Replication and Control of Circular Bacterial Plasmids

Gloria del Solar; Rafael Giraldo; María Jesús Ruiz-Echevarría; Manuel Espinosa; Ramón Díaz-Orejas

doi:10.1128/mmbr.62.2.434-464.1998

. 1998 Jun;62(2):434–464. doi: 10.1128/mmbr.62.2.434-464.1998

Replication and Control of Circular Bacterial Plasmids

Gloria del Solar ¹, Rafael Giraldo ¹, María Jesús Ruiz-Echevarría ¹, Manuel Espinosa ¹, Ramón Díaz-Orejas ^1,^*

PMCID: PMC98921 PMID: 9618448

Abstract

An essential feature of bacterial plasmids is their ability to replicate as autonomous genetic elements in a controlled way within the host. Therefore, they can be used to explore the mechanisms involved in DNA replication and to analyze the different strategies that couple DNA replication to other critical events in the cell cycle. In this review, we focus on replication and its control in circular plasmids. Plasmid replication can be conveniently divided into three stages: initiation, elongation, and termination. The inability of DNA polymerases to initiate de novo replication makes necessary the independent generation of a primer. This is solved, in circular plasmids, by two main strategies: (i) opening of the strands followed by RNA priming (theta and strand displacement replication) or (ii) cleavage of one of the DNA strands to generate a 3′-OH end (rolling-circle replication). Initiation is catalyzed most frequently by one or a few plasmid-encoded initiation proteins that recognize plasmid-specific DNA sequences and determine the point from which replication starts (the origin of replication). In some cases, these proteins also participate directly in the generation of the primer. These initiators can also play the role of pilot proteins that guide the assembly of the host replisome at the plasmid origin. Elongation of plasmid replication is carried out basically by DNA polymerase III holoenzyme (and, in some cases, by DNA polymerase I at an early stage), with the participation of other host proteins that form the replisome. Termination of replication has specific requirements and implications for reinitiation, studies of which have started. The initiation stage plays an additional role: it is the stage at which mechanisms controlling replication operate. The objective of this control is to maintain a fixed concentration of plasmid molecules in a growing bacterial population (duplication of the plasmid pool paced with duplication of the bacterial population). The molecules involved directly in this control can be (i) RNA (antisense RNA), (ii) DNA sequences (iterons), or (iii) antisense RNA and proteins acting in concert. The control elements maintain an average frequency of one plasmid replication per plasmid copy per cell cycle and can “sense” and correct deviations from this average. Most of the current knowledge on plasmid replication and its control is based on the results of analyses performed with pure cultures under steady-state growth conditions. This knowledge sets important parameters needed to understand the maintenance of these genetic elements in mixed populations and under environmental conditions.

Plasmids are extrachromosomal DNA elements with characteristic copy numbers within the host. These replicons have been found in species from the three representatives of the living world, namely, the domains Archaea, Bacteria, and Eukarya (318). Plasmids may constitute a substantial amount of the total genetic content of an organism, representing more than 25% of the genetic material of the cell in some members of the Archaea (127, 331). They can incorporate and deliver genes by recombination or transposition, thus favoring genetic exchanges in bacterial populations. Since plasmids can be introduced into new hosts by a variety of mechanisms, they can be considered to be a pool of extrachromosomal DNA which is shared among populations. The wealth of genetic information carried by plasmids, their impact in the microbial communities, and the potential of these elements to act as natural cloning vectors have stimulated research into plasmids not only from the fundamental but also from the clinical, biotechnological, and environmental points of view. Three main factors have contributed to the development of plasmid research: (i) the genetic organization of these elements is apparently simple, (ii) they can be easily isolated and manipulated in vitro, and (iii) since plasmids are dispensable, their manipulation does not appear, in principle, to have adverse consequences to the hosts.

The feature that better defines plasmids is that they replicate in an autonomous and self-controlled way. The analysis of plasmid replication and its control has led to milestone discoveries, such as the existence of antisense RNAs, and has contributed to the unraveling of mechanisms of DNA replication, macromolecular interactions, and control of gene expression. The ability of some plasmids to pass across the so-called genetic barriers among different living organisms has posed questions about general mechanisms governing replication and about the communication between plasmid replication components and the host machinery involved in DNA replication. This plasmid-host communication has attracted the attention of researchers working in environmental and in evolutionary fields. Plasmid host range studies also have clear implications in clinical microbiology and in biotechnology. Despite their autonomous replication, plasmids extensively use the replication machinery of the host, and therefore plasmid replication studies facilitate the exploration of the mechanisms involved in chromosome replication.

PLASMID REPLICATION MECHANISMS

There are three general replication mechanisms for circular plasmids, namely, theta type, strand displacement, and rolling circle (RC). Historical development of research on plasmids has led to the idea that theta replication is more frequent in replicons from gram-negative than from gram-positive bacteria whereas the opposite is found for plasmids replicating by the RC mode. This belief is probably wrong. It is true, however, that present knowledge on theta-replicating plasmids stems from replicons from gram-negative bacteria and that on RC-replicating plasmids derives from replicons from gram-positive hosts. Strand displacement replication has been associated with broad-host-range plasmids from the IncQ family. The molecular interactions and the functional relationships that take place in these three types of replication mechanisms are the focus of this review. Linear plasmids have been found in both gram-positive and gram-negative bacteria, and their structure can be of two types: those having a hairpin at each end, and those having a protein covalently bound at their 5′ ends. Linear plasmids of the first group replicate via concatemeric intermediates, whereas those of the second group seem to replicate by a protein-priming mechanism, similar to that of bacteriophage φ29 (264). However, initiation of replication from an internal origin in a plasmid with a terminal protein has been reported (48). Linear plasmids have been reviewed previously (123), and they will not be discussed here. Replication of plasmids from gram-negative bacteria has been specifically addressed (168a).

Concerning their genetic structure, plasmids have an essential region which contains the genes or loci involved in replication and its control. The organization of this essential region corresponds, in general, to the one described in the replicon model. In addition, plasmids may bear genes that could be considered dispensable, although they could actually play an important role for the plasmid itself and/or for the host. Some of these so-called dispensable genes are involved in processes such as plasmid transfer and spread among bacteria, resistance to antibiotics and heavy metals, resistance to radiation, and transfer of DNA to higher eukaryotes. Within the plasmid essential region, several genes and sequences can be considered. (i) The first is the origin(s) of replication (generically termed ori), which is characteristic of each replicon. (ii) Although this is not a general feature, many plasmids encode a protein involved in the initiation of replication, usually termed Rep protein. (iii) The third is the plasmid-borne genes involved in the control of replication. The requirement of a plasmid-encoded initiator is reflected by the presence of DNA cognate sites in the origin of replication, where protein-DNA interactions take place. These specific sites are the hallmark of a class of replicons that are different from replicons that do not require specific initiators.

Replication by the Theta Mechanism

Replication by the theta-type mechanism has been most extensively studied among the prototype circular plasmids of gram-negative bacteria, although this replication mode has also been described for plasmids isolated from gram-positive bacteria, namely, the streptococcal/enterococcal Inc18 group (40), some lactococcal replicons (152), and at least one Bacillus subtilis plasmid (192). DNA replication through the theta mechanism involves melting of the parental strands, synthesis of a primer RNA (pRNA), and initiation of DNA synthesis by covalent extension of the pRNA (163). DNA synthesis is continuous on one of the strands (leading strand) and discontinuous on the other (lagging strand), although synthesis of the two strands seems to be coupled (reviewed in references 148 and 326). Theta-type DNA synthesis can start from one or from several origins, and replication can be either uni- or bidirectional. Under electron microscopy (EM), the replication intermediates are seen as typical Θ (“theta”)-shaped molecules that, when digested with enzymes that cleave within the replicated region, yield Y-shaped molecules (“forks”). The replication intermediates can also be monitored by one- or two-dimensional electrophoresis. These analyses provide information on the nature of the replication intermediates, direction of replication, location of the origin and terminus, and degree of coupling between leading- and lagging-strand synthesis.

With some exceptions, plasmids using the theta mechanism of replication require a plasmid-encoded Rep initiator protein. Some replicons may require the host DNA polymerase I (DNA Pol I) during the early stages of leading-strand synthesis. Some features of various well-known replicons which are described here are depicted in Fig. 1.

FIG. 1 — Origins of replication and related regions of some representative theta-replicating plasmids from gram-negative bacteria. Plasmid names (left), origins of replication (center), and Rep initiator-binding sites (right) are indicated. The symbols used are as follows: solid boxed arrows correspond to repeated Rep-binding sequences (iterons); open arrows above maps indicate inverted repeats that have partial homology to the iterons; solid arrowheads indicate repeats found in AT-rich regions (A+T); for R1, arrows indicate the imperfect palindromes initially recognized by the RepA initiator protein. Promoters are indicated as open arrowheads below the maps. Other sites of interaction are as follows: IHF-binding sites (open rectangles), *dnaA* boxes (solid rectangles); FIS-binding sites (hexagons), *dam* methylation sites (solid circles), and primase assembly sites (pas). Other sites are indicated in the figure. Maps are based on the following references: pPS10 (104, 215); pSC101 (132, 284); P1 (6); RK2 (278); R6K (277); R1 (101); ColE1 (280, 301); ColE2-type plasmids (124).

Origins of replication.

Plasmid origins of replication can be defined as (i) the minimal cis-acting region that can support autonomous replication of the plasmid; (ii) the region where DNA strands are melted to initiate the replication process, or (iii) the base(s) at which leading-strand synthesis starts. Replication origins contain sites that are required for interactions of plasmid-encoded and/or host-encoded proteins.

(i) General features.

With some exceptions, initiation of plasmid DNA replication requires a specific plasmid-encoded Rep initiator protein. This is reflected by the presence, at the origin of replication, of specific sequences with which the Rep protein interacts. Additional features found in many origins of theta-replicating plasmids are (i) an adjacent AT-rich region containing sequence repeats, where opening of the strands and assembly of host initiation factors occur, and (ii) one or more sites (dnaA boxes) where the host DnaA initiator protein binds (30, 163). Multiple Dam methylation sequences, which are present in the origin of replication of the Escherichia coli chromosome, oriC, can also be found at the origin of replication of plasmids such as P1 (36, 38) and pSC101 (30). Methylation is not essential for replication, its role being primarily in postreplication (3). Dam methylation sequences are not present in other plasmid replicons.

Comparative analysis of the structural organization of the Pol I-independent origins of replication predicts that although the Rep-binding site is located within a potentially curved DNA region, the DNA within the repeats of the AT-rich region is essentially straight (81). Intrinsic DNA bends at the Rep-binding sites would favor additional curvatures of the origin induced by Rep proteins. The origins of replication can also contain sites for factors (e.g., the integration host factor, IHF, or the factor for inversion stimulation, FIS) that play an architectural role. These host-encoded proteins favor a topological proximity between different ori regions or even between different origins present in the same plasmid (as in plasmid R6K [see below]). The plasmid DNA sites are essential components of the origin of replication since they are required to organize a functional replisome (61, 62, 282). The presence of DNA sites for the binding of structural factors, found at the origin of replication of several plasmids (see below), resembles the situation found in oriC (317).

(ii) Iteron-containing origins.

In many cases, the origin of replication contains directly repeated sequences, termed iterons, which are the binding sites for the plasmid-encoded Rep proteins and which have control properties. As discussed below, iterons not only are essential for replication but also are key elements for the control of plasmid replication (reviewed in references 51, 87, 155, and 223). Among plasmids which restrict their establishment to a single or a few species of enterobacteria, iterons have been described for several replicons like P1 (5), F (209, 295), pSC101 (52), R6K (97, 98, 277, 278), Rts1 (144), and pColIV-K30 (247). Iterons are also found in theta-replicating broad-host-range plasmids such as RK2/RP4 (241, 279), pCU1 (164) and pSa (286), as well as in conditional broad-host-range plasmids such as pPS10 (85, 104, 215). It should be noted that the presence of directly repeated sequences to which Rep proteins bind is not restricted to plasmids replicating by the theta mechanism, since these sequences have been reported for plasmids using the strand-displacement mechanism or the RC mechanism (171, 176, 267). Iterons can also be found outside the origin region in some plasmids (P1, F, RK2, R6K, Rts1, and pColIV-K30). These iterons, unlike the origin iterons, are not required for initiation but play an important role in the control of replication, as the origin iterons do. In plasmids that do not have auxiliary iterons, the origin iterons are the only locus involved in control (see below).

Iterons can be adjacent or separated by intervening sequences. Iterons found in the origin region tend to be arranged as tandem repeats situated at a distance that is, in general, a multiple of 11 bp, i.e., close to the helical periodicity of the DNA double helix. This implies that the Rep-iteron nucleoprotein complexes roughly place the Rep molecules aligned on the same face of the DNA. In general, for a particular origin, the sequences of the different iterons are not identical, although they adjust to a consensus motif that defines the essential features of these sequences. However, the four 22-bp iterons of plasmid pPS10 are identical (215). Statistical analysis of the frequency of base changes within the iterons of plasmid P1, combined with the available footprinting data, have been performed (242). Three highly conserved sequence patches are found within the iterons of this replicon. The two outer patches are separated by one helix turn. Protection experiments indicated that the major groove sides of those patches are contacted by the RepA initiator protein of P1. The function of the middle patch is less clear, but it may contribute to a proper conformation of the RepA-binding site. It is remarkable that this pattern resembles the DNA-binding patterns of dimeric proteins, some of which are transcriptional repressors. Taking into account that some of these iterons are contacted by monomeric forms of the initiator proteins, this may reflect the presence of two DNA-binding domains in RepA (discussed in reference 51), a feature that may be extended to other plasmid-encoded Rep proteins (100). Alignment of iterons present in the origin of replication of different plasmids showed the conservation of the hexanucleotide TCAGPuG (86), which is directly involved in the binding of the π initiator protein to the ori-γ region of plasmid R6K (97, 98).

Multiple iterons are required for origin activity, although not all iterons present in a given origin have to be essential. For instance, removal of one of the seven iterons from ori-γ of plasmid R6K has no effect but deletion of two reduces the efficiency of replication and deletion of three or more abolishes plasmid replication (160). Interestingly, the deletions make ori-γ replication independent of DnaA (16a). In the case of P1, all five iterons seem to be required for replication in vivo, but deletion of one can be tolerated in an in vitro replication system (314).

Single iterons are present in the ori-α and ori-β origins of plasmid R6K and in the minimal origins of plasmids ColE2 and ColE3. In R6K, ori-α and ori-β contain just one iteron and half an iteron, respectively (87). This situation is compensated for by the presence of a cis-acting sequence (enhancer), which is located in a third origin (ori-γ) that contains seven iterons. The enhancer facilitates the transfer of the initiator π protein, assembled at the seven iterons of ori-γ, to ori-α and ori-β, and leads to initiation of DNA replication (see below). The smallest of all the prokaryotic origins described so far have been found in the ColE2 and ColE3 replicons (322). They consist of a stretch of 47 bp (ColE2) or 33 bp (ColE3) and contain two major directly repeated sequences.

(iii) Other origin configurations.

Origins of replication without iterons can be found in other well studied theta-replicating plasmids like R1 and ColE1, as well as in plasmid pLS20 from B. subtilis.

(a) Plasmid R1.

Initiation of replication of R1 is dependent on a plasmid-encoded initiator protein, RepA. The minimal region required for RepA-dependent replication (oriR) is included within a 188-bp DNA region (183) and comprises (i) a 9-bp dnaA box, (ii) a contiguous 100-bp region where RepA interacts, and (iii) an adjacent AT-rich region containing three 9-mers. A detailed study of the site of RepA interaction revealed two RepA-binding sites: a preferential RepA site, termed site 1 (5′-CAGTTAAATG-3′), which is adjacent to the dnaA box, and a related RepA binding sequence, site 2 (5′-TGTTTAAAAG-3′), for which the protein has a lower affinity. This second site is contiguous to the AT-rich region. Sites 1 and 2 share a core sequence (g/tTTAAA) that is an imperfect palindrome (101). The intervening sequence between the sites shows potential intrinsic curvature. The presence of the dnaA box optimizes the action of the DnaA protein at the origin, both in vivo and in vitro, but it is not absolutely required for the DnaA-dependent replication of R1 (233). EM of replicating intermediates obtained in vivo and in vitro shows that initiation of R1 replication occurs in a locus that is separated from the minimal origin region (78). A G-type priming signal, located 400 bp downstream of the RepA-binding sequences, has been identified as the site where initiation of the leading strand, primed by DnaG, occurs (186).

(b) Plasmid ColE1.

ColE1 is the prototype of a class of small multicopy plasmids that replicate by a theta-type mechanism. Unlike R1, ColE1 does not require a plasmid initiator protein but requires DNA Pol I to initiate replication. The origin of ColE1 replication spans a region of about 1 kb that includes (i) sequences promoting the synthesis of RNA II, the primer of the leading strand (298, 299); (ii) sequences that allow a stable hybridization of RNA II to DNA (139, 189); (iii) sequences that favor specific processing of this coupled complex by RNase H, which generates the 3′ end needed to prime leading-strand synthesis (122, 139); (iv) a primosome assembly site (pas or ssiA) that allows the loading of the DnaB helicase and DnaG primase to initiate the discontinuous priming of the lagging strand (28, 189, 220) (a dnaA box that is close to pas can be used as a DnaA-dependent DnaB-DnaC assembly site [269, 270]); and (v) a sequence for termination of lagging-strand synthesis, terH, which determines unidirectional replication (57, 198). The first two sequences are the most relevant, since they are required for ColE1 replication in the presence or absence of DNA Pol I and RNase H (57, 158, 211). The origin of ColE1 replication, defined as the transition point between RNA II primer and DNA synthesized by DNA Pol I, has been positioned 555 bp downstream of the start point of RNA II (24, 300). This transition point corresponds with data obtained in vivo for plasmid pMB1 (closely related to ColE1) (29). Analysis of replication intermediates of ColE1 by EM, located a single origin and showed that replication is unidirectional. At an early stage, leading-strand synthesis proceeds in the absence of lagging-strand synthesis (297, 298).

(c) Plasmid pLS20.

An interesting example of plasmid from gram-positive bacteria is the B. subtilis plasmid pLS20, for which a preliminary characterization has been reported (192). This plasmid replicates by the theta mechanism, and its replication is independent of DNA Pol I and of a Rep initiator protein. Several palindromes flanking a putative dnaA box are located within the origin of replication of pLS20.

Rep proteins.

Up to now, dozens of plasmids have been isolated from most bacteria, but not many of them have had their basic replicons dissected and characterized to the level of their nucleotide sequence, and even fewer replicons have been genetically and biochemically studied in detail. The classic way of classifying plasmids is to distribute them among incompatibility groups, whose members have very similar origin sequences and replication control mechanisms. However, due to the difficulty to cope with a complex experimentally based classification of each newly isolated replicon, a criterion based on sequence comparisons appears to be much more practical. Such a criterion could be the comparison of the amino acid sequences of Rep initiator proteins, since they are encoded by most of the plasmids and they share common functions. Rep proteins recognize specific sequences at the origin of replication, similar to the DnaA initiator protein in bacterial chromosomal replication, and they generate a nucleoprotein initiation complex in which essential macromolecular interactions take place (Rep-DNA, Rep-Rep, and Rep with other initiation proteins of the host) (30). In addition, many Rep proteins can generate complexes that negatively regulate their synthesis and the frequency of initiation.

Based on amino acid sequence alignments of multiple Rep proteins from theta-replicating plasmids, it is possible to construct phylogenetic trees like the one depicted in Fig. 2b. It must be considered that for plasmid sequences, evolution can occur not only by mutation and selection but also by horizontal gene transfer. This constitutes an additional difficulty in establishing evolutionary relationships among plasmids. The phylogenetic tree groups replicons with similar replicative features: plasmids with Rep proteins binding to iterons (like pPS10, pSC101, R6K, and F) cluster apart from others whose initiators bind to nonrepeated sequences (R1 and its relatives), whereas broad-host-range plasmids (RK2, RA1, RSF10110, and TF-FC2) and replicons with dissimilar initiation mechanisms (phage lambda and phasyl) cluster in separated branches. Figure 2a shows an amino acid alignment of a large family of iteron-binding Rep initiators (encircled in Fig. 2b), comprising most of the best-characterized plasmids. The use of such alignments has allowed us the identification of Rep protein motifs, involved in protein-protein interaction (leucine zipper [LZ]) and in DNA binding (αhelix-turn-αhelix [HTH]) (93, 94, 103). A recent in vitro study performed with pPS10-RepA has revealed the existence of two globular domains, joined by a flexible linker, in a region of the protein located C-terminal to the LZ motif (102). Protein conformational changes are coupled to the dissociation of RepA dimers (which have a compact package of both domains) into monomers (with an elongated arrangement of the domains). The LZ motif and, to a lesser extent, the first globular domain mediate RepA dimerization. In the compact dimer, the second domain (including the HTH motif) binds to each arm of the operator sequence. In the elongated monomers, the second domain binds to the 3′ end of each iteron sequence whereas a DNA-binding activity in the first domain (previously cryptic) is responsible for additional recognition of the 5′ half. The sequence alignments in Fig. 2a support a similar structural organization for other Rep proteins of theta-replicating plasmids.

FIG. 2 — Sequence alignment and phylogenetic tree of Rep initiator proteins from theta-replicating plasmids. (a) Sequence alignment of the Rep initiator proteins of nine related plasmids of the iteron-containing class. Sequences were aligned with the program CLUSTAL W (version 1.5) by using, for pairwise alignment, gap opening and extension penalties of 10.0 and 0.1, respectively, and the protein weight matrix BLOSUM30. For multiple alignment, the delay for divergent sequences was set to 40% (294). The degree of sequence identity to the pPS10 initiator sequence for conserved residues is shown: ∗, identical residues in eight or nine of the total sequences; •, identical residues in five to seven sequences. Over the sequences is shown a secondary-structure prediction performed by the neural network algorithm PHD (260) on the output from CLUSTAL W: predicted α-helical regions (boxes) and β-strands (arrows). The two characteristic motifs found in the Rep initiators, LZ (hydrophobic heptad residues pointed to by open arrowheads) and HTH, are indicated. The EMBL database accession numbers are as follows: pPS10, X58896; pECB2, Y10829; pRO1614, L30112; pCM1, X86092; pFA3, M31727; pSC101, K00828; pCU1, Z11775; RepFIA, Y00547; R6K, M65025. (b) Phylogenetic tree for theta-type replicons from gram-negative bacteria, based on sequence alignments of their Rep proteins (such as the one shown in panel a for the pPS10 family, encircled in the tree with a dashed line). The sequences for 35 initiators were retrieved from databases, and a preliminary alignment was performed with CLUSTAL W (data not shown). Sequences that were virtually identical (pairwise scores, ≥90) were discarded, and a refined alignment was used as input data for the DISTANCES program of the Genetics Computer Group software package (95). Distance matrixes were corrected for multiple substitutions by the method of Kimura. The phylogenetic tree was built up according to the UPGMA method with the program GROWTREE (95). For further discussion, see the text.

(i) Protein-protein interactions: the leucine zipper-like motif.

A protein-protein interaction motif resembling the LZ is present in several plasmid-encoded Rep proteins. The LZ motif is responsible for dimerization in several eukaryotic regulatory proteins, through formation of two-stranded coiled coils (172). LZ-like motifs have been detected in the N-terminal region of the Rep proteins of several plasmids (103, 215) (Fig. 3). A mutational analysis has been carried out in the LZ-like motif of the RepA protein of plasmid pPS10 (94). Substitutions of the first two Leu residues of the LZ-like motif (d position according to a coiled-coil nomenclature) with Val resulted in a 13-fold decrease in the RepA association constant (as determined by sedimentation equilibrium analysis of maltose-binding protein–RepA fusions). This finding indicates that the LZ-like motif is a protein-protein interaction interface that regulates the equilibrium between monomers and dimers of the RepA protein. A conservative Ala→Val change in a different residue of the motif (b position) has no effect on monomer-dimer equilibrium, which points to a relevant and specific role in dimerization for the Leu residues of the motif. RepA mutants having Leu→Val substitutions were still able to interact in vitro with the iterons of the origin, indicating that the LZ-like motif is not directly involved in the binding of RepA to DNA. Further analyses indicated that RepA monomers bind to the iterons of the origin of replication whereas dimers of the protein bind to the repA promoter region, pointing to the functional relevance of the two forms of the RepA protein. Similar results have been obtained with the RepA protein of plasmid pSC101. This protein exists in a monomer-dimer equilibrium, although it is mainly in the monomeric form at the protein concentration present in cells harboring wild-type pSC101 (134). However, when the repA gene is overexpressed, replication is inhibited (133, 308). Inhibition under overexpression conditions was explained by assuming that elevated concentrations of RepA would promote its dimerization and that the RepA dimers would hinder the interaction of the active RepA monomeric forms with the iterons of the origin (134). Since overproduction of host DnaA protein can reverse inhibition by excess of RepA (133), an alternative explanation to understand inhibition by excess of initiation proteins involves titration of host replication factors.

FIG. 3 — The theta-type replicon of the *Pseudomonas* plasmid pPS10. The iteron-containing origin (*oriV*) and motifs found in the replication initiator protein (RepA) are depicted. The minimal origin (*oriV*) of pPS10 plasmid is a good example for a “canonical” iteron-containing origin. It is composed of four contiguous and identical 22-bp iterons arranged as direct repeats (half arrows), flanked by a 9-bp *dnaA* box (hatched) and AT- and GC-rich sequences (215). The pPS10 replicon also contains the *repA* gene, encoding the RepA initiator protein (shadowed ovals). RepA is under a monomer-to-dimer equilibrium, which has consequences for protein activity: RepA dimers bind to an inverted repeat (with a sequence partially homologous to that of the iterons) that overlaps with the *repA* promoter (P), acting as self-repressor of *repA* expression, whereas RepA monomers bind to the iterons to form the initiation complex (94). Protein motifs found in RepA are depicted under the protein gene. The LZ motif, responsible for protein dimerization, is represented as a helical wheel projection, in which the hydrophobic spine of Leu residues and the chemical nature of the other displayed residues is indicated (103). For the HTH motif, involved in binding to DNA, the two proposed α-helices are underlined and a stretch of basic residues at the C end of the DNA recognition helix is indicated (+), whereas an arrow points to the Gly residue thought to start the turn (93).

A region of the pPS10-RepA protein probably involved in protein-protein interactions with host replication factors has been identified. This information was obtained from the analysis of mutations that allowed the establishment of pPS10 (originally a narrow-host-range plasmid from Pseudomonas savastanoi) in E. coli (85, 104). Three independently isolated mutations were located within the region encoding the LZ motif of RepA. They all resulted in the same Ala→Val change (I+5 position; Fig. 3). Other mutations that broaden the host range of pPS10 map in residues adjacent to the LZ-like motif (180), indicating that the RepA region responsible for this phenotype, although partially overlapping, is different from the LZ motif. Since some of the mutations broadened the pPS10 host range without altering RepA-RepA, RepA-oriV, or RepA-repA promoter interactions (94), it would appear that these changes in the pPS10 initiator should favor proper RepA interactions with host initiation proteins.

Genetic analyses revealed later that the LZ-like motifs found in other Rep proteins play a relevant role. For instance, a mutation that affects the LZ-like region of the R6K initiator protein π resulted in a protein that failed to activate the α or β origins of replication (199). Translation of the gene for the π protein of R6K, starting from an internal initiation codon, can give rise to shorter protein variants in which most of LZ is deleted. This could represent a mechanism for regulation of the level of active replication protein (87). Another example is found in the RepA protein of plasmid pSC101, in which a mutation located in the proximity of the region encoding the LZ-like motif increases the copy number of this plasmid (133).

The existence of protein-protein interfaces apart from the LZ motif in initiator proteins is supported by several lines of evidence. First, the initiator protein of plasmid R1, RepA, interacts cooperatively with DNA sequences at the origin of replication, oriR (see above). A mutation located at the 3′ end of repA results in a thermosensitive replication phenotype (232). The protein variant conserves its ability to interact specifically with oriR, but the mutation affects the cooperativity of these interactions (101). This indicates that the mutation has affected a protein-protein interface and suggests that this interface could be located in the C-terminal region of RepA. Mutations affecting RepA residues involved in binding to oriR have not been described. Second, a single-amino-acid change at the N-terminal end of the initiator protein π of plasmid R6K allows this protein to discriminate between palindromic and nonpalindromic binding sites (325). It has been proposed that the change alters a protein-protein interface which modulates interactions of π protein with differently arranged DNA target sequences. Third, RK2 is a broad-host-range plasmid, a characteristic that is related to the existence of two forms of the replication protein, TrfA-44 and TrfA-33. The larger form, TrfA-44, is required for replication in P. aeruginosa (80, 274). The shorter version, TrfA-33, starts in an internal initiation codon of trfA and promotes the establishment of RK2 in most of its hosts, including E. coli and P. putida. In addition, the origin requirements are different in the two cases (56, 142, 213, 241). A mutation at the 3′ end of the trfA gene (affecting the two versions of the protein) modifies the host range of RK2 without altering the binding of the protein to DNA (45, 175; also see reference 241). These results suggest that the C terminus of TrfA plays an important role in the interactions of the initiator protein(s) with host replication factors. Interactions of plasmid initiator proteins with host replication factors have been reported in different systems: (i) the DnaJ protein interacts with the initiation protein of plasmid P1 (312a) and with other chaperones, promoting the efficient binding of this initiator to the origin of replication; (ii) the DnaA protein requires a functional interaction with the RepA protein of plasmid R1 to enter the DnaA box present in the origin of replication (184) (this protein interaction seems to be sufficient to promote DNA replication in the absence of the DnaA box [233]); and (iii) most interestingly, the DnaA, DnaB, and DnaG proteins of the host interact with the π protein of plasmid R6K (16a, 258a) (mutations in the π protein that disrupt the interaction with the DnaA protein are defective in R6K replication [16a]; the specific regions of DnaB and π proteins involved in their interaction have been defined [258a]).

(ii) Specific binding of Rep proteins to DNA: the helix-turn-helix motif.

As mentioned above, Rep proteins are able to specifically recognize DNA sequences in the origin region. In addition to this, some of the Rep proteins autoregulate their own synthesis at the transcriptional level by binding to sequences in the rep promoter (operator) which show some degree of homology to those present in the origin region. When autoregulation exists, either a single species of the protein is involved in both regulation and replication, or different species of the protein, monomeric and dimeric, recognize the origin and the regulatory regions, respectively (discussed in reference 51). rep mutants leading to impaired Rep protein-DNA interactions have been found in various plasmids.

The Pseudomonas plasmid pPS10 contains four identical iterons in its origin of replication and an inverted repeat (IR) in the repA promoter region. The iterons and IR have partial sequence similarity (92). RepA variants that fail to repress the repA promoter had amino acid changes within or in the vicinity of an HTH motif located at the C-terminal end of the protein (93). This motif has been described in many prokaryotic DNA-binding proteins, where it is involved in binding to specific regulatory DNA regions (39, 235). The RepA proteins affected in the HTH motif failed to interact with both the repA promoter and the oriV, indicating that the motif is involved in interactions with both the DNA regions. A working model proposes that the RepA protein contacts the inverted repeats of the repA promoter region as a dimer, using the HTH motif (92, 93). This HTH motif also binds to a conserved 3′ region in the iterons, which in their 5′ ends would be bound by another region of the RepA protein (102). A similar model has been postulated for other plasmid Rep proteins, in which monomeric and dimeric forms of the protein are involved in replication and autoregulation, respectively (discussed in reference 51).

In plasmid RK2, mutations that lead to TrfA protein variants affected in binding to the origin were scattered over a trfA gene region encoding the 162-amino-acid C-terminal moiety of the protein (46). In plasmid pSC101, the last third of the RepA protein is not needed for binding to specific DNA sites (181), which contrasts with the role of the C-terminal region in other initiators.

Initiation and elongation of replication.

(i) DNA replication dependent on plasmid initiators.

Initiation of replication requires the assembly of the complete replication machinery including DNA polymerase III holoenzyme (DNA Pol III-HE), DnaB helicase, and primase at the plasmid origin. Once the checkpoint corresponding to the initiation of leading-strand synthesis is past, replication continues until completion, following a process catalyzed by DNA Pol III and other host proteins. Most of the theta-type replicons require, at least, a plasmid-encoded Rep protein and the host DnaA protein for the initiation step. The general organization of the origin region in these plasmids resembles the arrangement found in oriC (30). The plasmid ori includes not only the specific sequences where the Rep and DnaA proteins interact but also an AT-rich region containing direct repeats, analogous to the 13-mers in oriC, where the DNA strands are melted. The AT-rich repeats have also been involved in the transfer of the DnaB-DnaC complex to oriC. In the theta-replicating plasmids, the Rep protein binds to specific sequences in the origin, forming a nucleoprotein preinitiation complex analogous to the one formed by DnaA at oriC. The Rep-DNA complex, in combination with DnaA, facilitates the transfer of the DnaB-DnaC complex to the origin and the opening of the strands in the AT-rich region. The structural organization of the initiation complex could be facilitated by host factors such as HU, IHF, or FIS. The assembly of the preinitiation complex and details of the molecular interactions leading to the initiation of replication are well documented for a few theta-replicating plasmids (described below).

(a) Plasmid pSC101.

RepA, the initiator protein of plasmid pSC101, exists in a monomer-dimer equilibrium, which determines the efficiency of RepA in replication (134). Monomers and dimers of RepA are both functional, but they play different roles: monomers bind to the iterons at the origin, promoting initiation, whereas dimers bind to the adjacent inverted repeat, repressing transcription of the repA gene (181). However, interactions of the RepA dimers with the inverted repeat also play a role in replication in the absence of the par locus (197). Initiation of pSC101 replication requires, in addition to RepA (308), the DnaA host replication initiator (113), and IHF proteins (91, 283). Binding of IHF to its target, within the AT-rich region, leads to DNA bending (283), which promotes interactions between DnaA molecules bound to dnaA boxes separated by some 200 bp (282). Binding of RepA to the origin region further stabilizes DnaA contacts with the distant dnaA boxes (282). The RepA-DNA-DnaA complex plays a role in replication but also in partitioning of the plasmid (54). Stable plasmid inheritance requires the par locus, which is close to the origin region: this locus contains a site for DNA topoisomerase II and also determines the proper supercoiling at the origin region needed for initiation (53, 132).

(b) Plasmid P1.

Plasmid P1 replication is dependent, both in vivo and in vitro, on the specific initiator protein RepA (6, 313) and on the host DnaA protein (110, 313). Formation of the initiation complex requires the monomeric form of RepA (315), and RepA-DNA binding is stimulated by heat shock chaperones. The latter proteins could contribute to the dissociation of the RepA dimers into monomers, which is the form of the RepA protein that recognizes the five iterons of the origin (58, 315). However, growing evidence indicates that the chaperones are required to activate the monomers of RepA (50, 78a, 236). Binding of the activated RepA monomers to the five iterons of the origin results in wrapping of the DNA around RepA, presumably due to in-phase bending of DNA (206). RepA monomers contact each iteron through two consecutive major grooves on the same face of the DNA helix (242). RepA alone is unable to melt the origin; this role is performed or favored by DnaA, which also stimulates the DNA-binding activity of RepA (206). There is a set of two tandem dnaA boxes at one end and a set of three tandem dnaA boxes at the other end of the P1 origin. Although either of the sets, or even just one dnaA box that conforms exactly to the consensus, is sufficient to support DnaA-dependent replication (4, 36, 38), melting of the origin region by DnaA is maximally efficient when both sets are present, probably due to DNA looping mediated by DnaA bound to the two sets (206). The orientation of the dnaA boxes and the different sensitivity of the two strands to reagents specific for single-stranded DNA suggest that DnaA-dependent loading of DnaB preferentially occurs in one of the strands, which can account for the unidirectional mode of P1 replication (206). Efficient replication of P1 requires adenine methylation of the five GATC sites of the origin. These GATC sites are clustered in direct heptamer repeats which are separated from the RepA-binding site by a GC-rich spacer (1, 2, 37).

(c) Plasmid RK2.

Important information on the initial events of replication of plasmid RK2 has been obtained (161a). The ClpX chaperone yields monomers of the plasmid initiation protein, TrfA, which is the form that is active in binding to the five 17-bp iterons of the origin (161b). This binding promotes, in the presence of HU protein, local strand opening within the AT-rich region of the origin. Interactions of the DnaA protein of the host with four DnaA boxes present in this region are also required for initiation of plasmid replication. These interactions increase, but are not strictly required for, the opening of the strands. DnaA is required for the delivery of the DnaB helicase to the origin region and both DnaA and TrfA are required for DnaB-induced template unwinding. This suggests a role of TrfA in the repositioning and activation of the DnaB activity (79b). The requirement of particular DnaA boxes is host-dependent (79a). This is consistent with the plasticity of the RK2 origin with respect to structural requirements for replication in different bacterial hosts.

(d) Plasmid R6K.

As stated above, replication from the γ ori of R6K requires π protein (96, 160, 272, 278). This protein can recognize different types of DNA sequences: iterons, enhancer, inverted repeats in the promoter of the pir gene (encoding the π protein), and the AT-rich segment of the origin (174). The π initiator promotes the initiation of replication from three origins of replication: α, β, and γ (reviewed in reference 87). ori-γ is in a central position, separated by 3 and 1.2 kb, respectively, from ori-α and ori-β. ori-γ contains seven 22-bp iterons, flanked by two IHF-binding sites and two dnaA boxes. Contiguous to the iterons is an AT-rich region which contains one of the dnaA boxes and one of the IHF-binding sites (62). ori-β contains half an iteron, and ori-α one complete iteron that are essential for function. Under standard conditions, ori-α and ori-β are more active than ori-γ, but they depend on a distant enhancer for activity (146, 147). This enhancer partially overlaps ori-γ, but its activity and the origin function have been distinguished by mutational analysis.

A dimer of π seems to bind to each of the seven 22-bp iterons present in ori-γ (86). Protein π binds preferentially to one of the strands of the ori-γ (98) and bends the DNA, generating a wrapped nucleoprotein structure (205). DnaA protein is required for replication from this origin, and it can bind the two dnaA boxes that flank the iterons (146, 147). Although two IHF-binding sites are flanking the iterons of ori-γ, the preferential or unique binding site(s) is the one located within the AT-rich region (62, 145a). IHF protein binding to these sequences induces conformational changes that are important in the regulation of replication initiation (145a). This binding could also affect the interaction of the π protein with the DnaA initiation protein of the host (16a). In the presence of normal levels of π, IHF is required for replication from ori-γ (61). An active ori-γ requires the binding sites for π, DnaA, and IHF proteins in the correct geometrical alignment (147). Protein π binds efficiently to the iterons of the ori-γ but not to ori-β or ori-α. However, the enhancer favors the long-range activation of ori-β and ori-α by transfer of the initiator protein, and possibly other initiation factors, from ori-γ (199, 203, 204). The activation of ori-β, unlike ori-γ and ori-α, does not require DnaA protein (147). Three new R6K gene products that distort essential sequences of ori-α and ori-β have been described (89). However, these proteins have been identified as proteins needed for conjugative transfer rather than for plasmid replication (230a).

(e) Plasmid R1.

Plasmid R1 is the most extensively studied member of the IncFII family of plasmids. In vivo and in vitro replication of R1 requires the initiator protein, RepA (77, 159, 183, 305). oriR, defined as the minimal region required for RepA-dependent replication of R1 in vitro, is bound specifically by RepA (101, 183, 184). Unlike the above cases, oriR does not contain iterons. RepA protein, probably as a dimer, recognizes sequentially (albeit with different affinities) the cores of two partially palindromic sequences (Fig. 4) (101). These sequences are located on the same face of the DNA helix, within a region of potential curvature, and are 8 helical turns apart (79). Interactions between RepA molecules bound to each of the sites could be responsible for DNA looping at the ends of a 100-bp region within oriR that is protected by RepA against DNase I cleavage (101). Following formation of the initial complex, additional RepA molecules could bind to the intermediate region by cooperative protein-protein interactions, generating a high-order complex. These interactions are needed for replication, as indicated by the defective replication phenotype associated with a repA mutant that failed to generate high-order RepA-oriR complexes (101, 232). The RepA-DNA complex seems to melt the DNA strands in the AT-rich region (185). In vitro replication of R1 requires DnaA protein (184, 231). DnaA binds to a dnaA box that is adjacent to the RepA-binding region, but this binding does not occur, or is very inefficient, in the absence of RepA (184, 233). It is likely that RepA-DnaA contacts guide the entrance of the DnaA protein in oriR, because the dnaA box is not absolutely necessary for the DnaA-dependent replication of R1 (233). Surprisingly, although in vitro replication of R1 is dependent on DnaA, this protein is dispensable for the replication of IncFII plasmids in vivo (20, 290). In vivo replication in the absence of DnaA is inefficient, but plasmid copy mutants that increase the levels of RepA protein improve the efficiency of replication (20). These results show the essential role of RepA in origin activation and imply that RepA could promote melting of the DNA strands at the origin and loading of the DnaB-DnaC complexes.

FIG. 4 — RepA-*oriR* complexes in initiation of R1 plasmid replication. A 100-bp region in the *oriR* replication origin is continuously bound by the plasmid RepA protein to form the initiation complex. There are no iterons in *oriR*, but two partially palindromic 10-bp sequences (sites 1 and 2) are found at the ends of that 100-bp region. They are flanked, respectively, by a consensus *dnaA* box and by three AT-rich sequences. These three sequences are believed to be melted to allow the DnaBC complex access to the open origin. A RepA initiator (hatched oval) dimer binds with high affinity to site 1, and then, in a second binding event, a different RepA dimer would bind with lower affinity to the distal site 2 sequence. The DNA of the *oriR* region could be bent to facilitate the topological proximity of sites 1 and 2, which are disposed on the same face of DNA double helix. Binding of RepA dimers to sites 1 and 2 would generate a small DNA loop, held together by protein-protein interactions. The DNA loop would be filled afterwards with more RepA molecules that are brought to the complex mainly by protein-protein interactions. This model is based on experimental gel mobility shift assays and footprinting data with both wild-type and mutant RepA proteins and *oriR* sequences (101). Arrowheads indicate DNase I-hypersensitive sites (the size is proportional to the intensity of cleavage), whereas arrows point to strong cleavage sites for hydroxyl radicals. The interaction of DnaA host initiator with its DNA-binding site is dependent on the previous formation of the RepA-*oriR* nucleoprotein complex (233). A requirement for a DnaK chaperone has been described for R1 DNA replication (105). A hypothetical role for DnaK in modulating the aggregation and activation state of RepA dimers, inspired by the findings for P1 plasmid RepA initiator (316), is also shown.

Interactions of RepA protein with the oriR region promote, both in vitro and in vivo, initiation of leading-strand synthesis at a DnaG-priming site (the G site, resembling the bacteriophage G4 origin for complementary strand synthesis) (21, 186) that is located 400 bp downstream of oriR. It has been proposed that the G site is activated when synthesis of the lagging strand, which initiates at the AT-rich region, reaches the G4-like site. Initiation at the G site, promoted by DnaG, cannot progress toward the origin. This leaves a gap that is filled later in the replication cycle (186). The relevant role played by DnaG in initiation of R1 replication is consistent with the complete inhibition of the in vitro replication of the plasmid by anti-DnaG antibodies (231). Antibodies against the E. coli single-stranded DNA-binding protein (SSB) also block in vitro R1 replication (231), indicating the essential role of this protein at an early stage in R1 replication. Finally, it has been established, both in vivo and in vitro, that replication of R1 requires DnaK protein (105). RepA protein of R1 seems to interact initially with the two partially palindromic sequences at the oriR region as a dimer, but further binding to the intermediate region could be by monomers, since there are no symmetrical sequences (102). This could support the notion that DnaK plays a role in the activation of R1 replication similar to that proposed for P1 replication (58, 315).

(f) Plasmids ColE2 and ColE3.

As mentioned above, the smallest of all the prokaryotic origins described so far have been defined in the ColE2 and ColE3 replicons (322). These plasmids require for replication a plasmid-specific initiator, but, like ColE1 (see below), they also require DNA Pol I to initiate leading-strand synthesis. Initiation of ColE2 and ColE3 replication is dependent on the synthesis of specific primer by their Rep proteins (288). A single-strand initiation site (ssi) for the priming of DNA replication has been located near the ori of ColE2 (219).

(g) Plasmids of the pAMβ1 family.

Replication of the pAMβ1/pIP501 broad-host-range plasmid family from gram-positive bacteria (32, 41, 42) requires, like ColE2 and ColE3, DNA Pol I and Rep protein. A model based on the synthesis of a primer RNA catalyzed by the host RNA polymerase (RNAP), the specific cleavage of the primer at the origin region (perhaps mediated by the Rep protein), and the extension of the 3′ end catalyzed by DNA Pol I has been proposed (41).

(ii) Replication independent of plasmid-encoded initiator proteins.

The best-characterized replicon that is independent of plasmid-encoded initiator proteins is ColE1. Initiation of ColE1 replication involves the consecutive activities of RNAP, RNase H, DNA Pol I, and DNA Pol III-HE (reviewed in references 163, 182, 248, 280, and 301). Transcription mediated by the host RNAP is required to synthesize the preprimer for leading-strand synthesis. Specific cleavage by RNase H of the preprimer (termed RNA II) annealed to DNA provides a 3′ end that defines the starting point for leading-strand synthesis. This synthesis is initially carried out by DNA Pol I. Steric hindrance of the bulky DNA Pol III-HE by the folded RNA II (upstream of the RNA-to-DNA transition point) probably prevents extension of the primer by this polymerase (188). DNA Pol I synthesizes about 400 nucleotides of the leading strand, exposing, on the displaced strand, a primosome assembly site (pas). Once the primosome is assembled at the pas site, it translocates in the 5′→3′ direction, unwinding the helix and priming the discontinuous DNA synthesis. At this point, DNA Pol I is replaced by the highly processive DNA Pol III-HE. The switch between DNA Pol I and DNA Pol III-HE could be favored by helix-destabilizing proteins bound to the template of the leading strand, which has to be exposed by the DnaB helicase (discussed in reference 280). Leading-strand synthesis can occur uncoupled from lagging-strand synthesis, and DnaG, but not DnaB, is dispensable for this uncoupled synthesis (281). Lagging-strand synthesis, initiated at the pas site, extends toward the promoter of RNA II but is arrested at a site (terH) 17 bp upstream of the leading-strand initiation site (57). The mechanism of this arrest is unknown. These events determine the unidirectional pattern of ColE1 replication.

Termination of replication.

The points at which theta-type replication terminates can be actively determined by molecular interactions at particular sequences. The first replication arrest sequence, ter, was identified in plasmid R6K as a barrier to the unidirectional replication initiated in either ori-α or ori-β of this plasmid. Replication starts then from the initial origin in the opposite direction and progresses to completion (177). The R6K terminus acts as a temporal barrier to replication initiated in other replicons (160). The nucleotide sequence of the ter region was determined, and the replication terminus of R6K was cloned (17, 18). The organization of this sequence as two separable and polar terminus sites was recognized and verified (130). The recognition of the essential features of the ter site allowed the identification of similar sites in plasmids of the IncFII (R1 and R100) and IncFI (repFIC) groups, as well as in the chromosome of E. coli. The ter sequence is the binding site of Tus, a monomeric protein of E. coli that promotes the termination of plasmid replication (121, 275). The identification of ter homologous sites in the chromosome of E. coli triggered replication termination studies in this bacteria and also in B. subtilis, where DNA-arresting sequences (IR-I and IR-II) and a dimeric protein that promotes the termination of DNA replication, RTP, have been identified (reviewed in references 14, 19, and 120). Unlike Tus, which acts as a monomer, RTP acts as a tetramer of two dimers (261, 262).

A major step in understanding active termination of DNA replication was the determination of the three-dimensional structure of the RTP dimer at the atomic level by crystallographic methods (43). This determination allowed the identification of protein-protein interfaces in RTP and opened the way to comparative structural analyses which suggested that RTP folding is similar to the “winged-helix” domain found in a family of DNA-binding proteins (43, 285). The polar arrest of the replication fork caused by RTP protein has been proposed to be due to specific interactions between the terminator protein and the DnaB helicase (116, 151, 172a, 261). Like the B. subtilis RTP-IR complex, the E. coli Tus-ter complex interferes with the helicase activity of the replisome complex in an orientation-dependent manner (151). Following the determination of the RTP structure, the structure of the Tus-ter complex was also solved by X-ray crystallography (143). This structure provided information on the singular architecture of the Tus protein and on the Tus-ter protein-DNA interactions involved. In addition, these studies support the speculation that the polar arrest of the replication fork, occurring at the Tus-ter complex, could be due to the polar inaccessibility of the helicase to the protein DNA-binding site. The termination of DNA replication is a regulated process, as indicated by the identification of a small protein of E. coli that binds to a terminator site and prevents replication fork arrest mediated by the Tus protein (214).

Sequences arresting lagging-strand synthesis, called terH, have been found upstream of and close to the pas site of ColE1; the arrest seems to be caused by the nonhybridized portion of RNA II (57, 212). It has been found that in multimers containing ColE1-type replicons oriented head-to-head, one origin of replication acts as a polar pausing site for replication initiated in the other origin (307). It is possible that this pausing is due to the stalling of the replication fork by the unhybridized portion of RNA II. Alternatively, the replisome could be transiently stalled at a protein-DNA complex, such as the pas site (discussed in reference 307). The potential role of an initiator protein to arrest replication progressing toward the origin has been reported in plasmid R1 (168). Active stalling of replication forks could be important for determining the direction of replication and for accurate termination and may modulate the efficiency of replication or the coupling between replication and cell division.

During the final stages of plasmid replication, catenates containing gaps in both daughter strands can be originated (212). These catenates can be resolved by either type I or type II topoisomerases. Genetic analysis revealed the involvement of a specific type II topoisomerase, Topo IV, in the segregation of plasmids and bacterial nucleoids (145). A two-stage model for the segregation of the replication products has been proposed (8, 9), with DNA gyrase reducing the linking number during elongation of DNA synthesis (stage I) and Topo IV resolving the supercoiled catenates which are the products of replication (stage II). Although cross-activities of DNA gyrase in decatenation and of Topo IV in supporting fork progression can be detected, in vivo and in vitro data confirm the specialized role of Topo IV in unlinking daughter replicons (117, 245, 246, 327). Maturation of the open-circular forms arising from the decatenation by Topo IV into supercoiled molecules can be efficiently carried out by DNA gyrase. Data obtained with plasmid R1 indicate that maturation of newly replicated DNA molecules is a slow process, which prevents rapid reutilization of the last replicated molecules (221). As the result of an odd number of homologous recombination events, dimers can arise. Replication intermediates can provide ideal substrates for these homologous recombination events. The resolution of DNA dimers by specialized systems can also be considered part of the replication termination process (discussed in reference 19).

Synopsis.

Replication by the theta-type mechanism is widespread among plasmids from gram-negative bacteria and has also been reported in plasmids from gram-positive bacteria. EM shows that replicating intermediates appear as bubbles (early stages) that, when they increase in size, result in theta-shaped molecules. Two early events in this mode of replication are the opening of the strands at specific sequences (the origin of replication) and the synthesis of RNA primers. Opening of the strands is catalyzed by specific initiators (Rep and DnaA proteins) and/or by transcription by RNAP. Initiation proteins promote, at the origin of replication, the sequential assembly of components of the replisome complex. The main replicative helicase of the cell catalyzes further unwinding of the strands. RNA primers are synthesized either by RNAP or by bacterial or plasmid primases. DNA synthesis of both strands is coupled and occurs continuously on one of them (leading strand) and discontinuously on the other (lagging strand). DNA Pol III is required for elongation of plasmid DNA replication. In addition, DNA Pol I can participate in the early synthesis of the leading strand (ColE1 and pAMβ1). Theta-type replication is, in most cases, unidirectional. Topoisomers are originated at termination (right-handed catenates), and their resolution requires the participation of Topo IV. Termination of DNA replication is determined in some plasmids by specific protein-DNA complexes.

Strand Displacement Replication

The best-known examples of plasmids replicating by the strand displacement mechanism are the promiscuous plasmids of the IncQ family, whose prototype is RSF1010. Members of this family require three plasmid-encoded proteins for initiation of DNA replication. These proteins promote initiation at a complex origin region, and replication then proceeds in either direction by a strand displacement mechanism (266; reviewed in reference 263).