JOURNAL OF BACTERIOLOGY, June 1980, p. 992-1003 Vol. 142, No. 3 0021-9193/80/06-0992/12$02.00/0

Recombination Between Bacteriophage Lambda and Plasmid pBR322 in Escherichia coli KAY . POGUE-GEILE, SHILADITYA DASSARMA,t STEVEN R. KING, AND S. RICHARD JASKUNAS* Department of Chemistry and the Program in Molecular, Cellular and Developmental Biology, Indiana University, Bloomington, Indiana 47405

Recombinant A phages were isolated that resulted from recombination between the A genome and plasmid pBR322 in Escherichia coli, even though these deoxyribonucleic acids (DNAs) did not share extensive regions of homology. The characterization of these recombinant DNAs by heteroduplex analysis and re- striction endonucleases is described. All but one of the recombinants appeared to have resulted from reciprocal recombination between a site on A DNA and a site on the plasmid. In general, there were two classes of recombinants. One class appeared to have resulted from recombination at the phage attachment site that probably resulted from A integration into secondary attachment sites on the plasmid. Seven different secondary attachment sites on pBR322 were found. The other class resulted from plasmid integration at other sites that were widely scattered on the A genome. For this second class of recombinants, more than one site on the plasmid could recombine with A DNA. Thus, the recombination did not appear to be site specific with respect to A or the plasmid. Possible mechanisms for generating these recombinants are discussed.

There are two general classes of genetic re- the plasmid can undergo this type of recombi- combination. One class, which involves recom- nation. Thus, the mechanisms responsible for bination between homologous DNA molecules, generating these recombinants may not be site is called general recombination (24). The other specific. One possible mechanism for generating class, which involves recombination between these recombinants is reciprocal recombination DNA molecules or sites on DNA molecules that between short regions of homology. are not homologous, is called illegitimate recom- bination (31, 34). This second class is responsible MATERIALS AND METHODS for DNA rearrangements such as deletions, du- Bacteriophages, plasmids, and bacterial plications, insertions, transpositions, and inver- strains. Two derivatives of phage A were used; sions. Ab5l9b515cI857S7 (called Abb) and AcI857S7 (caUed This report concerns the illegitimate recom- A). bination that occurs between the bacteriophage Plasmid pBR322 encodes resistance to ampicillin A genome and the plasmid pBR322 in Esche- (Ampr) and tetracycline (Tetr) and is derived from richia coli. These DNA molecules do not share pMB1, a colicin -like plasmid (3). Plasmid pKPG10 extensive regions of homology. However, it is was isolated by cloning a HindIII restriction endonu- possible to isolate A clease fragment from AKan2 (1) that encoded kana- recombinant phages after mycin resistance (Kanr), using pBR322 as the vector. infection of strains containing pBR322 that All bacterial strains were derivatives of E. coli K- transduce the drug resistance genes encoded by 12. MO is an F- rpsL strain that is isogenic with this plasmid, tetracycline resistance and ampi- HfrHayes (20). Strains N747 and N761 are cillin resistance. The characterization of these MO(pKPG10) and MO(pBR322), respectively. E. coli recombinants by heteroduplex analysis and re- C600 was used as the host for A lysogens. striction endonucleases is described here. One Isolation of recombinant phages. Recombinant class of recombinant phages appears to have phages were isolated essentially as described by Berg resulted from A integration into secondary at- et al. (1). Fresh overnight cultures of N747 and N761 tachment (att) sites on the plasmid by recom- grown in LB broth (LB) plus 0.2% maltose plus 0.01 M bination MgSO4 were infected with Abb at a multiplicity of 10, with the phage att site. The other class incubated at 30°C for 30 min, diluted 1:10 with LB, has resulted from plasmid integration at other shaken at 30°C for 16 h, diluted 1:10, shaken at 300C regions of the A genome. Three or more sites on for 1 h, and induced thermally. Phages that transduced t Present address: Department of Biology, Massachusetts antibiotic resistance genes were isolated by infecting Institute of Technology, Cambridge, MA 02139. C600 and selecting resistant derivatives on LB plus 992 VOL. 142, 1980 A-pBR322 RECOMBINANTS 993 antibiotic agar plates. The drug concentrations were: lysogens on LB plus antibiotic plates at 30°C: LB plus kanamycin, 20 pg/mnl; tetracycline, 25 ,ug/ml; and am- kanamycin for AA6, LB plus tetracycline for AA4, and picillin, 100 ,ug/ml. All of the resistant derivatives of LB plus ampicillin for AT2, AK6, AK21, and AK47. C600 were found to be lysogens. LB medium and the Plasmids were purified from these revertants by the techniques for physically purifying A phages from the cleared-lysate technique described by Post et al. (23). C600 lysogens are described by Miller (21). The recom- Plasmids were recovered from C600 lysogens of binant phages we have characterized are listed in AKA2, AKA3, AKA8, AKAll, and ATA6 by isolating Table 1. The K (kanamycin), A (ampicillin), or T plasmid DNA from 10-ml cultures as described by (tetracycline) in the name refers to the antibiotic Post et al. (23), transforming C600 with the plasmid resistance genes transduced by each phage, and the preparation (19), and selecting transformants on LB number refers to an isolate number. The CsCl gradient plus kanamycin or LB plus tetracycline plates. The of the phage prepared from each of the lysogens con- of formation of these revertant plasmids tained two bands, the recombinant phage and another was estimated by comparing the number of transform- phage that appeared to be identical to the parental ants obtained with the plasmid preparations from the Abb phage. However, only AKA1, AKA2, AKA3, AKA8, lysogens and from C600(pKPG10). AKAll, and AK36 were found to be defective phages. Other techniques. DNA was extracted from the Nevertheless, we used the original lysogens as our physically purified phage as described by Miller (21), source of the recombinant phages because we found digested with restriction endonucleases as described that the yield of phage for some of the recombinants by Greene et al. (15), and electrophoresed on agarose was greater from these lysogens than from single ly- gels as described by Shinnick et al. (30). Sizes of sogens. The original lysogens were probably double restriction fragments were determined by using the lysogens of the recombinant phage and the parental EcoRI, HindIII, and BamHI restriction endonuclease phage that occurred because we used a high multiplic- fragments of A DNA as standards (13; also see Fig. 1). ity of infection (10 to 20) in isolating the drug-resistant The sizes determined in this way are expressed in lysogens. However, we cannot exclude the possibility percent A units (100% A units = length of wild-type A that the Abb-like phage in these lysogens arose by DNA). In some cases these lengths have been con- excision of the plasmid from the recombinant phage verted to base pairs (bp), using the factor 1% A unit after induction. We also do not know whether the = 490 bp (2). The sizes of restriction fragments are recombinant phage in these lysogens are integrated expressed in either percent A units or kilobase (kb) into the chromosome at the attA site, or whether they units (1 kb = 1,000 bp). are replicating as plasmid by using the pBR322 origin Heteroduplex molecules were prepared and ana- of replication. lyzed by electron microscopy as described by Davis et Recovery of plasmids from recombinant al. (7). The double-strand-length standard was nicked phages. Revertants that had recovered the inacti- circular pBR322 (4.36 kb), and the single-strand- vated gene were obtained by plating the original C600 length standard was the Ab519 deletion loop (3.0 kb)

TABLE 1. Summary ofrecombinant phagesa Site of insertion Orienta- Resistance en- Resistance tion of Class Mutant phage coded inactivated (%) P (kb) plasmidt I AA6 Abb::pKPG10 Amp Kan 57.3 ± 0.5 6.8 ± 0.3 II AKA5 -Abb::pKPG10 Kan, Amp 57.2 ± 0.3 0.8 ± 0.1 I AKA4 -Abb::pKPG10 Kan, Amp 57.4 ± 0.2 0.7 ± 0.1 I AA4 Abb::pBR322 Amp Tet 57.7 ± 0.2 (1.5) I AT2 =Abb::pBR322 Tet Amp 57.5 ± 0.2 (3.4) II AK6 -Abb::pKPG10 Kan Amp 57.3 ± 0.2 3.6 ± 0.3 II AK21 Abb::pKPG10 Kan Amp 57.2 ± 0.2 4.0 ± 0.2 I AK47 Abb::pKPG10 Kan Amp 57.3 ± 0.4 3.7 ± 0.1 I II AKA1 Abb::pKPG10 Kan, Amp 36.8 ± 0.3 5.7 ± 0.1 II AKA2 =Abb::pKPG10 Kan, Amp 99.1 ± 0.1 1.4 ± 0.2 I AKA3 = Abb::pKPG10 Kan, Amp 23.1 ± 0.7 0.9 ± 0.2 I XKA8 = Abb::pKPG10 Kan, Amp 7.0 ± 0.3 0.3 ± 0.05 II AKAll -Abb::pKPG10 Kan, Amp 14.0 ± 0.2 1.6 ± 0.2 I ATA6 Abb::pBR322 Tet, Amp 86.2 ± 0.7 (2.5) I AK36 Abb::pKPG10-A36 Kan Amp 8.1 ± 0.2 (1.0) I a Sites of insertion on A were determined from the A or B lengths given in Table 3. Results are given as percentage of the XPAPA genome. Sites of insertion on the plasmids were determined from the S lengths in Table 3 except for the values in parentheses, which were determined from sizes of restriction fragments in Table 2. Results are given in kilobases on the pBR322 or pKPG10 map in Fig. 1. Orientation I is defined as the one in which the plasmid sequences reading left to right with respect to the normal A genetic map is clockwise with respect to the plasmid map in Fig. 1. Orientation II has the opposite order. Class I phages resulted from recombination near the phage att site, 57.4%. Class II phages resulted from recombination at other sites on the A genome. 994 POGUE-GEILE ET AL. J. BACTERIOL. (13). Contour lengths were determined as described by Fiandt et al. (10). Measurements in each case were A based on 8 to 15 independent heteroduplex molecules. RESULTS Nature of recombinant phages. Most of the experiments described here were done with a derivative of pBR322, called pKPG10, that contains a 3.4-kb HindIII fragment from the transposon Tn5, which encodes resistance to kanamycin. The HindIlI sites that generate this fragment are in the 1.45-kb inverted repeats that flank Tn5 (1). Thus, pKPG10 has part of the inverted repeat ofTn5 (about 500 bp) in addition to the loop region of Tn5. The EcoRI, HindIII, BamHI, and PstI restriction endonuclease maps of pKPG10 are given in Fig. 1. pKPG10 encodes resistance to ampicillin and kanamycin, but not tetracycline. The Tetr genes of pBR322 were inactivated by the insertion of the Tn5 fragment at the HindIII site, which has been found pre- viously (3). We were originally trying to determine B whether the Tn5 HindIII fragment could trans- pose with its truncated inverted repeat. After infection of a strain containing pKPG10 (N747) with Abb, we found that it was possible to isolate recombinant X phages that transduced all pos- sible combinations of Kanr and Ampr. These phages occurred at a frequency of about 10-7 compared with the plaque-forming phages in the lysate. The approximate proportions ofthe three types of phages were as follows: AKanr Ampr, 80%; AKanr Amp8, 10%; AKan8 Ampr, 10%. The HindIII Tn5 fragment was not required for these recombinations because analogous recombinant phages could be isolated after infection of a strain that contained pBR322 (N761), i.e., AAmpr Tetr, AAmpr Tet8, AAmp8 Tetr. The fre- quency and relative occurrence of the three 2 types of phages were about the same as for FIG. 1. Restriction maps of A and pKPG10. (A) A. pKPG10. About 80% of the recombinant phages Restriction fragments are for Ab515b519 (10, 13). The transduced Ampr and Tetr, and 20% transduced sizes are in percent A units. The approxirnate loca- only Ampr or Tetr. tions of the A, J, attP, cI, and S genes are indicated. This report describes the characterization of (B) pKPG10. The coordinate system begins 0.03 kb 15 of the recombinant phages listed in Table 1 counterclockwise from the right HindIII site so that by restriction endonucleases and heteroduplex the coordinates ofpKPG10 are the same as pBR322 analysis. AT2, AA4, and ATA6 were obtained from 0.03 to 4.36 kb (32). The sizes of the fragments from a lysate ofXbb grown on N761 and selecting are given in kilobase units and percent A units. The one phage of each type; XTetr rectangles on the inner circle represent the truncated Amp8 (AT2), ATete inverted repeats from Tn5. The locations oftheAmp', Ampr (AA4), or XTetr Ampr (XTA6). Similarly, a Kan', and Tetrgenes are indicated (32). The location AAmpr Kan8 phage (AA6) and a AAmp8 Kanr ofthe Kanr gene was established from the remainder phage (AK6) were obtained from lysates of Abb ofpKPG10 that ispresent in AK36 and other mutants grown on N747. AKA1, AKA2, AKA3, XKA4, ofpKPG10 (Fig. 3 and our unpublished observations). AKA5, XKA6, XKA8, and AKAll were obtained by selecting one Kanr Ampr Abb::pKPG10 phage from a single lysate for their frequency of rever- from each of eight independent lysates. XK21, sion to Ampr at 300C. Eighteen ofthese lysogens AK36, and XK47 were obtained by screening 23 reverted to Ampr at a frequency of about l0', lysogens of Kanr Amp8 Xbb::pKPG10 phages as did the lysogen of AK6. 'Three reverted at a VOL. 142, 1980 A-pBR322 RECOMBINANTS 995 frequency of 106. Phages XK21 and AK47 were Heteroduplex molecules of each of the recom- obtained from two of these latter lysogens. Two binant molecules and A DNA were examined by lysogens did not revert to Ampr at a measurable electron microscopy. Examples of the hetero- frequency (<10-8), and AK36 was obtained from duplex molecules are given in Fig. 2. The site of one of these lysogens. insertion on the A genome was determined by The restriction fragments produced by diges- measuring the distance from the insertion to the tion of each of the recombinant DNAs with nearest bb deletion (either b515 or b519) or the EcoRI, HindIII, BamHI, and PstI are summa- nearest end (columns A and B, Table 3). The rized in Table 2. These data are presented in left and right ends could be identified from the terms of the fragments from the parental DNAs relative position and sizes of the bb deletions, that were missing from the recombinants and b519 being longer and on the left side of the b515 the new fragments that were produced. Restric- deletion (13; Fig. 1). The method for determining tion maps of the parental DNAs are given in Fig. the sites of insertion on the plasmid are dis- 1. cussed below. The sites of insertion on the A and

TABLE 2. Restriction fragments of recombinant phagesa EcoRI HindIII BamHI PstI Phage Missing New Missing New Missing New Missing X Plasmid A Plasmid A Plasmid Plasmid Class I AA6 11.3 16.0 18.0 19.7 7.0 21.0 37.1 4.5 -40 9.0 9.0 6.8 1.7 AKA5 11.3 16.0 17.5 19.7 9.0 21.0 37.1 11.5 _50b 9.0 9.9 7.7 AKA4 11.3 16.0 17.2 19.7 9.0 21.0 37.1 11.5 -50b 9.0 10.2 8.0 AA4 11.3 9.0 10.5 19.7 9.0 23.0 37.1 9.0 -40 9.0 9.5 6.4 4.3 AT2 11.3 9.0 10.2 19.7 9.0 19.5 37.1 9.0 -40 9.0 9.9 7.4 3.3 AK6 11.3 16.0 18.0 19.7 9.0 21.0 37.1 11.5 -30 5.4 9.0 8.0 5.1 AK21 11.3 16.0 27.0 19.7 9.0 -30b 37.1 11.5 -40 5.4 3.7 8.1 AK47 11.3 16.0 27.0 19.7 9.0 -30b 37.1 11.5 -40 5.4 3.9 7.4 Class II AKA1 44.7 16.0 -38 (end) 42.3 7.0 -38 (end) 37.1 11.5 -38 5.4 22.0 9.6 12.0 AKA2 6.9 16.0 12.0 8.8 9.0 14.0 13.5 11.5 21.0 9.0 (AKA6) 11.0 (end) 4.3 (end) 4.0 (end) AKA3 44.7 16.0 -32 42.3 9.0 -28 (end) 37.1 11.5 25.0 9.0 -28 (end) 22.0 20.0 AKA8 44.7 16.0 -46 42.3 9.0 -47 11.3 4.5 12.0 (end) 9.0 15.0 (end) 7.9 (end) 3.7 AKAll 44.7 16.0 -40 42.3 9.0 -40 37.1 11.5 -37 9.0 21.0 (end) 21.0 (end) 12.0 ATA6 12.1 9.0 12.0 13.4 9.0 12.2 15.2 9.0 19.8 9.0 9.0 9.8 4.7 AK36 44.7 16.0 -55C 42.3 9.0 -44 11.3 11.5 11.0 (end) 9.0 7.0 14.0 (end) 5.3 5.4 a Restriction fragments that were missing from the parental A and plasmid genomes in Fig. 1 and the new fragments that resulted are given. Sizes of restriction fragments are given in percent A units. We estimate the error in the sizes of the fragments to be +0.5% A units. Fragments identified with (end) are from the left or right end of the recombinant A genome. These fragments were easily identified because their intensity on the agarose gels was significantly less than the intensity of the internal fragments due to their partial fusion to the other end fragment. The fusion fragments were identified in the same way and were found to have sizes that were the sum of the sizes of the two end pieces. bAnother small fragment is presumably produced, but has not been detected. c Insert does not contain a restriction site for this enzyme. 996 POGUE-GEILE ET AL. J. BACTERIOL.

A...... ,.0 : ; :.

ISDO22~~~~~~~~

p.R322''',*' A

AO IV,x9f*, ' * .. .. . '.

.4' * !. *.,. s i,- g ."!'.i'59S< * :

'' '' 'C 'K'PG1O'. ' ' 5' fA .

.5iKpG519l.

l*3b512

FIG. 2. Heteroduplex molecules of the recombinant phages and A (A) ATA6 x A; (B) XKA5 X A; (C) AKAll x A. The b515 and b519 deletion loops are identified. The scale for (A) also applies to (B) and (C). VOL. 142, 1980 A-pBR322 RECOMBINANTS 997 plasmid genomes are summarized in Table 1 and duplex molecules for the phages that had recom- Fig. 3. bined with pBR322 contained a simple single- All the phages except AK36 appeared to have strand insertion loop that was about the same resulted from reciprocal recombination between size as pBR322 (column K, Table 3). a site on A and a site on the plasmid. That is, the (iii) There was one missing A restriction frag- recombinant phages appeared to contain all of ment that was replaced by two to four new both genomes, even for those phages in which fragments whose sizes approximately equaled one of the antibiotic resistance genes had been the size of the missing A fragment plus the size inactivated. The evidence for this conclusion is of the plasmid (9% for pBR322 and 16% for as follows. pKPG10). For example, AA4 is a recombinant of (i) Each of the heteroduplex molecules ap- Abb and pBR322. The 11.3% EcoRI fragment of peared to contain a simple insertion in the Abb Xbb was missing from the digest of AA4 DNA genome (Fig. 2). None of them contained two and was replaced by two fragments with sizes loops emanating from the site of the insertion, 9.5 and 10.5% units. The sum ofthe sizes ofthese which would indicate that some A DNA has been new fragments (9.5 + 10.5 = 20.0%) was about deleted or inverted. the same as the size of the missing Abb fragment (ii) For all the phages that had recombined plus pBR322 (11.3 + 9.0 = 20.3%). The two new with pKPG10, the insertion contained two sin- fragments were generated by the single EcoRI gle-strand loops. One was about the same size as site in pBR322 (Fig. 1). Similarly, the 19.7% the loop of Tn5, and the other was about the HindIII fragment was replaced by 6.4 and 23.0% same size as pBR322 (column K and column L fragments (6.4 + 23 = 29.4% _ 19.7 + 9.0). plus column S, Table 3). These two single-strand (iv) We have recovered plasmids from lyso- loops were connected by a double-strand stem gens of AA4, AA6, AT2, AK6, AK21, AK47, AKA3, that apparently resulted from the self-annealing AKA8, and ATA6 that appeared to be identical of the inverted repeat on pKPG10. The hetero- to the original plasmid and may have been gen-

TABLE 3. Heteroduplex analysis of recombinant phagesa

b519 L K Qb515

A - B -

Phage Nerestibbdeletion A B K L S Class I XA6 b515 1.30 + 0.25 3.7 ± 0.5 2.1 ± 1.0 0.5 ± 0.3 AKA5 b515 1.65 ± 0.12 2.3 ± 0.3 3.0 ± 0.4 0.8 ± 0.1 AKA4 b515 1.72 ± 0.07 2.2 ± 0.3 3.3 + 0.2 0.8 ± 0.1 AA4 b515 1.88 ± 0.10 3.8 ± 1.3 AT2 b515 1.81 ± 0.09 4.3 ± 0.1 AK6 b515 1.71 ± 0.10 2.0 ± 0.2 2.9 ± 0.2 0.7 ± 0.3 AK21 b515 1.68 ± 0.11 2.8 ± 0.7 4.3 ± 0.3 0.4 ± 0.2 AK47 b515 1.69 ± 0.21 2.1 ± 0.3 3.5 ± 0.7 0.7 ± 0.1 Class II AKA1 b519 2.3 ± 0.1 4.3 ± 0.3 1.6 ± 0.3 1.0 ± 0.1 AKA2 b515 0.4 ± 0.1 2.3 ± 0.2 3.0 ± 0.3 1.4 + 0.2 AKA3 b519 8.9 ± 0.4 2.0 ± 0.3 3.2 ± 0.3 0.9 ± 0.2 AKA8 b519 3.4 ± 0.1 1.8 ± 0.4 3.5 ± 0.3 0.25 ± 0.05 AKAll b519 6.8 ± 0.3 2.1 ± 0.2 2.3 ± 0.2 1.6 ± 0.2 XTA6 b515 6.7 ± 0.3 3.9 ± 0.4 AK36 b519 3.9 ± 0.3 4.0 ± 0.2 a Results of measurements of heteroduplex molecules between recombinant phage DNA and A DNA are given. The A, B, K, L, and S lengths are defined in the diagram at the top. A is the distance to the nearest bb deletion, either b515 or b519, which is identified in column 2. B is the distance to the nearest end, either left or right. The K loop is the single-strand loop at the end of the plasmid insertion except for AA4, AT2, ATA6, and AK36, where it is the entire insertion loop. These latter four phages do not have an L or S distance because the heteroduplex molecule contains a simple insertion loop. The K loop for XA6 is larger than for the other Xbb:: pKPG10 recombinants because recombination occurred in the Kan loop for XA6, and the K loop is pBR322. Lengths are given in kilobase units ± the standard deviation. 998 POGUE-GEILE ET AL. J. BACTERIOL.

0 10 20 30 40 50 60 70 80 90 100% I r ,AS

KA6

KAI KA4 TAS A T

KS

K21

i K47 I I FIG. 3. Restriction map of recombinant phages. The map is a summary of the data in Tables I through 3. The sites for the restriction enzymes have the following symbols: EcoRI, l; HindIII, T; BamHI, 1; PstI, $. Only the PstI sites in the plasmid inserts are given. The boxes on the inserts represent the inverted repeats of pKPG10. The locations of the Kanr, Ampr, and Tetr genes are indicated by K, A, and T. The insert in XK36 is assumed to be a continuous piece of DNA from pKPGl0, although this has not been proven. erated by a reversal of the process that formed the DNA in a cleared lysate from the lysogens the recombinants. For XA4, XA6, AT2, AK6, and selecting resistant transformants. Plasmids AK21, and XK47, in which one of the plasmid could be obtained from each of the lysogens antibiotic resistance genes had been inactivated, except that AKA1 lysogen. The frequency of strains containing plasmids were obtained by formation of plasmid was about lo-5. The plas- selecting revertants of lysogens that had re- mids in one to eight transformants from each covered the missing resistance gene. The inac- lysogen were analyzed by digestion with EcoRI, tivation ofthe antibiotic resistance gene appears HindIII, and BamHI. The plasmids from AKA8, to have resulted from the recombination within XTA6, and three of eight of the plasmids from the gene (see below). Thus, the ability to isolate XKA3 appeared to be identical to the original revertants indicates that the recombinant A plasmid. The HindIII and BamHI digests of phage contains a complete copy of the parental plasmids from ATA6 and AKA8 are shown in plasmid. Each ofthe revertants contained a plas- Fig. 4. However, none of the four plasmids from mid that was analyzed by digestion with EcoRI, either XKA2 or AKA11 was the same as the HindIII, and BamHI. The HindIII and BamHI parent plasmid. The plasmids from AKA2, digests of the plasmid from XK47 are shown in XKA3, and AKAll that were not the same as Fig. 4 along with the digests of pKPG1O. One pKPG10 were larger than the parental plasmid. revertant was analyzed for each lysogen except The ability to recover the parental plasmid from for the XK21 lysogen, for which we analyzed five nine of the recombinant phages strongly sup- independent revertants. All of the plasmids ap- ports the conclusion that the recombinants have peared to be the same as the parental plasmid resulted from recombination between a site on except for plasmids from four of the XK21 re- A and a site on the plasmid. vertants. These latter plasmids contained extra Orientation. The orientations of the plas- restriction fragments and were not analyzed fur- mids in the recombinants were determined from ther. We also attempted to recover the parental the size of the restriction fragments produced plasmids from AKA1, XKA2, AKA3, XKA8, and by knowing the site of recombination on the AKAll, and XTA6. Since neither ofthe antibiotic A genome as determined from the heteroduplex resistance genes of the parental plasmid had analysis. For example, ATA6 contained an inser- been inactivated in these recombinants, we tion of pBR322 at 86.2 ± 0.7% on the A genetic looked for plasmids by transforming C600 with map (Tables 1 and 3). The 15.2% BamHI frag- VOL. 142, 1980 A-pBR322 RECOMBINANTS 999 1 2 3 4 5 6 7 8 9 10 and AK47. Each of these appeared to have re- sulted from integration into a different second- ary att site in the Ampr gene. The evidence for this conclusion is as follows. The orientation of plasmid inserts in AT2 and AK6 is opposite to that of the inserts in AK21 and AK47. The latter two phages are clearly different because they produce slightly different EcoRI (3.7 and 3.9%) and BamHI fragments (7.4 and 8.1%) (Table 2). For AT2 and XK6, we would expect that the EcoRI and HindIII fragments containing the right junction would be the same if they resulted from recombination at the same site, even though one phage contains pBR322 and the FIG. 4. Restriction digests of plasmids recovered other contains pKPG10. However, this was not fr(am recombinant phages. Plasmids in lanes 1 the case. Thus, we concluded there are at least through 5 were digested with HindIII, andplasmids four secondary att sites in the Ampr gene of in lanes 6 through 10 were digested with BamHI. The pBR322 pllasmids in the lanes were as follows: 1 and 6, plas- pBR322. mlid from AKA8; 2 and 7, plasmid from AK47; 3 and The restriction enzyme analysis also indicated 8, pKPG10; 4 and 9, plasmid from XTA6; 5 and 10, that A had inserted in the vicinity of the Ampr P13R322. gene. The structural gene for 8l-lactamase has been sequenced and is located between 4,153 m4ent of A was replaced by 4.7 and 19.8% frag- and 3,295 on the pBR322 map (32). The PstI site m(ents (Table 2). The nearest BamHI sites on A at 3.6 kb is within this gene. For AK6, AK21, and weLre at 71.3 and 86.5%. The order of the new AK47 the 3.4-kb HindIII, 0.8-kb PstI, and 4.4-kb BczmHI fragments in ATA6 must be 19.8% - 4.7% PstI fragments of pKPG10 were intact. How- be,cause the distance between 71.3 and 86.2% is ever, the 2.6-kb PstI fragment was not, which allready more than 4.7%. Then the question be- indicates that the recombination occurred be- CO]mes: on which side of the BamHI site of the tween the PstI site at 3.6 kb and the HindlIl site pE3R322 insert are the plasmid EcoRI and at 4.3 kb on the pKPG10 map, i.e., clockwise HiindIII sites? If they are on the left, we calcu- from the PstI site at 3.6 kb. By contrast, the latLed that the new EcoRI fragments should be insertion in pBR322 to generate AT2 appears to 9.1I and 12.0% and its new HindIII fragments have occurred counterclockwise from the PstI shiould be 12.4 and 10.2%. If they are on the site in the Ampr gene. rig3ht, the new EcoRI fragments should be 10.5 We have also isolated phages in which A in- anLd 10.6%, and the new Hindlll fragments tegrated into secondary att sites in the Tetr gene sh ould be 13.8 and 8.8%. In fact, we found that of pBR322. Lysogens of XA4 are Tets Ampr, and the new EcoRI fragments were 9.0 and 12.0%, the restriction fragments of AA4 DNA indicated anLd the new HindIIl fragments were 12.2 and the integration had occurred near the Tetr gene. 9.E3%. Clearly, the first model provides a better The Tetr gene is already inactivated in AKA4 fit;to the data. Thus, we concluded that the and AKA5 because of the HindIII clone from pl;asmid BamHI site was to the right of the Tn5. However, restriction enzyme analysis indi- pl,asmid EcoRI site with respect to the conven- cated the insertions occurred in the vicinity of tic)nal A vegetative map, which we have called the Tetr gene. All three phages had the same orientation I. The orientations of the other plas- orientation (see Fig. 3), and in each case the miid inserts were determined similarly. The re- insertion of A occurred clockwise from the sul[lts are summarized in Table 1, and the restric- BamHI site in the Tetr genes. However, all three ticrn map of each of the recombinants is given in appeared to have resulted from integration at a FiLg. 3. different site because the HindIII fragments that Class I recombinants. Mapping the sites of contain the left A plasmid junction were differ- insertion on the A genome revealed that there ent; 8.0, 7.7, and 6.4% (Table 2). are two classes of recombinants. One class had The sites of insertion on the plasmids were unidergone recombination near the phage att site also determined for the Abb::pKPG10 phages by (5'7.4%), and the other class had undergone re- electron microscopy. For these phages we could COombination at other regions of the A genome use the inverted repeats of pKPG10 as a refer- (S1ee Fig. 3). The class I phages were AA6, AKA4, ence. The lengths of the single strands from the A}(A5, AA4, AT2, AK6, AK21, and AK47. pBR322 part of the pKPG10 insert are given in The Ampr gene of the plasmid was inactivated Table 3 as L (for the longest length between the in four of the class I phages, AT2, AK6, AK21, inverted repeat and the site of insertion) and S 1000 POGUE-GEILE ET AL. J. BACTERIOL. (for the shortest length). We have estimated the However, from the size of the various restriction sites of insertion from S. Essentially the same fragments and the site of insertion on the A results were obtained if the site of insertion was genome, we estimated the site of insertion to be estimated from the ratio S/(L + S), which about at 2.5 kb. Thus, there may be two or more amounts to using the pBR322 single-strand re- sites clockwise from the 0.3 kb BamHI site that gion (L + S) as an internal standard. These can be used to insert the plasmid into the X results have been converted to pKPG1O map genome. units in Table 1. In making this conversion, we Phage AK36 belongs to class II because it has used the restriction enzyme data to determine an insert at 8.1 ± 0.6% on the A genome. How- on which side of the Tn5 HindIII fragment the ever, unlike the other phages described here, the insertion had occurred. In general, these results insert does not contain all of pKPG10. The are in agreement with the other observations. heteroduplex molecule of AK36 and A had a Altogether, our observations suggest there are simple insertion loop at this position that did at least seven secondary att sites in pBR322. In not exhibit the characteristic pairing of the in- addition, there appears to be a secondary att site verted repeats on pKPG10. The size of the loop in the Kan gene of Tn5 because the Kanr gene was only 4.0 + 0.2 kb. The only restriction frag- of pKPG10 has been inactivated in forming XA6. ments of pKPG10 that we found were the 2.2-kb The restriction enzyme analysis and electron BamHI and 0.8-kb PstI fragments. The Ampr microscopy also indicated that A had inserted in gene had been deleted, which is consistent with the Tn5 loop on pKPG10 (Tables 2 and 3). the observation that Amps lysogens ofthe phage Class H recombinants. The class II recom- did not revert to Ampr. The simplest explanation binants that resulted from recombination at sites is that AK36 resulted from the insertion of a on the A genome other than at the att site were deletion mutant of pKPG10, or an internal dele- AKA1, AKA2, AKA3, AKA6, AKA8, AKAll, tion occurred in the plasmid after recombina- ATA6, and AK36. Two of these independent tion. phages appeared to be identical, AKA2 and AKA6. The other six had resulted from integra- DISCUSSION tion of the plasmid at widely scattered sites on The A genome normally integrates into a sin- the A genome (Fig. 3). gle site on the E. coli chromosome called the Because of the possibility that these recom- attA site (14). However, when this site is deleted, binants may have been cointegrate molecules the A genome can integrate into secondary att resulting from transposition processes, we were sites (29). The sequences of the phage and nor- interested in determining how many sites on the mal bacterial att sites, before and after recom- plasmid could be used for this type of recombi- bination, have been determined (17). The recom- nation. From the restriction fragments pro- bination that occurs at these sites involves a duced, it is clear that there are at least three crossover within a 15-bp core sequence that is sites on pKPG10, two of them on the pBR322 present on both att sites. sequences of pKPG10 and the other in the Tn5 Recent sequence studies have shown that sec- fragment. The site used to form AKA1 was be- ondary att sites contain a 15-bp sequence that is tween the HindIlI site at 4.3 kb and BamHI site partially homologous to the normal core se- at 6.0 kb in the Tn5 fragment because the 3.4-kb quence (6, 16). This core sequence for a second- HindIII, 5.6-kb BamHI, and 2.6-kb PstI frag- ary att site in the gal operon contains seven ments from pKPG10 were missing from the re- mismatches (16). We have done a combinant (Table 2). The electron microscopy search for possible secondary att sites on results indicated the insertion was about at 5.7 pBR322 in the vicinity of the Ampr and Tetr + 0.1 kb (Table 1). Another site was used to genes. Within the structural gene for /-lactam- form AKA8, which was between the HindIII site ase there are 23 potential core sequences that at 0.03 kb and the BamHI site at 0.3 kb. Electron contain no more than 7 mismatches: 1 with 4 microscopy indicated that the insertion was at mismatches, 5 with 6 mismatches, and 17 with 0.3 + 0.05 kb. The other class II phages resulted 7 mismatches. In the region between 375 and from recombination with sites that were clock- 1,500 bp, which is the region where recombina- wise from the 0.3-kb BamHI site. The electron tion has occurred for AKA4, AKA5, and AA4, microscopy results indicated that the sites were there are three potential core sequences with six at 1.4 + 0.2 kb for AKA2, 0.9 + 0.2 kb for AKA3, mismatches and six with seven mismatches. The and 1.6 + 0.2 kb for AKAll. The site for ATA6 orientation of these potential secondary att sites could not be determined by electron microscopy is also consistent with the hypothesis that they because there are no points of reference on were used to generate the class I recombinant pBR322 such as the inverted repeat on pKPG10. phages. VOL. 142, 1980 A-pBR322 RECOMBINANTS 1001 We were able to recover the parental plasmid tion (9). The precise excision of transposable from each of the class I phages in which an elements could also occur by this mechanism antibiotic resistance gene had been inactivated (12). Even though each of the rearrangements by isolating revertants of lysogens. The forma- mentioned above may be catalyzed by recombi- tion of these plasmids may have resulted from nation between short regions of homology, they recombination between the secondary att sites are still illegitimate in the sense that they do not after spontaneous induction, which is known to involve reciprocal crossing over between DNA occur in A lysogens (27). Lysogens of AK21 and molecules that have essentially the same se- AK47 reverted at a 10-fold-lower frequency than quence of bases. lysogens of AK6, which may indicate the second- Another possible mechanism for generating ary att site used to form AK6 more closely resem- the class II phages would be analogous to the bles the normal bacterial att site. formation of cointegrate molecules catalyzed by The class II phages were generated by recom- transposable sequences. The formation of coin- bination at several sites on the A genome. Three tegrate molecules results in the fusion of two or more sites on the plasmid were also used to DNA molecules, which is one of the character- generate these recombinants. Thus, most of the istic recombinations associated with transposa- evidence indicates the mechanism(s) responsible ble sequences (8, 11, 28). The cointegrate con- for the formation of these recombinants is not tains two copies of a transposable sequence that site specific. However, we cannot exclude the was present on one of the parental DNA mole- possibility that AKA2, AKA3, and AKAll re- cules in a single copy. The two transposable sulted from recombination at the same site on sequences are present as direct repeats at the pKPG10. junctions between the two fused molecules. There are three obvious possible mechanisms However, there are no known transposable se- for generating the class II phages. Since the quences on pBR322 or the A genome (3, 32). recombinants were isolated in a recA+ strain Nevertheless, the class II phages could be coin- with a red' A phage, one possibility is that they tegrate molecules whose formation was cata- were generated by homologous recombination. lyzed by short inverted repeats on either pBR322 However, pBR322 and A are not related in any or the A genome. For example, Ravetch et al. known way. The fact that these phages resulted (25) have recently identified a cointegrate mol- from recombination with widely scattered sites ecule formed between the plasmid pSC101 and on the A genome also indicates that there is no the bacteriophage Fl genome. As in the case for significant region of homology. Six out of seven pBR322, there are no known transposable se- class II phages were found at different sites on quences on pSC101. However, a 200-bp inverted A. Two independent class II phages, AKA2 and repeat on pSC101 was found at the junctions of AKA6, appeared to be identical, which might the pSClOl::F1 cointegrate molecule. Sequence indicate some significant homology at 99% on analysis revealed that the sequences at the ends the A genome where these recombinations oc- of this inverted repeat were homologous to the curred. However, four independent transposi- sequences at the ends of Tn3 (22, 33) and the tions of TnlO into A were found to be identical yS transposable sequence (26). The region of at the nucleotide level (18). Thus, recombination pBR322 involved in the formation of the class II catalyzed by transposable elements, which is phages is derived from pSC101 (32). A computer discussed below, might also generate two iden- search of the pBR322 sequence in this region tical recombinant phages. These considerations indicated that the pSC101 inverted repeat in- suggest that ifthe class II phages were generated volved in the formation of the pSC1O1::F1 coin- by reciprocal homologous recombinations, then tegrate is not present on pBR322. However, it the regions of homology were probably short seems conceivable that other inverted repeats (e.g., 10 bases or less). The possibility that some on either A or pBR322 could be used to form illegitimate recombination events result from cointegrate molecules. For example, there is a reciprocal recombination between short regions 32-bp sequence centered at 1,623 on the pBR322 of homology was also suggested from the obser- map which contains an 11-bp inverted repeat at the ends (32).

5'-T-T-C - C - G-T-G-T-T-T- C - G-T-A-A-A-G-T-C -T-G - G-A-A-A- C - G - C - G - G-A-A- - A-A-G -G - C-A- C -A-A-A-G- C-AI-T-T-T-C-A-G-A-C - C -T-T-T-G - C - G - C - C-T-T-5' vation that directly repeated sequences of five This sequence includes the sequence AC- or eight bases can act as the endpoints of dele- GAAAC (boxed above), which is similar to a 1002 POGUE-GEILE ET AL. J. BACTERIOL. sequence found near the ends of -y8, pSC101, A. Proc. Natl. Acad. Sci. U.S.A. 72:3628-3632. Tn3, and 2. Blattner, F. R., B. G. Williams, A. E. Blechl, K. Den- (26). Furthermore, this sequence niston-Thompson, H. E. Farber, L-A. Furlong, D. is located in the region where recombination J. Grunwald, D. 0. Kiefer, D. D. Moore, J. W. occurred to form XKA2 and AKA11. A computer Schumm, E. L. Sheldon, and 0. Smithies. 1977. search of the pBR322 sequence (32) between Charon phages: safer derivatives of bacteriophage residues 1 and 3,000 revealed 10 lambda for DNA cloning. Science 196:161-169. other sequences 3. Bolivar, F., R. L. Rodriguez, P. J. Greene, M. C. of fewer than 100 residues that are flanked by Betlach, H. L. Heyneker, H. W. Boyer, J. H. Crosa, inverted repeat sequences of at least 10 residues. and S. Falkow. 1977. Construction and characteriza- A third possible mechanism is that the class tion of new cloning vehicles II. A multipurpose cloning II phages resulted from random system. Gene 2:95-113. ligations; i.e., 4. Botchan, M., W. Topp, and J. Sambrook. 1976. The there are no specific sequences involved. The arrangement of simian virus 40 sequences in the DNA ability to recover plasmids from XKA3, XKA8, of transformed cells. Cell 9:269-287. and XTA6 that appear to be identical to the 5. Botchan, M., W. Topp, and J. Sambrook 1979. Studies parental plasmid indicates that this is on simian virus 40 excision from cellular chromosomes. probably Cold Spring Harbor Symp. Quant. Biol. 43:709-719. not the mechanism involved in the formation of 6. Christie, G. E., and T. Platt. 1979. A secondary attach- these recombinants because we would not expect ment site for bacteriophage A in trpC of E. coli. Cell 16: a totally random process to be precisely revers- 407414. ible. By contrast, the recovery of the parental 7. Davis, R. W., M. Simon, and N. Davidson. 1971. Elec- tron microscope heteroduplex methods for mapping plasmids can be easily understood by the first regions of base sequence homology in nucleic acids. two models. If homologous recombination can Methods Enzymol. 21:413-428. join the two replicons, then homologous recom- 8. Faelen, M., A. Toussaint, and J. de Lafonteyne. 1975. bination can separate them. Ifthe class II phages Model for the enhancement of A-gal integration into partially induced Mu-i lysogens. J. Bacteriol. 121:873- are cointegrates, then homologous recombina- 882. tion between the duplicated inverted repeats 9. Farabaugh, P. J., U. Schmeissner, M. Hofer, and J. could regenerate the original plasmid. H. Miller. 1978. Genetic studies of the lac repressor. Whatever the mechanism(s) responsible for VII. On the molecular nature of spontaneous hotspots in the lacI gene of Escherichia coli. J. Mol. Biol. 126: the class II phages, it should be noted that they 847-863. were generated with about the same frequency 10. Fiandt, M., W. Szybalski, F. R. Blattner, S. R. Jas- as the class I phages. About 80% of the recom- kunas, L. Lindahl, and M. Nomura. 1976. Organiza- binant phages in a Xbb lysate grown on tion of ribosomal protein genes in Escherichia coli. I. Physical structure of DNA from transducing A phages MO(pKPG10) were Kanr Ampr type, and six of carrying genes from the aroE-str region. J. Mol. Biol. eight of this type were class II phages. 106:817-835. Some aspects of the formation of the class II 11. Gill, R., F. Heffron, G. Dougan, and S. Falkow. 1978. recombinants resemble the integration ofsimian Analysis of sequences transposed by complementation oftwo classes oftransposition-deficient mutants ofTn3. virus 40 (SV40) DNA into chromosomal DNA J. Bacteriol. 136:742-756. (4, 5). Neither process appears site specific, yet 12. Grindley, N. D. F., and D. J. Sherratt. 1979. Sequences both are precisely reversible. Also, the excision analysis at IS1 insertion sites: models for transposition. of SV40 DNA sometimes generates DNAs that Cold Spring Harbor Symp. Quant. Biol. 43:1257-1261. are than 13. Gottesman, S., and S. Adhya. 1977. Genetic, physical, larger the parental DNA, which could and restriction map of bacteriophage A, p. 713-718. In be analogous to the plasmids we recovered from A. I. Bukhari, J. A. Shapiro, and S. L. Adhya (ed.), XKA2, AKA3, and AKAll that were larger than DNA insertion elements, plasmids, and episomes. Cold the parental plasmid (5). Either of the first two Spring Harbor Laboratory, Cold Spring Harbor, N.Y. models described above 14. Gottesman, M. E., and R. A. Weisberg. 1971. Prophage for the formation of the insertion and excision, p. 113-138. In A. D. Hershey class II recombinants could explain the integra- (ed.), The bacteriophage lambda. Cold Spring Harbor tion and excision of SV40 DNA. Laboratory, Cold Spring Harbor, N.Y. 15. Greene, P. J., H. LI Heyneker, F. Bolivar, R. L Rod- ACKNOWLEDGMENTS riquez, M. C. Betlach, A. A. Covarrubias, K. Back- We thank Doug Berg for AKan2, Barry Polisky for pBR322, man, D. J. Russel, R. Tait, and H. W. Boyer. 1978. Nancy Simonson and Paul Lipson for assistance with the A general method for the purification of restriction experiments, and Janice Hartman for preparation of the man- enzymes. Nucleic Acids Res. 5:2373-2380. uscript. 16. Landy, A., R. H. Hoess, K. Bidwell, and W. Ross. This work was supported by grants from the National 1979. Site-specific recombination in bacteriophage A: Science Foundation (grant PCM7714983) and the National structural features of recombining sites. Cold Spring Institutes ofHealth (Public Health Service grant GM2434702), Harbor Symp. Quant. Biol. 43:1089-1097. and by a career development fellowship from the National 17. Landy, A., and W. Ross. 1977. Viral integration and Institutes of Health (Public Health Service grant GM0040102) excision: structure of the lambda att sites. Science 197: to S.R.J. 1147-1160. 18. Kleckner, N. 1979. DNA sequence analysis of TnlO in- LITERATURE CITED sertions: origin and role of 9 bp flanking repetitions during TnlO translocation. Cell 16:711-720. 1. Berg, D. E., J. Davies, B. Allet, and J. D. Rochais. 19. Mandel, M., and A. Higa. 1970. Calcium-dependent bac- 1975. Transposition of R factor genes to bacteriophage teriophage DNA infection. J. Mol. Biol. 53:159-162. VOL. 142, 1980 A-pBR322 RECOMBINANTS 1003 20. Maquat, L E., and W. S. Reznikoff. 1978. In vitro 27. Roberts, J. W., and C. W. Roberts. 1975. Proteolytic analysis of the Escherichia coli RNA polymerase inter- cleavage of bacteriophage lambda repressor in induc- action with wild-type and mutant lactose promoters. J. tion. Proc. Natl. Acad. Sci. U.S.A. 72:147-151. Mol. Biol. 125:467-490. 28. Shapiro, J. A. 1979. Molecular model for the transposi- 21. Miller, J. H. 1972. Experiments in molecular genetics. tion and replication of bacteriophage Mu and other Cold Spring Harbor Laboratory, Cold Spring Harbor, transposable elements. Proc. Natl. Acad. Sci. U.S.A. 76: N.Y. 1933-1937. 22. Ohtsubo, H., H. Ohmori, and E. Ohtsubo. 1979. - 29. Shimada, K., R. A. Weisberg, and M. E. Gottesman. cleotide-sequence analysis of Tn3 (Ap): implications for 1972. Prophage lambda at unusual chromosomal loca- insertion and deletion. Cold Spring Harbor Symp. tions. I. Location of the secondary attachment sites and Quant. Biol. 43:1269-1277. the properties of the lysogens. J. Mol. Biol. 63:483-503. 23. Post, L. E., A. E. Arfsten, F. Reusser, and M. Nomura. 30. Shinnick, T. M., E. Lund, 0. Smithies, and F. R. 1978. DNA sequences of promoter regions for the str Blattner. 1975. Hybridization of labeled RNA to DNA and spc ribosomal protein operons in E. coli. Cell 15: in agarose gels. Nucleic Acids Res. 2:1911-1929. 215-229. in 24. Radding, C. M. 1978. Genetic recombination: strand 31. Starlinger, P. 1977. DNA rearrangements procaryotes. transfer and mismatch repair. Annu. Rev. Biochem. 47: Annu. Rev. Genet. 11:103-126. 847-880. 32. Sutciffe, J. G. 1979. Complete nucleotide sequence of 25. Ravetch, J. V., M. Ohsumi, P. Model, G. F. Vovis, D. the Escherichia coli plasmid pBR322. Cold Spring Har- Fischhoff, and N. D. Zinder. 1979. Organization of a bor Symp. Quant. Biol. 43:77-90. hybrid between phage Fl and plasmid pSC101. Proc. 33. Takeya, T., H. Nomiyama, J. Miyoshi, K. Shimada, Natl. Acad. Sci. U.S.A. 76:2195-2198. and Y. Takagi. 1979. DNA sequences of the integra- 26. Reed, R. R., R. A. Young, J. A. Steitz, N. D. F. tion sites and inverted repeated structure of transposon Grindley, and M. S. Guyer. 1979. Transposition of Tn3. Nucleic Acids Res. 6:1831-1842. the Escherichia coli insertion element -yS generates a 34. Weisberg, R. A., and S. Adhya. 1977. Illegitimate re- five-base-pair repeat. Proc. Natl. Acad. U.S.A. 76:4882- combination in bacteria and bacteriophage. Annu. Rev. 4886. Genet. 11:451-473.