Chapter 140

Transposable Element Tools for Microbial Genetics

CLAIRE M. BERG and DOUGLAS E. BERG

INTRODUCTION

Transposable elements are discrete DNA segments that can repeatedly insert into a few or many sites in a host genome (56). Transposition occurs without need for extensive DNA sequence homology or the rec gene functions required in classical homologous recombination. Transposable elements have proven to be superb tools for molecular genetics: (i) as insertion mutagens, resulting in stable loss-of-function mutations that are easy to map and manipulate; (ii) as sources of selectable markers that facilitate mapping, localized mutagenesis, in vitro cloning, and construction of recombinant strains; (iii) as agents for generating replicon fusions, deletions, and inversions; (iv) as portable restriction sites for in vitro DNA manipulation, gene mapping, and large-scale genome mapping; (v) as portable regions of homology for generating replicon fusions, deletions, duplications, and inversions by homologous recombination; (vi) as vectors for moving virtually any gene or sequence to fixed or random sites in the genome; (vii) as carriers of reporter genes for analyzing the control of gene expression, detecting genes that are expressed only under particular circumstances, protein localization, and localizing domains in membrane-spanning proteins; (viii) as carriers of transcriptional terminators for analyses of operon organization and transcriptional and translational regulation; (ix) as carriers of regulated promoters for generating conditional mutations; (x) for in vivo cloning; and (xi) as mobile primer binding sites for sequencing cloned DNAs.

Many uses of transposable elements were first developed and exploited in laboratory strains of Escherichia coli and Salmonella typhimurium (official designation, Salmonella enterica serovar Typhimurium), but the approaches and also the elements are well suited for use in clinical or other natural isolates and in many other bacterial species (17, 18, 19, 45a, 69, 176, 176a, 194a, 198). This chapter reviews uses of some transposable elements (Table 1) that are valuable in E. coli and S. typhimurium and aspects of their transposition behavior that contribute to their utility as tools. Chapter 124 provides a general review of transposable elements, with emphasis on biochemical mechanisms. Additional information can be found elsewhere (32, 199, 254).

Table 1Structure and properties of commonly used wild-type transposonsa

Transposable elements are diverse in size, structure, specificity of insertion, mechanism of transposition, and regulation of movement and may have several phylogenetic origins. The simplest elements, designated insertion sequence (IS) elements, generally contain only a transposase gene flanked by sites on which transposase and host factors act in moving the element. Most IS elements are less than 2 kb in size. E. coli K-12 contains at least eight distinct IS elements, none of which are related to the one IS element found in S. typhimurium LT2, although some are present in other S. typhimurium strains and in many clinical and environmental isolates of E. coli (38, 103; see chapter 111).

More complex transposable elements, designated transposons, contain genes encoding auxiliary traits such as antibiotic resistance, heavy metal resistance, or virulence determinants, in addition to genes and sites needed for transposition. Many transposons, such as Tn5, Tn9, and Tn10, are composites, consisting of direct or inverted repeats of an IS element bracketing the auxiliary gene(s), while others, such as Tn3 and γδ (Tn1000), do not contain functional IS elements.

Temperate integrating bacteriophages such as Mu and lambda (λ) are more complex than typical transposons in containing genes for DNA replication, phage development, and cell lysis, as well as the genes and sites needed for transposition (integration). Although mechanistic considerations now lead many to consider λ distinct from "true" transposable elements (see chapter 124), we include it for historical and operational reasons: transposable elements are defined as "specific DNA segments that can repeatedly insert into a few or many sites in the genome" (56). Furthermore, the ability of λ to insert into many different sites made it very important for in vivo gene cloning before in vitro recombinant DNA methods were available.

Despite the variety of structures, the minimum requirements for transposition are quite simple: the sites of action of transposition proteins (in many cases inverted repeats of only 18 to 38 bp at each end of the element), the cognate element-encoded and element-specific transposases, and one or more specific host factors. The simplicity of these requirements has permitted the engineering of many special-purpose derivatives of naturally occurring transposable elements. The most useful elements for molecular genetics in E. coli and S. typhimurium include transposons Tn3 (Amp^r) and the closely related γδ element (no resistance), Tn5 (Kan^r), Tn10 (Tet^r), and phages Mu and λ (no resistance).

SOME USEFUL HISTORY

Phage Mu, the first bacterial transposable element to be recognized, is a temperate phage that attracted attention because it occasionally generated mutations during lysogenization (285, 289). Mu was correctly interpreted as a prokaryotic analog of the eukaryotic transposable controlling elements that McClintock had discovered nearly 20 years earlier (194). The first use of Mu as a genetic tool was in the study of genes needed for lysine and diaminopimelic acid synthesis (51).

In the next dozen years other classes of transposable elements were discovered and their value as tools for genetic analysis was also immediately evident. IS elements were found a few years after Mu to be the cause of novel spontaneous mutations that were often associated with loss of function of adjacent genes but that mapped genetically as point mutations, that could revert spontaneously, and that were not suppressible or stimulated to revert by mutagens (147, 250). The initial characterizations of IS elements included recombining chromosomal gal mutant alleles into λ gal ⁺ transducing phage and determining the DNA content of the resultant phage particles. Many gal mutant phage exhibited higher-than-normal buoyant densities in CsCl equilibrium gradients, indicating that they contained additional DNA. Electron microscope heteroduplex studies of several such phages demonstrated two different types of insertions (called IS1 and IS2) in gal DNA (94, 138, 269). Thus, although it was clear that IS elements could cause mutations, IS elements could not yet be easily harnessed for molecular genetics because they were not marked genetically and no method for selecting their movement was known.

A few years later, phage λ was found to insert into many secondary chromosomal sites in strains that lacked the primary λ attachment site (255, 256). Aberrant excision of λ prophages at these sites yielded specialized transducing phages that carried chromosomal genes from one or the other side of the prophage (244, 255). This allowed the in vivo cloning of many different segments of the E. coli chromosome in λ phage vectors before in vitro cloning methods became widely available.

Several transposable elements carrying antibiotic resistance genes ("transposons") were recognized almost simultaneously in the mid 1970s as natural components of R-factor (antibiotic resistance) plasmids by their ability to insert into other DNA molecules (11, 31, 98, 108, 129, 131, 159, 167) (Table 1). For example, the Kan^r Tn5 element was discovered during an attempt to exploit λ integration and aberrant excision, described above, to clone resistance genes from R-factor plasmids. However, the segment containing kan mapped as a simple insertion at different sites in different isolates, indicating that the λ-kan phage had arisen by translocation, not by the expected λ integration adjacent to kan and aberrant excision (31).

The resistance genes in transposons made it easy to select for their transposition (e.g., from λ or P22 phage vectors to the E. coli or S. typhimurium chromosome), to isolate insertion mutations, and to begin analyzing transposition mechanisms. Interest in the resistance transposons was also high because, along with selection pressure caused by antibiotic usage in medicine and agriculture, their existence helped explain the disturbing spread of antibiotic resistance in bacteria.

Recombinant DNA methods became widely available a few years after the discovery of resistance transposons, and their resistance genes were frequently incorporated into plasmid cloning vectors (e.g., amp from Tn3, kan from Tn5 or Tn903, and cam from Tn9). The synergism between transposon-based in vivo and recombinant DNA-based in vitro approaches continues to this day in the numerous sophisticated special-purpose derivatives of several transposons (Tables 1 and 2), in the development of suicide approaches and vector systems for delivering transposons to target DNAs (Table 3), in the use of their selectable genes in a variety of applications, and in the development of transposon-based systems for in vivo and in vitro cloning (Table 2 and below).

Table 2Useful engineered transposonsa,b

Table 3Genes and strategies potentially useful for transposon deliverya

TRANSPOSITION MECHANISMS AND PRODUCTS IMPORTANT FOR TRANSPOSON USE

The structures of transposition products and the mechanisms of transposition are important in the use of transposons as molecular genetic tools. Early studies of the intermolecular transposition products of several different transposons revealed simple insertions of the element, free of vector DNA. The finding of such simple insertions after Tn5 and Tn10 transposition, plus the precedent provided by phage λ integration, led us to propose a conservative (nonreplicative) transposition model. In this model the element is separated from vector DNA by a double-strand break at each end. It is inserted into target DNA without being replicated, and the linearized vector is lost (Fig. 1A) (26, 28). Evidence of an alternative, replicative, transposition mechanism was provided by the finding of cointegrates consisting of vector and target DNAs joined by direct repeats of the element (Fig. 1B) with certain mutant derivatives of Tn3. Further studies showed that cointegrates are the primary products of Tn3 transposition, that they are normally broken down very efficiently by a transposon-encoded, site-specific recombination enzyme called resolvase, or more slowly by RecA-mediated homologous recombination, and that the mutants had lesions in either the resolvase gene or the resolution site (130, 253). Phage Mu was found to generate cointegrates and other rearrangements by transposition during lytic growth, although it does not encode a function corresponding to resolvase (62, 201).

Fig. 1Formation of simple insertions by conservative and replicative termolecular transposition. (A) Conservative transposition of Tn5 from ? to the bacterial chromosome on the (left) or to a Plasmid(right).Transposition is obtained by selecting for kanr colonies after infection with a ? phage that carries Tn5 and lacks the phage attachment site. Plasmids containing insertions are obtained by extracting DNA and transforming a new host. The linearized donor fragment is presumed to be destoryed during transposition (26, 28, 29, 33). (B) Replicative transpositiion of ?? from the F factor to a nonconjugative plasmid. The cointegrate product is resolved in a second recombination step (either site specific or generalized) (251, 253). Cointegrates formed between F and the garget nonconjugative plasmid are selected by conjugal transfer (119). The E.coli F factor, contains IS elements as well as ?? but ?? is themost active in this selection. A derivative of the F factor, pOX38, that is deleted for its naturally occurring transposable elements, including ?? (120), is used to deliver engineered ?? derivatives (24, 25). kept in mind when evaluating literature from that period. While intermolecular transposition yields simple insertions or cointegrates (Fig. 1), intramolecular transposition yields deletions and inversions (see below). These rearrangements are rare but can be recovered efficiently by selecting for loss of a gene that is near the transposon or that is between the ends of the two IS elements of composite transposons (2, 3, 97, 158, 231, 288, 297, 298).

Source:

In the replicative transposition model that was proposed to account for the findings with Tn3 and Mu, single-strand nicks at each element end allow the joining of element and target sequences, followed by copying of the element from the free 3' ends of target DNA (9, 111, 251; see chapter 124). The impact of the unexpected cointegrate structures and the appeal of a single transposition mechanism led some to assume in the early 1980s that all transposition was replicative (see references 103, 125, 155, 156, and 316). However, the insertion of Mu DNA after infection, like Tn5 transposition, seemed to be conservative (4, 177), and Tn9 (IS1) was found to make both simple insertions and cointegrates (37, 185). Accordingly, it seemed that the choice between conservative and replicative mechanisms might be based on whether transposition was initiated with double- or single-stranded cuts at each transposable element end to give conservative and replicative tranposition, respectively (30, 37). Widespread acceptance of two modes of transposition, either conservative or replicative, depending on the element or the conditions, began to emerge in the mid-1980s partly because of additional evidence of conservative transposition after Mu infection (124) and new studies indicating that Tn10 and IS903, like Tn5, transpose conservatively (160, 302). The assumption by some in the early 1980s that there is a single replicative mechanism should be kept in mind when evaluating literature from that period.

While intermolecular transposition yields simple insertions or cointegrates (Fig. 1), intramolecular transposition yields deletions and inversions (see below). These rearrangements are rare but can be recovered efficiently by selecting for loss of a gene that is near the transposon or that is between the ends of the two IS elements of composite transposons (2, 3, 97, 158, 231, 288, 297, 298).

Intermolecular Transposition

The very first studies of IS elements and transposons showed that transposition is rare, with most such elements moving at frequencies in the range of 10^–7 to10^–4 per cell, regardless of whether the host is rec+ or recA. Hence, much of the power of transposon-based methods depends on the ability to select transposition products—separating them from the many potential target DNAs that had not sustained transposition and also from the original donor DNAs or cells. Good selectable markers are essential, and antibiotic resistance genes have been used most often. Genes for heavy metal resistance, for biosynthetic or other metabolic enzymes, or for suppressor tRNAs have also been valuable in a number of instances (Tables 1 and 2; see chapter 139 for additional markers). An interesting exception to the general need for selectable markers in the transposon is provided by the strategy for selecting γδ transposition from the conjugative F factor to a nonconjugative plasmid (Fig. 1B).

As described above, the product of conservative intermolecular transposition is a simple insertion (Fig. 1A), while the product of replicative transposition is a cointegrate that can be resolved to yield also a simple insertion product (Fig. 1B). However, two types of conservative transposition yield structures that resemble cointegrates but that should not be mistaken for them: (i) "inverse transposition" (63)(described below) and (ii) "dimer transposition," in which transposition from a dimer that contains two transposon copies (and thus four IS elements for composite transposons such as Tn5) yields a product that contains vector sequences, as well as transposon DNA. However, unlike true cointegrates, these transposition products do not contain the entire donor molecule; the size of the donor fragment that is transposed depends upon which pair of IS elements participated in the transposition event (28).

Inverse Transposition (63).

Inverse transposition, which involves the inside, rather than the outside, ends of IS elements in composite, conservatively moving transposons such as Tn5 and Tn10, was discovered in studies of integrative suppresson (211) by a Tn10- containing R-factor plasmid (63). Most suppression events resulted from insertion of the R factor into the host chromosome, thereby allowing the R factor to take over chromosomal replication. The inside ends of the IS10 elements mediate inverse transposition so that the DNA segment that moves is the entire R factor, except the Tetr determinant. In such inverse transpositions the central portion of the transposon corresponds to the "vector." Thus, inverse transposition, like replicative transposition, can fuse different replicons, but the products differ from true cointegrates (generated by replicative transposition) in lacking the DNA from the central portion of the transposon and in not having the transposon duplicated(63, 207, 241).

Intramolecular Transposition

Transposition from one site to another in the same DNA molecule (intramolecular transposition) is mechanistically equivalent to intermolecular transposition, but the consequences are quite different and generally less familiar. Intramolecular transposition yields an inversion or a deletion product in a single step, independent of whether the element moves by a conservative or a replicative mechanism. The structures of intramolecular transposition products depend, however, on the number of twists between the donor and target sites and on whether transposition was conservative or replicative. In conservative intramolecular transposition, which is a form of inverse transposition, the inside ends of the element participate and the DNA between the transposon ends is lost (Fig. 2), while in replicative intramolecular transposition no intermediate corresponding to a cointegrate is formed (Fig. 3) and there is no resolution step.

Fig. 2Formation of deletion and inversion products by conservative intramolecular transposition. The inner, rather than the outer, IS element ends of these composite elements are involved in this intramolecular inverse transposition. The "vector" (marked by diagonal lines) is not found in transposition products. One of the two expected reciprocal deletion products is generally inviable because it lacks a replication origin. Inversion derivatives are smaller than the parent molecule by the size of the donor fragment (228; C. Chin and C. M. Berg, unpublished data).

Source:

Fig. 3Formation of deletion and inversion products by replicative intramolecular transposition. The transposon is duplicated and no DNA is lost during the transposition process, although one of the expected deletion products is generally inviable because it lacks a replication origin. Inversion derivatives are larger than the parent molecule by the size of the transposon (298, 303).

Source:

In both conservative and replicative transposition, the formation of a deletion versus an inversion depends on the relative orientation of donor and target sites. Deletions arise when there is an odd number of twists between the donor and target sites (Fig. 2 [left] and 3 [left]), and inversions arise when there are no or an even number of twists between the sites (Fig. 2 [right] and 3 [right]). The simple expectation that deletions and inversions with endpoints in a target gene should be equally frequent is borne out for replicative γδ transposition (298) but not for conservative Tn5 transposition (288), where deletions were severalfold more frequent than inversions, with the excess probably occurring by abortive transposition (146, 288).

It should be noted that, regardless of the frequency with which they arise, deletions always far outnumber inversions among recombinant products if loss of gene function is selected. This is because transposition to sites either in or beyond the counterselected gene results in loss of function in deletion derivatives, whereas only transposition to sites in the gene results in loss of function in inversion derivatives (288, 298).

Conservative Intramolecular Transposition.

If the outer ends of composite transposons such as Tn5 and Tn10, or of their IS components, participate in intramolecular transposition, linear, inviable products are expected. Therefore, findings of additional transposed copies of Tn5 or Tn10 or one of their component IS elements in a chromosome that already carries a copy of the element probably reflect transposition from one chromosome to its sibling chromosome—that is, an intermolecular event (as has been adduced for IS10-adjacent deletions [237a]). However, if the inner transposon ends participate in transposition, deletions and inversions result (15, 19, 158, 288). This inside-end intramolecular transposition is equivalent to inverse intermolecular transposition (63) (described above). In both situations the central segment of the transposon serves as the vector and is linearized and lost. It should be noted that the single inversion product retains both IS elements, while each of the two predicted deletion products retains one IS element; in both cases the transposition product(s) is smaller than the parent clone by the size of the interstitial vector fragment (Fig. 2) (19, 288).

The lucky circumstance that the functional tet gene of Tn10 can be selected against (Table 3) led to the widespread use of Tn10 to isolate deletions and other rearrangements adjacent to the insertion site (43, 97, 188). Rounds of Tn10 insertion and selection for loss of Tet^r have been used repeatedly to inactivate a series of genes in the construction of some laboratory strains. In these cases, one or both IS10 elements are expected to be retained after each conversion of a Tet^r to a Tet^s strain (Fig. 2). The retention of IS10, which was not always understood, can cause problems when Rec⁺ strains with a history of Tn10 insertion mutagenesis and Tet^s selection are used. For example, one study of a plasmid that happened to contain a hot spot for IS10 insertion revealed that the host strain retained three active IS10 elements, one from a Tn10-mediated deletion event and a pair from a Tn10-mediated inversion event, each associated with prior Tn10 mutagenesis and selection for Tet^s (44). In another study, most mutants selected for constitutive expression of the udp gene resulted from rearrangements mediated by a previously unrecognized IS10 element in the host (I. Brikum, A. Mironov, and D. E. Berg, unpublished data). Fortunately, chromosomal IS10 elements appear to be stable in recA strains (214a; J. P. O’Neil, P. J. L. Morris, and G. Bogosian, personal communication), even though Tn10 (IS10) transposition in other assays (e.g., from phage vectors) is RecA independent (157).

Replicative Intramolecular Transposition.

In replicative transposition the entire element is duplicated and no DNA is lost. Consequently, inversion products generated by replicative transposition are larger than the parent molecule by the size of the transposon (39, 291, 298, 303). In the deletion mode, two circular products should be produced, each containing a copy of the element, and their net size should also be larger than that of the parent molecule by the size of the transposon (Fig. 3). With the exception of Tn9, most elements known to undergo replicative transposition are not compound elements: they contain only one pair of ends capable of undergoing transposition.

The reciprocity of deletion products produced by either replicative or conservative transposition has not yet been examined at the molecular level because one of the two products is usually selected against or lacks a replication origin.

TARGET SPECIFICITY

Transposable elements differ greatly in insertion specificity, with none moving completely randomly. Most move at relatively low frequencies (∼10^–7 to 10^–4) and as readily to the chromosome as to plasmids. However, members of the Tn3 family, including γδ and Tn1721, transpose poorly into the chromosome, even though they insert readily and more randomly than most other elements into plasmid DNAs (130, 253). Insertions of wild-type Mu and its large engineered derivatives, in contrast, are generally not recovered in multicopy plasmids. This is probably due to selection against multiple copies of particular Mu genes, not to an inherent feature of the Mu transposition mechanism, because insertions of small mini-Mu elements into plasmids are easily obtained. Tn7 and phage λ insert preferentially into just one chromosomal site. However, when their primary chromosomal sites are deleted, each of these elements is found to insert at numerous secondary sites, albeit at lower frequencies (72, 255, 256).

Most elements generate short direct duplications of a sequence at the target site that is 5 bp long in the cases of Mu, Tn3, γδ, and Tn7 and that is usually 9 bp long in the cases of Tn5, Tn9, and Tn10 (phage λ inserts by site-specific recombination). How insertion sites are chosen is complex and different for each element. Except for Tn7 and λ, specific target sequences may often be less important than properties of the intracellular DNA milieu, including supercoiling, transcription, and methylation (Table 1) (29, 136, 157, 300).

Tn3 and γδ transpose most randomly to most plasmid targets, with relatively few hot spots or cold regions. Tn5 and Mu exhibit intermediate specificity for plasmid or chromosomal targets, with a majority of insertions into random sites, but some hot spots. Tn9 and Tn10 transpose nonrandomly, with most insertions into just one or a few hot spots per target gene (Table 1). However, reduced-specificity Tn10 transposase alleles (14) have been incorporated into several mini-Tn10 delivery systems (5a, 187a; J. Mahillon, C. K. Rode, and C. A. Bloch, personal communication) (Table 2).

The choice of transposon is affected by the purpose of the project at hand. Almost any transposon can be useful if just one or a few insertions in any given gene is needed. However, when a high density of insertions is needed in certain applications, e.g., to provide primer-binding sites for DNA sequencing, then one of the more randomly transposing elements should be used.

POLARITY AND DISTAL GENE EXPRESSION

Most IS element and transposon insertions in an early gene in an operon prevent transcription of distal genes from the operon promoter because of transcriptional stop signals or polarity caused by long stretches of untranslated mRNA (103). Some insertions appear incompletely polar, with distal genes being expressed constitutively. This expression can be due to the following: (i) outward facing promoters in the transposon and transcription across its end (Tn3 stimulates distal transcription by read-through from the transposase gene [130], both Tn5 and Tn10 stimulate distal transcription due to outward-facing promoters near their termini [35, 68, 184, 294], and the inside end of IS50L [Tn5] stimulates distal transcription by read-through from the promoter used for expression of the kan ble str resistance operon in Tn5 [27, 152]); (ii) new promoters formed by insertion (some elements contain potential "–35" sequences close to their ends and can form new promoters when they insert near potential "–10" sequences in target DNA); and (iii) internal promoters in the operon (transposons have been useful for ordering genes in operons and for detecting significant internal promoters [e.g., references 22 and 23] when element-associated promoters are low in activity [35, 68, 294]).

The turn-on of distal gene expression is not always due to transposon promoter activity: new insertions of IS1 and IS5 cause most of the mutations allowing expression of the cryptic bgl operon in E. coli K-12. These elements seem to act at a distance by providing an enhancer function, rather than a new promoter. This may be because of an effect on DNA supercoiling that allows the normally silent promoter to function either directly or by facilitating cyclic AMP receptor protein binding (103, 232). Conversely, clones containing Helicobacter pylori groES and groEL (chaperone protein) genes were found to acquire an upstream insertion of IS5 during propagation in E. coli, apparently because the IS5 insert decreased the normal high-level expression of these cloned genes, which was deleterious in E. coli (280). A complementary position effect has been found with a cloned avtA gene, in which some γδ insertions that knocked out avtA expression were downstream of the structural gene. AvtA⁺ revertants turned out to carry dimeric plasmids, probably reflecting changes in the local supercoiling by dimerization and its effects on gene expression (179; C. M. Berg and L. Liu, unpublished data).

STABILITY OF TRANSPOSON INSERTION MUTATIONS: EXCISION AND REVERSION

Two aspects of transposable element stability are important when using transposons as genetic tools: frequency of loss and the frequency of movement to new sites. Each of these parameters can be tested easily by using insertions in biosynthetic or catabolic genes. The results indicate that chromosomal insertions are quite stable, with revertants generally occurring at frequencies of less than 10^–8. In addition, revertants (or transductants to prototrophy) of a strain carrying Tn5 or Tn10 in a biosynthetic gene are almost always Kan^s or Tet^s, respectively, and devoid of all transposon sequences. This indicates that the elements rarely move to a second site in the genome (29, 157). The very few cases where Tn5 or Tn10 (or one of their respective IS elements) is found at a second genomic site may be due to prior intermolecular transposition between sibling chromosomes (or partially replicated chromosomes) because, as described above, these elements are not replicated during transposition and loss of the vector sequences would be lethal in intramolecular transposition.

Many of the current generation of engineered elements can transpose from their delivery vector with high efficiency but are then crippled for further movement because they depend on a transposase gene cloned in the vector or another plasmid (Table 2). However, many older, well-characterized insertions continue to be used, so secondary movement must still be considered.

Several cases of strains containing IS50 or IS10 at new sites have been found in Rec⁺ cells (44, 126; I. Brikun, A. Mironov, and D. E. Berg unpublished data). This accumulation of transposed elements is not found in RecA cells (214a; O’Neil et al., personal communication). It may be enhanced by prolonged storage of strains in room temperature stabs and probably on plates on the benchtop as well (8) but is generally not seen with freshly isolated insertion mutations or after transduction of an insertion mutation into a fresh strain. Strains that carry the transposon or component IS element at new sites also generally retain the complete element at its original site.

Most mutations due to insertion of Tn5 or Tn10 revert by loss ("excision") of the element. Revertants are generally Kan^s or Tet^s, respectively, and lack transposon sequences. In true revertants the element plus one copy of the 9-bp target sequence that had been duplicated during insertion must be lost in order to restore the ancestral gene sequence. These events depend on the inverted repeat structure but not the transposase functions of the elements and are not correlated with transposition to new sites. Rather, they are ascribed to mistakes in DNA replication: slippage of the growing DNA strand between the first and second copy of the 9-bp direct repeat, which may be triggered by hairpin formation between complementary sequences of the inverted repeats and then failure to copy the entire element. No actual DNA cleavage need be involved, nor is the excision process related to transposition per se. The following lines of evidence support this interpretation. (i) Insertions early in an operon are usually polar, and revertants selected for distal gene expression are often partial revertants, still mutant in the target gene. For example, in strains with Tn5, Tn10, or Mu insertions in the lacZ gene, selection for revertants in the downstream lacY gene yields many partial revertants that are antibiotic sensitive and have lost much, but not all, of the Tn or Mu element but have not regained lacZ function (26, 45, 50). Upon selection for LacZ⁺, some partial revertants give rise to true revertants: in these the initial deletion involved only part of the element (50, 91, 99, 239). Partial revertants that can give rise to true revertants are relatively common for Tn10 insertions, probably because there is a 24-bp direct repeat near each outside end of Tn10 that can serve as a substrate for slippage during replication. (ii) The frequency of Tn5 or Tn10 excision depends on the insertion site, even within a single gene (91), and is enhanced by certain mutations in DNA repair or recombination genes (182, 183). (iii) The excision of Tn5 and Tn10 is more frequent in F⁺ than in F^– cells and is strongly stimulated by moving the mutant allele from the chromosome to an F' plasmid. This excision-enhancing effect of F is exerted primarily in cis. It appears to depend on F-factor transfer between cells, probably because F transfer involves single-stranded DNA intermediates (which can form hairpin substrates that promote slippage during replication) (34).

Mutations due to Mu insertion are seen to revert even less frequently than those due to Tn5 or Tn10 insertion, whether selected for relief of polarity or for restoration of function of the mutated gene. This probably reflects the absence of long inverted repeats at the Mu termini and also the large size of Mu DNA. In fact, Mu insertion mutations are only seen to revert under conditions of partial induction of special Mu prophages that are themselves mutant in genes needed for full lytic development, and such reversion requires the MuA transposase protein (50). Mu excision may also reflect errors in DNA replication, in which slippage of the nascent DNA strand is triggered by MuA protein binding to Mu ends (206).

Insertions of Mu are also extremely stable if the Mu prophage is completely repressed. However, even partial induction results in cycles of Mu transposition to new sites and concomitant chromosomal rearrangement, reflecting the replicative nature of Mu transposition (Fig. 3). In the case of bacterial strains lysogenic for the widely used Mucts62 phage, which encodes a temperature-sensitive repressor, such partial induction is evident during brief periods at 37°C. Transposition to new sites is not observed, however, with engineered elements that lack a transposase gene (19, 113).

TRANSPOSON DELIVERY SYSTEMS

The strategy used to select for transposition depends on the target strain; on whether the target is the bacterial chromosome, a phage, or a plasmid; and, in some cases, on whether the element transposes conservatively or replicatively. Usually, a suicide vector that cannot survive is used (Table 3). In the last few years transposon-vector systems have been engineered that contain the transposase gene outside of the transposon ends (Table 2), so that the transposition product is stable. For phage and plasmid targets, a suicide vector is not required since transfer of the target replicon to a new host is sufficient to select for transposition. Strategies for transposon delivery include the following.

(i) For chromosomal targets, the transposon is usually delivered on a phage (usually λ for E. coli K-12 and P22 for S. typhimurium LT2) or plasmid suicide vector. Mu is itself a phage, and its DNA can be packaged into infectious virus particles, or Mu can be transferred on a plasmid. Bacteria containing chromosomal insertions are selected simply by their ability to grow in the presence of the antibiotic under conditions in which the donor vector cannot persist. Tn5, Tn10, Mu, and their derivatives are most often used for mutagenesis of the E. coli and S. typhimurium chromosomes. In contrast, Tn3, γδ, and related elements seem unable to insert efficiently into chromosomal DNA.

(ii) For phage targets, the transposon moves from a chromosomal or plasmid vector. Transposon-containing phage are recovered by packaging and infection of a new host and are selected after lysogenization or during lytic growth, depending on the marker used.

(iii) For plasmid targets, the transposon can be delivered from various heterologous vectors, as long as the transposition products can be selected. Common strategies include the following: (a) use of a suicide phage or heterologous suicide plasmid vector as the transposon donor and selection for the transposon marker (e.g., Fig. 1A); (b) extraction of plasmid DNA from pools of cells after transposition and transformation or electroporation, selecting for the transposon marker (e.g., Fig. 1A); (c) selection for hyperresistance after transposition from the chromosome to a high-copy-number plasmid; and (d) selection for mobilization of a normally nontransmissible target plasmid by a conjugative transposon-containing donor (e.g., the F factor) to form a cointegrate (e.g., Fig. 1B).

Phage used as suicide vectors usually have defects in integration or replication (26, 31, 159, 161). For example, transposition of Tn5 (kan) into a plasmid or the E. coli chromosome is readily obtained by selecting for Kan^r colonies after infection with a λ phage that carries Tn5 and lacks the phage attachment site (attP) (Fig. 1A). Lytic (and potentially lethal) phage replication is blocked by nonsense mutations in phage replication genes and the use of nonsuppressing host strains. Alternatively, replication is blocked by λ repressor synthesized by the infecting phage or by a resident prophage. Most transductants recovered under these conditions contain simple insertions of Tn5 in the genome and lack λ sequences. Equivalent strategies are used with transposon-containing derivatives of temperate phage P22 in S. typhimurium. In addition, λ suicide vectors can be used in S. typhimurium strains carrying a cloned constitutive allele of the E. coli λ receptor gene lamB (88).

A variety of suicide strategies have been useful with plasmid vectors (Table 3), including the following: (i) use of a mutant plasmid unable to replicate at high temperature; (ii) use of a mutant plasmid that lacks an essential trans-acting replication gene and an engineered host strain carrying this gene in the chromosome (genes and sites for DNA transfer during conjugation are generally included in such vectors, enabling their high-efficiency transfer to target cells in simple bacterial matings); (iii) use of a donor plasmid with a narrow host range in terms of replication but usually a broader host range in terms of conjugation and DNA transfer; (iv) use of a donor plasmid that replicates in gram-positive but not in gram-negative bacteria, and introduction into E. coli by transformation or electroporation; (v) use of a second incompatible plasmid to block replication of the transposon donor; and (vi) inclusion in the vector plasmid of a gene whose product is lethal to the cell under particular selective conditions.

Counterselectable genes have been used in suicide vectors and for a number of other purposes: to select for intramolecular transposition (discussed below), to select for the introduction of mutant alleles from cloned DNA to the bacterial chromosome (called marker exchange, allelic replacement, or gene targeting) (24, 41, 220, 226, , 239a, 240), to recover IS elements from the genome by "entrapment" (86, 106, 228, 247, 260), to cripple plasmids in genetically engineered microorganisms that are released into the environment (230), and for in vitro cloning (77, 133). A number of counterselectable genes are described in Table 3 and in chapter 139.

SPECIAL USES OF TRANSPOSONS

Once the utility of naturally occurring transposons for generating simple, polar, easily selected, loss-of-function mutations was established and recombinant DNA methods became available, many variants of these constructs were engineered to facilitate and extend molecular genetic analysis in laboratory, clinical, and environmental strains of E. coli, S. typhimurium, and many other bacterial species. Some of these elements and their uses are described here and in Table 2.

As Mobile Reporter Elements

Operon Fusions (Type I).

The first of the special engineered transposons, Mud1 (60), was the prototype of a diverse collection of elements that provide sensitive mobile reporters of transcription and, in later constructs, of translation and protein localization (Table 2). The use of these elements has revolutionized the study of genetic regulatory mechanisms and protein-membrane interactions and led to the discovery of many previously unknown genes. Mud1 is a derivative of phage Mu with an inward-facing promoterless lacZ gene near one end. Insertion of Mud1 in an operon in the correct orientation results in lacZ transcription from the operon promoter and hence β-galactosidase synthesis (this is a "type I" or "operon fusion") (60). The transcriptional regulation of the target genes is monitored by colony color on indicator medium or by a sensitive and easy enzyme assay. In an early implementation of this approach, a number of different genes whose transcription is induced in response to DNA damage were identified by screening a few thousand colonies containing new Mud1 insertions on medium under normal and damage-inducing conditions (96, 153). Some of these genes were previously known (e.g., uvrA, uvrB, and uvrC), but others were new. This approach was extended to identify genes that are turned on by shifting to medium of high osmolarity, shifting cultures to high temperature, etc. It also proved valuable in identifying new genes, identifying new functions for previously known genes, and studying the expression of genes whose products are difficult to assay (258, 265).

Reporter insertions can also be used to select mutations that alter expression of the reporter gene, thereby facilitating studies of gene regulation. The power of the promoter-probe approach has been extended by the construction of many derivatives of Mu, Tn3, Tn5, Tn10, and Tn917 (for gram-positive bacteria) with a variety of reporter genes. These include lac; uidA (gusA) (β-glucuronidase), a useful reporter gene in plant and animal cells (104); lux (luciferase), which is more sensitive than lac; cam (chloramphenicol transacetylase); and kan (aminoglycoside phosphotransferase) (see Table 2). cam and kan encode antibiotic resistances and allow direct selection, rather than just visual screening, for expression.

Protein Fusions (Type II).

Transposons with reporter genes that are missing the start sites for translation, as well as transcription initiation, are valuable for many purposes. To obtain mutants that express a hybrid fusion protein, the transposon must be in the correct orientation and reading frame (one-sixth of the insertions in a structural gene). The hybrid protein must also be in the proper milieu for expression: β-galactosidase (lacZ) is a cytoplasmic protein, while alkaline phosphatase (phoA) is active only if it has traversed the cytoplasmic membrane. Consequently, elements with one of these reporter genes can be used to identify the cellular compartment to which the target protein is localized. For transmembrane proteins, specific (surface exposed versus cytoplasmic) domains can also be identified (265).

A truncated phoA gene, lacking the signal sequence segment, as well as the promoter and the translational start site, was first used as a reporter for protein localization in a plasmid cloning vector (139). Soon thereafter, a comparable Tn5 reporter transposon, called TnphoA, was developed. TnphoA insertions result in PhoA⁺ phenotypes only in cases of in-frame insertion into genes for secreted proteins or in gene segments encoding surface-exposed domains (integral membrane protein genes) (190, 191). As with β-galactosidase, alkaline phosphatase can be scored by colony color on medium with a chromogenic substrate, as well as by a sensitive and easy direct enzyme assay.

In addition to its original use for probing protein localization and conformation, TnphoA has been of use in identifying genes in pathogens that are important for interaction with the host, since many proteins involved in colonization or virulence are surface exposed or secreted (150). For example, in one study, one-tenth of S. typhimurium colonies selected for a Pho⁺ phenotype after random TnphoA insertion were avirulent in mice (197a). This strategy is thought to sample only a subset of these genes, however, in part because it selects for genes that are strongly expressed on bacteriological medium (see references 176a and 176b).

Construction of MudphoA has permitted the phoA fusion strategy to be extended to species such as Legionella pneumophila, in which Tn5 transposition is rare and Mu transposition is quite efficient (5).

The complementary properties of β-galactosidase (active only in the cytoplasm) and alkaline phosphatase (active only if transported) were exploited to develop a strategy for examining membrane protein topology based on the activities of these two enzymes in sibling strains carrying TnphoA and Tn5lacZ (gene fusion) elements at exactly the same site in a target gene. This was accomplished by constructing pairs of TnphoA and Tn5lacZ elements with different antibiotic resistance determinants and homology at their ends, to allow replacement of one by the other by homologous recombination (189, 306).

Transposon tagging to analyze protein location has been extended with a Tn5 derivative containing a 5'-truncated bla (β-lactamase) gene instead of phoA. This element is useful in allowing in-frame insertions in genes for surface-exposed and secreted proteins to be selected directly by ampicillin resistance, rather than by screening for colony color (284).

As Mobile Promoter Elements

Transposons can be engineered to provide mobile promoters to turn on and also to turn off expression of adjacent genes. A Tn5 derivative called Tn5tac1 was constructed to allow isolation of insertion mutations with conditional mutant phenotypes. This element contains an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible outward-facing promoter (which is regulated by the lacI repressor gene), as well as the Tn5 transposase and kanamycin resistance genes (66). Tn5tac1 can be delivered to the E. coli genome from a λ suicide vector (66) and to nonenteric bacteria from a plasmid suicide vector (71). Tests with representative Tn5tac1 insertions showed that, in addition to standard insertion mutations, two kinds of conditional mutations were found: mutations whose effects are repaired by the addition of IPTG and are due to polarity in the absence of transcription from the element and mutations whose effects are only observed in the presence of IPTG and are due to IPTG-induced transcription. Transcription-induced mutant phenotypes were found for insertions upstream of the target gene, due to excess transcription (66), and for insertions downstream of the target gene, due to interference with normal transcription (209, 210).

A mini-Tn5 derivative, analogous to Tn5tac1, containing an outward-facing lacI-regulated promoter (Ptrc) has been constructed recently (83). This transposon is in the pUT broad-host-range conjugal suicide vector (135) that is useful for mutagenizing diverse enteric and nonenteric gram-negative bacteria.

For Multiplex Screening

It is appropriate to reemphasize here that many strains of E. coli and S. typhimurium are important pathogens and that transposon-based searches for potential pathogenicity determinants have been very useful. However, an elegant strategy for screening many insertion mutants simultaneously was recently developed to overcome the cost and impracticability of testing hundreds or thousands of insertion strains in animals, one at a time (134b). This method, termed "signature-tagged mutagenesis," entails mutagenesis of the pathogen en masse with a set of mini-Tn5 elements that contain thousands of different random oligonucleotide tags, arraying individual mutant colonies, and screening for tags that are missing after passage of the pathogen through an animal host. These missing tags represent mutants that could not proliferate in vivo. This strategy has been implemented with S. typhimurium, and a number of genes needed for infection of mice were identified (134b). This transposon-based multiplex tag approach should also be valuable for many other screening, mapping, cloning, and sequencing purposes. A complementary multiplex insertion and PCR process to identify genes required under specific growth conditions has been described for yeast (266a). It could readily be applied to bacteria.

For Genome Mapping

Transposable elements greatly facilitate genetic and physical chromosome mapping at several levels. For genetic mapping, the resistance transposons provide selectable markers at known sites that facilitate conjugational and transduction mapping of nearby genes (17, 19, 161). This is particularly well illustrated by a set of isogenic E. coli strains containing both Tet^r and Kan^r insertions every approximately 1 min along the chromosome at positions that are known with respect to both the E. coli genetic map and the Kohara library of ordered λ phage clones (263; see appendix to chapter 109). A rapid two-step mapping procedure was implemented with these strains. In the first step, any new mutation is located to a 5- to 15-min region of the chromosome by Hfr mapping with a set of Hfr strains carrying either Tn10 (tet) or Tn10 (kan) about 20 min from their respective origins of transfer (see chapters 127 and 137). In the second step, the mutation is localized to a 1-min region by P1 transduction, using a subset of the insertion mutant strains as donors. This strain collection is available from the E. coli Genetic Stock Center. A less complete, but still very useful, set of insertions is available from the Salmonella Stock Center (see chapter 141).

Large-scale genetic mapping has also been aided by the construction of Tn5 and mini-Mu derivatives containing plasmid origins of conjugational transfer (oriT). Insertion of these oriT transposons into the chromosome allows them to serve as origins of Hfr-like and F'-like transfer and thus allows efficient mapping of these chromosomal insertions and other markers by conjugation (229, 259, 311). Homologous recombination between a transposon on a fertility plasmid and a homologous transposon in the chromosome has also been used for conjugational mapping (67).

Physical mapping of genomic DNA by pulsed-field gel electrophoresis (PFGE) (see chapter 138) has also been facilitated by using transposon insertions. Standard PFGE analysis involves digesting genomic DNA with enzymes, such as NotI, that cut the DNA infrequently. The relatively few large fragments are separated by PFGE, and hybridization to probes for specific mapped genes is used to align the restriction fragments. Rare restriction sites in wild-type transposons Tn5 and Tn10 (180, 193, 238a, 265a, 308; Mahillon et al., personal communication), as well as additional sites engineered into mini-Tn5 and mini-Tn10 derivatives (128, 309), have been exploited by PFGE. A comparison of the PFGE restriction fragment profiles of strains containing insertions at known locations with the profile from the nonmutant parent or from a related strain reveals which parental fragment has been replaced by two fragments and also the approximate position of the insertion relative to one fragment end or the other.

This use of transposons containing rare restriction sites has been extended to the study of comparative genome content and organization of different isolates of the same species. This has been implemented with E. coli by moving mapped E. coli K-12 insertions to the same sites in clinical E. coli isolates by P1 transduction and selecting for the transposon marker. Comparison of the PFGE restriction fragment profile of the clinical isolate with that of E. coli K-12 helped locate and define differences in the physical maps; some of the genes unique to one or the other strain are likely to be involved in host colonization or disease (40a; Mahillon et al., personal communication).

Insertions can be used to anchor PCR for generating hybridization probes. For PCR, a second primer binding site within a few kilobases of the insertion has to be identified or created. This site can be created by using a PCR strategy, such as (i) inverse PCR (53, 213, 235); (ii) PCR between a transposon and naturally occurring "REP" (repetitive) sequences (279); or (iii) vectorette or "cassette ligation" PCR, in which a double-stranded oligonucleotide containing a unique primer binding site is ligated to restriction-digested DNA (142, 237). These PCR fragments can be used to probe Southern blots of ordered clone libraries, allowing rapid physical mapping.

With the recent development of long-distance PCR protocols that amplify cloned DNA fragments of up to 50 kb from phage λ templates (10, 64), it may soon be possible to map chromosomal insertions directly if they are within 50 kb of known markers.

For Mapping Cloned Fragments

Transposons have been exploited to localize genes in large cloned fragments and are particularly valuable when insertion in the gene can be detected by phenotype and then mapped by restriction analysis (17, 19, 79, 119), although insertions outside genes sometimes exert "position effects" (179).

A method for generating nested deletions in one direction in DNAs cloned in very large capacity (70 to 100 kb) P1 phage vectors (221, 222, 272) has been developed. This is based on the presence in P1 clones of the P1 loxP site for efficient site-specific recombination in the presence of Cre protein. Transposition of a loxP-containing mini-Tn10 element into Cre⁺ cells that contain a P1 clone and recombination between the two loxP sites generate sets of nested deletions in which DNA between the insertion and vector loxP sites is deleted (273). This system can be used to isolate deletion derivatives that contain the vector replication origin but not the reciprocal product lacking a replication origin. Nonetheless, this nested deletion strategy is particularly attractive for dissecting P1 clones, in which the DNA is too large for easy mapping of simple insertions.

Transposons can also be used to generate detailed restriction maps of the cloned fragment, in much the same way that they are used in generating PFGE (above) or "dropout" maps (described below).

For Sequencing Cloned DNA

Transposons provide mobile and easily mapped binding sites for universal PCR and DNA sequencing primers. They can give access to all regions of large cloned target DNAs without the need for (i) shotgun subcloning and attendant high-redundancy sequencing or (ii) extensive primer walking, which is slow and costly at present. In addition, transposons can be used to sequence DNA that contains repeated sequences. Both simple insertions, which allow sequences to be read from each transposon end (reading one strand in each direction), and transposon-mediated nested sets of deletions can be used to great advantage. The most effective transposon-based strategy for a given DNA segment depends on the size of the DNA segment and the vector in which it is cloned.

For plasmid clones, which are generally less than 10 kb, any element that inserts into many target sites and contains unique DNA sequences close to each end can be used (1, 179, 208). In cases of elements such as wild-type Tn5 or Tn10 with long inverted repeats, effective sequencing depends on separating the left and right inverted repeat, e.g., by digestion with a restriction endonuclease that cuts within the transposon and electrophoresis or subcloning (90). The best-developed method for transposon-based plasmid sequencing, however, exploits γδ and its relatives, in large part because of the many sites per gene in which these elements can insert and the simple conjugation method for obtaining insertions (Fig. 1B) (19, 76, 119, 299).

In vitro transposition of synthetic transposons is also very promising, as illustrated by AT-2, an "artificial transposon" based on the yeast Ty1 transposon (a retrovirus-like element). The addition of AT-2 DNA and integrase-containing "virus-like particles" specific for Ty1 to a target plasmid and then transformation of E. coli and selection for the transposon marker yield insertions throughout the cloned target DNA (87).

In all of these cases, insertions to be used for sequencing in each direction from any site of insertion can be chosen on the basis of map position, which can be determined easily by restriction endonuclease or probe mapping (277) or by PCR with primers specific for transposon ends and appropriate vector sequences (219). These approaches can be automated for large-scale sequencing (20, 192; R. Weiss and R. Gesteland, personal communication).

For DNAs cloned in most λ phage vectors, it is not possible to select for transposon insertions by using resistance markers because these phage cannot lysogenize E. coli and resistance to an antibacterial agent cannot be selected during lytic growth. This problem is overcome with Tn5supF, a 300-bp derivative of Tn5 marked with a suppressor tRNA gene (149, 170, 219). Tn5supF insertion into λ can be selected by plaque formation on one of two hosts: a standard nonsuppressing E. coli strain for vectors with amber mutant alleles of essential phage genes or a dnaB(Am) E. coli strain (the dnaB product is essential for λ growth) for vectors lacking amber mutations, such as those used for the Kohara E. coli library (163).

For large cosmid clones the effort needed to map many individual insertions makes a simple insertion strategy less appealing. In contrast, with nested deletions made in such clones by transposition, the position of the transposon end, and thus the region accessed, is indicated simply by the size of the plasmid remaining after the transposition event. The transposon-based cloning vectors described in the following section (2, 3, 20, 217, 296, 297) were developed for DNA sequencing and mapping.

For Cloning

In Vitro Cloning of Insertions.

Transposons can facilitate in vitro gene cloning in several ways.

(i) They provide resistance markers and an array of known restriction sites that can be inserted into or next to a gene and used to clone that gene (detailed restriction maps of useful natural and genetically engineered transposons can be obtained from public domain databases or from references cited in Tables 1 and 2).

ii) Insertions in a gene can be used to clone DNA to one or both sides of the insertion, depending on the restriction enzyme(s) used. In clones that contain DNA from both sides of the insertion, the wild-type allele can be recovered in the plasmid clone by homologous recombination with the chromosomal allele. This is facilitated by using a transposon, such as Tn5-rpsL, that carries both counterselectable and selectable markers (276). In addition, for insertions of elements in genes with selectable phenotypes, the wild-type allele can be obtained directly from the insertion mutant plasmid by selecting for function, if the element can undergo precise excision. Tn916 is especially useful for recovering revertants because it undergoes a high frequency of precise excision in E. coli (57, 105).

iii) Transposons that have been engineered to contain a plasmid replication origin can be used directly for in vitro cloning of segments bracketing the insertion. However, because insertion of a functional plasmid replication origin into the chromosome is often deleterious, this strategy is most effective when the origin does not function in the strain to be mutagenized. For example, Tn5-V, which contains the plasmid pSC101 origin of replication, can be used to mutagenize Myxococcus xanthus, in which pSC101 does not replicate. Insertions were cloned directly by restriction digestion of genomic DNA, self-ligation, and transformation of E. coli (102). Another derivative, Tn5-PV, has the origins of replication and of transfer but not the replication genes from plasmid RK2 (165). The replication origin in Tn5-PV insertions in RK2-free strains is nonfunctional; hence, this transposon can insert into the chromosome and be used to clone adjacent segments when the self-ligated genomic DNA mixture is transformed into a strain carrying RK2 replication genes.

Generally, the restriction endonuclease(s) used for cloning insertion mutations is one that does not cleave the transposon, but if DNA to only one side of the insertion is sought, an enzyme that cuts the transposon could be used.

Vectors for In Vivo Cloning.

As noted above, an early, and important, in vivo cloning method involved λ specialized transducing phages from aberrant excision of prophages inserted at dispersed secondary sites (244, 255). Other phage-based in vivo cloning strategies are discussed below.

(i) One strategy involves λ-Mu hybrid phage (λplacMu) that use Mu DNA homology to direct their integration by homologous recombination into Mu prophage (Mu inserts into more sites than λ), followed by aberrant excision and packaging into λ heads (58, 166, 186). More recently, this approach was greatly simplified to eliminate the homologous recombination step: other λ-Mu hybrid phage have been constructed that insert with the randomness of Mu, but replicate and package DNA as λ phage (47, 49, 290).

(ii) Another strategy involves Mu-P22 hybrid phage that insert into the S. typhimurium chromosome with the randomness of Mu. Upon induction, they replicate from the P22 origin in one direction and package sequential headfuls of DNA, starting from the P22 pac site. Therefore, chromosomal segments corresponding to a few percent of the genome to one side of the inserted phage are recoverable in high yield from any single lysogen (312).

(iii) The most generally useful approach to in vivo cloning involves small engineered derivatives of Mu (114, 116) and the Mu-related Pseudomonas aeruginosa phage D3112 (75). These phage replicate by transposing to different sites and they package DNA by the headful mechanism. Consequently, small derivatives of these phage also package a segment of adjacent host DNA, which sometimes contains part or all of a second phage copy (because of intramolecular replicative transposition). Such transducing phage give rise to clones containing the interstitial DNA by transduction of a Rec⁺ recipient and homologous recombination between the Mu ends. DNAs that do not contain a second Mu end cannot circularize.

Mini-Mu and D3112 derivatives contain phage ends, essential phage genes, sites that are essential for plasmid replication, and a selectable resistance determinant, but not a full set of morphogenetic genes, and thus require a helper phage to form phage particles. Some are small enough to package up to 34 kb of adjacent DNA into Mu or D3112. Many derivatives carrying various antibiotic resistance determinants and plasmid replication origins are available. Some of these engineered mini-Mu and mini-D3112 elements contain a promoterless lac reporter gene, a plasmid origin of transfer, a tac promoter, a λ cos site, and/or a phage T7 promoter (see Table 2 and reference 113).

In vivo mini-Mu cloning has some advantages over traditional in vitro methods for those species in which it can be used: the cloned piece is always colinear with the parental chromosome (there is no risk of obtaining chimeric clones), and the fragment endpoints are determined by Mu insertion, which is more random than cleavage by any one restriction enzyme (even a 4-bp cutter such as Sau3A), so that clones with a wide array of fragment endpoints can be obtained (299).

(iv) A recently described transposon, Tn5-rpsL oriR6K (Table 2), contains the crippled R6K origin and has been used to clone by P1 generalized transduction and recircularization by illegitimate recombination in a RecA cell expressing the R6K pir (replication) gene (275).

Transposon-Based Vectors for In Vitro Cloning.

Several cloning vectors have been constructed with transposons or transposon ends, which are used to generate nested deletions in the cloned fragment by intramolecular transposition. The deletion endpoints can be used to map and sequence the cloned DNA. The first implementation of a transposon-based strategy for making nested deletions in cloned DNA, which was also the first demonstration of transposon-based sequencing, involved the use of a vector containing Tn9 and galKT ⁺ counterselectable genes between the transposon and the cloning site (galE strains are galactose sensitive [Table 3]). Most Gal^r derivatives of strains carrying this plasmid arise by intramolecular transposition (either replicative or conservative) into or beyond galKT, producing deletions that extend into target DNA from a transposon end (2, 3, 217). The Tn9-based vectors have two disadvantages: Tn9 transposition is very nonrandom and deletions can be isolated in only one direction, so that two clones with oppositely oriented fragments are needed to access both strands of the cloned DNA.

The disadvantages of Tn9-based deletion vectors were addressed in a family of γδ-based "deletion factory" vectors called pDUAL or pDELTA (Fig. 4): (i) in these vectors, γδ intramolecular transposition is quite random, so nested deletions with near random distributions of endpoints can be obtained; and (ii) the vector was engineered so that deletions extending in either direction into the cloned fragment can be selected. This ability to isolate nested deletions in each direction, and thereby access each strand without recloning, is based on two features: (i) incorporating a plasmid replication origin within the synthetic γδ element, so that each of the two complementary circular DNAs generated by replicative transposition (deletion mode) (Fig. 3 [left]) is replication proficient; and (ii) placing different counterselectable genes (sacB and strA; sensitivity to sucrose and streptomycin, respectively [Table 3]) on each side of γδ. In addition, one selectable gene is placed on each side of the cloning site to increase the efficiency of selecting deletions that extend from either γδ end to sites within, rather than beyond, the cloned DNA segment (20, 296, 297). With these features, sets of nested deletions can be selected in either direction by using sucrose-tetracycline medium for clockwise deletions or streptomycin-kanamycin medium for counterclockwise deletions (Fig. 5). The selection against one counterselectable marker and for one selectable marker yields colonies with plasmids whose nested deletions extend in the selected direction for various distances into the cloned target DNA. After transposition, one transposon end always abuts a deletion endpoint, and the unique sequence just inside the transposon end can serve as a "universal" primer binding site for sequencing adjacent cloned DNA. All target regions can be sequenced by using a pair of universal primer binding sites engineered into γδ . Outward-facing SP6 and T7 promoters just within the γδ ends serve as sites for universal primers and also for riboprobe synthesis after γδ-mediated deletion formation (Fig. 4). pDUAL-1 (297) and pDELTA-1 (296) contain the pUC19 high-copy-number replication origin and amp gene but different unique restriction sites. pDUAL-3 contains an amplifiable low-copy-number replication origin to allow greater stability of segments not clonable in high-copy-number vectors (C. M. Berg, X. Liu, G. Wang, X. Xu, and D. E. Berg, unpublished data).

Fig. 4Generic pDUAL or pDELTA cosmid cloning vector for generating nested deletions in vivo. These vectors contain a synthetic ?? transposon containing 40 bp of inverted repeat copies of the ? end of ?? bracketing SP6 and T7 promoter/primer binding sites, rare restriction sites (R), a replication origin, and, in some cases, an amp gene. The counterselectable genes, sacB and strA (rpsL), immediately outside the transposon ends are used to select for deletioins that extend in either direction from a transposon end. The selectable genes kan and tet are between the counterselectable genes and the cloning site and are used to select against deletions that extent through the cloned fragment into vector sequences. pDUAL-1 (297) and its commercialized relative, pDELTA-1 (296), contain the replication origin and amp from pUC19 between the ?? ends but differ in their restriction sites. pDUAL-3 (C. M. Berg, unpublished data) contains an emplifiable R1 replication origin and no amp gene.

Source:

Fig. 5Expected restriction fragment profiles of pDUAL clone deletion derivatives. The four restriction sites (vertical lines 1 to 4) divide the clone into four fragments (A to D). Restriction sites 1 to 3 are in the cloned fragment; site 4 is at the cloning site. One fragment (A) contains both vector and cloned DNA; the other fragments. A? indicates a truncated fragment;underline indicates fusion fragments. (A) Linearized parent plasmid. (B and C) Deletion derivatives.Every ??- generated deletion reduces the size of the vector component (A), and most deletions generate fusion fragments (letters underlined on the left) between the remainder of the A fragment and one of the restriction fragments in the cloned portion. Note that the transposition target site (deletion endpoint) can be determined approximately by the size of the deletion plasmid and more accurately by the restriction fragment pattern (dropout mapping; 174a).

Source:

  Uses of nested deletions to generate restriction maps and for sample sequencing. Restriction maps of cloned fragments in vectors such as pDUAL can be generated with relative ease by aligning restriction data from single-enzyme digestions of the parent clone and of a small set of nested deletion derivatives, as depicted in Fig. 5. Consideration of the restriction fragments from the parent clone in comparison with those that drop out in the deletion derivatives permits the fragments and restriction sites to be ordered (174a). These deletion derivatives can also be used to obtain "sample sequence" information at known sites in the clone, so that regions of interest can be readily identified (169).

FUTURE PROSPECTS

Transposable elements continue to be superb as tools for mutational analysis and in vivo genetic engineering in laboratory strains of E. coli and S. typhimurium, in a diverse array of other prokaryotes, and in eukaryotes (18, 19, 32). Improved transposons and delivery systems continue to be developed for an expanding array of organisms. The next few years should bring even greater use of transposons: in direct genome analysis, especially in conjunction with long-distance PCR (10, 64); in mapping and sequencing >80-kb fragments cloned in large-capacity bacterial vectors (P1, BAC, and PAC) (141, 257, 272); in robust in vitro transposition systems (e.g., reference 87); and in primer walking with preformed oligonucleotide libraries (278, 283).

ACKNOWLEDGMENTS

We gratefully acknowledge the contributions of former and present colleagues to the work and ideas presented here. We are particularly indebted to Robert Blakesley, David Chaplin, Eduardo Groisman, Henry Huang, Katsumi Isono, Liz Jean, Dangeruta Kersulyte, Raja Krishnan, Frank LaBanca, Lin Liu, Xiayang Liu, Brooks Low, Chihiro Sasakawa, David Sherratt, Linda Strausbaugh, Gan Wang, and Xiaoxin Xu. Work in our laboratories was supported by grants from the Department of Energy (DE-FG02–89ER-60862 and DE-FG02–90ER-610), National Institutes of Health (HG-00820, DK48029, and AI38166), American Cancer Society (VM121), and University of Connecticut Research Foundation.