Plasmid deletion formation in Bacillus subtilis

Plasmid deletion formation in Bacillus subtilis

PLASMID 20,23-32 (1988) Plasmid Deletion Formation in Bacillus subtilis AD A. C. M. PEIJNENBURG,SIERDBRON, AND GERARDVENEMA Department of Genetics, ...

867KB Sizes 0 Downloads 82 Views

PLASMID

20,23-32 (1988)

Plasmid Deletion Formation in Bacillus subtilis AD A. C. M. PEIJNENBURG,SIERDBRON, AND GERARDVENEMA Department of Genetics, Center of Biological Sciences, Kerklaan 30, 9751 NN Haren (Gn), The Netherlands

Received February 22, 1988; revised May 18, 1988 Plasmid pCiP1 carrying a penP-1acZ fusion was used to study structural plasmid instability in Bacillus subtilis. In only one of 28 sequenceddeletion junction points in the penP-IacZ region short direct repeats ( 10 bp) and flanking imperfect inverted repeats (12 bp) were associatedwith endpoints. In 27 deletions no repeated sequencesof more than 3 bp were present at the endpoints. In 15 of these the sequenceS-T-G-T-A-3’ was found within 10 bp from the left endpoint. At the left cleavagesitesthe sequenceS-T-T-T-3, or the S-A-A-A-3’ complement thereof, was frequently observed. Most of the left deletion endpoints were located in potential stem-loop structures in the penP transcription/translation regulatory region, which is very rich in hyphenated dyad symmetry. Near the right deletion endpoints a sequenceconsisting of four G/C residues,followed by three or four A/T residues,was found in 15 cases.It is speculated that DNA topoisomerase I is involved in the formation of the deletions studied. o 1988 Academic RUSS. IIIC. Cloning experiments in Escherichia coli and Bacillus subtilis have frequently been hampered by structural instability, in particular deletions, of recombinant plasmids. Several studies of deletion junction points have indicated that plasmid deletions often result from recombination between nucleotide sequences of little or no homology (illegitimate recombination, for reviews seeEhrlich et al., 1986; Anderson, 1987). Two main classesof sequencesinvolved in illegitimate recombination have been described in both eucaryotes and procaryotes. The first comprises sequences presumed to constitute target sites for a variety of enzymes which cut and join DNA. Particularly, topoisomerase I (Bullock et al., 1985; Yarom et al., 1987) and topoisomerase II (Ike& et al., 1981; Ikeda, 1986; Naito et al., 1984) have been considered to play key roles in illegitimate recombination. In addition, enzymes involved in the initiation and termination of replication can result in deletions starting at origins and origin-like sequences (Horowitz and Deonier, 1985; Michel and Ehrlich, 1986a,b). The second main classof deletion-prone sequences, not known to be targets for cut-andjoin enzymes, consists of short homologous 23

sequences(usually 4 to 14 bp), in eucaryotes (Nalbantoglu et al., 1986), E. coli (Albertini et al., 1982; Jones et al., 1982; Schaaperet al., 1986), and B. subtilis (Lopez et al., 1984; Saunderset al., 1984). Two models have been proposed for illegitimate recombination between short homologous sequences.The.first, called “slipped mispairing,” involves aberrant replication of single-stranded DNA in the replication fork (Albertini et al., 1982). A newly replicated copy of one of the repeated templates is thought to misalign to the other, resulting in a structure in which the sequences between the repeats loop out. Deletion of the loop may occur either by a second round of replication (Albertini et al., 1982), or by excision repair (Efstratiadis et al., 1980; Glickman and Ripley, 1984). Planking inverted repeats are thought to stimulate this processby juxtaposing the deletion endpoints (Albertini et al., 1982; Glickman and Ripley, 1984; Peeters et al., 1988). Support for the idea of replication-stimulated deletion formation between direct repeatswas recently provided by Brunier et al. ( 1988), using precisetranspoaon excision from E. coli plasmids as a model system. The second model implies the introduction of a double-stranded break between the direct repeats (DasGupta et al., 1987). In this 0147-619X/88 $3.00 Copyright 0 1988 by Academic Prcs, Inc. All rights of reprduction in any form rserved.

24

PEIJNENBURG, BRON, AND VENEMA

breakage-and-reunion event, limited exonucleolytic resection would expose complementary sequencesof the repeats and allow annealing and repair synthesis. The primary aim of the present investigations was to study the mechanism of plasmid deletion formation in B. subtilis. Previously, we described a system that enabled the selection of deletions in a plasmid containing a penP-1acZ gene fusion (Peijnenburg et al., 1987). In the present studies we characterized 60 of these deletions by restriction and/or sequence analysis. The endpoints of only 1 out of 28 deletions appeared to be associatedwith direct and inverted repeats. Near most of the remaining deletion endpoints specific sequence patterns, other than repeats, were observed. MATERIALS

AND METHODS

Bacterial strains, phages, and plasmids. Plasmid pGP1 (CmR; EmR)’ consists of pGKl3 containing a 5.5kb BamHI insert with a fusion between the Bacillus licheniformis 749/C penicillinase gene(penp) and the E. coli ,&galactosidasegene (1acZ). Plasmid pGK13 is basedon the Streptococcuscremoris plasmid pWVO1 (Kok et al., 1984), which replicates in E. coli, B. subtilis, and streptococci. Expression of 1acZ in pGP1 is dependent on the S-fused penP regulatory signals. The easy detection of @-galactosidaseactivity enabled the isolation of plasmid deletions in the penP1acZregion (Peijnenburg et al., 1987). Deleted derivatives were denoted as pGPS-x, in which x is a suffix number. The recombination-proficient B. subtilis strain 8G5 (Bron and Venema, 1972) was used throughout. BacteriophagesM 13mp18 and -mp 19(Yanisch-Perron et al., 1983) servedas cloning vectors for DNA fragments containing the deletion junction points. Recombinant phageswere selectedin E. coli JM 109 and amplified in E. coli JM 101 (Yanisch-Perron et al., 1983). ’ Abbreviations used:Cm’, chloramphenicol resistance; EmR, erythromycin resistance; X-gal, Sbrom*4chioro3-indolyl 0-Bgalactopyranoside; IPTG, j3-D-thiogalactopyranoside.

Media and chemicals. TY and minimal media were as described before (Bron and Luxen, 1985). Restriction enzymes, T4 DNA ligase, and DNA polymerase I (Klenow fragment) were purchased from Boehringer (Mannheim, PRG) or New England Biolabs (Beverly, MA, USA) and used as recommended by the manufacturers. 5-Bromo-C chloro-3-indolyl P-D-galactopyranoside (Xgal) and isopropyl /3-D-thiogalactopyranoside (IPTG) were from Sigma (St. Louis, MO) and were used at 40 I.rglml and 1 mM, respectively. Isolation of DNA. Double-stranded Ml 3 phage DNA and plasmid DNA from E. coli and B. subtilis were purified following established procedures (Maniatis et al., 1982; Bron and Luxen, 1985). Preparation of singlestranded Ml 3 DNA for sequencing was as recommended by Amersham International (Buckinghamshire, UK). Selection and mapping of deletions. Cells of B. subtilis 8G5 (pGP1) were grown for ap proximately 20 generations on minimal agar supplemented with Cm (5 &ml), Em (5 gg/ ml), and X-gal. Subsequently, 175 blue colonies were grown separately for 15 to 20 generations in TY broth with antibiotics. After plating on minimal agar containing antibiotics and X-gal, 114 cultures appeared to contain white progeny, the frequency of white colonies varying from 0.5 to 4.5%. Plasmids from 114 independent white colonies were selected for further characterization. Deletions were mapped by electrophoresis of restricted plasmids on 0.5 to 2% horizontal agarosegels in TBE buffer (89 mM Tris, 89 mM boric acid, 2 mM EDTA, pH 8.0) containing 0.5 &ml ethidium bromide. BacteriophageSPP1 DNA, digested with EcoRI, served as a DNA size marker. Cloning and sequenceanalysis of deletion junction points. DNA manipulations were essentially as described by Maniatis et al. (1982) with minor modifications as recommended by Amersham International. Appropriate BamHI and/or DraI restriction fragments were purified from agarosegels by electrophoresis onto DEAE NA-45 membranes (Schleicher & Schuell, Dassel,PRG) as recommended by the

PLASMID DELETION FORMATION

supplier. Purified fragments were cloned in M 13mpl8 and -mp19, which had been linearized with BamHI or SmaI. Sequencing was performed by the dideoxy chain termination method (Sanger et al., 1977) using [(Y35S]dATP(2600 Ci/mmol, Amersham International). As primers the universal M 13 17mer, or appropriate internal 17-mers were used. Reaction mixtures were electrophoresed on 5% polyacrylamide denaturing gels in 0.5x TBE buffer. Labeled products were visualized by autoradiography on Fuji RX NIF films. Secondary structures of single-stranded DNA. The optimal and major suboptimal secondary structures of single-stranded DNA were predicted by the VAX computer program ASRNA, using the dynamic programming algorithms of Williams and Tinoco (1986). Thermodynamic parameters for the free energy were taken from Freier et al. (1986). Direct and inverted repeats were searched using the Microgenie program (Beckman, Palo Alto, CA). RESULTS

Selection and Mapping of Deletions A total of 114 pGP1 deletion derivatives, independently obtained in B. subtilis 8G5, was initially characterized by BamHI restriction. All carried deletions varying from 0.12 to 5.3 kb. In 60 plasmids the deletions were located within the 5.5-kb BamHI insert. Becausethe complete sequencesof penP (Neugebauer et al., 1981) and 1acZ(Kalnins et al., 1983) were known, thesedeletions were chosen for further characterization. Detailed restriction analysis showed that at least 46 different deletions could be distinguished (Fig. l), indicating that a great number of different sites were involved. In the following we define the left deletion endpoints as those having the lowest, and the right endpoints as those having the highest coordinate numbers in Fig. 1. Except for three (pGP5115/49/l 1l), all left deletion endpoints were located in the penP regulatory region. The majority of the right endpoints occurred in or downstream of the 1acZ coding region.

25

DNA Sequencesof Deletion Junction Points To examine whether specific sequencepatterns were present in the surroundings of the deletion termini, the junction points of 28 arbitrarily chosen deletions, shown in italics in Fig. 1, were sequenced. Also sequencedwere the B. licheniformis DNA fragments flanking the penP-1acZ fusion, since these (coordinates l-496 and 4835-5536) contained several deletion endpoints (Fig. 1). The sequences around the left and right deletion termini, as deduced from the junction point analysis, are presentedin Fig. 2. A map of the left endpoints is shown in Fig. 3. Several observations can be made. First, some deletions occurred more than once (Fig. 2, pGP5-92/110). Furthermore, several right endpoints (2/61; 10/108; 33/56; 92/l 10) occurred twice. Also severalleft terrnini occurred twice (8190; 34163; 66173; 92/110), or even three times (20/28/105). It appears that, despite the large variety in deletions, some sequences are preferred deletion targets. Second, the left deletion endpoints are located in the rather narrow region from nucleotides 20 to 665, in or just upstream of the penP regulatory signals (Fig. 3). This means that, without exception, penP promoter and/ or Shine-Dalgarno sequenceswere removed (the SD sequencestarts at position 668). These endpoints were not evenly distributed: clustering of at least four termini was observed in four small regions, defined by nucleotides 112 to 116; 339 to 347; 507 to 514; and 615 to 622, respectively. Third, although directly repeated sequences were present near many endpoints, deletions between direct repeats of more than 3 bp had occurred only in pGP5-56 (Fig. 4). The IO-bp direct repeat in pGP5-56 can even be extended to 21 bp, if two mismatches are included. Within this repeat an inverted repeat of 12 bp (including one mismatch) is present. Fourth, certain sequencepatterns were frequently observed at or near the other deletion termini: (a) At 15 (of 28) left termini cleavage had occurred within, or immediately next to, the sequence 5’-T-T-T-3’ or the 5’-A-A-A-3

26

PEIJNENBURG, BRON, AND VENEMA

A

pGP1 /DO/ypPG -35

;

0

A

E

H

S

C

-10 so

B

I

I

I I I

I I

I

I

I

I

!

I !

I

I

I

I

I

I

,

I

I

I I

I I

pGP5-X

A (kb)

26/70 56 83 65 33 57 90/b-6 71 80 69 6 63 96 15 30 34 56 lb 73 92/110 35/61 68/87 104 97/08/20 99 85/109/Z 4 42 38 108/79 105 50 IO/l1 84 28 8 115 13/27 49/111 86 102 95 12/18 43 78 39

0.15 0.17 0.22 0.25 0.27 1.70 2.00 2.40 2.50 2.60 2.70 2.80 3.00 3.10 3.15 3.20 3.30 3.40 3.45 3.45 3.50 3.55 3.55 3.65 3.70 3.80 3.85 3.90 3.90 3.95 4.00 4.05 4.10 4.35 4.40 4.60 4.60 4.65 4.70 4.75 4.75 4.85 4.90 5.00 5.05 5.10

FIG. 1. Location of deletions in the pet&IucZ region of pGP1. The upper part of the figure (A) shows the 5.5-kb BarnHI insert of pGP1. Numbers correspond to positions (in bp), the first nucleotide of the insert being 1. Restriction sitesare A, AafII; B, BarnHI; C, AccI; D, BraI; E, EcoRI; G, BgnI; H, BssHII; 0, OxuNI; P, PstI; S, SacI. In the lower part of the figure (B), the deleted regions in the various derivatives, denoted as pGP5 with a suffix, are indicated with horizontal lines. Vertical dashescorrespond to the restriction sites used for the fine-mapping of the deletion endpoints. The numbers in italics represent deletions of which the junction points were sequenced.O, penP sequences;I, lacZ sequences;-, B. Iicheniformis chromosomal DNA. The promoter and Shine-DaIgamo sequencesofpenP are designatedas -35, -10, and SD, respectively.

complement thereof (Fig. 2). In the 780-bp sequenceshown in Fig. 3,778 potential cleavage sites exist, 228 of which are within, or directly flank a T-T-T or A-A-A triplet. Given a frequency of 228/778 potential sites in or next to T-T-T or A-A-A sequences,the probability that, by chance, cleavage at such sites

would occur in 15 of 28 casesamounts to 4.2 X 10e3.This suggeststhat the high frequency of T-T-T or A-A-A sequencesat the left cleavage points might not be a matter of chance. (b) In 16 cases,the sequence 5’-T-G-T-A3’ could be localized within 10 bp from the left breakage point (Fig. 2). Interestingly, this

PLASMID DELETION FORMATION

r

27

LEFTpGP5----108 99 28 105 4

______________ 154~AAAGCGCCT 361-AGGTTTGCC lOl-TTTCGTCTA lOl-TTTCGTCTA lOl-TTTCGTCTA

EE 29 34 63 2 :i a4 90 a :i 66 ;: 110 68 102 12

3%TCATCCTATAC 451-CCGGTGGAAA BOZ-GCCATAACAC 255-AGAATGCTTC 44ZdATTTGTCCC 165-TTTAATGAGG _

FIG. 2. Nucleotide sequences(5’ to 3’) near deletion termini. The left and right sequencesrepresent the regions flanking the left and right deletion termini, respectively. Arrows indicate breakage points. Directly repeated sequencesat the endpoints are boxed. Breakage may have occurred at any position within the boxes. T-G-T-A sequencesare shown in small characters. Trinucleotide T-T-T or A-A-A boxes and the octanucleotide consensussequenceare indicated (0). The sequenceswere aligned according to these boxes. Nucleotide positions correspond to those in Fig. 1.

tetranucleotide is present in each of the 4 regions showing clustering of deletion endpoints (Fig. 3). T-G-T-A boxes were not frequent (5 of 28) near right deletion termini. A question of interest is whether the high frequency of TG-T-A boxes near left deletion termini might be more than a matter of chance. We estimated the probability that, by chance, any tetranucleotide sequence will occur in each of the 4 regions showing clustering of endpoints. The regions were considered to comprise 20 nucleotides (10 nucleotides to the left and to the right of the breakage points). This probability is ( 17/256)4, or 1.9 X 10w5.The probability that any of the 256 possible tetranucleotides will occur in the four stretches of 20 nucleotides is 4.9 X 10e3.This low probability suggeststhat T-G-T-A boxes might be involved in the generation of deletions in our system. (c) The right deletion endpoints did not reveal clear sequencepatterns. Only a loose sequence profile, consisting of four G or C followed by three A or T residues, and one A or G residue, could be identified in the immediate

surroundings of 15 right terxnini (Fig. 2). In this sequencethe trinucleotide G-G-A is preferred at the third, fourth and fifth positions. Near the left termini this octanucleotide was present only once (pGP5-79).

SecondaryStructuresin the penP Regulatory Region The clustering of left deletion endpoints in a narrow region of DNA containing the penP expression signals prompted us to study the potential of intrastrand secondary structure formation in this region. The regions studied were from positions 1 to 570 (Fig. 5A) and 560 to 780 (Fig. 5B). The thermodynamically stable structures corresponding to these regions, in which all 28 left deletion endpoints were located, have AG-values of - 127.3 (Fig. 5A) and -42.3 Kcal/mol (Fig. 5B). These structures reveal features of potential interest. Strikingly, the four most deletogenic regions containing the T-G-T-A boxes (I, II, III, IV in Fig. 5) are present near the tops of stem-loop structures.

28

PEIJNENBURG, BRON, AND VENEMA

BarnHI

DRlI 190 200 220 230 ----210 ATGGTTTGGATGGTTTTAGGGCTCCATGTACTGGTTTTGGATAACTCTTTAATTACCT~ 280 290 TGACTTTCATCACTTCTAATTCCGCATCAGAG

240 300

330 ATTTaGGTATTTTTTTCATTTTAATCATCCTATACTT@ &5:70 AGGT TGCCATTT

~",SO 390 400 410 AAGAAGTCAATTTAACTQGAGTGAACATGCAGTTTCTCGATTTTTA

420

ATTCCATCACTTCCCTCCGTTCATTTGTCCCC

580 550 560 570 590 ATTTGATGTTTGAGTCGGCTGAAAGATCGTACGTACCAATTATTGTTTCGTGATTGTTCA

600

610 m-(:3 AGCCATAACACTG AGGGA

660

6'"/8 630 640 650 A TGGAAAGAGTGCTTCATCTGGTTACGATCAATCAAATA

680 690 700 710 CWIRli~~AGACGATTTTG~AATTATGGTTCAGTACTTTAAAACTGAAAAAG LhI 740 730 750 760 770 GCTGCAGCAGTGTTGCTTTTCTCTTGCGTCGCGCTTGCAGGATGCGCTAACAATCAAACG --ET-

720

TTC

780

FIG. 3. Deletion end points in the penP regulatory region. The positions of the penP promoter (O), ShineDalgamo sequence(Bl), and translation initiation codon (a) were derived from McLaughlin et al. (1982). Arrows indicate junction points. In the boxed regions breakage may have occurred in between any of the nucleotides. T-G-T-A sequencesare indicated by broken lines. The sequenceis identical to that described for the corresponding parts of the B. lichenijbrmis strains 9945A (Himeno et al., 1986) and 749 (Nicholls and Lampen, 1987),except that the nucleotide pair G-A at positions 32-33 is a C-C pair in the latter strains.

DISCUSSION

In both E. coli and B. subtilisthe generation of deletions has been shown to be frequently associatedwith direct repeats (Albertini et al., 1982; Ehrlich et al., 1986; Jones et al., 1982; Lopez et al., 1984), and/or (quasi-)palin-

dromic sequences (Glickman and Ripley, 1984; Schaaper et al., 1986). The present studies show that such deletions constitute a minority classin a per&lucZ fusion on plasmid pGP 1 in B. subtilis.Whether this outcome is specific for the system used here is not known at present. In fact, only 1 of the 28

CCTATACTTACAA-ITTCAAA

left

__- - - - - ____- - --* 330-CCTATACTTACAA[Awi+TCATTAT --

right

____ - _________+ ATTTAAATCTTACAT,i,.ijTTCAAA-

525

PIG. 4. Sequencesat the junction point (top) and endpoints (bottom) of pGP5-56. The broken arrow representsa 2 1-bp direct repeat (including two mismatches).The solid arrows correspond to inverted repeats.

PLASMID DELETION FORMATION

29

FIG. 5. Optimal DNA folding in the 5’ region of the 5.5-kb BumHI fragment of pGP1, as predicted by computer analysis. The positions of the deletion termini are indicated with arrows. The tetranucleotide TG-T-A is indicated (Ca).(A) Positions I to 570; (B) positions 560 to 780; I, II, III, IV: deletogenic regions with multiple deletion endpoints at positions 112-l 16, 339-347, 507-5 14, and 6 15-622, respectively.

deletions analyzed had termini coinciding with direct repeats of more than 3 bp. In this deletion (pGP5-56), a direct repeat of 10 bp, flanked by a 12-bp inverted repeat (1 mismatch), was present at the deletion endpoint. Several other direct repeatsof 10 to 12 bp, not flanked by inverted repeats, were present in the 5.5-kb BarnHI insert of pGP1. None of these were found to be associated with the deletions studied here. This suggestsa role of the inverted repeatsin the generation of pGP556. That directly repeated sequencesflanked by inverted repeats favor deletion formation in other systems,such as the E. coli 1acIgene, has been suggestedby Glickman and Ripley (1984) and Schaaperet al. (1986). Also precise excision of transposons, a phenomenon re-

sembling plasmid deletion formation at direct repeats,is stimulated by inverted repeats,both in E. coli (Egner and Berg, 1981; Foster et al., 1981) and in B. subtilis (Peeterset al., 1988). Such deletions have frequently been considered to be the result of errors in replication, due to (“slipped”) mispairing of the direct repeats (Albertini et al., 1982; Ehrlich et al., 1986;Brunier et al., 1988;Peeterset al., 1988). pGP5-56 may well be caused by a similar mechanism. Regions of single-stranded DNA, assumedto promote slipped mispairing of the repeats,can easily be formed since, like many other plasmids in B. subtilis (te Riele et al., 1986a,b), plasmid pGP 1 replicates via singlestranded intermediates (our unpublished results).

30

PEIJNENBURG, BRON, AND VENEMA

Deletions not based on direct repeats are not easily explained by slipped mispairing mechanisms. This raises the question of how the majority of the deletions studied here might have arisen. So far, our data allow only speculations. We can conceive of at least two possible explanations for the observation that many left endpoints were located in the penP regulatory region. The first is that deletions preventing expression of the penP-la& fusion would be preferentially selected, possibly because cells carrying this kind of deleted plasmids would outgrow cells carrying the parental plasmid. The second, more speculative, possibility is that expression per sk would stimulate deletion formation. In this respect,it may be of interest that the region in which many of the left endpoints are located is divergently transcribed from the peril and penP promoters (Himeno et al., 1986; Nicholls and Lampen, 1987). Since pGP1 encodes inactive (truncated) Pen1repressor, expression of both peril and penP will be constitutive. It has been demonstrated that active transcription may affect superhelicity (Prussand Drlica, 1986),and an explanation was recently proposed by Liu and Wang (1987). These latter authors also postulated that in plasmid DNA the level of negative supercoiling is increased in between two divergently transcribed promoters. One could conceive of the possibility that increased negative supercoiling would stimulate the generation of appropriate secondary structures for deletion formation. The observation that many left deletion endpoints were associatedwith potential stemloop structures suggests that the cleavage specificity was, at least partially, determined by DNA secondarystructures. DNA structural specificity, resembling that observed here, has been described for E. coli DNA topoisomerase I (Kirkegaard et al., 1984). Although direct evidence is lacking, this suggeststhe possibility that topoisomerase I might be involved in the formation of deletions. In this respect,the observation that T-T-T and the complementary A-A-A trinucleotides were present at 15 of the left deletion endpoints may be of interest.

Bullock et al. (1985) and Yarom et al. (1987) have proposed that similar trinucleotides (Spy-T-T-3’, or the 5’-A-A-Pu-3’ complements) are preferred cleavage sites for eukaryotic DNA topoisomerase I. Also at bacterial topoisomerase I cleavagesites, T-T-T and A-AA sequenceswere frequently present (Kirkegaard et al., 1984). T-G-T-A and G/C-G/C-G/C-G-A/T-A/TA/T-A/G sequenceswere frequently observed in the vicinity of left and right deletion termini, respectively. The role, if any, of these sequences in deletion formation is not understood. We could not find similarities with postulated topoisomerase II recognition sites (Gellert et al., 1981; Lopez et al., 1984; Lockshon and Morris, 1985). Topoisomerase II (DNA gyrase) has been considered to be involved in illegitimate recombination events by severalauthors (Ikeda et al., 1981; Ikeda, 1986; Lopez et al., 1984). Our results, therefore, give no indication about a possible role of DNA gyrase in deletion formation. ACKNOWLEDGMENTS We thank Henk Mulder for preparing the figures, Peter Terpstra for assistancein the computer analysisof potential secondarystructures, Karl Drlica for carefully reading the manuscript, and Unilever ResearchLaboratories for synthesizing sequencing primers. This work was supported by the Netherlands Organization for the Advancement of Pure Research (ZWO) under the auspices of the Netherlands Foundation for Chemical Research (SON).

REFERENCES ALBERTINI, A. M., HOFER, M., CALOS, M. P., AND MILLER,J. H. (1982). On the formation of spontaneous deletions:The importance of short sequencehomologies in the generation of large deletions. Cell 9, 319-328. ANDERSON,P. (1987). Twenty years of illegitimate recombination. Genetics 115, 58 I-584. BRON,S., AND LUXEN, E. (1985). Segregationalinstability of pUB1 IO-derived recombinant plasmids in Bacillus subtilis. Plasmid 14, 235-244. BRON,S., AND VENEMA,G. (1972). Ultraviolet inactivation and excision repair in Bacillus subtilis. I. Construction and characterization of a transformable eightfold auxotrophic strain and two ultra-violet-sensitive derivatives. Mutat. Rex 15, l-10.

PLASMID DELETION FORMATION BRUNIER,D., MICHEL, B., AND EHRLICH,S. D. (1988). Copy choice illegitimate DNA recombination. Cell 52, 883-892. BULLOCK, P., CHAMPOUX, J. J., AND BOTCHAN, M. (1985). Association of crossover points with topoisomerase I cleavage sites: A model for nonhomologous recombination. Science 230,954-958. DASGU~A, U., WESTON-HAFER,K., AND BERG, D. (1987). Local DNA sequence control of deletion formation in Escherichia coli plasmid pBR322. Genetics U&41-49. EFSTRATIADIS,A., POSAKONY,J. W., MANIATIS, T., LAWN, R. M., O’CONNELL,C., SPRITZ,R. A., DERIEL, J. K., FORGET,B. G., WEISSMAN,S. M., SLIGHTOM, J. L., BLECHL, A. E., SMITHIES, O., BARALLE, F. E., SHOULDERS, C. C., ANDPROUDFOOT, N. J. (1980). The structure and evolution of the human j3-globingene family. Cell 21, 653-668. EGNER,C., AND BERG,D. E. (1981). Excision of transposon Tn5 is dependent on the inverted repeats but not on transposasefunctions of Tn5. Proc. Natl. Acad. Sci.

31

mediatesillegitimate recombination in vitro. Proc. Natl. Acad. Sci. USA 83,922-926.

JONES,I. M., PRIMROSE, S. B., ANDEHRLICH,S. D. (1982). Recombination between short direct repeats in a recA host. Mol. Gen. Genet. l&486-489. KALNINS,A., Oreo, K., RUTHER,U., AND MOLLER-HILL, B. (1983). Sequence of the IacZ gene of Escherichia coli. EMBO J. 2,593-597.

KOK, J., VAN DER VOSSEN,J. M. B. M., AND VENEMA, G. (1984). Construction of plasmid cloning vectors for lactic streptococciwhich also replicate in Bacillus subtilis and Escherichia coli. Appl. Environ. Microbial. 48,726731. KIRKEGAARD,K., PFLUGFELDER, G., AND WANG, J. C. (1984). The cleavage of DNA by type-l DNA topoisomerases. Cold Spring Harbor Symp. Quant. Biol. 49, 411-419. LIU, L. F., AND WANG, J. C. (1987). Supercoiling of the DNA template during transcription. Proc. Natl. Acad. Sci. USA 84,7024-7027.

LOCKSHON,D., AND MORRIS,D. R. (1985). Sites of reaction of Escherichia coli DNA gyrase on pBR322 in USA 18459-463. EHRLICH,S. D., NOIROT,P. H., PETIT,M. A., JANNIBRE, vivo as revealed by oxolinic acid-induced plasmid linL., MICHEL, B., AND TE RIELE, H. (1986). Structural earization. J. Mol. Biol. 181, 63-74. instability of Bacillus subtilis plasmids. In “Genetic En- LOPEZ,P., ESPINOSA,M., GREENBERG,B., AND LACKS, gineering” (J. K. Setlow and A. Hollaender, Eds.), Vol. S. A. (1984). Generation of deletions in pneumococcal mal genescloned in Bacillus subtilis. Proc. Natl. Acad. 8, pp. 71-83. Plenum, New York. Sci. USA 81,5 189-5 193. FOSTER,T. J., LUNDBLAD,V., HANLEY-WAY, S., HALLMANIATIS,T., FRITSCH,E. F., AND SAMBROOK,J. (1982). ING, S. M., AND KLECKNER,N. (1981). Three TnlOassociatedexcision events: Relationship to transposition “Molecular Cloning: A Laboratory Manual.” Cold and role of direct and inverted repeats. Cell 23, 215Spring Harbor Laboratory, Cold Spring Harbor, NY. 227. MCLAUGHLIN,J., CHANG,S.-Y., AND SHANG,S. (1982). FREIER,S., KIERZEK, R., JAEGER,J. A., SUGIMOTO,N., Transcriptional analyses of the Bacillus licheniformis penP gene. Nucleic Acids Res. 10, 3905-3919. CARUTHERS,M. H., NEILSON,T., AND TURNER,D. H. (1986). Improved free-energyparametersfor predictions MICHEL, B., AND EHRLICH, S. D. (1986a). Illegitimate of RNA duplex stability. Proc. Natl. Acad. Sci. USA 83, recombination at the replication origin of bacteriophage 9373-9377. M 13. Proc. Natl. Acad. Sci. USA 83,3386-3390. GELLERT, M., FISHER, L. M., OHMORI, H., G’DEA, MICHEL, B., AND EHRLICH, S. D. (1986b). Illegitimate M. H., AND MIZUUCHI, K. (1981). DNA gyrase: Siterecombination occurs between the replication origin of specific interactions and transient double-strand breakplasmid pC194 and a progressing replication fork. age of DNA. Cold Spring Harbor Symp. Quant. Biol. EMBO J. 5,369 l-3696. 33,39 l-398. NAITO, A., NAITO, S., AND IKEDA, H. (1984). Homology GLICKMAN, B. W., AND RIPLEY, L. S. (1984). Structural is not required for recombination mediated by DNA intermediates of deletion mutagenesis: A role for palgyrase of Escherichia coli. Mol. Gen. Genet. 193,238indromic DNA. Proc. Natl. Acad. Sci. USA 81, 512243. 516. NALBANTOGLU,J., HARTLEY,D., PHEAR,G., TEAR, G., HIMENO,T., IMANAKA, T., AND AIBA, S. (1986). NucleoAND MEUTH, M. (1986). Spontaneous deletion fortide sequenceof the penicillinase repressorgenepenl of mation at the aprt locus of hamster cells: The presence Bacillus licheniformis and regulation of penP and penl of short sequencehomologies and dyad symmetries at by the repressor.J. Bacterial. 168, 1128- 1132. deletion termini. EMBO .I. 5, 1I99- 1204. HOROWITZ,B., AND DEONIER,R. C. (1985). Formation NICHOLLS,N. J., AND LAMPEN,J. 0. (1987). Repressor of Delta-tra F’ plasmids: Specific recombination at oriT. gene, blal, for Bacillus licheniformis 749 /3-lactamase. .I. Mol. Biol. l&267-274. FEBS Left. 221, 179-183. IKEDA, H., MORIYA, K., AND MATSUMOTO,T. (198 I). In NEUGEBAUER,K., SPRENGEL,R., AND SCHALLER,H. vitro study of illegitimate recombination: Involvement (198 1). Penicillinase from Bacillus licheniformis: Nuof DNA gyrase. Cold Spring Harbor Symp. Quant. Biol. cleotide sequenceof the gene and implications for the 45,399-408. biosynthesis of a secretory protein in a gram-positive IKEDA, H. (1986). Bacteriophage T4 DNA topoisomerase bacterium. Nucleic Acids Res. 9, 2577-2588.

32

PEIJNENBURG, BRON, AND VENEMA

PEETERS,B. P. H., DE BOER, J. H., BRON, S., AND VENEMA,G. (1988). Structural plasmid instability in Bacillus subtilis: Effect of direct and inverted repeats.Mol. Gen. Genet., in press. PEIJNENBURG, A. A. C. M., BRON,S., AND VENEMA,G. (1987). Structural plasmid instability in recombinationand repair-deficient strains of Bacillus subtilis. Plasmid 17, 167-170.

PRUSS,G., AND DRLICA,K. (1986). Topoisomerase I mutants: The gene on pBR322 that encodes resistance to tetracycline affects plasmid DNA supercoiling. Proc. Natl. Acad. Sci. USA 83, 8952-8956. SANGER,F., NICKLEN, S., AND COULSON,A. R. (1977). DNA sequencing with chain terminating inhibitors. Proc. Natl. Acad. Sci. USA 74, 5463-5467. SAUNDERS,C. W., SCHMIDT, B. J., MIROT, M. S., THOMPSON,L. D., AND GUYER, M. S. (1984). Use of chromosomal integration in the establishment and expression of blaZ, a Staphylococcus aureus beta-lactamasegene, in Bacillus subtilis. J. Bacterial. 157,7 18726. S~HAAPER,

R. M., DANFORTH,

An analysis of the spectrum of spontaneous mutation in the Escherichia coli laclgene. J. Mol. Biol. 189,273284.

TE RIELE, H., MICHEL, B., AND EHRLICH,S. D. (1986a). Single-stranded plasmid DNA in Bacillus subtilis and Staphylococcusaureus. Proc. Natl. Acad. Sci. USA 83, 2541-2545.

TE RIELE, H., MICHEL, B., AND EHRLICH,S. D. (1986b). Are single-stranded circles intermediates in plasmid replication? EMBU J. 5, 63 I-637. WILLIAMS, A. L., AND TINOCO, I. (1986). A dynamic programming algorithm for finding alternative RNA secondary structures. Nucleic Acids Res. 14, 299-3 15. YANISCH-PERRON,

C., VIEIRA, J., AND MESSING, J. (1985).

Improved M 13 phage cloning vectors and host strains: Nucleotide sequencesof M 13mpI8 and pUC 19 vectors. Gene33, 103-119. YAROM, R., LAPIDOT, A., NEER, A., BARAN, N., AND MANOR, H. (1987). ‘Illegitimate’ recombination events

in polyoma-transformed rat cells. Gene 59, 87-98.

B. N., AND GLICKMAN,

B. W. (1986). Mechanism of spontaneousmutagenesis: Communicated by Kurt Nordstriim