a comparison of intramolecular rearrangements promoted by transposons tn5 and tn10

A Comparison of Intramolecular Rearrangements Promoted by Transposons Tn5 and Tn10Author(s): Asad AhmedSource: Proceedings: Biological Sciences, Vol. 244, No. 1309 (Apr. 22, 1991), pp. 1-9Published by: The Royal SocietyStable URL: http://www.jstor.org/stable/76642 .

Accessed: 08/05/2014 12:34

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

The Royal Society is collaborating with JSTOR to digitize, preserve and extend access to Proceedings:Biological Sciences.

http://www.jstor.org

This content downloaded from 169.229.32.137 on Thu, 8 May 2014 12:34:56 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=rsl

http://www.jstor.org/stable/76642?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


A comparison of intramolecular rearrangements promoted by transposons Tn5 and TnlO

ASAD AHMED

Department of Genetics, University of Alberta, Edmonton, Alberta T6G 2E9, Canada

SUMMARY

The bacterial transposon TnlO has previously been shown to move to other genomic sites by a conservative mechanism, whereby the transposon is excised by double-strand breaks and inserted between a pair of staggered nicks at the target. Other transposons, like Tn3, have been shown to transpose by a replicative mechanism that involves symmetrical nicking of the element and formation of the 'Shapiro intermediate', which can mature into either a cointegrate or a simple insert. The situation with respect to Tn5 is unclear; it was originally reported to use a conservative mechanism, but other evidence suggests that the mechanism might be replicative. In this paper, rearrangements of adjacent DNA promoted by TnlO and Tn5 have been compared using positive selection for galactose-resistance to detect such rearrangements. TnlO promoted the formation of adjacent deletions (that started from an inside end of TnlO), deletion/inversions and simple ISJO insertions, but no cointegrates. This behaviour is fully consistent with a conservative mechanism. In contrast, Tn5 was found to promote formation of adjacent deletions (that started mainly from an outside end of Tn5), IS50 insertions (that were frequently accompanied by inversions of adjacent DNA) and cointegrates. These characteristics seem compatible with a replicative, rather than a conservative, mode of transposition. Clearly, Tn5 and TnlO exhibit some significant differences in their transposition. These results, and results of some previous experiments, have been interpreted to mean that Tn5 could use a replicative mechanism for its transposition.

1. INTRODUCTION

Transposable elements are discrete segments of DNA that can move to many genomic sites and can cause rearrangements of adjacent DNA. Bacterial transposable elements often occur as composite structures, called transposons, consisting of two insertion sequence (IS) elements that flank genes conferring antibiotic resistance (reviewed by Galas & Chandler (1989)). Their mobility is responsible for giving rise to new combinations of multiple drug resistance often found on bacterial plasmids. Early studies on Tn3 and Mu led to the development of a model to explain the mechanism of transposition (Arthur & Sherratt 1979; Shapiro 1979). A key feature of this model is the formation of a branched structure, commonly referred to as the 'Shapiro intermediate', by the interaction of the donor and recipient replicons at the transposon (figure 1 a). Replication of the transposon within this intermediate results in the fusion of the two replicons to form a structure called the cointegrate (figure 1 c). The cointegrate is eventually resolved by site-specific or generalized recombination to regenerate the donor and recipient replicons, each harbouring a copy of the element. The recipient replicon thus acquires an insertion. If these reactions occur at an adjacent site, an intramolecular cointegrate is formed which results in the production of an insertion/inversion or an adjacent deletion. This model is referred to as the replicative model because the original element is

believed to undergo replication during the process. However, replication of the transposon is not obliga- tory as it has been shown that, if replication of the Shapiro intermediate is aborted, both DNA strands of the original transposon can appear at the target to give non-replicative or conservative transposition (Craigie & Mizuuchi 1985). This pathway is indicated by a dashed arrow between (a) and (b) in figure 1. Hence, it appears that processing of the Shapiro intermediate determines whether transposition is replicative or conservative.

There is strong experimental evidence to support the above model (reviewed by Mizuuchi & Craigie (1986); Sherratt (1989)). However, genetic and biochemical studies on TnlO (reviewed by Benjamin & Kleckner (1989); Kleckner (1989)) have revealed that this transposon moves by an exclusively conservative mechanism, and a 'cut-and-paste ' model has been proposed (figure 1 b). According to this model, the transposon is excised from the donor by double-strand breaks and inserted between a pair of staggered nicks at the target. In this case, cointegrates are not required for transposition, but their formation, and the formation of other rearrangements, can be explained by making certain assumptions.

The situation with respect to Tn5 is somewhat controversial. This element was originally reported to transpose by the conservative, cut-and-paste mechanism (reviewed by Berg (1989)). Other evidence, however, suggested that it may use a replicative

Proc. R. Soc. Lond. B (1991) 244, 1-9 Printed in Great Britain

Vol. 244. B (22 April 1991)



2 A. Ahmed Transposon-promoted rearrangements

(a) (b)

A B A B

C D C D

t t

A D |

C D

A D

CB

(c)

A l B A B C D

A B C D

.---- ---C-x----------.,

C D

A B

ICointegrate Resolution A D A D C _ .. , B

A B ) (C B )

--- - - - - - - - - -

Insertion Insertion/inversion Adjacent deletions

Figure 1. Two models of DNA transposition. (a) The replicative model states that the process is initiated by the appearance of symmetric nicks at the transposon termini (heavy lines) and staggered nicks at the target site. Target strands are transferred to the transposon termini and joined to form a branched structure (the 'Shapiro intermediate'). This intermediate can replicate (dashed lines) to produce two copies of the transposon that connect the donor arm A to the recipient arm D, and the arm B to C. This structure, in which the two replicons are fused together, is called a 'cointegrate'. The manner in which cointegrate formation can lead to various transposon-promoted rearrangements is illustrated under (c), below. If replication of the transposon is prevented, the Shapiro intermediate can still produce a simple insert (dashed arrow) by cleavage of the donor strands and gap-repair. (For details, see Shapiro (1979); Mizuuchi & Craigie (1986).) (b) The conservative (cut-and-paste) model proposes that the transposon is excised from the donor by double-strand breaks, and inserted between a pair of staggered nicks at the target. Terminal gaps are repaired (dashes) to produce a simple insert. (For details, see Berg (1989); Kleckner (1989).) (c) Interpretation of common transposon-promoted rearrangements as intermolecular or intramolecular cointegrates. As shown under (a), above, replication of the Shapiro intermediate produces two copies of the transposon fusing A to D and C to B. If the intermediate is formed between two separate replicons AB and CD, the result is an intermolecular cointegrate ADCB carrying two copies of the transposon (shown as rectangles). The cointegrate is subsequently resolved by generalized or site-specific recombination into replicons AB and CD, each harbouring a copy of the transposon. Thus, the recipient replicon CD gains an insertion. If the target ( x ) is located on the same replicon as the donor, an intramolecular cointegrate is formed. Depending on the orientation of the target CD with respect to the donor AB, the result can be an insertion/inversion or an adjacent deletion. Insertion/

mechanism (Ahmed 1986). Therefore, intramolecular rearrangements promoted by Tn5 and TnlO were compared to determine if the two transposons utilize similar mechanisms. Results presented in this paper show that the mechanisms of Tn5 and TnlO transposition are different. Tn5 is here interpreted to transpose by a replicative mechanism, in contrast to TnlO, which has been shown previously (Kleckner 1989) to transpose by a conservative mechanism.

2. MATERIALS AND METHODS

(a) Strains

The bacterial strains used were derivatives of Escherichia coli K12 (Ahmed 1986). The recA- strain A4 was F- recA trpC sup' strA A(gal-chlD-pgl-attA)4. The recA+ strain MB3 was F- thyA trpC sup' strA A (gal-chlD-pgl-attA)3.

(b) Media

The composition of growth media (LT broth, LT agar and MacConkey-galactose) has been described (Ahmed 1987). Antibiotic concentrations (per litre) were: ampicillin (Amp), 100 mg (or 500 mg in galactose-containing media); kanamycin (Kan), 50 mg; tetracycline (Tet), 12.5 mg.

(c) Transposons and plasmids

The source of Tn5 was a kb221 rex:: Tn5 c1857 phage supplied by D. E. Berg (Berg 1977). The source of TnlO was a TnlO variant (Element 2, Way et al. 1984). In this variant, a Hindlll KanR (R denotes resistance) fragment from Tn5 has been substituted for the HindIIJ TetR fragment of TnlO. The donor phage (X1 104) for this TnlO derivative was supplied by N. Kleckner.

The structure of pAA3B 101 has been described (Ahmed 1984). This 14.7 kilobase (kb) plasmid contains the trpl gene from yeast, cos-galK-gal T region from a kgal phage, and tet- ori-amp region from pBR322. The plasmid p4.1 (figure 2a) was constructed by transposing Tn5 from the k:: Tn5 donor to pAA3B 101. Similarly, p3T- 1 was constructed by transposing TnlO from X 1104 to pAA3B 101. The general structure of this plasmid (not shown), including the TnlO insertion site, is similar to p4.1. Plasmid pBgl28 (figure 2 c) was derived from p3T- 1 by partial BglII digestion, and carries only one IS1O element (namely, IS1O-R). p6A.1 (figure 2b) was constructed by transposing TnlO from 21104 to p5.6 (a derivative of pAA3B 101).

inversions carry a second copy of the transposon in an inverted orientation, and plasmid DNA between the two copies is also inverted. Deletions produce two circular products; plasmid AD contains a deletion of BC adjacent to the transposon, whereas plasmid CB contains a deletion of AD. Normally, only one of these products (that carries the origin of replication) is recovered. It may be noted that intermolecular cointegrates can be generated also by the conservative model. For example, it can be assumed that the donor replicon existed as a dimer A.BA.B (where the dot denotes the transposon) and, following double-strand breaks, two copies of the transposon together with the intervening segment BA were inserted between C and D. The result would be a cointegrate-like structure C .BA. D. It has also been suggested that monomeric DNA could form cointegrates by the action of a pair of sibling elements just after the passage of a replication fork. Details can be found in the references given above.

Proc. R. Soc. Lond. B (1991)



Transposon-promoted rearrangements A. Ahmed 3

Hpa I Hind III

Pst I /~~~PsIEo Hind III

Sph I Ps IL Eco IV s

Oil) Pst 1~~~~~~~~~~~Pt 20. BamH I Eco V

Sph I OP 15.5/0 ~ Bam I ~ ~ ~ ~ - BglItI Eco V / Eco V

Eco ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~~~c V ~~~P IEco V- : -Eco V

c BamH I

~~?/ SphphI

Eco ~ ~ ~ ~ ~ Bm V\ BgI ,Eco

a I Hind IIII Hnd II

EoVBgl 11CZ

(a) .~ ~ ~ IIBIIIHndSph(b

(c)

Figure 2. Structure of parent plasmids. (a) The plasmid p4. 1 carries Tn5, (b) p6A. 1 carries a KanR derivative of TnlO, and (C) pBgl28 carries an aSJO-R element. The transposons are shown as rectangles. Each transposon contains two IS elements (filled rectangles) flanking a region containing the gene (kan) for kanamycin-resistance. The letters 0 and I indicate the outside and inside ends of each IS element. Each plasmid also carries the amp and tet genes of pBR322 which confer resistance to ampicillin and tetracycline, respectively. The gal TK region, whose disruption is necessary for producing galactose-resistant (GalR) mutations, is shown by a sawtooth line. The map scale is in kilobases (kb).

(d) Selection and characterization of Gal mutants

A fresh bacterial colony was inoculated into 3 ml of LT broth+ Amp and grown overnight at 37 'C. The culture was spread on a sector of a MacConkey-galactose plate containing the appropriate antibiotic. In this selection system, galactose- resistant (GalR) colonies arise by inactivation of the gal genes caused by transposon-promoted rearrangements. After 24 h, one GalR colony was picked from each sector for further analysis.

GalR mutants were selected as GalR/TetR colonies from p4.1. From this plasmid, Tn5-promoted deletions were phenotypically AmpR, Kans (s denotes sensitivity) or KanR, and Trp-, whereas Tn5 insertions were AmpR, KanR, and Trp+. Cointegrates (that represented integrations of the whole plasmid into the host chromosome) and point mutations had a phenotype similar to insertions but showed weak GalR growth. GalR mutants were selected as GalR/AmpR from p6A. 1. In this case, ISiO-promoted deletions were phenotypically Kans, Tets and Trp+ or Trp-; deletion/ inversions were Kans, TetR and Trp+; ISJO insertions and point mutations were KanR, TetR and Trp+. Thus, most rearrangements could be identified from their phenotype alone.

GalR mutants from each plasmid were characterized by restriction of plasmid minipreparations. The sites of Tn5 and TnlO insertion, and the end-points of selected transposon- promoted deletions, were determined accurately by sequencing supercoiled plasmid DNA (Ahmed 1987). The primers

used for sequencing at the inside and outside ends of IS50, respectively, were: ATCTGATGGCGCAGGG (I-primer) and CATGTTAGGAGGTCACAT (0-primer). The IS1O primers used were: CTGCAACCCTACTAGCTC (I- primer) and GTGTATCCACCTTAACTT (0-primer). These sequences were obtained from the published sequences of IS50 and ISlO (Auerswald et al. 1980; Halling et al. 1982).

3. RESULTS (a) Principle of selection

Tn5 is a 5.8 kb transposon that contains a pair of 1533 base-pair (b.p.) IS50 elements (IS50-L and IS50-R) flanking genes coding for kanamycin, bleomycin, and streptomycin resistance. TnlO is a 9.3-kb transposon that contains a pair of 1329 b.p. IS1O elements (IS1O-L and ISlO-R) flanking genes conferring tetracycline resistance. In the present study, a 7.5 kb derivative of TnlO that carries KanR instead of TetR was used. Both of these elements are capable of transposing to new genomic sites and promoting rearrangements of adjacent DNA. These activities of Tn5 and TnlO have been reviewed by Berg (1989) and Kleckner (1989), respectively.

A system for positive selection of transposon- promoted rearrangements has been reported (Ahmed 1986). Briefly, it consists of a plasmid carrying the galK

Proc. R. Soc. Lond. B (1991) 1-2




AACAGTCGGT ACGGCTGACC ATCGGGTGCC AGTGCGGGAG TTTCGTTCAG CACTGTCCTG CTCCTTGTGA TGGTTTACAA ACGTAAAAAG TCTCTTTAAT ACCTGTTTTT GCTTCATATT GTTCAGCGAC ACGTTGCTGT ACGGCAGGCA CCAGCTCT'TC CGGGATCAGC GCGACGATAC AGCCGCCAAA TCCGCCGCCG GTCATGCGTA CGCCACCTTT GTCGCCAATC ACAGCTTTGA CGATTTCTAC CAGAGTGTCA ATTTGCGGCA CGGTGATTTC GAAATCATCG CGCATAGAGG CATGAGACTC CGCCATCAAC TCGCCCATAC GTTTCAGGTC GCCTTGCTCC AGCGCGCTGG CAGCTTCAAC GGTGCGGGCG TTTTCAGTCA GTATATGACG CACGCGTTTT GCCACGATCG GGTCCAGTTC ATGCGCAACA GCGTTGAACT CTTCAATGGT GACATCACGC AGGGCTGGCT GCTGGAAGAA ACGCGCACCG GTTTCGCACT GTTCACGACG GGTGTTGTAT TCGCTGCCAA CCAGGGTACG TTTGAAGTTA CTGTTGATGA TGACGACAGC CACACCTTTG GGCATGGAAA CTGCTTTGGT CCCCAGT5AG CGGCAATCGA TCAGCAAGGC ATGATCTTTC TTGCCGAGCG CGGAAATTAG CTGATCCATG ATCCCGCAGT TACAGCCTAC AAACTGGTTT TCTGCTTCCT GACCGTTAAG CGCGATTTGT GCGCCGTCCA GCGGCAGATG ATAAAGCT'C TGCAATACGG TTCCGACCGC GACTTCCAGT GAAGCGGAAG AACTTAACCC GGCACC'CTGC GGCACATTGC CGTGATCAAC CATGTCCACG CCGCCGAAGC TGTTGTTACG CAGTTGCAGA TGTTTCACCA CGCCACGAAC GTAGTTAGCC CATTGATAGT TTTJATGTGC GACAATGGGC GCATCGAGGG AAAACTCGTC GAGCTGATTT TCATAATCGG CTGCCATCAC GCGAACTTTA CGGTCATCGC GTGGTGCACA ACTGATCACG GTTTGATAAT CAATCGCGCA GGGCAGAACG AAACCGTCGT TGTAGTCGGT GTGTTCACCA ATCAAATTCA CGCGGCCAGG CGCCTGAATG GTGTGAGTGG CAGGGTAGCC AAATGCGTTG GCAAACAGAG ATTGTGTTTT TTCTTTCAGA CTCATTTCTT ACACTCCGGA TTCGCGAAAA TGGATATCGC TGACTGCGCG CAAACGCTCT GCTGCCTGTT CTGC4GTCAG GTCTCGCTGG GTCTCTGCCA GCATTTCATA ACCAACCATA AATTTACGTA CGGTGGCGGA CGGCAGCAGA GGCGGATAAA AGTGCGCGTG CAGCTGCCAG TGTTGATTCT CTTCGCCATT AAATGGCGCG CCGTGCCAGC CCATAGAGTA GGGGAAGGAG CACTGGAAGA GGTTGTCATA ACGACTGGTC AGCTTTTTCA ACGCCAGCGC CAGATCGCTG CGCTGGGCGT CGGTCAAATC GGTGATCCGT AAAACGTGGG CTTTGGGCAG CAGTAGCGTT TCGAACGGCC AGGCAGCCCt GTAAGGCACG ACGGCTAACC AGTGTTCGGT TTCGACAACG GTACGGCTAC CGTCT*GCCAG CTCGCGCTGA ACATAATCCt CCAGCATTGG TGATTTCTGT TCGGCAAAAT ATTCTTTTTG CAGGCGGTCT TCGCGCTCAG CTTCGTTAGG CAGGAAGCTA TTTGCCCAAA TCTGACCGTG CGGATtCGGG TTAGAGCAGC CCATCGCCGC GCCTTTGTTT TCAAAAACCT GCACCCATGG GTACGTTTTC CCCjAGTTCTG CGGTTTGCTC CTGCCAGGTT TTGACGATTT CCGTCAATGC TGCAACGSTG AGCTCTGGCA GCGTTTTACT GTGATCCGGT GAAAAGCAGA TCACCCGGCT GGTGCCGCGC GCGCTCTGGC AACGCATCAG CGGATCGTGA CTTTCTGGCG CATCTGGCGT GTCAGACATC AAAGCCGCAA AGTCATTAGT GAAAACGTAA GTCCCGGTGT AATCGGGGTT TTTATCGCCT GTCACCCGCA CATTACCTGC GCAGAGGAAG CAATCTGGAT CGTGCGCAG'G TAACACCTGT TTGGCTGGCG TTTCCTGCGC CCCCTGCCAG GGCSTTAGCC GGGTGCGGTG AAACCAGAAT CCATTGCCCG GTGAGCGGTT GTAGCGGCGA TGTGGATGAT CAACGGGATT AAATTGCGTC ATGGTCGTTC CTTAATCGGG ATATCCCTGT GGATGCCGTG ACTGCCAGTG CCAGGTGTCC TGCGCCATTT CATCGAGTGT GCGCGTTACG CAGTTCAGTT CACGGTCGGC TTTGCTGGCG TCCGCCAGTA GGCCGGAAGG TCGCCCTCGC GAGCGGTGCA AAATGATAAT TAACCGGTTT GCCGCACGTT TGCTGAAGGC ATTAACCACG TCCAGCACGC TGTTGCCTAC GCCAGCGCCG AGGTTGTAGA TGTGTACGCC TGGCTTGTTC GCCAGTTTTT CCATCGCCAC GACGTGACCG TCCGCCAGAT CCATTACGTG GATGTAATCG CGTACGCCAG TACCATCTTC GGTCGGATAA TCGTTACCAA AAATCGCCAG CGAGTCGCGA CGGCCTACAG CAACCTGGGC GATGTATGGC ATCAGGTTAT TCGGAATGCC TTGCGGATCT TCGCCCATAT CGCCCGACGG ATGCGCGCCA ACCGGGTTGA AGTAGCGCAC GAGGGCAATG CTCCAGTCCG GCTGGGCTTT TTGCAGATCG GTGAGGATCT GTTCCACCAT CAGCTTGCTT TTGCCGTAAG GGCTTTGCGG TGTGCCGGTC GGGA

Figure 3. Nucleotide sequence of the gal region (Ahmed 1987) showing the end-points of TnS-promoted Kans

--deletions isolated from p4.1. In each deletion, a fixed sequence (CAAGAGACAG), located at the inside end of IS50-L, was fused to a variable sequence to the right of each arrowhead. The sequencing primer ATCTGATGGCG- CAGGGGATC, corresponding to nucleotides 1494 to 1513 of IS5O (Auerswald et al. 1980), was a gift of New England Biolabs.

and T genes of E. coli cloned on pBR322, and an E. coli host deleted for the entire gal operon. The strain is extremely galactose-sensitive, so GalR mutants can be selected by plating cells on galactose-containing media. These spontaneous mutants are caused by inactivation of the gal genes and arise at a low (1. 1 x 1O-8 GalR cell-1 per generation) rate. However, if a transposon is placed on such a plasmid, GalR mutants arise at a higher frequency and most of the GalR events are transposon-promoted. In fact, by selecting GalR mutants in the presence of an appropriate antibiotic, it is possible to isolate insertions, insertion/inversions, cointegrates, deletions, deletion/inversions etc. simul- taneously. These rearrangements can be identified by their phenotypes, and by the size and structure of their plasmids. The parent plasmids p4.1, p6A.1, and pBgl28, whose structures are shown in figure 2, were constructed for isolating rearrangements promoted by Tn5, TnlO, and IS1O, respectively.

(b) Tn5-promoted rearrangements

The recA+ strain MB3/p4. 1 produced GalR/TetR mutants at a rate of 1.5 x 10- cell-1 per generation. Among 225 GalR mutants tested, 920% were Tn5- promoted deletions, 6 0 were IS50 insertions, and 2 0 were cointegrates. The essential features of Tn5- promoted rearrangements are summarized below.

1. Tn5-promoted deletions extended from the inside (Kans Trp-) and outside (KanR Trp-) ends of S5T0 to various sites in the gal region. Of the 87 Trp- deletions tested, 72%0 were KanR and 2800 were Kans. Their end-points were determined by restriction and DNA sequence analyses. Figure 3 shows the sequences in the

gal region where 20 Kans deletions had terminated. In each case, the sequence (CAAGAGACAG), located at the inside terminus of IS50-L, was fused to a variable sequence in the gal region.

2. Insertions of IS50 (or Tn5) occurred at a number of sites in the gal region. The striking observation was that, in nine out of ten insertions examined, the trp-cos segment of plasmid DNA (located between the original transposon and the new element) was also inverted. A close examination of these insertion/inversions indicated that three types of events had occurred. In type I, a new copy of IS50 was inserted in the gal region and the trp-cos segment located between the new IS50 and the original Tn5 element was inverted. In type II, a new copy of Tn5 was inserted in the gal region and the segment between the new and the original copies of Tn5 was inverted. In type III, a complete Tn5 element was inserted in the gal region, only IS50-L was left at the 'original' location, and the intervening segment was inverted. The frequencies of their occurrence were: six of type I, one of type II, and two of type III. Figure 4 shows the photograph of an agarose gel, and restriction maps, of the three types of insertion/ inversion events.

3. All of the insertion/inversions described above were extremely unstable, producing Trp- segregants at a high (ca. 2 oo) frequency. The structure of segregants suggested that they had arisen by recombination between directly-repeated IS50 elements that were present on the original/inversion. In their final structure, these segregants resembled the Tn5-promoted deletions. Insertion/inversions were stable when transferred into a recA- host.

The inversions found to be associated with insertions could have originated by recombination between two IS50 elements present in opposite orientations, so these experiments were repeated in a recA- host. In the recA- strain A4/p4. 1, GalR/TetR mutations occurred at a rate of 0.4 x 10- cell-1 per generation. Of these, 68 0 were Tn5-promoted deletions, 12 % were IS50 insertions, 15 % were cointegrates, and 5 0 were other spontaneous events. The important features of Tn5- promoted rearrangements from recA are summarized below.

1. Tn5-promoted deletions were identical in structure to those obtained from recA+, but their frequency was lower (1/5). Of the 143 Trp- deletions tested, 95 0

were KanR and 5 0 were Kans. 2. Insertions of IS50 or Tn5 occurred at a number of

sites in the gal region. Out of the 14 insertions examined, 11 had also acquired inversions of the trp-cos segment. The frequencies of their occurrence were: ten of type I, one of type II, and none of type III. The remaining three insertions did not carry inversions.

3. All insertion/inversions were stable, but became unstable when transferred into a recA+ host. The structure of segregants was similar to those described above.

(c) Tn 10-promoted rearrangements

The recA- strain A4/6A. 1 produced GalR/AmpR mutants at a rate of 0.9 x 10-7/cell per generation. In

Pros. R. Soc. Lond. B (1991)




(a)

1 2 3 4 5 6 7 8 9 10 11 12 13

(b)

ICO $ql Cco Ps 891 amP~~~~Bm s/ H

00 BOM263 4, cc , 20 2002 891 2

22.0/0 2 8cm 89 22 c4

8(b) 4 8 18 S

89/ \14 8 Bc,,, ~~~~~~~~~ 14 8 ,~~12 10 (95 16 10 129/

86g/ Sl Sq 14 12 P~ ' 1 0Cc', q 5 c

o~~~~~~~~~~~~c

I II III Figure 4. (a) Restriction analysis of Tn5-promoted insertion/inversions and deletions isolated from p4.1. Lanes 1-6 contain BamHI + EcoRI digests of the following plasmids: lane 1, p4.1 (parent); lane 2, 653 (an insertion/inversion of type I); lane 3, 465 (an insertion/inversion of type II); lane 4, 621 (an insertion/inversion of type III); lane 5, A2 (a Kan0 deletion); lane 6, A12 (a Kans deletion). Lanes 7-12 contain BglII digests of the same plasmids in the same order. Lane 13 contains marker DNA (1-kb DNA ladder supplied by Bethesda Research Laboratories). The 0.80% agarose gel was run at 23 V for 17 h. (b) Restriction maps of Tn5-promoted insertion/inversions 653 (type I), 465 (type II), and 621 (type III). In this figure, the gal-cos segment of p4.1 has been drawn as a sawtooth line. Map scale is in kb.

one experiment, 247 GalR mutants were classified and 9000 of these were spontaneous mutants (including point mutations, non-specific deletions, and IS] insertions), 400 were IS10-promoted deletions, 3 /0 were ISJO insertions, and 3 0 were IS10-promoted deletion/ inversions. The essential features of TnlO-promoted rearrangements are given below.

1. TnlO-promoted deletions started from the inside end of ISIO-L and extended to various sites in the gal region. By DNA sequencing, the fixed end-point of these deletions was located at the inside end of IS10 (data not shown). So far, no deletion has been found to start from the outside end of TnlO.

2. IS10 insertions occurred at a number of sites in the gal region. Both IS1O elements (L and R) were

found to insert independently, and in both orientations. However, these insertions were never accompanied by inversions of adjacent plasmid DNA.

3. The insertions were stable but, on transfer into recA+ cells, acquired instability. The structure of segregants (that were Tets) could be explained by assuming that they had arisen by recombination between directly-repeated ISIO elements. Thus, depending upon the orientation of the newly-inserted IS1O element, these segregants were either Kans or KanR.

4. A novel rearrangement promoted by TnlO, called deletion/inversion (Kleckner 1989), was found to occur as frequently as the TnJO-promoted deletions. In this event, the central part of TnlO (kan-ble-str) was lost and





I S10-R, together with the adjacent tet-gal T region, was inverted.

The failure to detect insertion/inversions and deletions starting from the outside end of TnlO could have been an artifact of the selection system. This was tested by selecting GalR events from the recA+ strain MB3/6A. 1. The results are summarized below.

1. In recA+, deletions appeared to start from both inside and outside ends of TnlO.

2. Insertions of IS1O (L and R) occurred at a number of sites in the gal region and in both orientations. Approximately one-fifth (7/37) of these insertions had also acquired inversions of the tet-gal segment located between the original TnlO and the newly-inserted IS1O.

3. All insertions were moderately unstable, and produced Kans or KanR segregants. The segregants appeared to have arisen by recombination between directly-repeated IS1O elements. In their final structure, these segregants were similar to the deletions that started from the inside and outside ends of TnlO described above.

4. Deletion/inversions were recovered as in recA-. The recovery, in recA+, of insertions accompanied by

inversions and deletions starting from the outside end of TnlO suggests that their absence in recA- was probably not a consequence of the selection system used. It rather appears to be a property of the TnlO element itself.

(d) IS10-promoted rearrangements

The structure of TnlO-promoted deletions and deletion/inversions has been explained (Kleckner 1989) by the assumption that the inside ends of IS10- L and ISJO-R interact with the target to produce the two events. This hypothesis can be tested by removal of one of the two ISJO elements from TnlO. If deletion formation requires the interaction of both inside ends, removal of one of them might prevent this process. pBgl28 is a plasmid that contains only ISJO-R, which is the active element in TnlO (figure 2c).

In a recA- host, pBgl28 produced GalR mutations at a rate of 0.7 x 10` cell-' per generation. Of the 115 GalR/AmpR mutants tested, 93 0 were spontaneous mutations, 3 0 were IS1O-promoted deletions, and 4 0

were IS1O insertions. By restriction analysis, a total of nine deletions were found to start from the inside end of IS1O. This observation was confirmed by sequence analysis of one (Bgl28/A6) of these deletions. In this deletion, the sequence GATCTCTCAG, located at the inside end of IS1O-R, was fused to residue 432 (TGCGCAACAG) of the gal sequence shown in figure 3. It may be added that a derivative of p4.1, that contains only the IS50-R element of Tn5, was also found to retain full deletion-forming activity.

4. DISCUSSION

The results presented above allow a comparison (figure 5, upper and lower rows) of the different types of rearrangements promoted by Tn5 and TnlO. The following observations can be made.

1. Both transposons promoted the formation of adjacent deletions in recA- hosts. In Tn5, these deletions started from both the inside and outside ends of the transposon and extended into adjacent DNA (figure 5 a). However, the outside end seems to be used preferentially since 95 % of these deletions started from the outside end and only 500 started from the inside end. In contrast, TnlO-promoted deletions started only from an inside end of IS1O (figure 5e). The difference in the behaviour of the outside and inside ends of Tn5 and TnlO is probably not attributable to methylation as the effect, and pattern, of methylation appear to be similar on both transposons.

2. Deletion/inversion, a characteristic rearrangement (figure 5g) that involves the two inside ends of the transposon (Kleckner 1989), was isolated as frequently as the deletions from TnlO. This rearrangement was never recovered from Tn5.

3. Insertions of IS50 and IS1O occurred at a number of sites in the gal region in recA- hosts. In almost 80 % of the cases, insertions of IS50 (or Tn5) were accompanied by inversions of DNA between the original transposon and the new element (figure 5 b-d). In contrast, IS1O insertions were never accompanied by inversions (figure 5f).

4. The pattern of Tn5-promoted deletions and insertion/inversions remained essentially unchanged in a recA+ host. However, TnlO-promoted deletions appeared to start from both the inside and outside ends of the transposon, and approximately 20 0 of the IS1O insertions were found to carry inversions of adjacent DNA. In all likelihood, these changes were the result of RecA-promoted recombination between two direct, or inverted, repeats of ISJO (figure 5f). It appears that these changes were not due to the activity of the transposon itself.

5. All IS50 and ISJO insertions were unstable in recA+ hosts. Their instability could be explained by recombination between directly-repeated IS elements (figure 5 b-d, f). The segregants, thus produced, resembled the deletions.

6. Previous reports show that intermolecular cointegrates, that are believed to involve the fusion of two replicons (figure 1 c), are produced by Tn5 (Hirschel et al. 1982; Ahmed 1986; reviewed by Galas & Chandler (1989)). In contrast, this intermediate has never been observed with TnlO (Harayama et al. 1984; Weinert et al. 1984; Bender & Kleckner 1986).

As stated earlier, the current view is that the mechanisms of Tn5 and TnlO transposition are identical (Berg 1989; Kleckner 1989). The comparison given above implies significant differences, and suggests that this view is not consistent with the observations. The mechanisms of transpositions of Tn5 and TnlO could be similar; they are clearly not identical.

The rearrangements observed with TnlO are fully compatible with a conservative, cut-and-paste mechanism. The mechanism explains simple insertions directly, and explains deletion/inversions and deletions (arising from an inside end of TnlO) by assuming that the two inside ends of TnlO can attack an adjacent target site (Kleckner 1989) . In addition, the observed absence of cointegrates, insertion/inversions, and dele-





(a) (b) (c) (d)

(e) (f) (g) (ht)

Figure 5. A comparison of intramolecular rearrangements promoted by Tn5 (upper row) and TnlO (lower row). (a) Adjacent deletions arising from the inside and outside ends of Tn5 on p.4. 1. (b) Tn5-promoted insertion/inversion of type I. (c) Insertion/inversion of type II. (d) Insertion/inversion of type III. (e) Adjacent deletions arising from an inside end of TnlO on p6A. 1. (f) Simple insertions of IS1O in both orientations. (g) Deletion/inversion, a unique TnlO- promoted rearrangement. (h) Adjacent deletions arising from a single IS1O element on pBgl28. Deletions and segregants (arising from directly-repeated IS elements) are shown by heavy arcs inside. Figures are not drawn to scale.

tions (starting from an outside end of TnlO) is also consistent with this mechanism. On the other hand, the rearrangements observed with Tn5 are best explained by a replicative rather than a conservative mechanism. If the Shapiro intermediate formed is intermolecular, the result is a cointegrate or a simple insert depending on whether the intermediate is replicated or not. If the intermediate is intramolecular, the result is an adjacent deletion (starting from an inside or outside end of the transposon) or an insertion/inversion depending on the orientation of the target. Perhaps the most persuasive evidence for a replicative mechanism is the spontaneous appearance of inversions when only insertions were selected. The three types (I-III) of insertion/inversion events depicted in figures 4 and 5 b-d can be explained by selective replication of IS50-R, Tn5, and IS50-L, respectively. Furthermore, the frequencies of their occurrence (16 of type I, only two of type III) are also in line with the relative activities of IS50-R and IS50-L. While the majority (80 %o) of the IS50 insertions apparently arose by the intramolecular pathway, the intermolecular pathway is expected to remain operational. The minority (20 %o) class of IS50 insertions, that was not accompanied by inversions, was probably the result of the latter pathway.

At this stage, it is necessary to review the evidence that has been presented to support a conservative mechanism for Tn5 transposition (Berg 1989). The first line of evidence is the absence of an IS50-specific resolvase. A resolvase is an enzyme that catalyses site- specific recombination between directly-repeated IS elements to resolve a cointegrate (figure 1 c). The present work shows that directly-repeated IS50 elements

are efficiently recombined by the host RecA function. Therefore, the presence of a specific Tn5-coded resolvase is not essential for its transposition by a replicative pathway. Moreover, stable cointegrates (as in recA-) would not block transposition as the Shapiro intermediate can also produce simple inserts directly (figure 1 a, b). The second line of evidence is that structures resembling cointegrates are formed frequently from Tn5-containing dimeric donor plasmids in recA+ cells, but these are not true cointegrates because they contain only a portion of the donor DNA. Hence, these structures were interpreted as simple inserts of fragments of dimeric donors into the recipient plasmids arising by a cut-and-paste mechanism. Dimeric plasmids are indeed common in recA+ cells and can act as donors in transposition, but all of the products from dimeric donors described by Berg can also be explained as breakdown products of true cointegrates formed between a dimeric donor and a recipient plasmid. The role of RecA in Tn5-promoted cointegrate formation is still unclear. Galas & Chandler (1989) have suggested that it may be more than just a source of dimeric DNA. They have proposed that either the formation of cointegrates or the stability of an intermediate is enhanced by RecA. To account for the formation of cointegrates from monomeric donors (as in recA- cells), it has been suggested that cointegrates could arise by conservative transposition from a pair of sister elements immediately after the passage of a replication fork (Lichens-Park & Syvanen 1988). This scheme provides the two active, hemimethylated copies of the element flanking the donor sequence that would be required if cointegrate formation could occur





K S a b ori

K 1 3 1 1 3 1

t ?- 0 ?0

K 1, A 3 A a0 O 0

S

a b

K(; a b art K 1 3 3 20

Inversion (Lost) Deletion

Figure 6. Analysis of a complex rearrangement occurring in a X-pBR322: : Tn5 cointegrate (Jsberg & Syvanen 1 985) . The cointegrate consisted of Tn5 (IS50 elements 1 and 2 (open boxes) flanking the central KanR segment KS), plasmid pBR322 (a, b, and ori), and an IS50 (element 3) joined by B. Loss of X sequence was selected by plating the strain at high temperature on a drug-containing medium. The products consists of a plasmid with three IS50 elements but with an inversion of some sequence between two of the elements, or a plasmid carrying a deletion. This intramolecular rearrangement can be explained by the replicative model involving the formation of a Shapiro intermediate. The structure of the cointegrate is such that the region bound by elements 1 and 3 can constitute a giant transposon that can transpose within itself. Elements 1 and 3 are nicked symmetrically and the target is nicked between a and b. Strand transfers lead to the formation of a Shapiro intermediate. Replication starting from the 3' end of the target (shown by dashes) can either stop after filling in the gaps or continue by strand displacement along elements 1 and 3 until stopped at terminal nicks shown by dashed arrows. Depending on the orientation of the target, the result would be a plasmid carrying three IS50 elements with an inversion, or a plasmid carrying an adjacent deletion. In both cases, X sequence would be lost. To simplify comparison, the figure is drawn and labelled as in Berg ( 1989) .

by a conservative mechanism. It is not known if these schemes (dimer donor and sister elements) actually operate in the formation of Tn5-promoted cointegrates; they do not seem to work in TnlO, an element that uses a conservative mechanism but forms no detectable cointegrates. The third line of evidence is derived from a complex intramolecular rearrangement from a B-pBR322: Tn5 cointegrate originally described by Isberg & Syvanen (1985). This cointegrate (figure 6) was found to break down, upon selection for

the loss of X genes, to produce (i) a plasmid containing three IS50 elements with an inversion of some sequences between two of the elements, or (ii) a plasmid containing one IS50 element with an adjacent deletion. On the basis of conservative transposition, this rearrangement was explained (see Berg 1989) in two steps: Excision by double-strand breaks at the outside ends of IS50 elements 1 and 3, and then joining of the ends to an internal site between a and b in either orientation. Figure 6 presents an alternate interpretation of the same rearrangement on the basis of a replicative mechanism (T. Mueller & A. Ahmed, un- published results). The formation of both inversion and deletion products can be explained in a single step. Thus, the evidence presented for a conservative mechanism of Tn5 transposition does not appear to be unequivocal. Clearly, physical experiments are now needed to test these ideas.

All of the TnlO-promoted rearrangements previously described by Kleckner (1989) were recovered in the present study. These included IS1O insertions, deletions starting from an inside end of IS1O, and deletion/ inversions. Both IS1O elements (L and R) were found inserted in the gal region in both orientations. Thus, unlike Tn5, there was no fixed orientation for IS1O insertions, and these insertions were never accompanied by inversions of adjacent plasmid DNA. This genetic evidence suggests that these insertions did not arise by the replication of the adjacent TnlO element and supports the findings of Kleckner that TnlO transposition is conservative. The IS1O insertions probably originated from other plasmid copies and their insertion was not accompanied by other rearrangements.

Conservative transposition can occur through either a cut-and-paste mechanism or a Shapiro intermediate that is unable to replicate (figure 1). On the basis of the cut-and-paste mechanism, the formation of deletion/ inversions and adjacent deletions can be explained (Kleckner 1989) by assuming that the inside ends of TnlO are cut by double-strand breaks and joined to a target carrying staggered nicks. This would produce either a deletion/inversion or a deletion starting from an inside end of TnlO depending on the orientation of the target. This scheme explains not only the formation of these two rearrangements but also the absence of insertion/inversions and deletions starting from the outside ends of TnlO. A prediction of this scheme is that the formation of adjacent deletions would be sup- pressed if one of the two inside ends was missing. However, the result obtained with pBgl28 shows that a single IS1O element can suffice to produce adjacent deletions. It implies that the other inside end may not be necessary. Kleckner has suggested that such an event could occur by post-replicational transposition in which two copies of IS1O, present on sister strands, would interact with the target. An alternate possibility is that TnlO transposition might also occur by symmetric nicking and formation of a Shapiro intermediate. Since TnlO transposition is conservative (Bender & Kleckner 1986), it must be assumed that the intermediate can repair the terminal gaps but cannot replicate. It would then mature into simple inserts by





the cleavage of the donor strands (figure 1 a, b). This alternate scheme can explain the formation of adjacent deletions from a single IS1O element, the formation of simple inserts, adjacent deletions, deletion/inversions, and excised transposon circles from TnlO, and the absence of cointegrates and insertion/inversions. The inactivity of the outside ends for deletion formation can probably be attributed to a factor such as methylation. There is as yet no evidence that a Shapiro intermediate is formed during TnlO transposition (Benjamin & Kleckner 1989). Nonetheless, it does raise the inter- esting possibility that, after all, Tn5 and TnlO transposition are alike, the only difference being that the Shapiro intermediate is replicated in one but not the other.

I acknowledge the important contributions of Alastair Hunter, M. S. Sidhu, Tom Mueller, Brendan Cormack, York Marahrens and Tom Clandinin at different stages of this work. I thank Professor D. Nash for reading the manuscript and for making constructive suggestions, and Dr D. E. Berg and Dr N. Kleckner for providing transposon stocks. This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada.

REFERENCES

Ahmed, A. 1984 A deletion analysis of transposon Tn9. J. molec. Biol. 173, 523-529.

Ahmed, A. 1986 Evidence for replicative transposition of Tn5 and Tn9. J. molec. Biol. 191, 75-84.

Ahmed, A. 1987 Use of transposon-promoted deletions in DNA sequence analysis. In Methods in enzymology (ed. R. Wu), vol. 155, pp. 177-204. San Diego, California: Academic Press.

Arthur, A. & Sherratt, D. 1979 Dissection of the transposition process: A transposon-encoded site-specific recombination system. Molec. Gen. Genet. 175, 267-274.

Auerswald, E.-A., Ludwig, G. & Schaller, H. 1980 Structural analysis of Tn5. Cold Spring Harb. Symp. quant. Biol. 45, 107-113.

Bender, J. & Kleckner, N. 1986 Genetic evidence that TnlO transposes by a nonreplicative mechanism. Cell 45, 801-815.

Benjamin, H. W. & Kleckner, N. 1989 Intramolecular transposition by TnlO. Cell 59, 373-383.

Berg, D. E. 1977 Insertion and excision of the transposable Kanamycin resistance determinant Tn5. In DNA insertion elements, plasmids, and episomes (ed. A. I. Bukhari, J. A.

Shapiro & S. L. Adhya), pp. 205-212. Cold Spring Harbor, New York: Cold Spring Harbor Laboratory.

Berg, D. E. 1989 Transposon Tn5. In Mobile DNA (ed. D. E. Berg & M. M. Howe), pp. 185-2 10. Washington, D.C.: American Society for Microbiology.

Craigie, R. & Mizuuchi, K. 1985 Mechanism of transposition of bacteriophage Mu: Structure of a transposition intermediate. Cell 41, 867-876.

Galas, D. J. & Chandler, M. 1989 Bacterial insertion sequences. In Mobile DNA, (ed. D. E. Berg & M. M. Howe), pp. 109-162. Washington, D.C.: American Society for Microbiology.

Halling, S. M., Simons, R. W., Way, J. C., Walsh, R. B. & Kleckner, N. 1982 DNA sequence organization of IS1O- right of TnlO and comparison with IS1O-left. Proc. natn. Acad. Sci. U.S.A. 79, 2608-2612.

Harayama, S., Oguchi, T. & lino, T. 1984 Does TnlO transpose via the cointegrate molecule? Molec. Gen. Genet. 194, 444-450.

Hirschel, B. J., Galas, D. J., Berg, D. E. & Chandler, M. 1982 Structure and stability of Transposon 5-mediated cointegrates. J. molec. Biol. 159, 557-580.

Isberg, R. R. & Syvanen, M. 1985 Tn5 transposes independently of cointegrate resolution. Evidence for an alternative model for transposition. J. molec. Biol. 182, 69-78.

Kleckner, N. 1989 Transposon TnlO. In Mobile DNA (ed. D. E. Berg & M. M. Howe), pp. 227-268. Washington, D.C.: American Society for Microbiology.

Lichens-Park, A. & Syvanen, M. 1988 Cointegrate formation by IS50 requires multiple donor molecules. Molec. Gen. Genet. 211, 244-251.

Mizuuchi, K. & Craigie, R. 1986 Mechanism of bacteriophage Mu transposition. A. Rev. Genet. 20, 385-429.

Shapiro, J. A. 1979 Molecular model for the transposition and replication of bacteriophage Mu and other transposable elements. Proc. natn. Acad. Sci. U.S.A. 76, 1933- 1937.

Sherratt, D. 1989 Tn3 and related transposable elements: Site-specific recombination and transposition. In Mobile DNA (ed. D.E. Berg & M. M. Howe), pp. 163-184. Washington, D.C.: American Society for Microbiology.

Way, J. C., Davis, M. A., Morisato, D., Roberts, D. E. & Kleckner, N. 1984 New TnlO derivatives for transposon mutagenesis and for construction of lacZ operon fusions by transposotion. Gene 32, 369-379.

Weinert, T. A., Derbyshire, K. M., Hughson, F. M. & Grindley, N. D. F. 1984 Replicative and conservative transpositional recombination of insertion sequences. Cold Spring Harb. Symp. quant. Biol. 49, 251-260.

Received 3 September 1990; revised 17 January 1991; accepted 11 February 1991




a comparison of intramolecular rearrangements promoted by transposons tn5 and tn10

Documents