IS50-mediated inverse transposition: specificity and precision

Gene, 34 (1985) 17-26 Elsevier

17

GENE 1196

ISSO-mediated inverse transposition: specificity and precision (Recombinant DNA; bacteriophage 1)

hotspots;

transposon

Tn5, insertion

element;

excision;

nucleotide

sequencing;

Dilip K. Nag, Ujjala DasGupta, Gabriela Adelt, and Douglas E. Berg* Departments of Microbiology and Immunology and Genetics, Box 8093, Washington UniversitySchool of Medicine St. Louis, MO 63110 (U.S.A.) Tel. (314) 362-2772 (Received August 8th, 1984) (Revision received October 26th, 1984) (Accepted October 30th, 1984)

SUMMARY

The IS50 elements, which are present as inverted repeats in the kanamycin-resistance transposon, Tn5, can move in unison carrying with them any interstitial DNA segment. In consequence, DNA molecules such as a 1: : Tn5 phage genome are composed of two overlapping transposons - the km segment bracketed by IS50 elements (Tn5), and I bracketed by IS50 elements. During direct transposition, mediated by IS50 “0” (outside) ends, the km gene is moved and the 1 vector is left behind. During inverse transposition, mediated by the “I” (inside) ends of the IS50 elements, the 1 vector segment is moved and the km gene is left behind. Direct transposition is several orders of magnitude more frequent than inverse transposition (Isberg and Syvanen, 1981; Sasakawa and Berg, 1982). We assessed the specificity and precision of the rare events mediated by pairs of I ends by mapping and sequencing independent inverse transpositions from a ii : : Tn5 phage into the amp and ret genes of plasmid pBR322. Using restriction analyses, 32 and 40 distinct sites of insertion were found among 46 and 72 independent inverse transpositions into the amp and tet genes, respectively. Eleven sites were used in two or more insertion events, and the two sites in tet used most frequently corresponded to major hotspots for the insertion of the Tn5 (by direct transposition). The sequences of 22 sites of inverse transposition (including each of the sites used more than once) were determined, in eleven cases by analyzing both pBR322-IS50 junctions, and in eleven others by sequencing one junction. The sequence of the “I” end of IS50 was preserved and 9-bp target sequence duplications were present in every case analyzed. GC pairs were found at each end of the target sequence duplication in ten of the eleven sites used more than once, and also in seven of the other eleven sites. Our data indicate that transposition mediated by pairs of “I” ends is similar in its specificity and precision to the more frequent transposition mediated by IS50 “0” ends.

* To whom correspondence and reprint requests should be sent. Abbreviations: Amp, ampicillin; bp, base pairs; “I”, inside ends of IS or Tn; IS, insertion element; Kan, kanamycin; kb, 1000 bp LN, see MATERIALS AND METHODS, section b; pfu 0378-1119/85/$03.30 0 1985 Elsevier Science Publishers

plaque-forming units; “O”, outside ends of IS or Tn; r, (superscript), resistance; recA, red, homologous recombination genes of E. coli and phage I, respectively; ‘, (superscript), sensitivity; Tet, tetracycline; Tn, transposon; :: , novel joint.

INTRODUCTION

The movement of a prokaryotic transposable element to new sites in a genome generally requires an element-specific protein called transposase and also short distinctive sequences at its ends which probably constitute sites of transposase action. These elements can move in cells which are deficient in generalized (homologous) recombination, and insert into many different sites, indicating that extensive homology between the target and either the element or the su~o~d~g vector sequences is not essential for transposition. Insertion sites are distributed nonrandomly, and in patterns that are element-specific, suggesting that transposase also participates in target selection (for reviews, see Berg and Berg, 1983; Heffron, 1983; Iida et al., 1983; Kleckner, 1983). Transposable elements are usually flanked by direct repeats of target sequences ranging from 3 to 20 bp, depending on the element. However, the duplications generated by certain elements are variable in size (Iida et al., 1981; Klaer et al., 1980; DasSarma et al., 1983) and with IS1 the choice of duplication size (8 or 9 bp) seems to be correlated with a base substitution in ISl’s terminal sequence (Iida et al., 1981). Target duplications probably result from staggered cutting of DNA strands and repair synthesis. Variation in the length of these duplications could therefore reflect flexible spacing of cuts or limited exonucleolytic degradation of vulnerable single-str~ded target DNAs before gap repair and splicing with the element. The bacterial Kan’ transposon Tn5 (Berg et al., 1975; for review, Berg and Berg, 1983) is a composite containing terminal inverted repeats of an insertion element called IS50 (see Fig. 1). During Tn5 and also during IS50 transposition 9-bp target duplications are generated (Auerswald et al., 1981; Bossi and Ciampi, 198 1; Berg et al., 1982a). Physical mapping of Tn 5 transpositions to plasmid pBR322 had revealed a broad dis~bution of in sertion sites - a few hotspots used repeatedly, and many other sites used only rarely (Berg et al., 1983b); G + C pairs were present at both ends of the target sequence duplication at each of the five hotspots, but other factors implicated in site selection by unrelated elements (A t T richness, matches to termini, or particular consensus sequences within the 9-bp

duplicated by insertion) seemed not to guide Tn5 to its preferred sites. The two ends of IS50 are called “0” and “I” corresponding to their positions (outside and inside) in Tn5. During direct transposition from a a: :Tn5 phage (mediated by IS50 “0” ends) the km gene is moved and the ,?vector is left behind. During inverse transposition from the same A: : Tn5 phage, but mediated by the “I” ends of the IS50 elements, the ,%vector segment is moved and the kan gene is left behind (see Fig. I). Interest in the termini of IS50 was heightened by findings that transposition using pairs of “0” ends or one “0” plus one “I” end (TnJ and IS50 transposition) are 100- to lOOO-foldmore frequent than inverse transposition using a pair of “I” ends (Isberg and Syvanen, 1981; Berg et al., 1982a,b; Sasakawa and Berg, 1982). Subsequent deletion analyses showed that about 19 bp at both the “0” and the “I” ends are required for IS50 movement, although these segments are rather weakly matched in DNA sequence 5’ ~GACT~ATACAC~GT . . . at the “0” end and 5 ’ CTGTCTCTTGATCAGATCT . . . at the “I” end (Sasakawa et al., 1983; Johnson and Reznikoff, 1983). To better understand the interactions between the IS50 encoded transposase protein, each of the ends of ISSO, and sequences in target DNAs chosen for insertion, we analyzed the inverse transpositions involving 1: : Tn5 phage and plasmid pBR322 at the DNA sequence level. The results reported here show precise conservation of “I” end termini, 9-bp target sequence duplications, a nonrandom distribution of insertion sites similar to that seen with direct Tn5 transposition, and a preference for insertion sites at which one or both ends of the segment to be duplicated is a G + C pair.

MATERIALS AND METHODS

(a) Bacteria, phage and plasmids All bacterial strains are derivatives of Escherichia c&K-12. DB1891 is F-AtrpE5, hfl-1, AdrecA306 (Hirschel and Berg, 1982); DB114 is F-, dtrpE5, hfl-1, supE (Egner and Berg, 1981). MC1061 is

19

F - A(aru-leu) hsr and was obtained from M. Casadaban via H. Huang. Bacteriophage hdis 1Dam15 b519 b515 intam redAl imm2lcr” Sam7 (Hirschel and Berg, 1982). IE:: Tn5 red- was generated by transpositions of Tn5 from a chromosomal site to the ared- phage. The 4363-bp plasmid pBR322 (Sutcliffe, 1979; Peden, 1983) is from our laboratory collection. (b) Media LN broth contained 10 g of H~o-She~eld NZ-amine, 5 g of Difco yeast extract and 10 g of NaCl per liter adjusted to pH 7.2. LN-agar contained 15 g of Difco agar per liter of LN broth. Antibiotics were added at the following concentrations unless specified: Amp 250 pg/ml; Tet, otherwise 12.5 pg/ml; Kan, 50 ,ug/ml. To select for inverse transposition of the ,? moiety of a 2: : Tn5 phage into plasmid pBR322 (Fig. 1), single-colony isolates of strain DB1891 carrying pBR322 and the 2 : : Tn5 red- prophage were grown to early exponential phase at 32’ C. Phage development was induced by thermal inactivation of the prophage repressor. Phage in the lysates were adsorbed to strain DB 114 and Amp’ or Tet’ transductants were selected at 32°C (Berg et al., 1982a). Transductants were characterized by replica plating, and those harboring inverse transposition products were recognized by their Kan” phenotype. (c) Molecular t~hniqu~ Plasmid DNA was prepared by the alkaline lysis procedure of Birnboim and Doly (1979). Restriction digests, ligations and agarose and polyacrylamide gel electrophoresis were performed according to standard procedures (Maniatis et al., 1982). DNA sequence analyses of IS5O::pBR322 junctions were carried out using DNAs of the small plasmids which had been generated by deletion of 1 and most of the IS.50 sequences following BgrII digestion and li gation (Fig. lC>. These plasmids were cleaved with Bg/II and a second suitable enzyme (usually PvuII or PstI), end-labeled at the BglII site with the Klenow fragment of DNA polymerase and [ a-32P]dCTP. End-labeled DNAs were sequenced by the basespecific chemical cleavage procedure (Maxam and Gilbert, 1980).

RESULTS

(a) Generation of inverse transposition products Inverse transpositions of I into plasmid pBR322 were obtained by prophage induction in a recA 1: : Tn5 red- lysogen harboring pBR322. Transductants containing the Amp’ or Tet’ marker of pBR322, but not Kan’ of Tn5 were found after infection of strain DB 114 at a frequency of about lo-* per pfu. Agarose gel electrophoresis of extracts that the ~~sduct~ts contained showed A: : pBR322 chimaeric plasmids in which replication of the repressed I prophage moiety was driven from the pBR322 origin (Berg et al., 1982a). About 34% of the transductants selected as Tet’ and Kan” were Amp”, and 64% selected as Amp’ and Kan” were Tet”; the Amp” and Tet” transductants contain inverse transpositions into the amp and tet genes, respectively. To ensure independence, no more than one isolate of any phenotype was analyzed from each lysate. Transductants which were Kan’ as well as Amp’ and/or Tet’ were also found at a frequency of about 3 x 10m8per pfu, and generally contained 1: : Tn5 : : pBR322 fusions with three instead of two copies of IS50. Many had probably resulted from the transposition of a single copy of IS50 and homologous but recA-independent recombination between pBR322: : IS50 and IE::Tn5 DNAs (Berg et al., 1982a; Doherty et al., 1983). Others may have resulted from replicative disposition (see Biel and Berg, 1984) instead of the apparently conservative mechanism by which ISSO-based elements usually transpose (Berg, 1983). In either case, these isolates were not the result of insertion mediated by pairs of “I” ends, and hence were not studied further. (b) Stability of mutations caused by inverse transposition Tr~sduct~ts carrying each of 118 ~dependent inverse transposition products (46 Amps and 72 Tet”) reverted to an Amp’ Tet’ phenotype at frequencies ranging from about 10“” to about 10e7, depending on the mutant (Table I). These results indicate that target sequences are not deleted during inverse transposition.

._ GGTCCKCGATCG . . .AGT-TGCTCT-TGCC .. . CTGGCCCCAGTGC . . . CGCCGCCCTATAC . . .ATAAGCI’ITAATG ._ AGGATCTTACCGC .. .ATCCAGTTCGATG ._ AGAACTTTAAAAG .. . GCCTCAACCTACI

pBR322

(5’ + 3’)

observeda

First junction

Sequences

duplications

Berg, 1981; Berg et al., 1983a). Each frequency

estimate

GTTGGATGATGAC

GAAATTTTCACGA

TCAAGCTACATTG..

is the median

correction

9 bases bracketing for the one bp addition

of the underlined

...

for at least three

separate

clones.

600-608

3941-3949

4015-4023

3995-4003

31-39

1210-1218

341 l-3419

3893-3901

3731-3739

Coordinates

(bp)

33 cell doublings

( x 1O’O)

(1983). (Egner and

. and . . .TGTC-5’)

23

3.6

2.8

1.4

1000

1100

1430

1430

2500

frequency

Revertant

as described

by Peden

(5’ CTGT..

at 30°C

to the tet gene shown

the IS50 sequences

locationb

phase clones grown from single cells for approx.

et al. (1982) without

9 bp in each case.

in young stationary

listed are those of Sutcliffe (1979) as listed in Maniatis

of Tet’ or Amp’ revertants

transposition,

b The coordinates were determined

. ..

AGAATGGCGATAA

tet

amp amp amp

...

CGGGATATGGAAC.. CGAAATTACGCCA

. ..

tet

. . .

GGGGTCACGACGT..

aw aw aw ret

_. _

CGAGAACGGGCCG..

pBR322

Gene

pBR322

GAGGCTAGCAACA

(3’ + 5’)

as in Fig. 2. The complementarity

..TGTC

c The frequencies

by inverse

were determined

generated

IS50

. _TGTC . .TGTC . .TGTC . .TGTC . .TGTC . .TGTC . .TGTC . .TGTC

sequence

duplication

CTGT . . .

CTGT...

CTGT...

CTGT...

CTGT...

CT’GT...

CTGT...

CTGT...

CTGT...

IS50

Second junction

reveals the target

of sites of inverse transposition

sequence

a Seouences

T-l

A-37

A-15

A-13

T-5

T-7

A-24

A-41

A-10

Isolate

Target

TABLE I

21

Electrophoresis of plasmid DNAs from revertants of several different insertion mutations showed that they comigrated with pBR322 wild-type controls in agarose gels, indicating that they had lost Izand IS50 sequences. Hue111 digestion of DNAs of revertants of the three isolates showing the highest revertant frequencies generated sets of DNA fragments indistinguishable from that of wild-type pBR322 (specifically the appearance of fragments 80, 267, and 587 bp long which had been altered in mutants A-24, A-10, and A-41, respectively; Table I). This suggests allele-specific differences in the probability of reversion by restoration of the ancestral DNA sequence. (c) Sequence

duplications

generated

by

of nascent DNA chains between short direct repeats, equivalent to the slippage invoked in the formation of other spontaneous deletions (Egner and Berg, 1981; Berg et al., 1983a; Farabaugh et al., 1978). The length of the target duplication should therefore influence the probability of such copy errors, and hence of reversion. To increase the chance of finding possible variability in target duplication length, the set of isolates analyzed included those exhibiting the lowest and the highest reversion frequencies (1O-‘o and lo-‘). To facilitate DNA sequence analysis of sites of inverse transposition, chimaeric DNAs were digested with Bg/II and ligated in vitro to delete 1 and most of IS50 as diagrammed in Fig. 1 (panels B and C). The DNAs of these simpler plasmids were then labeled at their remaining BglII site, and DNA sequences at the nearby IS50: :pBR322 junctions were determined. A 9-bp target sequence duplication was found in each of the eleven inverse transposition products examined (Fig. 2; Table I).

inverse

transposition

The reversion of mutations caused by insertion of ISSO-related transposons does not require transposition functions, and is thought to involve a slippage

In vitro deletion

A

u

B

Amp’ pBR322-A

TetS chimaera

C

pBR322

Fig. 1. Inverse transposition and in vitro deletion. H, B, E, Ps and Pv, indicate sites of cleavage with HindIII, BgZII, EcoRI, PstI and PvuII, respectively. Only the restriction sites critical for present analysis are indicated. (Panel A) Plasmid pBR322 and phage I::Tn5, the latter showing the overlapping I and Tn5 transposons (dashed lines). (Panel B) a pBR322::1 chimaera generated by inverse transposition. The “I” ends of the IS50 elements of TnJ were joined to pBR322 and the kan segment was left behind. This chimaera replicates as a large multicopy plasmid driven by its pBR322 component and is stably maintained in transductants provided that 2 genes remain repressed (Sasakawa and Berg, 1982). (Panel C) A simpler plasmid derived in vitro from the pBR322::L chimaera by BglII cleavage and ligation. The analysis described in this report shows that such plasmids contain a 16-bp inverted repeat of IS50 “I” end sequences and 9-bp target-sequence duplications. The arrows indicate the strategy of sequencing sites of inverse transposition in one or both directions from labeled BglII ends. The needed DNA fragments were generated by digestions with Bg!II + MI (for insertions in tet) or BglII + PvuII (for insertions in amp).

22

t

c f

c 7

T Fig. 2. Determination of nucleotide sequence at ISSQ::pBR322 junctions ofinverse transposition product A-41. The sequence is read 5’-to-3’ from top-to-bottom. The coordinates of the 9bp target sequence duplication in pBR322 generated by this inverse transposition event are indicated (3893 and 3901 on the right, and 3901’ and 3893’ for the complementary strand on the left).

(d) Hotspots and infrequently used sites To map positions at which inverse transposition had occurred, chimaeric DNAs were digested with HtndIII, an enzyme which cleaves pBR322, Iz and IS50 moieties (see Fig. 1). Of the six fragments generated (Fig. 3, lanes 1 to 9) two constant fragments contain only 1 DNA, and two contain 1: : IS50 fusions. The two variable fragments result from cleavage 340 bp from the “1” end of each IS50 element and at the beginning of the pBR322 tet locus. Their sizes permit estimation of the positions at

which insertion had occurred. Results such as those in Fig. 3 indicated that many sites were used for insertion, and that some sites or regions were used repeatedly (e.g., lanes 1,2). Better estimates of the positions of insertion were obtained using plasmids from which I and most IS50 sequences had been deleted, but which retained a Bg01 site and a remnant of IS50 sequences at the site of inverse transposition (Fig. IC). These DNAs were digested with BgEI and a second enzyme which cleaved pBR322 near the estimated site of insertion, generally EcoRI. Fig. 3, lanes 14 and 16, and 12 and 13 provide examples of insertions in amp which were separated by 20 and by 26 bp, respectively (positions known exactly from DNA sequence analysis detailed below) and which were distinguishable by such digests; also shown are insertions separated by 7 bp (lanes 18 and 22), which proved to be too close to resolve in these gels. To distinguish insertions separated by as little as a single base pair, either pBR322: : ,I chimaeras (Fig. 1B) or their simpler deletion derivatives (Fig. 1C) were digested with BgZII and a second enzyme chosen because it cleaved within 200 bp of the estimated site of insertion (&zdIII for insertions near the beginning of zet; HaeIII for most other insertions). These DNAs were labeled at Bg/II sites, and electrophoresed in a denaturing polyacrylamide gel. Fig. 4 shows examples of chimaeras containing insertions at pBR322 positions 3 l-39 (a hotspot, lanes 4,5, and 8), 70-78 (lane 3) and 72-80 (lane 6). Finally, precise sites of insertion of a number of inverse tr~sposition products were determined by sequencing one or both IS50: : pBR322 junctions (Table II). Many different sites of insertion were used during inverse transposition: The 46 inverse transpositions in amp were distributed among at least 32 sites, and the 72 in tet were distributed among at least 40 sites. However, eleven sites (single nucleotide positions) at which two or more independent insertion events had taken place were found (Table II), and the two sites used most frequently for inverse transposition into bet(positions 3 l-39 and 72-80) corresponded precisely to previously identified Tn5 insertion hotspots (Berg et al., 1983b). At ten of the eleven sites used in two or more insertion events, each end of the target sequence duplication consisted of a GC bp (the sole exception, used in two events, contained GC at one end and AT at the other). Among the eleven sites

23

A

I

2 3 4 5 6 7 8 9 IO II I2 I3 I4 I5 I6 I7 I8 I9 2021 22

Fig. 3. Agarose gel analysis of restriction endonuclease digestions of inverse transposition products. Lanes 1-9, HindIII-digested TetS and Amp’ chimaeric DNAs; lanes 10-22, BglII + EcoRI-digests of the small plasmids derived from inverse transposition products by deletion of 1 and most IS50 sequences as indicated in Fig. 1C. The precise sites of insertion of some ofthe inverse transposition products are known from DNA sequence analysis (Table II), and they are indicated below by lane, and then by pBR322 coordinates. 1, 31-39; 2,72-80; 6,1065-1073; 10,3731-3739; 12,3915-3923; 13,3941-3949; 14,3995-4003; 16,4015-4023; 18,3418-3426; 20,3730-3738; 22, 3411-3419.

used only once, seven contained GC at both ends and four contained GC at one end. No site contained AT at both ends. (e) Orientation of insertion and gene expression from IS50 promoters

The relative orientation of insertions can be deduced by digestion with a restriction endonuclease such as BumHI which cleaves pBR322 but not IS50 Barn HI digestion showed that of the seven chimaeras resulting from insertion into pBR322 position 3 l-39, four were in one orientation and three were in the other; of the five resulting from insertion into the nearby hotspot, pBR322 position 72-80, four were in one orientation and one was in the other (not shown). Cells harboring pBR322 formed healthy colonies on plates containing 12 pgTet/ml, whereas plasmid-

free cells failed to make colonies on plates containing 3 pgTet/ml. The hotspot at position 31-39 is between the tet promoter and the start of translation of the tet gene, and cells containing pBR322 with Tn5 insertions in this hotspot were also sensitive to 3 pgTet/ml (reflecting the transcriptional polarity of Tn5; Berg et al., 1980). However, inverse transpositions into the hotspot at position 3 l-39 resulted in a leaky Tet’ phenotype. One orientation permitted colony formation on agar containing 3 but not 6 pgTet/ml, and the other permitted growth on 6 but not 9 ~gTet/ml. In retrospect, the instability of Tet and the relative insensitivity of replica plating to small differences in Tet potency could have caused some inverse transpositions into this hotspot to be misclassified as fully resistant. If correct, the fraction 7/72 reported in Table II would underestimate the true frequency of use of this hotspot. The promoter for transposase synthesis found in

24

TABLE II Insertion

site distribution

Isolate*

Target

Insertions

duplicationb

pBR322 position”

Occurrenced

in ret

T-5

GCTTTAATG

31-39

7172

T-2

GTCAGGCAC

72-80

5112

T-19

GCATAAGGG

636-644

3172

T-26

GCCTTCAAC

670-678

3172

T-39

GCTCTTACC

1108-1116

3172

T-18

GGTAGATGA

1065-1073

2172

T-49

CCCTATACC

1211-1219

2172

T-63

CAGTCAGGC

70-78

l/72

T-l

CAACCTACT

600-608

l/72

T-45

CCATCAGGG

1077-1075

l/l2

T-l

GCCCTATAC*

1210-1218

l/72

Insertions

*

*

in nmp

A-20

GGATAATAC

3915-3923

7146

A-6

GCGCAGAAG

3495-3503

4146

A-37

CTTTAAAAG

*

3941-3949

3146

A-24

CCCCAGTGC

*

3411-3419

2146

A-l

GTGCTGCAA

3416-3424

l/46

A-19

GCTGCAATG

3418-3426

l/46

A-22

CCTCCGATC

3730-3738

l/46

A-10

CTCCGATCG*

3731-3739

l/46

A-4 1

GCTCTTGCC

3893-3901

l/46

A-13

TCITACCGC*

3995-4003

l/46

A-15

AGTTCGATG

4015-4023

l/46

a Isolate

numbers

insertions

which, although

beginning

* *

with T and with A indicate

independent,

inverse

transpositions

were at the same site depended

into ret and amp, respectively.

on the sequence

of restriction

endonuclease

The identification digestions

of

illustrated

in Fig. 3 and 4. In cases of two or more insertions b Asterisks

indicate

found at the same site only the isolate

data listed in Table I. Only one of two ISSO-pBR322

actually

junctions

sequenced

T-39, T-18, T-49, T-45, A-20, A-6, A-l, A-19, and A-22. In these isolates the 9-bp duplication at the one junction

analyzed.

’ pBR322

in bp as listed in Maniatis

position

d Frequency

of occurrence

among

is listed.

was determined

in the following isolates: T-19, T-26,

is inferred

as the 9 bp of pBR322 sequence

et al. (1982).

72 independent

insertions

in ret or 46 independent

each IS50 element directs transcription toward the “I” end of IS50, but is weaker than the additional promoter present only in ISWL, and normally used for expression of the kun gene (Rothstein et al., 1980). Hence, the two levels of weak resistance probably reflect transcription of tet directed from these two promoters.

DISCUSSION

The rarity of inverse transposition mediated by pairs of “I” ends of IS50 elements compared to

insertions

in amp.

direct transposition mediated by “0” ends, and the relatively weak match in sequence between the “I” and “0” ends suggested that transposase interacts more efficiently with an “0” than an “I” end. To help understand these reactions we characterized plasmids generated by inverse transpositions from a I : : Tn5 phage to plasmid pBR322. (i) The complex distribution of insertion sites was reminiscent of that seen during direct (Tn5) transposition (Figs. 3,4). (ii) The complete “I” end sequence was present in each of the 33 IS50: : pBR322 junctions examined (Table II). (iii) 9 bp of target sequence were duplicated in each of eleven insertions studied (including

25

I23456769

insertion during direct (Tn5) transposition, despite differences in growth temperature and changes in physiology caused by ,? prophage induction. This coincidence, coupled with the absence of sequences in pBR322 with extensive similarity to either the “I” or the “0” end of IS50 (see Berg et al., 1983a; 1984), suggests that hotspot selection may not depend on a specific direct interaction of target and IS50 DNAs. Rather, the distribution of insertion sites may reflect a specificity inherent in the transposase protein itself. ISSO-mediated transposition may require a GC pair at one or both ends of the segment duplicated by insertion: GC pairs were found at each end of the target duplication of ten of the eleven sites used more than once and also at both ends of the duplications at seven of the eleven sites which did not seem to be insertion hotspots (Table II); a GC pair was also present at least at one end of the target duplication of every other ISSO-mediated insertion analyzed to date (Berg et al., 1984, Sasakawa et al., 1985). That these G + C pairs cannot be the only signals for hotspot selection is evident from the hundreds of places which are infrequently used despite GCs 9 bp apart. Thus, other sequences may precisely position the transposition machinery and permit cleavage of the target only at particular nucleotide positions. The many low efficiency insertion sites may reflect relatively non-specific binding of the transposition machinery to any DNA sequence followed by a GC-dependent step, for example cutting or ligation.

ACKNOWLEDGEMENTS Fig. 4. Fine structure restriction analyses of inverse transposition products demonstrating repeated insertion into particular sites. Samples digested with BglII and Hind111 and labeled at the BgBI site with Klenow DNA polymerase and [ a-‘2P]dCTP were electrophoresed on a denaturing 10% polyacrylamide gel, and autoradiographed. The positions of live insertions are known from DNA sequence analyses: lanes 4,5, and 8 contain insertions at position 31-39; lanes 3 and 6 contain insertions that are just 2 bp apart at positions 70-78 and at 72-80, respectively.

We thank Dr. S. H. Phadnis for participation in the initial selection of inverse transposition products, and Dr. C.M. Berg for critical readings of the manuscript. This work was supported by U.S. Public Health Service research grants ROl-AI14267 and ROl-AI-18980 to D.E.B.

three with the highest and three with the lowest reversion frequencies; Table I). The two sites in tet most preferred for inverse transposition coincide precisely with the hotspots for

REFERENCES Auerswald, E-A., Ludwig, G. and Schaller, H.: Structural analysis of Tn5. Cold Spring Harbor Symp. Quant. Biol 45 (1981) 107-113.

26 Berg, D.E.: Structural r~~rement for ISSO-mediated gene transposition. Proc. Natl. Acad. Sci. USA 79 (1983) 792-796. Berg, D.E. and Berg, C.M.:The prokaryotic transposable element Tn5. Biotechnology 1 (1983) 417-435. Berg, D.E., Weiss, A. and Crossland, L.: Polarity ofTn5 insertion mutations in Escherichia coli. J. Bacterial. 142(1980)439-446. Berg, D.E., Davies, J., Allet, B. and Rochaix, J.-D.: Transposition of R factor genes to bacteriophage 1. Proc. Natl. Acad. Sci. USA 72 (1975) 3628-3632. Berg, D.E., Egner, C. and Lowe, J.B.: Mechanism of F factor enhanced excision of transposon Tn5. Gene 22 (1983a) l-7. Berg, D.E., Johnsrud, L., McDivitt, L., Ramabha~~, R. and Hirschel, B.J.: The inverted repeats of Tn5 are transposable elements. Proc. Natl. Acad. Sci. USA 79 (1982a) 2632-2635. Berg, DE., Lodge,J.K., Sasakawa,C.,Nag, D.K.,Phadnis, S.H., Weston, K. and Carle, G.F.: Transposon Tn5: specific sequence recognition and conservative transposition. Cold Spring Harbor Symp. Quant. Biol. 49 (1985) (in press). Berg, D.E., Lowe, J.B., Sasakawa, C. and McDivitt, L.: The mechanism and control of Tn5 transposition, in Gustafson, J.P. and Redei, G. (Eds.) Fourteenth Stadler Genetics Symposium, University of Missouri Press, Columbia, MO. (1982b) pp. S-28. Berg, D.E., Schmandt, M. and Lowe, J.B.: Specificity of transposon TnS insertion. Genetics 105 (198333)813-828. Biel, S.W. and Berg, D.E.: Mech~ism of ISI trans~sition in E. coli: choice between simple insertion and cointregration. Genetics 108 (1984) 319-330. Birnboim, H.C. and Doly, J.: A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nucl. Acids Res. 7 (1979) 1513-1523. Bossi, L. and Ciampi, MS.: DNA sequences at the sites of two insertions of the transposable element Tn.5 in the histidine operon ofSalmonella. Mol. Gen. Genet. 183 (1981) 406-408. DasSarma, S., RajBhandary, U.L. and Khorana, H.G., High-frequency spontaneous mutation in the bacteria-opsin gene in ~alobacte~um halobi~m is mediated by transposable elements. Proc. Nati. Acad. Sci. USA 80 (1983) 2201-2205. Doherty, M.J., Morrison, P.T. and Kolodner, R.: Genetic recombination of bacterial piasmid DNA: physical and genetic analysis of the products of piasmid recombination in Escherichia coli. J. Mol. Biol. 167 (1983) 539-560. Egner, C. and Berg, D.E.: Excision of transposon Tn5 is dependent on the inverted repeats but not the transposase function of Tn5. Proc. Natl. Acad. Sci. USA 78 (1981) 459-463. Farabaugh, PI., Schmeissner, U., Hofer, M. and Miller, J.H.: Genetic studies in the lac repressor, VII. On the molecular nature of spontaneous hotspots in the lad gene offkcherichia coli. J. Mol. Bioi. 126 (1978) 847-863.

Heffron, F.:Tn3 and its relatives,in Shapiro, J.A. (Ed.) Mobile Genetic Elements. Academic Press , New York, 1983, pp. 223-260. Hirschel, B.J. and Berg, D.E.: A derivative of Tn5 with direct terminal repeats can transpose. J. Mol. Biol. 155 (1982) 105-120. Iida, S., Marcoli, R. and Bickle, T.A.: Variant insertion element IS J generates 8-base pair duplications ofthe target sequence. Nature 294 (1981) 374-376. Iida, S., Meyer, S. and Arber, W.: Prokaryotic IS elements, in Shapiro, J.A. (Ed.) Mobile Genetic Elements. Academic Press, New York, 1983, pp. 159-221. Isberg, R.R. and Syvanen, M.: Replieon fusions promoted by the inverted repeats of Tn5: The right stem is an insertion sequence. J. Mol. Biol. 150 (1981) 15-32. Johnson, R.C. and Reznikoff, W.S.: DNA sequences at the ends of transposon Tn5 required for transposition. Nature 204 (1983) 280-282. Klaer, R., Kuhn, S., Fritz, H-F., Tillmann, E., Saint-Girons, I., Habermann, P., Pfeifer, D. and Starlinger, P.: Studies on transposition mechanisms and specificity of IS4. Cold Spring Harbor Symp. Quant. Biol. 45 (1980) 215-224. Kieckner, N.: Transposon TnlO, in Shapiro, J.A. (Ed.), Mobile Genetic Elements. Academic Press, New York, 1983, pp. 261-298. Maniatis, T., Fritsch, E.F. and Sambrook, J.: Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, 1982. Maxam, A. and Gilbert, W.: Sequencing end-labeled DNA with base specific chemical cleavages. Methods Enzymol. 65 (1980) 499-560. Peden, K.W.C.: Revised sequence of the tetracycline resistance gene of pBR322. Gene 22 (1983) 277-280. Rothstein, S.J., Jorgensen, R.A., Postle, K. and Reznikoff, W.S.: The inverted repeats of Tn5 are functionally different. Ceil 19 (1980) 795-805. Sasakawa, C. and Berg, D.E.: IS50 mediated inverse transposition: discrimination between the two ends of an IS element. J. Mol. Bioi. 159 (1982) 257-271. Sasakawa, C., Carte, G. and Berg, D.E.: Sequences essential for transposition at the termini of ISSO. Proc. Nati. Acad. Sci. USA 80 (1983) 7293-7297. Sasakawa, C., Phadnis, S.H., Carle, G.F. and Berg, D.E.: Sequences essential for IS50 transposition. J. Mol. Biol. (1985) in press. Sutcliffe, J.G.: Complete nucleotide sequence of the Escherichia coli plasmid pBR322. Cold Spring Harbor Symp. Quant. Biol. 43 (1979) 77-90. Communicated by A. Campbell.