Cell, Vol. 70, 199-199,

July 24, 1992, Copyright

0 1992 by Cell Press

Review

Bacterial Plasmids and Gene Flux Carlos F. Am&bile-Cuevas’t and Marina E. ChicurelS *Department of Molecular and Cellular Toxicology Harvard School of Public Health Boston, Massachusetts 02115 *Program in Neuroscience Department of Neurobiology Harvard Medical School Boston, Massachusetts 02115

The basis of evolutionary processes resides in the dynamic nature of genetic material: novel DNA sequences are continually being produced by mutation and recombination. Through the stabilization and transmission of these genetic products, evolutionary directions are established. Vertical transmission, the transfer of genes from parents to offspring, is most generally thought to determine these directions. However, it is becoming increasingly apparent that another form of transmission, horizontal transfer, contributes greatly to evolutionary processes in a wide variety of organisms. Horizontal gene transfer refers to the mobilization and stabilization of genetic information from one organism to another. Such exchange is a well-established phenomenon in the prokaryotic kingdom. Among bacteria, genetically plastic transmissible elements are mobilized into host organisms, where modifications and further radial transfer can occur. Mobilization primarily involves extrachromosomal elements, many of which encode systems for their own transfer. A consequence of such mobilization is the existence of a pool of genetic information that is accessible to virtually every bacterial cell. Indeed, it has been proposed that the prokaryotic community may be viewed as a single, heterogenetic, multicellular organism, containing replicons that are in continuous movement from one group of cells to another (Sonea, 1991). This genetic flux, however, is not confined to the Monera kingdom: genetic determinants carried by bacterial plasmids can be transferred to eukaryotic cells by mechanisms that seem to be similar to those employed between prokaryotic organisms. In the first part of this review, recently discovered aspectsof this broad genetic mobilization shared by bacteria, yeast, and plants are discussed. Two particularly striking examples are the transfer of genetic material from Escherichia coli to Saccharomyces cerevisiae and from Agrobacteria to plant cells. In the second part, a visible consequenceof lateral gene transfer isanalyzed: the emergence and evolution of antibiotic resistance in bacteria. We propose that this accumulated data represent the tip of the iceberg in terms of the role played by horizontal transfer in establishing evolutionary directions, and that the leading role in this process is played by autotransmissible bacterial plasmids.

tPresent address: Departamento de Microbiologia, tado Postal 102-006 08930, MBxico, D. F.

LUSARA,

Apar-

Intermolecular

Movement

of Genes

Transposition and other intermolecular mobilization events can group different genes together into a single replicon. Transposons can insert into and excise out of plasmids. The presence of multiple copies of a transposon, due either to multiple transposition events or to its integration in a multicopy plasmid, increases the probability of recombinational events involving not only the transposon but also neighboring sequences. Although, in general, transposons per se are not directly involved in the horizontal transfer of genes, their ability to incorporate into autotransmissible plasmids results in the potential for any of the genes they carry to contribute to the expansion of the gene pool involved in gene flux between organisms (see Table 1). Based on their functional structure, prokaryotic transposons have been classified into two major groups. Class I transposons include both insertion sequences (ISs), which are elements solely encoding the enzymatic machinery required for transposition, and composite transposons, which can be thought of as genes trapped between two identical ISs. Class II transposons are DNA sequences flanked by small inverted repeats that contain transposition and resolution enzyme genes in association with other genes. In both cases, the nature of the associated genes is generally irrelevant to the transposition process itself, suggesting that the prevalence of these associated genes is primarily due to the selective advantage they confer on their hosts. For example, they can encode the catabolism of unusual organic compounds or antibiotic resistances. The study of Tn21 and other related transposons and plasmids has uncovered another kind of genetic mobilization involving units found within the transposons themselves. This form of genetic integration is mediated by recently described elements named integrons(Stokes and Hall, 1989). lntegrons consist of a 59 bp sequence, the target for recombinational crossover, and a sequence encoding a polypeptide, possibly an integrase. Recombination at integron hot spots seems to result in the incorporation of genetic modules within individual transposons. Interestingly, the movement of genetic elements in a modular fashion seems to be characteristic of plasmid biology and evolution. For example, the RepFllA family of conjugative plasmids seems to have exchanged modules containing regulatory systems and incompatibility determinants (Lbpez et al., 1991). Multicopy plasmids of grampostive bacteria have also apparently exchanged genetic cassettes bearing resistance determinants, genes encoding replication proteins, and mobilization genes (Projan and Moghazeh, 1991). In addition to a mechanism for intermolecular transfer, ISs and integrons occasionally provide promoters that activate the expression of integrated or neighboring genes. Insertion sequences called ISp elements can activate genes that lack promoters. Activation can be due either to a complete promoter in the ISp or to the creation of a

Cell 190

Table

1. Pool for Interspecies

Gene

Flux Location

Trait

Tn

“Fertility

+

Mutagenic

DNA repair

in Bacterial

Genome

Plasmid

Chromosome

Horizontal

+

+

0-0”

+

+

B-B

Resistance to: Antibiotics Metal ions Bacteriophages

+ +

+ + +

+ +

B-B B-B B-0

Production of: Antibiotics Insecticidal toxins

+

+ +

+

B-0 B-B,

+

+

B-B

+

+

B-0,

+

+

B-0

Virulence Metabolic

pathways

Degradation

of halogenated

Plant tissue

proliferation

Opine

compounds

+

production

Endosymbiotic Nitrogen

organic

associations

fixation

Intercellular

Movement

of Genes

Among horizontal gene transfer mechanisms, those specified by conjugative plasmids seem to be the most sophisticated and widespread. Conjugative mechanisms permit gene flux among virtually all eubacteria, yeast, and even plant cells (Figure 1). Conjugative (or conjugative-like) mobilization is, in addition, the only known system in which the transferred DNA encodes the elements required for its own intercellular movement. Conjugation between Bacteria Conjugation is a process by which transfer of DNA from a donor to a recipient cell occurs during cell-cell contact. Almost without exception, it is mediated by plasmids. Conjugation has been observed in gram-negative and grampositive bacteria and even between members of the two groups. The conjugative mechanisms of gram-negative bacteria are the best understood, and our discussion focuses on this group, particularly the IncF plasmids (reviewed in Willets and Wilkins, 1984). For the initial mating contact to occur between conjugating bacteria, the donor cell must produce pili, long, thin, hollow protein cylinders. It is controversial whether these structures are merely involved in a contractile mechanism to bring the cells into direct contact to permit membrane

b-y

B-0,

B-P

+

B-B,

B-P

+

ND

+

+

ND

a B-B, bacteria to bacteria; B-P, bacteria to plant cell; B-Y, bacteria to yeast. Examples letters; lower case indicates the use of cloned genes and/or experimental conditions. ND, not determined.

promoter as a result of insertion (Labes and Simon, 1990). Alternatively, integron genes can insert downstream of a host promoter and consequently be cotranscribed with the host gene, as in the case of the sull gene (Stokes and Hall, 1989).

b-p

+

+

involving

natural

genes

and transfer

Transfer

are indicated

in capital

fusion and pore formation, or whether they act as actual canals for the transport of DNA. Although a recent study supports the canal hypothesis (Harrington and Rogerson, 1990) the alternative model is supported by the fact that DNA is not the only type of molecule transferred and that the genetic flux is somethimes bidirectional. Conjugational transfer of DNA is a sophisticated process that requires ~24 transfer genes (fra), organized into three operons, traM, traJ, and fraYZ, that span 33 kb in the plasmid. Fourteen genes (t&-G) are required for the formation and stabilization of cell-cell contact. In addition to the transfer origin oriT, five genes (fraMYD/Z) are required for DNA transfer. Transcriptional regulation is carried out by the products of fin0, finP, and rraJ (Paranchych et al., 1986). traS and traTencode surface exclusion factors that prevent conjugational acquisition of plasmids from the same incompatibility group. (Plasmids that cannot coexist stably in the same cell are grouped into incompatibility groups [lnc] and are thought to be evolutionarily related.) Establishment of cell-cell contact activates replication of the DNA intended for transfer. TraM is thought to be a mating sensor and the signal transducer for the initiation of conjugative DNA synthesis. First, a nucleoprotein complex, the relaxosome, is formed by the sequential binding of TraJ and Tral to oriT. This complex is recognized by TraH, which binds and stabilizes the entire structure (Pansegrau et al., 1990). Replication begins at oUTwith a single-stranded nick created by Tral, a DNA helicase I (Matson and Morton, 1991). The nicked strand begins to uncoil and move into the host cell, led by its Yend. Conjugative replication involves replacement of the transferred strand in the donor cell (by means of a rolling circle-like

$4ew:

Plasmids

and Gene

Flux

Figure

1. Range of PlasmidRelated

Gene Flux

Regions of overlap between major genetically related organisms represent gene exchange. Almost all this network is based on plasmid self-mobilization, although other means of lateral transfer may be involved. Arrows show the proposed origin and dispersion of different antibiotic resistance determinants. Some determinants arose from antibiotic-producing organisms and soil bacteria, others from mutant organisms.exposed to drugs during the antibiotic era.

mechanism) and synthesis of the complementary strand in the host. In both cases, primers are required for DNA synthesis. Several plasmids encode their own primases as well as single-stranded DNA-binding proteins that enter the host cell bound to the transferred DNA. Once transfer and synthesis have occurred, the plasmid DNA must recircularize. The membrane-associated activity of the traYZ products has been proposed to be involved in this process (Willets and Wilkins, 1984). In principle, fra genes acting in trans can mediate transfer of any molecule of DNA containing a compatible oriT. The consequences of this are P-fold. First, integration of a conjugative plasmid (or, experimentally, of an oriT in the presence of a conjugative plasmid) into a chromosome can result in transfer of the complete chromosome (as in Hfr cells). Second, in the presence of a compatible conjugative pldsmid, any plasmid containing an oriTcan beconjugationally transferred. Stable acquisition of chromosomal markers mobilized by Hfr cells is limited by the requirement for homologous recombination after transfer, and thus mobilization of only homologous sequences can be efficient (Smith, 1991). Recombination in vivo can be abolished by as little as 100/b-20% sequence divergence (Shen and Huang, 1986; Rayssiguier et al., 1989). Transmissible plasmids can now be classified in two groups: conjugative plasmids, which encode the complete conjugative apparatus, and mobilizable plasmids, which contain an oriT(and other mob genes) but require a conjugative plasmid to be comobilized. One of the bestdescribed mobilizable plasmids is RSFlOlO (In&t), which contains an oriT and at least three of the genes required for mobilization within a 1.8 kb mob region (in contrast to the 33 kb tra region found in conjugative plasmids) (Derbyshire et al., 1987). Recently, bidirectional transfer of DNA mediated by IncPl and IncQ plasmids has been described, This phenomenon, called retrotransfer, can include the transfer of conjugative and mobilizable plasmids as well as chromosomal markers. The kinetics of retro-

transfer appear to be similar to those of conventional conjugative events (Top et al., 1991). Plasmids also differ in the range of hosts they can colonize. Some plasmids have very narrow host ranges, including only a few closely related species. Others, in contrast, can exist within a large variety of cell types and are therefore referred to as promiscuous or wide host range plasmids. The mechanisms underlying promiscuity are beginning to be understood. In some cases, proteins that have been previously synthesized in the donor cell must be “injected” into the recipient’s cytoplasm to stabilize the new extrachromosomal element. For example, the Sog proteins (DNA primases) responsible for the promiscuity of lncll plasmids initiate synthesis of the complementary strand of transferred DNA in a manner that is independent of host cell primases (Wilkins et al., 1991). Another mechanism underlying promiscuity involves preventing restriction enzyme attack by the host. The ard (alleviation of restriction DNA) gene, for example, encodes antirestriction proteins. Since this gene is one of the first to be transferred during conjugation, it is possible that it accomplishes its protective role by being expressed even before conjugation has been completed (Delver et al., 1991). Conjugation occurs both naturally and in experimental settings between very distantly related bacterial species. For example, plasmids have been constructed that are capable of conjugative transfer from E. coli to Streptococcus, Staphylococcus, Bacillus, Listeria (Trieu-Cuot et al., 1987), Corynebacterium (Schafer et al., 1990), and Streptomyces (Mazodier et al., 1989). An example involving a naturally occurring plasmid is the transfer of the mobilizable E. coli plasmid RSFl 010 to Streptomyces and to Mycobacterium species. In both cases the antibiotic resistance phenotype specified by RSFlOlO was fully expressed (Gormley and Davies, 1991). Significantly less information is available on conjugation between gram-positive bacteria. Studies in Streptococcus have revealed at least two interesting characteristics of

Cell 192

this process: the existence of conjugative transposons and the secretion of sexual pheromones by potential recipient cells. A chromosomal determinant has been described for tetracycline resistance in S. faecalis that can transfer conjugationally in the absence of plasmids. This chromosomal marker can incorporate into plasmids in Ret cells, suggesting the involvement of a transposon, later identified as Tn916. The first step in this kind of mobilization is thought to be the excision and circularization of the transposon, which is then followed by a plasmid-like conjugational transfer (Clewell and Gawron-Burke, 1966). Tn916 and other related transposons seem to be capable of moving into other cells of the same genus (S. mutans, S. pyogenes, S. faecalis) as well as into Lactobacillus, Leuconostoc, and partially Staphylococcus aureus. A new conjugative transposon identified in Lactococcus lactis contains genes encoding a bacteriocin (Horn et al., 1991). Another characteristic of S. faecalis is the production and secretion of pheromones by potential recipient cells. Recipient cells produce at least five hydrophobic peptide pheromones that induce synthesis of an aggregation substance in the donor cell. The aggregation substance is a proteinaceous adhesin that covers the cell surface and facilitates the formation of cellular aggregates in liquid media that facilitate conjugation. The plasmids transferred in this process are specific and cannot be transferred to other species present. Once the plasmid is acquired by the recipient cell, specific pheromone secretion ceases. At this time, production is initiated of a competitive inhibitor of the pheromone, which is encoded by the transferred plasmid. The secretion and action of other pheromones are not affected (Clewell et al., 1966). Anaerobic bacteria are informative for yet other conjugative mechanisms. Bacteroides, for example, carries chromosomal conjugative elements coding for tetracycline, clindamycin, and erythromycin resistance that can be transferred between different Bacteroides species. As previously described for conjugative transposons, these elements are able to excise out from the chromosome, forming plasmid-like structures. Interestingly, both excision and transfer are increased by preincubation in low concentrations of tetracycline (Shoemaker and Salyers, 1968; Stevens et al., 1990). These conjugative elements are also capable of mobilizing coresident plasmids in very distantly related bacteria, such as E. coli. Conjugation between Bacteria and Yeast The horizontal transfer of DNA by conjugation is not restricted to bacteria. The transfer of conjugative plasmids from E. coli to the budding yeast S. cerevisiae has been described (Heinemann and Sprague, 1989). The mechanisms underlying this transfer are strikingly similar to those involved in bacterial conjugation. Both physical and genetic manipulations known to disrupt bacterial conjugation (but not other forms of transfer) were effective in disrupting the bacteria-yeast transfer. For example, bacteria-yeast transfer is dependent on cell-cell contact, similar to conjugation but in contrast to transformation and transduction. Additionally, genetic manipulations revealed a strict requirement of the conjugational oriT, mob, and fra functions

(Heinemann and Sprague, 1991). Interestingly, not only broad host range plasmids (R751 and its derivative pDPT51) but also a modified limited host range plasmid (pFLEU2, a derivative of F) were capable of being transferred from E. coli to S. cerevisiae. Therefore, at least two different bacterial transfer systems (including distinct cellcell contact, transfer, and mob and oriT functions) are capable of mediating prokaryotic-eukaryotic gene transfer. T-DNA: Conjugation between Bacteria and Plants? One of the most striking instances of horizontal transfer of genetic information occurs naturally between individuals of two extremely distant taxonomical groups: bacteria and plants. Transfer of bacterial DNA to plant cells occurs in certain species of Agrobacterium that have the ability to mobilize a DNA segment, T-DNA, from a bacterial plasmid into a susceptible plant’s genome. Agrobacterium is capable of transferring T-DNA into a broad range of dicots in addition to some gymnosperms and even a few monocots (Binns and Thomashow, 1988). Perhaps as remarkable as the taxonomical distance that separates Agrobacteria from plants is the apparent similarity of the transfer to conventional bacterial conjugation (Zambryski et al., 1989). The mechanisms involved in DNA transfer between bacteria and plants are only beginning to be understood (reviewed in Zambryski et al., 1989). As in conjugation, T-DNA transfer requires direct cell-cell contact. Agrobacterium detects and (through chemotaxis) binds to wounded plant cells. A group of Vir proteins encoded by Ti plasmids mediates these processes. VirA is most probably the membrane protein involved in the detection of these chemical signals. The stimulus is then transmitted to VirG, which is directly responsible for activation. Evidence suggests that following attachment, VirB in the bacterial envelope interacts with plant membrane proteins to stabilize this initial surface-surface contact resulting in localized membrane fusion with formation of a pore (Ward et al., 1988). The interactive role of the VirB proteins is analogous to the role proposed for the TraG and TraN proteins of the bacterial F transfer system (Manning et al., 1981). The vir operons of Ti plasmids are subject to positive transcriptional control by the VirA and VirG proteins. VirA and VirG are homologous to CpxA and SfrA, respectively, which control the expression of the WaJ gene in conjugative F plasmids (Miller et al., 1989). It is the process of preparing the T-DNA for transfer that appears to share the strongest similarity with bacterial conjugal DNA transfer. The initial step in T-DNA transfer involves nicking at both ends of the strand of the T-DNA whose 5’end maps to the right T-DNA border. The nicking of T-DNA is analogous to the nicking that occurs at oriT sites during conjugal DNA transfer. More importantly, the nick regions of IncP plasmid origins contain consensus sequences that are identical to those found at the T-DNA borders of Agrobacterium tumefaciens, suggesting a relationship between broad host range bacterial conjugation and T-DNA transfer (Waters et al., 1991). In both conjugation and T-DNA transfer, nicking results in the production of a linear single-stranded donor molecule.

y;giew:

Plasmids

and Gene

Flux

In addition, in both cases, the transferred DNA strand is led by its 5’ end. In the case of T-DNA, this polarity is achieved by a difference in nicking susceptibility at the right and left borders: right borders seem to be more easily nicked than left borders (Peralta et al., 1988). Sequences responsible for this preferred nicking have been found close to the region of the right border nick. In conjugation, it has been suggested that analogous nicking-transfer enhancing sequences may be present near oriT in RK2 and Rp4 plasmids (Waters et al., 1991). Another important similarity between conjugal DNA transfer and T-DNA transfer is the association of the transferred DNA with specific proteins. Like many conjugative plasmids, the Ti plasmid encodes single-stranded DNAbinding proteins that are thought to stabilize and protect newly synthesized T strands. In particular, the VirEP protein seems to coat the length of the T strand (Christie et al., 1988) whereas the VirDP protein binds the 5’ end to target the T-DNA to the plant nucleus (Herrera-Estrella et al., 1990; Howard et al., 1992). An elegant and convincing demonstration of the evolutionary similarity of T-DNA transfer and conjugal DNA transfer has made use of a hybrid transfer system-oriT from the E. coli RSFlOlO plasmid can substitute effectively for T-DNA borders in directing transfer of DNA from Agrobacterium into plant cells (Buchanan-Wollaston et al., 1987). In these experiments, the bacterial mob genes and the vir-encoded proteins required for transfer of T-DNA are capable of working together to mediate transfer. It is interesting to note that vir functions can mediate plasmid transfer not only in this experimental situation but also in nature to transmit the Ti plasmid itself between agrobacteria (Steck and Kado, 1990; Dessaux et al., 1989). The T-DNA transfer system not only illustrates the wide range of action of conjugative or conjugative-like transfer systems but indicates the surprising potential for widespread horizontal gene transfer. Since the T-DNA transfer system is not specific for the transferred DNA-VirE2 binds single-stranded DNA molecules nonspecificallythe variety of transferable DNA sequences appear to be practically unlimited. In addition, the hybrid transfer experiments(Buchanan-Wollaston et al., 1987) indicate, at least in principle, the potential for gene flow from gram-negative bacteria into plants. Finally, the targeting and incorporation of transferred DNA appears to be compatible with higher eukaryotes other than plants. The nuclear targeting VirD2 protein contains a nuclear localization motif that is common to higher eukaryotes (Howard et al., 1992) and T-DNA integration is accomplished by illegitimate recombination, the same process by which foreign DNA is incorporated into animal cells during transfection (Mayerhofer et al., 1991; Gheysen et al., 1991). Nonconjugative Mechanisms of Horizontal Gene Transfer Prokatyotes rely on at least two other mechanisms in addition to conjugation for horizontal gene transfer: transformation and transduction. For several reasons, these mechanisms generally have been considered to be of less evolutionary importance than conjugation. First, their nat-

ural occurrence is restricted to bacterial cells. Furthermore, the genetic pools contributed by these systems are available only to fairly small groups of bacteria. Finally, the conditions required for transformation and transduction are encountered less frequently in nature than those required for conjugation. However, these nonconjugative mechanisms may have a greater evolutionary impact than previously suspected. The mechanisms of intergeneric transfer between bacteria (Mazodier and Davies, 1991) and the potential consequences of these gene fluxes on the dispersal of engineered extrachromosomal elements (Sayre and Miller, 1991) have been reviewed recently. Natural transformation, the uptake and incorporation of exogenous DNA by bacterial cells, has been observed in only ~15 bacterial genera (Stewart and Carlson, 1986). Transformation of chromosomal markers seems to result not in the incorporation of novel genes but rather in local homologous recombination events that produce variations in preexisting genes (e.g., Smith et al., 1991). However, because transformation by plasmids results in the incorporation of the complete replicon, it might play an important role in the mobilization of nonconjugative plasmids under certain circumstances. For example, transformationcompetent organisms may have served as bridges for the transfer of genes between very distant, conjugationally incompatible groups. In addition, transformation may be important in the transfer of nonconjugative plasmids that are common in bacteria that inhabit marine sediment environments (Paul et al., 1991). Plasmids that are released by lysis or secretion during certain growth phases may remain stably adsorbed to mineral surfaces that protect them from degradation (Romanowski et al., 1991). In this manner, the extended extracellular half-life of DNA would enhance the opportunity for transformational uptake. Transduction, the transfer of bacterial genes by phage particles, has two basic advantages over transformation. First, transferred DNA is protected within the phage capsid during extracellular transport. Second, transferred DNA does not depend on the recipient cell’s adsorption mechanisms for its entry. However, bacteriophages in general have a high degree of infective specificity and, therefore, their transfer range is significantly limited. The low frequency and limited range of transduction events may be countered to some extent by the very large size of natural bacteriophage populations-concentrations up to 2.5 x lOa phage particles per ml are present in sea water samples (Bergh et al., 1989). As in transformation, transduction requires homologous recombination for the stable acquisition of chromosomal markers, whereas genes carried by small plasmids do not. In S. aureus, for example, cointegrates are formed by the recombinational activities of the host and phage on small areas of homology (Novick et al., 1984; Dyer et al., 1984). Another example is the E. coli Pl temperate phage. In its prophage stage, Pl exists as a low copy number plasmid instead of being incorporated into the host chromosome. Although it is incapable of replicating lytically or as a plasmid in all the cells it infects, the range of Pl-susceptible organisms is very broad; transduction of Pl-mediated resistance has been observed in 16 gram-negative bacterial

Cell 194

genera (Scott and Froehlich, 1986). It is interesting to note that, like plasmids, prophages can confer selective advantages to their bacterial hosts (e.g., Barondess and Beckwith, 1990). In addition to conjugation, transformation, and transduction, other forms of horizontal gene transfer have been described. For example, prokaryotic elements encoding reverse transcriptase have been found in myxobacteria and in some strains of E. coli. These elements, named retrons, seem to insert into specific chromosomal sites by a mechanism other than transposition (Lim, 1991). Retrons also seem to be involved in intercellular mobilization. The (pR73 prophage, which is capable of being mobilized by the P2 phage, is a retron (in fact, a retronphage) capable of integrating into the gene coding for selenocysteinyl tRNA (se/C; lnouye et al., 1991). The extent of mobility of these elements is currently unknown, but these observations foster several questions. What are the evolutionary consequences of a migratory retroelement? Do retrons exist in other bacteria? Does retrotransposition exist in prokaryotes? What are the requirements and limitations of this process? Even less is known about events involved in horizontal gene transfer in eukaryotes. One example is the apparent mobilization of P element transposons from Drosophila melanogaster to D. willistoni. It has been proposed that the vector mediating this transfer is the mite Proctolaelaps regalis (Houck et al., 1991). Transposon transfer between eukaryotes might even occur between organisms from different genera. The transposable element mariner seems to have been transferred from Drosophila to Zaprionus (Maruyama and Hartl, 1991). Another example is group I introns found in mitochondrial, chloroplast, cyanellar, or nuclear genomes as well as bacteriophages; these have been described as transposable elements. There is evidence suggesting horizontal transfer of these sequences between very distantly related species, such as fungi and algae (Dujon, 1989). An unusual kind of extrachromosomal genetic element found in the spirochete Borrelia burgdorferi (the etiological agent of Lyme disease) provides evidence for an extraordinary event in horizontal transfer. The Borrelia element is a double-stranded linear plasmid. Its covalently closed ends contain telomeric fragments required for replication that are almost 70% identical to those of an African swine virus, which is carried by the same tick vector that carries the spirochete (Hinnebusch and Barbour, 1991). This may be the first report of an actual link between the satellite genomes of bacteria and mammals. Horizontal gene transfer also appears to have occurred within eukaryotic cells-genetic determinants from mitochondria and chloroplasts have been transferred to the nucleus during the course of evolution. Such mobilization, in addition to transfer between mitochondria and chloroplasts, continues to occur (Thorsness and Fox, 1990). The evolutionary impact of such mobilizations is difficult to estimate; perhaps they constitute vestiges of the horizontal transfer abilities of the preendosymbiotic ancestors of presentday organelles. The origin and maintenance of organelle genomes re-

main controversial. Assuming an endosymbiotic origin of chloroplasts and mitochondria, an interesting possibility is that the functions required for such endosymbiosis were originally encoded by plasmids, as those of present-day Rhizobiaceae bacteria (Long, 1989). The ribosomal RNA of plant mitochondria may have a common ancestor with Agrobacteria and Rhizobacteria (Singer and Berg, 1991). The symbiotic association between Bradyrhizobium japonicum and the soy plant seems to have resulted in the transfer of the glutamine synthetase gene from the plant to the bacteria (Carlson and Chelm, 1986). Interestingly, based on some degree of similarity in the hydrophobic profiles of mitochondrially encoded proteins and the plasmid encoded Hok protein-a protein involved in plasmid stabilization-it has been proposed that the mitochondrial genome itself is a plasmid (Jacobs, 1991). Emergence and Evolution to Antibiotics

of Bacterial

Resistance

Antibiotics are secondary metabolites produced largely by fungi and soil bacteria. Because antibiotics were present in several natural environments long before their discovery and mass production by humans, the selection of various antibiotic-resistant organisms could have occurred long ago. However, the presence of a few resistance determinants in strains isolated before the antibiotic era is not sufficient to account for the rate at which resistance mechanisms have appeared and evolved, especially to semisynthetic and synthetic drugs. It appears that horizontal transfer of genetic determinants has played the decisive role in this evolution. The first possible sources of antibiotic resistance genes were apparently the bacterial antibiotic producers themselves. Actinomycetes, especially the Streptomyces genus, are the producers of almost all clinically used aminoglycosides. These bacteria, however, are intrinsically susceptible to their own metabolites and thus must carry resistancedeterminants(Cundliffe, 1984). Itwasproposed that the evolutionary origin of resistance genes in R plasmids could be traced to the Streptomyces producers, since the enzymatic mechanisms for aminoglycoside inactivation are similar (Benveniste and Davies, 1973). Further studies based on DNA sequence data support this proposal by revealing significant homology between resistance determinants in E. coli plasmids and the gene that encodes 6’-aminoglycoside-phosphotransferase [APH(G’)] in Streptomyces fradiae (Foster, 1983). The Streptomyces genus has physiological and genetic characteristics that are very different from those of most higher organism pathogens. S. coelicolor A3(2), for example, carries a transposon that is occasionally found as a minicircle (Henderson et al., 1989) and may play an important role in the genetic mobility that characterizes this species. These mobile elements might have been involved in the transfer of genes encoding resistance to aminoglycosides and macrolides from producer to nonproducer strains. However, it is important to realize that the regulation of resistance gene expression in Streptomyces is particularly complex and very different from that of resistance

T;giew:

Plasmids

and Gene

Flux

genes in other eubacteria (Bibb and Janssen, 1986). More than simple horizontal gene transfer is required to explain the presence of resistance determinants in bacteria that don’t produce antibiotics. The ability of certain ISs to act as promoters of neighboring genes could offer the necessary component required for the expression of Streptomyces genes in very diverse cellular backgrounds. Another group of organisms that have evolved in continuous contact with antibiotics are the soil bacteria (e.g., Bacillus) other than Actinomycetes. Small populations of Bacillus may have evolved to a resistant phenotype under the selective pressure of antibiotics released into the soil by Streptomyces. Moreover, both Bacillus and Streptomyces species attain natural competence during certain phases of their growth (Stewart and Carlson, 1986) which in some cases involves an induction of DNA secretion into the environment (Lorenz et al., 1991). This suggests the dispersal of Streptomyces resistance genes with Bacillus as an intermediary host. Resistance determinants from Streptomyces could have been obtained by Bacillus via transformation and then passed on to a variety of other bacteria via conjugation (Figure 1). Consistent with this possibility, divergence of the APH(3’) gene from a common ancestor is indicated by amino acid sequence comparisons of APH(3’) encoded by transposons of gram-negative bacteria, Streptococcus, Staphylococcus, Bacillus, and that of the neomycin producer S. fradiae (Trieu-Cuot and Courvalin, 1986). The source of resistance to synthetic or semisynthetic drugs and antibiotics produced by fungi (i.e., fl-lactamics) clearly must be different from resistance to antibiotics produced by bacteria. Two plasmidsncoded properties found in strains of the Murray collection (1917-1954) mutagenic DNA repair (Sedgwick et al., 1989) and conjugative transfer (Hughes and Datta, 1983) permit the elaboration of a model for the appearance and distribution of these antibiotic resistance genes. Mutagenic repair could initially have generated a variety of cells resistant to antimicrobial drugs. (For example, mutations in penicillin-binding proteins or dihydropteroate synthetase could have conferred resistance to f%lactamics or sulfonamides, respectively.) This mutagenic repair system may play the same role as the inducible hypermutator states proposed to be a source of adaptive mutations during starvation (Hall, 1991). For example, the appearance of a plasmid-related resistance to nalidixic acid in Shigella dysenteriae strains seems to be the result of the action of mutagenic plasmids. The plasmid involved initially conferred only minimal tolerance to the drug, but the bacteria carrying it showed a lOOO-fold increase in the frequency of mutation to the Nal’ phenotype after two exposures to nalidixic acid (Ashraf et al., 1991). A possible second step in this process involves the intermolecular transfer of these new determinants, possibly through transposition, to a conjugative plasmid. Finally, the lateral mobilization of these plasmids among other bacterial types would permit the dissemination of the resistance determinants as well as the evolution of the replicons by acquisition of new genes from different cellular backgrounds. In this manner, although plasmids of the preanti-

biotic era seem generally to have lacked resistance genes, the determinants required for their rapid appearance and diffusion were present well before the massive introduction of antimicrobial drugs. Mobilization of Resistance Genes between Replicons The possible origins of resistance genes, their location on plasmids, and their intergeneric mobilization by means of conjugation have been discussed. The accumulation of several resistance genes in a single plasmid, multiresistance (MR), is a still more striking consequence of lateral mobility. MR plasmids can carry resistance determinants that are homologous to determinants found in evolutionarily distant or unrelated plasmids. In addition, resistance determinants on MR plasmids can be associated with other genetic determinants, such as virulence. Attempts to explain MR as an accumulation of individual mutations fail, owing to the low probability of multiple events, which is not compatible with the high frequency of appearance of these phenotypes. Some important exceptions to this notion exist-the mutationally acquired MR specified either by the marA locus or by SOXQ, as well as the adaptive response to the superoxide radical governed by the soxRS system (reviewed in Demple and AmabileCuevas, 1991), and the adaptive response to chemotactic repellants, such as salicylates (Rosner et al., 1991). In the first three cases and possibly the fourth, outer membrane permeability is decreased and, consequently, resistance to xenobiotics, including antibiotics, is increased. All other MR phenotypes are encoded by plasmids. How have MR determinants accumulated in single replicons? MR plasmids have evolved as collections of transposons (Figure 2). TnlO, the best understood of the class I resistance transposons, codes for tetracycline resistance and contains two inverted IS10 sequences at its ends (Kleckner, 1989). A well-characterized member of the class II group is Tn21, which contains resistance genes to sulfonamides (sull) and to streptomycin and spectinomycin (aadA). Tn21 is very broadly distributed. The majority of resistance transposons identified in gram-negative bacteria are related to Tn21, and many MR plasmids contain resistance determinants flanked by fragments of the Tn21 integron (reviewed in Grinsted et al., 1990). It is clear that this system of intermolecular transfer has been highly successful, with homologous transposons found in even very distant genera, such as Mycobacterium (Martin et al., 1990). A large group of Tn21-related transposons differ only in the aadA resistance gene or in the genes adjacent to aadA. As previously discussed, it seems as though resistance genes have moved as integron cassettes within these mobile elements. That the ancestral integron included the sull gene is suggested by its consistent presence in a common region of more than ten plasmids and transposons (including Tn21) that contain integrons or integron fragments (Hall et al., 1991). Dispersal of Bacterial Resistance to Antibiotics The dispersal of bacterial resistance to antibiotics has been an extraordinarily rapid phenomenon: antibiotic re-

Cell 196

Figure Effect

2. The

“Matrioshka”

or Russian

Doll

MR plasmids act as collections of transposons. Some transposons. such as the Tn21 group, bear an element called integron that allows for the integration of gene cassettes. These cassettes may be exchanged between different DNA molecules.

sistance has been selected for and catalyzed by the massive introduction of antibiotics that began only slightly over 50 years ago. It seems that resistance to the first antibiotics used in the antibiotic era has reached an equilibrium, while the evolution and distribution of resistance to more recently introduced compounds are dynamic processes that can be followed within hospital environments. An example can be found in the different resistance genes for aminoglycosides. aadA and aphA (streptomycin’) are broadly distributed, the first in association with Tn21 -like elements, and the second in association with small, nonconjugative ubiquitous plasmids. aacC and aadS (gentamicin’), on the other hand, are present in epidemic plasmids, in a regional manner, and only after several years of antibiotic usage. Finally, aacA (amikacirr) is just beginning to be observed in plasmids. It constitutes a relatively minor fraction of the population resistant to other antibiotics and is apparently present in only a limited range of plasmids (Hopkins et al., 1991). That bacterial resistance to antibiotics is located in plasmids was discovered in Japan in 1959 as a consequence of MR transfer from Shigella to E. coli during an infection. Two important conclusions derive from this initial observation: transfer occurs in vivo (and not only in the laboratory), and bidirectional transfer occurs between both pathogenic and nonpathogenic organisms. This bidirectional transfer results in conversion of nonpathogens into resistance gene pools available to pathogens during colonization of the infected organism. E. coli, Hemophilus parainfluenzae, and Staphylococcus epidermidis are examples of such reservoirs (Levy, 1986). Another consequence of horizontal gene mobilization is the observed association of resistance genes with other

clinically or environmentally relevant determinants. For example, a conjugative plasmid that specifies resistance to tetracycline, streptomycin, and sulfonamides, in addition to two enterotoxins, has been isolated from E. coli (Gyles et al., 1977). Other virulence factors have been found in enteropathogenic E. coli associated with MR plasmids (Reynaud et al., 1991). Metal ion resistance determinants are also frequently associated with antibiotic resistance genes. Resistance to mercury and organomercurial compounds is of both clinical importance, owing to their use as components of dental fillings and disinfectants, and environmental relevance in areas contaminated by mercuric ions. These resistances are carried by the Tn21 transposon and many groups of penicillinase plasmids in S. aureus. Frequently, these plasmids also carry arsenate and tellurite resistance genes. In addition, some conjugation determinants, such as WaT, are themselves related to pathogenicity (Sukupolvi and O’Connor, 1990). Most horizontal transfer of bacterial resistance to antibiotics seems to be the result of conjugation. However, other transfer mechanisms may develop among bacteria resident in multispecific colonies called biofilms (Costerton et al., 1987). These aggregations, the clinical importance of which is now becoming apparent, may constitute ideal environments for conjugation events that do not occur in laboratory conditions as well as for transformation and transduction, which were previously considered to be of only minimal relevance in vivo. Solutions to an Undesirable Consequence of Gene Flux The rapid evolution of bacterial resistance to antibiotics

$iew:

Plasmids

and Gene

Flux

poses a serious challenge to medicine and pharmacology. The solutions considered so far have been a search for new antimicrobial agents and the temporary suppression in the use of certain antibiotics, especially in hospitals. The first solution is relevant only for the short term. Bacteria usually develop resistance faster than scientists can discover and produce new antibiotics. The rationale for the second solution is based on the alleged disadvantage conferred by plasmids when there is no selective pressure, which should lead to the eventual preponderance of cells that have spontaneously lost their plasmids. However, there are at least three facts that make this strategy ineffective. First, coevolution of the plasmid-bacteria association has resulted in the greater fitness of plasmid-containing bacteria as compared with those lacking plasmids, even in the absence of selective pressure by antibiotics (Bouma and Lenski, 1988). Second, several kinds of plasmids code for a system that kills plasmid-free segregants, thus guaranteeing the replicon’s stability in the bacterial population (Gerdes et al., 1988). Finally, for an antibiotic suppression strategy to work, the suspension must include all antibiotics for which resistances are encoded by a given plasmid; if the use of a single one of these antibiotics is sustained, it will act as a selective pressure for the conservation of the complete MR plasmid. For example, the prevalence of sulfonamide resistance, despite the significant reduction in its use, can been attributed to the association of the sull gene with other resistance genes in Tn21-derived transposons. An alternative approach is the elimination or inhibition of the plasmids themselves. A variety of compounds are capable of curing or interfering with the expression of bacterial plasmids(Hahn, 1978; AmBbile-Cuevas et al., 19%). Other agents can prevent conjugative transfer (Viljanen and Boratynski, 1991). Although a hyperstable plasmid could eventually be selected as a consequence of the use of a curing agent, the selection of these varieties becomes less probable because nonvital bacterial structures are attacked. (The rapid selection of antibiotic resistant strains has occurred because the massive release of these drugs exerts a constant lethal pressure on bacterial populations in a variety of different environments.) In addition, if curing agents with potential clinical use are discovered, virulence plasmids as well as MR plasmids could be targeted. Considering the adaptive advantages gained by the acquisition and maintenance of plasmids, the ability to control the establishment of these associations in the microbial world is of particular importance. Concluding

Remarks

Based on the ubiquity and the apparent parallel nature of the evolution of horizontal transfer mechanisms, it can be concluded that the reception and transfer of genes represent a significant evolutionary advantage, particularly for prokaryotes. At least some eukaryotic cells are also involved in this genetic flux through mechanisms that share significant similarities with bacterial conjugation. Indeed, the autotransmissible plasmid appears to be one of the most versatile and ubiquitous mobilization systems known.

Satellite elements, as opposed to chromosomes, seem to be privileged locations for horizontal mobilization of genes by conjugation, transformation, or transduction. The ease with which plasmids acquire mobile genetic elements makes any gene contained in such an element a potential candidate for horizontal transfer. The evolutionary impact of interkingdom genetic transfer is difficult to assess. However, judging from the extent to which antibiotic resistance genes have evolved and spread during only a few decades, it seems likely that, in thousands of years of evolution, horizontal transfer has been a cornerstone in the organization of all prokaryotic genomes. Horizontal gene transfer may have been even more extensive in early evolutionary times. An interesting possibility is that a significant number of modern genes are the result of early horizontal transfer. The involvement of eukaryotic cells in gene transfers thought to be exclusively prokaryotic implies that gene flux has had an impact on the evolution of all living beings. It is not a process restricted to prokaryotic ancestors of the distant past but a currently active mechanism. Further studies should reveal whether parallel mechanisms of horizontal transfer exist in eukaryotes, and whether bacterial conjugative plasmids will continue to play a leading role in the genetic flux between organisms. Acknowledgments The authors wish to especially thank Bruce Demple and Maria Luisa Ruiz for many helpful comments and suggestions. Spencer B. Farr, Dindial Ramotar, Mark Boothby, Huntington Potter, Kenneth Kosik, and Bruce Peters are also thanked for their suggestions. M. E. C. is a Howard Hughes predoctoral fellow. References AmBbile-Cuevas, C. F.. Pifia-Zentella, R. M., and Wah-Laborde, M. E. (1991). Decreased resistance to antibiotics and plasmid loss in plasmid-carrying strains of S~ephylococcus aureus treated with ascorbic acid. Mutat. Res. 264, 119-125. Ashraf, M. M., Ahmed, 2. U., and Sack, D.A. (1991). Unusual association of a plasmid with nalidixic acid resistance in an epidemic strain of ShigeIa dysenteriae type 1 from Asia. Can. J. Microbial. 37, 59-63. Barondess, J. J.. and Beckwith, J. (1990). A bacterial virulence determinant encoded by lysogenic coliphage h. Nature 346, 871-874. Benveniste, R., and Davies, J. (1973). Aminoglycoside antibiotic-inactivating enzymes in actinomycetes similar to those present in clinical isolates of antibiotic-resistant bacteria. Proc. Natl. Acad. Sci. USA 70, 2276-2280. Bergh, a., Borsheim, K. Y., Bratbak, G., and Heldal, M. (1989). High abundance of viruses found in aquatic environments. Nature 340,487468. Bibb, M. J., and Janssen, G. R. (1986). Unusual features of transcrip tion and translation of antibiotic resistance genes in antibiotic-producing Streptomyces. In Antibiotic Resistance Genes: Ecology, Transfer, and Expression, S. B. Levy and R. P. Novick, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 323334. Binns, A. N., andThomashow, M. F. (1988). Cell biology of Agrobacterium infection and transformation of plants. Annu. Rev. Microbial. 42, 575-606. Bouma, J. E., and Lenski. R. E. (1988). association. Nature 335, 351352.

Evolution

of a bacteria/plasmid

Buchanan-Wollaston, V., Passiatore. J. E., and Cannon, mob and oriT functions of a bacterial plasmid promote plants. Nature 326, 172-175.

F. (1987). The its transfer to

Cdl 198

Carlson, T. A., and Chelm, E. K. (1986). Apparent eukaryotic origin of glutaminesynthetase II from the bacterium8radyrhizobiumjaponicum. Nature 322, 566-570.

Heinemann. J. A., and Sprague, G. F.. Jr. (1991). plasmid DNA to yeast by conjugation with bacteria. 194, 187-195.

Christie, P. J., Ward, J. E., Winans, S. C., and Nester, E. W. (1988). The Agrobactarium tumefaciens virE2 gene product is a singlestranded DNA-binding protein that associates with T-DNA. J. Bacteriol. 7 70, 2659-2667.

Henderson, D. J.. Lydiate, D. J., and Hopwood, D.A. (1969). Structural and functional analysis of the mini-circle, a transposable element of Streptomyces coelicolor A3(2). Mol. Microbial. 3, 1307-1318.

Clewell, D. B., Ehrenfeld, E. E., Kessler, Ft. E.. Ike, Y., Franke, A. E., Madion, M., Shaw, J. H., Wirth. R., An, F., Mori, M., Kitada, C., Fujino, M.. and Suzuki, A. (1986). Sex-pheromone systems in Streptococcus faecalis. In Antibiotic Resistance Genes: Ecology, Transfer, and Expression, S. 6. Levy and R. P. Novick, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 131-142. Clewell, D. E., and Gawron-Burke, C. (1966). Conjugative transposons and the dissemination of antibiotic resistance in streptococci. Annu. Rev. Microbial. 40, 635-659. Costerton, J. W., Cheng, K. J., Gesey, G. G.. Ladd, T. I., Nickel, J. C., Dasgupta, M.. and Marrie, T. J. (1987). Bacterial biofilms in nature and disease. Annu. Rev. Microbial. 47, 435-464. Cundliffe, E. (1984). Self defence Brit. Med. Bull. 40. 61-67.

in antibiotic-producing

organisms.

Transmission of Meth. Enzymol.

Herrera-Estrella, A., Van Montagu, M., and Wang, K. (1990). A bacterial peptide acting as a plant nuclear targeting signal: the aminoterminal portion of Agrobacterium VirDP protein directs a j%galactosidase fusion protein into tobacco nuclei. Proc. Natl. Acad. Sci. USA 87, 9534-9537. Hinnebusch, J., and Barbour. A. G. (1991). Linear plasmids of Borrelia burgdorferi have a telomeric structure and sequence similar to those of a eukaryotic virus. J. Bacterial. 173, 7233-7239. Hopkins, J. D., Flores, A.. Pla, M. P., Lester, S., and O’Brien, T. F. (1991). Nosocomial spread of an amikacin resistance gene on both a mobilized, nonconjugative plasmid and a conjugative plasmid. Antimicrab. Agents Chemother. 35, 1605-1611. Horn, N., Swindell, S., Dodd, H., and Gasson, M. (1991). Nisin biosynthesis genes are encoded by a novel conjugative transposon. Mol. Gen. Genet. 228, 129-135.

Delver, E. P., Kotova, V. U., Zavilgelsky, G. B., and Belogurov, A. A. (1991). Nucleotide sequence of the gene (ard) encoding the antirestriction protein of plasmid Collb-P9. J. Bacterial. 173, 5887-5892.

Houck, M. A., Clark, J. B., Peterson, K. R., and Kidwell. M. G. (1991). Possible horizontal transfer of Drcsophia genes by the mite Proctolae laps regalis. Science 253, 1125-l 129.

Demple. B., and Am&bile-Cuevas, trol of oxidative stress response.

Howard, E. A., &pan, J. R., Citovsky, V., and Zambryski, P. C. (1992). The VirD2 protein of A. tumefaciens contains a C-terminal bipartite nuclear localization signal: implications for nuclear uptake of DNA in plant cells. Cell 68, 109-I 18.

C. F. (1991). Redox Cell 67, 837-839.

redux:

the con-

Derbyshire, K. M., Hatfull, G., and Willets, N. (1987). Mobilization of the non-conjugative plasmid RSFlOlO: a genetic and DNA sequence analysis of the mobilization region. Mol. Gen. Genet. 206, 161-166. Dessaux, Y.. Petit, A., Ellis, J. G., Legrain, C., Demarez, J.-M.. Popoff, M., and Tempe, J. (1989). Ti plasmid-controlled some transfer in Agrobacterium tumefaciens. J. Bacterial. 6366. Dujon, B. (1969). Group I introns and mechanistic speculations-a

M., Wiame, chromo177,6363-

as mobile genetic elements: review. Gene 82, 91-114.

facts

Dyer, D. W., Rock, M. I., and landolo, J. J. (1984). Autogenous transduction of ~11 de in Staphpcoccus aureus: transfer and genetic properties. J. Bacterial. 158, 669-696. Foster, T. J. (1983). Plasmid-mediated drugs and toxic metal ions in bacteria.

resistance to antimicrobial Microbial. Rev. 47, 361-409.

Gerdes, K., Rasmussen, P. B., and Molin, S. (1986). Unique type of plasmid maintenance function: postsegregational killing of plasmidfree cells. Proc. Natl. Acad. Sci. USA 83, 31163120. Gheysen, G., Villarroel. R., and Van Montagu, M. (1991). Illegitimate recombination in plants: a model for T-DNA integration. Genes Dev. 5, 287-297. Gormley, E. P., and Davies, J. (1991). Transfer of plasmid RSFlOlO by conjugation from Escherichia co/i to Streptomyces ffvidans and Mycobacterium smagmatis. J. Bacterial. 7 73, 6705-6708. Grinsted, J., de IaCruz, of bacterial transposable

F., and Schmitt, R. (1990). TheTn27 elements. Plasmid 24, 163-189.

subgroup

Gyles, C. L., Palchaudhuri, S., and Maas, W. K. (1977). Naturally occurring plasmid carrying genes for enterotoxin production and drug resistance. Science 198, 196-199. Hahn, F. E. (1976). Experimental Chemother. 20, 196-226. Hall, 8. G. (1991). Adaptive evolution ous mutations: mutations involving Acad. Sci. USA 88,5882-5886.

elimination

of R factors.

Antibiot.

that requires multiple spontanebase substitutions. Proc. Natl.

Hall, R. M., Brookes, D. E., and Stokes, H. W. (1991). Site-specific insertion of genes into integrons: role of the 59base element and determination of the recombination crossover point. Mol. Microbial. 5, 1941-1959. Harrington, L. C., and Rogerson, A. C. (1990). The F pilus of Escherichia co/i appears to support stable DNA transfer in the absence of wall-to-wall contact between cells. J. Bacterial. 172, 7263-7264. Heinemann, J. A., and Sprague, G. F., Jr. (1989). Bacterial conjugative plasmids mobilize DNA transfer between bacteria and yeast. Nature 340.205-209.

Hughes, V. M., and Datta, N. (1983). Conjugative of the “pre-antibiotic” era. Nature 302, 725-726.

plasmids

in bacteria

Inouye, S., Sunshine, M. G., Six, E. W., and Inouye, M. (1991). Retronphage 1pR73: an E. co/i phage that contains a retroelement and integrates into a tRNA gene. Science 252, 969-971. Jacobs, H. T. (1991). Structural similarities between a mitochondrially encoded polypeptide and a family of prokaryotic respiratory toxins involved in plasmid maintenance suggest a novel mechanism for the evolutionary maintenance of mitochondrial DNA. J. Mol. Evol. 32,333339. Kleckner, N. (1969). Transposon M. M. Howe, eds. (Washington, ogy), pp. 163-184.

Tn70. In Mobile DNA, D. E. Berg and DC: American Society for Microbiol-

Labes. G., and Simon, R. (1990). Isolation of DNA insertion elements from Rhizobium mefiloti which are able to promote transcription of adjacent genes. Plasmid 24, 235-239. Levy, S. B. (1986). Ecology of antibiotic resistance determinants. In Antibiotic Resistance Genes: Ecology, Transfer, and Expression, S. 8. Levy and R. P. Novick, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 17-30. Lim, D. (1991). Structure of two retrons of Escherichia co/i and their common chromosomal insertion site. Mol. Microbial. 5, 1863-1872. Long, S. R. (1969). Rhizobium-legume underground. Cell 56, 203-214.

nodulation:

life together

in the

Lopez, J.. Delgado, D., Andreis, I., Ortiz, J. M.. and Rodriguez. J. C. (1991). Isolation and evolutionary analysis of a RepFVlB replicon of the plasmid pSU212. J. Gen. Microbial. 737, 1093-1099. Lorenz, M. G., Gerjets, D., and Wackernagel, W. (1991). Release of transforming plasmid and chromosomal DNA from two cultured soil bacteria. Arch. Microbial. 156, 319-326. Manning, P. A., Morelli, G., and Achtman, M. (1961). TraG protein of the F sex factor of Escherichia co/i K-12 and its role in conjugation. Proc. Natl. Acad. Sci. USA 78, 7487-7491. Martin, C., Timm, J., Rauzier, quel, B. (1990). Transposition teria. Nature 345, 739-743. Maruyama, onus.

Matson,

J., Gomez-Lus, of an antibiotic

K.. and Ham, D. L. (1991).

Evidence

R., Davies, J., and Gicresistance in mycobacfor interspecific

trans-

J. Mol. Evol. 33, 514-524. S. W.. and Morton,

8. S. (1991).

Escherichia

co/i DNA helicase

ys$iew:

Plasmids

and Gene

Flux

I catalyzes asite- and strand-specific nicking oriT. J. Biol. Chem. 266, 16232-16237.

reaction

at the F plasmid

Mayerhofer, R., Koncz-Kalman, Z., Nawrath. C., Bakkeren, G., Crameri, A., Angelis, K., Redei, G. P., Schell, J., Hohn, B., and Koncz, C. (1991). T-DNA integration: a mode of illegitimate recombination in plants. EMBO J. 10, 697-704. Mazodier, P., and Davies, J. (1991). related bacteria. Annu. Rev. Genet.

Gene transfer 25, 147-171.

Mazodier, P., Petter, Ft., and Thompson, C. (1989). gation between Escherichia co/i and Streptomyces riol. 171, 3593-3595.

between

distantly

lntergenericconjuspecies. J. Bacte-

Miller, J. F., Mekalanos, J. J., and Falkow, S. (1989). Coordinate regulation and sensory transduction in the control of bacterial virulence. Science 243,916-922. Novick, R. P., Projan, S. J., Rosenblum, W., and Edelman, I. (1984). Staphylococcal plasmid cointegrates are formed by host- and phagemediated general ret systems that act on short regions of homology. Mol. Gen. Genet. 195, 374-377. Pansegrau, W., Balzer, D., Kruft, V., Lurz, R., and Lanka, E. (1990). In vitro assembly of relaxosomes at the transfer origin of plasmid RP4. Proc. Natl. Acad. Sci. USA 67, 6555-6559. Paranchych, W., Finlay, B. B.. and Frost, L. S. (1986). Studies on the regulation of IncF plasmid transfer operon expression. In Antibiotic Resistance Genes: Ecology, Transfer, and Expression, S. B. Levy and R. P. Novick, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 117-129. Paul, J. H., Frischer, M. E., and Thurmond, J. M. (1991). Gene transfer in marine water column and sediment microcosms by natural plasmid transformation. Appl. Environ. Microbial. 57, 1509-1515. Peralta, E. G., Hellmis, R., and Ream, R. (1966). overdrive, transmission enhancer on the A. tumefaciens tumor-inducing EMBO J. 5, 1137-1142.

a T-DNA plasmid.

Singer, M., and Berg, P. (1991). Genes California: University Science Books). Smith, G. R. (1991). Conjugational mechanisms. Cell 64, 19-27.

recombination

Smith, J. M., Dowson, C. G., and Spratt, in bacteria. Nature 349, 29-31. Sonea, S. (1991). Bacterial as a Source of Evolutionary (Cambridge, Massachusetts:

and Genomes

(Mill Valley,

in E. coli: myths and

B. G. (1991).

Localized

sex

evolution without speciation. In Symbiosis Innovation, L. Margulis and R. Fester, eds. MIT Press), pp. 95-105.

Steck, T. R., and Kado, C. I. (1990). Virulence genes promote tive transfer of the Ti plasmid between Agrobacterium strains. riol. 772, 2191-2193.

conjugaJ. Bacte-

Stevens, A. M., Shoemaker, N. B., and Salyers, A. A. (1990). The region of a Bacteroides conjugal chromosomal tetracycline resistance element which is responsible for production of plasmidlike forms from unlinked chromosomal DNA might also be involved in transfer of the element. J. Bacterial. 772, 4271-4279. Stewart, G. J., and Carlson, C. A. (1986). formation. Microbial. Rev. 40, 21 l-235.

The biology

of natural

trans-

Stokes, H. W., and Hall, R. M. (1989). A novel family of potentially mobile DNA elements encoding site-specific gene-integration functions: integrons. Mol. Microbial. 3, 1669-1683. Sukupolvi, S., and O’Connor, D. (1990). TraT lipoprotein, a plasmidspecified mediator of interactions between gram-negative bacteriaand their environment. Microbial. Rev. 54, 331-341. Thorsness, P. E., and Fox, T. D. (1990). dria to the nucleus in Saccharomyces 379.

Escapeof DNAfrom cerevisiae. Nature

mitochon346, 376-

Top, E., Mergeay, M., and Springael, D. (1991). Retrotransfer: a specific trait of certain broad-host-range plasmids (which allows the host to pick up new genes from other species and genera). Plasmid 25, 238.

Projan, S. J., and Moghazeh, S. L. (1991). Termination of replication and its potential role in cassette assembly of the “single-stranded” plasmids of gram-positive bacteria. Plasmid 25, 241.

Trieu-Cuot, P., and Courvalin, P. (1986). Evolution and transfer of aminoglycoside resistance genes under natural conditions. J. Antimicrab. Chemother. (Suppl. C) 78, 93-102.

Rayssiguier, C., Thaler, D. S., and Radman, M. (1989). The barrier to recombination between Escherichia co/i and Salmonella typhimurium is disrupted in mismatch-repair mutants. Nature 342, 396-401.

Trieu-Cuot, P., Carlier, C., Martin, P., and Courvalin, P. (1987). Plasmid transfer by conjugation from Escherichia co/i to gram-positive bacteria. FEMS Microbial. Lett. 46. 289-294.

Reynaud, A., Federighi, M., Licois, D., Guillot, J.-F., and Joly, (1991). R plasmid in Escherichia co/i 0103 coding for colonization the rabbit intestinal tract. Infect. Immun. 59. 1888-1892.

Viljanen, P., and Boratynski, J. (1991). Thesusceptibility of conjugative resistance transfer in gram-negative bacteria to physicochemical and biochemical agents. FEMS Microbial. Rev. 86, 43-54.

8. of

Romanowski, G., Lorenz, M. G., and Wackernagel, W. (1991)Adsorp tion of plasmid DNA to mineral surfaces and protection against DNase I. Appl. Environ. Microbial. 57, 1057-1061. Rosner, J. L., Chai. T.J., and Foulds, J. (1991). Regulation of OmpF porin expression by salicylate in Escherichia co/i. J. Bacterial. 173, 5631-5638. Sayre, P., and Miller, R. V. (1991). Bacterial importance in assessing the environmental neered sequences. Plasmid 26. 151-171.

mobile genetic elements: fate of genetically engi-

Schafer, A., Kalinowski, J., Simon, R., Seep-Feldhaus, A.-H., and Piihler, A. (1990). High-frequency conjugal plasmid transfer from gram-negative Eschericbia co/i to various gram-positive coryneform bacteria. J. Bacterial. 172, 1683-1886. Scott, J. R., and Froehlich, 8. J. (1986). Regulation of replication of the plasmid prophage PI. In Antibiotic Resistance Genes: Ecology, Transfer, and Expression, S. 8. Levy and R. P. Novick, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory Press), pp. 153-164. Sedgwick, S. G., Thomas, S. M., Hughes, V. M., Lodwick, Strike, P. (1989). Mutagenic DNA repair genes on plasmids “pre-antibiotic era.” Mol. Gen. Genet. 276, 323-329. Shen, P.. and Huang, H. V. (1986). Homologous Escherichia co/i: dependence on substrate length netics 112, 441-457.

D., and from the

recombination and homology.

in Ge-

Shoemaker, N. 8.. and Salyers, A. A. (1988). Tetracycline-dependent appearance of plasmidlike forms in Bectemides uniformis 9961 mediated by conjugal Bactem/des tetracycline resistance elements. J. Bacteriol. 770, 1851-1657.

Ward, J. E., Akiyoshi, D. E., Regier, D., Datta, A., Gordon, M. P.. and Nester, E. W. (1988). Characterization of the virS operon from an Agrobacterium tumefaciens Ti plasmid. J. Biol. Chem. 263, 58&t5814. Waters, V. L., Hirata, K. H., Pansegrau, W., Lanka, E., and Guiney, D. G. (1991). Sequence identity in the nick regions of IncP plasmid transfer origins and T-DNA borders of Agrobacterium Ti plasmids. Proc. Natl. Acad. Sci. USA 88, 1456-1460. Wilkins, 8. M., Rees, C. E., Thomas, A. T., and Read, T. D. (1991). Conjugative events in the recipient cell affecting plasmid promiscuity. Plasmid 25, 227. Willets, N., and Wilkins, B. (1984). Processing of plasmid bacterial conjugation. Microbial. Rev. 48, 24-41.

DNA during

Zambryski, P., Tempe, J., and Schell, J. (1989). Transfer and function of T-DNA genes from Agrobacterium Ti and Ri plasmids in plants. Cell 56, 193-201.