Spread and survival of promiscuous IncP-1 plasmids

Plasmids classified to the IncP-1 incompatibility group belong to the most stably maintained mobile elements among low copy number plasmids known to date. The remarkable persistence is achieved by various tightly controlled stability mechanisms like active partitioning, efficient conjugative transfer system, killing of plasmid-free segregants and multimer resolution. The unique feature of IncP-1 plasmids is the central control operon coding for global regulators which control the expression of genes involved in vegetative replication, stable maintenance and conjugative transfer. The multivalent regulatory network provides means for coordinated expression of all plasmid functions. The current state of knowledge about two fully sequenced plasmids RK2 and R751, representatives of the IncP-1 and IncP-1 subgroups, is presented.

IncP-1 plasmids (IncP in Escherichia coli) have been of interest to molecular, environmental and industrial microbiologists since the 1970s.Plasmids classified within the IncP-1a subgroup were originally isolated in a Birmingham hospital (U.K.) in 1969 from Pseudomonas aeruginosa and Klebsiella aero-genes strains, and found to harbor genetic information that allowed their hosts to survive contact with multiple antibiotics (penicillin, kanamycin and tetracycline).The IncP-1 plasmids can replicate and be stably maintained in almost all Gram-negative bacteria and may be transferred by conjugation to Gram-positive bacteria, yeasts and eukaryotic cell lines (Heinemann & Sprague, 1989;Waters, 2001).The IncP-1 plasmid group is an ex-ample of highly potent, self-transmissible, selfish DNA molecules with a complicated regulatory circuit, which utilise very efficient strategies for stable maintenance in different host cells.Recent studies have provided evidence that plasmid maintenance mechanisms have a lot in common with the systems for chromosome segregation in bacteria.New data concerning the nature and properties of these DNA units also provide information about their hosts adaptation and versatility.Some questions about IncP-1 plasmids address their role as vehicles in the gene exchange phenomenon within bacteria, their functions in interactions between plant and animal pathogens and their hosts and the characterization of the intricate mechanisms of plasmid inheritance.
There is a growing list of naturally occurring IncP-1 plasmids identified in clinical and environmental bacterial from all over the world (Smith & Thomas, 1989).IncP-1 have been identified in soil from areas contaminated by industry, in pig manure, waste water of industrial origin, river epithelion and freshwater (Top et al., 1994;Pettigrew et al., 1990;Gotz et al., 1996).
IncP-1 plasmids are divided into three subgroups: a, b and g.Comparative sequence analysis of plasmids belonging to the subgroups a and b has not only shown a similar gene organisation pattern but has also facilitated the definition of the indispensable backbone, identifying so called "load" genes as well as monitoring their evolutionary pathways (Smith & Thomas, 1987).
Black box, control and stability functions; dark grey, replication, mating pair formation, surface exclusion; light grey, relaxosome assembly and auxiliary transfer functions.Triangles indicate positions of insertion (for detailed information see Table 1).
The plasmids assigned to the IncP-1a subgroup confer wide spectrum of antibiotic resistances (Smith & Thomas, 1989;Droge et al., 2000) while IncP-1b members carry multi-antibiotic resistance determinants and degradative cassettes (Burlage et al., 1990;Droge et al., 2000).There are no conclusive data about resistance to inorganic mercury conferred by plasmids of the latter subgroup.All IncP-1b plasmids seem to share homology with at least part of the Tn501/Tn21-related mercury resistance operon (R751, R772, R906, pJP4, pSS50, pSS60, pBR60, pTSA, pB10, pADP-1), although some lack the resistance genes and display a mercury sensitive phenotype (R751, R772, pB12).The IR element fused to the remnants of the merR gene appears to be homologous to the mer-proximal IR in Tn501 and Tn21.Such fusion of this element was recently reported for other mercury resistance operons, such as Tn5041, Tn5053, plasmid pKLH2 and Tn5718.In the case of R751, the sequence homologous to the IR element within transposon Tn4321 is inverted compared to that in R906, R772 and pJP4.The presence of sequence relics of the mercury resistance transposon Tn501 may reflect inheritance of the sequence from a common ancestor of IncP-1b plasmids.
Comparison of the physical maps and DNA sequences of IncP-1a and IncP-1b plasmids has lead to the conclusion that they are very similar and share a common structural organization (Pansegrau et al., 1994;Burlage et al., 1990;Martinez et al., 2001).The regions of replication control and stability, and two Tra1/Tra2 regions responsible for conjugative transfer constitute the plasmid "backbone".Tra1 and Tra2 in IncP-1b plasmids are always separated by regions of clustered restriction sites, which may represent hotspots for the integration of various determinants (Thorsted et al., 1998).Interestingly, a similar insertion site occurs between the oriV and trfA operons splitting the replication cassette into two parts.In IncP-1b plasmids, the backbone segments are in the same order, the only ex-ception being R772 in which a region containing the trb genes (Tra2) is inverted compared to the other members of this subgroup.
The number of insertions in the "backbone" differs among IncP-1 plasmids (Fig. 1).pJP4, pSS50 and pSS60 backbones are interrupted only once to accommodate the insertion of catabolic genes, invariably located between the oriV and trfA loci.In contrast, R751 has two such interruptions.The tsa degradative genes present in plasmid pTSA (which shares 99% identity with the backbone of R751) are integrated within a transposon-like structure located in a region where R751 carries a transposon conferring trimethoprim resistance (Tn402/5090).In the case of pADP-1, the "insertions" are found between the origin of replication oriV and the replication activating gene trfA, and between the Tra1 and Tra2 regions.The only plasmid analysed to date with three insertions in the main backbone is pB4.In common with plasmids containing only one interruption it has its degradative operon placed near oriV.The pB4 antibiotic resistance determinants are located proximal to both ends of the Tra2 region.Even when a resistance plasmid has only one interruption, such as pJP4, it usually has clusters of restriction sites at other potential sites of integration.
The "load" of IncP-1 plasmids presented in Table 1 is a "protective arsenal" of genes carried by individual plasmids which may influence the host survival and thus benefit the plasmids indirectly.The genetic information in the backbone sequence is much more significant because it ensures the existence of the plasmids.

REPLICATION DETERMINANTS
The ability of IncP-1 plasmids to replicate efficiently in a wide range of bacterial hosts is the major property of these autonomy units.They rely on a theta replication mode.The replication of plasmid RK2 in E. coli is now Table 1.Characteristics of IncP-1a and IncP-1b plasmids and their "load" genes.
Analysis of the oriV nucleotide sequences from RK2 and R751 revealed conserved features in this region.The minimal DNA segment providing oriV activity is approximately 393bp long and consists of nine direct repeats called iterons (17-mers) to which TrfA binds, DnaA boxes and A+T-rich and G+C-rich regions (Fig. 2) (Pansegrau et al., 1994).
The A+T rich repeats (13-mers) designated L, M1, M2 and R are the initial sites of helix destabilization.Iterons, to which the monomeric TrfA molecules bind are organized into three groups: containing one, three and five copies of the direct repeats, respectively (Fig. 2).Iterons 5 to 9 form a cluster upstream of the A+T region and they represent minimal origin activated by the TrfA initiation protein.
TrfA exists in two forms TrfA44 and TrfA33, which are expressed from the trfA mRNA using separate in-frame translation start sites (Pansegrau et al., 1994).The carboxyl terminus of TrfA, which constitutes the dimerization domain, is highly conserved in the form of the protein encoded by both RK2 and R751 plasmids and exhibits homology to the sub-strates of E. coli ClpX.It has been shown that the ClpX chaperone is involved in TrfA monomerization (Konieczny & Helinski, 1997a).TrfA dimers are believed to play a key role in the negative control of replication ("handcuffing model") (Toukdarian & Helinski, 1998).When present in excess, TrfA monomers bound to the iterons of the two replication ori interact with each other and form a sandwich-like structure, making the ori inaccessible for other components of the replication complex.Thus, the TrfA protein has a dual role in RK2 replication: as a positive initiator of replication and a negative regulator of plasmid copy number when in excess.
The positive action of TrfA as an initiator of replication is modulated by the DnaA protein.This factor binds to four DnaA boxes arranged as inverted pairs of a 9 bp consensus sequence separated by a few base pair which are located immediately upstream of the block of five iterons.The DnaA box cluster is located approximately 200 bp upstream of the A+T-rich region which provides the origin duplex opening site (Fig. 2).The binding of DnaA to DnaA boxes enhances or stabilizes the formation of the TrfA-mediated open complex, which requires ATP and proceeds in the presence of HU (Konieczny et al., 1997).In E. coli a mutation in DnaA box 4 abolishes replication and disturbs DnaA binding at all four boxes, whereas inversion of DnaA box 4 has no effect on binding to the other boxes (Doran et al., 1999).DnaA bound to its recognition sites interacts with the DnaB helicase to form a DnaABC complex in which DnaB is responsible for interaction between the TrfA protein and a prepriming form of the DnaBC complex (Konieczny & Helinski, 1997b).In the case of RK2, one molecule of the DnaB helicase is loaded at the A+T rich 13-mer repeat region and replication proceeds toward the G+C-rich region.The mechanism for helicase loading on ssDNA in the open duplex region and helix unwinding are not yet well understood.The process of DnaB loading and activation at oriV RK2 may be different in various hosts (Caspi et al., 2001).
The recent observation that the four DnaA boxes are not required for the replication of RK2 in P. aeruginosa (Fig. 2) raises the possibility of an alternative mechanism for the stimulation of open complex formation and DnaB loading at oriV (Doran et al., 1999).
Such a mechanism might also be utilised in other hosts like E. coli or Pseudomonas putida under certain conditions.It has been shown that mutation of DnaA boxes 3 and 4 (indispensable for replication in these hosts) or deletion of all four DnaA boxes may be compensated for by over-production of the DnaA protein (Doran et al., 1999).
It also remains to be determined whether the preinitiation role of the ClpX chaperone is performed by ClpX homologues in the wide range of Gram-negative bacteria that can support the replication of RK2 plasmid.A comparison of the nucleotide sequences of oriV regions of RK2 and R751 revealed an overall homology of 65% (Smith & Thomas, 1985).However, all the important features of oriV RK2 are also present in oriV R751 .In oriV of R751 a single copy of the iteron is found to be separated from the group of three repeats by a sequence relic of Tn501.This is not the case in plasmids R906 and R772 (64% overall homology) which have the single repeat adjacent to oriV and the other highly conserved features (Thomas & Smith, 1987).
The trfA gene of RK2 has been shown to activate replication from cloned oriV locus of R751, indicating that there is functional conservation between the two plasmids.Interestingly, while the TrfA33 protein is highly conserved in both subgroups (78% of aa identity, 91% homology), the N-terminal extensions of the TrfA44 forms are less similar (41% identity, 54% homology).Genetic drift in the N-terminus of TrfA1 is probably related to the adaptative ability of the two plasmid subgroups to specific hosts.The full length TrfA is required for replication in P. aeruginosa whereas TrfA33 is sufficient in E. coli and P. putida (Thorsted et al., 1996).

THE CONTROL OF COPY NUMBER
IncP-1 plasmids exist in bacterial cells at a copy number of 5-7 per chromosome.The number of replication initiation events depends on the level of the initiation protein TrfA.The control of copy number is achieved by tight regulation of trfA operon expression.The trfA promoter is classed as a very strong promoter so it may drive expression at a sufficient level in a wide spectrum of hosts.It is regulated by at least two plasmid-encoded regulators: KorA and KorB (described in the Regulatory Network Section) which bind to sites that overlap the -10 (O A -KorA binding site) and precede the -35 sequence (O B -KorB binding site) of the promoter.Whereas KorA represses trfAp approximately 10 fold and KorB on its own 3-5 fold, acting cooperatively they decrease the activity of trfA promoter about 1000fold (Smith et al., 1984).
The level of active TrfA is also regulated by its ability to form inert dimers.These dimmers, as mentioned previously, are dissociated into monomers active in binding to iterons and initiating replication by the action of the ClpX and ClpB chaperones (Konieczny & Helinski, 1997b;Konieczny & Liberek, 2002).While the group of five iterons is essential and sufficient for oriV function it has been demonstrated that the mini-replicon copy number is negatively dependent on the total number of iterons present, suggesting an accessory role of "extra" iterons in decreasing the level of TrfA available for the initiation of plasmid replication (Thomas et al., 1984).
It has also been demonstrated that the mini-replicon copy number is not proportional to the increase in the concentration of TrfA (Pansegrau et al., 1994).This prompted speculation concerning the role of TrfA as a negative regulator of copy number when present in excess ("handcuffing model" mentioned in the previous section).

CONJUGATIVE TRANSFER
IncP-1 plasmids have a broad-host-range with respect to their vegetative replication, which originates at oriV and also with respect to conjugation, which originates at oriT.The conjugative replication functions are different from those required for vegetative replication.Conjugative replication proceeds by the rolling circle mode.The range of recipient bacteria is broader than the host range in which the plasmids may replicate.Although conjugation is defined as bacterium-to-bacterium DNA transfer, in principle any type of cell can serve as the recipient as the process is typically dependent on factors encoded by the plasmid and expressed in the donor cell.It has been shown that plasmid RK2 mediates the transfer of DNA from bacteria to yeast (Heinemann & Sprague, 1989) and to Chinese hamster mammalian cells (Waters, 2001), although it cannot replicate in those eukaryotic organisms.The recent discovery that at least 41 genes are shared exclusively by humans and bacteria implies that some lateral transfer of DNA into the human genome may have occurred (Waters, 2001).
All IncP-1 plasmids are conjugationally highly promiscuous, with the exception of R906, isolated from Bordetella bronchiseptica R906.This plasmid has a reduced recipient host range although examination of its restriction profile reveals no drastic differences from other IncP-1b plasmids.The fact that IncP-1b plasmids share a common arrangement of transfer loci with the IncP-1a plasmid RK2 (Pansegrau et al., 1994) supports the assumption that they rely on similar conjugational transfer mechanisms and the extensively studied RK2 may serve as a paradigm for the whole group.

Mating pair formation (Mpf) apparatus
The strategy for plasmid spread among bacteria is reliant on intimate molecular associations between the surfaces of cells, which are plasmid-determined and in the case of IncP-1 plasmids promoted by extracellular filaments called sex pili.After initial contact has been established the DNA is thought to be transported in a complex with associated proteins through a channel at the mating bridge between the donor and the recipient cells.Formation of this channel as well as pilus assembly depends on gene products of the Mpf system (Pansegrau et al., 1994;Haase et al., 1995;Daugelavicius et al., 1997;Eisenbrandt et al., 2000).
Mpf genes (trbB, C, D, E, F, G, H, I, J, K and L) are located in the Tra2 region except for traF, which is located in the Tra1 region.These genes products are essential for intergeneric mating except for TrbK, the only surface exclusion protein encoded by IncP-1a subgroup plasmids (Giebelhaus et al., 1996).TrbK is located in the inner membrane and probably prevents formation of the DNA entry pore in the recipient cell envelope (Haase et al., 1996).The same genes are also required for phage reproduction (Haase et al., 1995).Most Mpf components are likely to be integral membrane or membrane-associated proteins as indicated by their aminoacid sequences and some experimental data (Grahn et al., 2000).It has been shown that over-expression of the IncP-1 Tra2 proteins increases permeability of the OM to lipophilic compounds and ATP and decreases intracellular K + content (Daugelavicius et al., 1997).Genes from the so-called trb region of RK2 and R751 have counterparts in bacterial species with well-studied transport systems, e.g. the type IV secretion system engaged not only in conjugative processes but also in the transfer of virulence factors to eukaryotic cells.These include Agrobacterium tumefaciens which uses the T-DNA transfer machinery to deliver nucleoprotein complexes to plant cells, and human pathogens such as Bordetella pertussis, Legionella pneumophila and Helicobacter pylori that use type IV systems to export effector proteins to the extracellular milieu or the mammalian cell cytosol.Homologous genes within functional clusters are often arranged co-linearly in the respective operons (Pansegrau et al., 1994;Vogel et al., 1998;Krause et al., 2000b).
The first gene in the trb operon, trbA, encodes a transcriptional regulator which controls expression of the tra and trb operons (see Regulatory Network).TrbA is non-essential for Mpf functions.
Six TrbB molecules associate to form a ring-shaped hexameric complex which is loosely associated with the inner membrane.There are four conserved domains in the TrbB homologues found in other DNA transport systems.These are a highly conserved Walker A box, an aspartate box, a Walker B box and a histidine box.TrbB possesses weak ATPase and kinase activities.The NTPase activity of VirB11-type protein (TrbB homologue of the Ti plasmid) (Krause et al., 2000a) might play a role in a chaperone-like function during the pilus assembly process.
The pilus subunit is encoded by the gene trbC.IncP-1 pilin is composed of a 78-amino acid TrbC polypeptide forming a ring structure via an intramolecular head-to tail peptide bond (Eisenbrandt et al., 1999).Three-step processing of the prepilin termini precedes the cyclization reaction.LepB, a chromosomally encoded protease cleaves 36 aa from the N-terminus of TrbC and another (so far unidentified) host-encoded protease removes 28 C-terminal residues dispensable for pilin maturation.The cyclization of TrbC occurs only in the presence of TraF.TraF protein (encoded in the tra region) catalyzes the intramolecular cyclization of TrbC by removing a further 4 aa from the C-terminus (Eisenbrandt et al., 2000).These modifications are consistent with data obtained by mass spectrometry.The resulting pili of RP4 and R751 seen in the electron microscope appear to be very rigid, inflexible structures measuring 10 nm in diameter (Eisenbrandt et al., 1999).
Recently, mating pair formation and plasmid R751 establishment in live E.coli cells have been visualized by fluorescence micros-copy using donor cells containing R751::lacO and recipient cells expressing a GFP-LacI fusion protein (Lawley et al., 2002).Evenly dispersed GFP-LacI fusion protein in the recipient forms foci upon the entry and replication (establishment) of R751::lacO.The localization of the GFP-LacI fusion protein enabled the authors to monitor the distribution of the plasmid in the recipient cell.Interestingly, the plasmid DNA was not randomly distributed throughout the recipient cells but rather localized either at the cell centre or at quarter positions in the cell similar to the localization observed during plasmid replication/cell cycle events (see Stable Inheritance of IncP-1 Plasmids).Another finding was that the junctions between mating cells occurred at various positions along the polar and lateral sides of both donor and recipient cells indicating that conjugative pores have no specific pre-programmed locations.

DNA transfer and replication (Dtr) functions
The preparation of IncP-1 plasmid DNA for transfer includes relaxosome formation and initiation of rolling circle replication.A nick within the plasmid oriT sequence followed by extension of the 3¢ end displaces the single strand that is to be transferred into the recipient cell.The proteins encoded by the tra region are involved in Dtr functions.
Tra1 of IncP-1a consists of three operons: the relaxase operon containing genes traJIH, the primase operon of genes traGFEDCBA (transcribed in the same direction) and the leader operon in which genes traK, traL and traM are transcribed in the opposite direction (Pansegrau et al., 1994).IncP-1b R751 lacks the two first genes of the primase operon (traA and traB) and shows poor conservation of traE and traD (59% and 56%, respectively) (Thorsted et al., 1998).The greatest divergence is displayed in the traJ and traK genes encoding essential relaxosome-proteins.Yakobson & Guiney (1983) showed that although both R751 and R772 could mobilize a plasmid carrying a segment of RK2, including the oriT and traHIJK genes, they could not mobilize plasmids carrying oriT RK2 alone.Thus, the initiation of conjugative replication via relaxosome binding and nicking of oriT is highly specific for plasmids of different IncP-1 subgroups.
The transfer origin (oriT) sequence of 250 bp, which is also located within Tra1, is the only cis-acting DNA element that is essential for transfer (Pansegrau et al., 1994).The nick region is highly conserved in oriT of the RK2, R751, R64 (IncI1), pTF-FC2 (IncQ) and in both the left and right T-DNA borders of the plant tumor-inducing Ti plasmids of A. tumefaciens (Pansegrau et al., 1994;Pansegrau & Lanka, 1996).This conservation emphasizes the basic similarity between IncP-1 mediated conjugation and T-DNA transfer to plants.The transfer origin of both R751 and RK2 lies in the intergenic region between the divergently transcribed traJ and traK genes and consists of their two promoter sequences located to the one side of the nick sequence (Fig. 3) where TraK binds and 19 bp inverted repeats recognized by TraJ on the other side of the nick site.A TraI binding site overlaps the nick site.Binding of TraJ has been proposed as the first step in functional relaxosome assembly (Pansegrau et al., 1994).In the second step TraI interacts with the TraJ-oriT DNA complex.An altered DNA conformation is thought to allow TraI to interact with the nick region.TraI access to its target site is also facilitated by TraK.TraK binds in the divergent promoter region and autoregulates the expression of both operons with the help of other relaxosome proteins (Pansegrau et al., 1994;Zatyka et al., 1994).In vitro studies on TraK-oriT nucleoprotein complexes have revealed that TraK induces local changes in DNA superhelicity.TraK wraps DNA and helps to melt the nick region.
The last player in the relaxosome assembly that has been identified so far is TraH protein.
TraH has been proposed to have chaperonelike activity, which might stabilize the relaxosomal nucleoprotein complex by specific interactions with both TraI and TraJ.When relaxosome formation has been completed, single-strand nicking occurs within the 6 bp nic site and TraI becomes covalently bound to the 5¢-terminal nucleotide (G) of the nicked strand (Pansegrau & Lanka, 1996).Rolling-circle replication may then start and the donor cell is ready to transfer DNA to the recipient cell.Two products of the primase operon are essential for conjugative transfer between E. coli cells: TraG and TraF (previously men-tioned for its involvement in Mpf apparatus formation).These proteins are highly conserved, fully exchangeable between the R751 and RK2 sytems and are thought to temporarily link the DNA processing and transfer reactions.Indeed, TraG is also known as "coupling protein".Unlike the relaxosome components, which are specific to their cognate oriT sequences, the coupling proteins of different conjugative plasmids belonging to distinct incompatibility groups (TrwB of IncW plasmid R388, TraD of IncF plasmids, VirD4 of Ti plasmids) show high levels of homology, suggesting that they have similar properties (Cabezon et al., 1994).These proteins have transmembrane a-helices in their N-terminal regions and their C-terminal domains are cytoplasmic and all contain conserved Walker motifs, although no NTPase activity has yet been reported.Direct interactions between the coupling proteins and relaxosome components have been demonstrated, both in vitro and in vivo.
TraG of IncP-1, which was shown to localize to the cytoplasmic membrane, may also bind DNA and could be a transporter protein interacting with the Mpf apparatus (the direct interactions between TraG and Mpf proteins have been demonstrated (Grahn et al. 2000).It has been observed that TraG coupling protein is essential for transfer of mobilizable plasmids.In vivo interaction between TraG and the Mob relaxase protein of pBBR1 has been reported (Szpirer et al., 2000) suggesting that the relaxosome may be delivered to the Mpf apparatus by TraG.There is also evidence that inhibition of IncP-1 plasmid conjugation and IncP-mediated mobilization of RSF1010 to cells already carrying an F factor or plasmid pKM101 is due to the action of PifC and FipA, respectively, on TraG (Santini & Stanisich, 1998).No functions have yet been assigned to the products of traA, traB, traD, and traE which are thought not to be essential for conjugation (Pansegrau et al., 1994).TraE of R751 is predicted to be a DNA topoisomerase on the basis of sequence analysis only (Lin et al., 1997;Thorsted et al., 1998).Two products of the traC gene, TraC1 and TraC2, possess primase activity.During conjugation, TraC1 and Ssb (single strand DNA binding protein encoded by trfA operon) are transferred to the recipient cells as part of the DNA-protein complex (Pansegrau et al., 1994).

STABLE INHERITANCE OF IncP-1 PLASMIDS
Conjugative transfer as the tool to spread IncP-1 plasmids across a wide range of bacteria and vegetative replication as the mean to establish these plasmids in new intracellular environments are accompanied by systems providing stable inheritance in the host populations.

The active partitioning mechanism
Low copy number plasmids encode functions for stable maintenance that are in many respects very similar to the strategies used by their hosts.The best example as illuminated by recent data is the active partitioning process of bacterial chromosome segregation to daughter cells at cell division.The best understood chromosomal Par systems include: sojspoOJ of Bacillus subtilis (Mysliwiec et al., 1991;Ireton et al., 1994;Lewis & Errington, 1997;Sharpe et al., 1998), parAB of Caulobacter crescentus (Mohl & Gober, 1997;Mohl et al., 2001) and P. putida (Ogasawara & Yoshikawa, 1992;Lewis et al., 2002;Godfrin-Estevenon et al., 2002).These systems share many features with the active partitioning of plasmid molecules between dividing cells (Bignell & Thomas, 2001).
All Par systems (plasmid and chromosomal) which have been studied to date require three components to drive the partitioning process: these are two trans-acting factors, ParA and ParB, and a cis-acting centromere-like site (by analogy to the role of the centromere during eukaryotic mitosis) on which a nucleoprotein complex is formed.ParB is a DNA-binding protein that recognizes and binds to the centromere-like site and then interacts with ParA whose ATPase activity (energy supply) is essential to the segregation process.The plasmid molecules may form pairs after replication that takes place at mid-cell (at the division plane).Although the consensus of opinion is that plasmid pairs have to be separated and that this is an active process, there is still controversy on the localization of the pairs as well as the timing of the whole process.Some have argued that the pairing and separation of plasmids are not strictly related to the cell cycle but that specific localization of plasmid clusters at polar or so called ¼ and ¾ positions must occur before the host cell divides.Others (Li & Austin, 2002) support the idea that the plasmid molecules are moving freely within the cytoplasm until the host cell reaches a certain stage of its cycle and then the plasmid pairs/clusters attach to the future site of division and become separated actively at cell division.
The protein components of the active partitioning process are always encoded by a single operon and are usually transcribed from an autoregulated promoter.The virulence plasmid pSLT (Cerin & Hackett, 1989;1993) and QpHI plasmid (Lin & Mallavia, 1994) whose parA and parB genes have separate promoters are the only exceptions to date.The structure of the centromere-like site, the length and DNA sequence involved in ParB binding are specific for different plasmids.The binding sequences are repetitive and a single locus may contain many ParB-binding sites (Hayakawa et al., 1985;Helsberg & Eichenlaub, 1986;Lane et al., 1987) or alternatively single binding sites may occur in multiple locations as exemplified by N15 phage (Ravin & Lane, 1999;Grigoriev & Lobocka, 2001) and IncP-1 plasmids (Williams & Thomas, 1992).On the basis of amino-acid sequence comparisons of ParA and ParB as well as centromere-like site location, the par loci of different plasmids have been classified into two main groups (Gerdes et al., 2000).The first group (I) consists of the par systems in which ParA contains typical ATPase motifs which match the Walker motifs for NTP binding.Group I has been further divided into two subtypes Ia and Ib on the basis of ParB homology as well as location of the centromere.The second group (II) consists of systems in which the ParA homologue is an actin-like ATPase (Fig. 4).The best studied representatives of type Ia par are the P1 and F plasmids.All putative chromosomal partitioning proteins also fall into this category.Type Ib is represented by par of the pTAR plasmid and type II by par of plasmid R1.
The IncC and KorB proteins of the IncP-1 plasmids are equivalent to ParA and ParB of type Ia plasmids.KorB and IncC are highly conserved between the IncP-1a and IncP-1b subgroups (77% and 72% nucleotide sequence identity, respectively).They are encoded by one operon, which is negatively autoregulated by the KorB protein.IncC enhances the repression exerted by KorB (Jagura-Burdzy et al., 1999b) (Fig. 6).
IncC, like other members of the ATPase superfamily, contains two motifs involved in NTP hydrolysis: the A motif (Walker A or P-loop) and the B motif (Walker B), which is always found adjacent to the former (Motallebi-Veshareh et al., 1990;Koonin, 1993).berculosis, H. pylori, P. putida, Streptomyces coelicolor and C. crescentus (Hayes, 2000).
The partitioning model suggests that a conformational change in ParA as it converts from the ATP-bound to ADP-bound form is a key driver of the partitioning cycle (Bignell & Thomas, 2001;Watanabe et al., 1992;Davey & Funnell, 1997;Bouet & Funnell, 1999;Quisel & Grossman, 2000).As the ParA proteins are known to be membrane bound, this model proposes that the ParA-ATP form promotes cell wall attachment of the partitioning factory (Lin & Mallavia, 1998) while ATP hydrolysis leads to its dissociation from the membrane.
IncC inactivation causes a severe instability phenotype of mutant plasmids (Bignell et al., 1999;Siddique & Figurski, 2002).When IncC function was impaired in vivo, the loss of stability of a minireplicon, composed of the P 7 replication cassette plus the incCkorB partitioning cassette, was due to a defect in the separation of plasmid molecules (observed as immunofluorescent foci linked in pairs) and the mislocalization of non-separated plasmid pairs to one cell pole (Bignell et al., 1999).
ParB proteins share a similar architecture and properties within the family.Proteins classified as ParB type Ia have a helix-turn-helix motif in their central region, that is likely to be involved in DNA recognition (Bignell & Thomas, 2001).These factors can bind DNA and act as transcriptional repressors and silencers.There are many experimental data demonstrating the dimerization and multimerization properties of ParBs.The dimerization domain is usually located in the C-terminus of the protein (Lobocka & Yarmolinsky, 1996;Jagura-Burdzy et al., 1999a;Kim & Shim, 1999).Typical ParB, as a component of the partitioning complex, has an ability to interact with ParA.The location of the domain which interacts with ParA seems to be more varied and may be found in the N-terminal, central or C-terminal parts of the ParB proteins (Surtees & Funnell, 1999;Kim & Shim, 1999;Lukaszewicz et al., 2002).
The ParB equivalent protein of IncP-1 plasmids, KorB RK2 , possesses DNA binding activity and also forms dimers and multimers.The dimerization domain has been localized to the C-terminus, with that responsible for multimerization probably located in the central part of KorB (Lukaszewicz et al., 2002;Delbruck et al., 2002).The central region is also responsible for interaction with IncC (Lukaszewicz et al. 2002).
The cellular location of IncP-1 plasmids has been studied by immunofluorescence microscopy using anti-KorB antibodies (Bignell et al., 1999).KorB localization is KorB DNA binding site-dependent.The KorB foci reflected a symmetrical plasmid distribution pattern which matches that of other type Ia plasmids (F, P1), being coupled to the replication zone in the center of the cell and then moving to the 1/4 and 3/4 positions before division (Erdmann et al., 1999;Bignell et al., 1999).This pattern of localization is remarkably well conserved among the Gram-negative bacteria P. aeruginosa, Vibrio cholerae and E. coli.One focus of RK2 represents clustered plasmid molecules (Pogliano et al., 2001;Pogliano, 2002).RK2 replication is coordinated with the host growth rate, so that the number of symmetrically localized foci per cell increases with cell length and decreases when the cells enter stationary phase.The correct RK2 plasmid distribution in filamentous cells proves that cell division is not required for plasmid segregation (Pogliano et al., 2001;Ho et al., 2002).
Although the IncP-1 par system appears to be closely related to that of the type Ia plasmids on the basis of ParA/ParB homologies, the cis-acting centromere-like sequence has not yet been precisely identified.The 13 bp sequences (5¢TTTAGCCGCTAAA3¢) specifically recognised by KorB (O B ) are dispersed throughout the whole plasmid.There are twelve such candidates (O B 1-O B 12) for the partitioning site in RK2 and eleven in R751 (see Regulatory Network).O B 3 which is present in both plasmids has been suggested to function as a centromere on inspection of the available data (Williams et al., 1998).Firstly, KorB may bind to all O B s but only binding to O B 3 is not potentiated by the presence of IncC (Kostelidou & Thomas, 2000).Secondly, the removal of O B 3, leaving O B 1 and O B 2 intact, leads to a strong destabilization effect of the test plasmid (greater than random loss) (Williams et al., 1998).However, the observation that this destabilization phenotype could be reversed to the loss rate of an unstable control by the deletion of O B 1 or inactivation of korB indicates that the situation is more complicated and prevents the drawing of firm conclusions as to the role of O B 3 in the resolution of RK2 partitioning pairs.So the question remains, whether one particular O B acts as the centromere site in IncP-1 plasmids or all sites are equivalent and chosen for this function at random.It is also possible that a few or all of these sequence elements participate in the formation of plasmid pairs.
The mechanism by which plasmids are transferred to the desired locations within the cell remains unknown, besides the fact that it is dependent on the ParA ATPase activity (Bignell et al. 1999;Erdmann et al., 1999).A recent double labelling study on P1, F and RK2 subcellular positioning indicated that each plasmid possesses its own unique strategy (Ho et al., 2002).The positions in which plasmids normally reside as well as the timing of segregation differ.It has been previously suggested that the timing of segregation is regulated at the stage of replication and the resolution of plasmid dimers (Austin et al., 1981).Recent studies correlate plasmid active partitioning with the cell cycle (Li & Austin, 2002).Some important features of the segregation mechanism of plasmids using the type II active partitioning system (Fig. 4) have been elucidated recently.Studies on R1 plasmid revealed that ParM, an actin-like ATPase which is equivalent to ParA, forms pole-to-pole axial filaments which facilitate R1 plasmid DNA segregation (Moller-Jensen et al., 2002).Filament formation requires the ParR protein (ParB homologue) and the centromere-like sequence parC, located upstream of the genes encoding ParM and ParR, and containing direct repeats to which ParR binds (Dam & Gerdes, 1994).ParR binding is important both for plasmid partitioning and for autoregulation of the par promoter.The new model implies that the ParM ATPase generates mechanical force that drives separation and subsequent movement of plasmid molecules to opposite cell poles after replication (Moller-Jensen et al., 2002;Van den Ent et al., 2002).In contrast to F, P or IncP-1, R1 plasmid molecules seem to be tethered at the cell poles rather than at the 1/4 and 3/4 positions (Jensen & Gerdes, 1999).The R1 example has provided the first evidence that active daughter plasmid segregation is reliant upon a plasmid encoded analogue of the eukaryotic mitotic spindle apparatus.The fact that the ATPase activity is essential for regular intracellular localization of R1 argues that it has analogous function to those of other par families.
In the light of the above data concerning R1 plasmid segregation, it seems reasonable to look for mitotic spindle structures involved in the active partitioning of type I plasmids.Studies on R751 revealed that the kfrA-upf54.8-upf54.4region adjacent to the active partitioning operon plays an important role in the stability of the plasmid (Adamczyk & Jagura-Burdzy, unpublished).It has been suggested previously (Jagura-Burdzy & Thomas, 1992) that KfrA of RK2 might be involved in the active partitioning process (see kfrA-upf54.8-upf54.4Genes as Part of the Central Control Region).KfrA RK2 was hypothesized to provide a filamentous bridge for the plasmid to travel along toward the cell poles after paired molecules separation (Williams & Thomas, 1992).The immunofluorescence studies on R751 plasmid confirmed the existence of such structures under conditions of KfrA over-production (Adamczyk, unpublished).Further experiments are needed to visualize intracellular structures without the necessity to over-produce the KfrA protein.
The efficiency of the KorB/IncC active partitioning system is reportedly improved by functions encoded by the klc/kle region in at least two hosts: P. aeruginosa and E. coli (Wilson et al., 1997;Thorsted et al., 1998;Bignell et al., 1999).When the kle/klc region was fused to the central control region (which on its own does not confer complete stability (Macartney et al., 1997) no loss of a test plasmid carrying this region was observed in E. coli (Thorsted et al., 1998).Inactivation of incC in the presence of the kle/klc region causes more severe plasmid destabilization than when the klc/kle region is absent.There are some differences in the gene content within the klc/kle region of plasmids RK2 and R751.The kleCD genes, which seem to represent duplicated kleAB genes in RK2, are absent from R751.On the other hand plasmid RK2 apparently lacks functional klcB, the second gene in the klcA operon, due to a Tn1 transposon insertion.The klcA gene product of both plasmids shows homology to the ArdB anti-restriction proteins of the IncN plasmid pKM101, suggesting a protective role after conjugative transfer (Larsen & Figurski, 1994).The KorC proteins encoded by this region (korC has been recently re-named klcC) have no assigned function apart from being regulatory factor for the klcA, kleA and kleC promoters (see Regulatory Network) (Larsen & Figurski, 1994).Recently, additional elements have been described that may contribute to the stability effect provided by the klc/kle region.Between the korC and kleA genes two 9 bp direct repeats (DR) have been identified, which when placed in trans significantly destabilize the RK2 plasmid in P. aeruginosa.This effect has not been observed in E. coli.These DR elements appear to be a target for a plasmid-encoded protein and, interestingly, the actual sequences differ between RK2 and R751 (Wilson & Figurski, 2002).

Post-segregational killing (psk) and multimer resolution systems (mrs) as accessory stabilization mechanisms
Plasmids of the IncP-1b subgroup seem to rely only on active partitioning as the stability function, while IncP-1a plasmids possess additional stability determinants that are also confusingly termed par region (Gerlitz et al., 1990).The locus occurs between the Tra1 and Tra2 gene blocks in RK2.In R751 this locus is occupied by the Tn402 mercury resistance transposon belonging to the Tn5053/Tn402 family whose members are known as res site hunters."Par" of RK2 encompasses two dis-tinct stability functions: post-segregational killing (psk) and multimer-resolution system (mrs), encoded by two divergently arranged operons, parDE and parCBA, respectively.parDE operon Deletion of the parDE genes was found to produce an unstable plasmid phenotype in E. coli and in P. aeruginosa (Sia et al., 1995).This operon was identified as the psk system.The presence of psk results in the death of plasmid-free segregants (Roberts & Helinski, 1992).The system ensures that plasmids are maintained in their natural hosts for a number of generations in the absence of selection.The post-segregational killing effect is based on a toxin-antidote principle.The plasmid encodes two components; a stable toxin and an unstable antidote, which prevents the lethal action of the toxin by direct interaction (proteic psk) or by posttranscriptional inhibition of its synthesis (by anti-sense RNA).Both elements of psk remain in the cytoplasm of plasmid free segregants and such cells are quickly deprived of the unstable antidote leaving the active toxin behind.
The parDE operon of RK2 codes for a proteic psk system.ParD is a 9 kDa protein which forms a homodimer with two functions: inhibition of ParE toxicity and an autoregulatory function through binding to a 33 bp region overlapping the -10 sequence of the parDE promoter.The N-terminal part of ParD has been found to possess DNA binding properties and anti-toxic activity (Roberts et al., 1993).Recent data obtained from NMR analysis have shown that the DNA-binding domain of ParD belongs to the ribbon-helix-helix fold family like the prokaryotic transcriptional repressors of E. coli: Arc, MetJ, Mnt and the streptococcal plasmid pMV158-encoded CopG (Oberer et al., 2002).The C-terminal part of ParD is very flexible.As this region appears to be responsible for ParE binding it may become structurally ordered upon anti-toxintoxin complex formation.
The 12 kDa ParE protein also exists in solution as a homodimer which interacts with the ParD dimer.ParE alone is unable to bind to the parDE promoter region but can bind there when complexed with ParD (Johnson et al., 1996).The specific cellular target for ParE toxicity has been recently identified as DNA gyrase (Jiang et al., 2002).Inactivation of DNA gyrase by ParE leads to DNA replication inhibition and cell filamentation.It is possible to reverse the replication arrest in vitro by the addition of the ParD antitoxin to the ParE-inhibited gyrase (Jiang et al., 2002).
Although the ParDE system was found to act post-segregationally it could also have some pre-segregational effect by providing a homeostatic control loop between host cell and plasmid replication.In this case impaired plasmid replication could lead to the inhibition of host replication via the toxin activity and thus render plasmid replication more efficient.
Similar killer gene systems are widely spread among plasmids and are also found in chromosomes of E. coli K-12 (Masuda et al., 1993), V. cholerae (Heidelberg et al., 2000) and the plant pathogen Xylella fastidiosa (Frohme et al., 2000).It is still unclear as to how the chromosomally encoded psk could benefit bacteria but a possible role in programmed cell death under stress conditions has been discussed (Engelberg-Kulka & Glaser, 1999).

parCBA operon
The mrs system present in almost all plasmids and chromosomes ensures that after replication or recombinational events each copy of the plasmid/sister chromosome will function as a separate entity.Theoretically, an absence of this system would cause catenation of circular molecules at each replication cycle that could lead to so-called dimer catastrophy in the case of plasmid DNA (Summers et al., 1993).In IncP-1a plasmids the parABC operon seems to provide the mrs functions.
The parA gene codes for a resolvase which acts on the plasmid multimer resolution site (res) to resolve multimers.This protein functions as a site-specific recombination system homologous to the Tn3 family of resolvases.The res site is located within a 100 bp region between the divergent parABC and parDE promoters.Mutational analysis within the parABC region indicates that the parA gene is absolutely essential for the stabilization of mini-RK2.Single deletions in the parB or parC genes have relatively little effect on the stabilization.However, deletion of both the parB and parC genes significantly reduces the effectiveness of the whole operon and the mrs function becomes strain dependent (Easter et al., 1998).
ParB is a Ca 2+ -dependent endo-and 5¢-3¢ exonuclease (Johnson et al., 1999)  It is worth mentioning that plasmid RK2 carries a functional Tn1 transposon with an active ParA-like resolvase, although deletion of this transposon has little effect on the stability of the plasmid compared to deletion of the parABC region.
Plasmids from IncP-1b subgroup do not possess the parABCDE locus however their stability seems to be unaffected.The question of how the lack of the parABCDE locus can be compensated for by IncP-1b plasmids is still unanswered.

THE REGULATORY NETWORK OF IncP-1 PLASMIDS
IncP-1 plasmids could easily become a burden to the host metabolism unless there is some regulation of plasmid gene expression.
The regulation of gene expression is achieved by multiple factors and each operon on the plasmid is controlled by more than one A linear map of the RK2 plasmid.Vegetative replication functions are marked red, light grey boxes correspond to transfer functions, bluish, stability functions; dark grey, phenotypic markers and transposable elements.General transcriptional organization of the replication, transfer and stable inheritance functions are shown by arrows.The scheme shows in colours the regions from which regulators are originated: orange, korA, blue, incC, green, korB, black, trbA, yellow, korC.Sites of action for repressors are showed as filled circles (for R751 some aspects of the regulatory network are hypothetical, extrapolated from the experimental data for RK2 and marked here by broken external lines).The O A and O B operators are numbered on RK2 map from right to left.
repressor.It is possible that in this way broad-host range plasmids respond to variable concentrations of repressors in different hosts.In addition, cooperativity in the action of different repressors potentiates the regulatory effect that a single repressor can exert.Overlapping regulatory circuits provide coordination of regulatory response that is tight but sensitive.
Besides the local control circuits mentioned in the previous sections (e.g.relaxosome operon, parABCDE), a common feature of IncP-1 plasmids is the existence of regulatory circuits in which at least four global repressors KorA, KorB, KorC and TrbA are involved.
KorA and KorB are encoded by the same operon (Figs. 4 and 6), located within the replication/stability segment in a region termed the central control region (ccr) which was mentioned earlier (Fig. 4) due to its function in active partitioning.
The korA RK2 gene encodes a 101 amino-acid protein (KorA R751 -100 aa).This is a potent global regulator that co-operates with KorB (Kostelidou et al., 1999) and exists as a homo-dimer in solution (Jagura-Burdzy & Thomas, 1995).The KorA monomer consists of three functional domains: the N-terminal region I acts as the dimerization domain (Kostelidou et al., 1998), region II containing a H-T-H motif facilities binding to O A operators (5¢TTTAGCTAAA3¢) (Smith et al., 1984), while the C-terminal domain is required for interaction with KorB (Kostelidou et al., 1999), although some data suggest that it also plays a role in dimerization (Bhattacharayya & Figurski, 2001).Regions I and II are separated from the C-terminal domain by a 4 aa linker sequence which might result in physical (spatial) demarcation in the KorA tertiary structure (Kostelidou et al., 1998)  O A s precede the -35 sequence of the regulated promoters.The affinity of binding depends on how perfect the palindromic sequence is and on the nature of flanking sequences.KorA has been shown to bind to and repress all promoters in which O A operators are present.The genes whose expression is regulated by KorA are involved in vegetative replication and stable maintenance.The strength of these regulated promoters is similar in both RK2 and R751.korA promoter, containing O A 1, is the most strongly KorA-repressed promoter in both plasmids (Kostelidou & Thomas, 2002).Through binding to the O A 3 operator, KorA mediates the switch in gene expression between vegetative replication and conjugation transfer that takes place at the trfAp/trbAp face-to-face divergent promoters.The strong transcriptional activity of trfAp inhibits RNA polymerase (RNAP) proceeding from trbAp.KorA plays a dominant role in the repression of the trfA promoter, causing elevation of the trbA promoter activity at the same time (Jagura-Burdzy & Thomas, 1994;Jagura-Burdzy & Thomas, 1995).
KorB RK2 (358 aa) binds to O B s with different affinities (Kostelidou & Thomas, 2000).Operator O B 10 in trfAp is bound by KorB with the highest affinity, which implies that it is a key element in controlling replication.O B 2 and O B 6 carrying a one base pair change (A®G) at position +4 from the centre of symmetry of the operator, which disrupts the sequence symmetry compared to the O B consensus sequence (5¢TTTAGC G / C GCTAAA3¢), have been shown to bind KorB with the lowest affinity.Additionally, flanking sequences modulate KorB binding to the O B operators (Kostelidou & Thomas, 2000).Among global regulators, KorB has the most general function in the IncP-1 plasmid biology.It represses operons involved not only in the vegetative replication and stable maintenance but also those required for conjugative transfer (Motallebi-Veshareh et al., 1992).
As mentioned earlier there are eleven and twelve highly conserved operator sequences recognized by KorB in the IncP-1 plasmids R751 and RK2, respectively.The location of these sequences in respect to the different  The hypothetical dimer is modelled because a-helical proteins have a tendency to form dimeric coiled-coil structures.
points, whereas Class III is comprised of six (intergenic/intragenic) sites (O B 3 to O B 8) located at least 1 kb from the promoters (Balzer et al., 1992;Williams et al., 1993;Kostelidou & Thomas, 2000) (Fig. 5).It has been established that KorB represses transcription by binding to the Class I and II operators (Jagura-Burdzy et al., 1999a), but not to Class III sites.In the case of Class II O B s, KorB acts at a distance and formation of KorB dimers/ multimers is required for the repressor function (Jagura-Burdzy et al., 1999a).KorB repression at a distance from the site of binding is facilitated by co-operation with other repressors KorA (Kostelidou et al., 1999) and TrbA (Zatyka et al., 2001).
IncC-1 which is encoded by the same operon as KorA and KorB and is involved with KorB in active partitioning as a ParA homologue, is also an important factor in the regulation of gene expression.Although it does not bind to DNA, IncC-1 enhances KorB repression at Class I and II operators.It also stimulates KorB binding and the formation of higher order complexes at all operators except for O B 3 (Jagura- Burdzy et al., 1999b).A third global regulator, KorC (85 aa) acts at three promoters, klcAp, kleAp and kleCp of RK2 and at two promoters, klcAp and kleAp, of R751 in the regions providing both plasmids with additional stability functions.It recognizes the palindromic O C sequence (5¢TAGGGCATAATGCCCTA3¢), located at the -10 region of both promoters (Thomas et al.,1988;Kornacki et al., 1990).
The last global repressor to be considered, TrbA RK2 (121 aa) (TrbA R751 -120 aa), has been shown to repress transcription from promoters of operons involved in conjugative transfer: traGp, traJp, traKp and trbBp (Zatyka et al., 1994).The binding site for TrbA (O T ) (consensus 5¢CGATATATCG3¢) is poorly conserved but is located close to the -10 region of regulated promoters (Zatyka et al., 1994, Bingle & Thomas, unpublished).The C-terminal part of TrbA shares similarity with the C-terminal domain of KorA and is proba-bly responsible for interaction with KorB.The co-operativity between TrbA and KorB in trbAp regulation takes place despite the significantly long distance (150 bp) between O B and O T operators (Zatyka et al., 1994;2001).The organization of the essential ccr of plasmid RK2 and R751 is basically the same.It contains the central control operon cco and three genes transcribed in the same direction: kfrA, upf54.8 and upf54.4.The genes encoding KorA and KorB, the key global regulators, as well as incC and the specific recognition sites O A and O B are highly conserved (Fig. 6).Clustering of the regulatory functions into a single operon is unique among plasmids.Besides korA, incC and korB, the cco of RK2 also contains two additional regulatory genes korF and korG which encode putative histone-like proteins.KorF (173 aa) and KorG (175 aa) repress transcription from kfrAp and trfAp, the two strongest promoters on plasmid RK2 (Jagura-Burdzy et al., 1991).Plasmid R751 lacks the korF and korG genes.
The ccr region of RK2 and R751 contains three O B sites: O B 1 (Class I) is in the korA promoter, O B 2 (Class II) within the kfrA promoter and O B 3 (Class III) in the intergenic region between upf54.8 and upf54.4 (Fig. 6).O B 2 in RK2 is unique, being the only KorB binding site without the perfect consensus sequence (it has an A®G transition at +4 position) (Macartney et al., 1997).In O B 2 R751 a perfect palindrome core is extended for 15 bp rather than 13 bp.The intergenic O B 3 site was postulated to play a centromere-like role in the active partitioning process (Williams et al., 1998).
While the cco gene products are conserved between plasmids RK2 and R751 and have assigned functions, upf54.8 and upf54.4 are orphan genes encoding proteins of high gluta-mine content.The Upf54.4 amino acid sequence contains the purine/pirimidine phosphoribosyl transferase signature (Pansegrau et al., 1994), whereas Upf54.8 shows the highest homology (58%) to the anthranilate synthase component II (Blast search).The sequence conservation of these genes between RK2 and R751 is significantly high (Fig. 6) although Upf54.4RK2 , in comparison with Upf54.4R751 , contains an additional 231 aa at the N-terminus (Pansegrau et al., 1994).
kfrA RK2 encodes a protein of 308 amino acids whereas kfrA R751 codes for a protein of 342a.a.Both have high alanine content.At the DNA level kfrA RK2 and kfrA R751 show only 50% identity.High G+C content and presence of G+C-rich repeats in this gene promote illegitimate recombination resulting in local deletions or duplications.Although the homology between KfrA RK2 and KfrA R751 is low at both the nucleotide and amino-acid levels, both proteins have a conserved tertiary structure.KfrA RK2 was predicted to possess an unusually long a-helical tail and a globular head domain.The a-helical tails might form coiled-coil structures organized in a filamentous network, which has been hypothesized to facilitate the plasmid partitioning process during cell division (Jagura-Burdzy & Thomas, 1992;Williams & Thomas, 1992).The predicted structure for KfrA R751 shows the same coiled-coil folding (Fig. 7).
Experimental data have revealed that both KfrA RK2 (Jagura-Burdzy & Thomas, 1992) and KfrA R751 (Adamczyk & Jagura-Burdzy, unpublished) repress their own transcription and that this function is dependent on the N-terminal region.Both KfrAs can form dimers and higher order complexes in vitro.kfrA RK2 has been postulated to form a monocistronic operon (Thomas et al., 1990) although there are some observations which suggest that expression of the two upfs in an RK2 derivative is dependent on transcription from kfrAp (Williams & Thomas, unpublished).Recent studies on KfrA R751 revealed that upf54.8 and upf54.4belong to the kfrA operon and the genes have been renamed kfrB and kfrC, respectively (Adamczyk & Jagura-Burdzy, unpublished).KfrA interacts with KfrB (Upf54.8)whereas KfrC (Upf54.4)has the ability to interact with KfrB implicating that all three proteins may form a multiprotein complex.A role for such complex in the stable inheritance of IncP-1 plasmids has been postulated (Adamczyk & Jagura-Burdzy, unpublished).
The steadily increasing volume of plasmid sequence data has shown that KfrA-like proteins are encoded by many plasmids of different incompatibility groups (Kwong et al., 1998;2000;Tauch et al., 2002;Burland et al., 1998;Close & Kado, 1992;Haneda et al., 2001).The fact that kfrA-related genes are frequently found close to the par locus or the oriV region rises the possibility that the biological role of this protein is more general and important for key survival functions.

CONCLUSIONS
The IncP-1 plasmids represent extremely well organized genetic units.Their genomes consist of a basic backbone disrupted by mobile elements with different phenotypic traits.The regions coding for "survival" functions may be physically separated but the genes remain co-regulated due to the global regulatory network The network of at least four overlapping regulons is supported by autoregulatory circuits.Such multivalent regulatory system provides unique coordination of all plasmid functions: vegetative replication, stable maintenance and conjugative transfer, facilitates the fine-tuning of gene expression and at the same time creates a safety device to minimize the effects of genetic changes or cell-cycle fluctuations.Assembly of the regulatory module (central control operon coding for at least three global regulators) is the unique feature of IncP-1 plasmids.It is believed that flexibility in the replication initiation, multiple stability mechanisms and coordinate regulation of all plasmid backbone functions guarantee the enormous adaptability and stable maintenance of the IncP-1 plasmids in such a wide spectrum of hosts.

Figure 1 .
Figure 1.Schematic presentation of gene organization in the IncP-1 incompatibility group plasmids based on available sequence data.

Figure 2 .
Figure 2. The oriV region from plasmid RK2.TrfA (replication initiator protein) binds to directly repeated iteron sequences, ClpX (chaperone) monomerizes TrfA.The orientation of iterons, the DnaA boxes to which DnaA binds as well as orientation of 13-mers in A/T rich region are indicated by arrowheads.

Figure 3 .
Figure 3. oriT of plasmid RK2.Proteins involved in nick site recognition as well as the nic (cleavage site) sequence are shown.Shaded box indicates the binding site for TraI.
IncC shows the highest similarity to ParA of the P. putida pM3 (IncP-9) and Rhodococcus erythropolis pFAJ2600 plasmids.The closest homologues of IncC among chromosomal ParA proteins are those of the Soj family of B. subtilis and ParA of Mycobacterium tu-

Figure 4 .
Figure 4. Comparison of the par loci of P1, pTAR, R1 and IncP-1 plasmids.The proteins involved in active partitioning process and a schematic presentation of cis-acting sites are shown.parS, cis-site, parC --centromeric sequences; O B 1, O B 2, O B 3 -KorB IncP-1 (ParB) binding sites; O B 3 -putative centromeric-like sequence; O A 1, O A 2 -KorA binding site, O K --binding site for KfrA.Partitioning proteins ParA and ParB, global regulator KorA and the KfrA protein are shown bound with promoters.Promoters are indicated by black open squares and direction of transcription by black arrows.
with homology to certain nucleases of Staphylococci and RuvC of E. coli.ParB has been shown to be a monomer in solution and display only weak sequence specificity.It is speculated that ParB plays a role in recombination, DNA repair and conversion of catenated plasmid dimers to the monomeric form.A putative signal sequence and a cleavage site found in the N-terminus of ParB suggest localization of the protein in the cell envelope.The function of ParC has not yet been determined.

Figure 5 .
Figure 5.A comparison of regulatory networks of the RK2 and R751 plasmids.
. There are seven KorA binding sites in RK2 and five in R751.In plasmid RK2 the O A operators have been assigned to class I(O A 1 korAp, O A 3 trfAp, O A 7 klaAp) or II (O A 2 kfrAp, O A 4 klcAp, O A 5 kleAp, O A 6 kleCp) (Jagura-Burdzy &Thomas, 1995).This classification is based on the affinity of KorA binding which coincides with the localization of O A within the promoter sequences, class I O A s overlap the -10 sequences whereas the low affinity class II

Figure 6 .
Figure 6.Comparison of the central control regions of RK2 and R751.ORFs are shown as open boxes, deletions in R751 are indicated by the delta symbol, transcripts (continuous arrows) and putative transcripts (broken lines) are marked, Rho-independent terminators are shown as stem loops and binding sites for KorA (O A ), KorB (O B ) and KfrA (O K ) are listed.The numbers in the shaded box correspond to % nucleotide and amino acid identities between the two plasmids.
promoters is the basis for the classification of KorB RK2 operators into three classes.Class I sites (O B 1 korAp, O B 10 trfAp, and O B 12 klaAp) lie immediately upstream of the -35 region of each promoter, Class II sites (O B 2 kfrAp, O B 9 trbBp, O B 10 trbAp and O B 11 kleAp) are located up to 189 bp upstream or downstream of the transcriptional start