Vol. 48 No. 4/2001

1003–1023

QUARTERLY

Review

Mechanisms of stable maintenance with special focus on plasmid addiction systems.

½ Urszula Zielenkiewicz and Piotr Ceg³owski

Institute of Biochemistry and Biophysics, Polish Academy of Sciences Received: 5 November, 2001, accepted: 24 November, 2001

Key words: plasmid addiction, post-segregational killing, partition; multimer resolution

The stable inheritance of bacterial is achieved by a number of different mechanisms. Among them are resolution of plasmid oligomers into monomers, active plasmid partitioning into dividing cells and selective killing of plasmid-free segre- gants. A special focus is given to the last mechanism. It involves a stable toxin and an unstable antidote. The antidotes neutralize their cognate toxins or prevent their syn- thesis. The different decay rates of the toxins and the antidotes underlie molecular mechanisms of toxin activation in plasmid-free cells. By eliminating of plasmid-free cells from the population of plasmid-bearing ones the toxin-antidote couples therefore act as plasmid addiction systems.

Plasmids are separate, autonomous genetic burgdorferi, Fraser et al., 1997; Bacillus cereus, elements present in a cell independently of Carlson & Kolstø, 1994). It is commonly ac- chromosomes. Most plasmids are small: from cepted that plasmid do not encode infor- several to 100 kb, but sometimes they are so mation indispensable for the functioning of large that using the size criteria their distinc- the host cell. However, plasmids specify nu- tion from the chromosome is difficult (e.g. in merous features advantageous for the host in Vibrio cholerae, Yamaichi et al., 1999; in specific environments, such as resistance to Rhizobium meliloti, Honeycutt et al., 1993). harmful agents, ability to degrade rare com- Different plasmids can constitute even up to pounds, pathogenicity, toxin production, ni- 50% of bacterial DNA (e.g. in Borrelia trogen assimilation etc.

. This work was supported by the State Committee for Scientific Research (KBN, Poland) grant No. 6 P04B 008 20 ½ Corresponding author: Doc. Piotr Ceg³owski, Institute of Biochemistry and Biophysics, A. Pawiñskiego 5a, 02-106 Warszawa, Poland; tel.: (48 22) 658 4723; fax.: (48 22) 658 3646; e-mail: [email protected] 1004 U. Zielenkiewicz and P. Ceg³owski 2001

Plasmids are very widely distributed among A—site-specific recombination systems Prokaryota and, in general, are inherited with ensuring that plasmid multimers arisen dur- a high degree of stability. In special environ- ing replication and(or) recombination will be mental conditions some plasmid genes confer resolved and thus every monomer copy will be a selective advantage. But, in many cases, independently subjected to random distribu- plasmids are retained over generations with- tion; out any selective pressure. Thus, there have to B—active partition process which pre- exist mechanisms which enable the mainte- cisely distributes plasmid copies to each nance of the plasmid during cell growth in daughter cell at division; nonselective conditions. Systems that contrib- C—plasmid addiction systems, e.g. func- ute to this stability are encoded by DNA cas- tions that kill or reduce growth of plasmid- settes and are, in most cases, independent of free descendant cells. one another. A particular plasmid can carry While mechanisms described in class A lead different stabilizing cassettes. Even more, to the optimization of random plasmid distri- cassettes from different plasmids may be com- bution, class B and C mechanisms ensure bined to give a stable replicon. better than random plasmid inheritance.

RANDOM- AND BETTER-THAN- SITE-SPECIFIC RECOMBINATION RANDOM PLASMID DISTRIBUTION SYSTEMS

During the process of cell division plasmid Identical copies of a given plasmid fre- copies are distributed between two descen- quently form oligomers. This reduces the dants. If plasmid distribution within a grow- number of independent units during segrega- ing cell is random, in the “ideal replicon”, it is tion. Moreover, since origins of plasmid repli- the number of plasmid molecules inside the di- cation are selected at random (Summers et al., viding cell that determines the probability 1993), the probability that plasmid monomers (P0) that one of the daughters will be plasmid will be selected for replication is, for instance, (1–n) free (Nordström & Austin, 1989): P0 =2 two-fold lower than for plasmid dimers. Con- , where n is the number of plasmid copies. sequently, dimers will out-replicate mono- Therefore, a cell having 30 plasmid copies at mers. The outcome of this ‘dimer catastrophe’ the time of division would produce in 109 cells (Summers et al., 1993) would be a consider- only two plasmid-free ones, whereas in the able plasmid loss from the host cells. To en- case of 5 copies, the chance of giving a sure a plasmid distribution process to be plasmid-free cell is as much as 1 per 16. highly efficient, there is a need for every As long as the plasmid copy number remains plasmid copy to be accessible for distribution. high, the subpopulation of plasmid-free cells is This is achieved by site-specific recombination extremely limited. For low copy number systems, also called mrs — multimer reso- plasmids obeying the random distribution law lution systems. In this process plasmid oligo- would mean that a significant fraction of mers naturally formed during replication or daughter cells will be plasmid free. This is in recombination are resolved into monomers in- contrast to the observed maintenance of creasing the number of independent mole- plasmids in the host cells. Several mecha- cules accessible for distribution. nisms have been proposed, and shown to oper- The multimer resolution systems consist ate, to explain this discrepancy (see below). of a site-specific recombinase, so called resol- They can be divided into three classes: A, B vase, and the defined nucleotide sequence res and C. located on the plasmid. By specific recombina- Vol. 48 Plasmid stable maintenance 1005 tion between repeated res sequences, the is very efficient (for example: the probability recombinase resolves oligomers to mono- of loss of P1 prophage, which is present as one mers. The mrs systems can be encoded en- extrachromosomal unit in the cell, is <1in tirely by the plasmid (e.g. loxP-cre of P1; Aus- 105 per generation; Abeles et al., 1985). tin et al., 1981) or combine the res site of the Models for the active partitioning process in- plasmid with a host encoded recombinase clude the pairing of plasmid molecules at (e.g. cer of ColE1 and host xerC, xerD, argR, cis-acting sites and subsequent separation of pepA; Summers & Sherrat, 1984; Sharpe et single plasmid copies to the opposite poles of al., 1999). the cell. The involvement of cellular compo- The multimer resolution systems of the nents in the process of plasmid separation is inc18 plasmid family from Gram-positive bac- proposed (Watanabe et al., 1989; Bignell et teria, although encoded by plasmids, require al., 1999). the host factor Hbsu to mediate resolution The partition process is carried out by two (Rojo & Alonso, 1994; Alonso et al., 1995; proteins usually designated as A and B. The Janniére et al., 1996). genetic organization of various partition sys- In the majority of high-copy number tems is similar, with the for the A pro- plasmids, the multimer resolution system is tein preceding the protein B-encoding gene the sole additional mechanism influencing and an adjacent centromere-like DNA se- their stable inheritance (for a review see quence. As an example, the structure of the Nordstrom & Austin, 1989; Hiraga, 1992; partition of P1 plasmid is shown Summers, 1998). In combination with the (Fig. 1). copy number and cell division control it en- The partition are negatively sures a very low frequency of plasmid loss. autoregulated by protein A (i.e. plasmids P1, F) or protein B (i.e. plasmids NR1, R1, pTAR). The repressor activity of one protein can be ACTIVE PARTITION enhanced by the other. The tight regulation of par operons expression is very important Many plasmids, especially those having mod- since an excess of protein B leads to plasmid erate or low copy number, possess an active destabilization (i.e. SopB of F plasmid; (i.e. energy consuming) mechanism for pre- Kusukawa et al., 1987; ParB of P1; £obocka &

Figure 1. Schematic representation of the par operon organization of P1 plasmid. The promoter–operator parPO and parS regions are blown up. Binding sites for ParB protein are boxed. Broken ar- row indicates the promoter. cise distribution prior to cell division of avail- Yarmolinsky, 1996). Protein B excess can also able plasmid copies to prospective daughter silence genes flanking the cis-acting sites (up cells. This process, named by analogy to chro- to several kilobases away) by polymerization mosome partitioning, as active partitioning, along the DNA starting from a certain point, 1006 U. Zielenkiewicz and P. Ceg³owski 2001 as has been shown for SopB (Lynch & Wang, , including Bacillus subtilis, Caulo- 1995) and ParB (Rodionov et al., 1999). bacter crescentus and Pseudomonas putida. All three components, partition proteins A, Their role in chromosome segregation has B and par site, are required for partition. The been shown for example in B. subtilis (locus ParB proteins bind specifically to their cog- soj-spo0J; Sharpe & Errington, 1996; 1998) or nate cis-acting DNA sequences termed centro- C. crescentus (genes parA, parB and parS; mere-like regions (Mori et al., 1989; Watanabe Mohl & Gober, 1997). In spite of a functional et al., 1989). These cis-acting sites are located similarity between chromosome and plasmid downstream (plasmids P1, F, RK2) or up- partitioning these two processes proceed dis- stream (NR1, R1, pTAR) from par genes. tinctly (Gordon et al., 1997). Their structure is unique for different plasmids, but always contains a characteristic number of iterated repeats which are the sites PLASMID ADDICTION SYSTEMS for specific binding of protein B. A motif characteristic for proteins showing Independently of the processes that increase ATPase activity is found in partition A pro- the probability of receiving a plasmid by a di- teins suggesting their role in providing energy viding cell there are very special strategies necessary for the process (Motallebi-Veshareh adopted by plasmids preventing plasmid-free et al., 1990; Davies et al., 1991). On the basis segregants from surviving. of sequence similarity ParA proteins from dif- The idea of an “addiction” mechanism lead- ferent plasmids can be grouped into two quite ing to very efficient plasmid maintenance co- distinct families: those with conserved mes from Koyama (Koyama et al., 1975). In Walker motif of ATPases and those showing his considerations he pointed out that if cells homology to actin-like ATPases (Gerdes et al., that lose an established plasmid die, the popu- 2000). Based on sequence analysis of proteins lation would never contain viable cured cells. related to the ParF protein from the Terms like killer system, killing-anti- multidrug resistance plasmid TP228, Hayes killing, post-segregational killing, toxin- proposed recently (Hayes, 2000) the existence antitoxin, poison-antidote, plasmid addic- of a third distinct subgroup of the ParA tion system or programmed cell death are superfamily, evolutionarily related to the all used to describe the situation when the MinD subgroup of cell division proteins. Pro- host cell is selectively killed if it has not re- teins of the ParB group are significantly more ceived any copy of the plasmid. diverse and difficult to classify. The molecular basis of this killing requires Recent experiments with fluorescence mi- the existence of at least two plasmid genes: croscopy have visualized bona fide plasmid one specifying a stable toxic agent and an- positioning during the time of cell growth at other coding for an unstable factor which pre- one- and three-quarter distance to cell poles, vents the lethal action of the toxin. While the (plasmids F, Kim & Wang, 1998; P1, toxin is always a protein, the antidote is either Erdmann et al., 1999; and RK2, Bignell et al., antisense RNA (which inhibits translation of 1999) or at the mid-cell with subsequent move- toxin mRNA) or a protein (Jensen & Gerdes, ment to opposite cell poles (plasmid R1, 1985). Jensen & Gerdes, 1999), correlated with the presence of ParB proteins. The factors in- Antisense RNA regulated plasmid addiction volved in the process of plasmid DNA move- systems ment are still unknown. Chromosomal homologues of plasmid parti- Plasmid stabilization systems regulated by tion proteins have been identified in many antisense RNA constitute a well conserved Vol. 48 Plasmid stable maintenance 1007 group called the hok-sok family (the name re- Sok-RNA is very labile (half-life about 30 sec), flecting functions of the host killing and hok mRNA is stable for hours (Gerdes et suppression of killing genes from plasmid R1) al.,1988). extensively studied and described in detail by The hok mRNA exists in a plasmid-carrying Gerdes (Gerdes et al.,1997). This family has cell in two forms: the inert full-length and a been found only in Gram-negative bacteria shorter active one. In full-length hok mRNA (see Table 1). the “fold-back-inhibition” element (fbi) present

Table 1. Plasmid- and chromosome-encoded hok homologues

A model of the genetic organization of the at 3¢ end pairs with the 5¢ end giving an RNA hok-sok system is presented in Fig. 2. structure inactive both in translation and The locus consists of three juxtaposed antisense RNA binding (Thisted et al., 1994a). genes: hok specifying a toxin, mok In this form hok mRNA is accumulated inside (modulation of killing) required for expres- the cell. The full-length stable form is slowly sion and regulation of hok translation and sok processed at the 3¢ end (Thisted et al., 1994b). coding for antisense RNA complementary to After removal of the fbi element, a refolded the hok mRNA leader region. Whereas structure of hok mRNA is accessible for antisense RNA binding and for translation. The constitutively expressed 67nt Sok-RNA binds to the leader sequence of hok mRNA. This complex is immediately cleaved by RNase III (Gerdes et al., 1992), thus impeding translation of hok mRNA. In a plasmid-free Figure 2. Schematic representation of the hok-sok cell, the absence of Sok-RNA (degraded by system organization. RNaseE) enables the translation of hok The shadowed box represents Hok toxin; open box, the mRNA. mok regulatory gene; Sok-RNA, antisense RNA, and Hok-like proteins are small (about 50aa), sokT, Sok-RNA target region. The thick, thin and bro- membrane-associated polypeptides. Over ken arrows represent hok mRNAs, sok antisense RNA - and promoter, respectively. In the full length mRNA expression of Hok leads to a decrease of cell the presence of fbi, the fold-back-inhibition element, is membrane potential, arrest of respiration and indicated. efflux of small molecules, resulting in cell 1008 U. Zielenkiewicz and P. Ceg³owski 2001 death (giving rise to characteristic “ghost” in cell death (Weaver & Clawell, 1989). The le- phase-contrast microscopy). thal action of Fst can be prevented by supply- The hok-sok systems may be very efficient. ing RNA II in excess (Weaver et al., 1996). The stabilization of unrelated replicons ranges between 50 (for high copy number plasmid) and 10000 fold (for low copy number PROTEIC PLASMID ADDICTION plasmids), as compared to random distribu- SYSTEMS (PPAS) tion (Wu & Wood, 1994). A number of homologues of hok-sok genes Common features of the proteic plasmid ad- from R1 plasmid system have been identified diction systems on other plasmids (see Table 1). If tested, all of them were able to post-segregational killing in As mentioned above, in PPAS both the toxin a manner similar to R1. In contrast, homo- and its antagonist (preventing agent) are pro- logues of this family found on bacterial chro- teins. Its functioning as the post-segregational mosomes are not functioning as post-se- killing system is based on different decay gregational killing agents. However, when rates of two proteins involved: the toxin is placed on a plasmid they clearly showed all much more stable than the antidote. This can characteristics of the hok-sok genes products be shown by provoking synchronous plasmid including the active toxic transmembrane pro- curing by inhibiting plasmid replication with- tein (e.g. hocC (=gef) (Poulsen et al., 1989) and out affecting cell growth and division (Jensen hocD (=relF) (Gerdes et al., 1986). The origin et al., 1995). and possible functions of these genes, widely The antidote prevents the lethal action of distributed in bacterial chromosomes, remain its cognate toxin by forming a tight complex mysterious. with it. So, as long as the antidote protein re- The antisense RNA-regulated stability deter- mains in excess relative to the toxin, the activ- minant has been also found in Gram-positive ity of the toxin is blocked. In a cured cell, bacteria. This determinant, designated par, when de novo synthesis of plasmid encoded stabilizes the enteroccocal plasmid pAD1 proteins is impossible, the concentration of (Weaver et al., 1996) and has no homology to antidote decreases quickly due to degradation the hok-sok family. The 400nt long par region by cellular proteases. The toxin, now released encodes two small, convergently transcribed from the complex, executes its toxicity caus- RNAs (210nt long RNA I and 65nt long RNA ing the cell death. Thus, in a population only II), with 33 codons for the fst (faecalis cells carrying the plasmid will exist (Fig. 3). plasmid-stabilizng toxin) peptide inside the All PPAS have similar genetic organization, longer RNA I. The smaller RNA II shows high shown schematically in Fig. 4. degree of complementarity to RNA I. Complex In all described cases the genes constituting formation between RNA I and the antisense proteic killer systems are organized in RNA II was recently demonstrated in vitro operons. With one exception of pRts1 (Tian et (Greenfield & Weaver, 2000). In this complex al., 1996), the first cistron encodes the anti- RNA II inhibits fst translation by preventing dote and the second the toxin. ribosome binding to the SD sequence seques- An important characteristic of these tered between complementary direct repeats. operons is their autoregulation at the trans- The half-life of RNA II (about 10 min) is sub- criptional level. The antidote protein itself or stantially shorter then that of RNA I (>1 h) in a complex with the toxin is a repressor of (Greenfield et al., 2000) as expected for anti- the promoter in question. In the majority of sense-regulated stabilizing systems. Overpro- cases, toxins enhance the promoter repres- duction of RNA I (or Fst peptide) causes host sion by the antidote (e.g.: plasmid P1, Vol. 48 Plasmid stable maintenance 1009

Proteins of PPAS are generally rather small (about 70aa–130aa), the toxin usually being slightly bigger than the corresponding anti- toxin. The only exceptions are the hig locus of Rts1 plasmid with a 104aa antidote versus a 90aa toxin and a large toxin Zeta protein (287aa) in the w-e-z system of pSM19035 (see Fig. 4). All PPAS toxins act from within the host cell. In spite of the structural and functional simi- larity of various PPAS systems there is no sig- nificant sequence identity among their genes. The only weak amino-acid sequence similarity was detected between CcdA of F plasmid and Kis of R1 (Ruiz-Echevarría et al., 1991a) as well as between ParD of RK2 and PasA of pTF-FC2 (Smith & Rawlings, 1997). Instead, it is very common to find highly conserved se- quences responsible for plasmid stability on Figure 3. The principle of PPAS functioning. many related plasmids which probably illus- A. In a plasmid-bearing cell: both proteins, the antidote trates horizontal transfer of genes between and the toxin, are produced. The antidote (constantly different plasmids and bacteria (Gerdes, present in excess) neutralizes the toxin by complex for- 2000; Hayes, 1998). B. mation. In cell after plasmid loss: the synthesis of There is no uniform pattern for the place- both, the antidote and toxin proteins does not occur. The antidote becomes degraded by cellular protease(s) ment of stability cassettes on plasmids, al- and the toxin remains free. The uncomplexed toxin though in a few cases a position close to the or- acts against its cellular target. igin of replication (e.g. ccd of F plasmid and phd/doc of P1 plasmid) could be pointed out. Names given to particular stabilizing loci are Magnuson & Yarmolinsky, 1998; R1, Ruiz- not consistent. In the majority of cases they Echevarria et al., 1991b; pTF-FC2; Smith & try to describe the observed behaviour of cells Rawlings, 1998). In case of F1 plasmid, a com- when stabilizing genes are activated. As, his- plex of both the CcdA and CcdB proteins is torically, the function of this type of genes was needed for the autoregulation of the ccd difficult to distinguish from partitioning, operon (Tam & Kline, 1989). In contrast, the quite often the misleading name par is used autoregulation of the PPAS encoded by for stability genes. The stb abbreviation of sta- pSM19035 plasmid does not include neither bility becomes frequent in recent literature. the antidote nor the toxin; it is achieved by a third component that exclusively plays a role of the transcriptional repressor (de la Hoz et EXAMPLES OF THE BETTER al., 2000). Precise autoregulation seems to be CHARACTERIZED PROTEIC a prerequisite for proper functioning of the PLASMID ADDICTION SYSTEMS stabilizing systems. The antidote proteins are present in excess ccd locus of plasmid F to the toxins in bacterial cells. Their half-life times in vivo are significantly shorter than The ccd locus of plasmid F is the best charac- those of the toxins due to degradation by cel- terized post-segregational proteic killer sys- lular proteases, as indicated in Table 2. tem. The large, 95 kb conjugative plasmid F 1010 U. Zielenkiewicz and P. Ceg³owski 2001

Figure 4. Genetic organization of PPAS operons. Bars represent: open — antidotes, grey — toxins and hatched — additional elements, respectively. Arrows indicate promoters. Autoregulatory circuits are shown for every system. Continuos and discontinuos lines indicate genes di- rectly involved in autoregulation and enhancement of autoregulatory activity, respectively. See text for details. exists in E. coli cells in only one or two copies The name of the ccd locus (coupled cell per chromosome. Yet it is extremely well division) originates from the observation maintained in a growing population. This high (Ogura & Hiraga, 1983; Miki et al., 1984) that degree of stability plasmid F achieves by com- cell division is dependent on previous replica- bining the effects of multiple control ele- tion of F plasmid. This inadequate description ments: two functional replication systems, was later deciphered using the same letters as several site-specific recombinases, the active control of cell death (Van Melderen et al., partition region sop and three post-segre- 1994). Indeed, it was shown (Jaffé et al., 1985) gational killing systems (Nordström & Austin, that genes of the ccd locus mediate plasmid 1989). The flm (Fleading maintenance) locus maintenance primarily by killing plasmid-free and srnB (stable RNA degradation) locus are segregants. homologous to the hok/sok family regulated Theccd locus is composed of two genes: ccdA by antisense RNA (Golub & Panzer, 1988; and ccdB (also called H and G or letA and Ohnishi et al., 1977; Nielsen et al., 1991). letB, respectively) encoding the correspond- Vol. 48 Plasmid stable maintenance 1011

Table 2. Examples of proteic plasmid addiction systems

ing proteins: the CcdA antidote and the CcdB breaks in the DNA, in turn, induces the SOS toxin (Bex et al., 1983; Miki et al., 1984). These response. genes are organized in an operon together In addition, the CcdB protein is a direct in- with the resolvase-encoding gene resD (Lane hibitor of gyrase (Miki et al., 1992). In the ab- et al., 1986). Expression of the operon is nega- sence of covalently bound DNA CcdB binds to tively controlled by a complex of the two Ccd free GyrA subunit making gyrase catalytically proteins (de Feyter et al., 1989). The domains inactive. Both CcdB activities are suppressed of both proteins involved in autoregulation by the same mutations in GyrA (substitution are different from those responsible for the R462C leading to resistance to CcdB; Bernard killing-antikilling effect (Salmon et al., 1994; et al., 1993) and in CcdB (mutations in the Bahassi et al., 1995). In the absence of the three carboxy-terminal amino acids that an- CcdA antidote, CcdB causes a reduction in nul the lethal activity of CcdB; Bahassi et al., DNA synthesis (Jaffé et al., 1985), activation 1995). of the SOS regulon (Karoui et al., 1983; Reece The recently determined crystal structures & Maxwell, 1991), cell filamentation (Miki et of a large, 59 kDa, fragment of the amino-ter- al., 1984 ), formation of anucleate cells and, as minal region of GyrA (Berger et al., 1996) a consequence, cell death (Jaffé et al., 1985). have led to a model for the GyrA–CcdB com- Bacterial mutants resistant (Bernard & Cou- plex formation (Lorris et al., 1999). The model turier, 1992) and tolerant (Miki et al., 1992 ) to accommodates the critical C-terminus of the the lethal activity of CcdB were altered in the toxin dimer in the central hole of a GyrA gyrA gene. The product of gyrA constitutes a dimer during its cycling on DNA, impeding part of an essential enzym — gyrase, responsi- the closing stage and therefore the religation ble for introducing negative supercoils in of nicked DNA. The CcdA antidote not only DNA. During the breaking and rejoining reac- prevents this complex formation, but also re- tion, the 5¢-ends of substrate DNA are cova- leases CcdB from inactive complexes by form- lently linked to the A subunit of gyrase. The ing a new tight CcdA–CcdB complex (Bernard CcdB toxin traps the cleaved DNA–gyrase et al., 1993; Maki et al., 1996). complex and impedes the resealing of the The ATP-dependent serine protease Lon is nicked DNA (Bernard & Couturier, 1992; Ber- responsible for degradation of CcdA. The nard et al., 1993). Formation of double-strand half-life of the CcdA protein was estimated to 1012 U. Zielenkiewicz and P. Ceg³owski 2001 be about 1 h, whilst CcdB remained stable for PemK/Kid protein is DnaB helicase. The pri- over 2 h (Van Melderen et al., 1994). mary effect of the pem/parD system in a ma- Among the F plasmid encoded stability cas- jority of strains is cell division inhibition settes, the ccd system is a rather inefficient rather than killing of the plasmid-free segre- one: only 10-fold stabilisation is achieved, gants (Tsuchimoto & Ohtsubo, 1989; Jensen compared to 100-fold for flm or 1000-fold for et al., 1995) thus explaining its modest, less sop (Jensen et al., 1995; Boe et al., 1987). Per- than 10-fold, plasmid stabilization. haps, as the CcdB toxin primarily induces the SOS regulon, rather than immediate cell phd-doc locus of bacteriophage P1 death, part of its role may be to generate ge- netic diversity in those bacteria which do not Bacteriophage P1 lysogenizes E. coli cells as have the F plasmid after division (Couturier an extremaly low-copy number plasmid. Its et al., 1998). stable maintenance within bacterial cells is ensured, in part, by an addiction mechanism mediated by two specific proteins: the anti- pem/parD locus of plasmids R100/R1 dote Phd (prevent host death) and the toxin This is the second (and so far the last) PPAS Doc (death on cure), (Lehnherr et al., 1993). with a known cellular target of lethal action. The corresponding genes phd and doc are or- Loci described as pem (plasmid emergency ganized in an operon, the transcription of maintenance) for plasmid R100 (Tsuchimoto which is negatively autoregulated by Phd (two et al., 1988) and as parD for plasmid R1 dimers of the Phd protein bind to the opera- (Bravo et al., 1987) encode the antitoxin PemI tor, Gazit & Sauer, 1999) and enhanced by the identical to Kis (killing suppressor) and toxin Doc protein (Magnuson & Yarmolinsky, PemK identical to Kid (killing determinant). 1998). The pem/parD operon possesses the canoni- Free Phd protein (unbound to the operator cal features of PPAS: it encodes two small pro- or Doc) exists predominantly in an unfolded teins with the gene for the antitoxin preceding conformation and is subject to degradation by that for the toxin; it is autoregulated at the the ATP-dependent serine protease ClpXP level of transcription by a complex of the two (Lehnherr & Yarmolinsky, 1995). Physical proteins involved and the decay time for binding between the prophage stabilizing pro- these proteins is different. The degradation of teins was shown for a nontoxic mutant of Doc the antidote PemI/Kis is due to the Lon prote- in vitro (Gazit & Sauer, 1999). The cellular tar- ase (Tsuchimoto et al., 1992). get of the action of Doc remains unknown but In vitro, it has been shown that the Kis and under conditions of its action the protein syn- Kid proteins form a complex. Also, thesis is inhibited. Recently it was demon- DnaB-dependent initiation of plasmid ColE1 strated that post-segregational killing induced replication was specifically inhibited by the by phd-doc addiction genes requires mazEF addition of Kid. The replication was restored chromosomal system (Hazan et al., 2001). The when the antidote Kis was added. In vivo,in- phd-doc system stabilizes mini P1 plasmid hibition of lytic l prophage induction (whose about 7-fold (Lehnherr et al., 1993). replication initiation is DnaB dependent) was observed after inactivation of the antagonist parDE locus of plasmids RK2/RP4 protein Kis. Furthermore, overproduction of DnaB suppressed the lethal action of Kid in The broad-host-range plasmid RK2 (60kb, vivo (Ruiz-Echevarría et al., 1995). Taking this identical to the plasmid RP4, IncP incompati- into account, it seems well documented that bility group) is maintained at the relatively the cellular target of the action of the low number of 5–8 copies per chromosome Vol. 48 Plasmid stable maintenance 1013 and is very stable in a wide range of in the case of conjugal plasmids may be addi- Gram-negative bacteria. tionally masked by the simultaneous transfer The investigation of separate loci involved of the plasmid to the plasmid-free cells (Easter in the stable maintenance of plasmids from et al., 1997). the IncP incompatibility group is extremely The RK2 plasmid also carries several poten- difficult because of the interrelated co-ordi- tially host-lethal kil loci (kilA, B, C and E also nated regulation of the vital functions by the denominated as kla, klb, klc and kle). kilB is plasmid central control region ccr. The involved in conjugal transfer (Motallebi- juxtaposited influence of the regulatory pro- Veshareh et al., 1992). The function of other teins KorA and KorB on transcription of kil genes is unknown, but their unregulated many promoters makes the final effect of the expression can lead to the death of host E. coli propagation of IncP plasmids particularly de- cells (Figurski et al., 1982). pendent on the host strain and conditions of An intriguing question whether the parDE cultivation. operon works in concert with the second par The parDE operon responsible for operon parCBA comes from the comparison post-segregational killing of plasmid-free cells with the resolution of dimer chromosomes is located inside the par locus together with coupled to cell division (Steiner & Kuempel, the parCBA operon encoding a site-specific 1998). recombination system (Grinter et al., 1989; Roberts et al., 1990; Roberts & Helinski, pas locus of plasmid pTF-FC2 1994). Expression of the ParE protein in the ab- An unusual proteic plasmid addiction sys- sence of ParD leads to growth inhibition and tem has been described for the broad host filamentation of the cells. An uncharacterized range plasmid pTF-FC2 from the biomining region psa has been postulated to be responsi- bacterium Thiobacillus ferrooxidans. ble for growth arrest not involving filamen- The pas (plasmid addiction system) locus tation (Jovanovic et al., 1994). The cellular tar- consists of three genes: pasA, B and C. PasA get of the toxin is not known. The effect of the encodes an antitoxin, pasB a toxin, and pasC a functioning of the parDE operon on bacterial protein which enhances the ability of the PasA growth and morphology is dependent on the antidote to neutralise the toxic effect of PasB. host strain and the type of medium in a man- Nevertheless, the pair PasA and PasB is ner which is not well understood (Roberts et equally efficient in plasmid stabilization in the al., 1994). presence or absence of PasC (Smith & Both proteins, ParD and ParE, exist as Rawlings, 1997). dimers in solution and form tetrameric com- The proteins PasA and PasB, but not PasC, plexes in vitro (Johnson et al., 1996). The anti- are required for the full repression of the pas dote ParD negatively regulates transcription operon (Smith & Rawlings, 1998b). Precise of the parDE operon (Roberts et al., 1993). autoregulation of the pas operon is very im- In its natural context, the expression of the portant for its functioning — placing the pas parDE operon might be subject to global regu- genes under a heterologous promoter results lation by the KorA and KorB proteins in a complete loss of the test plasmid stability (Pansegrau et al., 1994). Perhaps this is the (Smith & Rawlings, 1998b). reason that, unexpected for proteic killer sys- The efficiency of the pTF-FC2 pas poi- tem, when tested in a control system (Jensen son-antidote stability system in E. coli is et al., 1995) containing the parDE operon greatly affected by the host strain: from only, this cassette exerts a 100-fold stabilizing 100-fold in CSH50 to no increase in JM109. effectiveness. The sole effect of stabilization The Lon protease is responsible for degrada- 1014 U. Zielenkiewicz and P. Ceg³owski 2001 tion of the antidote PasA (Smith & Rawlings, The locus relBE, at the moment of discovery 1998a). assigned only to pathogenic bacteria, now Although the specific action of the PasB seems unexpectedly widely distributed among toxin is still unknown and the stabilization ef- prokaryotic chromosomes. ficiency is modest, the pasABC system un- doubtedly falls into the bactericidal category hig locus of plasmid Rts1 as shown in in vivo experiments (Smith & Rawlings, 1997). Compared with other proteic killer genes, The pTF-FC2 PasA antitoxin is poorly but the hig (host inhibition of growth) locus of clearly related (31% aa identity) to the ParD plasmid Rts1 is unique in that the toxic part antitoxin of the parDE system of RK2 higB lies upstream of the antidote gene higA. plasmid. The protein products of the two overlapping genes have been identified: the 92aa killer fac- tor of higB and the 104aa antitoxin of higA. relBE/stbDE locus of plasmids P307/R485 HigA suppresses the HigB function both in cis Recently discovered on enteropathogenic and in trans. Two promoters were shown to be plasmid P307 from E. coli, the stability system active in the hig operon: a stronger one up- relBE shares about 50% sequence identity stream of higB, negatively regulated by the with stbDE genes on R485 plasmid from HigA antidote, and a weaker but constitutive Morganella morganii (Grønlund & Gerdes, one upstream of higA, within the higB coding 1999; Hayes, 1998). region (Tian et al., 1996b). Its structural organization is canonical for The cellular target of the action of HigB re- the toxin-antitoxin genes family with the mains unknown. autoregulatory relB gene encoding an anti- dote and the overlapping relE gene encoding w-e-z locus of plasmid pSM19035 a toxin. When overproduced, the relE gene causes immediate decline in viable counts The w-e-z operon of the low-copy-number, over 5 orders of magnitude, indicating that broad-host-range plasmid pSM19035 from RelE is a very potent toxin. The killing by RelE Streptococcus pyogenes encodes the only can be counteracted by the induction of RelB proteic addiction system discovered in Gram expression. The protease responsible for RelB positive bacteria. It consists of three genes: w, degradation is Lon. The relBE operon is able e and z. Two of them, e and z, encode PPAS to stabilize a mini-P307 replicon. with Zeta (the product of z) acting as a toxin Homologues of the relBE/stbDE operon have and Epsilon (the product of e) as an antidote. been found on Gram-negative bacterial chro- This system is unique in that neither the anti- mosomes. The relBEF operon of E. coli chro- dote nor the antidote–toxin complex are in- mosome was found to be affected in stringent volved in regulation of their own synthesis. control during amino-acid starvation (so The autoregulatory functions are exclusively called “delayed relaxed response”, Lavallé, played here by a third component of the 1965). RelB mutants had an increased level of operon, w gene. It has been shown that w is relBEF mRNA and the growth recovered very also implicated in regulation of expression of slowly after the termination of amino acid two other pSM19035 genes: d and copS. Delta starvation (Bech et al., 1985). In this context belongs to a family of ATPases involved in ac- the proposed action for the RelE toxin is inhi- tive partitioning and CopS regulates plasmid bition of translation (Gotfredsen & Gerdes, copy number. Hence, Omega-repressor links 1998). random (plasmid-copy-number control) and Vol. 48 Plasmid stable maintenance 1015 better-than-random (PPAS, putative partition- teins interact directly. Expression of mazEF is ing) inheritance of pSM19035 (de la Hoz et al., regulated by the cellular level of ppGpp, prod- 2000). uct of the activity of the protein encoded by The w-e-z operon was shown to stabilize relA gene. As shown in recent papers (Sat et heterologous plasmids in Gram-positive bacte- al., 2001; Hazan et al., 2001) inhibition of rium B. subtilis and less efficiently in Gram- transcription and/or translation caused by negative E. coli. In both organisms the toxic stress conditions (starvation, action of some effects of the excess of protein Zeta were coun- antibiotics) or the activity of another PPAS, teracted by the proper expression of the gene phd-doc, triggers mazEF lethality by prevent- e.InB. subtilis the activity of protein Zeta was ing de novo synthesis of the short-lived anti- bactericidal whereas in E. coli it was mainly toxin. bacteriostatic (Zielenkiewicz et al., 2000). The Irrespective of only partial amino-acid se- interaction between Epsilon and Zeta pro- quence identity, the ChpAI (MazE) or ChpBI teins was detected in vivo in yeast two-hybrid (the second homologue of PemI) antitoxin can system (Gerdes et al., 2000). It is important to functionally interact with the PemK/Kid note the broad spectrum of the toxicity of pro- toxin (Santos-Sierra et al., 1997). tein Zeta: it acts against Gram-positive and Recently, a new widely distributed family of Gram-negative bacteria as well as eukaryotic genes, homologous to E. coli relBE has been S. cerevisiae (Zielenkiewicz, unpublished). classified as toxin-antitoxin modules (Got- The cellular target of the toxin Zeta remains fredsen & Gerdes, 1998). Again, by use of a unknown. plasmid model, it was shown that relB speci- fies an antitoxin that prevents the lethal ac- tion of relE-encoded toxin; the protein RelB CHROMOSOMALLY LOCATED autoregulates the relBEF operon at the level TOXIN-ANTITOXIN GENES of transcription and the genes relBE stabilize a mini-R1 test plasmid. The third gene of this Many of the addiction genes found in operon, relF, potentially codes for a toxin be- plasmids have homologues on bacterial chro- longing to the Hok family (Gerdes et al., mosomes. Systematic inspection of data-bases 1986), although the cistron is not translated. reveals that they are present in every system- Additionally, the relBE genes are clearly ho- atic group of Prokaryota, including Archaea. mologous to many operons located on differ- Moreover, several paralogues of toxin-anti- ent plasmids, some of them with a well docu- toxin genes are frequently repeated on the mented function of an addiction system (e.g. same chromosome (Gerdes, 2000). P307; Grønlund & Gerdes, 1999, R485; When cloned into plasmids, these genes are Hayes, 1998, pTF-FC2; Smith & Rawlings, able to stabilize an unstable replicon by 1997). post-segregational killing. For example, one of the two E. coli homologues of the pem locus, called chpA (or mazEF) located in the rel BACTERIOCINS AND OTHER OUT operon was tested in detail for its killer prop- ACTING AGENTS ENCODED BY erties (Aizenman et al., 1996). After cloning PLASMID on a plasmid it was clearly demonstrated that these genes possess all of the properties re- Antagonistic interactions between related quired for an addiction module. MazE is a la- microorganisms have been noticed since the bile protein degraded by the ClpAP protease very beginning of bacteriological studies. In and protects the bacterial cell from the toxic time, substances produced by bacteria that in- effect of the stable MazF protein. Both pro- hibit the growth of other vicinal bacteria from 1016 U. Zielenkiewicz and P. Ceg³owski 2001 closely related species received the name at specific sequences and will be cleaved by bacteriocin. the restriction enzyme present in the cell. But, Production of many bacteriocins is encoded this “cellular defense” hypothesis can hardly on plasmids together with correspondent explain the exceptional diversity and specific- gene(s) specifying the immunity to them. The ity of recognition sequences. For some “rare toxic protein is secreted outside the cell, ad- cutter” restriction enzymes it is difficult to sorbs to specific outer membrane receptors of find foreign DNA containing such sequences. the sensitive cell and, consequently, inhibits or It is also true that the R–M pairs are not eas- kills this cell. The bacteriocin producing cells ily lost from their host cell. Those of the de- are resistant to their own toxin thanks to the scendant cells that lose an RM gene cluster immunity gene present on a plasmid and ex- are unable to modify a sufficient number of pressed at the same time with the gene for the recognition sequences in their chromosome to toxin. The detailed information on the function protect them from the lethal attack by the re- of bacteriocins can be found in recent reviews maining restriction enzyme. (Jack et al., 1995; Jack & Jung, 2000). When present or moved on a plasmid, R–M The mechanism of stabilization achieved by gene pairs can increase apparent plasmid sta- bacteriocin killing is only superficially similar bility in a post-segregational host killing man- to post-segregational stabilization by addic- ner. Naito et al. (1995) have demonstrated that tion systems. Whereas plasmid addiction sys- plasmids carrying EcoRI or PaeR7 R–M gene tems do not permit the progeny without a pairs (r+m+) are considerably more stable than plasmid to arise, plasmids producing toxins the vector alone or plasmids with the modifica- acting from outside protect plasmid-bearing tion gene only (r–m+). When the replication of cells from competition by eliminating incom- r+m+ plasmids was temperature arrested, the ing plasmid-free cells. number of viable cells ceased to increase as To some extent, this is similar to many other well as the number of plasmid carrying cells. phenomena like antibiotic production, patho- The proposed explanation for the cell death af- genicity, yeast killer viruses and the Kluyver- ter r+m+ plasmids loss is the cleavage of the in- omyces killer plasmids. sufficiently methylated host chromosome. It was confirmed by direct observation of chro- mosomal DNA degradation in cured cells. RESTRICTION-MODIFICATION It has been experimentally demonstrated SYSTEMS (R–M) AS STABILITY that two R–M systems of the same sequence DETERMINANTS specificity cannot simultaneously force their maintenance on a host (Kusano et al., 1995). A restriction endonuclease (R) recognizes a That could be the basis of the quick evolution- specific short sequence on duplex DNA and ary diversification of the R–M genes. makes double-strand break within or close to From this perspective the restriction en- it, unless the modification enzyme (M) zyme and its cognate methylase are like a methylates this sequence protecting it from toxin-antidote pair. The only difference is that the cleavage. R–M gene complexes are abun- the restriction enzyme is not interacting with dant in Prokaryota, including Archaea, pres- the methylase directly. However, as the ent equally on chromosomes and plasmids. methylase is protecting the cellular target One bacterial cell can contain multiple R–M from the restriction enzyme attack, the latest gene pairs. effect of the R–M pair presence is apparent R–M systems are believed to have evolved to stabilization of plasmid carrying them. protect cells from foreign DNA like viruses or All these observations allow the R–M genes plasmids. The incoming DNA is not modified to be seen as selfish elements, evolved and Vol. 48 Plasmid stable maintenance 1017 fixed as programmed cell death modules, for- self parasites, suicide cassettes for pro- mally similar to plasmid addiction systems grammed bacterial cell death, elements of (Yarmolinsky, 1995; Kobayashi, 1998). stress response or even beneficial in genetic diversification is still to judge.

PPAS EVOLUTION REFERENCES The origin of potentially lethal genes re- mains open to speculations. As the addiction Abeles, A.L., Friedman, S.A. & Austin, S.J. (1985) cassettes are so frequent in prokaryotic chro- Partition of unit-copy miniplasmids to daugh- mosomes and plasmids, the hypothesis of the ter cells. III. The DNA sequence and func- concentration of multiple and independent tional organization of the P1 partition region. 185 smaller replicons to give a new bigger plasmid J. Mol. Biol. , 261–272. or even a bacterial chromosome has been pro- Aizenman, E., Engelberg-Kulka, H. & Glaser, G. posed. This is supported by an observation (1996) An chromosomal “ad- that the presence of multiple stability determi- diction module” regulated by guanosine ¢ ¢ nants on a given plasmid is often paralleled by 3 ,5 -bispyrophosphate: A model for pro- multiple replicons (e.g. plasmids F and RK2). grammed bacterial cell death. Proc. Natl. Acad. Sci. U.S.A 93, 6059–6063. Another hypothesis is proposed basing on sequence comparison among dispersed homo- Alonso, J.C., Weise, F. & Rojo, F. (1995) The Bacil- logues (Grønlund & Gerdes, 1999) assumed lus subtilis histone-like protein Hbsu is re- that ubiquity of relBE genes is likely due to quired for DNA resolution and DNA inversion vertical transmission of an ancestor that have mediated by the beta recombinase of plasmid pSM19035. J. Biol. Chem. 270, 2938–2945. appeared early in evolution before the branch- ing of Bacteria and Archaea. Austin, S., Ziese, M. & Sternberg, N. (1981) A Toxin-antitoxin cassettes present on chro- novel role for site-specific recombination in mosomes could have become the source of maintenance of bacterial replicons. Cell 25, genes for plasmid addiction systems (Raw- 729–736. lings, 1999). Once captured on a plasmid Bahassi, E.M., Salmon, M.A., Van Melderen, L., these genes would be infinitely maintained in Bernard, P. & Couturier, M. (1995) F plasmid population owing the ability to kill plasmid- CcdB killer protein: ccdB gene mutants coding free cells. At the same time the diversification for non-cytotoxic proteins which retain their and horizontal transfer of the toxin-antitoxin regulatory functions. Mol. Microbiol. 15, genes could have occurred. The opposite di- 1031–1037. rection of moving genes (i.e. from plasmid to Bahassi, E.M., O’Dea, M.H., Allali, N., Messens, the chromosome) cannot be yet ruled out. J., Gellert, M. & Couturier, M. (1999) Interac- The abundance of killing genes on tions of CcdB with DNA gyrase. Inactivation prokaryotic chromosomes also suggests their of GyrA, poisoning of the gyrase-DNA com- different that stabilizing functions. Since plex, and the antidote action of CcdA. J. Biol. Chem. 274, 10936–10944. some of these genes (e.g. relBE and chpA) are linked to the stringent control and their acti- Bech, F.W., Jorgensen, S.T., Diderichsen, B. & vation reduces the rate of translation (RelE) Karlstrom, O.H. (1985) Sequence of the relB or replication (ChpA), it is now proposed that transcription unit from Escherichia coli and the toxin-antitoxin loci constitute a part of the identification of the relB gene. EMBO J. 4, global stress response (Gerdes, 2000). 1059–1066. The question whether toxin-antitoxin loci, Berger, J.M., Gamblin, S.J., Harrison, S.C. & plasmidic or chromosomal, are a kind of DNA Wang, J.C. (1996) Structure and mechanism 1018 U. Zielenkiewicz and P. Ceg³owski 2001

of DNA topoisomerase II. Nature 379, to daughter cells. Mol. Microbiol. 21, 225–232. 1029–1036. Bernard, P. & Couturier, M. (1992) Cell killing by de Feyter, R., Wallace, C. & Lane, D. (1989) the F plasmid CcdB protein involves poisoning Autoregulation of the ccd operon in the F of DNA- topoisomerase II complexes. J. Mol. plasmid. Mol. Gen. Genet. 218, 481–486. Biol. 226, 735–745. de la Hoz, A.B., Ayora, S., Sitkiewicz, I., Bernard, P., Kezdy, K.E., Van Melderen, L., Fernandez, S., Pankiewicz, R., Alonso, J.C. & Steyaert, J., Wyns, L., Pato, M.L., Higgins, Ceg³owski, P. (2000) Plasmid copy-number P.N. & Couturier, M. (1993) The F plasmid control and better-than-random segregation CcdB protein induces efficient ATP-dependent genes of pSM19035 share a common regula- DNA cleavage by gyrase. J. Mol. Biol. 234, tor. Proc. Natl. Acad. Sci. U.S.A97, 728–733. 534–541. Easter, C.L., Sobecky, P.A. & Helinski, D.R. (1997) Bex, F., Karoui, H., Rokeach, L., Dreze, P., Garcia, Contribution of different segments of the par L. & Couturier, M. (1983) Mini-F encoded pro- region to stable maintenance of the broad- teins: identification of a new 10.5 kilodalton host-range plasmid RK2. J. Bacteriol. 179, species. EMBO J. 2, 1853–1861. 6472–6479. Bignell, C.R., Haines, A.S., Khare, D. & Thomas, Erdmann, N., Petroff, T. & Funnell, B.E. (1999) C.M. (1999) Effect of growth rate and incC mu- Intracellular localization of P1 ParB protein tation on symmetric plasmid distribution by depends on ParA and parS. Proc. Natl. Acad. the IncP-1 partitioning apparatus. Mol. Sci. U.S.A 96, 14905–14910. Microbiol. 34, 205–216. Figurski, D.H., Pohlman, R.F., Bechhofer, D.H., Boe, L., Gerdes, K. & Molin, S. (1987) Effects of Prince, A.S. & Kelton, C.A. (1982) Broad host genes exerting growth inhibition and plasmid range plasmid RK2 encodes multiple kil genes stability on plasmid maintenance. J. Bacteriol. potentially lethal to Escherichia coli host cells. 169, 4646–4650. Proc. Natl. Acad. Sci. U.S.A 79, 1935–1939. Bravo, A., de Torrontegui, G. & Diaz, R. (1987) Fraser, C.M., Casjens, S., Huang, W.M., Sutton, Identification of components of a new stability G.G., Clayton, R., Lathigra, R., White, O., system of plasmid R1, ParD, that is close to Ketchum, K.A., Dodson, R., Hickey, E.K., the origin of replication of this plasmid. Mol. Gwinn, M., Dougherty, B., Tomb, J.F., Gen. Genet. 210, 101–110. Fleischmann, R.D., Richardson, D., Peterson, J., Kerlavage, A.R., Quackenbush, J., Salz- Carlson, C.R. & Kolsto, A.B. (1994) A small (2.4 berg, S., Hanson, M., van Vugt, R., Palmer, N., Mb) Bacillus cereus chromosome corresponds Adams, M.D., Gocayne, J. & Venter, J.C. to a conserved region of a larger (5.3 Mb) Ba- (1997) Genomic sequence of a Lyme disease cillus cereus chromosome. Mol. Microbiol. 13, spirochaete, Borrelia burgdorferi. Nature 390, 161–169. 580–586. Couturier, M., Bahassi, E. & Van Melderen, L. Furuya, N. & Komano, T. (1996) Nucleotide se- (1998) Bacterial death by DNA gyrase poison- quence and characterization of the trbABC re- ing. Trends Microbiol. 6, 269–275. gion of the IncI1 Plasmid R64: existence of the Davis, M. A., Martin, K.A. & Austin, S.J. (1992) pnd gene for plasmid maintenance within the Biochemical activities of the parA partition transfer region. J. Bacteriol. 178 , 1491–1497. protein of the P1 plasmid. Mol. Microbiol. 6, Gazit, E. & Sauer, R.T. (1999a) The Doc toxin and 1141–1147. Phd antidote proteins of the bacteriophage P1 Davis, M.A., Radnedge, L., Martin, K.A., Hayes, plasmid addiction system form a hetero- F., Youngren, B. & Austin, S.J. (1996) The P1 trimeric complex. J. Biol. Chem. 274, ParA protein and its ATPase activity play a di- 16813–16818. rect role in the segregation of plasmid copies Vol. 48 Plasmid stable maintenance 1019

Gazit, E. & Sauer, R.T. (1999b) Stability and DNA Plasmids and Gene Spread. (Thomas, C.M., ed) binding of the phd protein of the phage P1 pp. 66–67, Harwood Academic Publishers. plasmid addiction system. J. Biol. Chem. 274, Golub, E.I. & Panzer, H.A. (1988) The F factor of 2652–2657. Escherichia coli carries a locus of stable Gerdes, K., Bech, F.W., Jorgensen, S.T., plasmid inheritance stm, similar to the parB Lobner-Olesen, A., Rasmussen, P.B., Atlung, locus of plasmid RI. Mol. Gen. Genet. 214, T., Boe, L., Karlstrom, O., Molin, S. & von 353–357. Meyenburg, K. (1986) Mechanism of Gordon, G.S., Sitnikov, D., Webb, C.D., Teleman, postsegregational killing by the hok gene prod- A., Straight, A., Losick, R., Murray, A.W. & uct of the parB system of plasmid R1 and its Wright, A. (1997) Chromosome and low copy homology with the relF gene product of the E. plasmid segregation in E. coli: Visual evidence coli relB operon. EMBO J. 5, 2023–2029. for distinct mechanisms. Cell 90, 1113–1121. Gerdes, K., Helin, K., Christensen, O.W. & Gotfredsen, M. & Gerdes, K. (1998) The Esche- Lobner-Olesen, A. (1988) Translational con- richia coli relBE genes belong to a new trol and differential RNA decay are key ele- toxin-antitoxin gene family. Mol. Microbiol. ments regulating postsegregational expres- 29, 1065–1076. sion of the killer protein encoded by the parB locus of plasmid R1. J. Mol. Biol. 203, Greenfield, T.J., Ehli, E., Kirshenmann, T., 119–129. Franch, T., Gerdes, K. & Weaver, K.E. (2000) The antisense RNA of the par locus of pAD1 Gerdes, K., Poulsen, L.K., Thisted, T., Nielsen, regulates the expression of a 33-amino-acid A.K., Martinussen, J. & Andreasen, P.H. toxic peptide by an unusual mechanism. Mol. (1990) The hok killer gene family in Microbiol. 37, 652–660. gram-negative bacteria. New Biol. 2, 946–956. Grinter, N.J., Brewster, G. & Barth, P.T. (1989) Gerdes, K., Nielsen, A., Thorsted, P. & Wagner, Two mechanisms necessary for the stable in- E.G. (1992) Mechanism of killer gene activa- heritance of plasmid RP4. Plasmid 22, tion. Antisense RNA-dependent RNase III 203–214. cleavage ensures rapid turn-over of the stable hok, srnB and pndA effector messenger Grønlund, H. & Gerdes, K. (1999) Toxin-antitoxin RNAs. J. Mol. Biol. 226, 637–649. systems homologous with relBE of Escherichia coli plasmid P307 are ubiquitous in Gerdes, K., Gultyaev, A.P., Franch, T., Pedersen, prokaryotes. J. Mol. Biol. 285, 1401–1415. K. & Mikkelsen, N.D. (1997) Antisense RNA-regulated programmed cell death. Annu. Hayes, F. (1998) A family of stability determinants Rev. Genet. 31, 1–31. in pathogenic bacteria. J. Bacteriol. 180, 6415–6418. Gerdes, K., Moller-Jensen, J. & Bugge, J.R. (2000) Plasmid and chromosome partitioning: Sur- Hayes, F. (2000) The partition system of prises from phylogeny. Mol. Microbiol. 37, multidrug resistance plasmid TP228 includes 455–466. a novel protein that epitomizes an evolutionarily distinct subgroup of the ParA Gerdes, K. (2000) Toxin-antitoxin modules may superfamily. Mol. Microbiol. 37, 528–541. regulate synthesis of macromolecules during nutritional stress. J. Bacteriol. 182, 561–572. Hazan, R., Sat, B., Reches, M. & Engelberg-Kulka, H. (2001) Postsegregational killing mediated Gerdes, K., Aroya, S., Canosa, I., Ceglowski, P., by the P1 phage “Addiction Module“ phd-doc Diaz-Orejas, R., Franch, T., Gultyaev, A.P., requires the Escherichia coli programmed cell Bugge Jensen, R., Kobayashi, I., Macpherson, death system mazEF. J. Bacteriol. 183, C., Summers, D., Thomas, C.M. & Zielen- 2046–2050. kiewicz, U. (2000) A novel proteic plasmid sta- bilization system from Gram-positive bacteria. Hiraga, S. (1992) Chromosome and plasmid parti- in The Horizontal Gene Pool. Bacterial tion in Escherichia coli. Annu. Rev. Biochem. 61, 283–306. 1020 U. Zielenkiewicz and P. Ceg³owski 2001

Honeycutt, R.J., McClelland, M. & Sobral, B.W. Kim, S.K. & Wang, J.C. (1998) Localization of F (1993) Physical map of the genome of Rhizo- plasmid SopB protein to positions near the bium meliloti 1021. J. Bacteriol. 175 , 6945– poles of Escherichia coli cells. Proc. Natl. Acad. 6952. Sci. U.S.A 95, 1523–1527. Jack, R.W., Tagg, J.R. & Ray, B. (1995) Kobayashi, I. (1998) Selfishness and death: Rai- Bacteriocins of gram-positive bacteria. Micro- son d’etre of restriction, recombination and biol. Rev. 59, 171–200. mitochondria. Trends Genet. 14, 368–374. Jack, R.W. & Jung, G. (2000) Lantibiotics and Koyama, A.H., Wada, C., Nagata, T. & Yura, T. microcins: Polypeptides with unusual chemi- (1975) Indirect selection for plasmid mutants: cal diversity. Curr. Opin. Chem. Biol. 4, 310– Isolation of ColVBtrp mutants defective in 317. self-maintenance in Escherichia coli. J. Bacteriol. 122, 73–79. Jaffe, A., Ogura, T. & Hiraga, S. (1985) Effects of the ccd function of the F plasmid on bacterial Kusano, K., Naito, T., Handa, N. & Kobayashi, I. growth. J. Bacteriol. 163, 841–849. (1995) Restriction-modification systems as genomic parasites in competition for specific Janniere, L., McGovern, S., Pujol, C., Petit, M.A. & sequences. Proc. Natl. Acad. Sci. U.S.A 92, Ehrlich, S.D. (1996) In vivo analysis of the 11095–11099. plasmid pAM beta 1 resolution system. Nucleic Acids Res. 24, 3431–3436. Kusukawa, N., Mori, H., Kondo, A. & Hiraga, S. (1987) Partitioning of the F plasmid: Overpro- Jensen, R.B. & Gerdes, K. (1995) Programmed duction of an essential protein for partition in- cell death in bacteria: proteic plasmid stabili- hibits plasmid maintenance. Mol. Gen. Genet. zation systems. Mol. Microbiol. 17, 205–210. 208, 365–372. Jensen, R.B., Grohmann, E., Schwab, H., Lane, D., de Feyter, R., Kennedy, M., Phua, S.H. & Diaz-Orejas, R. & Gerdes, K. (1995) Compari- Semon, D. (1986) D protein of miniF plasmid son of ccd of F, parDE of RP4, and parD of R1 acts as a repressor of transcription and as a using a novel conditional replication control site-specific resolvase. Nucleic Acids Res. 14, system of plasmid R1. Mol. Microbiol. 17, 9713–9728. 211–220. Lavalle, R. (1965) New mutants for regulation of Jensen, R.B. & Gerdes, K. (1999) Mechanism of RNA synthesis. Bull. Soc. Chim. Biol. (Paris) DNA segregation in prokaryotes: ParM parti- 47, 1567–1570. tioning protein of plasmid R1 co-localizes with its replicon during the cell cycle. EMBO J. 18, Lehnherr, H., Maguin, E., Jafri, S. & Yarmolinsky, 4076–4084. M.B. (1993) Plasmid addiction genes of bacteriophage P1: doc, which causes cell death Johnson, E.P., Strom, A.R. & Helinski, D.R. (1996) on curing of prophage, and phd, which pre- Plasmid RK2 toxin protein ParE: Purification vents host death when prophage is retained. J. and interaction with the ParD antitoxin pro - Mol. Biol. 233, 414–428. tein. J. Bacteriol. 178, 1420–1429. Lehnherr, H. & Yarmolinsky, M.B. (1995) Addic- Jovanovic, O.S., Ayres, E.K. & Figurski, D.H. tion protein Phd of plasmid prophage P1 is a (1994) Host-inhibitory functions encoded by substrate of the ClpXP serine protease of Esch- promiscuous plasmids. Transient arrest of erichia coli. Proc. Natl. Acad. Sci. U.S.A 92, Escherichia coli segregants that fail to inherit 3274–3277. plasmid RK2. J. Mol. Biol. 237, 52–64. £obocka, M. & Yarmolinsky, M. (1996) P1 plasmid Karoui, H., Bex, F., Dreze, P. & Couturier, M. partition: A mutational analysis of ParB. J. (1983) Ham22, a mini-F mutation which is le - Mol. Biol. 259, 366–382. thal to host cell and promotes recA-dependent induction of lambdoid prophage. EMBO J. 2, Loh, S.M., Cram, D.S. & Skurray, R. A. (1988) Nu- 1863–1868. cleotide sequence and transcriptional analysis Vol. 48 Plasmid stable maintenance 1021

of a third function (Flm) involved in F-plasmid range plasmid RK2 are coregulated with vege- maintenance. Gene 66, 259–268. tative replication. Mol. Microbiol. 6, 907–920. Loris, R., Dao-Thi, M.H., Bahassi, E.M., Van Naito, T., Kusano, K. & Kobayashi, I. (1995) Self- Melderen, L., Poortmans, F., Liddington, R., ish behavior of restriction-modification sys- Couturier, M. & Wyns, L. (1999) Crystal struc- tems. Science 267, 897–899. ture of CcdB, a topoisomerase poison from E. Nielsen, A.K., Thorsted, P., Thisted, T., Wagner, coli. J. Mol. Biol. 285, 1667–1677. E.G. & Gerdes, K. (1991) The Lynch, A.S. & Wang, J.C. (1995) SopB pro- rifampicin-inducible genes srnB from F and tein-mediated silencing of genes linked to the pnd from R483 are regulated by antisense sopC locus of Escherichia coli F plasmid. Proc. RNAs and mediate plasmid maintenance by Natl. Acad. Sci. U.S.A 92, 1896–1900. killing of plasmid-free segregants. Mol. Micro- biol. 5, 1961–1973. Magnuson, R. & Yarmolinsky, M.B. (1998) Corepression of the P1 addiction operon by Nordstrom, K. & Austin, S.J. (1989) Mechanisms Phd and Doc. J. Bacteriol. 180, 6342–6351. that contribute to the stable segregation of plasmids. Annu. Rev. Genet. 23, 37–69. Maki, S., Takiguchi, S., Horiuchi, T., Sekimizu, K. & Miki, T. (1996) Partner switching mecha- Ogura, T. & Hiraga, S. (1983) Mini-F plasmid nisms in inactivation and rejuvenation of genes that couple host cell division to plasmid Escherichia coli DNA gyrase by F plasmid pro- proliferation. Proc. Natl. Acad. Sci. U.S.A 80, teins LetD (CcdB) and LetA (CcdA). J. Mol. 4784–4788. Biol. 256, 473–482. Ohnishi, Y., Iguma, H., Ono, T., Nagaishi, H. & Miki, T., Chang, Z.T. & Horiuchi, T. (1984) Control Clark, A.J. (1977) Genetic mapping of the F of cell division by sex factor F in Escherichia plasmid gene that promotes degradation of coli. II. Identification of genes for inhibitor stable ribonucleic acid in Escherichia coli. J. protein and trigger protein on the 42.84–43.6 Bacteriol. 132, 784–789. F segment. J. Mol. Biol. 174, 627–646. Ono, K., Akimoto, S. & Ohnishi, Y. (1987) Nucleo- Miki, T., Park, J.A., Nagao, K., Murayama, N. & tide sequence of the pnd gene in plasmid R483 Horiuchi, T. (1992) Control of segregation of and role of the pnd gene product in chromosomal DNA by sex factor F in Esche- plasmolysis. Microbiol. Immunol. 31, richia coli. Mutants of DNA gyrase subunit A 1071–1083. suppress letD (ccdB) product growth inhibi- Ono, K., Akimoto, S., Ono, T. & Ohnishi, Y. (1986) tion. J. Mol. Biol. 225, 39–52. Plasmid genes increase membrane permeabil- Mohl, D.A. & Gober, J.W. (1997) Cell cy- ity in Escherichia coli. Biochim. Biophys. Acta cle-dependent polar localization of chromo- 867, 81–88. some partitioning proteins in Caulobacter Pansegrau, W., Lanka, E., Barth, P.T., Figurski, crescentus. Cell 88, 675–684. D.H., Guiney, D.G., Haas, D., Helinski, D.R., Mori, H., Mori, Y., Ichinose, C., Niki, H., Ogura, T., Schwab, H., Stanisich, V.A. & Thomas, C M. Kato, A. & Hiraga, S. (1989) Purification and (1994) Complete nucleotide sequence of Bir- characterization of SopA and SopB proteins mingham IncP alpha plasmids. Compilation essential for F plasmid partitioning. J. Biol. and comparative analysis. J. Mol. Biol. 239, Chem. 264, 15535–15541. 623–663. Motallebi-Veshareh, M., Rouch, D.A. & Thomas, Pedersen, K. & Gerdes, K. (1999) Multiple hok C.M. (1990) A family of ATPases involved in genes on the chromosome of Escherichia coli. active partitioning of diverse bacterial Mol. Microbiol. 32, 1090–1102. plasmids. Mol. Microbiol. 4, 1455–1463. Poulsen, L.K., Larsen, N.W., Molin, S. & Motallebi-Veshareh, M., Balzer, D., Lanka, E., Andersson, P. (1989) A family of genes encod- Jagura-Burdzy, G. & Thomas, C.M. (1992) ing a cell-killing function may be conserved in Conjugative transfer functions of broad-host- 1022 U. Zielenkiewicz and P. Ceg³owski 2001

all gram-negative bacteria. Mol. Microbiol. 3, Sakikawa, T., Akimoto, S. & Ohnishi, Y. (1989) 1463–1472. The pnd gene in E. coli plasmid R16: Nucleo- tide sequence and gene expression leading to Roberts, R.C., Burioni, R. & Helinski, D.R. (1990) cell Mg2+ release and stable RNA degradation. Genetic characterization of the stabilizing Biochim. Biophys. Acta 1007, 158–166. functions of a region of broad-host-range plasmid RK2. J. Bacteriol. 172, 6204–6216. Salmon, M.A., Van Melderen, L., Bernard, P. & Couturier, M. (1994) The antidote and Roberts, R.C. & Helinski, D.R. (1992) Definition of autoregulatory functions of the F plasmid a minimal plasmid stabilization system from CcdA protein: A genetic and biochemical sur- the broad- host-range plasmid RK2. J. vey. Mol. Gen. Genet. 244, 530–538. Bacteriol. 174, 8119–8132. Santos-Sierra, S., Giraldo, R. & Diaz-Orejas, R. Roberts, R.C., Spangler, C. & Helinski, D.R. (1993) (1997) Functional interactions between ho- Characteristics and significance of DNA bind- mologous conditional killer systems of ing activity of plasmid stabilization protein plasmid and chromosomal origin. FEMS ParD from the broad-host-range plasmid RK2. Microbiol. Lett. 152, 51–56. J. Biol. Chem. 268, 27109–27117. Sat, B., Hazan, R., Fisher, T., Khaner, H., Glaser, Roberts, R.C., Strom, A.R. & Helinski, D.R. (1994) G. & Engelberg-Kulka, H. (2001) Programmed The parDE operon of the broad-host-range cell death in Escherichia coli: Some antibiotics plasmid RK2 specifies growth inhibition asso - can trigger mazEF lethality. J. Bacteriol. 183, ciated with plasmid loss. J. Mol. Biol. 237, 2041–2045. 35–51. Sharpe, M.E. & Errington, J. (1996) The Bacillus Rodionov, O., Lobocka, M. & Yarmolinsky, M. subtilis soj-spo0J locus is required for a (1999) Silencing of genes flanking the P1 centromere-like function involved in prespore plasmid centromere. Science 283, 546–549. chromosome partitioning. Mol. Microbiol. 21, Rojo, F. & Alonso, J.C. (1994) A novel site-specific 501–509. recombinase encoded by the Streptococcus Sharpe, M.E., Chatwin, H.M., Macpherson, C., pyogenes plasmid pSM19035. J. Mol. Biol. Withers, H.L. & Summers, D.K. (1999) Analy- 238, 159–172. sis of the CoIE1 stability determinant Rcd. Mi- Ruiz-Echevarria, M.J., Berzal-Herranz, A., crobiology 145, 2135–2144. Gerdes, K. & Diaz-Orejas, R. (1991a) The kis Sia, E.A., Roberts, R.C., Easter, C., Helinski, D.R. and kid genes of the parD maintenance system & Figurski, D.H. (1995) Different relative of plasmid R1 form an operon that is importances of the par operons and the effect autoregulated at the level of transcription by of conjugal transfer on the maintenance of in- the co- ordinated action of the Kis and Kid pro- tact promiscuous plasmid RK2. J. Bacteriol. teins. Mol. Microbiol. 5, 2685–2693. 177, 2789–2797. Ruiz-Echevarria, M.J., de Torrontegui, G., Smith, A.S. & Rawlings, D.E. (1997) The poi- Gimenez-Gallego, G. & Diaz-Orejas, R. (1991b) son-antidote stability system of the broad- Structural and functional comparison be- host-range Thiobacillus ferrooxidans plasmid tween the stability systems ParD of plasmid pTF-FC2. Mol. Microbiol. 26, 961–970. R1 and Ccd of plasmid F. Mol. Gen. Genet. 225, 355–362. Smith, A.S. & Rawlings, D.E. (1998a) Auto- regulation of the pTF-FC2 proteic poison-anti- Ruiz-Echevarria, M.J., Gimenez-Gallego, G., dote plasmid addiction system (pas) is essen- Sabariegos-Jareno, R. & Diaz-Orejas, R. (1995) tial for plasmid stabilization. J. Bacteriol. 180, Kid, a small protein of the parD stability sys- 5463–5465. tem of plasmid R1, is an inhibitor of DNA rep- lication acting at the initiation of DNA synthe- Smith, A.S. & Rawlings, D.E. (1998b) Efficiency of sis. J. Mol. Biol. 247, 568–577. the pTF-FC2 pas poison-antidote stability sys- tem in Escherichia coli is affected by the host Vol. 48 Plasmid stable maintenance 1023

strain, and antidote degradation requires the R100 in various Escherichia coli hosts. Mol. lon protease. J. Bacteriol. 180, 5458–5462. Gen. Genet. 215, 463–468. Steiner, W.W. & Kuempel, P.L. (1998) Cell division Tsuchimoto, S., Nishimura, Y. & Ohtsubo, E. is required for resolution of dimer chromo- (1992) The stable maintenance system pem of somes at the dif locus of Escherichia coli. Mol. plasmid R100: Degradation of PemI protein Microbiol. 27, 257–268. may allow PemK protein to inhibit cell growth. J. Bacteriol. 174, 4205–4211. Summers, D. (1998) Timing, self-control and a sense of direction are the secrets of multicopy Van Melderen, L., Bernard, P. & Couturier, M. plasmid stability. Mol. Microbiol. 29, (1994) Lon-dependent proteolysis of CcdA is 1137–1145. the key control for activation of CcdB in plasmid-free segregant bacteria. Mol. Summers, D.K. & Sherratt, D.J. (1984) Multi- Microbiol. 11, 1151–1157. merization of high copy number plasmids causes instability: CoIE1 encodes a determi- Watanabe, E., Inamoto, S., Lee, M.H., Kim, S.U., nant essential for plasmid monomerization Ogua, T., Mori, H., Hiraga, S., Yamasaki, M. & and stability. Cell 36, 1097–1103. Nagai, K. (1989) Purification and character- ization of the sopB gene product which is re- Tam, J.E. & Kline, B.C. (1989) The F plasmid ccd sponsible for stable maintenance of mini-F autorepressor is a complex of CcdA and CcdB plasmid. Mol. Gen. Genet. 218, 431–436. proteins. Mol. Gen. Genet. 219, 26–32. Weaver, K.E. & Clewell, D.B. (1989) Construction Thisted, T., Sorensen, N.S., Wagner, E.G. & of Enterococcus faecalis pAD1 miniplasmids: Gerdes, K. (1994) Mechanism of post-se- Identification of a minimal pheromone re- gregational killing: Sok antisense RNA inter- sponse regulatory region and evaluation of a acts with Hok mRNA via its 5¢-end sin- novel pheromone-dependent growth inhibi- gle-stranded leader and competes with the tion. Plasmid 22, 106–119. 3¢-end of Hok mRNA for binding to the mok translational initiation region. EMBO J. 13, Weaver, K.E., Jensen, K.D., Colwell, A. & Sriram, 1960–1968. S.I. (1996) Functional analysis of the Enterococcus faecalis plasmid pAD1-encoded Thisted, T., Nielsen, A.K. & Gerdes, K. (1994) stability determinant par. Mol. Microbiol. 20, Mechanism of post-segregational killing: 53–63. Translation of Hok, SrnB and Pnd mRNAs of plasmids R1, F and R483 is activated by 3’-end Wu, K., Jahng, D. & Wood, T.K. (1994) Tempera- processing. EMBO J. 13, 1950–1959. ture and growth rate effects on the hok/sok killer locus for enhanced plasmid stability. Tian, Q.B., Hayashi, T., Murata, T. & Terawaki, Y. Biotechnol. Prog. 10, 621–629. (1996b) Gene product identification and pro- moter analysis of hig locus of plasmid Rts1. Yamaichi, Y., Iida, T., Park, K.S., Yamamoto, K. & Biochem. Biophys. Res. Commun. 225, Honda, T. (1999) Physical and genetic map of 679–684. the genome of Vibrio parahaemolyticus: Pres- ence of two chromosomes in Vibrio species. Tian, Q.B., Ohnishi, M., Tabuchi, A. & Terawaki, Mol. Microbiol. 31, 1513–1521. Y. (1996a) A new plasmid-encoded proteic killer gene system: Cloning, sequencing, and Yarmolinsky, M.B. (1995) Programmed cell death analyzing hig locus of plasmid Rts1. Biochem. in bacterial populations. Science 267, Biophys. Res. Commun. 220, 280–284. 836–837. Tsuchimoto, S., Ohtsubo, H. & Ohtsubo, E. (1988) Zielenkiewicz, U., Sitkiewicz, I., Kern-Zdanowicz, Two genes, pemK and pemI, responsible for I. & Ceg³owski, P. (2000) Activity of a post- stable maintenance of resistance plasmid segregational killing system of the plasmid R100. J. Bacteriol. 170, 1461–1466. pSM19035 in Bacillus subtilis cells. Plasmid 45, 156. Tsuchimoto, S. & Ohtsubo, E. (1989) Effect of the pem system on stable maintenance of plasmid