Proc. NatL Acad. Sci. USA Vol. 78, No. 2, pp. 1090-1094, February 1981

A mechanism ofDNA transposition ( Mu/transposable genetic elements/integrative replication/DNA amplification/chromosomal translocations) BRASIKA M. HARSHEY AND AHMAD 1. BUKHARI Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 11724 Communicated byJ. D. Watson, September 10, 1980 ABSTRACT Bacteriophage Mu and many other transposable elements undergo transposition by a process that involves replica- tion ofthe element. We describe here a mechanism by which such integrative replication maytake place. We have examined electron microscopically the DNA structures generated in host cells after Mu induction and have deduced the following steps in the trans- position process. (i) Association. A protein-mediated association is brought about between the and the target '4' 6 DNA. (ii) Attachment. One end of the element is nicked and at- tached to a site that undergoes a double-stranded cleavage. (iii) Roll in replication. While one strand ofthe target DNAis linked to the nicked strand ofthe element, the complementary strand ofthe target DNA is used as a primer for replication into the element such that the replicating DNA is threaded through the replication complex. (iv) Roi-in termination. When the distal end of the ele- ment arrives at the replication complex, replication is terminated. The roll-in replication mechanism can also explain laying down of tandem repeats-i.e., amplification ofcircular DNA sequences. It has become evident in the last few years that spatial arrange- ments of genetic material may undergo frequent changes both in and . Specific DNA sequences, called "transposable elements," are the primary mediators of such rearrangements (see ref. 1). Studies on bacteriophage Mu, a giant transposable element, suggested that the mechanism oftransposition may involve rep- lication ofthe element (2). The general model ofintegration de- veloped from these studies can be summarized in the scheme 1TE2 + 34 -. 1TE2 + 3TE4 in which 1TE2 is the donor host DNA containing the transpos- able element TE linked to sequences 1 and 2; 3 and 4 represent the recipient DNA sequences. The element appears in both products of the reaction-i.e., is duplicated during transposi- tion. The above scheme can also be written as FIG. 1. Interaction ofmini-Mu with host DNA. (A) Mini- Mu in contact with E. coli DNA. (B) Mini-Mu plasmid with 1TE2 + 34 -- 1TE4 + 3TE2. a tail (key structure). After Mu induction, total cellular DNA was extracted and prepared for electron microscopy by the Kleinschmidt Sequence 1 is now linked to sequence 4 and sequence 3 is linked technique (8). to sequence 2, and the element is present at each linkage. Note that the scheme is formally equivalent to reciprocal chromo- serve interactions between a transposon and its target sites on somal translocations, in-which 1 and 2 represent sequences on the host by electron microscopy using small plas- one chromosome and 3 and 4 are sequences on another. If se- mids containing internally deleted Mu (mini- quences 1 and 2 and sequences 3 and 4 are present on circular Mus) as model systems. We were able to show previously that DNA molecules, then the transposable element-mediated ille- these plasmids made physical contact with the chromosomal gitimate recombination between the two molecules will result DNA (8). We have also examined both the DNA structures gen- in fusion of the two molecules (replicon fusion). Such fused erated in Escherichia coli cells upon Mu induction structures, called "cointegrates," have been observed with al- and those arising from infection by Mu particles. Based on our most alltransposable elements (5-7). observations, we propose here a mechanism for transposition in To study the process of transposition we have sought to ob- which the transposable element is replicated into a new site by roll-in replication. Certain features ofthis mechanism are simi- The publication costs ofthis article were defrayed in part by page charge lar to those proposed earlier (9). payment. This article must therefore be hereby marked "advertise- ment" in accordance with 18.U. S. C. ยง1734 solely to indicate this fact. Abbreviation: kb, kilobase(s). 1090 Downloaded by guest on September 23, 2021 Genetics: Harshey and Bukhari Proc. NatL Acad. Sci. USA 78 (1981) 1091

IL~~~~~~~~~~~~~~~~~~~

D E~~~~~~~~~~~~~~

01~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~1

FIG. 2. Electron micrographs ofkey structures after induction ofMu prophage. The size ofthe circles and the tails attached~to them vary consid- erably. Circle sizes: A, 2.6kb; B, 5.1 kb; C, 17kb; D, 29kb; E, 117 kb. Transposition intermediates and key structures standing the mechanism ofMu transposition. The number ofthe key structures is apparently increased by Pronase treatment and lysogenic for Mucts62, a temperature-inducible mu- phenol extraction. The observation that the circular parts ofthe tant, and containing in addition a mini-Mu plasmid (8) were in- key structures can be smaller than one Mu length implies that duced by the standard procedures (10). The association between only one end of Mu attaches at a target site at one time. Yet we the mini-Mu plasmid and the E. coli chromosomal DNA, after know that both ends ofMu are required for full-scale-Mu repli- induction ofthe Mu prophage, was'investigated by electron mi- cation. Deletion ofone end drastically affects the replication of croscopy (8). We saw structures in which the plasmid seemed to Mu DNA (12). Furthermore, no freely replicating Mu DNA, be attached to the chromosome at a single point. We have also separate from the chromosomal DNA, can be seen. Based on seen structures in which the circle had a standard plasmid size these observations, we will divide the replication/integration but had a tail of an indeterminate length. We will refer to the cycle ofa mini-Mu plasmid into four discrete steps. The model structure in which a circle is attached to a tail as a "key struc- for transposition based on these steps is shown in Fig. 3. This ture. " These structures are shown in Fig. 1. mechanism of DNA transposition can explain the generation of Key structures are also readily seen when a Mu lysogen with- all DNA rearrangements associated with Mu as well as those as- out plasmid is induced. These have been reported (11). They sociated with some other DNA sequences as discussed below. consist of circles of all sizes and have tails of various lengths. Step 1. Association. It has been shown (13) that Mu DNA be- They are seen as early as 10 min after induction, and their num- comes associated with host DNA after infection and that this as- ber increases with time. Their presence is dependent on the sociation requires A and B proteins of Mu. Similarly, the func- products of A and B of Mu. The number of these struc- tioning ofat leastA and B genes is required for interaction ofthe tures is at least 5 times greater when the DNA samples are Mu-containing plasmids with the chromosome (8). Further- treated with Pronase and extracted with phenol forexamination, more, there is some evidence that the association also requires compared to the DNA samples observed without Pronase and active replication of host DNA (14). Thus, Mu proteins must phenol treatments. recognize the ends ofMu or sequences near the ends and must The lengths of the circles in these keys varied from 1 to 200 have affinity for some component ofthe replication complex of kilobases (kb) (Fig. 2). Ofthe 25 molecules examined, more than E. coli DNA. Either one end or both ends of Mu may be in- 50% had a circle size ofless than 38 kb, the length ofMu DNA. volved in bringing about this association. A possible explanation ofhow these structures arise is inherent Step 2. Attachment. Because mini-Mu plasmids always show to the mechanism oftransposition proposed here. single-point contact with host DNA, we postulate that attach- ment occurs at one end ofthe element. A simple way to achieve Mechanism oftransposition this is to have a single-stranded nick at one end ofthe element and a double-stranded break at the target DNA and to link the Because the key structures are generated as a consequence of nicked end ofthe element to one host DNA strand such that the Mu induction, we can assume that they are important in under- complementary host DNA strand provides a primer for DNA Downloaded by guest on September 23, 2021 1092 Genetics: Harshey and Bukhari Proc. Nad Acad. Sci. USA 78 (1981)

Target + a R

Donor

1. Protein-mediated association L R L R -k C O __ _ _ w~~~~~~~~~~~~~~~~~~~~~~~~pu _ _ _ 0 N N R

L 2. Attachment b

L

L R 3. Roll-in replication -l N _

L

4. Roll-in termination FIG. 4. Cointegrate formation versus simple transposition. When the replication complex arrives at the distal end ofthe element, the 3'- end ofthe donor parental strand could be cut and ligated to the target strand (a). This would result in the insertion ofthe entire plasmid car- L rying the transposable element into the target site, generating a coin- tegrate structure. However, if the 3' end of the newly synthesized LIf strand is recognized for ligation (b), this would result in the simple transposition ofthe element into the target site, regenerating the orig- inal plasmid. FIG. 3. Mechanism of transposition. Arrowheads indicate 3'-hy- droxyl ends ofDNA strands; solid circles indicate 5'-phosphate ends. 1. Donor (ww transposable element) and target molecules are brought Step 3. Roll-in replication. Replication of the element then together by a protein that recognizes sequences at both ends of the proceeds by a roll-in mechanism on one strand by extending the transposable element and a sequence in the target molecule. (L and R 3-OH group ofthe exposed complementary target strand and on define the ends ofthe transposable element.) 2. The target site under- the other by discontinuous DNA synthesis (Fig. 3). The whole goes a double-stranded cut. Replication ofthe transposable element is replication complex is held in place while the replicated DNA initiated at one ofits ends by placement of a nick in one strand and li- attached to host DNA is being pulled through and the transpos- gating it to the exposed 5'-P ofthe target strand. The 3'-OH ofthe com- plementary target strand is used as a primer to extend into the ele- ing DNA rolls in. ment; the free target strand is held in place by the replication complex, Step 4. Roll-in termination. When the replication fork fixing the replication point. 3. DNA is replicated at this fixed complex reaches the end ofthe element, replication must be terminated. as it reels through it. DNA synthesis may be discontinuous on the op- The termination step accounts at least partially for the require- posite strand of the element. 4. Replication is terminated when the ment for both ends ofan element during transposition. This ter- other end ofthe element passes through the replication complex. mination must result in cointegrate formation for a mini-Mu plasmid and for several other transposable elements (ref. 7; un- replication into the element as shown in Fig. 3. The double- published data). However, we must make an allowance for an stranded breaks in host DNA are staggered to account for small integration event in which a simple transposition rather than duplications in host sequences generated at the sites of inser- cointegrate formation takes place-for example, transposition tion, as hypothesized by Grindley and Sherratt (9) and Calos et from linear Mu DNA during lysogenization. Termination is thus a! (15). a difficult step to define and may have alternative modes. Downloaded by guest on September 23, 2021 Genetics: Harshey and Bukhari Proc. Nad Acad. Sci. USA 78(1981) 1093 Cointegrate Formation and Simple Transposition. At the Formation ofKey Structures. It is not clear exactly how the completion ofthe roll-in replication, the end ofthe element pre- replication complex is held together, but we assume that pro- sents itselffor termination. Two possibilities arise at this point. teins in some manner stabilize the complex and that it is fragile. The 3' end ofthe donor parental strand could be cut and ligated Disruption ofthis complex during isolation ofDNA would yield to the target site (Fig. 4a). Thus the entire donor plasmid will a structure in which the plasmid has a tail attached to it (Figs. 1 insert into the chromosome, resulting in acointegrate. If, on the and 3). Similarly, when transposition occurs within the same re- other hand, the 3' end ofthe newly synthesized strand would be plicon (Fig. 5), disruption of the replication complex would recognized for ligation, the termination products would segre- yield circles with tails. The size of the circles would vary, de- gate, leaving a copy ofthe element at the target site (Fig. 4b). pending on the distance between the transposing end ofthe ele- If transposition were occurring within the same replicon, the ment and the site on the host DNA where transposition occurs. cointegrate mode ofinsertion would result in inversion ofDNA The size ofthe tails would also similarly vary. (Fig. 2). sequences between the two copies of the element (ref. 16; see Heterogeneous Circle Formation. Covalently closed circles Fig. 5). containing Mu DNA and host DNA are formed during Mu rep- Resolution ofthe termination structure is an intriguing prob- lication in E. coli (11, 17). The orientation ofthe ligation step will lem. We know that cointegrates are the end products ofMu in- determine the manner in which these circles are generated as a sertion during the lytic cycle and that there is no efficient mech- consequence oftransposition within the same replicon as shown anism for resolving these cointegrate structures (ref. 8; in Fig. 5. Ifreplication is completed, both newly formed circles unpublished data). We also know that, upon infection ofthe host will have one copy ofMu each. The result would be deletions of cells with Mu, linear Mu DNA integrates into the chromosomal host DNA immediately adjacent to the ends ofthe inserted ele- DNA to give stable lysogens. In this case there can be no coin- ment. This is implicit in a model proposed by Shapiro (18). Be- tegrate mode of integration of linear DNA because that would cause in our proposal the initiation complex looks like the ter- disrupt the chromosome. Thus, lysogeny demands a simple mination complex, the transposition process may be aborted at transposition event. It would seem therefore that resolution of the very beginning. We predict, therefore, that heterogeneous the termination structure can be modulated between cointe- circle formation sometimes would result in a circle with Mu and grate formation (fusion of replicons when two replicons are in- a circle without Mu. The net result would be a simple excision volved) in one case and the segregation of donor and recipient ofMu with adjacent host sequences. molecules in the other case. The decision between the two path- Formation of Substitution Deletions. Occasionally, both ways may be made by specific proteins. ends of the element may be able to attack independent target a b Target site Target site A E C BD R L Donor sequence Donor sequence Protein-mediated association Protein-mediated association A C BOtQ

Roll-in replication Attachment and roll-in replication L E ABWLGJPD RI '---LED

R oel Ait A L R D C

By -- L Roll-intermination Protease treatment FIG. 5. Consequences oftransposition within the same replicon. (a) The basic steps are similar to those outlined in Fig. 3. Cointegrate formation would lead to inversion ofmarkers between the transposed sites (bottom left structure). Ifthe process is disrupted during DNA extraction, it would result in the appearance ofkey structures (bottom right structure). (b) Depending on the orientation ofthe ligation reaction, the transposition prod- ucts would be circles with one copy ofthe element each, ifthe replication goes to completion and circles without any element on it if replication is aborted at the beginning. These circles would now represent deletions ofhost DNA sequences from either end ofthe transposon. Downloaded by guest on September 23, 2021 1094 Genetics: Harshey and Bukhari Proc. Nad Acad. Sci. USA 78 (1981) licated semiconservatively by extension of the exposed 3'-OH IN - Host DNA group ofthe host strand. D( A The basic model oftransposition by roll-in replication can be extended to several modes ofDNA-DNA interaction. The yeast transposable element Tyl containing two a sequences (25) may C B give rise to a circular intermediate with only one 6 sequence. This 6 sequence may again be duplicated at a new site by the Attachment and roll-in replication roll-in replication. Any circular molecule can integrate itselfby such a mechanism if it has the necessary proteins to associate and attach to the host DNA. This particularly applies to circular molecules such as simian 40. The mode ofexcision ofthe circular molecule containing a transposable element, either through a completed transposition event or through an aborted event, can be used to pick up new host sequences by the trans- posable element. 1Termination |[Termination We thank Frank Stahl and Max Delbruck for discussions and encour- agement. This work was supported by grants from the National Science ABCDAB ABCDABCDAB Foundation (7826710) and the National Institutes of Health (GM23566) and a Career Development Award to A. I. B. from the National Institutes ofHealth. FIG. 6. Integration of a circular DNA element into the chromo- some and formation of tandems. The replication scheme is similar to 1. Bukhari, A. I., Shapiro, J. & Adhya, S., eds. (1977) in DNA Inser- the one outlined in Fig. 3. A random termination site would result in tion Elements, Plasmids and Episomes (Cold Spring Harbor Lab- the formation ofpartial tandems. oratory, Cold Spring Harbor, NY). 2. Ljungquist, E. & Bukhari, A. I. (1977) Proc. Natl Acad. Sci. USA sites simultaneously. Two double-stranded breaks of such type 74,3143-3147. would result in loss of DNA between these sites with a substi- 3. Bukhari, A. I. (1977) in Genetic Interaction and Transfer, tution ofa copy ofthe element (see ref. 19 for deletions accom- Brookhaven Symposia in Biology, ed. Anderson, C. W. 29, 218-232. panied by Mu insertions; unpublished data). This phenomenon 4. Bukhari, A. I., Ljungquist, E., deBruijn, F. & Khatoon, H. (1977) may be similar to the behavior ofmating-type cassettes in which in DNA Insertion Elements, Plasmids and Episomes, eds. Buk- a copy of the silent locus is transposed to the active locus with hari, A. I., Shapiro, J. & Adhya, S. (Cold Spring Harbor Labora- consequent loss ofresident information at the active locus (20). tory, Cold Spring Harbor, NY), pp. 249-261. Formation of Tandem Repeats and Amplification of DNA 5. Toussaint, A. & Faelen, M. (1973) Nature (London) New Biol 242, Sequences. An important consequence of the roll-in model is 1-4. 6. Gill, R., Heffron, F., Dougan, E. & Falkow, S. (1978)J. Bacteriol laying down oftandem repeats while one end is attached to the 136, 742-756. target site. This would happen if the process of termination is 7. Starlinger, P. (1980) Plasmid 3, 241-259. inefficient (Fig. 6). Circular insertion elements that have no 8. Chaconas, G., Harshey, R. M. & Bukhari, A. I. (1980) Proc. Natl fixed initiation or termination sites for integrative replication Acad. Sci. USA 77, 1778-1782. would behave in this manner. A noteworthy example is simian 9. Grindley, N. D. F. & Sherratt, D. J. (1978) Cold Spring Harbor virus 40 which requires the appropriate functioning of the A Symp. Quant. Biol 43, 1257-1262. 10. Bukhari, A. I. & Ljungquist, E. (1977) in DNA Insertion Elements, gene for replication and integration and generates tandem re- Plasmids and Episomes, eds. Bukhari, A. I., Shapiro, J. & Adhya, peats (21). S. (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY), pp. 749-756. 11. Schr6der, W., Bade, E. B. & Delius, H. (1974) Virology 60, General discussion 534-542. 12. Van de Putte, P., Giphart-Gassler, M., Goosen, T., van Meeteren, Since the original proposal of transposition by replication was A., & Wijfffelman, C. (1978) in Integration and Excision ofDNA advanced (2), various structures have been proposed to recon- Molecules, eds. Koyschneider, P. & Starlinger, I. (Springer, Hei- cile the of and integration. Particularly delberg), pp. 33-40. processes replication 13. Ljungquist, E. & Bukhari, A. I. (1979)J. Mot BioL 133, 339-358. noteworthy are the intermediates suggested by Grindley and 14. Ljungquist, E., Khatoon, J., DuBow, M. Ambrosio, L. deBruijn, Sherratt (9) and Shapiro (18). Based on the work of Calos et al. F. & Bukhari, A. I. (1978) Cold Spring Harbor Symp. Quant. Biol. (22) and Grindley (23), Grindley and Sherratt (9) adopted the 43, 1151-1158. idea ofstaggered cuts at the host target site to account for small 15. Calos, M., Johnsrud, L. & Miller, J. F. (1978) Cold Spring Harbor duplications ofthe host sequences present at thejunctions ofthe Symp. Quant. Biol 43, 1257-1262. element. The insertion element then attaches to the 16. Faelen, M. & Toussaint, A. (1980)J. Bacteriol. 142, 391-399. inserted 17. Waggoner, B. T., Gonzalez, M. S. & Taylor, A. L. (1974) Proc. nicks and replicates from one end to the other such that the end Natl Acad. Sci. USA 71, 1255-1259. result is a conservative segregation ofthe daughter strands. Ar- 18. Shapiro, J. A. (1979) Proc. Natl Acad. Sci. USA 76, 1933-1937. thur and Sherratt (24) later favored a different structure pro- 19. Bukhari, A. I. (1975)J. Mol Biol 96, 87-99. posed earlier by Shapiro (18). In the Shapiro structure, both 20. Hicks, J., Strathern, J. & Klar, A. (1979) Nature (London) 282, ends of the element attack the nicked target site and get at- 478-483. tached to this site. of the element then invariably 21. Botchan, M., Topp, W. & Sambrook, J. (1976) 9, 269-287. Replication 22. Calos, M. P., Johnsrud, L. & Miller, H. H. (1978) Cell 13, results in a cointegrate structure. To best account for the struc- 411-418. tures reported here, we propose a roll-in replication in which 23. Grindley, N. D. F. (1978) Cell 13, 419-426. one end of the element is attached to the target site, the repli- 24. Arthur, A. & Sherratt, D. (1979) Mol Gen. Genet. 175, 267-274. cation complex is fixed, and the DNA is reeled through and rep- 25. Roeder, S. & Fink, G. R. (1980) Cell 1,239-249. Downloaded by guest on September 23, 2021