Nucleic Acids Research, Vol. 19, No. 3 497

k.) 1991 Oxford University Press

Physical and biochemical characterization of recombination-dependent synthesis of linear plasmid multimers in Bacillus subtilis Heinrich Leonhardt, Rudi Lurz and Juan C.Alonso* Max-Planck-lnstitut fur Molekulare Genetik, IhnestraBe 73, D-1000 Berlin 33, FRG Received November 12, 1990; Revised and Accepted January 14, 1991

ABSTRACT The synthesis and structure of linear multimeric plasmid molecules (hmw DNA) in Bacillus subtilis were investigated. The replication of covalently-closedcircular supercoiled (form 1) DNA requires the ratelimiting plasmid-encoded replication initiation protein. Unlike form 1, hmw DNA synthesis is partially resistant to inhibition of cellular transcription or translation and requires the host DnaB protein. In addition, hmw DNA synthesis involves host recombination and repair functions (RecE and Poll). Analysis of hmw DNA by electron microscopy displayed linear DNA molecules up to 100 kb in size, which were either single-stranded, double-stranded or double-stranded with single-stranded ends. Structural features of hmw DNA molecules were mapped by means of heteroduplex studies using defined strandspecific probes. The results suggest that a recombination intermediate, but not plasmid-encoded replication, is involved in the initiation of hmw DNA synthesis. INTRODUCTION Early genetic experiments already suggested a tight interrelation between the three basic processes of DNA metabolism, replication, recombination and repair (1, 2). The involvement of recombination functions was later demonstrated for several DNA replication systems (3-5). All these systems, however, involve complex genomes which renders the analysis difficult. Recently, linear multimeric plasmid DNA forms were observed in Escherichia coli (6) and Bacillus subtilis (7) which apparently require recombination and repair functions for their synthesis. This DNA form was characterized as consisting of linear headto-tail plasmid multimers and was termed 'hmw DNA' due to its high molecular weight in comparison to normal plasmid DNA forms. Increased amounts of hmw DNA were observed when the exonuclease V (ExoV) was inactivated (6, 7). This enzyme is also termed AddABC (B. subtilis) and RecBCD (E. colt). Hmw DNA has been found with all plasmid replicons tested so far, i.e. with plasmids replicating via a theta (Cairns) mode as well as sigma (also sometimes termed rolling circle) mode (7, 8; *

To whom correspondence should be addressed

Alonso and Stiege, unpublished). The synthesis of hmw DNA, therefore, seems to reflect a general feature of bacterial cells rather than a replicon specific one. In B. subtilis accumulation of hmw DNA was also detected in ExoV-proficient cells (7, 9, 10). So far it is not known how the replication of circular molecules can be shifted to the synthesis of linear multigenomiclength DNA molecules (rolling circle replication), as has been observed for example with late phage X replication. The analysis of hmw DNA may provide information on the synthesis of linear DNA in bacteria and on the cellular interplay of replication, repair and recombination functions. The plasmids used in this study were based on the replicon of pUBl O, which belongs to a group of small high-copy-number plasmids found in Gram+ bacteria (11). These plasmids replicate via a sigma mechanism similar to that of single-stranded enterobacteriaceae phages (12). As in the phage system, plasmid replication proceeds via a single-stranded circular intermediate termed plus- or leading-strand, which then serves as a template for the minus- or lagging-strand synthesis (13). In this case, mainly circular unit-length plasmid molecules are generated (sigma replication). In this communication we present a simple approach for the isolation of hmw DNA molecules and their subsequent characterization by electron microscopy. The initiation of hmw DNA synthesis was characterized by testing the involvement of the plasmid-encoded initiation protein as well as host-encoded DNA replication, recombination and repair functions.

MATERIALS AND METHODS Bacterial strains, plasmids, phages and media B. subtilis strains used were: YB886 (trpC2 metB5 amyE xin-1 attSP,B sigB) and its isogenic derivatives BG190 (recE6), BG189 (addA5 addB72), BG213 (recE6 addA5 addB72) (14), BG193 (diaB37ts) (9), BG199 (&daF33ts), BG200 (dnaN5ts) and BG201 (dnaX5lts) (15). Furthermore we used YB965 (metB hisB leuA8 xin-l attSP, polA5) (16) and its isogenic derivative BG219 (recE6 polAS) (14). A rec+ derivative (BG83) of PSL1 (recE4 leuA8 arg-15 thrA5 attSPb hsmM hsrM stpl) (17) was constructed (this report).

498 Nucleic Acids Research, Vol. 19, No. 3 Plasmid pEBi 11 and pEB1 14 are shuttle vectors derived from pUBIIO (18). Plasmid pEB114copl is a low-copy-number derivative of pEBi 14 (19). The hybridization probes were derived from phage M 13. A 1047 bp DNA fragment (coordinates 2467 to 3514 in pEB1 14) was cloned into the multicloning site of M13mpl8/19 (20) so that the ssDNA form contains either the plus- or the minus-strand of pEB114 (Ml3p and M13m, respectively). Cells were grown in TY medium (21) supplemented with 5 isg/ml neomycin. The recE gene in strains BG190, BG213 and BG219 was disrupted by insertion of a chloramphenicol acetyltransferase gene. Such strains were grown in TY medium containing 5 jig/ml chloramphenicol (see 14).

Preparation of hmw DNA B. subtilis strain BG83 containing plasmid pEB1 14 was grown up to an OD560 of about 0.8. The cells were washed and resuspended in 1/100 of the original volume of lysis buffer A (50 mM glucose, 10 mM EDTA, 25 mM Tris-HCl pH 8) at 37°C. LGT agarose (BioRad, Richmond) was added to a final concentration of 0.8% and a cell density of about 1010. Solidified agarose blocks were incubated for 2 h at 37°C in lysis buffer A containing 0.5 mg/ml lysozyme, 0.1 mg/ml RNase A and subsequently for 20 h in lysis buffer B (1% SDS, 0.2 M EDTA and 3 mg/ml proteinase K). The agarose blocks were then dialysed and stored in TE buffer (5 mM Tris-HCl pH 8 and 10 mM EDTA). These blocks were incorporated into normal 0.8% agarose gels, and the different DNA forms were separated by electrophoresis at 4°C for 20-30 h at 4 volts/cm. Bands containing hmw DNA were excised and DNA was prepared according to (22) with modifications. The agarose gel was dissolved by NaI, DNA was bound to glassmilk (BiolOl, La Jolla) and directly eluted in TE buffer. Electron microscopy of hmw DNA Heteroduplex molecules of the hmw DNA with M13p or Ml3m were formed without denaturation by incubation overnight at 37°C in 50% formamide, 100 mM phosphate buffer, pH 7.0, 200 mM Na-perchlorate. Hmw DNA and heteroduplex molecules were spread with carbonate buffer on a water hypophase as previously described (23). The size of the hmw DNA molecules was determined by comparison with the lengths of internal standards (Ml3p or Ml3m for ssDNA and pEBi 14 for dsDNA).

RESULTS Initiation of hmw DNA synthesis is independent of a rate limiting effector DNA replication is usually regulated at the level of initiation by a rate-limiting factor. Accordingly, sigma replication of pUBi 10 and its derivatives used in this study is regulated via the plasmid encoded replication initiation protein (RepU). Inhibition of cellular transcription or translation [by rifampicin (Rf) and chloramphenicol (Cm), respectively] leads to a rapid shut down of the synthesis of covalently closed circular (form I) DNA (15). Furthermore, initiation of pUBi 10 replication requires negatively supercoiled DNA, since it is sensitive to DNA gyrase inhibition by novobiocin (Nv) (24). To further elucidate how hmw DNA synthesis is initiated, we tested the effect of these antibiotics. For this purpose, we determined the relative amount of form I and hmw DNA after addition of Rf, Cm and Nv. Total DNA from cells containing plasmid pEB 1l4copl was prepared at different times after addition of antibiotics. DNA forms were separated by gel electrophoresis, and plasmid specific DNA forms were highlighted by Southern hybridization and quantified. In the absence of antibiotics, form I and hmw DNA increased about 4-fold within 120 min (Fig. 1). After inhibiting transcription or translation, the relative amount of form I DNA increased about 1.3-fold within the first 10 min and then levelled off. In the same experiment chromosomal DNA increased at most 2-fold during 120 min (data not shown). Hmw DNA synthesis, however, still continued in the presence of Cm or Rf and increased during 120 min 1.7-fold and 2.5-fold, respectively. The addition of Nv caused an immediate stop of form I, hmw (Fig. 1) and chromosomal DNA synthesis (data not shown), indicating the requirement of DNA gyrase for all three types of replicative molecules. Unlike the DNA synthesis of form I plasmid DNA



300 -300 0


Preparation of total DNA, Southern hybridization and relative quantification of DNA forms Total DNA was prepared as previously described (7). DNA forms were separated in an 0.8% agarose gel and transferred to a nylon membrane (GeneScreen, Dupont). The transfer and subsequent hybridization at 600C were carried out according to the supplier's recommendation. As hybridization probe we used 32P-labelled pUBI 10 DNA, which was prepared with a nick-translation kit (Boehringer, Mannheim). For relative quantification of the plasmid DNA forms the corresponding signals in autoradiographs were scanned in two dimensions with a laser densitometer (LKB Ultroscan XL). Data were analysed with the LKB GelScan XL. Chromosomal DNA was quantified by scanning photographic negatives from ethidium bromide stained agarose gels. Measurements were carried out at a range of exposition times with linear increasing signal intensities.

















Time (min) Fig. 1. Effect of various antibiotics on hmw DNA synthesis. The relative amounts of form I (A) and hmw DNA (B) was determined after addition of antibiotics. Antibiotics were used at the following final concentrations: rifampicin (10 tg/ml, A), chloramphenicol (100 yg/ml, 0) and novobiocin (10 j&g/ml, *). Antibiotics were added in the early logarithmic growth phase (time zero) and samples were taken at indicated times. As a control, cells were grown without antibiotic added (V7). These experiments were performed with B. subtilis strain YB886 containing plasmid pEB I 14cop l.

Nucleic Acids Research, Vol. 19, No. 3 499 (see 15), the synthesis of hmw DNA is not controlled by a ratelimiting factor. This is consistent with the fact that hmw DNA synthesis occurs in the absence of active plasmid-encoded Rep protein (7, 8). Furthermore, the partial resistance to Rf also indicates that RNA polymerase is not involved in priming hmw DNA synthesis. Involvement of recombination proteins in hmw DNA synthesis Previous analyses revealed that a high amount of hmw DNA accumulates in a B. subtilis strain (addA) deficient in the exodeoxyribonuclease V (ExoV) enzyme (7), whereas in the recE4 and polA5 strains such accumulation is markedly reduced when compared to a wild-type strain (9). The recE and polA genes are the B. subtilis counterparts of the E. coli recA and polA genes, respectively. The reduced amount of hmw DNA in recE strains may either indicate an involvement of RecE in hmw DNA synthesis as proposed by (9) or may simply be due to an elevated rate of DNA degradation by the ExoV enzyme in a recE genetic background. The latter interpretation would be supported by the observation that in E. coli the RecA protein protects singlestranded DNA or gapped duplex DNA from degradation by the ExoV enzyme (25). To elucidate the role of RecE in hmw DNA metabolism we used a recE null mutation (recE6) in combination with an ExoV double mutation (addA5 addB72) or the polA5 mutation, and compared the amount of hmw DNA in these strains. Equal amounts of total DNA from these strains containing plasmid pEB 111 were separated by agarose gel electrophoresis and plasmid-specific DNA forms were highlighted by Southern hybridization (Fig. 2). In all six strains similar amounts of form cn







ci c5 ~~~oo u)

Involvement of host DNA replication functions in hmw DNA synthesis and sigma plasmid replication To differentiate further between form I and hmw DNA synthesis and to define their specific requirements, we searched for host functions required for one but not the other replication mode. The genetic approach to identify host encoded proteins involved in hmw DNA synthesis in B. subtilis is, however, limited to a few characterized mutations.

c U)10 QJQj

_3 < CL.


Origin _


I DNA were detected. This indicates that the addA5 addB72, polA5 and recE6 mutations have no effect on formation or degradation of form I DNA (see 9). The comparison of wildtype and recE6 mutant strains (Fig. 2, lane 1 and 3, respectively) shows that in the absence of recE activity the amount of hmw DNA was reduced at least 100-fold. A precise measurement was not possible since the observed difference exceeds the range of linearity. Furthermore the leaky polA5 mutation (26) causes a 7-fold reduction in hmw DNA (Fig. 2, lane 4). In the polA5 recE6 double mutant strain no hmw DNA could be detected, even after longer exposition times (Fig. 2, lane 4). The inactivation of ExoV (addA5 addB72; Fig.2, lane2) caused at most a twofold increase in the amount of hmw DNA in comparison to the wildtype strain. This indicates that most of the hmw DNA is a poor substrate for the major cellular exonuclease (ExoV). Moreover the inactivation of ExoV enzyme by the addA5 addB72 mutations in a recE6 genetic background did not have a significant effect on the accumulation of hmw DNA (Fig. 2, lane 3 and 5). The more than 100-fold reduced amount of hmw DNA in recE6 strains, which is also observed upon inactivation of the major cellular exonuclease (recE6 addA5 addB72) argues against simply an elevated degradation rate of hmw DNA in recE strains (reckless phenotype). Hence we favour the first model, in which hmw DNA synthesis is initiated by a recombination intermediate. In this respect hmw DNA synthesis is similar to T4 recombination-dependent (secondary) replication, which is also resistant to Rf (3).



_ 600





< 400


@ 300 0 C-

Form I -r Fig. 2. Effect of inactivation of RecE, Poll and ExoV on hmw DNA synthesis. Cells containing pEB 111 were grown to a titer of about 1-2 x 108. Plasmid specific DNA forms were visualized as described. The following strains were compared: YB886 (rec+, lane 1), BG189 (addA5 addB72, lane 2), BG190 (recE6, lane 3), YB965 (polA5, lane 4), BG213 (addA5 addB72 recE6, lane 5) and BG219 (recE6 polA5, lane 6).





90 Time(min )




Fig. 3. Differential effect of thermal inactivation of DnaB on cellular DNA syntheses. The relative amount of chromosomal (M), form I (0) and hmw (A) DNA after thermal inactivation of DnaB was determined. Cells were shifted to

nonpermissive temperature (47°C) at time zero. Levels of synthesis observed at permissive temperature (30°C) are indicated by the corresponding open symbols. The data shown in this figure were obtained with BG193 containing pEBI 14copl. Identical results were obtained with plasmid pEBl 11.

500 Nucleic Acids Research, Vol. 19, No. 3

5 kt)

Fig. 5. Electron micrograph of hmw DNA molecules. The photograph shows two hmw DNA molecules with a total size of 22 and 56 kb. Both molecules are double-stranded with a single-stranded end. Putative transition points from singleto double-strand are marked by arrowheads.

Fig. 4. Isolation and subsequent characterization of hmw DNA. Agarose gel slices containing total DNA were embedded in the agarose matrix (lane 2). The part which was excised for the isolation of hmw DNA is indicated by a dotted line. Hmw DNA, which had been isolated like that in a preceeding experiment is shown undigested (lane 3) and digested with EcoRI and BgEI (lane 6 and 8, respectively). For comparison, restriction fragments obtained after digestion of purified form I DNA with EcoRI (lane 5) and BglII (lane 7) are included. DNA from bacteriophage SPP1 was used as a size marker (undigested, 46 kb, lane 1 and digested with EcoRI, lane 4).

The inactivation of different subunits of DNA polymerase III (PoIIII) in dnaF, dnaN and dnaX strains blocked form I, chromosomal (15) and hmw DNA synthesis (data not shown). Hence, hmw DNA synthesis is performed by PollIl. Furthermore, the role of the host DnaB protein was investigated. Mutations in the dnaB gene have been characterized as an initiation mutation of chromosomal DNA replication (see (27) and references therein). Sigma replication of pUB 110-derived plasmids does not require DnaB (9). The involvement of DnaB in hmw DNA synthesis was tested by using strain BG193 carrying the dnaB37 mutation, which encodes a thermosensitive DnaB protein. A shift up to non-permissive temperature (47°C) inactivates the DnaB protein and consequently DnaB-dependent DNA synthesis ceases. The effect of DnaB inactivation on the synthesis of chromosomal, form I and hmw DNA is shown in Fig. 3. As expected for an initiation mutation, the amount of chromosomal DNA increases after inactivation of DnaB about 1.4-fold and then levels off. Isolation and electron microscopy of hmw DNA molecules Hmw DNA molecules were isolated and analysed by electron microscopy to investigate the structure of this DNA form. In order to avoid shearing of DNA, plasmid-containing cells were embedded in LGT agarose prior to cell lysis. With this method the chromosomal DNA remained intact and could be separated from hmw DNA by normal gel electrophoresis (Fig. 4, lane 2). The chromosomal DNA did not enter the gel. The major DNA band obtained had a slower electrophoretic mobility than SPP1 phage DNA (45 kb; see Fig. 4, lanes 1 and 2). This band was

excised as indicated by a dotted line in Fig. 4 and the DNA was extracted. Digestion products of this material were indistinguishable in size from those of purified form I plasmid DNA (Fig.4, lanes 5 to 8), confirming that indeed hmw DNA and not chromosomal DNA had been isolated. The size and structure of 316 hmw DNA molecules were determined by electron microscopy and statistically analysed. As previously predicted (7) three types of linear DNA molecules with a length of up to 100 kb were observed with about similar frequency: single-stranded (ss), double-stranded (ds) and dsDNA molecules with ssDNA ends. The latter molecules were analysed in more detail and two of them are shown in Fig. 5. The singlestranded tails extend up to 35 kb with a mean length of about 10 kb and the double-stranded part extends up to 90 kb. The average total length of these molecules was about 41 kb. Nine out of these 316 molecules had ssDNA regions at both ends and one had an internal ssDNA region. We failed to detect branched or circular molecules. Altogether the length distribution was random with no bias for multiples of plasmid unit-length. From the analysis of these 316 hmw DNA molecules the portion of ssDNA was estimated to be on average in the range of 20 to 30%.

Physical characterization of hmw DNA molecules by heteroduplex studies The structural features of hmw DNA molecules observed by electron microscopy (Fig. 5) cannot be correlated directly with specific plasmid DNA sequences. We therefore used strandspecific probes to tag hmw DNA molecules. This technique allows us to discriminate which of the DNA strands is present in the single-stranded region and to map structural features. As probes we used derivatives of bacteriophages Ml3mpl8 and M13mpl9, which contain in their single-stranded form a fragment of either pEB 1 14 plus- or minus-strand (M I 3p and M 13m, respectively). These probes were separately hybridized with hmw DNA molecules under non-denaturing conditions. Hybridization at the ssDNA end of hmw DNA molecules was observed with

Nucleic Acids Research, Vol. 19, No. 3 501



3500' 2 kb

2kb ---


Fig. 6. Heteroduplex studies on hmw DNA molecules. Electron microscopic pictures of heteroduplex molecules obtained with M13p (A) and M13m (B) are shown together with the corresponding schematic drawings below (C and D). For didactic reasons very small molecules were chosen. DNA strands are distinguished as follows: M13mpl8/19 DNA (dotted line), plus-strand (thin line) and minus-strand (thick line). The length of DNA molecules was determined by comparison with molecules of known size, which renders it possible to calculate, e.g., the position of the ds end of hmw DNA molecules in pEB1 14. The 5'-3' polarity of DNA strands was attributed according to the basic rules of DNA replication: DNA synthesis requires a template and a primer and proceeds in 5'-3' direction.

both strand-specific probes. Altogether 37 hybrids with M13p and 25 hybrids with M13m were identified and analysed. One example of each experiment together with a schematic representation, is shown in Fig. 6. These data show that no strand selectivity occurs in hmw DNA synthesis, whereas sigma plasmid replication specifically proceeds via a circular plus-strand intermediate (28). Since it is known that the hybridizing fragment in the DNA probes extends from position 2467 to 3514 in pEBi 14 (see Fig. 6), it is possible to correlate structural features in the hmw DNA molecules with the DNA sequence of plasmid pEBi 14. The analysis of the above 52 heteroduplices revealed no significant correlation between the ssDNA end points of the hmw DNA molecules and DNA sequences in plasmid pEB1 14. Also the transition points between ss- and dsDNA were randomly distributed in pEB1 14. The location of the dsDNA end points of hmw DNA molecules, which hybridize with Ml3m and which therefore have a plus-strand specific end, is shown in Fig. 7A. The maximum of this distribution curve lies in the region between positions 1000 and 2000 in pEB1 14. The corresponding results obtained from hmw DNA molecules, which hybridize with M13p and have a minus-strand specific end, are shown in Fig. 7B. The two highest values were found in regions (positions 1 to 1000 and 2000 to 3000) which contain the two origins for plasmid replication (oriU and oriL, see Fig. 7C). Only one of these origins (oriL) would be in an active orientation, and hmw DNA synthesis is independent of sigma plasmid replication. We therefore assume that this preference might be due to a particular DNA structure in this region (hot spot), which renders it more susceptible to single- or double-strand breaks. A similar preference for the origin region was observed under certain conditions for X roilingcircle replication (29). For both types of hmw DNA molecules, however, values over the average were also found within a nonorigin region, (Fig. 7A and 7B, positions 8000 to 9000).





@ 2E z n-







Nucleotde coordinates of pEB1 14

oniU -*



Fig. 7. Location of starting points for hmw DNA synthesis in pEB 1 14. Doublestranded endpoints (i.e. putative starting points) of 52 hmw DNA molecules were mapped as outlined in Fig. 6. The respective region in pEBl 14 is indicated in lOOObp intervals on the abscissa and the number of hmw DNA molecules with a double-stranded endpoint mapping in a given interval is indicated on the ordinate. Hmw DNA molecules having a plus-strand specific single-stranded end (A) and those having a minus-strand specific end (B) were separately analysed. A physical

map of pEBl 14 with the location of the pUBl 10 derived part (hatched box) and the origin and direction of leading- and lagging-strand syntheses is outlined below


502 Nucleic Acids Research, Vol. 19, No. 3

DISCUSSION This communication describes the synthesis and structure of hmw plasmid DNA in B. subtilis. Electron microscopy of isolated hmw DNA displayed linear DNA molecules up to 100 kb in size, which were either ss, ds or ds with a ssDNA end. In E. coli linear hmw DNA molecules with ssDNA tails and in addition single-branched molecules were also observed (30). The presence of such molecules in B. subtilis cannot be ruled out, since they might have escaped our isolation procedure (see Material and Methods). Restriction analysis of isolated hmw DNA confirmed the headto-tail multimeric structure of this material. Altogether, these data on the structure of hmw DNA molecules are in agreement with predictions derived from enzymatic degradation studies (6, 7, 31). The accumulation of hmw DNA even in ExoV-proficient cells could be explained by the fact that the ExoV enzyme poorly binds to long ssDNA tails (32). Heteroduplex studies with strand-specific probes confirmed that the ssDNA ends of hmw DNA molecules can be of either plasmid-strand (7; this report). This means that no strandselectivity occurs in hmw DNA synthesis. By contrast, sigma plasmid replication specifically proceeds via a plus-strand specific intermediate (28). In addition, these heteroduplex studies allow structural features of hmw DNA molecules to be determined and mapped (Fig. 6 und 7). The ssDNA end points of hmw DNA molecules were found to be randomly distributed in the pEB1 14 genome, which indicates that hmw DNA molecules might be released by random disruption. Furthermore, the transition points from ss- to dsDNA showed a random distribution, which might be due to the existence of several unresolved points at which priming of complementary strand synthesis can occur. The dsDNA end points most likely represent putative starting points of hmw DNA synthesis. This starting point would be the 5'-end of the discontinuous strand, which is extended at its 3'-end. The starting points of hmw molecules were distributed all over the plasmid (Fig. 7), which is consistent with the hypothesis that hmw DNA synthesis starts from random nicks (7). The synthesis of hmw DNA was found to require DnaB, RecE and DNA polymerase I (Poll), which are not required for sigma plasmid replication (9). Moreover, unlike sigma plasmid replication, the synthesis of hmw DNA continues after inhibition of cellular transcription and translation with Rf and Cm, respectively. From these data we infer that hmw DNA is synthesized by a mechanism which is distinct and independent of sigma plasmid replication. This conclusion is further supported by the previous observation that hmw DNA synthesis continued after thermal inactivation of the replication initiation protein of the related plasmid pE194 (7, 33). In theory, three different mechanisms could account for hmw DNA synthesis: i) rolling circle replication as postulated for phage X (34-36), ii) replication via branched structures like phage T4 secondary replication (3), or iii) bubble migration replication (37). The first mode was proposed to be initiated by any strand discontinuity, which then is converted by DNA repair enzymes into a rolling-circle replicative molecule (7, 36). For this mechanism the requirement of a recombinase and a DNA gyrase function is not obvious. By contrast, in the second and third mode a single-stranded 3'-end invades a homologous dsDNA molecule in a recombinase-mediated process. This 3'-end provides a primer for DNA synthesis, which in the case of circular templates can generate


multimers. With


DNA synthesis the

invasive DNA strand is extruded at the 5'-end of the D-loop and therefore represents a conservative mode of DNA replication (see

37). The requirement of a recombinase (RecE) indicates that hmw DNA is mainly generated by the second or third mode. The requirement of DNA gyrase may further support this, because strand invasion is greatly enhanced by negative superhelicity (38). Branched structures, as observed in secondary T4 replication, were not detected in our electron microscopic studies on hmw DNA. Altogether, these data suggest that hmw DNA is mainly synthesized by a bubble-migration-like mechanism (see 39). The hmw DNA synthesis described here parallels in several aspects the synthesis of linear plasmid concatemeres in phage infected cells (phage-mediated plasmid transduction). In that case, however, the host factors required for hmw DNA synthesis are substituted by phage-encoded product(s) (40, 41). Phage-mediated plasmid transduction also allows two further predictions for a recombination-dependent replication mode to be tested directly. Firstly, if a phage derived 3'-end invades a homologous region on a plasmid molecule, then chimeric molecules should be generated, which were indeed found in phage-infected cells (41, 42). Secondly, this replication mode should be sensitive to mismatches within the minimal region of homology. In agreement with this prediction, plasmid transduction frequency was found to be reduced by about 30-fold when one or two mismatches were introduced within that region (K.Wilke, A.Bravo and J.C.Alonso, unpublished results).

ACKNOWLEDGEMENTS We are grateful to T.A.Trautner for discussions during the preparation of the manuscript. We are indepted to B.Dobrinski and C.A.Stiege for excellent technical assistance. This work was supported in part by the Deutsche Forschungsgemeinschaft

(SFB344). REFERENCES 1. Hershey, and Rotman, (1949) Genetics, 34, 44-71. 2. Hershey, and Chase, (1951) Cold Spring Habor Symp. Quant. Biol., 16, 471-479. 3. Mosig,G. (1983) In Mathews, C., Kutter,E., Mosig,G. and Berget,P.(eds.), Bacteriophage T4, American Society for Microbiology, Washington D.C.,

pp.120-130. 4. Smith,G.R. (1983) General recombination. In: Hendrix,R.W., Roberts,J.W., Stahl,F.H. and Weisberg,R.A. (eds) Lambda II, 175-209. 5. Kogoma,T., Skarstad,K., Boye,E., von Meyenburg,K. and Steen,H.B. (1985) J. Bacteriol., 163, 439-444. 6. Cohen,A. and Clark,A.J. (1986) J. Bacteriol., 167, 327-335. 7. Viret,J.-F. and Alonso,J.C. (1987) Nucl. Acids Res., 15, 6349-6367. 8. Silberstein,Z. and Cohen,A. (1987) J. Bacteriol., 169, 3131-3137. 9. Viret,J.-F. and Alonso,J.C. (1988) Nucl. Acids Res., 16, 4389-4406. 10. Gruss,A. and Ehrlich,S.D. (1988) J. Bacteriol., 170, 1183-1190. 11. Alonso,J.C. (1989) Microbiol., 5, 5-12. 12. Gruss,A. and Ehrlich,S.D. (1989) Microbiol. Rev., 53, 231-241. 13. Novick,R.P. (1989) Ann. Rev. Genet. 43, 537-565. 14. Ceglowski,P., Luder,G. and Alonso,J.C. (1990) Mol. Gen. Genet., 222, 441 -445. 15. Alonso,J.C., Stiege,C.A., Tailor,R.H. and Viret,J.-F. (1988) Mol. Gen. Genet., 214, 482-489. 16. Friedman, B.M. and Yasbin, R.E. (1983) Mol. Gen. Genet., 190, 481-486. 17. Ostroff,G.R. and Pene,J.J. (1983) J. Bacteriol., 156, 934-936. 18. Leonhardt,H. and Alonso,J.C. (1988) J. Gen. Microbiol., 134, 605-609. 19. Leonhardt,H. (1990) Gene, 94, 121-124. 20. Norrander,J., Kempe,T. and Messing,J. (1983) Gene, 26, 101-106. 21. Biswal,N., Kleinschmidt,A.K., Spatz,H.C., Trautner,T.A. (1967) Mol. Gen. Genet., 100, 39-55. 22. Vogelstein,B. and Gillespie,D. (1979) Proc. Natl. Acad. Sci. USA, 76, 615 -619. 23. Burkardt, H. and Lurz, R. (1984) Electron microscopy. In: Puhler, A., Timmis, K.N. (eds) Advanced molecular genetics.Springer-Verlag, Berlin, 281 -302.

Nucleic Acids Research, Vol. 19, No. 3 503 24. Alonso,J.C., Leonhardt,H. and Stiege,C.A. (1988) Nucl. Acids Res., 16, 9127-9145. 25. Williams, J.K.G., Shibata, T. and Radding, C.M. (1981) J. Biol. Chem., 256, 7573-7582. 26. Martinez,S, Lopez,P., Espinosa,M. and Lacks,S.A.(1987) Mol. Gen. Genet., 210, 203-210. 27. Winston,S. and Sueoka,N. (1982) DNA replication in Bacillus subtilis. In: Dubnau,D. (ed) The molecular biology of bacilli. Academic Press, New York, 35-69. 28. te Riele,H., Michel, B. and Ehrlich, S. (1986) Proc. Natl. Acad. Sci. USA, 83, 2541-2545. 29. Bastia, D., Sueoka, N. and Cox, E. (1975) J. Mol. Biol., 98, 305-320. 30. Silberstein,Z., Maor,S., Berger,I. and Cohen,A. (1990) MoI. Gen. Genet., 223, 496-507. 31. Kusano,K., Nakayama,K. and Nakayama,H. (1989) J. Mol. Biol. 209, 623-634. 32. Taylor,A.F. and Smith,G.R. (1985) J. Moi. Biol., 185, 431-443. 33. Alonso,J.C., Luder,G. and Trautner,T.A. (1986) EMBOJ., 5, 3723-3728. 34. Eisen,H., Pereira da Silva,L. and Jacob,F. (1968) Cold Spring Harbor Symp. Quant. Biol., 33, 755-764. 35. Gilbert,W.and Dressler,D (1968) Cold Spring Harbor Symp. Quant. Biol., 33, 473-484. 36. Skalka,A.M. (1977) Curr. Top. Microbiol. Immunol., 78, 201-237. 37. Formosa,T. and Alberts,B.M. (1986) Cell, 47, 793-806. 38. Radding,C.M. (1989) Biochem. Biophys. Acta, 1008, 131-145. 39. Niki,H., Ogura,T. and Hiraga,S. (1990) Mol. Gen. Genet 224, 1-9. 40. Kreuzer,K.N., Yap,W.Y, Menkens,A.E. and Engman,H.W. (1988) J. Biol. Chem., 263, 11366-11373. 41. Bravo,A. and Alonso,J.C. (1990) Nucl. Acids Res., 18, 4651-4657. 42. Novick,R.P., Edelman,I. and Lofdahl,S. (1986) J. Mol. Biol., 192, 209-220.

Physical and biochemical characterization of recombination-dependent synthesis of linear plasmid multimers in Bacillus subtilis.

The synthesis and structure of linear multimeric plasmid molecules (hmw DNA) in Bacillus subtilis were investigated. The replication of covalently-clo...
2MB Sizes 0 Downloads 0 Views