Tobacco Documents Online

Mut~t~¢.~ Reze~c~ 214 (1~39) 233-252 EIs~r MUT 047~3 233 Overlapping direct repeats stimulate deletions in specially designed derivatives of plasmid pBR325 in Escherichia cell Elias Balbinder, Cheryl Mac Vean and Robert E. Williams Department of Biochemistry, Bio~h),sics and Genetics and Colorado Cancer Center, UnioersiO, of Colorado Health Sciences Center, 4200 E. 9th Av~, Box B.12~ .Denver, CO 80262, and Internaffonal Center.for Cancer Research and Developmental Biology, Denver, CO (U.S.A.) (Received 19 December 1988) (Revision received 6 March 1989) (Accepted 6 April 1989) Keywords: Deletions on plasmids; Plasmids, deletions on; DNA, slngle-stranded, misaligrtrnents; Chloramphcnicol acetyltransferase Summary Current misalignment mutagenesis models have identified certain sequences such as direct and inverted repeats, which can stabilize transient misalignments on single-stranded DNA, as major structural parame- ters for deletions. We have constructed derivatives of the plasmid pBR325 to investigate further the relative roles of such sequences under controlled conditions. The plasmid derivatives pOCE15, pRS1 and pRS4, were obtained by cloning fragments of the same approximate size (60-64 bp) but different sequence into the unique EcoR1 site of the ebloramphenicol aeetyl transferase (cat) gene of plasmid pBR325. The ins.eat of pOCE15 is a perfectly palindromic lac operator fragment. Both pRS1 and pRS4 carry the same non-palindromic fragment but differ from each other in the sequence at the right (3') end of the insert. Plasmid pRS4 differs from pRS1 by a 9-bp duplication containing an additional EcoR! site at the 3' end of the insert. This arrangement yields overlapping imperfect 17-18 bp and perfect 11 bp direct repeats at the deletion termini mad creates multiple opportunities for the staWflizafion of misaligned intermediates. The deletion rate, measured from the reversion of chloramphenicol sensitivity (Cms) to resistance (Cmr), was always highest in pRS4, intermediate in pOCE15 and lowest in pRS1, with an approximate ratio of 100:10 : 1. We have also obtained evidence for the participation of ReeA in the genesis of deletions in these pBR325-derived plasmids. Deletions are an important class of genetic rearrangements and have been found to be at the basis of a number of cancers and other genetic diseases (Yunis, 1983; Eseot et al., 1986; Monaco Correspondence: Dr. Elias Balbinder, Department el Biochem- istry, Biophysie~ and Genetics% B-121, University el Colorado Health Seie~zes Centex, 4200 E. 9th Ave., Denver, CO 80262 et al., 1985). They have been extensively studied at the molecular level, mainly in prokaryotes, and most of them are explained by misalignment mutagenesis models. These propose that deletions arise from the resolution of transient misalign- ments formed by slippage of single-stranded stretches of DNA during replication and stabi- lized by sequence homologies such as terminal direct repeats, intervening inverted repeats (palin- dromes) or a combination of both (Albertini ctal., 0027-5107/89/$03.50 © 19~9 Elssvier Se~ene~ Publishers B.V. (BiomeclL~al Division) 40000087

234 1982; Ripley and Glic'kman, 1983; Drake, Glick- man and Ripley, 1983; Glic.kmma and Ripley, 1984). Although the existence of tnisaligned struc- tures has not been demonstrated, these models are supported by a large body of indirect evidence; i.e. most reported deletions occur between direct repeats (Albertini et al., 1982; Jones et al., 1982; Meulien et at., 1982; Goldberg and Mekalanos, 1986; Das Crupta et al., 1987; Singer and Westlye, 1988), are facilitated by internal palindromes (A1- bertini et al., 1982; Foster et al., 1981; Das Gupta et al., 1987), can occur at the end of palindromes in the absence of direct terminal repeats (P, ipley and Glickman, 1983; Glickman and Ripley, 1984). In addition, it has been shown that palindromes are unstable in bacteriophages and plasmids (Col- lins, 1980; Das Gupta et al., 1987; a,Villiams and Muller, 1987) and are lethal to carrier plasmlds unless deleted (Hagan and Warren, 1983). Accord- ing to these models, deletion incidence should be determined by two major parameters: (1) the frequency of misalignment occurrence, and (2) the stability of the.misaligned structures. This predict- ion is supported by experimental evidence: dele- tion frequency between direct repeats decreased inversely with the distance between the repeats (Albertini et al., 1982; Singer and Westlye, 1988) and increased in direct proportion to the hydrogen bonding potential of the putative misalignements (Albcrtini et al., 1982; Williams and Mailer, 1987; Das Gupta et al., 1987; Singer and Westlye, 1988). While these models have identified certain DNA sequences as playing major roles in deletion for- marion, they have left entirely open one major question: how are the transient secondary struc- tures formed between direct and/or inverted re- peats resolved to )~eld deletions? There has been considerable speculation about the mechanisms responsible for deletions, but these have not yet been defined. Among those proposed are "ille- gitimate" recombination (Franklin, 196'0, site- specific recombination mediated by gyrases and toposiomerases (Anderson, 1987), homologous re- combination (Anderson and Roth, 1977; Singer and Westlye, 1988), misreplieation across the base of transient deletion intermediates (Albertini et al., 1982; Gliekman and Ripley, 1984; Drake et al., 1983; Dos Gupta et al., 198"0, and nuclease excision of protruding hairpin or unpaired loops (refs. above). A consensus seems to be developing that deletions are a heterogeneous class of re- arrangements and can be brou~ht about by a variety of pathways involving both DNA repli- cation and recombination mechanisms (Anderson, 1987; Das Gupta et al., 1987; Singer and Westlye, 1988) and it has been suggested that the same deletions can be the produets of different mecha- nisms (Balbinder, 1988). "Illegitimate" recombi- nation was originally proposed by Franklin (1967) based on the observation that spontaneous dele- tions in bacteriophage lambda were equally pres- ent in recA + and recA - cells, and while this was subsequently found to be true for a variety of other deletions in plasmids and the E. colt chro- mosome (Inselburg, 1967; Anderson, 1970; Coukell and Yanofsky, 1970; Foster et al., 1981; Collins, Volckaert and Nevers, 1982; Jones, Prim- rose and Ehrlich, 1982; Das Gupta, Weston-Hafer and Berg, 1987), it was also reported that deletion frequency could be erthaneed by large (Sommer et al., 1981; Albertini et al., 1982) or small but reproducible factors (Williams and Mtiller, 1987) by recA +. Thus, some deletions are stimulated by the RecA protein and others are not. In this report we describe the construction of derivatives of plasmid pBR325 obtained by insert- ing fragments of the same size but different se- quence into the unique gcoR1 site in the cat (chloramphenicol acetyltransferase) gene of this high copy number plasmid. These pBR325 deriva- tives were used to (a) determine the effect on deletion incidence of eertaiax sequences within lind around the insert, and (b) to determine whether rec.4 plays a role in the production of these dele- tions. Our results suggest that, while deletion potential does indeed depend on base sequence at and around the deletion site, deletion frequency may sometimes be a direct result of the stability of a misaligned intermediate and other times an ex- pression of the frequency at which such inter- mediates arise. 3Ne also found evidence for the participation of ReeA ~ the formation of dele- tions in pBR325-derived plasmids. ]'vlateriais and methods CultUre media L-broth and L-agar were prepared according to standard recipes (Silhavy, Berman and Enquist, 400OO088

1984). To detect colonies with constitutive lac operon function, a diagnostic feature for plasmids carrying lac operator inserts, we used X-Gal plates (Smith and Sadler, 1971), prepared as described by Betz et al. (1986). Chloramphenicol and ampl- cillin (Sigma) and tetracycline (Aldrich) were ad- ded when necessary at final concentrations of 25 /~g/ml. Se/ection of Trp+ revertants was on minimal medium E (Vogel and Bonner, 1956) supplemented as indicated in Table 3. Bacterial strains and plasmids The strains employed in this investigation are listed in Table 1. Plasmids were introduced by transformation using the procedure of Sadler et al. (1977). All plasmids employed in tiffs investigation were constructed as described under Results. Preparation of plasmid DNA for restriction enzyme analysis and sequencing The procedure employed to obtain large amounts of plasmid DNA was essentially the one described by Sadler et al. (1977) with minor mod- ifications. 3-liter L-broth cultures were gown in a small fermentor at 37°C with the addition of chlorampherdcol to amplify synthesis of plasmid DNA (Hersb_field et al., 1974). Cleared lysates were prepared by the method of Clewell and FIelinsld (1969). After phenol extraction and ethanol precipitation, the DNA was resuspended in TEN buffer (10 mM Tris, 1 mM EDTA, 50 mM NaCI, pH 8.0), brought to 51.5% (w/v) CsC1 and 200/~g/ml of ethidium bromide, and centrifuged in a type Ti-50 fixed angle rotor at 43 000 rpm for 48 h. The lower, brightly fluorescing (in UV light) plasmid band was withdrawn with a syringe. Ethidium bromide was removed by several extrac- tions with n-butanol, followed by dialysis vs. 1/2 strength TEN buffer and ethanol precipitation. For rapid plasmid screening we used the proce- dure of Birnboim and Doly (1979). In some experiments (see tex0 where we wanted to establish that Cmr revertants had lost the in- serted fragment from the EcoR1 site, we harvested all revertants recovered in a single experiments by washing the selection plates with 5 ml of L-broth containing chloramphenieol, using this to inoc- ulate 1 liter of the same medium and, after grow- ing overnight at 37°C with aeration, harvesting 235 the cells and purifb4mg the plasmids. This proce- dure allov;ed us to sample the entire plasmid population in revert,ant cells. Since pBR325 has a colE1 origin of replication and thus a high copy number (ClewelL 1972; Bolivar and Backanan, 1979) and the revertants selected had a dominant phenotype (Cmr), this population consisted of both mutant and revertant plasmids. Restriction enzyme analysis and gel electrophoresis The restriction endonueleases EcoR1, Alul, HindlI and gaelII were purchased from Boeh- ringer-Mannheim and used according to the manufacturer's directions. Plasmid DNA was anal3~zed in horizontal 0.8% agarose gels (Sigma), and restriction enzyme digests on 5 or 15% aerylamide gels (see text). The buffer was 89 mM tris borate, 2 mM EDTA, pH 8.3 (Maniatis, Fritseh and Sambrook, 1982). EtlMdium bromide (60 t~g/ml) was added to the agarose gels and glycerol (10%) to the aerylamide gels. Gels were run at 7-10 V/cm and DNA bands visualized under ultraviolet light illumination and photographed on Polaroid Type 55 film. Acrylamide gels were stained with ethidium bromide (4/~g/ml in dec- trop.horesis buffer) prior to DNA visuatization. Sequencing Sequencing was carried out by the dideoxy chain-termination procedure of Sanger, Nicklen and Coulson (1977). AluI restriction fragments to be sequenced (see rex0 were eluted from prepara- tive aerylamide gels (Maxam and Gilbert, 1980), ethanol precipitated, resuspended in TEN buffer or 10 mM Tris, pH 7.4 and cloned into the Hinc H site of digested double-stranded (RF form) M13mpl8 or M13mpl9 (Messing, 1983) by blunt end ligation with T4 DNA ligase (Collaborative Research). Following transformation of strain JM103 (ref. above) single-stranded M13 DNA was recovered from infective phage in the culture fluid and used as a template for the sequencing reac- tions. These were carried out using the triphos- phate mixture kit of New England Biolabs and [35S]ATP purchased from Amersham Corp. Se- quencing acrylamide gels were run on a Sequl-Gen nucleic acid sequenelng cell (BioRad) according to the instructions of the manufacturer. 40000089

236 TABLE I BACTEP, IAL STAINS USED IN THIS INVESTIGATION Strain Relevant genotype Source HB101 re~413 hsdS20(rBmB) lac )7 Boyer and Roulland-Dussoix 1£69 D1206 recA ÷ revertant of HB101 Betz and Sadler, 1£81(a) E,KI'945 ~recA306 srlC3OO::TnlO/Ap reeA ÷ art +el~ Tessman and Peterson, 1£85 SC30-V197 recM 730 trpE65 uurA155 lonl I sulal E. Wilkin SC30RP-V260 recA + srlC3OO::TnlO alaS3 trpE65 v.vrA153 Ionll E. Witldu EB301 srl¢ +, derived from SC30RP-'V260 by loss of Tnl 0 This laboratory EB305 as EB301, but z~recA306 srlC3OO::TnlO, This laboratory by transduction of EB301 with P1 phage grown on EST 945 EB252 HB101/pRS1 This study EB244 HB101/pOCE15 This study EB250 HBI01/pRS4 This study EB259 D1206/pRSI This study EB246 D1206/pOCE15 This study EB261 D1206/plL.q4 This study EB306 $C~I0-V197/pOCE15 This study EB309 EB301/pOCE15 This study EB312 EB305/pOCE15 This study EB496 $C30-V197/pRS1 This study EB497 EB301/pR$1 This study EB498 EB305/pRS1 This study EB499 SC30-V197/pR.S4 This study EB500 EB301/pRS4 This study EB501 EB305/pRS4 This study TABLE 2 Cms ~ Cmr DELETION RATES AND FREQUENCIES Strain Plasmid Oenotype Number of Number of Po N m R Deletion frequencies cultures culture~ ×10s x10-9 X10-9 with Be Cmr EB252 pR$1 recA 13 85 78 (*) 0.92 2.6 0.086 0.3 (*) 0.6 EB244 petE15 re.cA 13 85 5 0.06 6.0 1.7 2.8 EB250 pRS4 reeA 13 85 0 < 0.01 1.2 6.9 58 260 EB259 pKSI rec~÷ 106 79 (*) 0.74 2.4 0.29 12 2.4 EB246 pOCE15 tee& ~ I00 II 0.11 2,1 1.7 8.1 33 EB261 pR$4 recA÷ 96 0 < 0.01 2.5 18.7 74.8 362 Ddetion rates (R) were ealculatezl by the method of Lufia and Delbruek (1943) as modified by Lev. and Coulson (1949) (see Materials and methods). "Fne values represent the frequency of deletion per cell per generation and were obtained from the expression R = m/N, where ra is the mean number of mutatio~x~ per sample and N is the final viable cell count per euhure. When most of the cultures contained no Cmr revertants (*) the value of m was obtained from the zero term o.~ the Pois~on distribution (P0 = e-'n)• in all other eases m was determSr~ed from the median of the distribution of the number of erar revertants per culture (re)- By this procedure, m can be calculated from the expression ro/ra using the values given in Table 3 of Lea and Coulson (1949). N was always obtained by direct viable counts. Deletion frezlUeaeies represent the number of Cmr revertants per number of cells plated on selective agar, and were calculated from the average numb~ of revertants in the entire sample of 85-106 cdtures divided byN. 40000090

Measuring deled'on rates and frequencies Deletions were measured by r~'er~on from Cms -> Cm~ (chloramphenicol sensitivity to resis- tance). Deletion rates were calculated by the method of Luria and Delbruck (1943, see legend to Table 2). In this procedure, a large number (85-106) of 0.5 ml cultures were seeded with a few hundred ceils (0.1 ml of a 10-6 dilution of an overnight culture), the cultures were grown over- night without shaking at 37~ C, and the entire contents of each culture plated on L broth plus chloramphenicol plates. These were incubated 24-48 h and the number of revertants on each plate counted. Deletion frequencies were obtained as described in the legend to Table 2. Results Plasmid construction and the analysis of revertants The work which we report here was carried out with derivatives of pBR325 obtained by inserting two fragments of the same approximate size (60-64 bp), but different sequence, into the unique EcoR1 site in the cat gene of this plasmid. The cat gene has been completely sequenced (Alton and Vap- nek, ] 979), and the unique EcoR1 site is located within an AluI fragment of 129 bp (Fig. 1A), which is easily indentified in AluI-EcoR1 double digests of pBR325 (Fig. 2A, e and d). Insertion of a DNA fragment within the gcoR1 site inactivates the cat gene giving a chloramphenieol sensitive (Cm~) phenotype and generates two EeoR1 sites at each end of the insert (Fig. 1B and C). Plasmid pOCE15 was constructed by Betz and Sadler (1981a) and contains the 60 bp inverted repeat of a lae operator fragment. This is a palindromic sequence and is capable of forming a hairpin structure as illustrated in Fig. l-B1 and shown in vitro by Sinden, Broyles and Pettijotm (1983). The use of lac operator sequences as palindromic DNA provides an additional ad- vantage in giving us a visible marker. Ceils con- raining multiple copies of the lax: operator on plasmids titrate repressor off the chromosomal/ac operon and produce ~-galactosidase resulting in blue colonies on X-Gal plates. Loss of functional lac operators gives normal, white colonies (Betz and Sadler, 1981a, b). Plasmids pRS1 and pKS4 were obtained as independent transformants in 237 the same experimenq when the 64-bp non- palindromic HaelII fragment of PBR322 with EcoR1 Ym.kers added was inserted into the unique EcoR1 site of pBR325 (Sinden, personal com- munication, see legend to Fig. 1). This fragment is not palindromic (Fig. l-B2). Although both pRS1 and pRS4 contain the same non-palindromic in- sert in the same orientation they differ from each other in the sequences of the insert termini (Figs. 1-C and 3). In contrast to pRS1, where the insert is flanked on each side by the same 8-bp sequence which includes an EcoRI site, the insert of pRS4 carries an additional copy of the 8bp repeat on the 3' side as a consequence of a 9-bp duplication. Also, because of the orientation of the insert, the terminal homologies include adjacent sequences to the left of each EcoR1 linker. In pRS1 this gener- ates terminal 17-bp repeats with a 15/17-bp ho- mology. In pRS4 the 9-bp tandem duplication creates a second perfect ll-bp direct repeat which overlaps with the imperfect 18-bp one (Fig. 1C). In addition, there is an extra G added to the right hand 17-bp repeat so the homology becomes 15/18 bp at this position. These differences in terminal sequences between pRS1 and pRS4 have dramatie effects on the rate of deletion, as we will see. The extended fight hand 17-18-bp homologies of pRS1 and pRS4 have 2-3-bp substitutions which can result in a Cm~ phenotype, so deletions which retain the entire right terminus of the insert will not be recovered by our selection system. This will be discussed later. AIuI digests of the derived plasrnids showed in acrylamide gels that the 129-bp AluI fragment of pBR325 had been replaeed by larger fragments: about 195 bp in pOCE15 and pRS1, and 204 bp in pRS4 (Fig. 2). The 9-bp difference between the inserts in pRS1 and pRS4 is dearly resolved on 5% and 15% aerylarnlde gels (Fig. 2B and C). Since these plasmids have been designed for rapid measurements of deletion frequency by a simple reversion test, it is important to establish that selection for a Cm~ phenotype is always strin- gent for the restoration of the wild-type cat gene. For this to be true, Cmr revertants would have to satisfy the following 5 criteria: (1) they should be fully resistant to ehloramphenicol, (2) they should have lost the insert at the gcoRI site, (3) this loss should coincide ~th the disappearance of the 40000091

195-204-bp A/u/ fragmv.nt and th~ reappr.a.rance of the 129-bp fragment of pBR325, (4) they should contain a single EcoRl site, (5) the wild-type sequencc at and around the EcoR1 s~tc should be restorcd. All 5 criteria were satisfied without ex- ception. On the basis of work done by others, we know that the two codons contained in the EcoRl site of c~zt are e~scnd~1 for a functional chlo- ramph~Licol acctyltz~nsfer~se, since loss or p~xtial alteration of the EcoRl sequence gives a Cm~ phcnotypc (Sasmor, personal communication, and our unpublished observations). There is a certain de~ce of flexibility in the general re,on of the EcoP, J site, however, since Betz and Sadlcr (1991b) rcportcd that the addition of 13 extra amino acids A 10 20 30 40 ZiO 60 70 80 A......~.~,TAT TACGGCCTrT TTAI~GACCG T.~.Af.~AJ~ TAAGCAtJ~,G T'TrFATCCGG CCTTI"A'rI'C.Jt CA"I'TCTfG4:C 90 11~0 .llO 120 130 CGCCTGATGA ATGCTCATCC G~CGT ATGGCAATGA A.'tGACGGTG AGC___T a ~._dCA~AO ~0 30 4O SO eo .70 A~TTG'T~ATC CGCTCACAAT TCCACATGTG GAATI'GTG.~G CGGATAqCAA 40 50 60 70 TGTGGAATTGTGAGCGGATAACAATTTG~ ACA CCTTA~,CACT C GCCTA'rrG'ri'AAACA~ 30 20 10 10 20 30 40 $0 60 70 ~CGG GGGACTGTTG GGCGCCA'TCT CCTTGCATGC ACCATTCCTT GCGGCP-GCGG TGCTCA~CGI~ -- 10 20 30 40 50 60 70 b ~'~'~CC~,T TGAe..-CACCGC Cr~CCGCAAGG ART~'TGC~'T (~C.I~.GGA~,T G~CfiCCCAAC AGTCCCCCG~A'-~ pRS1 C ]0 ~ __ 20 30 40 50 60 7{} ,80__ 90 TGCTCATCCG ;G~c~.~C~CCGG GGGACTGTTG GGCGCCATCT CCTTGCATGC ,~CCATTCCTT GCGGC6GCGG TGCTCAACGG ~GTA TGGCA,~I' pRS4 rev.pEB8 10 __ 20 3D 40000092

A b c d ~ B ab C abcd ef 195~ 129. I00~ 66-- Fig. 2. Acrylamide gel ¢lcctrophoresis of plasmid digests. (A) Single (A/u/) and double (Alul-EcoRl) digests of plasmids pOCE15 (a and b) and pBR325 (c and d), run on a 5%, 17 x22 cm acrylamide gel. (a) pOCE15, Alul digest; (b) pOCEIS, AIuI-EcoRI digest; (c) pBR325, AluI digest; (b) pOCEIS, AluI-EcoRI digest; (c) pBR325, Alu~ dis~st; (d) pBR325, Alul-.Ecol~J digest; (e) ~X174 Haelll molecular size ladder. Note that EeoRI digestion of AluT-dig~sted pOCE15 leads only to the disappearance of the 195-bp rcplac~ment fragment and the appearance of a 100-bp and a 66-bp fragment. The form~ corze~ponds to the segment between the first AluI and the ~coRl sit~s (Fig. 1A), the second is the 66-bp insert. Double digests of pBR325 (d) show the disappearance of the 129-bp Alul fragment and its replacement by the 100-bp Almr-EeoRl fragment also found with pOCE15, but the 66-bp inse~ cannot be se~n. (B) AluI digests of plasmids pRfil (a) and pRS4 (b) run alongside each other on a 5% acrylamide, 17 × 22 era gel. The insert-carrying fragment of pRS4 runs behind the one from pRS1. The difference in size between these fragraen~.~ is a 9-bp duplication con~alnlng one EcoR1 linker in pRS4 (Fig. 1C). (C) AluI digests of Cms plasmids and two Cmr revertants ran on a 17×22 era, 15% a~rylamide gel. (a) pOCE15, (b) pRS4, (c) pRS1, (d) pOCE15 rev.pEB1, (e) pOCE15 rev.pEB8 (dlm~r), (0 OX174 molecular size ladder. This gel increases resolution for the smaller fragraenls over the 5% gel (compare separation of the 118- and 129-bp fragra~nts with Fig. ZA). Both revetments of pOCE15 have lost the insert and recovered the 129-bp fragment, but pEB8 (e) is a dimer consisting of one Cmr and one Cm~ plasmid and thus also shows the presence of the 195-bp fragment carying the cloneai insert. Fig. 1. Relevant DNA sequence. (A) Sequence of the 129 bp Alul fragment of pBIL325 containing the unique EcoPJ site (boxed). AluI sites are underlined. (B) Sequences of fragments cloned into the EcoR[ site of pBR325 generating at least one such site (boxed) at each end. (1) Sequence of the lac operator fragment in plasmid pOCElS. Th~s plasmid was constructed and sequenced by Betz and Sad]et (19gla). The cloned fragment has two lac operator sequences forming an inverted repeat 60 bp long (72 bp including the EcoRI sites). This constitutes a perfect palindrome with a 36-bp stern, as shown. (2) Sequence of the 64 bp HaellI [ragment of pBR322 plus added terminal EcoRl linkers used to ennstruet plasmids pR$1 and pRS4. Both orientations are shown. This fragment is not palindromic. (C) Relevant sequences o1~ plasmids pRS1, pRS4 and revertant pEB8 (see Fig. 3) showing presence of inserted fragment (pRS1 and pRS4) and its absence (pEB$). Plasmids pRS1 and pRS4 were constructed by blunt end ligation of oetamer EcoRl linkers to the tYaelll 64-bp fragment of pBR322 followed by EcoRl digestion to generate active EcoRI terminals, and cloning into EcoR[ linearized pBl~25. Each was recovered as an independent isolate. Note that in both plasmids the insert is in the same orientation (B-2a, above). As a result terminal homologies which include adjacent sequences arc generaled, as follows. For pl{$1, in addition to the 8-bp direct repeats containing the 6-bp Eco2~1 sites there is a 9-bp sequence extending leftwards from each EcoR1 terminus giving 17-bp direct repeats with a 15/17-bp homology. These sequences are underlined with a broken llne and the mlsmatehes indicated by asterisks. P]asmid pRS4 differs from pRS1 by a 9-bp tandem duplic-afion which includes the 8-bp sequence containing the EcoP.-I site and adds an extra G to the 3~ end of the in.~.ert. This creates an extra mismatch in the right hand 17-bp ~tended dkeet repeat and the extent of homv!ogy now becomes 15/18 bp (broken underlining). In addition, the 9-bp duplication creates a second set of ll-bp direct repeats with perfect (11/1 l-bp) homology' (solid underlining). These overlap with the 17-18-bp imperfect direct repeats. The mismatches are, again, indicated by asterisks. Deletlons in pRS1 and pRS4 which retain the tight-hand 17-18-bp d~reet repeat V.~ll result irt a Cm~ phenotype and are not recovered by our selection s~stem. From the sequence of the cat gene (Alton and Vapnek, 1979) and its position on pBR32.5 relative 1o the origin of replication (Soberon et al, 1982) we know that the polarity of transcription is fi'om left to right and that of replication is ~rom right to/eft as the sequences are shown. 40000093

A E, C D ACGT I CATGC A G G T ACGT TG GAGTO CGGG TCCG GT AT G CG Fig, 3. Sequencing gels for pRS4 (both strands, A and B), pRSI (C'), and revertant pEB8 (D). For the latter, the sequenc, is from the 129 bp AluI fragment. Note the two tandem EcoRl sites at the 3" and of the insert in pRS4 vs. a single EcoR1 site at the same position in pRSL between a -1 and a +1 frameshift directly up- stream of EcoR1 could be tolerated resulting in a partially acdve enzyme. In this case, howevex, the bacteria were only partially resistant to the drug: a wild-type cat gene makes bacteria resistant to 200 ~tg/mI of chloramphenicol while the addition of 13 amino acids in Betz and Sadler's plasmid re- sulted in resistance to only 25-50 #g/ml and sensitivity to 200 /~g/ml (ref. above). In our ex- periments Cmr revertants were always selected for resistance to 25 #g/m] of chloramphenicol and tested for resistance to 200 /~g/ml by replica plating on L-near containing that concentration of the drug. All revertants were tested and found io bc resistant to ~e ~gh concentration of chlor- amphenicol. In the analysis of revertant plasmids, we have to take into account the high copy number of pBR325, which has been estimated at about 24 per cell (Clewell, 19"/2; Bolivar and Ba -ckman, 1979). Since Cm~ is a dominant phenotype, Cm' re- vertant cells should contain about 20 copies of mutant Cms to one copy of revertant Cmr plas- mid. Recovery of cells carrying revertant plasmid exclusively was possible with pOCE15, since the cloned insert is a lac operator fragment whose presence or absence can be detected visually on X-Gal plates, but not with pRS1 or pRS4. Most Cmr revertants of pOCE15 gave blue colonies on X-Gal plus chloramphenieol agar, but these segregated stable white colonies when restreaked on the same medium after growth in liquid. Occa- sionally, however, we obtained stable blue Cmr revertants which carried dimers consisting of one copy of mutant (Cms) and one copy of revertant plasmid (Cmr). White colonies carried only re- vertant plasmids. AIuI digests of these showed that, as expected, the 195 bp fragments had been replaced by the 129-bp fragment of pBR325 (Fig. 2C-d). AluI digests of dimer plasmids showed both the 195 bp as well a.~ the 129 bp fragments (Fig. 2C-e). Acrylamide gel electrophoresis of 40000094

a b c d ~. 4. Alul di~ests of plasmids recovered from total revertant populations (see text) dcctrophorcsed on a ]5~, 17X22 cm acrylamidc gcl. (a) revertants of pOCE15, (b) revertants of pRS4 (30 pg), (c) same, but only I0 pg, (d) mixturo of (a) and (c). Because of the hight copy number of pBR325 we expect revertant populations to carry both Cms and Cmr plasmids. The presence of the 195-20d,-bp replacement fragment is used here as a built-in control to show that fragments which differ by 9 bp can be easily resolved. EcoR1 digests of mutant and revertant plasmids sho~ved that all the mutants carded the fragments while revertants had lost it, except in the case of directs, and all revertants contained intact single EcoR1 sites (except dimers, which had three). Since pRS1 and pRS4 do not car~ fragments which provide a visible marker, Cmr revertants were analyzed by acrylamide gel electrophoresis. AluI digests of plasrnid preparations obtained from cultures of all the Cmr revertants recovered in one experiment (see Materials and Methods) and con- taining both mutant and revertant plasmids show both the wild-type 129 bp and the 195-204 sub- stitute bp fragments, while digests of mutant plasmids show only the latter (compare Fig. 4 to Fig. 2). Since plasmid pRS4 carries an extra copy of EcoR1 at the 3' end of the insert, it is conceiva- ble that some revertants of this plasmid may re- tain two tandem EcoRl sites. Such plasmids should show an Altar fragment of 137 bp in 15% 241 acrylamide gels. As we see in Fig. 4, in mixtures of digests from pOCE15 and prS4 revertants (lane d) the 195-bp and 204-bp insert-carrying fragments are dearly separated, and we see even better sep- aration between the smaller llS-bp and 129-bp fragments. Thus, a 137-bp fragment would be detected in these gels if present in any significant amount. As shown in fig. 4 we never observed a 137-bp fragment in these revertant plasmid prep- arations but always found only the wild-type frag- ment of 129 bp, in agreement with the expectation that we will always recover the results of deletion events which restore the wild-type cat gene. Neither in pOCE15-pRS4 mixture (lane d) nor when we used 3 times the amount of digested plasmid from pRS4 revertants (lane b) were we able to detect any trace of a 137-bp band. This means that, if such a band was present, it would have been less than 4% of the 129-bp band. A deletion event in pRS4 which would leave behind two tandem EcoR1 sites would probably go unde- tected, because either the 3-bp substitution on the fight 18-bp repeat or the frameshift created by the two C-G-G-C bp separating the EcoR1 sites (Fig. 1C) would result in a Cms phenotype. Also, from preliminary experiments (unpublished) in which we inserted extra l~coRl linkers into the l~coR1 site of pBR325, it seems that a tandem duplication of this sequence is not tolerated in any case. The possibility that deletions in our pBR325- derived plasmids could leave behind an altered EcoR1 site was tested by aerylamide gel eleetro- phoresis o~ AluI-EcoR1 double digests of plas- raids recovered from whole revertant populations. This would show up as the incomplete digestion of the 129 bp and complete digestion of the 195-204 bp AluI fragments by EcoR10.e., see Fig. 2A). In all cases the 129-bp fragment was totally digested showing that the overwhelming majority, if not 'all, of revertant plasmids have an intact ~EcoR1 site. Finally, purification o~ the 129 bp fragment from the AluI digests of these revertant plasmid prep- arations and its subsequent cloning into M13mpl8 or M13mpl9 for sequencing, would give us a sample of fragments derived from different re- vertant pla.,~nids. If a sequence different from that of the wild-type cat gene were to occur w~th any significant frequency it may show up. We have 40000095

242 sequenced about 30 such fragments from all re- vertants and thus far have found no sequence different in any respect from the ~qd-type (Fig. 3D). In conclusion, all the evidence is consistent in showing that reversion to Cmr always restores a normal, wild-type cat gene. We have found no indication of the presence of pseudorevertants. Measurement of deletion rates and frequencies We used the plasmids describes in the preced- ing section to ask two questions: (1) what is the effect on deletion frequency of the sequence dif- ferences of each plasmid, and (2) does recA func- tion affect these frequencies to a detectable de- gree. All plasmids were introduced into HB101 (recA-) and its recA + derivative D1206 (Table 1) by transformation, and deletion rates and'frequen- cies determined for all strains thus obtained (Ta- ble 2). We always found that deletions were highest in pRS4, intermediate in pOCEI5 and lowest in pRS1 with an approximate ratio of 100:10:1. These results were consistent and reproducible also in many experiments when deletion frequen- cies were determined by simply plating aliquots from overnight cultures on selective agar (data no shown). Since secondary structures formed by misalignment can interfere with plasmid rep- lication (La Duca et at., 1983; Baumel et al., 1984; Weaver and De Pamphilis, 1934) it is possible that the observed differences in numbers of Cm~ re- vertants were caused by a more rapid replication of revertant over mutant plasmids. This possibility can be eliminated by comparing the sizes of the respective P0 fractions (fractions of cultures with no Cmr revertams) in a fluctuation test performed with liquid cultures (Table 2). Differential plasmid replication would not affect the size of the P0 fraction. Similar results were obtained with fluctuation tests performed on solid medium by the procedure of Das Gupta et al. (1987). Table 2 also shows that, while the occurrence of deletions in the 3 plasrnids does not require the participa- tion of recA÷, both the deletion rates (R) as well as the average deledon frequencies for each plas- mid were consistently higher by a small factors (between 1.5 and 4) in recA ~ than in reeA- cells. The fact that these differences between recA + and recA- were reproducible and observed indepen- dently with each plasmid suggested that recA+ was a participant in the processes giving rise to TABLE 3 EFFECTS OF DIFFERENT recA ALLELES ON DELETION FREQUENCY (Cm* .o Cmr) Strain Plasmid RecA (Cm~ ~ Cmr) trt~E - --. trpE ÷ Freq. × 10-~ Number of revertants/plate SEM~+} SEM+A~*) E1~96 pRS1 recA730 3.6(3.3) 89 (19) 196 (30) EB497 pRSt recA * 1.5 (1.4) 15 (3.2) 25 (3.8) EB498 pRS1 AreeA306 1.1 (1) 4.7 (1) 6.5 (1) EB306 pOCE15 recA730 128 (3.9) 104 (12) 202 (25) EB309 pOCE15 reeA * 44 (1.3) 10 (1.2) 16 (2) EB312 pOCE15 ArecA306 33 (1) 8.6 (1) 7.8 (1) EB499 pRS4 recA730 1433 (3.6) 32 (3.2) 57 (6.3) EB500 pRS4 recA+ 670 (1.7) 11 (~) 20 (2) EBS01 pRS4 Arec,4306 402 (1) 10 (1) 9 (1) (+) SEAM, Bonner+Vogel minim,-d medium E enriched with L-broth (20 ml/liter). (*) SEM+A, SEM+ adenine (75 ~g/ml). To obtain these deletion frequencies, overnight L-broth cultures were started from single-colony inoculation, 0.1-0.5 ml aliquots plated on L-agar containing 25/~g,/ml of chloran~ph-'w,.ico! and 0.1-ml aliquots of an appropriate dilution plated on L-agar to obtain a viable count. Reversion frequeney is ex."pressed as the number of Cmr revertants per number of viable cells plated. Relative frequencies are shown in parenthes~. These results are the average of 3-5 independent determinations. Frequencies varied by less than a factor of two. To obtain Trp÷ revertants, 0.1-ml aliquots of each culture were plated on SEM and SEM+A agar. The plat~ were incubated at 30°C and counted. The numbers sho~m in the table are the average number ofTrp+ revertants per plate O,Vitldn et al., 1982). 40000098