Gyrase Inhibitors and Thymine Starvation Disrupt the Normal Pattern of Plasmid RK2 Localization in Escherichia coli

Multicopy plasmids in Escherichia coli are not randomly distributedthroughout the cell but exist as defined clusters that are localizedat the mid-cell, or at the 1/4 and 3/4 cell length positions.To explore the factors that contribute to plasmid clusteringand localization, E. coli cells carrying a plasmid RK2 derivativethat can be tagged with a green fluorescent protein-LacI fusionprotein were subjected to various conditions that interferewith plasmid superhelicity and/or DNA replication. The varioustreatments included thymine starvation and the addition of thegyrase inhibitors nalidixic acid and novobiocin. In each case,localization of plasmid clusters at the preferred positionswas disrupted but the plasmids remained in clusters, suggestingthat normal plasmid superhelicity and DNA synthesis in elongatingcells are not required for the clustering of individual plasmidmolecules. It was also observed that the inhibition of DNA replicationby these treatments produced filaments in which the plasmidclusters were confined to one or two nucleoid bodies, whichwere located near the midline of the filament and were not evenlyspaced throughout the filament, as is found in cells treatedwith cephalexin. Finally, the enhanced yellow fluorescent protein-RarAfusion protein was used to localize the replication complexin individual E. coli cells. Novobiocin and nalidixic acid treatmentboth resulted in rapid loss of RarA foci. Under these conditionsthe RK2 plasmid clusters were not disassembled, suggesting thata completely intact replication complex is not required forplasmid clustering.

INTRODUCTION

Top

Introduction

The development of methods for tagging bacterial plasmids withspecific fluorescent fusion proteins or by fluorescence in situhybridization has greatly facilitated the microscopic analysisof the location and movement of plasmids in a bacterial cellduring growth and division. Contrary to the generally acceptedview that plasmids randomly diffuse throughout the cell, ithas now been shown that the low-copy-number plasmids F (16,18, 49), P1 (16, 45), and R1 (31, 63) and the multicopy plasmidsRK2 and pUC (4, 25, 53, 54) are not randomly distributed throughoutout an Escherichia coli cell but are present as clusters atpreferred locations. Using differentially labeled probes andfluorescence in situ hybridization analysis, it has been shownthat except for plasmid R1, which is located at mid-cell orat the cell poles (31), clusters of plasmids F, P1, RK2, andpUC generally are located at the mid-cell position in newbornE. coli cells and at the 1/4 and 3/4 positions in larger cells(16, 25, 44). For these plasmids it has been shown further thatthe movement of newly duplicated plasmid clusters from the mid-cellto the quarter-cell positions occurs with relatively rapid kinetics(16, 18, 25, 44, 49).

Much has yet to be learned about cell- or plasmid-specifiedfactors that are responsible for the localization and movementof plasmid clusters. It has been shown for plasmids F (18, 49),P1 (6, 44), R1 (63), R27 (39), and pB171 (5) that the mutationof plasmid-encoded partition loci perturbs the regular patternof plasmid localization or, in the case of plasmid R1, resultsin an interference with the separation of plasmid pairs andthe movement of plasmids to opposite poles in the cell (31).Interestingly, the ParM component of the R1 partitioning systemforms double helical protofilaments with features similar toF-actin (48, 61). It has been proposed that ParM moves newlyduplicated plasmid R1 molecules by an actin-like insertionalpolymerization mechanism (15).

Virtually nothing is known regarding the factors that are responsiblefor the clustering of individual plasmid molecules. One modelthat has been considered is that the clusters of individualplasmid molecules are tethered to the cellular replisome (53).The finding that DNA replication occurs in stationary factoriesin the regions of the mid-cell and quarter-cell positions inBacillus subtilis (41) and in E. coli (36) at least raises thepossibility that plasmid duplication occurs at the mid-cellposition and that after duplication the replication complexeswith the attached plasmid clusters segregate to quarter-cellpositions prior to the next round of cell division. Using atemperature-sensitive replication initiation protein for theF plasmid, evidence has been obtained in support of duplicationof this plasmid at the mid-cell immediately followed by movementof each of the duplicated plasmids to their respective quarter-cellpositions (51). Consistent with the notion that DNA replicationproteins play a role in the segregation of plasmids, or plasmidclusters, to new fixed positions is the report of Miller andCohen (46) that DNA replication proteins contribute to plasmidpSC101 segregation in addition to their role in replicationof this plasmid. It has been argued, however, from studies withderivatives of plasmid pSC101 that are temperature sensitivefor replication that declustering of plasmids can occur in theabsence of plasmid replication (21, 50). It is also clear thatwhile plasmid-encoded partition proteins play a role in thelocalization of plasmid elements in E. coli, par^– derivativesof plasmid R1 continue to form clusters, indicating that althoughthe par system contributes to localization of plasmid clustersit is not the binding force for clustering of individual plasmidmolecules (64).

A number of different antibacterial agents have been shown todirectly or indirectly inhibit DNA replication and/or cell division,but little is known about the effects of these agents on plasmidlocalization or movement. Cephalexin, which inhibits cell divisionbut not DNA replication, produces filaments with evenly spacednucleoids (30). Interestingly, fluorescence microscopy analysisof the location of tagged plasmids in cephalexin-treated cellshas demonstrated that plasmid clusters are also evenly spacedalong the length of the filaments within the nucleoid bodies(25). Nalidixic acid addition rapidly interferes with DNA replicationin E. coli by causing the covalent attachment of gyrase to theDNA substrate, with the subsequent production of double-strandbreaks in the DNA (12, 22, 23, 58). Novobiocin also inhibitsDNA gyrase (13) but, unlike nalidixic acid, this inhibitiondoes not result in the covalent attachment of gyrase to DNA(43, 57) and, therefore, there is no extensive formation ofdouble-strand DNA breaks. The addition of either nalidixic acidor novobiocin also results in filament formation.

In this work, we have examined the effect of the addition ofnalidixic acid and novobiocin as well as the inhibition of DNAsynthesis by thymine starvation on plasmid RK2 cluster formationand localization in E. coli. Plasmid RK2 is present at fiveto eight copies per chromosome (9, 55, 59, 60). These copiesform one or several clusters in E. coli, Pseudomonas aeruginosa,and Vibrio cholerae, depending on cell size and growth conditions(25). For this analysis, a plasmid RK2 construct was used thatboth expresses the green fluorescent protein (GFP)-LacI fusionprotein and contains an array of lacO operator sequences thatserve as a binding site for GFP-LacI (16, 53). In addition,an enhanced yellow fluorescent protein (EYFP)-RarA fusion proteinwas used to localize DNA replication complexes (38). Evidencehas been obtained for an association of the RarA protein withthe E. coli replication factory (2, 38). Inhibition of DNA synthesisby thymine starvation or the addition of nalidixic acid or novobiocinresults in filament formation. In each case the filaments containextended nucleoids as measured by 4',6'-diamidino-2-phenylindole(DAPI) staining and a dislocation of plasmid clusters that arecontained within the extended nucleoid region. Furthermore,there is a relatively small increase in the number of plasmidfoci, suggesting a partial breakdown of plasmid clusters. Interestingly,both novobiocin and nalidixic acid addition results in a rapidloss of RarA foci without a significant increase in RK2-taggedfoci, indicating that an intact replication complex is not requiredfor plasmid clustering.

MATERIALS AND METHODS

Top

Materials and Methods

Strains, media, and growth conditions. Strains MG1655(Seq) (F^– ${lambda}$ ^– ilvG rfb-50 rph-1; CGSC7740) or DG76 [F^– leuB6(Am) ${lambda}$ ^– thyA47 rpsL153 (strR)deoC3] were grown at 30°C in M63 medium (52) prepared with0.2% glycerol or in Lennox L broth (LB; Sigma, St. Louis, MO).DG76 cultures were supplemented with thymine at 50 µg/ml.Strains containing plasmid pZZ15 were grown in the presenceof penicillin (250 µg/ml) and tetracycline (15 µg/ml),while MG1655(pBS8) was grown in the presence of penicillin.Plasmid pZZ15 (53) is an RK2 derivative containing GFP fusedto LacI expressed by the pBAD promoter (19) and approximately256 lacO sequence repeats. Plasmid pZZ15 contains the partitioningand conjugal transfer functions of the native plasmid. Uponinduction with L-arabinose, GFP-LacI is produced from pZZ15and specifically binds to the lacO repeats to provide a fluorescenttag of the plasmid. Plasmid pBS8 is derived from pBAD24 andexpresses an EYFP-RarA fusion protein after L-arabinose induction(38).

Microscopy. To achieve GFP tagging of plasmids, strains were grown overnightat 30°C in LB and then diluted 1:100 into LB or M63 with0.2% glycerol as a carbon source (time zero). After growth at30°C to an optical density at 600 nm (OD₆₀₀) of 0.1 overinoculum, GFP-LacI expression was induced by adding L-arabinoseto a final concentration of 0.6% and then, after the appearanceof tagged plasmid foci, repressed by adding glucose to a finalconcentration of 0.6%. Foci typically became visible approximately1 h after induction with L-arabinose. MG1655(pZZ15) was treatedwith 15 µg/ml nalidixic acid at 3 h, with 1 mg/ml novobiocinat 4 h, or with 200 µg/ml cephalexin immediately afterinoculation. DG76(pZZ15) cells were grown either with or withoutthymine at 50 µg/ml as a supplement. Samples were harvestedat various time points for visualization by fluorescence microscopy.

For tagging of replication complexes, MG1655(pBS8) was grownovernight in LB at 30°C and then diluted 1:100 into 20 mlof M63 with 0.2% glycerol as the carbon source. After reachingan OD₆₀₀ of 0.2 over preinoculated culture, cultures were diluted1:4 and EYFP-RarA expression was induced by adding L-arabinose(in M63) to a final concentration of 1%. After an additional3 h of growth, the culture was split into three portions: oneportion was untreated and the other two were treated with 15µg/ml nalidixic acid or 1 mg/ml novobiocin (final concentration).

For visualization by fluorescence microscopy, 8 µl ofcells were stained with DAPI (5 µg/ml) and added to poly-lysine-treatedslides without fixation. Cells were visualized with a BX60 fluorescencemicroscope (Olympus), and images were captured with a C-5050digital camera (Olympus). Cells and foci were measured usingpublic domain NIH Image software (National Institutes of Health;http://rsb.info.nih.gov/nih-image/).

Thymidine incorporation assay. E. coli strain MG1655 was grown overnight at 30°C in LBmedium supplemented with 50 µg/ml thymidine. After 1:100dilution with fresh medium, cells were grown at 30°C toan OD₆₀₀ of 0.1 and 100 µCi [³H]thymidine was added tothe medium. After an additional 90 min of growth, cells wereeither untreated or treated by adding at a final concentrationof 15 µg/ml nalidixic acid or 1 mg/ml novobiocin to thecultures. At 15- or 30-min intervals, 200-µl samples wereremoved and incubated for 5 min at 4°C in 5 ml of prechilled5% trichloroacetic acid with 1 mg/ml cold thymidine to precipitateDNA (8). Samples were then passed through GF/C filters and washedtwice with 5 ml of 5% trichloroacetic acid at 4°C, threetimes with 1 ml of water, and twice with 1 ml of 95% ethanol.Filters were dried and subjected to liquid scintillation counting.

RESULTS

Top

Results

Effects of nalidixic acid, novobiocin, and cephalexin on RK2 localization. Nalidixic acid and novobiocin are known to inhibit E. coli gyraseactivity (12, 13, 58). As a consequence of gyrase inhibition,there is a rapid inhibition of DNA synthesis (23) and eventuallycell division, leading to the formation of filamentous cells.A thymidine incorporation assay was performed to determine thetiming and extent of DNA inhibition by nalidixic acid and novobiocinunder the conditions of our experiments. As shown in Fig. 1A,there was an apparently immediate inhibition of [³H]thymidineincorporation after the addition of either drug compared tountreated cells.

View larger version (19K):
[in this window]
[in a new window]
FIG. 1. Treatment of MG1655(pZZ15) cells with nalidixic acid, novobiocin, and cephalexin. A. Thymidine incorporation during growth of MG1655(pZZ15) cells either untreated or treated with na-lidixic acid or novobiocin. The black arrow indicates time of addition of inhibitor. B. Foci per cell versus cell length for cells either untreated or treated with nalidixic acid for 3 h or cephalexin for 5 h. Each point represents one cell. A total of 325 cells were counted. C. Foci per cell versus cell length for cells either untreated or treated with novobiocin for 4 h. Each point represents one cell. A total of 325 cells were counted. D. Histogram of cells with a given number of foci. The y axis is expressed as the percentage of total cells with a given number of foci. Cells were either untreated or treated with nalidixic acid for 3 h or novobiocin for 4 h. A total of 254 cells were counted. Over 90% of the nalidixic acid-treated cells and 98% of the novobiocin-treated cells that were stained with DAPI contained a single nucleoid; the remaining cells contained two nucleoids. The results presented are from a single experiment for each treatment, but each treatment was carried out in more than one experiment with similar results.

The RK2 derivative pZZ15 was used to localize this plasmid inMG1655 cells grown in minimal medium. Plasmid pZZ15 encodesan L-arabinose-inducible GFP-LacI fusion protein and a lacObinding site cassette. Addition of L-arabinose results in expressionof GFP-LacI and specific binding of this fluorescent fusionprotein to the plasmid. After approximately 1 h of treatment,nalidixic acid-treated cells began to filament. By 3 hours aftertreatment, some nalidixic acid-treated cells achieved a lengthof approximately 10 µm. Some of the elongated cells containedas many as eight plasmid foci, compared with no more than fivein untreated cells (Fig. 1B and D). This increase in foci numberupon treatment with nalidixic acid under conditions of DNA replicationinhibition suggests some destabilization of plasmid clusters.However, the possibility that this increase in the number ofplasmid foci is due to less-than-complete inhibition of plasmidreplication cannot be ruled out.

It was also observed that in cells treated with nalidixic acid,foci were often grouped near the midline of the elongated cell(Fig. 3B and H). In untreated cells with two foci, the fociwere located at approximately the 1/4 and 3/4 positions, asexpected (Fig. 2A and 3A and G). In nalidixic acid-treated cellswith two foci, however, the foci were not well spaced but tendedto be centered around the cell midline (Fig. 2B and 3B and H).Of the nalidixic acid-treated cells in our study, 65% were filamentouswith a midline grouping of two or more foci. The remaining cellswere filamentous with foci dispersed throughout the cell (7%)or were cells of normal length with foci dispersed randomlythroughout the cell (28%). Staining of the nucleoid by DAPIrevealed that foci were almost always confined to the regionof the nucleoid in nalidixic acid-treated cells (Fig. 3E and H).

View larger version (56K):
[in this window]
[in a new window]
FIG. 3. Effect of inhibitors on the positions of nucleoids and RK2 foci. MG1655(pZZ15) cells were untreated (A, D, and G) or treated with 15 µg/ml nalidixic acid for 3 h (B, E, and H), 200 µg/ml cephalexin for 4 h (C, F, I, and J), or 1 mg/ml novobiocin for 4 (K) or 24 (L) hours. Cells were visualized by fluorescence microscopy, and plasmid foci were visualized by tagging with LacI-GFP (A, B, C, and J, i panels). Chromosomal DNA was stained with DAPI (D, E, F, and J, ii panels). GFP and DAPI images were merged to give the images shown in panels G, H, I, J (panel iii), K, and L.

View larger version (12K):
[in this window]
[in a new window]
FIG. 2. Nalidixic acid interference with RK2 localization. Position of individual foci in MG1655(pZZ15) cells with two foci (A and B) and untreated (A) or treated with nalidixic acid (B). The y axis is the length in micrometers of the host cell, while the x axis is the fractional distance of the foci from the cell pole. Each point represents a single focus and is represented by the first focus and second focus from a pole.

Novobiocin had a similar effect on the distribution of plasmidfoci. Like nalidixic acid-treated cells, many cells were elongatedwith an approximately twofold increase in the number of foci4 hours after the addition of novobiocin (Fig. 1C and D and3K). In addition, as with nalidixic acid treatment, the additionof novobiocin resulted in the grouping of foci near the midlineand within the region of the nucleoid as revealed by DAPI staining.After 4 h of treatment 65% of novobiocin-treated cells had oneor two foci at locations typical of untreated cells, while 31%demonstrated mid-cell grouping of foci (Fig. 3K, panel i). Theseeffects of drug treatment were even more pronounced after 24h (Fig. 3L, panel i). There were, however, some differencesbetween nalidixic acid- and novobiocin-treated cells. In additionto cells with midline-grouped foci, 4% of novobiocin-treatedcells had foci with a "spaced" grouping 4 h after treatment(Fig. 3K, panel ii). In these cells, foci were predominatelyin two groups, one on either side of the cell midline. After24 h of treatment, the number of cells with spaced groupingincreased to 14% of the total (Fig. 3L, panel ii).

The localization of plasmid foci after treatment with nalidixicacid or novobiocin is unlike that observed with cephalexin addition.Growth of cephalexin-treated MG1655(pZZ15) cells in either M63prepared with glycerol (Fig. 3C, F, and I) or LB medium (Fig.3J) resulted in filaments with evenly spaced nucleoids withthe foci located within the nucleoid bodies. Since cephalexindoes not inhibit DNA replication, the filaments contained largenumbers of evenly spaced foci (Fig. 1B and 3C and I). This suggeststhat the mislocalization of plasmid clusters seen with nalidixicacid and novobiocin treatment, and not seen with cephalexintreatment, is directly or indirectly due to the inhibition ofDNA replication and not the inhibition of cell division alone.In 35% of cephalexin-treated cells, the filament contained aregion with a large number of somewhat less intense foci (Fig.3C, panel i, and I, panel i), suggesting declustering of fociin these regions, but these foci were evenly distributed throughoutthe region and did not show the localization pattern characteristicof cells that were treated with the gyrase inhibitors.

Effect of thymine starvation on RK2 localization. In E. coli strains auxotrophic for thymine, thymine starvationresults in the inhibition of DNA replication as well as a numberof other direct and indirect effects, including DNA breakageand the induction of the SOS response (1). DG76, a thyA46 strain,was subjected to thymine starvation to determine the effectof thymine-mediated replication inhibition on plasmid localization.DG76(pZZ15) cells were grown overnight in LB supplemented withthymine and then diluted into M63 with or without thymine. DG76(pZZ15)cells grown in thymine were somewhat longer, up to 6 µmlong, and had more foci than MG1655(pZZ15) cells (Fig. 1C; seealso Fig. 5A). After 6 h without thymine, however, many cellshad become highly filamentous, reaching as much as 25 µmin length (Fig. 4A and 5A and B). These cells also containedup to 14 plasmid foci, with an average of 5.5 per cell comparedto 3.2 for cells supplemented with thymine, again suggestingsome destabilization of plasmid clusters, as seen with nalidixicacid and novobiocin treatment. Like MG1655(pZZ15) cells treatedwith nalidixic acid or novobiocin, DG76(pZZ15) cells grown withoutthymine displayed dispersed foci (56%) (Fig. 4A, panel i, andG, panel i) and mid-cell grouping of foci (33%) (Fig. 4A, panelii, and G, panel ii). A further 11% of thymine-starved cellsdisplayed the spaced grouping phenotype seen after novobiocinaddition (Fig. 4A, panel iii, and G, panel iii). As with cellstreated with either gyrase inhibitor, plasmid foci were largelyconfined to the region occupied by the nucleoids (Fig. 4D and G).

View larger version (27K):
[in this window]
[in a new window]
FIG. 5. Effect of thymine starvation of cells on foci number. (A and B) DG76(pZZ15) cells were supplemented with thymine (blue) or were grown without thymine for 6 h (red). (C and D) Cells grown without thymine for 6 h (red) and with thymine added after 6 h and then grown for an additional 3 h (green). Graphs show foci per cell versus cell length in micrometers (A and C) or the percentage of cells with a given number of foci (B and D). In panels A and C, each point represents one cell. A total of 154 thy⁺, 152 thy^–, and 180 recovering thy^– cells were counted. Of the thymine-starved cells that were stained with DAPI, 60% contained a single nucleoid. The remaining cells contained two nucleoids. The results are from a single experiment, but the thymine starvation was carried out a second time with overall similar results.

View larger version (53K):
[in this window]
[in a new window]
FIG. 4. Plasmid localization after thymine starvation and recovery. DG76(pZZ15) cells were grown without thymine for 6 h (A, D, and G), and then thymine was added to the growth medium. Cells were then grown for an additional 1 h (B, E, and H) or 2 h (C, F, and I). (A, B, and C) Visualization with GFP and bright field; (D, E, and F) staining with DAPI; (G, H, and I) merged GFP, bright field, and DAPI images. White arrows show the positions of cell division septa.

Redistribution of plasmid foci in thymine-starved cells after the reintroduction of thymine. To observe the effect of cell recovery after thymine starvationon plasmid localization, thymine was added back to cells after6 h of starvation. One hour after the addition of thymine, focihad largely become dispersed throughout the cell (Fig. 4B and H).At this time point, only 5% and 2% of cells counted displayedmid-cell and spaced groupings of foci, respectively. Instead,the plasmid foci in most cells were more evenly distributedthroughout the cells and within the nucleoid regions (Fig. 4E and H).At 2 h after the addition of thymine, foci appearedto be evenly distributed throughout the cell (Fig. 4C and I,panels i and ii). In addition, cell division had apparentlyresumed, as single cells were observed separating from the endsof the bacterial filaments and there was a high proportion ofcells (48%) of unit length, with the remainder being filamentsof various lengths.

Analysis of foci number and cell length at 3 h after the additionof thymine indicated the emergence of a large proportion ofcells (Fig. 5C) that were of a length and foci count similarto cells that have not been deprived of thymine (Fig. 5A). Thus,the addition of thymine to thymine-starved cultures appearsto result, in a large proportion of cells, in the redistributionof nucleoids and plasmid foci to their characteristically normallocations and the resumption of cell division.

Localization of the replisome with EYFP-RarA. It has been reported that novobiocin addition to Caulobactercrescentus results in a rapid disassembly of the replisome (33).In view of the possible role of the replication complex in plasmidcluster formation and localization, it was of interest to determineif novobiocin and/or nalidixic acid cause a similar disassemblyof the replisome in E. coli MG1655. Evidence has been obtainedfor an association of the E. coli RarA protein with the replisomeand can be used as an EYFP-RarA fusion protein to localize thereplication complex by fluorescence microscopy (2, 38). MG1655cells carrying pBS8, which expresses EYFP-RarA in response toL-arabinose, were treated with either nalidixic acid or novobiocinafter the appearance of RarA foci (usually 1 to 2 h after theaddition of L-arabinose to exponentially growing cells). Sampleswere removed over a 60-min period to determine the effect ofaddition of each of these two gyrase inhibitors on the EYFP-RarAfoci. Novobiocin reduced the proportion of cells that exhibitedone or more foci from 76% to less than 4% within 5 minutes afterthe addition of the inhibitor (Fig. 6A). Under the same growthconditions, novobiocin did not reduce the number of cells exhibitingfoci consisting of GFP-LacI-tagged RK2 plasmid (Fig. 6B). Greaterthan 95% of the cells displayed RK2 foci before and after treatmentwith novobiocin over the 60-min period. In addition, the numberof RK2 foci per cell did not increase substantially (Fig. 6C and E)despite the disassembly of the replisome, suggestingthat a fully intact replication complex is not a critical factorin the clustering of individual RK2 molecules. A similar resultwas obtained after nalidixic acid addition, except that theRarA-tagged replication complexes reappeared in most cells after30 to 60 min of treatment (Fig. 6A). Again, the addition ofnalidixic acid did not result in a change in the proportionof cells exhibiting RK2 foci, or in a substantial change inthe number of foci per cell, during the 60-min treatment despitethe substantial loss of RarA foci during the initial 30 minafter nalidixic acid addition (Fig. 6C and D). The reappearanceof RarA foci after 30 to 60 min of nalidixic acid treatmentmay be due to repair synthesis of double-strand breaks, whichare known to occur after addition of this gyrase inhibitor.As discussed previously, both nalidixic acid and novobiocincause some RK2 cluster destabilization after 3 or 4 h of treatment,but neither shows significant destabilization within the firsthour of treatment.

View larger version (29K):
[in this window]
[in a new window]
FIG. 6. RarA and RK2 foci in nalidixic acid- and novobiocin-treated cells. (A) Histogram of the percentage of cells with RarA foci at a given time after treatment in MG1655(pBS8) cells either untreated or treated with nalidixic acid or novobiocin. (B) Histogram of the percentage of cells with plasmid RK2 foci at a given time after treatment in MG1655(pZZ15) cells either untreated or treated with nalidixic acid or novobiocin. (C to E) Percentage of MG1655(pZZ15) cells with one, two, three, four, or greater than four foci at a given time in untreated cells (C), cells treated with nalidixic acid (D), or in cells treated with novobiocin (E).

DISCUSSION

Top

Discussion

Both the chromosomal replication origin and plasmids exhibitspecific localization and dynamic movement during bacterialcell growth and division (for reviews, see references 17, 24,and 56). After duplication, clusters of plasmids P1, F, R1,and RK2 have been shown to separate rapidly to specific locationsat some time prior to cell division (16, 18, 31, 44, 49, 53).It has been estimated that plasmids P1 and F in E. coli separatefive times faster than oriC (18). While the machinery for chromosomalorigin movement and plasmid segregation may be the same or sharecommon elements, it is clear that the separation of chromosomalorigins and plasmid clusters are not coordinated with each otherat a specific time in the cell cycle (15, 18, 25). Little isknown about the role of plasmid- or host-encoded factors inclustering, localization, and movement of plasmids, but severalrecent studies have begun to provide some insight. Evidencehas been obtained for a role of the MreB and the SetB proteinsin chromosome segregation (7, 37). In the case of plasmid R1the plasmid-encoded ParM protein, a member of the F-actin family,is likely to play a key role in segregation (32, 47). The partitionsystems of plasmids P1, F, and RK2 do not appear to encode actin-likeproteins. The partition systems for these plasmids, however,do play a role in their specific localization (4, 6, 49). Onthe other hand, plasmid cluster formation occurs with or withouta functional partition system (25, 64). While the par regionof a plasmid does not appear to play a role in plasmid clustering,it is not unreasonable to expect that the plasmid replicationorigin with or without an associated plasmid-specified initiationprotein is important in plasmid clustering.

It is likely that both plasmid-encoded and host factors contributeto plasmid clustering. There is accumulating evidence that bothin E. coli and B. subtilis the DNA replication complexes arefixed in the region of the mid- and quarter-cell positions (36,41, 42). It has also been demonstrated for the E. coli chromosomethat in fast-growing cells, new replisomes are assembled andreplication is initiated at the quarter positions, even beforetermination at the mid-cell replisome and before cell division(38, 40). For plasmids, however, it is not yet known if thereis movement of the fully formed replication assembly from themid- to the quarter-cell positions or if the assembly is reformedat the quarter-cell positions after disassembly at the mid-cellposition. Nor is it known whether or not one or more componentsof the replication factory play a role in plasmid clustering,localization, and movement. Studies with plasmid pSC101 in E.coli supported a role for DNA replication proteins in segregationof this plasmid (46). In addition, it is conceivable that aplasmid cluster nucleates a replication complex that is dedicatedspecifically to the replication of that plasmid element. InC. crescentus, using a fusion of the ${delta}$ ' component of DNA polymeraseIII to GFP to identify the position of the replication complex,additional foci were observed in cells that harbored a plasmidelement (33). Similarly, it was recently reported that B. subtiliscells containing clusters of a multicopy plasmid also containedadditional foci that were tagged with GFP fused to a subunitof DNA polymerase (62). The fact that the clusters of differentreplicons occupy different positions at the mid- or quarter-celllocations (25) may be the result of targeting of individualreplication complexes with associated plasmid clusters to distinguishablelocations within these regions of the cell.

The present study was directed at determining the effect ofgyrase inhibition and/or the inhibition of DNA synthesis and,possibly, the disassembly of replication complexes on plasmidclustering and localization. It has been shown in vitro thatthe inhibition of gyrase activity prevents the initiation ofplasmid RK2 replication (34). The fact that both of the gyraseinhibitors used in this study at best cause only partial declusteringof plasmid pZZ15, which is present at five to eight copies perchromosome (9, 55, 59, 60), suggests that a normal level ofplasmid superhelicity is not a requirement for plasmid clustering.It is of course possible that there is sufficient residual superhelicityin plasmid pZZ15 after addition of the gyrase inhibitors toretain plasmid clustering. It is well established that DNA gyraseinhibition and thymine starvation of an E. coli thymine auxotrophinhibit DNA synthesis and indirectly inhibit cell division.After several hours the filaments that are produced by thesetreatments contain one or two extended nucleoids along the mid-lineof the cell (Fig. 3 and 4). In the present study, using GFP-LacIto tag RK2 containing an array of lacO sequences, it was foundthat shutting down DNA synthesis interferes with the localizationof plasmid foci at their characteristic mid- or quarter-cellposition as the cell elongates prior to cell division. Underthese conditions the foci are disordered but remain confinedto the region of the nucleoid body, as revealed by DAPI staining.This disruption of the orderly spacing of both nucleoids andRK2 foci is not simply due to filamentation, since cephalexintreatment, which inhibits cell division but not replication,results in orderly spacing of nucleoids and foci clusters (Fig.3). In addition, gyrase inhibition by the addition of nalidixicacid or novobiocin and thymine starvation result in a smallincrease in the number of foci per cell without a significantincrease in the number of nucleoids, suggesting some declustering.On the basis of RK2 copy number, complete declustering of theplasmid would yield 5 to 8 foci in young cells and 10 to 16foci in older cells prior to cell division. Interestingly, thereaddition of thymine to thymine-starved cells results in arelocalization of plasmid foci over time and a reduction inthe number of foci per cell, suggesting a reversal of the partialdeclustering and movement of the plasmid clusters to their normalfixed positions (Fig. 4).

Because of the direct and indirect effects of nalidixic acid,novobiocin, and thymine starvation on cell processes, the immediatecause of foci delocalization and partial declustering is notclear. Direct effects of thymine starvation, in addition toinhibition of DNA synthesis, are the formation of single- anddouble-strand breaks (10, 65), the induction of the SOS system(28, 29), and activation of DNA repair processes (see reference1 for review). Indirect effects include inhibition of cell division(29) and filamentation (3). In the cases of novobiocin and nalidixicacid treatment, an immediate effect is the inhibition of gyraseand, therefore, a reduction of negative superhelical turns inplasmid and chromosomal DNA (12, 13, 58). In addition, as shownin Fig. 1A and reported by others (23, 57), there is a rapidinhibition of DNA synthesis by these two inhibitors and eventuallyan inhibition of cell division. Both nalidixic acid and novobiocinalso inhibit E. coli topoisomerase IV (20, 26, 35). By virtueof its mode of action on E. coli gyrase, nalidixic acid alsoproduces an irreversible lesion in DNA, which can lead to double-strandedbreaks (22, 27). Unlike nalidixic acid, the coumarin novobiocinis a competitive inhibitor of the ATPase reaction of gyrase,and this inhibition does not result in the extensive formationof double-stranded breaks in DNA. It is of interest that despitethe different modes of action and consequences with respectto the intactness of the DNA, similar effects on localizationand clustering were obtained. This may suggest that inhibitionof DNA synthesis in growing cells, a consequence of each ofthese treatments, is responsible for the changes observed inthe localization of plasmid ZZ15. This difference in the extentof DNA breakage depending on the gyrase inhibitor used is consistentwith the differential effect of these two gyrase inhibitorson the appearance of RarA foci (2, 38). Inactivation of therarA gene in otherwise-wild-type cells does not result in amutant phenotype; however, there is evidence that the RarA proteinis associated with the replication complex in addition to beingimplicated in recombination events at the replication fork (2,38). As shown in Fig. 6, the addition of either nalidixic acidor novobiocin causes a dramatic loss of RarA foci within 5 minutesof treatment. However, unlike novobiocin treatment, most ofthe nalidixic acid-treated cells began to redisplay RarA focibetween 30 and 60 min after addition of the inhibitor. Thisreappearance of RarA foci in nalidixic acid-treated cells maybe due to the production of double-stranded breaks and the attemptsof the cell to repair these DNA breaks. At least for bacteriophageT4 there is a tight linkage between DNA replication and double-strandbreak repair in the life cycle of this bacteriophage in E. coli(14). The reappearance of RarA foci may thus also be relatedto attempts on the part of the cell to restart stalled replicationforks, a process that may involve the induction of the SOS systemand an accumulation of enzymes at the replication fork, includinga replication complex.

The rapid disappearance of RarA foci after novobiocin or nalidixicacid treatment is interpreted as a partial or complete disassemblyof the replisome as a direct or indirect result of inhibitionof gyrase. In C. crescentus it was observed that novobiocintreatment caused the loss of replisomes tagged by a fusion proteinconsisting of GFP and the core ${delta}$ ' subunit of DNA polymerase III(33), suggesting that treatment with this gyrase inhibitor resultsin more than just a loss of the RarA component of the replicationfactory in our experiments. Significantly, although there isdisordering of the GFP-LacI-tagged RK2 foci, treatment witheither gyrase inhibitor does not result in a significant increasein the overall number of foci or the distribution of foci percell within 60 min (except for perhaps a somewhat higher numberof cells with four or more foci after 60 min of treatment withnovobiocin). Three to four hours of treatment is required tosee an approximate doubling of the average number of foci, whichis far less than a five- to eightfold increase that would beobserved if the plasmid were completely declustered. These resultssuggest that while a fully intact replisome may play a rollin the targeting of the plasmid cluster to a specific location,it is not an essential factor in the clustering of plasmids.The disassembly of the replication factory, however, could resultin an untethering of the plasmid cluster at its receptor site.Nevertheless, while the plasmid foci become disordered theyremain confined to the region of the nucleoid body, even insubstantially elongated filaments. This confinement may be dueto a specific or nonspecific attachment of the plasmid clusterto the chromosomal matrix or a physical barrier to movementof detached plasmid clusters beyond the nucleoid body. The recentreport of the reorganization of chromosomal DNA into orderedtoroidal structures that are highly condensed and exhibit crystallographicfeatures when E. coli cells enter the stationary state (11)may be relevant in this regard. It is possible that not unlikethis effect of the stationary phase, a highly condensed stateof chromosomal DNA occurs after gyrase inhibition and thyminestarvation that may prevent the free movement of detached plasmidclusters and their confinement to the nucleoid body. Studieswith plasmids P1 and R1 that lack functional partitioning genesand an oriC plasmid have shown the presence of plasmid focioutside of the nucleoid under normal growth conditions, indicatingthat plasmids are not in every case confined to the nucleoidbody (6, 31, 49). It is of course possible that the partitioninggenes of plasmid RK2 similarly are required for the normal positioningof this plasmid within the nucleoid body. The continued useof inhibitors that disrupt plasmid dynamics and the analysisof host and plasmid mutants that are defective in plasmid localizationand movement should eventually provide a mechanistic insightinto these complex processes.

ACKNOWLEDGMENTS

We thank Aresa Toukdarian for her insights regarding the projectand Jennifer Han and Elizabeth Long for assistance with dataanalysis.

This work was supported by research grant A1-07194 from theNational Institute of Allergy and Infectious Diseases.

FOOTNOTES

* Corresponding author. Mailing address: Center for Molecular Genetics and Division of Biological Sciences, University of California at San Diego, 9500 Gilman Drive, La Jolla, CA 92093-0322. Phone: (858) 534-3638. Fax: (858) 534-0559. E-mail: dhelinski{at}ucsd.edu.

Present address: Quest Diagnostics Nichols Institute, 33608Ortega Highway, San Juan Capistrano, CA 92675.

REFERENCES

Top