution (Fig. 6 and 7). Largeramounts of plasmid DNA may similarly beseparated by preparative centrifugation,usually in a fixed angle rotor at 95,000 x g for60 hr at 25 C. The contents of the centrifugetube can then be fractionated by collectingdrops through a hole punctured in the bottom.The DNA is assayed either by its ultravioletadsorption at 260 nm or by measuring atritium or "IC-label incorporated in the DNAby the addition of radioactive thymidine (orthymine) to the culture during growth (Fig. 8).Further purification and enrichment can beachieved by column chromatography withmethylated albumin kieselguhr (MAK), whichfractionates by both molecular weight and basecomposition, lower GC DNA, and higher molec-ular weights eluting with increasing ionicstrength (88). Plasmid DNA tends to elutewith the first samples (73).

Early studies on the DNA of bacterial plas-

371VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

I.742' 1.698[1.709

FIG. 6. Microdensitometer tracing of ultravioletadsorption photograph after analytical ultracentrif-ugation for 27 hr at 44,000 rev/min in a cesium chlo-ride density gradient of DNA from a Proteus (R15)+host strain (1.742 g/cm3 is the density of a referenceDNA [Bacillus subtilis phage SPOJ DNA]). Themajor peak at 1.698 g/cm3 represents the chromo-somal DNA of the Proteus host strain (176).

mids used the technique of Marmur (162) forDNA isolation which leads to its breakage intofragments of approximately 10 million molec-ular weight. A more gentle method of DNAextraction, applied by Helinski and co-workers(115), consists, first, in the formation of spher-oplasts by the addition to the bacteria of amixture of lysozyme, ribonuclease, and ethyl-enediaminetetraacetic acid (EDTA) in sucrose,followed by lysis with an ionic detergent, e.g.,sodium lauryl (dodecyl) sulfate (SLS = SDS),after which protein is removed by extractionwith carefully buffered phenol. After removalof phenol by dialysis, the DNA is centrifugedas above. Preparations adequate for configura-tion studies and molecular weight assays maybe obtained from this improved method ofDNA isolation by using centrifugation only,without recourse to column chromatography.Molecules of over 100 x 106 molecular weighthave been isolated by this method (see Tables3 and 4).

From Minicells

Several E. coli mutants have been identifiedas having an abnormal cell division that re-

1j742jb1.699 L-1.708

L1.717FIG. 7. Microdensitometer tracing of ultraviolet

adsorption photograph after analytical untracentrif-ugation for 27 hr at 44,000 rev/min in a cesium chlo-ride density gradient of DNA from a Proteus (222)+host strain (1.742 g/cm3 is the density of a referenceDNA [Bacillus subtilis phage SPOJ DNA]). Themajor peak at 1.699 g/cm3 represents the chromo-somal DNA of the Proteus host strain (176).

0.4

o V3E0(D

z 0.20

a-0a,)m 0.1I

0 10 20 30 40 50FRACTION NUMBER

FIG. 8. Ultraviolet absorption (260 nm) of frac-tions obtained after centrifugation of DNA from aProteus (222)+ strain in cesium chloride in a prepar-ative ultracentrifuge for 60 hr at 95,000 x g (176).

372 CLOWES


mbr.asm

.org/D

ownloaded from




sults in the segregation of progeny cells whichdo not contain any appreciable amounts ofchromosomal DNA. One such mutant of E.coli K-12, P678-54, has been extensively inves-tigated. Its DNA-less progeny, termed "mini-cells," are produced by a variety of growthconditions in a ratio to normal cells of up to3: 1. Minicells are about one-tenth the volumeof normal cells and contain no detectable DNAand greatly reduced amounts of three presum-ably DNA-associated enzymes (ribonucleicacid [RNA] polymerase, DNA methylase, andthe photoreactivating enzyme); however, nearnormal amounts of DNA polymerase I arepresent (3, 47). Minicells can be separatedfrom normal-sized parental cells by differen-tial centrifugation, followed by two successivesucrose-gradient centrifugations (3) or bygrowth in penicillin (155). The fraction of pa-rental cells is thereby reduced to less than 1 to106 minicells, which are concentrated as highas 1011 per ml. Minicells may be lysed bystandard methods effective with normal cells.It has been shown that the DNA of many bac-terial plasmids such as ColEl (127, 129), ColB(195), R1 (157), R222 (128, 158), N3 (156, 157),and R64 (156, 157, 195) is segregated effi-ciently into minicells, almost all minicellsharboring one or more plasmid copies. In con-trast, F segregates very inefficiently to lessthan 1% of minicells (140). However, minicellsderived from an F- parental strain can act asefficient recipients for the conjugal transfer ofplasmids from normal-sized donor cells. F anda variety of F' factors can be transferred, al-though only a limited amount of chromosomaltransfer occurs, and even DNA transfer of theF' factor is limited, only shorter F' factors beingtotally transferred (47). Minicells harboringsex factor plasmids have also been shown to beefficient donors (158, 195). Thus, the DNA ofexconjugant or segregant minicells is almostentirely plasmid DNA and therefore providesan efficient biological separation of plasmidfrom chromosomal DNA.

Lysis with Nonionic Detergent (Brij 58)Several methods of gentle lysis of bacterial

cells have been developed, during which chro-mosomal DNA remains attached to a cellularcomponent during low-speed centrifugation.Helinski and his co-workers (36) have appliedone of these techniques using the nonionic de-tergent, polyoxyethylene cetyl ether (Brij 58;reference 96), to the isolation of plasmid DNA.They have found that, when cell spheroplastsprepared in the usual way in buffered sucrose

and lysozyme are lysed by the addition of Brij58 in the presence of sodium deoxycholate(DOC) and EDTA, much of the plasmid DNAremains in the supernatant fluid, even thoughthe bulk (c. 95% of a radioactive thymidinelabel) of the chromosomal DNA continues tobe sedimented by low-speed centrifugation. Itmay be concluded that, since a decrease inconcentration of Mg2+ permits molecules ofincreasing size to be released (96), plasmidDNA might be released in the complete ab-sence of Mg2+ when the membrane is suffi-ciently disorganized, although chromosomalDNA would remain associated with the mem-brane. Alternatively, since DOC dissociatesmembrane lipoprotein or lipopolysaccharidecomponents (96), plasmid DNA may be boundby a DOC-sensitive attachment to the mem-brane. The supernatant fluids from such Brij-lysed preparations have been termed "clearedlysates" and have been extensively used forthe separation of plasmid DNA. This tech-nique has also been adapted for the isolation ofplasmid DNA from S. aureus by using theenzyme lysostaphin in place of lysozyme (179).

Conjugal TransferA technique for specifically labeling plasmid

DNA was developed by Freifelder and Frei-felder using a system in which only those DNAmolecules transferred to a recipient cell areable to incorporate radioactive label (90). Theconjugal system uses an E. coli K-12 donorstrain which cannot incorporate exogenouslabeled thymine due to a genetic block, and arecipient which is prevented from replicatingits own DNA due to a large number of lesionsproduced in the DNA by a prior heavy dose ofultraviolet light. During conjugation, theplasmid is transferred to the ultraviolet-killedrecipient with near normal efficiency and canreplicate in the irradiated cells. During thisreplication, it can take up labeled thymine,and its DNA can therefore be identified andmeasured in recipient cell lysates. A modifica-tion uses a thymidine kinase-deficient (tdk)mutation in the recipient strain, preventingincorporation of exogenous thymidine to lessthan 0.1% of that by the tdk+ donor strain.Thus, specific labeling of DNA in the donorwithout inhibiting DNA synthesis in the recip-ient is possible, and the transfer of a labeledplasmid can be followed (228). The use of arecipient strain, carrying a mutation to resist-ance to phage T6, and a T6-sensitive donorstrain allows selective lysis of donor cells, afterwhich recipient cells can be lysed by lysozyme

VOL. 36, 1972 373


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

and detergents in the usual manner.

Isolation of CCC Plasmid MoleculesEarly experiments separating plasmid DNA

by virtue of its difference in base ratio fromhost DNA (49, 115, 176, 197) had consistentlyindicated a proportion of molecules with "su-percoiled" CCC structures. It is thus possibleto apply techniques developed by Vinogradand others for the isolation of molecules ofsimilar configuration from polyoma virus tostudies of plasmids.

Cellulose nitrate adsorption. If a mixtureof chromosomal and plasmid DNA is subjectedto shearing forces, conditions can be chosensuch that the chromosomal DNA can be frag-mented, whereas plasmid DNA (because of itssmaller size and CCC configuration [79]) isleft intact. (Shearing may be accomplished bydrawing the DNA through a capillary pipette[e.g., 0.1 ml] or fine-needled syringe [14] or byblending or vortexing [89].) If such a mixtureof sheared chromosomal and plasmid DNA isbriefly exposed either to high temperature orpH (>11.5), the hydrogen bonds between thestrands of the DNA duplex are broken and thestrands of the linear (chromosomal) DNA areseparated (denatured), leaving the CCCplasmid DNA strands interlocked. [If thetemperature is rapidly reduced, or if the alka-linity is maintained, the strands of linear DNAremain separated, although with this tempera-ture reduction, the interlocked CCC strandscan reform the duplex. If the pH is returned toneutrality, or if the preparation is allowed tocool down slowly, the double-stranded linearstructure can be reformed (renaturation, reas-sociation, annealing).] Since single-strandedDNA will bind to nitrocellulose whereasdouble-stranded DNA will not (182), thesingle-stranded chromosomal DNA can beseparated from the plasmid DNA duplex, ei-ther by passage through a nitrocellulosecolumn (228) or by filtration through a cellu-lose nitrate membrane (250). This method hasbeen used to separate a plasmid DNA that hasa higher GC composition than its host andwhich would, even in the linear form, havebeen more stable to denaturation (250). It hasalso been used in early studies separatingplasmid DNA from S. aureus (205).Alkaline sucrose sedimentation. DNA re-

leased by lysis with lysozyme and detergent(SDS) is first sheared to degrade the chromo-somal DNA into fragments small enough so asnot to interfere with separation by centrifuga-tion. After shearing, the degraded lysate is

denatured in 0.3 M NaOH, layered on a 5 to20% sucrose gradient containing 0.3 M NaOH,and sedimented to separate the CCC DNA asa peak which sediments three to four timesmore rapidly than linear or open circular (OC)DNA (see Fig. 4).Ethidium bromide-cesium chloride cen-

trifugation. A crude cell lysate made by theaction of detergent on spheroplasts is shearedas above and is added directly to a mixture ofcesium chloride and ethidium bromide. SinceSDS is insoluble in cesium chloride, the deter-gent used is sodium sarcosinate (Sarkosyl) (14).The mixture is immediately centrifuged, re-sulting in the separation of the CCC DNA as adenser band below the less dense open circularor linear DNA, the bulk of which is chromo-somal DNA (see Fig. 9).

METHODS OF MOLECULAR WEIGHTDETERMINATION

Sedimentation AnalysisThe molecular weights of plasmid DNA iso-

lated by methods described above can be esti-mated from the sedimentation coefficients of

100-^

50

>~ E

) 10 2o :30 40 50 60

Fraction numberFIG. 9. Radioactivity of fractions obtained after

ultracentrifugation in ethidium bromide-CsCI solu-tion of crude lysates of E. coli strains grown in radio-active thymidine. From an R- E. coli host (0); fromthe same host carrying an R factor (0) (177).

374 CLOWES


mbr.asm

.org/D

ownloaded from




the various configurational forms in neutralsucrose gradients. A mixture of plasmid DNAin the CCC, OC, and linear double-strandedforms would sediment in neutral sucrose withsedimentation coefficients in the ratios shownin Fig. 5. The separation of a CCC peak char-acteristic of plasmid DNA can be achieved inthe presence of chromosomal or nonplasmidlinear fragments. The OC form can alsousually be identified as a sharp peak, butlinear plasmid DNA is usually poorly resolvedand is not normally isolated by most tech-niques. In practice, therefore, two peaks (CCCand OC) with relative sedimentation coeffi-cients of 1.33 to 1.7: 1 (depending on molecularweight of plasmid, Fig. 5) are usually observed(Fig. 10). (In those cases where deviation fromthe expected ratio is observed, the quantita-tion below must obviously be viewed with cau-tion.) The identity of the CCC and OC peakscan be confirmed by taking advantage of thefact that a single scission or nick in just one ofthe strands of a CCC molecule produces anOC molecule (Fig. 4). This may be done by oneof several methods, including limiting deoxyri-bonuclease treatment (76, 79, 128, 229) or X-irradiation (85), and leads to a decrease in thesize of the CCC peak and a corresponding in-crease in the size of the OC peak which arethereby defined. The absolute value of the sed-imentation coefficient for the OC peak may bedetermined by cosedimentation of a carefullypurified DNA standard of known sedimenta-tion coefficient and by application of the for-mula, S1IS2 = D1/D2 (26), where S1, S2 arethe sedimentation coefficients and D1, D2 thedistances of the peaks from the meniscus ofthe OC and standard peaks, respectively. Thesedimentation coefficient (53) of the linearform of the plasmid is then approximated bydividing the calculated value of S1 by thevalue expected from the ratio OC: linear at therough value of molecular weight (Fig. 5). If themolecular weight of the standard DNA isknown, the molecular weight of the linear formof the plasmid may now be estimated accord-ing to the formula S2/S3 = (M2/M3) 0.38 (26),where M2, M3 are the molecular weights ofstandard and linear form, respectively. Expo-nent 0.38 is an empirical figure derived from anumber of carefully measured values of M, andis a revision (87) of the previous figure of 0.35derived by Burgi and Hershey (26) and used inmost calculations referred to in this review.

Loss of Supercoil Structure by IrradiationAs noted above, a single nick in just one of

the strands of the duplex is sufficient to con-vert a CCC into an OC molecule. Freifelderapplied this technique to obtain some of theearliest measurements of the molecularweights of the CCC molecules of F and F' fac-tors by measuring the relative losses of CCCDNA and corresponding gains in OC DNA in aseries of DNA samples exposed to increasingdoses of X-irradiation (86). The logarithm ofthe fraction of surviving CCC molecules is alinear function of the irradiation dose, theslope of the survival curve being proportionalto the molecular weight. Freifelder estimated

B4

I 50FRACTIONI

II 1.- Do .la

I* - DC el

FIG. 10. Diagrammatic representation of a neutralsucrose gradient profile. The radioactive DNA prep-aration, together with a differentially labeled refer-ence DNA of known sedimentation coefficient andmolecular weight, is added to the top of a 5 to 20%osucrose gradient in a centrifuge tube. After centrifu-gation for 8 hr at 75,000 x g, 50 equal-volume frac-tions taken through a hole punctured in the bottomof the tube were assayed for radioactivity. The refer-ence DNA is distinguished by its differential ra-dioactivity, and the covalently closed circular (CCC)and the open circular (OC) peaks are recognized ei-ther by their ratio of sedimentation coefficients or bya parallel experiment where the unknown DNA isexposed to limiting deoxyribonuclease action re-sulting in a diminution of the CCC and an increasein the OC peaks. The sedimentation coefficients ofthe CCC and OC peaks can be calculated by substi-tution of the distances (DO, DR, DC) of each peakfrom the top of the gradient into the formula DJ/D2= Si/S2.

VOL. 36, 1972 375


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

the molecular weights of a number of F ele-ments in this way by comparing the rates ofsurvival of their CCC molecules to that of XCCC DNA, whose molecular weight is accu-rately known (61).

Reassociation KineticsIf a preparation of DNA molecules is

sheared to produce fragments of about 5 x 106molecular weight, heat-denatured to separatethe two strands of the double helix, and thenslowly cooled, each single-stranded fragmentwill find and reassociate with its homologouscomplementary strand to form a double-stranded fragment. The time taken for thisreassociation will depend upon the probabilityof a single-stranded fragment finding its hom-ologue. This in turn will depend upon the con-centration of homologues. Clearly, fragmentsof large DNA genomes will contain less homo-logues for the same total DNA concentrationthan those of small DNA genomes, and reasso-ciation will be slower. Thus, measurements ofrates of reassociation give a measure of genomesize (24). Since the absorption of ultravioletlight by double-stranded DNA is less than thatby single-stranded DNA, the change from thesingle- to double-stranded form by reassocia-tion can be followed by ultraviolet absorptionmeasurement. The fraction of reassociatedmolecules is then plotted against the logarithmof the product of the initial DNA concentra-tion (C0) and the time (t), (i.e., Cot) measuredin moles of nucleotide per liter second.The 50% Cot value being proportional to theoriginal molecular weight is about 8 for mole-cules of E. coli (2.5 x 101 molecular weight)and about 0.1 for a plasmid of 40 x 106 molec-ular weight (24).

Electron MicroscopyIf DNA is allowed to diffuse to an air/water

interface at which there is a monolayer of aprotein such as cytochrome c, the acidicgroups of the nucleic acid molecules and thebasic groups of the protein molecules interactin such a way as to adsorb the long, thin DNAmolecule to the surface of the liquid (143). If asample is then removed from the protein mon-olayer film and examined by electron micros-copy, the DNA is found to be extended in in-dividual molecules (143, 150), in contrast tothe usual aggregated and oriented structures(148; see Fig. 11). Plasmid DNA isolated by avariety of methods consists largely of circulardouble-stranded molecules. Many of these

molecules show the expected supertwistedstructure, but some will always have an opencircular configuration (Fig. 12a). When theDNA molecules seen in the electron micro-scope are photographed and projected on ascreen, the contours of the open circular mole-cules can be traced and their lengths accu-rately measured by a map measurer (149).Most plasmid DNA is found to contain mole-cules of one size. (A particular advantage of acircular molecular species is that fragmenta-tion or degradation of the DNA results in lossof the circularity, and thus all circular mole-cules are by nature intact.) From a knowledgeof the magnification, the physical length of themolecule (1) can be determined with a repro-ducibility of about +2% (147). If the value ofthe ratio of molecular weight to length (M/A) isknown for the conditions of electron micros-copy preparation, the molecular weight of theDNA concerned can be derived. The assump-tion is made that MA is independent of thebase composition of the DNA; since the basecompositions of most plasmids do not varymore than about 45% to 55% GC (Table 4), thisassumption is not unreasonable. The most re-cent calibration for preparations in ionicstrength of 0.3 M is (2.07 ± 0.04) x 106 daltonsper gm (147). (In an attempt to relate molec-ular weight determinations from a number oflaboratories, this conversion figure will be usedfor all molecular weight calculations, eventhough many published data have used otherM/1 values.) Alternatively, standard DNAmolecules of uniform size and known mole-cular weight (e.g.. X DNA) may be added be-fore sampling and serve for calculating M (227).

SIZE AND CONFIGURATION

F and F' Sex FactorsFreifelder isolated the DNA of an F'lac

plasmid specifically labeled after conjugaltransfer and first confirmed it as being that ofthe plasmid by its ability to bind on an agar-gel column to Proteus PM1(F'kac)+ DNAwithout showing binding to DNA of the pa-rental Proteus PM1 strain, not carrying theFlac factor. The sedimentation coefficient ofthe transferred DNA, which formed a broadband on sucrose sedimentation and was there-fore assumed to consist of both OC and linearmolecules, was measured by cosedimentationwith a pure nonfragmented preparation of T7DNA. Calculation by application of the for-mula above gave a molecular weight of 55 i 8

376 CLOWES


mbr.asm

.org/D

ownloaded from




FIG. 11. a, Electron micrograph of DNA dried onto a specimen grid and shadowed with platinum. Lateralaggregation and preferential orientation in one direction are typical drying artifacts. The bar represents 1Mm. x51,000 (148). b, Electron micrograph of double- and single-stranded DNA prepared on the same gridby the diffusion method (148). Single-stranded DNA has less contrast and is more kinked. Bar indicates 1um. x57,000 (148). (Electron micrographs kindly provided by D. Lang.)

VOL. 36, 1972 377


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

FIG. 12. Electron micrographs prepared by the diffusion method of R-factor DNA taken from satelliteband after dye-buoyant density centrifugation of E. coli R+ cultures. a, Nicked (open) circular and super-twisted forms of R6K (12.8 Am) are shown. Bar indicates 1 Am. x 27,000 (145). b, Electron micrograph of R6Kdimer. Catenated dimer consisting of one supertwisted monomer interlocked with nicked circular monomer.Bar indicates 1 Am. x 27,600 (145).

378 CLOWES


mbr.asm

.org/D

ownloaded from



MOLECULAR STRUCTURIE OF BACTERLAL PLASMIDS

Mdal [molecular weights will be expressed indaltons x 106 (megadaltons Mdal) I for F'lacand a preliminary estimation of the F factor of35 7 Mdal (91). Later experiments (85) ex-

ploited the separation of CCC DNA of F' ele-ments by alkaline sucrose and ethidium bro-mide-cesium chloride sedimentation. X-irra-diation inactivation curves of such CCC DNAusing X CCC DNA as a standard resulted inthe estimation of the molecular weights of a

series of F' elements (see Table 2).Bazaral and Helinski later employed Sar-

kosyl lysis, followed by ethidium bromide-ce-sium chloride centrifugation to isolate F-DNA which sedimented in sucrose gradients intwo peaks at 80S and 48S (15). Assuming thatthese two peaks corresponded, respectively, tothe CCC and OC forms (ratio CCC/OC = 1.66),a molecular weight of 75 Mdal was estimated,using the 48S value. Electron microscopy of F-DNA isolated by the same procedure by sev-

eral workers and measured by contour lengthhas led to a molecular weight of 61 to 64 Mdal(44, 144, 188). F-DNA also isolated as a CCCmolecule after conjugal transfer and measuredby electron microscopy using an internal X

DNA standard was calculated as 62 Mdal (227)The electron microscopic measurements of

the contour length of F from at least fivesources in four laboratories have led to remark-ably similar results of contour length measure-

ments of 30.8 to 31.7 gm, conforming to a mo-

lecular weight of 64 Mdal, if an M/A ratio of2.07 Mdal per Jim is assumed (Table 2).

F'lac DNA (165) has also been isolated by asimilar procedure and measured by electronmicroscopy.

Colicinogenic PlasmidsDeWitt and Helinski (67) showed that DNA

from a strain of Proteus harboring the ColElfactor had a satellite band at a density of 1.710g/cm3 which was not present in the parentalstrain. Other attempts to isolate Col factorDNA from Proteus have been unsuccessful,probably owing to the instability of the plas-mids in Proteus and the lack of a selectivemechanism. (It has recently been claimed thatI-like R factors [but not F-like or other non-I-like fi- factors] cannot be transferred to Pro-teus strains [59].) The ColEl plasmid from P.mirabilis was later isolated and identified by a

technique including alkaline denaturation ofthe DNA. This was followed by CsCl centrifu-gation and electron microscopy of the satelliteDNA to show circular molecules of 2.3 gm and4.7 jim in a 2: 1 ratio (and a number of su-

pertwisted CCC molecules) corresponding tomolecular weights of 4.5 and 9.2 Mdal for themonomeric and dimeric forms of ColEl, re-spectively (198). (Using the corrected value ofM/1, the monomeric form corresponds to 4.8Mdal [Table 3.]) Separation of ColEl from E.coli by the ethidium bromide method led tothe isolation of the DNA as a 23S CCC and a17S OC form corresponding to a molecularweight of 4.8 Mdal as previously found. In con-trast to Proteus, no multimeric forms werefound in E. coli. Application of a similar tech-nique to ColE2 and ColE3 resulted in sedi-mentation peaks at 25S leading to an estima-tion of molecular weights of approximately 5Mdal for each of these factors (14). DNA fromColE1+ minicells was found by Inselburg (127)to sediment as a 24S peak, concluded to beformed of CCC molecules, since the kinetics oftreatment with pancreatic deoxyribonucleasegave rise to its exponential conversion to an18S (OC) form (127). Separation of ColEl mole-

TABLE 2. Molecular weights of F sex factors

Method of Molecular

Plasmid molecular wt wt (Mdal)b Referencedetermina-tiona Q C

F X 45 42 85S 35 34 91S 75 15EM 61 66 144EM 64 64 44('EM 62 64 227EM 64 188

F-Lacd X 74 69 85S 55 52 91EM 72 78 165

F-Gal X 51 48 85F'-Galxatt X 72 67 85

a S, Neutral sucrose sedimentation; X, X-ray inac-tivation; EM, contour length measurement by elec-tron microscopy.

b Megadaltons (Mdal) equivalent to atomic massunits x 106; Q, value quoted in reference; C, cor-rected value; using 0.38 (87) for sedimentation indexin place of 0.35 (26); using 30.8 Mdal for molecularweight of X standard (61) in place of 33 Mdal used inX-irradiation data (85) or 30 Mdal used as electronmicroscopy standard (227), and assuming MA ratiofor double-stranded DNA of 2.07 Mdal per pm (147).

(- Including unpublished data by author in collabo-ration with M. Mitani using F from E. coli K-12strain W1485F-; based on 32 molecules of 30.7 + 0.2pin, and 23 of 31.0 + 0.5 gm (two independent iso-lates).

d F-Lac from Jacob strain 200P.

VOL. 36, 1972 379


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

TABLE 3. Molecular weights of colicinogenic factors

Method of MolecularPlasmid molecular wt wt (Mdal)b Reference

determinationa Q C

ColEl EM 4.5 4.8 197EM 4.8 129

ColE2 S 5.0 6.0 14ColE3 S 5.0 6.0 14ColIb S 61.5 53.5 38

EM 68 68 44'ColB2 EM 70 70 44dColV2 EM 94 94 44FVBtc' EM 107 113 115

aS, Neutral sucrose sedimentation; EM, contourlength measurement by electron microscopy.

' Q, quoted molecular weight in reference; C, cor-rected molecular weight: electron microscopy by useof 2.07 Mdal per pm (147) in place of 1.96 Mdal perpum (161).

('Including unpublished data by author in collabo-ration with M. Mitani based on measurement of 32molecules of 31.8 + 0.4 gm, 13 of 33.8 + 0.7 pm, and6 of 35.5 ± 0.6 pm.

dIncluding unpublished data by author in collabo-ration with M. Mitani based on measurement of 29molecules of 33.9 + 0.5 ,m.

" Including unpublished data by author in collabo-ration with M. Mitani based on measurement of 4molecules of 47.0 + 0.7 pm, 7 of 46.1 ± 0.3 pm, and 4of 43.0 + 2.5 pm.

' Fredericq F* ColV ColB trp cys plasmid (84).

cules replicating in minicells was achievedby their incorporation of a heavy (5-bromoura-cil) label, and their subsequent examination byelectron microscopy led to an average contourlength measurement of 2.31 + 0.06 pm (4.8Mdal; reference 129), corresponding veryclosely to the previous measurements above.The large F ColV ColB trp cys plasmid ofFredericq (84) was isolated from a Proteus hoststrain using lysozyme, SDS, and phenol ex-traction followed by density gradient centrifu-gation in CsCl (115). A satellite band at 1.710g/cm3 was identified from which electron mi-crographs showed open and supertwisted mole-cules with contour length of 54.5 1.7 ,m cor-responding to a molecular weight of 107 (cor-rected 113) Mdal. Dye-buoyant density gra-dient centrifugation of DNA from E. colistrains harboring one of a number of colicinfactors has also been used to isolate a numberof different Col-factor DNA species in thislaboratory, the measurement of contour lengthby electron microscopy resulting in molecularweight determinations also summarized inTable 3.

Drug-Resistance Factors

Monomolecular factors. Wild-type R fac-tors have been transferred to strains of Proteusand E. coli (including minicell-producingstrains) from which the DNA has been isolatedand plasmid DNA separated by a variety ofmethods, including base-ratio differences, ethi-dium bromide-cesium chloride centrifugation,and alkaline sucrose sedimentation. The DNAwas found in all instances to be represented bycircular molecules, both CCC and OC, thecontour lengths of the OC molecules beingusually of uniform size. The densities, methodsof isolation, molecular weight determination,and the sizes of these R factor molecules aresummarized in Table 4 and range in densitiesfrom 1.704 to 1.711 g/cm3 and in sizes from 26to 78 million molecular weight.

Segregant R factors have been derived bytransfer of wild-type factors to strains of Sal-monellk typhimurium in which it has beenshown (240) that many R factors are unstableand tend to lose some, and occasionally all, oftheir drug resistances. Two such segreganttypes of 222/R4 (conferring resistance to sul-fonamides [Surn, streptomycin [Smr], chloram-phenicol [Cmr], and tetracycline [Tcr]) wereisolated, one, 222/R3(N), having lost Tcr andthe other, 222/Ri, having lost Sur, Smr, andCmr (177). In all, only one R3 segregant typewas isolated, but five independent R1 segre-gants were isolated and designated RiAthrough R1E. The factors were transferred toE. coli hosts and DNA was isolated. All segre-gants showed genetic loss to have been accom-panied by physical loss of DNA, since DNA ofsegregant factors was shorter in contour lengththan the parental 222/R4 factor DNA (Table 5).Among the R1 segregants, two groups werefound, R1A having lost more DNA than R1B,and R1C, D, and E resembling RiB in the lossof similar amounts of DNA, assumed to be dueto the same genetic deletion. R1A and RiBwere selected as typical of these two groups.Molecular recombination between these segre-gants was demonstrated in E. coli by selectingfrom a mixed culture of host strains (onecarrying 222/Ri and the other, 222/R3),"recombinant" strains carrying all four drugresistances. (Recombination between R fac-tors has previously been demonstrated geneti-cally [108] and has been shown to depend onthe rec system of the host [81].) A number ofcolonies were isolated, all of which were stableand could transfer concommitantly all fourdrug resistances. Isolation of DNA from these

380 CLOWES


mbr.asm

.org/D

ownloaded from




"recombinant" R factor strains indicated thatmolecular recombination had taken place,since the DNA of the "recombinant" R factorcomprised a unimolecular species of sizeequivalent to that of parental 222/R4 fromwhich the two segregant R factors were de-rived (Table 5 and Fig. 13).

Derivative R factors retaining transfer factorproperties, which have lost all drug resistancemarkers (and therefore inferred to be the sex-

factor element [RTF] responsible for thetransfer of drug resistance), have been isolatedfrom factors R1 and R6. That of R1 was shownto have a density of 1.709 g/cm3 (51, 103) andsedimented in sucrose with peaks of 64S and44S; the linear monomer was therefore calcu-lated as 39S, equivalent to 50 Mdal (216). TheRTF from R6 also sedimented at 1.709 g/cm3and, when isolated by ethidium bromide andmeasured by electron microscopy, its contourlength was 26 to 34 gm, the mean value of 31,m corresponding to a molecular weight of 63Mdal (50).

Staphylococcal plasmids have recently beensubjected to a number of physical studies byNovick and co-workers (179). Using a methodof alkaline denaturation and nitrocellulose

chromatography, the DNA of a wild-type peni-cillinase plasmid (PI) which also carries a

number of other genes (see Fig. 2), and that oftwo of its segregants showing genetic deletions,were shown to be circular molecules of contourlength 9.4 ,um, 7.3 Jtm, and 8.1 Atm, respec-

tively (205). Thus, the PI penicillinase plasmidis a unimolecular species of molecular weightabout 19 Mdal and undergoes deletions to pro-

duce segregants of about 17 and 15 Mdal. Twofurther independent S. aureus plasmids haverecently been examined by an adaptation ofthe "cleared lysate" method with Brij 58 (36)and have led to contour length measurementsof 1.4 gm for a plasmid (tet) controlling tetra-cycline resistance and 1.6 um for anotherplasmid (cml) controlling chloramphenicol re-

sistance. These two plasmids thus differ fromthe penicillinase plasmid PI not only in thatthey control only one type of resistance (Tcand Cm, respectively), but they are also muchsmaller, with molecular weights of 2.9 Mdal(tet) and 3.1 Mdal (cml) (179).Unstable R factors. The DNA species of a

number of R factors have been shown to becomprised of more than one molecular species.When DNA of the R1 and R6 factors was iso-

TABLE 4. Molecular weights of monomolecular R factors

Molecular wt

Plasmid fir Isolation" Density GCd Method of (Mdal)' Reference(g/cm3) ratio molecular wtCQ C

R15 - BR 1.709 49 EM 35 389 176EB 1.708 48 EM 46 46 177

R6K - EB 1.704 45 EM 26 26 145R28K - EB 1.710 50 EM 44 44 145R64 - CT EM 76 78h 227R538 + CT EM 49 50h 227222/R4 + EB 1.710 50 EM 70 70 177

mini EM 62 649 128222/R3W + EB 1.710 50 EM 69 69 177R1 + EB 1.710 50 S 63 59' 49

CT EM 65 689 50S 65 611 215

R6 + EB 1.711 51 EM 64 |67 50S 64 60i 50

a Type of R factor indicated by fi group (239). All fi- factors are F-like; some fir factors are I-like; othersare not I-like (58).

BR, By base ratio difference from Proteus host; EB, by ethidium bromide-cesium chloride from E. coli;CT, by conjugal transfer method in E. coli; mini, by segregation into E. coli minicells.

"Measured to an E. coli DNA standard of 1.710 g/cm3 (164).d Conversion from (210).S, Neutral sucrose sedimentation; EM, contour length measurement by electron microscopy.Q, Quoted molecular weight in reference; C, corrected molecular weight.

' By use of MA factor of 2.07 Mdal/gm (147).h By use of corrected X MW of 30.8 (61).'S value by use of sedimentation index of 0.38 (87).

381VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

TABLE 5. Contour length measurements andmolecular weights of R factor 222, its segregants,

and its recombinants in E. coli (177)

No. of MlcContour mole- Mean Molec-

Plasmid length cules length ularSSDa (Mm) meas- (Am) (Mdal)ured

Parental222/R4 33.5 ± 0.5 21 33.6 70

33.8 ± 0.7 1233.8 ± 0.4 14

Segregant222/R3N 29.5 ± 0.4 33 30.3 63

30.5 ± 0.4 2231.5 ± 0.8 17

222/RlA 22.3 ± 0.8 25 22.3 46222/RiB 25.9 ± 0.3 30 25.6 53222/RlC 25.7 ± 0.3 35222/RID 25.4 ± 0.3 28222/RlE 25.8 ± 0.3 20

RecombinantRlAxR3N-1 32.6 ± 0.9 29 33.5 69

-2 34.3 ± 0.3 17-3 33.3 ± 0.5 20-4 33.9 ± 0.4 22

RlBxR3N-1 32.9 ± 0.7 29 33.2 69-2 34.1 ± 0.4 24

a SSD, sample standard deviation.

lated from E. coli, in addition to the moleculesof 60 to 70 Mdal found as a majority species(Table 4), a minority of smaller molecules was

found (49, 50). DNA isolated from host strainsof Proteus mirabilis carrying either of thesetwo factors (50) contained three sizes of mole-cules with different densities (Table 6). Verysimilar results had previously been reportedfor the independent R factor 222/R3W (seealso Fig. 14). From the experiments with 222,it was concluded that, since the sum of thesizes of the two smaller molecular species was

equal to that of the largest species (which alsohad a density intermediate to the two smallerones), that the two smaller molecules had re-sulted from segregation from the larger (i.e.,composite) molecules in the host strains ofProteus (176). This composite molecule was

concluded to be equivalent to the single molec-ular species later found in E. coli of size 33.5,gm (69 Mdal) (177 and Table 4; see Fig. 15).Similar conclusions were later drawn for theother factors R1 and R6 which, like 222, are

also fi+ factors (52, 215). It is interesting tonote that, in spite of the similarity of thesesegregation patterns in Proteus and particu-

larly in the molecular sizes of all three R fac-tors and their segregants, their origins are dis-tinct. Factor 222 (234), isolated in Japan (174),has also been termed NR1 (201) and R100 (72)by different workers and carries resistances toSu, Sm, Cm, and Tc. R1 originated in theUnited Kingdom (8) and determines resistanceto Su, Sm, Cm, Km, and Ap, whereas R6, iso-lated in Switzerland (152), carries Su, Sm,Cm, Tc, Km, and Nm resistance. Homologystudies of these geographically and phenotypi-cally distinct R factors examining the DNA:DNA heteroduplex formation by electron mi-croscopy would obviously be of interest, sinceit seems possible that at least the sametransfer factor moiety may be involved. (Apreliminary report [P. A. Sharp, N. Davidson,and S. N. Cohen, 1971, Fed. Proc., 1054/8]shows homology of ColV2, R1 and R6 with thesame region of F. R1 and R6 show extensivehomologies over their total length.) It has beensuggested that E. coli or Enterobacteriaceaemay be the normal hosts of these R factors(i.e., in which their evolution by the accretionof extrachromosomal drug resistance plasmidsto a transfer factor may have occurred), wherethe stable composite form is preserved. Pro-teus may be considered as a foreign host inwhich the R factor breaks down to its evolu-tionary component molecules.

Multimolecular factors. Another system ofR factors recently studied, the ASAT systemof Anderson (7), shows that even in E. colihosts some R factors may be comprised of (sta-ble) multimolecular species. From an originalstrain carrying infectious resistance to strepto-mycin, ampicillin, and tetracycline, three se-gregant strains were isolated by Anderson andco-workers: one termed A+S+ carries infectiousresistance to streptomycin, another referred toas A+A+ transmits infectious resistance toampicillin, and a third, (A-T)+, shows infec-tious resistance to tetracycline (Fig. 16). Fromhost cells carrying the first two factors, furthersegregants were isolated. The A+S+ strain gaverise to two segregant strains: one carried non-infectious resistance to streptomycin (termedS), and another was not drug resistant but car-ried a transfer factor (A), shown by virtue ofthe fact that, when grown in mixed culturewith the S+ strain, it could mobilize thetransfer of streptomycin resistance to a recip-ient strain. From the A+A+ strain, two similarsegregant types were isolated, one carrying thetransfer factor A and the other with noninfec-tious ampicillin resistance, A. From (A-T)+strains, no further segregants were isolated (5,7, 9, 11). DNA was separated from each of the

382 CLOWES


mbr.asm

.org/D

ownloaded from




plasmid-containing strains by Sarkosyl lysisfollowed by ethidium bromide-cesium chloridedensity centrifugation and in each case wascomposed of a mixture of CCC and OC mole-cules. The contour lengths of a number of OCmolecules were measured as shown in Table 7(168a). The DNA from a strain carrying A com-prises molecules of one size only (29.1 gm con-tour length, equivalent to 60 Mdal). From astrain carrying S, again a monomolecularspecies was isolated of approximately one-tenth the size of A, and of molecular weightestimated therefore as 6 Mdal. In A+S+strains, a bimodal distribution of moleculeswas found, corresponding in sizes to the mole-cules found in A+ and S+ strains, respectively.Similarly, DNA extracted from &+A+ strainsformed a similar bimodal distribution of mole-cules. In contrast, (A-T)+ strains comprised amonomolecular species of 32.3 gm contourlength (67 Mdal).From a comparison of length measurements

of small molecules inferred to be S, isolated

from A+S+ strains, we concluded that there isno real difference in size from those isolatedfrom S+ hosts. Moreover, the size distributionof large molecules from both A+A+ and A+S+strains corresponds with that of moleculesfound in DNA of A+ strains, except for asmall number of molecules of the size onemight expect of a recombinant molecule be-tween A and either A or S. However, thisnumber is less than 5% of the total number oflarge molecules. It was concluded that in A+S+strains the majority of molecules act as inde-pendent A and S molecules, and little if anyrecombination has taken place, although re-combination cannot be completely excluded. Asimilar conclusion was drawn from the AAdata. In contrast, from data with A-T DNA, itwas concluded that this plasmid represents astable union of A and T (a tetracycline-re-sistant determinant) and takes the form of amonomolecular species larger than A by anamount about the size of A or S (168a). In thisway, the physical data supported the conclu-

(III)

(v)

FIG. 13. Schematic diagram representing possible alternative pathways of recombination of two segregantR factors to reconstitute the parental type. Structure I is the parental 222/R4 factor giving rise to the222/R3N segregant (II) having lost region A, and the 222/R1 segregant (III) having lost region D. The parentalR4 structure could be reconstituted by either (i) two sequential crossing-over events (the first, between B andC, gives rise to structure IV by "addition" of I and III, followed by a further reciprocal crossing-over eventbetween E and F, eliminating the duplicated region BCEF to release the "left-hand" part of VI, identicalwith 1), or alternatively, (it) two simultaneous crossing-over events (in similar regions as above could occurafter the formation of the double-looped structure, V, to regenerate a structure identical to 1) (177).

383VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

TABLE 6. Densities and molecular sizes of R factormolecules found in strains of Proteus carrying

222/R3W, R1, or R6a

lestisCngthu Molecular Refer-lsid \ e(g/cm3) length wtb (Mdal) ence(pm) ±SSD

222/R3W" 1.708 28.5 + 0.3 59 1761.711 35.8 i 0.3 74d1.717 6.4 ± 0.1 13

R1 1.709 28 ± 1.0 58 501.711 33 ± 0.8 681.717-8 5 ± 0.5 10

R6 1.709 26 to 31 54-64 501.711 31 to 38 64-791.717-8 4 to 7 8-14

a DNA was isolated by either SLS (176) or Brij 58+ SDS (50). It was then centrifuged in CsCl (176and 50 [R1]) or CsSO4 + Hg2+ (50 [R6]). The threemolecular sizes of 222 separated from logarithmic-phase cultures are shown in Fig. 14. R1 and R6 frac-tions at 1.709 to 1.711 density were isolated fromlogarithmic-phase cultures and 1.717 to 1.718 frac-tions from stationary cultures (50).

h Calculated on the basis of 2.07 Mdal per gm ofDNA (147).

c Plasmid 222/R3W is a tetracycline-sensitive seg-regant of 222/R4 (74,236). From contour lengthdata, it appears to be either a point mutant or asmall deletion mutant of 222/R4 (177).

d Based on 16 molecules, other experimental datameasuring a further 22 molecules of mean contourlength 33.3 pm (177) leads to an overall mean of 34.3gm or 71 Mdal.

sions previously drawn from physiological andgenetic experiments (7, 9; see Table 1).

PHYSICAL INTERRELATIONSHIPSThe similarity of a number of bacterial plas-

mids, particularly in their sex factor proper-ties, has led to frequent speculation on theirinterrelationship, origins, and evolution. Ge-netic homologies have been inferred from re-ports of the detection of recombinants betweennonisogenic plasmids. In some cases, thesereports are based on tenuous evidence of co-transfer or, what may now be suspect, cotrans-duction data (102); in other cases, notably thecomplex F * ColV ColB * trp * cys, the genetic

58MdoI

7OMdaI

E Co/i ProteusFIG. 15. Alternate molecular forms of 222/R3W

factor as a composite (70 Mdal) molecule in E. coliwhich separates in Proteus into two component rep-licons of 58 Mdal and 12 Mdal.

IlPROTEUSCHROMOSOMAL DNA

(1.699 gCm 3)

6/1m (I 2 Mdal) MOLECULES 28MLm(58 Mdal) MOLECULES(1.717 gcm-3) (1.708 gcm-3)

34~m( 7OMdao) MOLECULES(1.711 gcm 3)

FIG. 14. Diagrammatic representation of satellite peaks of 222/R3WDNA (enlargement of part of Fig. 7 or8) isolated from a Proteus host strain after preparative CsCI density-gradient centrifugation, indicating den-sities and contour lengths (in micrometers) of molecules measured, and calculated molecular weights inmegadaltons (from 176).

384 CLOWES


mbr.asm

.org/D

ownloaded from




(AAST AS -S

ATFIG. 16. Derivation of strains in ASAT system. A

denotes a strain with transfer factor properties; S, A,and T represent noninfectious resistance to strepto-mycin (plus sulfonamide), ampicillin, and tetracy-cline, respectively (9).

TABLE 7. Contour lengths and molecular weights ofa number of plasmids of the USA T system of

Anderson and Lewis (9)a

Contour length No. of MolecularPlasmid4- SSD (tum) meaured wt (Mdal)

29.1 i 0.8 33 60S 2.9 i 0.13 63 6.0AS 29.3 ± 1.1 51 61

2.7 ± 0.13 296 5.6AA 29.5 ± 1.4 32 61

2.7 i 0.15 210 5.6A-T 32.3 ± 1.1 80 67

a Data from Milliken and Clowes (168a).

evidence (84) has been supported by the phys-ical isolation of a unitary DNA structure (115).Good evidence has also been indicated of phys-iological similarity of a number of different sexfactors through the cross complementation offertility characteristics (1, 2, 184, 185, 248).

Physical demonstration of genetic homologyis possible through DNA:DNA hybridizationstudies (19). Radioactive DNA of one plasmidis sheared into small fragments of about300,000 daltons and is then denatured intosingle strands. The proportion of this labelwhich can bind to sheared, denatured DNA,originating from a different plasmid and im-mobilized in an agar gel, is measured (73, 75).Early binding experiments led to overestima-tions of homology, probably because of non-

homologous binding at regions having partialhomologies (73). More recent experiments haveshown that much of this binding is unstable attemperatures well below the temperature ofdenaturation of the DNA of either parent,whereas a proportion, presumably representingregions of complete homology, is as stable as a

homologous control. Under these more ex-acting conditions, Falkow and co-workers haveshown that R1 has a homology of 74% with222, 35% with F, and 16% with ColIb (76).More recently, a technique has been de-

scribed for examining by electron microscopythe regions of hybridization between twostrands of DNA from different origins (64, 218).Two plasmid DNA .species of distinctive sizeare denatured, mixed, and allowed to reanneal.Homologous regions on heterologous duplexescan be seen as double-stranded regions ofDNA, whereas the nonhomologous segmentsremain as single strands which can be distin-guished by electron microscopy both by thick-ness and conformation (see Fig. lib). By usingthis technique, Cohen and associates (52) haveshown that the homologous region of R6 and Fcomprises 44% of the R factor and is in onecontiguous segment. Clearly the use of thistechnique is destined to play a major role infuture studies of relationships between plas-mids, particularly if correlations are madebetween the location of the homologies and thegenetic map as has been already so success-fully accomplished in the case of X (78, 218).

MOLECULAR STRUCTURE

Monomolecular PlasmidsIt is clear that all plasmids so far examined

are circular, double-stranded DNA moleculesand exist intracellularly for at least a portionof their time as covalently closed structures.The circularity of plasmids had been antici-pated as the structure that would be consistentwith their integration as episomes within thehost chromosome in a manner analogous tothat originally proposed by Campbell for the Xmolecule (29), later confirmed to be physicallycircular (114). However, in view of the factthat integration may not in fact be a generalproperty of plasmids, circularity seems morelikely to represent an essential structure forreplication (133).

It has been suggested (112) that the covalentring closure of plasmids may be an artifactproduced subsequent to cellular lysis by theaction of a bacterial ligase (100), but this ap-pears to be at variance with a number of well-established data (see 89 for discussion), and wewill conclude that the CCC form is establishedintracellularly. The supercoiled structure ofplasmids would appear to have no significanceof any special plasmid quality but to be merelya consequence of covalent ring closure. Since itseems likely that the number of superhelicaltwists in all CCC molecules (including plas-

VOL. 36, 1972 385


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

mids) is proportional to the contour length (25,55), the argument has been made (25) thatsupertwisting is the extracellular end result ofring closure of intracellular DNA which, per-haps due to the ionic conditions existingwithin the cell, has a helical pitch slightly lessthan that measured in the original "B" config-uration (159).

In CCC DNA isolated from mitochondriaand animal viruses, and double-stranded repli-cating forms of single-stranded bacterial vi-ruses, a further degree of configurational com-plexity has been described which results fromthe interlocking of several CCC molecules("catenation") forming products termed caten-anes, which are usually dimers, but may alsobe trimers or higher multimers (see Fig. 12b).These structures can be identified since one ofthe interlocked molecules may be "relaxed" or"open," whereas others in the same catenaneremain covalently closed. Such moleculeswould have a density in ethidium bromide-ce-sium chloride gradients intermediate betweenCCC and OC molecules and may thereby beisolated and identified (Fig. 17). CatenatedDNA was first described in bacterial plasmidsisolated from S. aureus (205), and later for anenteric bacterial fi- factor, R6K, controllingpenicillinase (145). However, this latter factorproduces unusually high amounts of plasmidDNA, and the proportion (5% to 10%) of caten-ated molecules may thus be more easily de-tected in this strain, but may be characteristicof all plasmid DNA species. This conclusionhas been made more probable by a recent re-port on the isolation of catenated DNA of R1and R6 factors (both fi+) from minicells (48,52) and of catenated ColEl, also from mini-cells (93, 130). It has not been possible to cor-relate the catenation of mitochondrial DNAwith an origin by either replication or recombi-nation, and it may well occur subsequent toring closure by random breakage and rejoiningof double-stranded ring structures (124). Re-cent evidence from kX174 (17) leads to theconclusion that, although circular multimersarise as a result of replication errors, this is nota source of most catenanes. These are formed'by recombination errors which, if nonrecipro-cal, would be equivalent to the breaking of acircular molecule and its rejoining after inter-lock with another circular molecule. Theamount of catenated DNA found in R6K DNAwas remarkably constant, corresponding to onedimer per chromosome, irrespective of thephysiological state of the cells, in contrast tothe amount of noncatenated CCC DNA whichincreased threefold in cultures moving into

resting phase (145). Catenation of bacterialplasmids may thus be related to a replicationor a segregation mechanism involving a mem-brane structure (133).A summary spectrum of the molecular sizes

of bacterial plasmids is shown in Fig. 18,where it can be seen that they range from ap-proximately 3 Mdal to greater than 100 Mdal.One clear distinction emerges: as might beexpected from the identification of a dozen ormore genes associated with sex-factor fertility(1, 2, 80, 132, 184, 185, 246, 248, 249), plasmidsthat are noninfectious are smaller (less than 20Mdal) than those possessing sex-factor activ-ity, and all plasmids greater in size than 26Mdal are capable of self-transfer and have fer-tility characteristics.The size of the R factors 222, R6, and AT

relative to the DNA content of the generalizedtransducing phages P1 (126) and P22 (193) isconsistent with the fact that each factor can

4j.4

'U~~~~~~~~~

20( .,,, 0j, ,,1

Frdction number

FIG. 17. Sedimentation profile of DNA from anE. coli (R6K)+ culture after ethidium bromide-CsCldensity centrifugation. Fractions are numbered ascollected in five-drop samples from the bottom ofthe centrifuge tube. Radioactivity in 0.006-ml sam-ples was determined by liquid scintillation countingof the sample after filtration and washing through amembrane filter. Electron microscopy showed thatpeak I contained CCC DNA; peak II contained pre-dominantly catenated DNA. (145).

386 CLOWES


mbr.asm

.org/D

ownloaded from




be transduced in toto by P1 (10, 237, 242) (mo-lecular weight 78 Mdal), but only transductionof parts of each R factor is found with P22 (11,237, 242) (molecular weight 28 Mdal).The molecular relationships found for var-

ious plasmids do not so far lead to any obviousinterpretation for the mechanism of curing.Clearly, all plasmids are supercoiled struc-tures, and since many plasmids can be curedby drugs that intercalate, it has been proposed(104) that curing may depend upon the CCCform. However, this does not seem to take intoaccount differences found in susceptibility tocuring agents; for example, the two colicino-genic plasmids ColIb and ColEl are com-pletely refractory to curing by the intercalatingdye, AO, in contrast to the sensitivity of F (46),although ColIb and F are very similar in sizeand ColEl is much smaller.

Curing seems to result from an inhibition ofplasmid replication rather than any effect onsegregation (57, 120). Furthermore, since theinitiation of chromosomal replication becomesmore susceptible to AO inhibition when itcomes under the control of an integrated Ffactor (175), it may be concluded that thegreater sensitivity to AO of F replication com-pared to chromosomal replication (thusleading to curing) may lie in the initiationevent. Moreover, rifampin, which appears toprevent the initiation of RNA transcription bybinding to RNA polymerase (244), has alsobeen reported to lead to curing in S. aureus(137) and more recently to ColEl in E. coli(35). Since the act of transcription rather thanthe product(s) has been suggested to be a nec-essary prerequisite for the initiation of DNAreplication (68), it may be concluded that''curing" originates in this event and that anycompound that preferentially interferes withthe initiation of plasmid DNA replicationmight be an effective curing agent. A similarinterpretation would arise if in fact an RNAprimer were necessary to initiate the replica-tion ofDNA (24a).

A Revised Map for R222The establishment of a monomolecular DNA

species for R factor 222 in E. coli (177) con-firms its genetic structure as a plasmid cointe-grate (see Fig. 1, Table 1). Values for the molec-ular weights of the segregant molecules of222 (177 and Table 5) permit the construc-tion of a revised physical map (Fig. 19) withthe chloramphenicol (Cm), streptomycin (Sm)-sulfonamide (Su) elements located withinsegment CD and the tetracycline (Tc) resist-ance gene within DE (a similar map can be

drawn for R1 and R6 with the amp, kan, andneo markers located in the CDE region). Sinceboth the segregant factors R1A (AXFED) andR3N (FXABCD) appear to have normaltransfer factor properties, all genes controllingtransfer of these plasmids, sometimes referredto as the RTF region, must be located withinthe AXF segment. If 222 resembles R1 and R6so that its 12 Mdal segregant in Proteus (CDEsegment) can be identified as the segregantcarrying the drug-resistance determinants ("r"segregant) and the 58 Mdal segregant(CBAXFE segment) as the drug-sensitive se-gregant carrying transfer factor properties(RTF), then the latter clearly has more DNAthan is required for RTF function, if only be-cause it is larger than R1A which retains RTFproperties and also has a Tc resistance region.If RTF occupies a large part of the AXF seg-ment, because of the size of P22 DNA (ca. 28Mdal), it would not be possible for P22 trans-ductants selected for either Cm or Tc to in-clude region AXF, and they would therefore benoninfectious (237). What is difficult to ex-plain is that, although the Sm-Su, Cm, and Tcdeterminants (segment CDE) may be con-cluded to segregate as the 12 Mdal molecule,they are not picked up by P22 to give stablenoninfectious Cmr, SuSmr, Tcr transductants.(Nor are those transductants that are found-CmrSmrSur or Tcr-apparently able to persisteven as noninfectious replicating plasmids [70,237].) This could be due to the inability of the12 Mdal molecule (or parts of it) to be stablyreplicated in the absence of the 58 Mdal mole-cule. Either the 12 Mdal molecule has a repli-cator locus but no regulation locus, the larger58 Mdal plasmid molecule providing eithersome form of positive control in the shape ofan initiator, or, as previously suggested (146,191), a negative (repressor) control which pre-vents unregulated synthesis of the 12 Mdalelement which would otherwise lead to thedeath of a transductant carrying this element.Alternatively, there may always be a breakbetween Cm, Sm-Su and Tc in the linear mole-cules that are perhaps necessary for incorpo-ration in P22 for transduction. If these mole-cules were assumed to arise during replication,it might be concluded that the replicator locusof the composite 70 Mdal 222 molecule wassituated between Tc and Cm,Su-Sm.

Alternative Molecular Structures of RFactors

Figure 20 summarizes diagrammatically themolecular nature of R factors which can be

387VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



--- Sex factor activity _

"Relaxed"____________ "Stringent"-Replication Replication

R6K

RIA R;R28K R3N Ib

RI5 RIB A4 ATI3WR482 V2

MW I11 20 3II0 6I0 7IIoIoIlbM lals) 1,,11 l iii Iipliili i ia''tXlii.atlail111111111 llal'ala''(Mdals)I11 -R58'R4I)- RI(CR64-~

pen

tF42lT7 X

R538 r'-4(j Rl (C) R64Ib(H) F{vLl)

R6(C)

T5 PiFIG. 18. Size spectrum of bacterial plasmids. Data of plasmids shown above the scale are those of the au-

thor and colleagues, and below the scale those of other workers (detail in Tables 2, 3, and 4; note that RIA,RIB, R3N, R3W, and R4 are abbreviations for derivatives of the 222 R factor; Ib, B2, and V2 are abbreviationsfor the colicin factors ColIb-P9, ColB2-K77, and ColV2-K94, respectively); pen, tet, and cml are plasmids ofS. aureus (179). Ib(H) is size of ColIb from Clewell and Helinski (38), F( VLI) is value ofF in reference 227, R4(I)is value of 222/R4 by Inselberg (128). RJ(C) and R6(C) are data of Cohen and Miller (50) for the factors R1 andR6. The sizes of a number of phage DNA species are shown by arrows below: OX174 (79), T7 and T5 (87), P22(193), A (61), and P1 (126).

C

,'~~~~~~1224

'CSM ,Su'

~~~~~E

A-----|- F

39 -

x

FIG. 19. Revised physical map of 222/R4 based on

contour lengths of DNA from deletion mutants (177)showing sizes in megadaltons. Cm, Sm, Su, and Tcrepresent the genes controlling resistance to chlor-amphenicol, streptomycin, sulfonamide, and tetracy-cline, respectively. From the parental molecule222/R4 (70 Mdal), deletion AD (24 Mdal) gives RIA(46 Mdal), deletion BD (17 Mdal) gives RIB (53Mdal), deletion DF (7 Mdal) gives R3N (63 Mdal),and deletion CE (12 Mdal) would give "resistancetransfer factor" assuming similar data are found as

for Ri and R6 (50, 51).

+

STABLEPLASMID

COIN TEGRATE

eg R222 in Eco/iRI in EcollR6 in Eco/hA-T

UNSTABLEPLASMID

COINTEGRATE

eg. R222 in PROTEUSRI in PROTEUSR6 in PROTEUS

PLASMIDAGGREGATE

eqg A +AA + SF + El

FIG. 20. Alternative molecular structures of R fac-tors. TF represents genes controlling conjugal trans-fer, which can exist as an independent structureknown as a transfer factor. NIP represents genescontrolling other properties which can exist inde-pendently as noninfectious plasmids.

represented by several alternative structures.The major R factor type appears to have a

composite structure, in which a number ofelements are cointegrated in the same circularplasmid molecule to form a stable "plasmidcointegrate." This structure is clearly appro-

priate for the R factors 222, R1, R6, and AT as

they exist in E. coli, since both genetic studies

A

cml 1j2|EI!tet E3

#X174

CLOWES BACTERIOL. REV.388


mbr.asm

.org/D

ownloaded from




(9, 10, 233, 237, 242) and physical studies (Ta-bles 4 and 7) are consistent with a single struc-ture. It is also consistent for other plasmidssuch as the F - ColV ColB -trp - cys plasmid(from both genetic studies [Fig. 2; reference84] and physical studies [115]) and for the P1penicillinase plasmid which has a circular ge-

netic map (Fig. 3; reference 179) and takes theform of a single 19 Mdal circular DNA mole-cule (179).

In some cases, the cointegrate has incorpo-rated more than one element capable of inde-pendent replication and, although stable inone host, it may break down into its independ-ently replicating molecules in another host.Examples of this are the factors 222, Ri, andR6, which are stable in E. coli, but when trans-ferred to a Proteus host strain they act as an

unstable plasmid cointegrate and segregatetwo of their component replicons (see Fig. 15)(described by some workers as an "association-dissociation" phenomenon [52, 76, 111]). SinceProteus would not appear to be the normalhost for these plasmids, it is perhaps not sur-

prising that a measure of control is lost andthe plasmid may revert to what may be sup-

posed as a more primitive evolutionary type.Another type of R factor structure is repre-

sented by the "plasmid aggregate," as found inAA and AS, which is similar to the artificialcomplex F ColEl. In these systems, thetransfer factor properties are carried on a largemolecule and the drug resistance (or colicino-genic properties) on one (or more) noninfec-tious small molecule(s) which may frequentlybe transferred as a result of being resident inthe same cell as the infectious plasmid. At themoment, one can only speculate on the phys-ical basis for the conjugal cotransfer of theseelements (which has also been referred to as

"association" [5, 7]) but, because of the lack ofrecombination between them in strains whichreceived their plasmids by this transfer (168a).The probability of transient covalent linkagewould appear to be unlikely. It has previouslybeen concluded that chromosomal segmentsmay be transferred from one bacterial strainto another by virtue of a conjugating systemset up by the ColIb sex factor, which appearsto be unable to integrate into the chromosome(45, 173), and is therefore not likely to be co-

valently linked to the chromosomal segmentsin conjugal transfer. Similarly, transfer of a

noninfectious element such as S, A, El, or E2by mobilization with either A, ColIb, or F (9,43, 185) may occur in a parallel way and beindependent of any physical association be-

tween the two elements, both being presumedto be transferred as single-stranded DNAmolecules with a 5'-3' orientation (183, 204).

Speculations on R-Factor EvolutionThe idea that R factors are comprised of

independent replicons, and therefore that theymay arise by de novo mutation of moleculeswhich are initially extrachromosomal, leads tosome interesting speculations on their evolu-tion. It seems clear at least that, as an alterna-tive evolutionary origin of R factors by a "genepick-up" mechanism, we must consider thatsome drug-resistance genes originate (aftermutation?) from elements which were extra-chromosomal at the time that selective anti-biotic pressure evolved the R factors in thevariety and profusion that we see today.Whether drug-resistance genes originatechromosomally or extrachromosomally, thesame problem exists of establishing what isthe "normal" function of genes that mutate togive resistance to antibiotics that the bacteriais not likely to have had contact with in thecourse of normal evolutionary development.Since many R-factor resistances are due tosuch standard enzymatic processes as acetyla-tion, phosphorylation, adenylation, etc. (62, 63,212) (as opposed to chromosomal antibioticresistance which results from alterations inribosomal structure), it is possible to conceiveof R factor evolution via the evolution of en-zymes through an increasing affinity for an.antibiotic as substrate, the original enzyme ac-tivity being directed towards a different sub-strate that is more commonly encountered. In-deed, increasing substrate affinity by evolu-tion has already been proposed by Shaw (212)on the basis of differences in Michaelis (Km)constants of a number of Cm-transacetylatingenzymes from Cm-resistant bacteria.Since in enteric bacteria most extrachromo-

somal elements can be mobilized by one oranother transfer factors and, indeed, transferfactors of either F-like or I-like type can mob-ilize the same non-infectious plasmid, thesimultaneous presence of transfer factors inthe same interfertile population as smallernoninfectious drug-resistance elements wouldbe likely to ensure the rapid dissemination ofthese smaller resistance molecules, leading tosystems such as A+S+ and A+A+. This wouldsuggest that the plasmid aggregates may beearly evolutionary forms. Later perhaps, someof these plasmid aggregates may have devel-oped into plasmid cointegrates which are

VOL. 36, 1972 389


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

likely to be a more efficient system for boththe transfer and the stability of drug resist-ance.

ASPECTS OF PLASMID REPLICATION

Plasmid Copy Number-Relaxed andStringent Replication

Since we know the molecular weights of anumber of plasmids, and the molecular weightof the host chromosome is also known (themost commonly used value for E. coli being2500 Mdal [53]), an assay of the ratio ofplasmid to chromosomal DNA would permit acalculation of the relative numbers of copies ofplasmid to chromosome in those cases whereonly one size of plasmid molecule exists in acell. A number of methods depend upon themeasurement of plasmid molecules as repre-senting the only CCC DNA in the cell on theassumption that most plasmid DNA exists in-tracellularly in this state. Freifelder et al. con-cluded that 25 to 100% F'lac DNA exists asCCC molecules (89). Clewell and Helinski (39)isolated an unstable plasmid DNA moleculewhich they argue is lost in the standard proce-dures for isolating CCC molecules and con-clude that CCC DNA accounts for between 27and 44% of the total plasmid DNA. Hence, themeasurement of CCC DNA to chromosomalDNA gives only a rough estimation of thenumber of plasmid copies, and this will clearlybe a lower limit. The percentage of CCCplasmid to chromosomal DNA has been esti-mated in various ways. Some of the results areshown in Table 8. In spite of the fact thatmany of these methods are subject to certainapparently uncontrollable experimental varia-tions (89; probably reflecting the proportion ofplasmid DNA isolated in CCC fractions), oneimportant feature emerges from this type ofdata. Table 8 shows that for many plasmidsthe number of copies correspond to betweenone and two per chromosome. For the re-maining plasmids, this relationship clearlydoes not hold, and the relative number ofplasmid copies per chromosome is muchgreater and is generally between 10 and 40.These plasmids include the noninfectious ele-ments ColEl, A, and S and, in addition, thesmallest factor so far found with sex factor ac-tivity, R6K. Thus, although the data are crude,two distinct classes of plasmids can be recog-nized. First, those preserving a near unitaryrelationship in copy number between thefactor and the chromosome, which have been

termed "stringent" in the regulation of theirDNA replication (145), and secondly, thosewhere there are 10-fold or so more copies of theR factor than the chromosome, in which thereplication of the DNA has been termed "re-laxed" (198). A further feature of at least somerelaxed-regulated plasmids is that the relativenumbers of copies appears to increase somethreefold if lysates are made from stationaryrather than from logarithmic cultures (139,145, 198). Moreover, the control of plasmidreplication appears to be size-related (Fig. 18),only the smaller plasmids being relaxed (al-though the discrepancy should be noted be-tween the stringent pen plasmid [-20 Mdal]and the relaxed R6K [26 Mdal], which maypossibly be due to differences in the staphylo-coccal and enterobacterial systems). Relaxa-tion thus appears to result from a lack of con-trolling genes and implies that stringent con-trol may be effected via repressor genes.Clearly, a stringent relationship would appear

TABLE 8. Amounts of plasmid and chromosomalDNA and ratio of number of copies of plasmid

per chromosome"

tioRn Plasmid' Method" S/Cd Copy ReferencecontrolSC no

Relaxed ColEl Brij 1.3 to 2.3 7 to 12 39S EB 3.1 14 168a

15' 13 145R6K EB 409 38 145

Stringent R28K EB 4.1 2.2 145R15 EB 2.5 1.4 177R538 AS 3.3 1.7 227X EB 1.8 0.8 168aF EB 2.2 0.9" 15

AS 1.7 0.7' 227222/R4' EB 5.0 1.8 177R64 AS 3.7 1.2 227FVBtc BR 5.0 1.1 115

,,All plasmids isolated from E. coli K-12, except FVBtc(Fredericq's F CoIV ColB trp cys) from Proteus.

hArranged in order of increasing molecular weight.t EB, Ethidium bromide-cesium chloride of crude lysate;

AS, lysis with Sarkosyl, alkaline sucrose centrifugation(227); Brij, Brij/DOC lysis (36); BR, Sarkosyl lysis of Proteushost strain (115).

dRatio of DNA as percent thymine label in satellite peakto chromosomal peak.

e Calculated from d using molecular weights in Tables 2,3, and 4 and molecular weight for DNA of E. coli 2,500Mdal, Proteus, 2,300 Mdal.

' From logarithmic-phase cultures." From stationary-phase cultures.'Assuming molecular weight of F of 64 Mdal.'DNA from the 222 segregants R3W, R3N, RiA, RiB

give rise to a plasmid chromosome copy ratio of 2.2, 1.2, 3.3,and 1.5, respectively (177).

390 CLOWES


mbr.asm

.org/D

ownloaded from




to be more stable and could be a stage in de-velopment from relaxed replication towardsstabilization by chromosomal integration.Table 8 shows data only or enteric bacterial

plasmids. More limited data for staphylococcalplasmids, obtained by Novick and co-workers(179) by the use of "cleared lysates" and there-fore possibly more reliable than that de-pending on the isolation of CCC DNA, show astriking parallel. The penicillinase plasmid ofabout 20 Mdal is estimated as being present intwo to three copies per cell, whereas thesmaller tetracycline plasmid (about 3 Mdal)exists in 30 to 32 copies per cell, and prelimi-nary studies suggest that the small chloram-phenicol plasmid (also about 3 Mdal) may alsoexist in "many copies per cell" (179).From a limited number of experiments, it

also appears that each plasmid copy is active,to give a constant "gene-dosage" response.This appears to be true for the F' elementcarrying fi-galactosidase activity (134), for thepenicillinase gene of R6K (145), and for theCm-transacetylating activity of R222 (199).

Mechanism of Relaxed ReplicationA strain of Proteus containing the 222 R

factor grown for several generations in thepresence of Cm has an R-factor DNA levelabout 16% that of chromosome. Rownd as-sumed the R-factor molecular weight to be 50Mdal which would be equivalent to about 10plasmid copies per chromosome, and posed thequestion of how these 10 copies replicated(198). (If in fact the preponderant R factormolecule is the 12 Mdal molecule [see Fig. 15],this would lead to a copy number of about 32but would not influence the following argu-ments.) Two possible models were proposed.One, a "master copy" model, suggested onlyone of these 10 molecules replicated, and didso 10 times for each replication of the chromo-some, to produce the 20 copies required at celldivision. The alternative model, which wouldbe called the "democratic" model, suggeststhat each of the 10 copies replicated once foreach chromosomal replication. Rownd per-formed a Meselson-Stahl density-shift experi-ment moving the host strain from a '5N(heavy) medium to a 14N (light) medium andanalyzing R-factor DNA at various generationtimes after density shift. The master copymodel would predict that, one generation aftershift, nine of the molecules would remainheavy (HH), two would be hybrid (HL), andnine would be light (LL). In contrast, the dem-

ocratic model would produce all (20) hybrid(HL) R-factor molecules. Rownd analyzed thedistribution of DNA in the R-factor satelliteband by means of a curve resolver and showedthat, after one generation, approximately one-fourth of the DNA was HH, one-half was HL,and one-fourth was LL, satisfying neither ofthe suggested models and being consistentwith the idea of a random replicating pool;after density shift, one molecule from the poolwould begin to replicate; during its replication,other molecules in the pool would not repli-cate; after replication, the two daughterplasmid molecules would return to the poolwhere the probability of their further replica-tion would be equal to the probability of repli-cation of any one of the other unreplicatedmolecules (198). It should be noted that theseresults would not lead to this interpretation ifaccount is taken that the system studied com-prises three distinct molecular species (176)and if it is not assumed that the bulk of theDNA is represented by only one of thesemolecular species. For example, the data arealso consistent with 50% of the DNA of either astringently controlled molecular species thatreplicates once, or a relaxed controlled speciesthat replicates "democratically," with theremaining 50% of the DNA replicating in arelaxed way from a master copy. However, asimilar experiment by Bazaral and Helinski(15) observing the replication of the monomo-lecular ColEl species in E. coli, with approxi-mately 10 identical plasmid molecules perchromosome, led to an essentially similar con-clusion as that of Rownd, favoring a randomlyreplicating pool.

If we accept the idea of a random replicatingpool, these experiments rule out a simple ideaof membrane attachment and replicationwhich in its simplest form would have led to amaster copy model. Membrane attachment isof course not excluded, but if all moleculeswere membrane bound, an initiator moleculewould be required in limiting amounts so as topermit only one of the attached molecules toreplicate at any one time. Alternatively, inter-mittent attachment and replication of arandom molecule from the pool, the subse-quent detachment of the two daughter mole-cules and their return to the pool before thenext signal for attachment, though cumber-some, would also preserve the idea of mem-brane attachment. Kasamatsu and Rownd(139) also concluded from the observationsthat growth under one of several physiologicalconditions rapidly led to a characteristic

VOL. 36, 1972 391


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

number of replications per generation and thatreplication was under a simple, positive con-trol.

"Transition" of Drug ResistanceIn strains of Proteus carrying R factor 222,

Rownd also reported a phenomenon which hetermed "transition" (199, 202). When such222+ strains are grown for a number of genera-tions in either Cm-, Sm-, or Su-containingmedium, to which the R factor confers resist-ance (but not in Tc-containing medium towhich it also confers resistance), the density ofthe satellite band shifts, so that after 200 gen-erations it is almost entirely of density 1.718g/cm3 and of an amount corresponding to ap-proximately 30% that of the chromosome.When the culture medium is shifted to excludeany antibiotic, sequential samples show a de-crease in the 1.718 g/cm3 satellite and theappearance of a 1.711 g/cm3 satellite. Afterapproximately 200 generations in drug-freemedium, there is little if any satellite DNA ofdensity 1.718 g/cm3, and the total satelliteDNA is restricted to density of 1.711 g/cm3and is reduced to about 6% that of the chromo-some (199, 200). Other workers have obtainedsimilar data for growth in drug-free and drug-containing medium (146, 191). As previouslyshown, in Proteus (222)+ cultures grown di-rectly from a single colony, three sizes of mole-cules of 12 Mdal (1.717 g/cm3), 58 Mdal (1.708g/cm3), and 70 Mdal (1.711 g/cm3) are present(see Fig. 15; 176), and similar conclusions havebeen proposed from studies with two inde-pendent R factors R1 and R6 (50, 75). Sincethe 1.717 g/cm3 peak (Fig. 14) is equal to orgreater in size than the other density peaksand is comprised of smaller molecules thanthose in the other peaks, the 12 Mdal mole-cules must be present in more copies than the58 Mdal and the 70 Mdal molecules (176).These results have been confirmed by otherworkers (146) and have also been found for R1or R6 (50, 75). Other small plasmid moleculesare relaxed in their DNA replication, in con-trast to larger molecules which are stringent(Table 8 and Fig. 18). These data are thus con-sistent with the idea that the 12 Mdal mole-cule is relaxed and the 58 Mdal and 70 Mdalmolecules are stringent. Moreover, if the 58Mdal molecule is identified in 222, as it hasbeen in R1 or R6, as a drug-sensitive segregantcarrying RTF (51, 216), then the 12 Mdal mole-cule would carry all four drug resistances.With the R6K plasmid, the level of drug resist-

ance increases in parallel with R factor copynumber (145). An increase in the copy numberof the 12 Mdal plasmid would thus be ex-pected to lead to increased resistance to thosedrugs where resistance is mediated through anantibiotic-inactivating mechanism (e.g., Cm,Sm, and Su) but not to an increase in resist-ance which may be due to a mechanism de-pending on a cell wall permeability change(Tcr; see references 62 and 212). "Transition"would then be consistent with an increase inthe relative numbers of copies of the 12 Mdalmolecule in drug-containing medium leadingto a preponderance of DNA at the density ofthis molecule. (1.717 g/cm3; variations in ex-perimental reproducibility of density peakmeasurements are about _0.001 g/cm3)."Back-transition" in drug-free medium wouldbe due to a reduction in the number of copiesof the 12 Mdal molecules until they were equalin number to the 58 Mdal molecule (1.708g/cm3) or perhaps even integrated as part ofthe 70 Mdal (1.711 g/cm3, composite) mole-cule, when they would be subject to the strin-gent control imposed by the other (58 Mdal)component of this molecule. The interpreta-tions of this phenomenon by Rownd and co-workers (63, 199, 200, 202) appear to ignorecertain aspects of the molecular studies with222 (176, 177) and of the related factor R1 (49,50, 51, 103, 215, 216) and R6 (50, 52) and arenot consistent with what is known of the regu-latory control of other plasmids (Table 8 andFig. 18). Rownd's interpretation similarlyidentifies the RTF component with a largemolecule of 1.711 g/cm3 and the drug resist-ance (r) elements as small molecules of density1.718 g/cm3. However, r is concluded to bestringently controlled in the cell, whereas RTFis assumed to be relaxed and present in mul-tiple copies; the explanation of the densitychange in "transition" being based upon theelaboration of large polymeric molecules com-prised of a single copy of the RTF (1.711g/cm3:70 Mdal) determinant linked to a re-peating sequence of a large number of r (1.718g/cm3: 12 Mdal) determinants. These mole-cules would need to approach molecularweights of several hundred megadaltons toaccount for the density shift and plasmid DNAfractions found, and they have not so far beenidentified in structural studies, at least not asa common circular DNA molecule. Rownd alsoconcludes that control of R-factor replicationis positive and is made through initiator mole-cules (198, 199).

Kopecko and Punch (146, 191) have claimed

392 CLOWES


mbr.asm

.org/D

ownloaded from




that a similar increase in the ratio of the 1.718g/cm3 to 1.710 g/cm3 DNA also occurs undervarious conditions of protein inhibition. Theyalso interpret their data on the basis of a 1.710g/cm3 (RTF) molecule and a 1.718 g/cm3 (r)molecule and conclude that both are subject tonegative control by a repressor produced by agene on the RTF molecule. In addition, theRTF molecule is proposed to be membrane-attached and subject to positive control. Bothmolecules are concluded to be present as mul-tiple copies, the r molecule always outnum-bering the RTF molecule. Their model accom-modates the molecular structural data of 222and is not inconsistent with what is known ofthe regulatory control of other plasmids. Thelack of a segregant harboring only the 1.718g/cm3 element (or perhaps more forcibly, thelack of any transductant harboring Cm, Sm,Su, and Tc resistance [237] which would beexpected to arise were the 1.718 element ableto exist independently as an r element) wasinterpreted as due to lack of a repressor pro-duced by RTF and necessary to prevent un-controlled replication of r leading to its le-thality for the cell.

(This, however, could be equally well ex-plained by assuming the production of an ini-tiator by the RTF molecule which acts on RTFas well as on the r molecule and can give riseto a number of initiating events. This initiatorwould normally be controlled by a cellular re-pressor so that, in E. coli, for example, therepressor would inactivate initiator in excess ofthat required to permit one R-factor replica-tion per cell generation and would thus effec-tively limit the R factor to the composite mole-cule. In Proteus, this cellular repressionwould be less effective, so that more initiationevents could occur per cell division. Thus, ifinitiation of molecules was random, segrega-tion into the RTF and r molecules would belikely to lead to a preferential increase in thenumber of copies of the r molecules, since itstime of replication is only about one-fifth thatof the larger molecule. Under conditions ofdrug-resistance selection [transition], thosecells where a greater proportion of R-factorDNA was in the form of the r molecules would,because of higher resistance, tend to be se-lected. Under conditions of protein inhibition,repression would be further reduced and morecopies of both molecules would result.)The replication behavior of 222 discussed

above is paralelled by that of ColEL. Recentevidence (34) has been presented to show that,in the presence of Cm, ColEl continues to rep-

licate for 10 to 15 hr, long after chromosomalreplication has ceased, approaching a max-imum rate of eight times normal and leadingto about 3,000 copies per cell.

DNA-Protein "Relaxation Complexes"Further work by Helinski and colleagues on

the CCC molecules isolated in cleared lysatesusing Brij and DOC indicates that the molec-ular structure of such CCC molecules differsfrom those isolated by ethidium bromide (36,39). The Brij-isolated CCC molecules from(ColEl)+ cells sedimented in neutral sucrose at24S (the OC derivative sedimenting at 18S)compared with values of 23S (and 17S) forsimilar molecules isolated after Sarkosyl lysisand ethidium bromide-cesium chloride centrif-ugation (197). When the 24S CCC moleculesisolated from cleared lysates were exposed toone of a number of compounds including ethi-dium bromide, proteolytic enzymes, ionic de-tergents, or alkali, a large fraction of the CCCDNA was converted to OC DNA, whereas sim-ilar treatments were without effect on the 23SSarkosyl-isolated CCC molecules. Furtheranalysis led to the conclusion that, in the intra-cellular state, a protein was attached to one ofthe DNA strands of many of the CCC mole-cules, this CCC DNA-protein complex beingtermed a "relaxation complex," since removal(or activation) of the protein by one of theabove agents led to the breakage of the DNAstrand attached followed by "relaxation" ofthe CCC form into the OC form. It was sug-gested that the relaxation complex might playa role in the normal replication of the plasmidand that the protein could be an inactive en-donuclease (initiator?), activated by the var-ious treatments to produce a nick in one of thestrands of DNA (36). Separation of the relaxed(17S) complex by alkaline sucrose sedimenta-tion into circular (C) and linear (L) singlestrands shows that the molecular weights andamounts of both are similar and indicates thatonly one nick occurs in one of the strands ofeach molecule. DNA:DNA hybridization of Cand L strands from two independent prepara-tions indicated that a specific strand wasnicked. More recent work has investigated thesedimentation of the C and L strands in thepresence of poly(U,G) (39). Poly(U,G) is knownto bind preferentially to DNA strands rich inAT bases (12), and when X DNA is denaturedand centrifuged in poly(U,G), it can be sepa-rated into heavy and light strands which arecomplementary to each other (123). A similar

VOL. 36, 1972 393


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

resolution of a heavy and a light strand wasfound with DNA from ColEl (39). Relaxationcomplexes have also been isolated from ColE2and ColE3 (18, 37), ColIb (38), and F (144),and in all cases so far studied (ColEl, ColE2,and F) the L strand was heavy and the Cstrand was light. Thus the heavy strand is ineach case the unique strand that is nicked as aresult of relaxation.

Replication in MinicellsIf a DNA precursor with a radioactive label

is incubated with a preparation of minicellsthat have been separated from a parentalstrain carrying a colicin factor or R factor, in-corporation of label occurs into the DNA of theminicell, previously shown to be entirelyplasmid DNA (127, 128, 129, 156-158). Theextent of incorporation varies with the plasmidunder observation. If a density (5-bromodeox-yuridine, "heavy") label was used,then at 3 hrafter the addition of label to minicells contain-ing the 222 R factor, only a small fraction oftheir DNA was fully dense (consistent withtwo sequential plasmid replications) and thebulk of the plasmid DNA was found at a hy-brid density indicating that most of the 222 R-factor molecules replicate only once in mini-cells (128). In contrast, minicells containingColEl appear to replicate DNA more exten-sively, so that a greater proportion of the DNAis fully labeled although the majority of theplasmid DNA still remains at the hybrid level(127-129). Thus, growth in minicells reflectsthe relaxed and stringent replication of theColEl and 222 R plasmids, respectively, in theparental (maxi) cells (see Table 8).

In the presence of AO (50 ,g/ml), which canefficiently cure maxicells of the F factor butwhich does not cure ColEl (46) and has littlecuring effect on the R factor (236), replicationof both ColEl and R DNA species was inhib-ited to the same relative extent, both beingreduced to about 25% of the level of replicationfound in minicells in the absence of AO (128).(It should be noted that the cell division ofmaxicells and thus presumably chromosomalreplication is also inhibited at this level of AO[116]. The effect of AO on DNA replication inminicells of the F factor or some other plasmidmore susceptible to AO curing would appear tobe a more crucial experiment.)

Density labeling of minicells has also beenused to isolate ColEl molecules in the processof replication, fractions being selected fromeach side of the hybrid peak, i.e., that con-

taining only hybrid ("half-heavy") molecules.On examination by electron microscopy, ap-proximately 3% of the circular molecules fromthese fractions were 0-like in appearance. In allcases, two arms of the (replication) loop wereof equal length, and the sum of the length ofone of these arms together with the length ofthe rest of the molecule was a constant, equiv-alent in contour length to that of the circularColEl molecule previously measured, i.e., 2.3zm. The size of the (replication) loop variedbetween approximately 3 and 95% that of thetotal molecule (129). The data are thereforeconsistent with the idea that the 0 structuresare in fact replicating molecules consistentwith a Cairns (28) symmetrical model. How-ever, a minority of more complex structureswere also found as circular 2.3-,um moleculeswith linear tails of an equal or multiple length(93, 129, 130), consistent with the intermedi-ates of "rolling circle" replication (95).

Transfer ReplicationA further aspect of the replicon model (133)

proposed that the formation of cellular con-tacts during conjugation triggers replicationand consequent transfer of the donor F sexfactor due to its attachment to the membrane.In Hfr cells, the ensuing transfer would thus bedue to replication initiated on the sex factorand continued along the continuity of the Hfrchromosome. This idea has recently been vin-dicated in elegant experiments by Vapnek andRupp with F+ donors (228). Using a modifica-tion of the Freifelder (90) technique, they wereable to isolate CCC DNA selectively from ei-ther donor or recipient cells before or aftermating. The CCC DNA was then heated for ashort time in alkali so as to produce a smallnumber of single-strand breaks in each of thestrands and to denature the DNA duplex intoits component single strands. The limitednumber of breaks resulted in the circularstrands being broken once only to form intactlinear strands. The linear strands were thencentrifuged in CsCl in the presence of poly-(U,G), under which conditions F-DNA also wasfound to separate into heavy and light strands.Subsequent analysis of DNA from donor andrecipient cells was consistent with the ideathat on conjugation a single break occurred inthe heavy strand of the CCC F-DNA moleculein the F+ donor; this heavy linear strand wasthen transferred to the recipient (by analogywith Hfr transfer, presumably with a 5'-3' ori-entation [183, 294]), where the light comple-

394 CLOWES


mbr.asm

.org/D

ownloaded from




ment to it was synthesized, at the same timethat the heavy complement to the light cir-cular strand remaining in the donor was alsosynthesized (228). Later experiments haveshown that in the case of two R factors, R538(F-like) and R64 (I-like), similar events occur,in both cases infectious plasmid transfer re-sulting in the heavy strand being transferredto the recipient (227).The transfer of the heavy strand in each of

these three cases could be coincidence. How-ever, breakage of the heavy strand was foundon relaxation of the DNA-protein complex ofeach of three factors examined by Helinski andcolleagues. These data, taken together, wouldsuggest that relaxation of the complexes asseen by Helinski may be the first step in repli-cation which in the case of sex factors also oc-curs prior to their transfer as well as in normalvegetative replication.

It would be of interest to determine whethertransfer of the hpavy strand also occurs fromthose plasmids, such as ColEl or A, which arenoninfectious and are transferred only whencoresident with a transfer factor such as F,ColIb, or A. If this were so, then cotransfercould be thought to arise by the act of cell con-tact triggering breakage of the heavy strand ofboth plasmids, transfer of any single-strandedDNA being semiautomatic under the condi-tions favoring transfer effected by the sexfactor. The quantitative differences of co-transfer observed (e.g., ColEl by F > ColE2 byColIb > ColEl by ColIb > ColE2 by F) in pre-vious experiments (43) might perhaps dependon the relative proximity of the sites of cellularattachment of the plasmids so as to lead to theprobability of breakage by the same contactevent, or perhaps on the lack of absolute speci-ficity of an endonuclease, which is released bycell contact, acting on the sex factor to initiatetransfer replication. A proposal to explainchromosomal transfer by nonintegrating sexfactors such as ColIb (45) could also be accom-modated within such a model.Some recent experiments by Falkow and co-

workers (76) throw more light on the kineticsof transfer replication and its relation tomembrane attachment. Using a fertility-dere-pressed R1 factor and a conjugation systemcomprising a non-thymine incorporating donorwith an ultraviolet-irradiated, non-DNA-syn-thesizing recipient (90), specific labeling of atransferred plasmid could be accomplished,and the plasmid DNA could be isolated aftereither SDS or Brij treatment. Immediately

after conjugation, the sex factor in the recip-ient was found to be membrane-bound, withlittle evidence of circularity. Shortly there-after, the label was released from the mem-brane and began to appear in a sedimentationfraction characteristic of an OC structure andlater in a fraction characteristic of CCC mole-cules. The data are consistent with the transferof a linear molecule which attaches to themembrane of the recipient, where its comple-mentary strand is synthesized. The double-stranded structure is then circularized and re-leased from the membrane into the cytoplasmas an OC structure, after which covalent ringclosure occurs. These experiments were ex-tended to show that entry exclusion by an iso-genic or closely related plasmid in the recip-ient is due to an inhibition early in the replica-tion process so as to prevent membrane at-tachment, and, even under circumstanceswhere entry exclusion was circumvented byuse of aeration to produce R- "phenocopies,"incompatibility was manifest as a reduction intotal DNA replication of the transferredplasmid and by the appearance of uncharacter-istic sedimentation fractions, not representa-tive of the OC and CCC structures usuallyfound in F-R- recipient cells.

CONCLUSIONSThe study of bacterial plasmids can be justi-

fied at two levels. In the control of infectiousdiseases, it is obviously necessary to under-stand the basis of the rapid proliferation ofsuch elements as infectious drug-resistancefactors with a view to curtailing their furtherspread and attempting to reduce both theirincidence and their limitations of antibioticchemotherapy. More importantly perhaps,plasmids may be a fundamental entity ofmany bacterial species, involved in "protec-tive" activities other than infectious drug re-sistance, and they may well have their coun-terparts in cells of higher organisms.

Central to the further study of bacterialplasmids is an understanding of their replica-tion and segregation. The molecular nature ofbacterial plasmids is now sufficiently well es-tablished so that more meaningful questions ofreplication can be posed. Such information iscritical to an understanding of curing and in-compatibility relationships. The further estab-lishment of cross-homologies between plasmidsthat are likely to come from electron micros-copy studies of heteroduplex formation (64),

VOL. 36, 1972 395


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

together with related studies on fine geneticstructure, will ultimately play a major role inthe unraveling of the evolutionary relation-ships. The key problem, however, remains: therole of replication and the role played by thecell and the cell membrane. The relationshipof stringent and relaxed control to membranecontrol is at present difficult to assess. Forexample, apparent differences in membrane orcell attachment could be inferred by differ-ences in the segregation into minicells of dif-ferent stringently regulated plasmids includingF. F acts as might be expected of a plasmidwith a fixed cellular site and is not efficientlysegregated to minicells (140), whereas otherfactors apparently segregate so efficiently as toappear in all minicells (127, 128, 156, 157, 195,196). The control of F replication appears to bedifferent from that of chromosomal replica-tion, its initiation being uncoordinated withchromosomal initiation, so that over a range ofgenerations times, irrespective of the numberof chromosomal initiations and their incidencein time during the course of the cell cycle, Finitiation is unchanged and occurs in themiddle of the cell cycle (253). Furthermore, thereplication of newly transferred penicillinaseplasmids in S. aureus is not immediately ac-companied by hereditary stabilization, and thetwo events are separated in time and thus maybe distinct events (180). Observations on"cleared lysates" obtained by the use of Brij-DOC may indicate that all plasmids so farexamined, including F, can be freed of attach-ments to cell components (111), often inferredto be membranes (96), under conditions wheremembrane attachment of the chromosome ismaintained. Conflicting evidence of "mem-brane-attachment" has been reported using adifferent method of preparation (the M' bandmethod [226]-by direct lysis of lysozyme-EDTA spheroplasts with Sarkosyl on sucrosegradients. DNA in minicells derived from an Rfactor, or from F ColV ColB trp cys (neitherplasmid investigated in "cleared lysates") orfrom an F+ parent was concluded to be mem-brane associated from its co-sedimentationwith the M band (213).Promising new approaches have come from

studies (57, 179, Kingsbury and Helinski, Bac-teriol. Proc., 1970, p. 55; 247) of temperature-sensitive mutants of plasmid-carrying strains,i.e., a class unable to continue plasmid replica-tion at an elevated temperature (40 C) but ableto replicate normally at a temperature belownormal (30 C). Some of these mutations arelocated in the plasmid and others in the chro-

mosome. The ability of other wild-type plas-mids to complement the plasmid mutants, orbe affected by the same chromosomal muta-tion, offers a key to the understanding of theinterrelationship of plasmids of various groupsand its influence on incompatibility, fertilityrepression, and other characteristics. Mainte-nance of the small ColEl plasmid has alsobeen shown to depend on the Kornberg poly-merase (142), and it may also need a diffusible,chromosomally determined (by the dnaA lo-cus) initiator for replication (97). More strik-ingly perhaps, bacterial mutants temperaturesensitive in their DNA replication, which havebeen shown to belong to the fast-stop dnaEgroup and which are now known to produce atemperature-sensitive DNA-polymerase IIIenzyme (94) (suggested to be the bactericalreplicase), may continue to replicate ColEl atthe nonpermissive temperature (98). The spe-cific regulation of some (or all) bacterialplasmid replication through the use of specificreplicating enzymes (DNA polymerases I, II, orIII, and others unknown) is thus an intriguingpossibility.One of the more interesting recent develop-

ments in the study of bacterial plasmids hasbeen the establishment of the class of R-factorplasmid aggregates. Although many R factorsare unitary DNA molecules (plasmid cointe-grates) controlling both drug resistance andtransfer properties, others are represented bythe simultaneous presence in the same cell oftwo or more autonomous replicons differing inphysical properties and in the genes carried,exemplified by the A (resistance transfer)factor and the r (drug-resistance determinant)elements A or S and probably present in othernatural systems (220). Moreover, the break-down of certain cointegrates into componentreplicons indicates an intermediate stage be-tween these two types and suggests an evolu-tion whereby aggregates may evolve into coin-tegrates.The model of R-factor plasmid aggregates

evolving through mutation of extrachromo-somal elements might have some virtue as ageneral mechanism for bacterial evolution.Most bacterial genes are normally not active.They are switched off, except in special envi-ronments where their functions are needed.The frequency with which different functionalactivities are required is likely to vary enor-mously. If a particular activity (determined bya number of sequentially acting enzymes) isrequired very infrequently, the necessity toreplicate the determinant gene(s) at every rep-

396 CLOWES


mbr.asm

.org/D

ownloaded from




lication of the chromosome may in fact imposemore of an evolutionary burden on the speciesthan is compensated for by the rare need forthe function. However, if the determinantgenes were carried on a plasmid, present inonly a few cells per population, their replica-tion would be necessary in only those rarecells. Yet, with the imposition of a need for theactivity, these genes could be acquired by allthe cells in a progeny population, if theplasmid also had sex factor properties (i.e.,was a plasmid cointegrate), or, more economi-cally still, if other cells contained transfer fac-tors which could interreact to establish a"plasmid aggregate." The control of a numberof activities in this way by their distribution toa number of different plasmids carried by dif-ferent cells in the population would enor-mously increase the total gene pool and extendthe overall metabolic potential of the species.It would appear to represent a logical exten-sion of genetic regulation, from the presentconcept of cellular control of protein synthesisto the species control of gene-pool synthesis.The ability of such a control system to func-tion by responding to selective pressure ap-pears to be well demonstrated in the case ofinfectious antibiotic resistance in pathogenicSalmonella (6). Moreover, enzymes which areconcerned in catabolic systems rather thanprotective systems have recently also beenshown to be located on plasmids in Pseudo-moras-a species noted for its nutritional ver-satility. A CAM plasmid carrying inducibleenzymes for the conversion of D-camphor toisobutyrate has been described in Pseudo-monas putida and is self-transmitted to otherpseudomonads, including P. aeruginosa, P.fluorescens, and P. oleovorans (Chakrabartyand Gunsalus, Bacteriol Proc., 1971, p. 46) andmore recently, a similar SAL plasmid control-ling salicyclic acid breakdown (Chakrabarty,Abstr. Meeting Amer. Soc. Microbiol. 1972, p.60). These systems may be concerned withrarely required functions or, as Chakrabartyand Gunsalus (32) describe it, "enzymes con-cerned with peripheral metabolism." The pres-ence of fertility factors has been known formany years in pseudomonads (121, 122), and Rfactors have recently been noted in this genus(60, 224), which would further implicate it as apossible reservoir of plasmids.Bacterial plasmids have often been regarded

as a virus subclass of rather lesser importance,but their properties are now divergent enoughto consider them as a class within their ownright. Although they resemble integrating

temperate phages by their ability to establish"symbiosis at the genetic level," they may bethought to be more successful in this role.Their genetic symbiosis, being frequently ofthe nonintegrated kind, may be less intimate,but it is probably less restrictive and is limitedperhaps only to a DNA sequence able to existas a stable replicon. Their dissemination doesnot depend upon the destruction of their host,and their transfer may be thought to be moresubtle yet more direct. The elaboration of acomplex system of proteins, often abandonedonce a virus achieves its end of transferring itsDNA into a new host, is replaced by the syn-thesis of those relatively few proteins requiredfor the conversion of the host strain into adonor for the plasmid and is further reservedfor emergency use so that the host is not di-verted into this activity without purpose. Moreimportantly, the benefits brought to the hostmay well outweigh those of mere defensivesurvival and may embrace such a range of ca-tabolic activities as to qualify bacterial plas-mids as true small "supernumerary chromo-somes" (110), and as such a vital part of thegene pool of the species.

ACKNOWLEDGMENTSThe basis of this review was presented at a meeting on

Bacterial Plasmids convened by the Royal Society, London,on 10 and 11 June, 1971. Much of the work reported wasperformed in collaboration with P. Kontomichalou, C. Milli-ken, M. Mitani, and T. Nisioka, with support from PublicHealth Service research grants from the National Institutesof General Medical Sciences (GM-14394 and GM-13234) andfrom the Institute of Allergy and Infectious Diseases (AI-10468).

I am grateful to Michiko Egel-Mitani for a number ofpreviously unpublished electron microscopy measurements.I acknowledge with thanks the helpful comments and criti-cisms of my colleagues, D. J. McCorquodale and M. H. Pa-trick, in the preparation of this article. I especially thankDimitrij Lang, to whom I am indebted for his expert adviceon electron microscopy and other aspects of macromolecularphysical chemistry.

LITERATURE CITED1. Achtman, M., N. Willetts, and A. J. Clark.

1971. Beginning a genetic analysis of conjuga-tional transfer determined by the F factor inEscherichia coli by isolation and characteri-zation of transfer-deficient mutants. J. Bac-teriol. 106:529-538.

2. Achtman, M., N. Willetts, and A. J. Clark.1972. Conjugational complementation anal-ysis of transfer-deficient mutants of Flac inEscherichia coli. J. Bacteriol. 110:831-842.

3. Adler, H. I., W. D. Fisher, A. Cohen, and A. A.Hardigree. 1967. Miniature Escherichia colicells deficient in DNA. Proc. Nat. Acad. Sci.U.S.A. 57:321-326.

4. Alfoldi, L., F. Jacob, E. L. Wollman, and R.

VOL. 36, 1972 397


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

Maze. 1958. Sur le determinisme genetiquede la colicinogenie. C. R. Acad. Sci. 246:3531-3533.

5. Anderson, E. S. 1966. Influence of the Atransfer factor on the phage sensitivity ofsalmonellae. Nature (London) 212:795-799.

6. Anderson, E. S. 1968. Drug resistance in Sal-monella and its implication. Brit. Med. J. 3:333-339.

7. Anderson, E. S. 1969. Ecology and epidemi-ology of transferable drug resistance, p. 102-119. In G. E. W. Wolstenholme and M. O'-Connor (ed.), Bacterial episomes and plas-mids. J. & A. Churchill Ltd., London.

8. Anderson, E. S., and N. Datta. 1965. Resistanceto penicillin and its transfer in Enterobac-teriaceae. Lancet 7382:407-409.

9. Anderson, E. S., and M. J. Lewis. 1965. Char-acterization of a transfer factor associatedwith drug resistance in Salmonella typhimu-rium. Nature (London) 208:843-849.

10. Anderson, E. S., and E. Natkin. 1972. Trans-duction of resistance determinants and R fac-tors of the A transfer systems by phage Plkc.Mol. Gen. Genet. 114:261-265.

11. Anderson, E. S., and D. Perret. 1967. Mobiliza-tion of transduced tetracycline resistance bythe A transfer factor in Salmonella typhimu-rium and S. typhi. Nature (London) 214:810-811.

12. Arber, W. 1960. Transduction of chromosomalgenes and episomes in Escherichia coli. Vi-rology 11:273-288.

13. Bauer, W., and J. Vinograd. 1968. The interac-tion of closed circular DNA with intercalativedyes. I. The super-helix density of SV40 DNAin the presence and absence of dye. J. Mol.Biol. 33:141-171.

14. Bazaral, M., and D. R. Helinski. 1968. CircularDNA forms of colicinogenic factors El, E2and E3 from Escherichia coli. J. Mol. Biol.36:185-194.

15. Bazaral, M., and D. Helinski. 1970. Replicationof a bacterial plasmid and episome in Esche-richia coli. Biochemistry 9:399-406.

16. Bazzicalupo, P., and G. P. Tocchini-Valentini.1972. Curing of an Escherichia coli episomeby rifampicin (acridine orange/F+/F-/Hfr/lac). Proc. Nat. Acad. Sci. U.S.A. 69:298-300.

17. Benbow, R. M., M. Eisenberg, and R. L. Sinsh-eimer. 1972. Multiple length DNA moleculesof bacteriophage Ox174. Nature N. Biol.237:141-144.

18. Blair, D. G., D. B. Clewell, D. J. Sheratt, andD. R. Helinski. 1971. Strand-specific super-coiled DNA-protein relaxation complexes:comparison of the complexes of bacterialplasmids ColEl and ColE2. Proc. Nat. Acad.Sci. U.S.A. 68:210-214.

19. Bolton, E. T., and B. J. McCarthy. 1962. Ageneral method for the isolation of RNAcomplementary to DNA. Proc. Nat. Acad.Sci. U.S.A. 48:1390-1397.

20. Bouanchaud, D. H., and Y. A. Chabbert. 1969.Stable co-existence of three factors (fi-) inSalmonella panama and Escherichia coliK12. J. Gen. Microbiol. 58:107-113.

21. Bouanchaud, D. H., M. R. Scavizzi, and Y. A.Chabbert. 1968. Elimination by ethidiumbromide of antibiotic resistance in enterobac-teriaceae and staphylococci. J. Gen. Micro-biol. 54:417-425.

22. Brenner, D. J., and S. Falkow. 1971. Molecularrelationships among members of the Entero-bacteriaceae. Advan. Genet. 16:81-118.

23. Brinton, C. C., Jr. 1965. The structure, func-tion, synthesis and genetic control of bac-terial pili and a molecular model for DNAand RNA transport in gram-negative bac-teria. Trans. N.Y. Acad. Sci. 27:1003-1054.

24. Britten, R. J., and D. E. Kohne. 1968. Repeatedsequences in DNA. Science 161:529-540.

24a. Brutlag, D., R. Schekman, and A. Kornberg.1971. A possible role for RNA polymerase inthe initiation of M13 DNA synthesis. Proc.Nat. Acad. Sci. U.S.A. 68:2826-2829.

25. Bujard, H. 1968. Studies on circular DNA. II.Number of tertiary turns in Papilloma DNA.J. Mol. Biol. 33:503-505.

26. Burgi, E., and A. D. Hershey. 1963. Sedimen-tation rate as a measure of molecular weightofDNA. Biophys. J. 3:309:321.

27. Burton, A., and R. L. Sinsheimer. 1965. Theprocess of infection with bacteriophagekx174. VII. Ultracentrifugal analysis of thereplicative form. J. Mol. Biol. 14:327-347.

28. Cairns, J. 1963. The chromosome of Escherichiacoli. Cold Spring Harbor Symp. Quant. Biol.28:43-46.

29. Campbell, A. M. 1962. Episomes. Advan.Genet. 11:101-145.

30. Carlton, B. C., and D. R. Helinski. 1969. Heter-ogeneous circular DNA elements in vegeta-tive cultures of Bacillus megaterium. Proc.Nat. Acad. Sci. U.S.A. 64:592-599.

31. Caro, L. G., and C. M. Berg. 1968. Chromo-somal replication in some strains of E. coli.Cold Spring Harb. Symp. Quant. Biol. 33:559-573.

32. Chakrabarty, A. M., and I. C. Gunsalus. 1969.Defective phage and chromosomal mobiliza-tion in Pseudomonas putida. Proc. Nat.Acad. Sci. U.S.A. 64:1217-1223.

33. Clark, A. J., and A. D. Margulies. 1965. Isola-tion and characterization of recombination-deficient mutants of Escherichia coli K12.Proc. Nat. Acad. Sci. U.S.A. 53:451-459.

34. Clewell, D. B. 1972. Nature of Col E, plasmidreplication in Escherichia coli in the presenceof chloramphenicol. J. Bacteriol. 110:667-676.

35. Clewell, D. B., B. Evenchik, and J. W. Cran-ston. 1972. Direct inhibition of ColElplasmid DNA replication in Escherichia coliby rifampicin. Nature N. Biol. 237:29.

36. Clewell, D. B., and D. R. Helinski. 1969. Super-coiled circular DNA-protein complex in

398 CLOWES


mbr.asm

.org/D

ownloaded from



VOL. 36, 1972 MOLECULAR STRUCTURE

Escherichia coli: purification and inducedconversion to an open circular DNA form.Proc. Nat. Acad. Sci. U.S.A. 62:1159-1166.

37. Clewell, D. B., and D. R. Helinski. 1970. Evi-dence for the existence of the colicinogenicfactors E2 and E3 as supercoiled DNA-pro-tein relaxation complexes. Biochem. Biophys.Res. Commun. 40:608-613.

38. Clewell, D. B., and D. R. Helinski. 1970. Exist-ence of the colicinogenic factor-sex factorColIb-P9 as a supercoiled circular DNA-pro-tein relaxation complex. Biochem. Biophys.Res. Commun. 41:150-156.

39. Clewell, D. B., and D. R Helinski. 1970. Prop-erties of a supercoiled deoxyribonucleic acid-protein relaxation complex and strand speci-ficity of the relaxation event. Biochemistry 9:4428-4440.

40. Clewell, D. B., and D. R. Helinski. 1972. Effectof growth conditions on the formation of therelaxation complex of supercoiled ColE,DNA and protein in Escherichia coli. J. Bac-teriol. 110:1135-1146.

41. Clowes, R. C. 1961. Colicin factors as fertilityfactors in bacteria: Escherichia coli. Nature(London) 190:988-989.

42. Clowes, R. C. 1963. Colicin factors and epi-somes. Genet. Res. 4:162-165.

43. Clowes, R. C. 1964. Transfert genetique desfacteurs colicinogenes. (Symp. on Bacterio-cinogy, Paris, May, 1965). Ann. Inst. Pasteur107: suppl. no. 5:74-92.

44. Clowes, R. C. 1970. Molecular weight andnumber of copies of various sex factor DNAmolecules in Escherichia coli. In A. PerezMiravete and D. Peliez (ed.), Xth Interna-tional Congress for Microbiology, Mexico,1970, 58(Cd-1).

45. Clowes, R. C., and E. E. M. Moody. 1966.Chromosome transfer from "recombination-deficient" strains of Escherichia coli K12.Genetics 53:717-726.

46. Clowes, R. C., E. E. M. Moody, and R. H. Prit-chard. 1965. Elimination of extrachromo-somal elements in thymineless strains ofEscherichia coli K12. Genet. Res. 6:147-152.

47. Cohen, A., W. D. Fisher, R. Curtiss III, and H.I. Adler. 1968. The properties of DNA trans-ferred to minicells during conjugation. ColdSpring Harbor Symp. Quant. Biol. 33:635-641.

48. Cohen, S. N., R. P. Silver, A. E. McCoubrey,and P. A. Sharp. 1971. Isolation of catenatedforms of R factor DNA from minicells. Na-ture N. Biol. 231:249-252.

49. Cohen, S. N., and C. A. Miller. 1969. Multiplemolecular species of circular R-factor DNAisolated from Escherichia coli. Nature (Lon-don) 224:1273-1277.

50. Cohen, S. N., and C. A. Miller. 1970. Non-chromosomal antibiotic resistance in bac-teria. II. Molecular nature of R-factors iso-lated from Proteus mirabilis and Escherichiacoli. J. Mol. Biol. 50:671-687.

)F BACTERIAL PLASMIDS 399

51. Cohen, S. N., and C. A. Miller. 1970. Non-chromosomal antibiotic resistance in bac-teria. III. Isolation of the discrete transferunit of the R-factor R1. Proc. Nat. Acad. Sci.U.S.A. 67:510-516.

52. Cohen, S. N., R. P. Silver, P. A. Sharp, and A.E. McCoubrey. 1971. Studies on the molec-ular nature of R factors. Ann. N.Y. Acad.Sci. 182:172-187.

53. Cooper, S., and C. E. Helmstetter. 1968. Chro-mosomal replication and the division cycle ofEscherichia coli B/r. J. Mol. Biol. 31:619-540.

54. Cozzarelli, N. R., R. B. Kelly, and A. Kornberg.1968. A minute circular DNA from Esche-richia coli 15. Proc. Nat. Acad. Sci. U.S.A.60:992-999.

55. Crawford, L. V., and M. J. Waring. 1967. Su-percoiling of polyoma virus DNA measuredby its interaction with ethidium bromide. J.Mol. Biol. 25:23-30.

56. Curtiss, R. 1969. Bacterial conjugation. Annu.Rev. Microbiol. 23:69-136.

57. Cuzin, F., and F. Jacob. 1967. Existence chezEscherichia coli K12 d'une unite genetiquede transmission formee de diffarents repli-cons. Ann. Inst. Pasteur 112:529-545.

58. Datta, N., and R. W. Hedges. 1971. Compati-bility groups among fi- R factors. Nature(London) 234:222-223.

59. Datta, N., and R. W. Hedges. 1972. Host rangesof R factors. J. Gen. Microbiol. 70:453-460.

60. Datta, N., R. W. Hedges, E. J. Shaw, R. B.Sykes, and M. H. Richmond. 1971. Propertiesof an R factor from Pseudomonas aeruginosa.J. Bacteriol. 108:1244-1249.

61. Davidson, N., and W. Szybalski. 1971. Physicaland chemical characteristics of lambda DNAp. 45-82. In A. D. Hershey (ed.), The bacteri-ophage lambda. Cold Spring Harbor Labora-tory, N.Y.

62. Davies, J., M. Brzezinska, and R. Benveniste.1971. R factors: biochemical mechanisms ofresistance to aminoglycoside antibiotics. Ann.N.Y. Acad. Sci. 182:226-233.

63. Davies, J. E., and R. Rownd. 1972. Transmis-sible multiple drug resistance in enterobac-teriaceae. Science 176:758-768.

64. Davis, R. W., M. Simon, and N. Davidson.1971. Electron microscope heteroduplexmethods for mapping regions of base se-quence homology in nucleic acids, p. 413-428.In L. Grossman and K. Moldave (ed.),Methods in enzymology, vol. 21. AcademicPress Inc., New York.

65. DeHaan, P. G., fand A. H. Stouthamer. 1963. F-prime transfer and multiplication of sex-duced cells. Genet. Res. 4:30-41.

66. Demerec, M., E. A. Adelberg, A. J. Clark, andP. E. Hartman. 1966. A proposal for a uni-form nomenclature in bacterial genetics. Ge-netics 54:61-76.

67. DeWitt, W., and D. R. Helinski. 1965. Charac-terization of colicinogenic factor El from anon-induced and a mitomycin C-induced


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

Proteus strain. J. Mol. Biol. 13:692-703.68. Dove, W. F., H. Inokuchi, and W. F. Stevens.

1971. Replication control in phage lambda, p.747-771. A. D. Hershey (ed.), The bacterio-phage lambda. Cold Spring Harbor Labora-tory, New York.

69. Dubnau, E., and W. K. Maas. 1968. Inhibitionof replication of an F'lac episome in Hfr cellsof Escherichia coli. J. Bacteriol. 95:531-539.

70. Dubnau, E., and B. A. D. Stocker. 1964. Ge-netics of plasmids in Salmonella typhimu-rium. Nature (London) 204:1112-1113.

71. Dulaney, E. L., and A. I. Laskin (ed.). 1971."The problems of drug-resistance pathogenicbacteria." Ann. N.Y. A ad. Sci. 182.

72. Egawa, R., and Y. Hirota. 1962. Inhibition offertility by multiple drug-resistance factor(R) in Escherichia coli K-12. Jap. J. Genet.37:66-69.

73. Falkow, S., and R. V. Citarella. 1965. Molecularhomology of F-merogenote DNA. J. Mol.Biol. 12:138-151.

74. Falkow, S., R. V. Citarella, J. A. Wohlheiter,and T. Watanabe. 1966. The molecular natureof R factors. J. Mol. Biol. 17:102-116.

75. Falkow, S., K. K. Haapala, and R. P. Silver.1969. Relationships between extrachromo-somal elements, p. 136-162. In G. E. W. Wol-stenholme and M. O'Connor (ed.), Bacterialepisomes and plasmids. J. and A. ChurchillLtd., London.

76. Falkow, S., L. S. Tompkins, R. P. Silver, P.Guerry, and D. J. LeBlanc. 1971. The replica-tion of R-factor DNA in Escherichia coli K-12following conjugation. Ann. N.Y. Acad. Sci.182: 153-171.

77. Falkow, S., J. A. Wohlheiter, R. V. Citarella,and L. S. Baron. 1964. Transfer of episomicelements to Proteus. I. Transfer of F-linkedchromosomal determinants. J. Bacteriol. 87:209-219.

78. Fiandt, M., Z. Hradecna, H. A. Lozeron, andW. Szybalski. 1971. Electron micrographicmapping of deletions, insertions, inversions,and homologies in the DNAs of coliphageslambda and phi 80, p. 329-354. In A. D. Her-shey (ed.), The bacteriophage lambda. ColdSpring Harbor Laboratory, New York.

79. Fiers, W., and R. L. Sinsheimer. 1962. Thestructure of DNA of bacteriophage Xx174. III.Ultracentrifugal evidence for a ring structure.J. Mol. Biol. 5:424-434.

80. Finnegan, D. J., and N. S. Willetts. 1971. Twoclasses of Flac mutants insensitive to transferinhibition by an F-like R factor. Mol. Gen.Genet. 111:256-264.

81. Foster, T. J., and T. G. B. Howe. 1971. Recom-bination and complementation between Rfactors in Escherichia coli K12. Genet. Res.18:287-297.

82. Franklin, N. C. 1971. Illegitimate recombina-tion, p. 175-194. In A. D. Hershey (ed.), Thebacteriophage lambda. Cold Spring HarborLaboratory, New York.

83. Fredericq, P. 1954. Transduction genetique desproprietes colicinogenes chez Escherichia coliet Shigella sonnei. C. R. Soc. Biol. 148:399-402.

84. Fredericq, P. 1969. The recombination of coli-cinogenic factors with other episomes andplasmids, p. 163-174. In G. E. W. Wolsten-holme and M. O'Connor (ed.), Bacterial epi-somes and plasmids. J. and A. ChurchillLtd., London.

85. Freifelder, D. 1968. Studies on Escherichia colisex factors. III. Covalently closed Flac DNAmolecules. J. Mol. Biol. 34:31-38.

86. Freifelder, D. 1968. Studies on Escherichia colisex factors. IV. Molecular weights of theDNA of several F' elements. J. Mol. Biol. 35:95-102.

87. Freifelder, D. 1970. Molecular weights of coli-phages and coliphage DNA. IV. Molecularweights of DNA from bacteriophages T4, T5and T7 and the general problem of determi-nation of M. J. Mol. Biol. 54:567-577.

88. Freifelder, D. 1970. Isolation of extrachromo-somal DNA from bacteria, p. 153-163. In L.Grossman and K. Moldave (ed.), Methods inenzymology, vol. 21. Academic Press Inc.,New York.

89. Freifelder, D., A. Folkmanis, and I. Kirschner.1971. Studies on Escherichia coli sex factors:evidence that covalent circles exist withincells and the general problem of isolation ofcovalent circles. J. Bacteriol. 105:722-727.

90. Freifelder, D. R., and D. Freifelder. 1968.Studies on Escherichia coli sex factors. I.Specific labeling of F'lac DNA. J. Mol. Biol.32:15-23.

91. Freifelder, D. R., and D. Freifelder. 1968.Studies on Escherichia coli sex factors. II.Some physical properties of F'lac DNA and FDNA. J. Mol. Biol. 32:25-35.

92. Frydman, A., and E. Meynell. 1969. Interac-tions between de-repressed F-like R factorsand wild type colicin B factors: superinfec-tion immunity and repressor susceptibility.Genet. Res. 14:315-332.

93. Fuke, M., and J. Inselburg. 1972. Electron mi-croscopic studies of replicating and catenatedcolicin factor D1 DNA isolated from mini-cells. Proc. Nat. Acad. Sci. U.S.A. 69:89-92.

94. Gefter, M. L., Y. Hirota, T. Kornberg, J. A.Wechsler, and C. Barnoux. 1971. Analysis ofDNA polymerases II and III in mutants ofEscherichia coli thermosensitive for DNAsynthesis. Proc. Nat. Acad. Sci. U.S.A. 68:3150-3153.

95. Gilbert, W., and D. Dressler. 1968. DNA repli-cation: the rolling circle model. Cold SpringHarb. Symp. Quant. Biol. 33:473-484.

96. Godson, G. N., and R. L. Sinsheimer. 1967.Lysis of Escherichia coli with a neutral deter-gent. Biochim. Biophys. Acta 149:476-488.

97. Goebel. W. 1970. Studies on extrachromosomalDNA elements. Replication of the colicino-genic factor ColEl in two temperature-sensi-

400 CLOWES


mbr.asm

.org/D

ownloaded from




tive mutants of Escherichia coli defective inDNA replication. Eur. J. Biochem. 15:311-320.

98. Goebel, W. 1972. Replication of the DNA of thecolicinogenic factor El (Col El) at the restric-tive temperature in a DNA replication mu-tant thermosensitive for DNA polymerase III.Nature N. Biol. 237:67-70.

99. Gottesman, M. E., and R. A. Weisberg. 1971.Prophage insertion and excision, p. 113-138.In A. D. Hershey (ed.), The bacteriophagelambda. Cold Spring Harbor Laboratory,New York.

100. Goulian, M. 1971. Biosynthesis of DNA. Annu.Rev. Biochem. 40:855-898.

101. Grindley, J. N., and E. S. Anderson. 1971. I-like resistance factors with the fi+ character.Genet. Res. 17:267-271.

102. Grubb, W. B., and R. J. O'Reilly. 1971. Jointtransduction of separate extrachromosomaldrug resistance determinants in Staphylo-coccus aureus E169. Biochem. Biophys. Res.Commun. 42:377-383.

103. Haapala, D. K., and S. Falkow. 1971. Physicalstudies of the drug-resistance transfer factorin Proteus. J. Bacteriol. 106:294-295.

104. Hahn, F. E., and J. Ciak. 1971. Elimination ofbacterial episomes by DNA-complexing com-pounds. Ann. N.Y. Acad. Sci. 182:295-304.

105. Harada, K., M. Kameda, M. Suzuki, and S.Mitsuhashi. 1963. Drug resistance of entericbacteria. II. Transduction of transmissibledrug-resistance (R) factors with phage ep-silon. J. Bacteriol. 86:1332-1338.

106. Harada, K., M. Kameda, M. Suzuki, and S.Mitsuhashi. 1964. Drug resistance of entericbacteria. HI. Acquisition of transferability ofnontransmissible R(TC) factor in cooperationwith F factor and formation of FR(TC). J.Bacteriol. 88:1257-1265.

107. Harada, K., M. Kameda, M. Suzuki, S. Shige-hara, and S. Mitsuhashi. 1967. Drug resist-ance of enteric bacteria. VmI. Chromosomallocation of non-transferable R factor in Esch-erichia coli. J. Bacteriol. 93:1236-1241.

108. Hashimoto, H., and Y. Hirota. 1966. Gene re-combination and segregation of resistancefactor R in Escherichia coli. J. Bacteriol. 91:51-62.

109. Hausmann, C., and R C. Clowes. 1971. ColB2-K77; a fertility-repressed F-like sex factor. J.Bacteriol. 107:900-906.

110. Hayes, W. 1969. The genetics of bacteria andtheir viruses. Wiley, New York.

111. Helinski, D. R., and D. B. Clewell. 1971. Cir-cular DNA. Annu. Rev. Biochem. 40:899-942.

112. Herschberger, C. L., S. Michel, and R. Rownd.1969. Structure of R factor DNA in situ. Fed.Proc. (Abstr.) 28:466.

113. Hershey, A. D. (ed.). The bacteriophagelambda. Cold Spring Harbor Laboratory,New York.

114. Hershey, A. D., and E. Burghi. 1965. Comple-mentary structure of interacting sites at theends of the X DNA molecules. Proc. Nat.

Acad. Sci. U.S.A. 53:325-328.115. Hickson, F. T., T. R. Roth, and D. R. Helinski.

1967. Circular DNA forms of a bacterial sexfactor. Proc. Nat. Acad. Sci. U.S.A. 58:1731-1738.

116. Hirota, Y. 1960. The effect of acridine dyes onmating type factors in Escherichia coli. Proc.Nat. Acad. Sci. U.S.A. 46:57-64.

117. Hirota, Y., J. Mordoh, and F. Jacob. 1970. Onthe process of cellular division in Escherichiacoli. ImI. Thermosensitive mutants of Esche-richia coli altered in the process of DNA ini-tiation. J. Mol. Biol. 53:369-387.

118. Hirota, Y., A. Ryter, and F. Jacob. 1968. Ther-mosensitive mutants of E. coli affected in theprocesses of DNA synthesis and cellular divi-sion. Cold Spring Harbor Symp. Quant. Biol.33:677-693.

119. Hoar, D. I. 1970. Fertility regulation in F-likeresistance transfer factors. J. Bacteriol. 101:916-920.

120. Hohn, B., and D. Kom. 1969. Cosegregation ofa sex factor with the Escherichia coli chromo-some during curing by acridine orange. J.Mol. Biol. 45:385-395.

121. Holloway, B. W. 1969. Genetics of Pseudo-monas. Bacteriol. Rev. 33:419-443.

122. Hollowayj B. W., V. Krishnapillai, and V. Stan-isich. 1971. Pseudomonas genetics. Annu.Rev. Genet. 5:425-445.

123. Hradecna, Z., and W. Szybalski. 1967. Frac-tionation of the complementary strands ofcoliphage DNA based on the asymmetric dis-tribution of the poly U,G-binding sites. Vi-rology 32:633-543.

124. Hudson, B., D. A. Clayton, and J. Vinograd.1968. Complex mitochondrial DNA. ColdSpring Harbor Symp. Quant. Biol. 33:435-442.

125. Ikeda, H., M. Inuzuka, and J. Tomizawa. 1970.P1-like plasmid in Escherichia coli 15. J.Mol. Biol. 50:457-470.

126. Ikeda, H., and J. Tomizawa. 1968. Prophage P1,an extrachromosomal replication unit. ColdSpring Harbor Symp. Quant. Biol. 33:791-798.

127. Inselburg, J. 1970. Segregation into and replica-tion of plasmid deoxyribonucleic acid in chro-mosomeless segregants of Escherichia coli. J.Bacteriol. 102:642-47.

128. Inselburg, J. 1971. R factor deoxyribonucleicacid in chromosomeless progeny of Esche-richia coli. J. Bacteriol. 105:620-628.

129. Inselburg, J., and M. Fuke. 1970. ReplicatingDNA: structure of colicin factor El. Science169:590-592.

130. Inselburg, J., and M. Fuke. 1971. Isolation ofcatenated and replicating DNA molecules ofcolicin factor El from minicells. Proc. Nat.Acad. Sci. U.S.A. 68:2839-2842.

131. Inuzuka, N., S. Nakamura, M. Inuzuka, and M.Tomoeda. 1969. Specific action of sodiumdodecyl sulfate on the sex factor of Esche-richia coli K-12 Hfr strains. J. Bacteriol. 100:827-835.

401VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



BACrMuIOL. REV.

132. Ippen-Ihler, K., M. Achtman, and N. Willetts.1972. Deletion map of the Escherichia coli K-12 sex factor F; the order of eleven transfercistrons. J. Bacteriol. 110:857-863.

133. Jacob, F., S. Brenner, and F. Cuzin. 1963. Onthe regulation of DNA replication in bacteria.Cold Spring Harbor Symp. Quant. Biol. 28:329-348.

134. Jacob, F., P. Schaeffer, and E. Wollman. 1960.Episomic elements in bacteria. Symp. Soc.Gen. Microbiol. 10:67.

135. Jacob, F., and E. L. Wollman. 1958. Les6pisomes, elements genetiques ajoutes. C.R.Acad. Sci. 247:154-156.

136. Jacob, F., and E. L. Wollman. 1961. Sexualityand the genetics of bacteria. Academic PressInc., New York.

137. Johnston, J. H., and M. H. Richmond. 1970.The increased rate of loss of penicillinaseplasmids from Staphylococcus aureus in thepresence of rifampicin. J. Gen. Microbiol. 69:137-139.

138. Kahn, P., and D. R. Helinski. 1964. Relation-ship between colicinogenic factors E, and Vand an F factor in Escherichia coli. J. Bac-teriol. 88:1573-1579.

139. Kasamatsu, H., and R. Rownd. 1970. Replica-tion of R-factors in Proteus mirabilis: replica-tion -under relaxed control. J. Mol. Biol. 51:473-489.

140. Kass, L. R., and M. B. Yarmolinsky. 1970. Seg-regation of a functional sex factor into mini-cells. Proc. Nat. Acad. Sci. U.S.A. 66:815-822.

141. Khatoon, H., and R. V. Iyer. 1971. Stable coex-istence of Rfi- factors in Escherichia coli.Can. J. Microbiol. 17:669-675.

142. Kingsbury, D. T., and D. R. Helinski. 1971.DNA polymerase as a requirement for themaintenance of the bacterial plasmid colicin-ogenic factor E,- Biochem. Biophys. Res.Commun. 41:1538-1544.

143. Kleinschmidt, A., and R. K. Zahn. 1959. UberDesoxyribonucleinsaure-Molekeln in Protein-Mischfilmen. Z. Naturforsch. 14b:770-779.

144. Kline, B. C., and D. R. Helinski. 1971. F, sexfactor of Escherichia coli. Size and purifica-tion in the form of a strand-specific relaxa-tion complex of supercoiled deoxyribonucleicacid and protein. Biochemistry 10:4975-4980.

145. Kontomichalou, P., M. Mitani, and R. C.Clowes. 1970. Circular R-factor moleculescontrolling penicillinase synthesis, replicatingin Escherichia coli under either relaxed orstringent control. J. Bacteriol. 104:34-44.

146. Kopecko, D. J., and J. D. Punch. 1971. Regula-tion of R-factor replication in Proteus mira-bilis. Ann. N.Y. Acad. Sci. 182:207-216.

147. Lang, D. 1970. Molecular weights of coliphagesand coliphage DNA. HI. Contour length andmolecular weight of DNA from bacterio-phages T4, T5, and T7, and from bovine pap-illoma virus. J. Mol. Biol. 54:557-565.

148. Lang. D. 1971. Individual macromolecules:

preparation and recent results with DNA.Phil. Trans. Roy. Soc. London B Ser. 261:151-158.

149. Lang, D., H. Bujard, B. Wolff, and D. Russell.1967. Electron microscopy of size and shapeof viral DNA in solutions of different ionicstrengths. J. Mol. Biol. 23:163-181.

150. Lang, D., and M. Mitani. 1970. Simplifiedquantitative electron microscopy of biopoly-mers. Biopolymers 9:373-379.

151. Lavalle, R., and F. Jacob. 1961. Sur lasensibilite des episomes sexuel etcolicinogene d'Escherichia coli K12 a ladesintegration du radiophosphore. C.R. Acad.Sci. 252:1678-1680.

152. Lebek, G. 1963. Ueber die Enstehung mehrfach-resistanter Salmonellen. Ein experimentellerBeitrag. Zentralbl. Bakteriol. Parasitenk. In-fektionskr. Abt. I Orig. 188:494-505.

153. Lederberg, J. 1951. Streptomycin resistance: agenetically recessive mutation. J. Bacteriol.61:549-550.

154. Lee, C. S., and N. Davidson. 1968. Covalently-closed minicircular DNA in Micrococcus ly-sodeikticus. Biochem. Biophys. Res.Commun. 32:757-762.

155. Levy, S. B. 1970. Resistance of minicells topenicillin lysis: a method of obtaining largequantities of purified minicells. J. Bacteriol.103:836-839.

156. Levy, S. B. 1971. Studies on R factors segre-gated into E. coli minicells. Ann. N.Y.Acad. Sci. 182:217-225.

157. Levy, S. B. 1971. Physical and functional char-acteristics of R factor deoxyribonucleic acidsegregated into Escherichia coli minicells. J.Bacteriol. 108:300-308.

158. Levy, S. B., and P. Norman. 1970. Segregationof transferable R factors into Escherichia coliminicells. Nature (London) 227:606-607.

159. Luzzati, V., A. Mathis, F. Masson, and J. Witz.1964. Structure transitions observed in DNAand polyA in solution as a function of tem-perature and pH. J. Mol. Biol. 10:28-41.

160. Macfarren, A. C., and R. C. Clowes. 1967. Acomparative study of two F-like colicin fac-tors, ColV2 and ColV3, in Escherichia coli K-12. J. Bacteriol. 94:365-377.

161. MacHattie, L. A., K. I. Berns, and C. A.Thomas, Jr. 1965. Electron microscopy ofDNA from Hemophilus influenzae. J. Mol.Biol. 11:648-649.

162. Marmur, J. 1961. A procedure for the isolationof deoxyribonucleic acid from micro-organ-isms. J. Mol. Biol. 3:208-218.

163. Marmur, J., S. Falkow, and M. Mandel. 1963.New approaches to bacterial taxonomy.Annu. Rev. Microbiol. 17:329-372.

164. Marmur, J., R. Rownd, S. Falkow, L. S. Baron,C. Schildkraut, and P. Doty. 1961. The na-ture of intergeneric episomal infection. Proc.Nat. Acad. Sci. U.S.A. 47:972-979.

165. Martuscelli, J., A. L. Taylor, D. J. Cummings,V. A. Chapman, S. S. DeLong, and L.

402 CLOWES


mbr.asm

.org/D

ownloaded from




Cinedo. 1971. Electron microscopic evidencefor linear insertion of bacteriophage MU-1 inlysogenic bacteria. J. Virol. 8:551-563.

166. Matsubara, K., and A. D. Kaiser. 1968. A dv:An autonomously replicating DNA fragment.Cold Spring Harbor Symp. Quant. Biol. 33:769-775.

167. Meynell, E., and M. Cooke. 1969. Repressor-minus and operator-constitutive de-repressedmutants of F-like R factors: their effect onchromosomal transfer by HfrC. Genet. Res.14:309-313.

168. Meynell, E., G. G. Meynell, and N. Datta.1968. Phylogenetic relationships of drug-re-sistance factors and other transmissible bac-terial plasmids. Bacteriol. Rev. 32:55-83.

168a. Milliken, C. E., and R. C. Clowes. 1973. Molec-ular structure of an R factor, its componentdrug resistance determinants and transferfactor. J. Bacteriol., in press.

169. Mitsuhashi, S. 1971. Epidemiology and geneticsof R factors, Ann. N.Y. Acad. Sci. 182:141-152.

170. Mitsuhashi, S., M. Kameda, K. Harada, andM. Suzuki. 1969. Formation of recombinantsbetween nontransmissible drug-resistancedeterminants and transfer factors. J. Bacte-riol. 97:1520-1521.

171. Monk, M., and R. C. Clowes. 1964. Transfer ofthe colicin I factor in Escherichia coli K12and its interaction with the F fertility factor.J. Gen. Microbiol. 36:365-384.

172. Monk, M., and R. C. Clowes. 1964. The regula-tion of colicin synthesis and colicin factortransfer in Escherichia coli K12. J. Gen Mi-crobiol. 36:385-392.

173. Moody, E. E. M., and R. Runge. 1972. The in-tegration of autonomous transmissible plas-mids into the chromosome of Escherichia coliK12. Genet. Res. 19:181-186.

174. Nakaya, R., A. Nakamura, and Y. Murata.1960. Resistance transfer agents in Shigella.Biochem. Biophys. Res. Commun. 3:654-659.

175. Nishimura, Y., L. Cargo, C. M. Berg, and Y.Hirota. 1971. Chromosome replication inEscherichia coli. IV. Control of chromosomereplication and cell division by an integratedepisome. J. Mol. Biol. 55:441-456.

176. Nisioka, T., M. Mitani, and R. C. Clowes. 1969.Composite circular forms of R-factor deoxyri-bonucleic acid molecules. J. Bacteriol. 97:376-385.

177. Nisoka, T., M. Mitani, and R. C. Clowes. 1970.Molecular recombination between R-factordeoxyribonucleic acid molecules in Esche-richia coli host cells. J. Bacteriol. 103:166-177.

178. Novick, R. P. 1969. Extrachromosomal inheri-tance in bacteria. Bacteriol. Rev. 33:210-257.

179. Novick, R. P., and D. Bouanchaud. 1971. Ex-trachromosomal nature of drug resistance inStaphylococcus aureus. Ann. N.Y. Acad. Sci.182:279-294.

180. Novick, R. P., and R. Brodsky. 1971. Studies

on plasmid replication. I. Plasmid incompati-bility and establishment in Staphylococcusaureus. J. Mol. Biol. 68:285-302.

181. Novick, R. P., and M. H. Richmond. 1965.Nature and interactions of the genetic ele-ments governing penicillinase synthesis inStaphylococcus aureus. J. Bacteriol. 90:467-480.

182. Nygaard, A. P., and B. D. Hall. 1964. Forma-tion and properties of RNA-DNA complexes.J. Mol. Biol. 9:125-142.

183. Ohki, M., and J. Tomizawa. 1968. Asymmetrictransfer of DNA strands in bacterial conjuga-tion. Cold Spring Harbor Symp. Quant. Biol.33:651-658.

184. Ohtsubo, E. 1970. Transfer-defective mutantsof sex factors in Escherichia coli. II. Deletionmutants of an F-prime and deletion mappingof cistrons involved in genetic transfer. Ge-netics 64:189-197.

185. Ohtsubo, E., Y. Nishimura, and Y. Hirota.1970. Transfer-defective mutants. of sex fac-tors in Escherichia coli. I. Defective mutantsand complementation analysis. Genetics 64:173-188.

186. Ou, J. T., and T. F. Anderson. 1970. Role of piliin bacterial conjugation. J. Bacteriol. 102:648-654.

187. Ozeki, H., B. A. D. Stocker, and S. M. Smith.1962. Transmission of colicinogeny betweenstrains of Salmonella typhimurium growntogether. J. Gen. Microbiol. 28:671-687.

188. Palchaudhuri, S. R., A. J. Mazaitis, W. K.Mass, and A. K. Kleinschmidt. 1972. Charac-terization by electron microscopy of fused F-prime factors in Escherichia coli. Proc. Nat.Acad. Sci. U.S.A. 69:1873-1876.

188a. Pinney, R. J., and J. T. Smith. 1972. R factorelimination during thymine starvation:effects of inhibition of protein synthesis andreaddition of thymine. J. Bacteriol. 111:361-367.

189. Pritchard, R. H., P. T. Barth, and J. Collins.1969. Control of DNA synthesis in bacteria.Symp. Soc. Gen. Microbiol. 19:263-297.

190. Puig, J., and R. Nagel de Zwaig. 1964. Etudegenetique d'un facteur colicinogene B et deson influence sur la fertilite des croisementschez Escherichia coli K12. Ann. Inst. Pasteur107 suppl., no. 2:115-131.

191. Punch, J. D., and D. J. Kopecko. 1972. Positiveand negative control of R-factor replication inProteus mirabilis. J. Bacteriol. 109:336-349.

192. Radloff, R., W. Bauer, and J. Vinograd. 1967. Adye-buoyant-density method for the detec-tion and isolation of closed circular duplexDNA: the closed circular DNA in HeLa cells.Proc. Nat. Acad. Sci. U.S.A. 57:1514-1520.

193. Rhoades, M., L. A. MacHattie, and C. A.Thomas, Jr. 1968. The P22 bacteriophageDNA molecule. I. The mature form. J. Mol.Biol. 37:21-40.

194. Romero, E., and E. Meynell. 1969. Covert fi- Rfactors in fi+ R+ strains of bacteria. J. Bac-

VOL. 36, 1972 403


mbr.asm

.org/D

ownloaded from



BACTERIOL. REV.

teriol. 97:780-786.195. Roozen, K. J., R. G. Fenwick, Jr., and R. Cur-

tiss III. 1971. Isolation of plasmids and spe-cific chromosomal segments from Escherichiacoli K-12. In L. G. H. Ledoux (ed.), Informa-tive molecules in biological systems. North-Holland Publishing Co., Amsterdam.

196. Roozen, K. J., R. G. Fenwick, Jr., and R. Cur-tiss III. 1971. Synthesis of ribonucleic acidand protein in plasmid-containing minicellsof Escherichia coli K-12. J. Bacteriol. 107:21-33.

197. Roth, T. F., and D. R. Helinski. 1967. Evidencefor circular DNA forms of a bacterialplasmid. Proc. Nat. Acad. Sci. U.S.A. 58:650-657.

198. Rownd, R. 1969. Replication of a bacterial epi-some under relaxed control. J. Mol. Biol. 44:387-402.

199. Rownd, R., H. Kasamatsu, and S. Mickel. 1971.The molecular nature and replication of drugresistance factors of the enterobacteriaceae.Ann. N.Y. Acad. Sci. 182:188-206.

200. Rownd, R., and S. Mickel. 1971. Dissociationand reassociation of RTF and r-determinantsof the R-factor NR1 in Proteus mirabilis.Nature N. Biol. 234:40-43.

201. Rownd, R., R. Nakaya, and A. Nakamura. 1966.Molecular nature of the drug-resistance fac-tors of the enterobacteriaceae. J. Mol. Biol.17:376-393.

202. Rownd, R., H. Watanabe, S. Mickel, R. Na-kaya, and B. Gargan. 1971. The molecularnature and control of replication of R factors,p. 536-544. In Progress in antimicrobial andanticancer chemotherapy. Int. Congr. Chemo-therapy, Proc. VI. Univ. of Tokyo Press.

203. Rubin, S. J., and E. D. Rosenblum. 1971. Ef-fects of the recipient strain and ultravioletirradiation on transduction kinetics of thepenicillinase plasmid of Staphylococcus au-reus. J. Bacteriol. 108:1192-1199.

204. Rupp, W. D., and G. Ihler. 1968. Strand selec-tion during bacterial mating. Cold SpringHarbor Symp. Quant. Biol. 33:647-650.

205. Rush, M. G., C. N. Gordon, R. P. Novick, andR. C. Warner. 1969. Penicillinase plasmidDNA from Staphylococcus aureus. Proc. Nat.Acad. Sci. U.S.A. 63:1304-1310.

206. Rush, M. G., C. N. Gordon, and R. C. Warner.1969. Circular deoxyribonucleic acid fromShigella dysenteriae Y6R. J. Bacteriol. 100:803-808.

207. Sanderson, K. E. 1967. Revised linkage map ofSalmonella typhimurium. Bacteriol. Rev. 31:354-372.

208. Salisbury, V., R. W. Hedges, and N. Datta.1972. Two modes of 'curing' transmissiblebacterial plasmids. J. Gen. Microbiol. 70:443-452.

209. Scaife, J. 1967. Episomes. Annu. Rev. Micro-biol. 21:601-638.

210. Schildkraut, C. L., J. Marmur, and P. Doty.1962. Determination of base composition of

deoxyribonucleic acid from its buoyant den-sity in CsCl. J. Mol. Biol. 4:430-443.

211. Sharp, P. A., M-T. Hsu, E. Ohtsubo, and N.Davidson. 1972. Electron microscope hetero-duplex studies of sequence relations amongplasmids of E. coli. I. Structure of F-primefactors. J. Mol. Biol., in press.

212. Shaw, W. V. 1971. Comparative enzymology ofchloramphenicol resistance. Ann. N.Y. Acad.Sci. 182:234-242.

213. Shull, F. W., Jr., J. A. Fralick, L. P. Stratton,and W. D. Fisher. 1971. Membrane associa-tion of conjugally transferred deoxyribonu-cleic acid in Escherichia coli minicells. J.Bacteriol. 106:626-633.

214. Signer, E. R. 1969. Plasmid formation: a newmode of lysogeny by phase A. Nature (Lon-don) 233:158-160.

215. Silver, R. P., and S. Falkow. 1970. Specific la-beling and physical characterization of R-factor deoxyribonucleic acid in Escherichiacoli. J. Bacteriol. 104:331-339.

216. Silver, R. P., and S. Falkow. 1970. Studies onresistance transfer factor deoxyribonucleicacid in Escherichia coli. J. Bacteriol. 104:340-344.

217. Silver, S., and H. Ozeki. 1962. Transfer ofdeoxyribonucleic acid accompanying thetransmission of colicinogenic properties bycell mating. Nature (London) 195:873-874.

218. Simon, M. N., R. W. Davis, and N. Davidson.1971. Heteroduplexes of DNA molecules oflambdoid phages: physical mapping of theirbase sequence relationships by electron mi-croscopy, p. 313-328. In A. D. Hershey (ed.),The bacteriophage lambda. Cold SpringHarbor Laboratory, New York.

219. Skalka, A., E. Burgi, and A. D. Hershey. 1968.Segmental distribution of nucleotides in theDNA of bacteriophage lambda. J. Mol. Biol.34:1-16.

220. Smith, H. W., and C. L. Gyles. 1970. The rela-tionship between different transmissible plas-mids introduced by F into the same strain ofEscherichia coli K12. J. Gen. Microbiol. 62:277-285.

221. Smith, H. W., and M. A. Lingood. 1970.Transfer factors in Escherichia coli with par-ticular regard to their incidence in entero-pathogenic strains. J. Gen. Microbiol. 62:287-299.

222. Sonstein, S., and J. N. Baldwin. 1972. Loss ofthe penicillinase plasmid after treatment ofStaphylococcus aureus with sodium dodecylsulfate. J. Bacteriol. 109:262-265.

223. Stocker, B. A. D., S. M. Smith, and H. Ozeki.1963. High infectivity of Salmonella typhi-murium newly infected by the colI factor. J.Gen. Microbiol. 30:201-221.

224. Sykes, R. B., and M. H. Richmond. 1970. Inter-generic transfer of a f,-lactamase gene be-tween P. aeruginosa and E. coli. Nature (Lon-don) 226:952-954.

225. Taylor, A. L. 1970. Current linkage map of

404 CLOWES


mbr.asm

.org/D

ownloaded from




Escherichia coli. Bacteriol. Rev. 34:155-175.226. Tremblay, G. Y., M. J. Daniels, and M. Schae-

chter. 1969. Isolation of a cell membrane-DNA-nascent RNA complex from bacteria. J.Mol. Biol. 40:65-76.

227. Vapnek, D., M. B. Lipman, and W. D. Rupp.1971. Physical properties and mechanism oftransfer of R factors in Escherichia coli. J.Bacteriol. 108:508-514.

228. Vapnek, D., and W. D. Rupp. 1970. Asym-metric segregation of the complementary sex-factor DNA strands during conjugation inEscherichia coli. J. Mol. Biol. 53:387-403.

229. Vinograd, J., J. Lebowitz, R. Radloff, R. Wat-son, and P. Laipis. 1965. The twisted circularform of polyoma viral DNA. Proc. Nat. Acad.Sci. U.S.A. 53:1104-1111.

230. Watanabe, T. 1967. Infectious drug resistance.Sci. Amer. 217/6:19-27.

231. Watanabe, T. 1967. The concept of R factors. InS. Falkow (ed.), Symp. Infect. Multiple DrugResistance. U.S. Government Printing Office5-16.

232. Watanabe, T. 1969. Transferable drug resist-ance: the nature of the problem, p. 81-101. InG. E. W. Wolstenholme and M. O'Connor(ed.), Bacterial episomes and plasmids. J. &A. Churchill Ltd., London.

233. Watanabe, T. 1971. The origin of R factors.Ann. N.Y. Acad. Sci. 182:126-140.

234. Watanabe, T., and T. Fukasawa. 1960. "Resist-ance transfer factor" an episome in Entero-bacteriaceae. Biochem. Biophys. Res.Commun. 3:660-665.

235. Watanabe, T., and T. Fukasawa. 1961. Epi-some-mediated transfer of drug resistance inEnterobacteriaceae. I. Transfer of resistancefactors by conjugation. J. Bacteriol. 81:669-678.

236. Watanabe, T., and T. Fukasawa. 1961. Epi-some-mediated transfer of drug resistance inEnterobacteriaceae. II. Elimination of resist-ance factors with acridines. J. Bacteriol. 81:679-683.

237. Watanabe, T., and T. Fukasawa. 1961. Epi-some-mediated transfer of drug resistance inEnterobacteriaceae. III. Transduction of re-sistance factors. J. Bacteriol. 82:202-209.

238. Watanabe, T., C. Furuse, and S. Sakaizumi.1968. Transduction of various R factors byphage P1 in Escherichia coli and by phageP22 in Salmonella typhimurium. J. Bacteriol.96:1791-1795.

239. Watanabe, T., H. Nishida, C. Ogata, T. Arai,and S. Sato. 1964. Episome-mediatedtransfer of drug resistance in Enterobac-

teriaceae. VII. Two types of naturally occur-ring R factors. J. Bacteriol. 88:716-726.

240. Watanabe, T., and Y. Ogata. 1970. Genetic sta-bility of various resistance factors in Esche-richia coli and Salmonella typhimurium. J.Bacteriol. 102:363-368.

241. Watanabe, T., and Y. Ogata. 1970. Abortivetransduction of resistance factor by bacterio-phage P22 in Salmonella typhimurium. J.Bacteriol. 102:596-597.

242. Watanabe, T., C. Ogata, and S. Sato. 1964.Episome-mediated transfer of drug resistancein Enterobacteriaceae. VII. Six drug-resist-ance R factors. J. Bacteriol. 88:922-928.

243. Watanabe, T., and T. Takano. 1963. Episomicresistance factors in Enterobacteriaceae. 18.Inactivation of four drug-resistance R factorsby the decay of incorporated P32. Med. Biol.(Tokyo) 65:111.

244. Wehrli, W., and M. Staehelin. 1971. Actions ofthe rifamycins. Bacteriol. Rev. 35:20-309.

245. Weisblum, B., and J. Davies. 1968. Antibioticinhibitors of the bacterial ribosome. Bacte-riol. Rev. 32:493-528.

246. Willetts, N. S. 1971. Plasmid specificity of twoproteins required for conjugation in E. coliK12. Nature N. Biol. 230:183-185.

247. Willetts, N. S. 1972. The genetics of transmis-sible plasmids. Annu. Rev. Genet. 6, in press.

248. Willetts, N. S., and M. Achtman. 1972. Geneticanalysis of transfer by the Escherichia colisex factor F, using P1 transductional comple-mentation. J. Bacteriol. 110:843-851.

249. Willetts, N. S., and D. J. Finnegan. 1971. Agenetic analysis of conjugational transfer andits control. 1st International Symposium onInfectious (transferable) Antibiotic Resist-ance, Smolenice, Czechoslovakia. Springer-Verlag. Berlin, in press.

250. Wohlheiter, J. A., S. Falkow, and R. V. Citar-ella. 1966. Purification of episomal DNA withcellulose nitrate membrane filters. Biochim.Biophys. Acta 129:475-481.

251. Wolf, B., M. L. Pato, C. B. Ward, and D. A.Glaser. 1968. On the origin and direction ofreplication of the E. coli chromosome. ColdSpring Harbor Symp. Quant. Biol. 33:575-584.

252. Wolstenholme, G. E. W., and M. O'Connor(ed.). 1969. Bacterial plasmids and episomes.J. & A. Churchill, London.

253. Zeuthen, J., and M. L. Pato. 1971. Replicationof the F'lac sex factor in the cell cycle ofEscherichia coli. Mol. Gen. Genet. 111:242-255.

405VOL. 36, 1972


mbr.asm

.org/D

ownloaded from



plasmid

Documents