Proc. Nati. Acad. Sci. USA Vol. 75, No. 3, pp. 1490-1494, March 1978

Chromosomal integration of phage X by means of a DNA insertion element (IS1 insertion sequence/chloramphenicol resistance transposon Tn9/integrative recombination) L. A. MACHATTIE AND J. A. SHAPIRO Department of Microbiology, University of Chicago, Chicago, Illinois 60637 Communicated by Albert Dorfman, January 10, 1978

ABSTRACT Phage Xcamll2, which contains the chlor- carries a of the gal-attB-bio region of the Escherichia amphenicol resistance transposon Tn9 and has a deletion of attP coil . MGBO is a gal+bio+ transductant of and the int , will lysogenize Escherichia coli K-12. Pro- phage integration occurs at different chromosomal sites, in- MADO. MS6 is a galE indicator for detection of XgalE + T- cluding lacYand maiB, but not at attB. All Xcamll2 prophages transducing particles and S1653 is a gal deletion strain for de- are excised from the chromosome after induction but with tection of Xgal + particles (3). Strain 200PS is a thi lacY strain various efficiencies for different locations. Heteroduplex from the Pasteur collection. Strain QL carries a complete lac analysis of XplacZ transducing phages isolated from a lacY:: deletion, strain X9003 carries the nonpolar M15 lacZ deletion, Xcamll2 prophage reveals an insertion sequence 1 (IS1) element and either will serve as indicator for XplacZ phage in a blue at theloint of viral and chromosomal DNA. Two lines of evi- dencekdicate that Xcamll2 encodes an excision activity that plaque assay (8). recognizes the ISI element: (i) prophage derepression increases Media. Our basic minimal medium and complete TYE the frequency of excision from IacYto yield lac+ revertants, and medium have been described (9). For selecting chlormpheni- (i) Xcamll2 infection increases reversion of a gaIT::ISl col-resistant (Cmr) clones, we supplemented media with 50Mug about 50-fold. Our results indicate that the ISI termini of chloramphenicol per ml. Our tetrazolium indicator medium of Tn9 can replace attP as a site for X insertion in the bacterial has been described (7). chromosome and that excision events are catalyzed by an ISI- Genetic Methods. Basic techniques are described elsewhere encoded protein under X repressor and N gene control. (3, 7). Tn9-containing phage are distinguished from cat phage We have argued that DNA insertion elements and specific re- by a turbid plaque assay (5); we have improved this by use of combinases provide a mechanism for joining unrelated chro- TYE agar, 30 ml per plate, containing 4,ug of chloramphenicol mosome segments (1, 2). The smallest insertion elements are per ml. Survival of Xcamll2 lysogens at 420 was determined the insertion sequences (IS elements) that causepolar by plating appropriate dilutions of TYE-glucose (0.4%) cultures (3, 4). To study the role of IS elements in integration and exci- grown at 32' on TYE-glucose agar at 320 and 420. Unless sion events, we have developed a model system using bacter- specified otherwise, all cultures were grown at 32°. iophage X and the transposable Tn9 element, which encodes Electron Microscopy. We used a Siemens 101 microscope resistance to chloramphenicol. Tn9 contains the structural gene to photograph heteroduplex DNA preparations which were for chloramphenicol acetyltransferase (cat+) bracketed by made and measured as described (11). For separate calibration direct repeats of the IS1 element (5). Xcamll2 is a Tn9-con- of duplex and single-stranded length measurements, we used taining phage that has lost the attP region and integrase gene as internal length standards the duplex length from the right so that it has no viral integration system (Fig. 1). In this paper, X cohesive end to the imm2l substitution (assumed to represent we report the isolation of stable Xcamll2 lysogens, show that 0.206 of the X+ DNA molecule) and the single-stranded immX the IS1 element is the joint between host and viral DNA seg- portion of the immunity nonhomology bubble (assumed to ments in one of these lysogens, and present genetic evidence represent 0.083 X+). that the Xcamll2 genome directs the synthesis of an IS1 exci- sion activity. RESULTS Transduction by Xcamll2. We observed that Xcamll2 will MATERIALS AND METHODS transduce E. coil K-12 to a stable chloramphenicol-resistant Bacteriophages. XcamlO5 and Xcamll2 (5) are deletion (Cmr) phenotype despite the absence of a viral integration mutants of the original Xcam phage, which had acquired Tn9 system. Comparison of transduction by Xcamll2 and its from PlCm (6). XMGB22 is a derivative of int+attP+ sibling XcamlO5 (Fig. 1) reveals that loss of the Ximm434ctslb5l5b519, which acquired Tn9 from a chromo- attP-int segment depresses transduction at a low multiplicity somal site and simultaneously lost integrase activity. XCI857S7 by only 14-fold in an attB + strain (MGBO) and not at all in an free of Xgal was obtained by induction of a lysogen carrying attB-deleted strain (MADO) (Table 1). Although the conditions the S165 gal deletion (7). We have previously described are different, these effects are much less severe than the effect XdgalS188 (3), Xcamll2imm2l and Xcamll2imm2l Acat-IS1 of an int on X insertion into the bacterial chro- (5). mosome. The Wnt6 mutation decreases lysogenization of an Bacteria. MADO is an F-thi rha trkA trkD strain which attB + strain by more than 10,000-fold and of an attB-deleted strain by 500-fold (12). This difference is not explained by Tn9 The costs of publication of this article were defrayed in part by the transposition from the X genome to the bacterial chromosome payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate Abbreviations: Cmr, chloramphenicol resistance phenotype; Cms, this fact. chloramphenicol sensitivity phenotype; IS, insertion sequence. 1490 Downloaded by guest on October 1, 2021 Genetics: MacHattie and Shapiro Proc. Natl. Acad. Sci. USA 75 (1978) 1491

434 Table 1. Transduction by XcamlO5 and Xcamll2 Tn9~~ imm A |@JN SAR Mean relative 0.1 0.2 0.3 0.4 0.5 ..6 0.7. .08 0.9 1.0 attP- transduction /05_ 21tP112 imm Phage Host int attB frequency 104 _ 108 XcamlO5 MGBO + + 100 XcamlO5 MADO + - 1.2 i 0.4 FIG. 1. The genome ofAcam and its derivatives. The coordinates Xcamll2 MGBO - + 7.0 ± 1.4 of the various A markers are their fractional distances from the left Xcamll2 MADO - - 1.3 ± 0.4 end of the X+ DNA molecule. The Tn9 element is inserted at position 0.498 and measures 0.053 unit (5). Endpoints of deletions adjacent All transductions were done at multiplicities of infection less than to Tn9 are: 105,0.466; 104,0.561; 108,0.525; and 112,0.608. The attP 1 plaque-forming unit per 50 host cells. The last column gives the site is at position 0.574 and the nt gene extends from 0.576 to 0.60 mean of transduction frequency (Cmr transductants formed per in- (10). fected cell) relative to XcamlO5 on MGBO (integrase-promoted transduction) in the same experiment repeated three times. In these experiments (one on growing cells, two on starved cells), the absolute because tests on repurified independent Cmr transductants frequencies of integrase-promoted transduction varied from 0.3 to showed that more than half of them are stable lysogens (5/7 5%. MGBO transductants and 9/12 MADO transductants). Thus, Xcamll2 appears to have an alternative mechanism for chro- boxes that bound the A sequences and the cat + gene indicate mosome integration. None of the Cmr transductants carries a ISI elements or a shorter sequence that recombines more or less detectable auxotrophic mutation (267 clones examined), but efficiently with the end of an ISI element in a specific excision selection on tetrazolium indicator media has permitted us to event. We postulate the presence of these elements based on isolate lac- and mabl Cmr Acamll2 lysogens of MGBO. Re- the structure of Xcamll2 and to explain the following results: version of the lac and mal mutations is accompanied by loss of (i) efficient prophage excision to yield two sharp density species Cmr and of phage production. Hence, the specific mutations of cat + and cat phage progeny (Table 2), (ii) the occurrence result from Xcamll2 insertion. (Subsequent experiments with of both Cmr and Cms cured survivors at comparable frequen- a different strain have yielded Xcamll2 insertions in glpR.) cies (Table 2), and (iii) the fact that most cured survivors revert Properties of Xcamll2 Lysogens. We have examined to lac +. Xcamll2 lysogens of strain MGBO in a number of different Structure of XplacZPhages Derived from MGB18. In order ways. Table 2 summarizes our results. All of the lysogens pro- to examine the connection between host and Xamll2 se- duce chloramphenicol-transducing phage when induced, and quences directly, we isolated blue-plaque-forming XlacZ phages all are stably lysogenic by two criteria: (i) the efficiency of among induced progeny of X9003 lac+ tranductants plated killing by prophage induction at 42° (>99%) and (ii) phage on lawns of X9003 or QL on plates containing a chromogenic production by isolated subclones (>99%). Among the first eight substrate for fl-galactosidase (8). We thus isolated four inde- lysogens isolated, only strains MGB5 and MGB7 appear to be pendent XplacZ phages (XXJS18, XXJS19, ASJX20, and XXJS21). double lysogens and segregate clones with reduced phage yields We also obtained a transducing particle for the lacY point (between 4 and 12% of all clones tested). The single-lysogen mutation in strain 200PS. This plaque-forming phage (XXJS25) segregants (such as MGB155 and MGB156) still produce bursts also carries the lacZ gene; so the lacY mutation must lie between greater than one per induced cell. The various lysogens differ the Z gene and Xcamll2 (Fig. 2). Heteroduplexes of DNA from from one another in several properties: (i) burst size, (if) density these XplacZ against Xcamll2imm2l or Xcam- distribution of induced progeny phage, (iii) proportion of in- 112imm2l Acat-ISI DNA show clearly that there is a common duced phage that have lost the cat + gene, and (iv) retention point at which host DNA joins prophage DNA. This point lies of the cat + gene in cured survivors of thermal induction. Al- within Tn9, about 0.015 A equivalent (the length of IS1) from though the MGBO strain yielded more transductants after its right end (Fig. 3). The critical measurement in these mole- Xcamll2 infection than the attB-deleted strain MADO (Table cules is the distance from the fmm2l substitution to the lac 1), this result may reflect only physiological differences rather substitution. The accuracy of our measurements of this distance than the presence or absence of attB. We observe that none of allow probabilities of 0.003 or less that the host-X joint could the Xcamll2 lysogens derived from MGBO will produce Xgal lie outside of Tn9. Fig. 3 shows that all the lacZ phages exhibit transducing phage. So there is no evidence for integration of a Xdg-like substitution of host for viral DNA, in agreement with Xcamll2 into attB. Fig. 2. We have examined the lac- strain MGB18 in some detail. Evidence for ISI Excision Activity. If Xcamll2 encodes an Deletion mapping and fl-galactosidase assays show that the ISI-specific integrative recombinase under control of the X prophage insertion is in lacY. Reversion to lac+ occurs at frequencies between 0.5 and 8 per 107 late-exponential-phase cat+.. cells, and the revertants are Cms and nonlysogenic. In cells i Po Z Y- SR Ndc y a plated at 420, loss of lethal A DNA (curing) has occurred about * 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4Ji a, 1000 times more frequently than reversion, and more than 200 PS MGB 152 :: one-third of the survivors retain the cat + gene (Table 2). Most of the cured survivors are still capable of reverting to lac+ MGB 41------15- (12/14 Cmr and 5/8 Cms clones tested produced revertants). One of the nonreverting Cmr survivors (MGB41) has a deletion FIG. 2. The Xcamll2 prophage of MGB18. The order of markers of most or all of lacZ but retains a functional i gene. XlacZ + based on the characters of the MGB152 and MGB41 deletions agrees with the results of heteroduplex analysis of lacZ transducing phages phages isolated from induced MGB18 lysates by transduction (Fig. 3). MGB152 has lost the cat+ gene and at least the K, L, M, I, of X9003 (about 2 to 6 per 106 plaque-forming phage in four and J A markers but retains A immunity, N, and lysis functions, and replicate experiments) are Cm5. All of these results are most will revert to lac+. MGB41 has lost lethal A functions, A immunity, consistent with the prophage structure depicted in Fig. 2. The and all lacZ markers tested but retains lacI and cat+. Downloaded by guest on October 1, 2021 1492 Genetics: MacHattie and Shapiro Proc. Natl. Acad. Sci. USA 75 (1978) Table 2. Some properties of Xcamll2 lysogens % induced % No. % Proportion Proportion cells giving Burst phage bands Survival survivors clones clones Lysogen infectious centers size cat- CsCl (420) Cmr nonlysogenic Cm" MGB1 73 74.5 <4 1 8.5 X 10-4 13.7 2/458* 0/458 MGB4 99.5 129.5 11.5 3 4.5 X 10-4 95.5 0/401 0/401 MGB5 100 281 0.3 1.2 X 10-3 49 0/296 0/296 MGB6 77.5 42 2.9 X 10-3 9.5 0/306 0/306 MGB7 94 253 0.5 2.3 X 10-3 47 0/197 0/197 MGB18 (lacY) 93.5 85.5 67.5 2 2.65 X 10-4 36.5 1/460t 1/460t MGB88 (maiB) 60.5 1.5 79 - 0/376 0/376 MGB91 (mal-) 74.5 2.5 62 0/276 0/276 0/276 MGB155 40 17.5 MGB156 37 17 The percentage of induced cells giving infectious centers is the ratio of infectious centers 15 min after induction to viable cells before induction. The burst size is the total progeny phage 120 min after induction per initial viable cell. The number of bands in CsCl was determined by visual inspection of gradients containing more than 1012 polyethylene glycol (PEG 6000)-concentrated phage particles from induced lysates (Sorvall TV865 vertical rotor, 16 hr at 30,000 rpm). Except for 420 survival of MGB5, MGB6, and MGB7, all experiments reported here were repeated at least once with similar results. * These two nonlysogenic clones are Cmr and lack X immunity. t This nonlysogenic Cm" clone is MGB152 (Fig. 2).

repressor which is capable of catalyzing the excision of a at 42° (40-100% survival) followed by expression at 32° results Xcamll2 prophage, then we expect to see an effect of prophage in a dramatic increase in the frequency of lac+ cells. (These derepression on reversion of the lacY mutation in MGB18. To data have been corrected for differential survival by deducting prevent killing of nonimmune cells by free phage, we used the maximum enrichment due simply to complete survival of defective lysogens that arose as spontaneous mutants of MGB18. pre-existing lac + revertants.) Curing of lethal X functions re- MGB152 (Cms, Fig. 2) and MGB176 (Cmr) lyse upon incuba- veals an even greater increase (data not shown). In contrast, the tion at 420 and so retain all early X functions. MGB 159 is an N- N- strains show neither loss of viability nor increase in lac + mutant of MGB18, and MGB165 is an N- mutant of MGB152. frequencies after 20 min of derepression. (A 30- to 60-min Fig. 4 shows that brief derepression of MGB152 and MGB176 derepression will result in some increased excision activity, and an even longer derepression results in cell death.) These results 0.1 0.2 0.3 0.4 0.61 0.7 0.8 0.9 1.0 indicate the synthesis of a precise excision activity from an early (0.083) X promoter. 0.498 (0.206) To examine this possibility further, we measured precise excision of an IS1 element from the gal operon. We chose the 112i21 6 i2 () S188 mutation in galT (3, 7) because we have this IS1 insertion '(+0.001) XJS 18 0 _075,_ 1 112 i21 0.4294Z 0Ao.120. (b) .Fa (±0.004) ;(±0.001) 0'0.1+ XJS 25 X 0.01 .~~~~~0-56- 112 i2la 0.409 0c087-0121- 4001 (±0.01 1) '(±0.004) XJS 21 L- A.. r- -- 0)IJ 300 112 O2IA o (±0.007) (±0.021) 0.096 I- XJS 20 4- ' (e) 'c 200- 112 i2Ia 0.434 0.070 as (±0.014) (±0.003)(003 + 100 FIG. 3. Schematic representation of heteroduplex molecules Q0 involving XplacZ and Xcamll2imm2l DNA. (a) Regions of DNA heterology between Xcamll2imm2l and wild-type X with the internal size standards used (figures in parentheses). The 0-1.0 scale indicates 5 10 wild-type X sequences according to position on the wild- Time at 420, min type physical map. (b-e) Measured structures of DNA heteroduplexes FIG. 4. Effect of prophage derepression on reversion of MGB18 of Xcamll2imm2l (or its Acat-IS1 deleted derivative) with four dif- derivatives. Growing cultures of defective lacY::Xcaml12 lysogen ferent XplacZ phages derived from the Xcamll2 lysogen MGB18. The mutants were incubated at 420 for the times indicated on the abscissa, various DNA segments depicted are: X (-), cat+ (...... ), IS1 (o), diluted, and grown overnight at 320. (Lower) Results of measuring E. coli (lac region) (------). The mean dimensions for each segment lac+ reversion frequencies after overnight growth. (Upper) Survival are given in fractional X+ lengths ±SE. The different heteroduplexes of the cultures measured immediately after removal from 420 incu- and the numbers of molecules analyzed are as follows: (a) \+:Xcam- bation. 0, MGB152; O, MGB176; X, MGB159; A, MGB165. The lac+ 112imm2l (95); (b) XXJS18:Xcamll2imm21 (36); (c) XXJS25:X- frequencies have been multiplied by the survival values to correct for camll2imm2lAcat-IS1 (8); (d) \XJS21:Xcamll2imm21Acat-IS1 enrichment due to the fact that lac+ clones are not temperature- (6); (e) XXJS20:Xcamll2imm21Acat-IS1 (5). sensitive. Downloaded by guest on October 1, 2021 Genetics: MacHattie and Shapiro Proc. Nati. Acad. Sci. USA 75 (1978) 1493

Table 3. Quantitative assay of IS1 excision from XdgalS188 Xdgal+ revertants/ A cam 112 Helper phage 104 Xdgal ACI867S7 0.7 + 0.4 Xcamll2 33.4 i 7.6 XMGB22 78.4 + 10.7 E. coil In five replicate experiments, a single-colony culture of a defective lysogen for XdgalS188 was superinfected with each helper phage and cat* NdC J induced to produce high frequency transducing (HFT) lysates (3). 0-C3 (o) The Xcamll2 used had been induced from MGB1 and the XMGB22 cat+ N J from MGB22; neither helper lysate contained Xdgal particles. cZ (b) Nd j cat+ on a Xdgal phage and can titer transducing progeny for both O- E---4 (c) gal + revertants and parental mutants (3). Xcamll2 infection N CZ J cota leads to a significant increase in the proportion of Xdgal + .-.-A (d) revertants when compared to the control infection (Table 3). Another Tn9 phage, XMGB22, shows an even greater effect on FIG. 5. Possible structures of Xcamll2 prophages generated by IS1 excision to yield Xdgal + particles. Because neither of these integrative recombination at each of the four IS1 termini in the viral phages has itself any gal homology, we conclude that they di- genome. O, IS1 element; $, integration sequence. rect the synthesis of an enzyme that catalyzes precise excision of the ISI insertion in galS188. mosensitivity of adjacent deletion formation involving IS1 mutations in the gal operon. Second are the lacY precise exci- DISCUSSION sion results illustrated in Fig. 4 and the galT::IS1 precise excision results in Table 3. These data provide evidence for synthesis of Our results show that Xcamll2, which lacks the X integration an excision activity (which does not require homology) after system, can form stable lysogens. Prophage can be found at derepression of an integrated or superinfecting Xcamll2 ge- different locations, but not at attB. In one case, we have been nome. Synthesis of excision activity from the MGB152 prophage able to establish that an ISI element serves as the connection (Fig. 2) indicates that the recombinase is encoded by the ISI between viral and chromosomal DNA. Although integration element itself and not by the rest of Tn9. The results in Fig. 4 apparently occurs at an IS1 element in the viral genome, the suggest that N-dependent transcription from the X PL promoter isolation of insertions into functional suggests that prior crosses the ISI transcription barriers and results in synthesis of presence of an ISI may not be required at the chromosomal an ISI-encoded enzyme. The IS1 excision activity could explain integration site. Moreover, one of our main results is that in- the high rate of rec- and red- independent cutout of cat-IS1 dependent Xcamll2 lysogens vary significantly in their excision segments during lytic Xcamll2 growth (5). The lower back- properties (Table 2). This is difficult to explain if Xcaml l2 al- ground level of excision and deletion activity involving ISI and ways integrates by homologous recombination into a resident Tn9 elements in the absence of transcription from a phage ISI element. promoter could reflect the absence of a promoter in the IS1 We can account for both quantitative and qualitative dif- element itself. If correct, this means that synthesis of the re- ferences in the excision behavior of independent lysogens on combinase would result only from occasional read-through the following model: Integration occurs via a reciprocal cross- across the IS1 transcription barriers. It may be that insertion over between the end of an ISI element and a suitable "inte- elements contain termination signals to reduce synthesis of gration sequence" on the chromosome (Fig. 5). The nature of integrative recombinases and consequent genetic instability. "integration sequences" is unknown, but we postulate that they function similarly to secondary A attachment sites as substrates Note Added in Proof. The following recent results are relevant to the for a specific nonhomologous integrative recombination event precise excision experiment described in Fig. 4 and Table 3. (i) The (12). The two ends of ISI are very similar (13, 14). So the model XMGB22 phage has a simple Tn9 insertion at position 0.592 on the X predicts four possible prophage structures, and we would expect molecule. This is within the integrase gene. The insertion is oriented them to behave differently with respect to excision of A and the same way as Tn9 in Xcamll2, so that transcription of Tn9 from cat + sequences during curing or after induction. The differ- the A PL promoter in either XMGB22 or Xcamll2 would produce the same Tn9 message. (ii) The method we used to determine survival of ences would result from various efficiencies of different "in- Xcamll2 lysogens after heat pulses does not adequately reflect the tegration sequences" (and of ISI termini at different locations) effects of thermal derepression on these cultures. We have found that as substrates for the excision enzyme. We know that secondary survival values of heat-pulsed lysogens are 2- to 10-fold lower several X attachment sites show such variability as excision substrates hours after 420 incubation than they are immediately after heating. (12). The MGB18 prophage would correspond to Fig. 5c, and Hence the data shown in Fig. 4 are insufficiently corrected for en- the prophage in Fig. 5d might be especially poor for excision richment of pre-existing lac + revertants. because derepression would not result in early ISI transcrip- tion. Our model postulates the function of a specific integrative We thank L. Yamamoto and G. Bock for technical assistance, W. While Epstein for bacterial strains, and A. Campbell and S. Adhya for useful recombinase during Xcamll2 integration and excision. discussions. This research was supported by grants from the University the present data do not exclude homologous recombination of Chicago Cancer Research Center, the Cancer Research Foundation functions (rec or Xred) during lysogenization and production of the University of Chicago, the National Science Foundation, and of progeny phage, there are reasons to believe that such a spe- the Louis Block Fund of the University of Chicago. J.S. is the recipient cific IS1 recombinase exists. First are the results of Reif and of a U.S. Public Health Service Research Career Development Saedler (15) showing rec-independence of reversion and ther- Award. Downloaded by guest on October 1, 2021 1494 Genetics: MacHattie and Shapiro Proc. Natl. Acad. Sci. USA 75 (1978)

1. Shapiro, J. A., Adhya, S. L. & Bukhari, A. I. (1977) in DNA In- 9. Nieder, M. & Shapiro, J. (1975) J. Bacteriol. 122,93-98. sertion Elements, Plasmids and Episomes, eds. Bukhari, A. I. 10. Landy, A. & Ross, W. (1977) Science 197, 1147-1160. Shapiro, J. A. & Adhya, S. L. (Cold Spring Harbor Laboratory, 11. Gill, G. S. & MacHattie, L. A. (1976) J. Mol. Biol. 104, 505- Cold Spring Harbor, NY), pp. 3-12. 515. 2. Shapiro, J. A. (1977) Trends Biochem. Sci. 2, 176-180. 12. Shimada, K., Weisberg, R. & Gottesman, M. (1973) J. Mol. Biol. 3. Shapiro, J. A. (1969) J. Mol. Biol. 40,93-105. 63,483-503. 4. Starlinger, P. & Saedler, H. (1972) Biochimie 54, 177-185. 13. Grindley, N. D. F. (1977) in DNA Insertion Elements, Plasmids 5. MacHattie, L. A. & Jackowski, J. B. (1977) in DNA Insertion and Episomes, eds. Bukhari, A. I., Shapiro, J. A. & Adhya, S. L. Elements, Plasmids and Episomes, eds. Bukhari, A. I., Shapiro, (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY), pp. J. A. & Adhya, S. L. (Cold Spring Harbor Laboratory, Cold Spring 595-596. Harbor, NY), pp. 219-228. 14. Ohtsubo, H. & Ohtsubo, E. (1977) in DNA Insertion Elements, 6. Scott, J, R. (1973) Virology 53,327-336. Plasmids and Episomes, eds. Bukhari, A. I., Shapiro, J. A. & 7. Shapiro, J. A. & Adhya, S. L. (1969) Genetics 62, 249-267. Adhya, S. L. (Cold Spring Harbor Laboratory, Cold Spring 8. Malamy, M. H., Fiandt, M. & Szybalski, W. (1972) Mol. Gen. Harbor, NY), pp. 591-594. Genet. 119,207-222. 15. Reif, H. J. & Saedler, H. (1975) Mol. Gen. Genet. 137, 17-28. Downloaded by guest on October 1, 2021