DNA recognition sites activate MuA transposase to perform transposition of non-Mu DNA.

Mu transposition occurs within a large protein-DNA complex called a transpososome. This stable complex includes four subunits of MuA transposase, each contacting a 22-base pair recognition site located near an end of the transposon DNA. These MuA recognition sites are critical for assembling the transpososome. Here we report that when concentrations of Mu DNA are limited, the MuA recognition sites permit assembly of transpososomes in which non-Mu DNA substitutes for some of the Mu sequences. These "hybrid" transpososomes are stable to competitor DNA, actively transpose the non-Mu DNA, and produce transposition products that had been previously observed but not explained. The strongest activator of non-Mu transposition is a DNA fragment containing two MuA recognition sites and no cleavage site, but a shorter fragment with just one recognition site is sufficient. Based on our results, we propose that MuA recognition sites drive assembly of functional transpososomes in two complementary ways. Multiple recognition sites help physically position MuA subunits in the transpososome plus each individual site allosterically activates transposase.

Transposons are found in all the biological kingdoms, and some perform specialized functions. For example, the machinery that initiates V(D)J recombination likely evolved from a transposon (1,2), and the cDNA of HIV and other retroviruses integrate into host cell DNA through mechanisms nearly identical to transposition (3). The genome of bacteriophage Mu is a transposon that uses transposition both to integrate into the DNA of a new host cell and to replicate before lysis. Like most DNA rearrangements, transposition is a complex, multi-step process, requiring numerous DNA sequence elements. Studies of bacteriophage Mu have been central to our understanding of both the fundamental mechanisms and the complexities of DNA transposition.
Phage Mu encodes a transposase, MuA, that transfers the Mu genome from one DNA location (the transposition donor) to a new location (the transposition target) (4,5). During transposition, transposase performs two principle reactions: DNA cleavage and DNA strand transfer. During cleavage, the donor DNA is nicked twice, once at each 3Ј end of the Mu genome. During strand transfer, the cleaved transposon ends are in-serted into neighboring sites on the two target strands.
Little or no specific sequence information is needed on the target DNA (6), but the Mu DNA provides many sequence cues for transposition (Fig. 1). For example, the last two nucleotides at either 3Ј end of the Mu DNA, the cleavage sites, have the sequence 5Ј-CA. Also near each end of the Mu DNA are three recognition sites, distinct from the cleavage sites, which share a 22-base pair consensus sequence. The recognition sites are referred to as R1, R2, and R3 on the right end and L1, L2, and L3 on the left end ( Fig. 1) (7). The recognition sites are bound specifically by the N-terminal domain of MuA (8), whereas the cleavage sites must be engaged by the protein's active site, contained in a different region of the protein (9). Both the recognition sequences and the 5Ј-CA cleavage sequences are required for transposition (10 -13).
Both the cleavage and strand transfer reactions occur within a stable MuA⅐DNA complex called a transpososome (Fig. 1). Because the transpososome binds both transposon ends simultaneously, it is also referred to as a synaptic complex. The complex contains three MuA subunits tightly bound to recognition sites (the R1, R2, and L1 sites) plus a fourth subunit tightly bound in the complex but weakly bound to the L2 recognition site (14 -16). At least two of the four MuA subunits individually bridge the transposon ends: the subunit bound to the right-end's R1 site engages the left cleavage site, and the subunit bound to the left-end's L1 site engages the right cleavage site (17)(18)(19)(20). This crisscross structure helps coordinate reactions at the two DNA ends, ensuring that the transposon moves as a single unit. Given this intertwined structure, and also the intimate involvement of multiple DNA sites in the complex, it is unknown to what extent protein-protein versus protein-DNA interactions contribute to the stability and functionality of the synaptic complex.
In vitro, transpososomes can also assemble on ϳ50-base pair DNA fragments, containing the R2, R1, and cleavage site sequences. Two of these "donor fragments" are synapsed by a MuA tetramer, mimicking synapsis of the two ends of a transposon ( Fig. 2A) (21). The fragments are then cleaved at the proper cleavage site and transferred together to a target DNA. If the fragments are synthesized to appear precleaved, cleavage by MuA is unnecessary prior to strand transfer (Fig. 2B). Because the fragments are small relative to the target, the resulting transposition product comigrates with linear target. Transpososomes formed on donor fragments, like those formed on larger DNA molecules, can resist competition from additional recognition sites for hours or perhaps days. In contrast, monomeric MuA has rapid on and off rates from its recognition sites (21,22). 1 How does transposase change from a form that is monomeric, with a rapid dissociation rate from DNA, into a stable, active synaptic complex? This process involves interactions between at least four multi-domain subunits and multiple DNA sequences, and little is understood about the role of each component in the complex. For example, although the recognition sites are essential, their precise role during assembly is uncertain.
Here, we use a promiscuous activity of MuA to elucidate the functions of MuA recognition sites during transposition. We find that MuA recognition sites can activate MuA to transpose non-Mu DNA. Two covalently linked MuA recognition sites are the strongest activators of transposition, but unlinked sites can also serve in this capacity. Our results suggest that MuA recognition sites perform at least two complementary functions during transposition. Covalently linked sites promote transpososome assembly by spatially constraining two of the four MuA subunits. In addition, each individual recognition site activates transposition, probably by inducing conformational changes in the transposase.

EXPERIMENTAL PROCEDURES
Proteins-In some experiments (see Figs. 7, B, C, and 8) the MuA truncation 77-663 was substituted for full-length MuA. This was done because, in the absence of the Internal Activating Sequence (a DNA sequence on the Mu phage that activates transpososome assembly), the truncation protein is hyperactive compared with the full-length protein.
DNA-X174 RFI was purchased either from Life Technologies, Inc. or New England Biolabs. Oligonucleotides (fragment donors) were purchased either from Massachusetts Institute of Technology/Howard Hughes Medical Institute (MIT/HHMI) biopolymers laboratory or from GeneLink and were purified by denaturing PAGE, except for those less than 30 nucleotides, which were purified by reverse phase cartridges. Most oligonucleotide sequences are described in Fig. 2B. However, over the course of this study, we used several different "unjoinable fragments" that differed at three non-essential nucleotide positions and/or in the length of the 5Ј overhang on the non-transferred strand. We saw no qualitative difference in the behavior of these various fragments (data not shown). The unjoinable fragment used in Fig. 3A was: agtgaagcggcgcacgaaaaacgcgaaagcgtttcacgaaaaacgcgaaagcg/cgctttcgcgtttttcgtgaaacgctttcgcgtttttcgtgcgccgcttc. The fragment used in Fig. 3B was: gcatgaagcggcgcacgaaaaacgcgaaagcg tttcacgataaatgcgaaaac/gttttcgcatttatcgtgaaacgctttcgcgtttttcgtgcgccgcttc. All other figures used the unjoinable fragment listed in Fig. 2B. The full sequence of the cleaved strand of the uncleaved fragment is: cgttttcgcatttatcgtgaaacgctttcgcgtttttcgtgcgccgcttcactagacgcttggcgtaatcgggcgtaatgc. The experiment in Figs. 4C and 6B (squares) used the following DNA fragment to maintain the total fragment DNA at 1440 nM: gccggtatctttccagcactgggccggtatctttccagcactggcg/cgccagtgctggaaagataccggcccagtgctggaaagataccggc. The experiment in Fig. 6B (triangles) contained the following fragment to maintain total fragment DNA at 1440 nM: gccggtatctttccagcactgg/ ccagtgctggaaagataccggc.
Transposition Reactions-Unless indicated otherwise, reactions were conducted in a 25-l volume containing 25 mM Tris-HCl (pH 8 at room temperature), 140 mM NaCl, 10 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mg/ml bovine serum albumin, 15% glycerol, 12% Me 2 SO, 0.1% Triton, 2 mM ATP, 340 nM MuB, 250 ng of X174 DNA, and variable amounts of Mu DNA fragments and MuA. Proteins were prepared by dilution of concentrated stock into the following buffers. MuA: 600 mM NaCl, 25 mM Hepes-KOH, 0.1 mM EDTA, 10% glycerol, and 1 mM dithiothreitol; MuB: 1 M NaCl, 25 mM Hepes, 0.1 mM EDTA, 20% glycerol, and 1 mM dithiothreitol. The reactions were incubated at 30°C for 20 -60 min unless otherwise indicated. They were stopped by addition of 0.2 volumes of a stop solution (ϳ0.1% bromphenol blue, 8% SDS, 50 mM EDTA, 30% glycerol) and electrophoresed through a 0.9% HGTagarose gel, in 1ϫ TAB buffer (40 mM Tris, pH 8, 3.6 mM EDTA, 27 mM sodium acetate). The gels were then analyzed by one of the following methods: 1) stained in 1 g/ml ethidium bromide and photographed with a Polaroid camera, 2) stained in a 1:10,000 dilution Vistra Green in TAB buffer and visualized on a Molecular Dynamics Fluorimager 595, and 3) for radioactive samples, the gels were pressed and dried on DEAE paper and exposed to a Molecular Dynamics phosphorimaging cassette. The plasmid assays, described in Fig. 8, were performed in two steps. The first incubation was as described (23) (reactions contained neither Me 2 SO nor Triton), except that the reaction volume was 50 l and incubation time was 2 h. For the second step, either 1 l of restriction enzyme or 1 l of 50% glycerol was added to reactions, and they were incubated for another 2 h at 37°C. The second incubation period increased the amount of intramolecular X174 products.
Two-dimensional Gels-In the first dimension, electrophoreses was as described above. A single lane was then excised and positioned horizontally across the top of a 0.8% HGT-agarose gel in alkaline buffer (30 mM NaOH, 1 mM EDTA) for electrophoreses in the second dimension. The gel was analyzed by Southern blotting: DNA was transferred to a PerkinElmer Life Sciences Genescreen hybridization membrane and probed with labeled DNA fragments that had been randomly primed off of X174 RFI DNA.

RESULTS
A Non-Mu DNA Molecule Serves As a Transposition Donor-In transposition reactions containing limiting concentrations of Mu-derived DNA, non-Mu DNA molecules were seen to transpose (Fig. 3). There were three principle requirements for this promiscuous activity. The first requirement is MuA protein. The second requirement is a large, non-Mu DNA molecule to serve as a transposition donor. We focused on the 5386-base pair X174 RFI, but two molecules unrelated to X174, pBR322 or pUC19, worked about equally well (data not shown). The third requirement is MuA recognition sites supplied in trans on another DNA molecule.
In our initial experiments, the MuA recognition sites were supplied on a fragment that is itself defective in the strand transfer step (called "unjoinable," see Fig. 2B). This unjoinable fragment has two MuA recognition sites (R1 and R2) and an incomplete cleavage site. Thus the unjoinable fragment and the non-Mu DNA complemented each other. The fragment supported transpososome assembly but failed to transpose due to its defective cleavage site (see below, and Ref. 17). The non-Mu DNA cannot on its own support transpososome assembly (11), but it did transpose.
Incubation of the unjoinable Mu fragment, MuA, the MuB protein (see below), and a target DNA (X174 RFI) resulted in the formation of two types of transposition products. One family of products electrophoresed between the nicked and linear target DNAs (Fig. 3A, lanes 1 and 2, labeled intramolecular). The second product ran slightly above dimeric target (Fig. 3A, lane 1, labeled intermolecular). Both were true recombination products rather than non-covalent rearrangements trapped by proteins; the samples in Fig. 3A were treated with 1% SDS, and the products were also unaffected by treatment with proteinase K or by phenol extraction (data not shown). The Mu fragment was a critical component of the reaction (lane 5).
Formation of the slower-migrating product (labeled intermolecular) required the MuB protein (Fig. 3A, lane 2). MuB is a DNA-binding protein that stimulates MuA and controls MuA's target selection (26). For example, when transposition of Mu donor plasmids is assayed in the absence of MuB, the products are usually intramolecular; the Mu sequences define the donor FIG. 2. Mu donor fragments. A, fragments containing 50 base pairs from the right end of Mu support transpososome assembly. Mimicking transposition of a complete transposon, MuA can synapse two fragments and join them to opposite strands of a target DNA. The DFT (double fragment transfer) product results from successful transposition of both fragments; the SFT (single fragment transfer) product results if one fragment fails to transpose. B, sequences of fragments used in this study. The fragments are shown with the transferred strand on top, but that strand is listed 3Ј to 5Ј. The important differences among most of these oligonucleotides is the length of the transferred (top) strand on the 3Ј end. The darker box highlights the sequence of the native R1 site, lighter box the native R2 site, and bold lettering the cleavage site. Some experiments used fragments with the same names but slightly different sequence than those shown here (see "Experimental Procedures"). Additional fragments are also listed under "Experimental Procedures," as is the complete sequence of the uncleaved fragment.

FIG. 3. Transposition of a non-Mu DNA molecule (X174 RFI).
A, ethidium bromide stain of an agarose gel showing non-Mu transposition products. Transposition of X174 RFI DNA was activated by an unjoinable (UJ) Mu fragment. On this gel, intramolecular transposition products migrated near linear X174. But with TBE running buffer, as opposed to the TAB buffer used in the gel shown here, the same products very clearly separated from linear DNA (data not shown). A reaction with a pre-cleaved (PC) fragment is shown for comparison. The DFT product of that reaction migrates with linear X174. Fragment and MuA concentrations were each 50 nM. B, two-dimensional gel electrophoreses supported the identification of the intra-and intermolecular transposition products. Transposition products were formed with an unjoinable fragment and X174 DNA. Electrophoreses in the first dimension was on a native agarose gel as in Fig. 2A; the second dimension was alkaline (denaturing) agarose, and was visualized by Southern blotting. Here, an ethidium bromide stain of the first dimension is shown positioned horizontally across the autoradiogram of the second dimension. The ethidium stain is not of the actual gel-slice that was used for the second dimension, but it is the same DNA sample. Circular and linear strands are labeled based on standards (specifically, yield from supercoiled and nicked DNA). Dimer is labeled based on its slow migration. sites, and another site on the same plasmid serves as target site. In the presence of MuB, transposition is usually intermolecular; the Mu plasmid serves as donor, a separate non-Mu DNA molecule serves as target, and the two molecules become joined to each other through transposition. Here, the slow electrophoretic mobility of the MuB-dependent product (Fig.  3A, labeled intermolecular) strongly suggested that it contained two X174 molecules. This product did not contain Mu DNA (data not shown, results of radiolabeling the Mu donor fragment). Therefore, this MuB-dependent product was most likely the result of an aberrant intermolecular reaction, in which two molecules of X174 were joined together. The MuBindependent product (labeled intramolecular) could have resulted from intramolecular recombination, in which a single X174 molecule provided both the donor and target sites.
Two-dimensional gel electrophoresis confirmed the structures of the new products (Fig. 3B). Transposition reactions were performed with MuA, MuB, X174, and a Mu fragment. The DNA products were resolved on an agarose gel similar to that shown in Fig. 3A. A lane was then excised from the gel (shown lying horizontally in Fig. 3B), and electrophoresed through alkaline (denaturing)-agarose. In the denaturing dimension, the intermolecular product yielded a slow-migrating species and a species that co-migrated with single-stranded circular X174 DNA. These are the expected components of a figure-eight-shaped transposition product, in which two fulllength, double-stranded X174 molecules are joined by a single strand. The MuB-independent products also yielded a slowmigrating component in the second dimension. This long single-stranded species could result from an intramolecular transposition in which a strand was joined to its complement. Intramolecular transposition products should also yield smaller components of varying mobilities, depending on the exact insertion site. Since these smaller products do not migrate as a distinct band, they are difficult to distinguish on the two-dimensional gel. Together, these data show that a Mu DNA fragment permits a non-Mu DNA molecule (in this case X174) to participate as a donor DNA during transposition.
Hybrid Transpososomes Synapse One Mu Fragment and a X174 Molecule-It is likely that Mu fragments promote non-Mu transposition, at least in part, by promoting assembly of synaptic complexes. In support of this argument, unjoinable Mu fragments are known to support assembly of MuA tetramers that are stable to high salt and to competitor DNA, a diagnostic of a synaptic complex (17,20). X174 DNA alone does not support assembly of detectable levels of MuA tetramer (11).
We considered two models for synaptic complexes that might transpose non-Mu DNA. In the "two-fragment model," two Mu fragments form a synaptic complex, similar to the one shown in Fig. 2A. A X174 molecule associates with this pre-assembled synaptic complex, contributing a few nucleotides to serve as the cleavage site. In the alternative "hybrid complex" model, synaptic complexes form between a single Mu fragment and a site on X174 (Fig. 4A). This model suggests that X174 DNA functions both structurally and chemically as one of the two transposon ends.
The two-fragment model posits that synapsing two Mu DNA fragments is a prerequisite for non-Mu transposition. By contrast, the hybrid-complex model predicts that synapsis of two fragments competes with assembly of hybrid transpososomes. In support of the hybrid complex model, pre-incubating MuA with an unjoinable Mu fragment inhibited X174 transposition (Fig. 4B, lane 4). The reactions shown here were primarily intermolecular, but in reactions without MuB, intramolecular transposition was also inhibited by preincubation with the Mu fragment (data not shown). Pre-incubating MuA alone (lane 1) or with X174 (lane 3) did not significantly inhibit subsequent reactions. Most likely, then, functional complexes are hybrids of a single Mu fragment and a X174 molecule.
The hybrid-complex model also predicts that high concentrations of Mu fragments should always inhibit non-Mu transposition, due to assembly of complexes on pairs of fragments. To test this prediction, we performed transposition assays in which we varied the concentration of a Mu fragment (in this case R1-R1, see Fig. 2B) but maintained the total DNA concentration constant using a DNA fragment of unrelated sequence but identical length. The yield of transposition products peaked sharply at 8 nM Mu fragment (Fig. 4C), substantially below the MuA concentration of 50 nM. These results further support the hybrid complex model. This experiment also showed that it is the Mu recognition sequence, as opposed to a general feature of short DNA fragments, that both activates and inhibits non-Mu transposition.
Two additional observations support the hybrid complex model. First, stable complexes of unjoinable fragment, X174 DNA, and MuA had been seen previously (17). X174 was assumed to be bound as target DNA in such complexes, but we now suggest that it is also bound as a donor partner. Second, in the accompanying paper we demonstrate that the regions of X174 that participate in transposition, when viewed as a group, bear For this figure and all that follow, unless noted otherwise, products were visualized by Vistra-green staining of an agarose gel. C, graph of product concentration versus fragment concentration shows a fragment optimum. The total DNA concentration was maintained constant at 1440 nM, by balancing the Mu fragment with a fragment of identical length but unrelated sequence. The Mu fragment was R1-R1 (see Fig. 2B), and the incubation time was 4 h to permit this reaction to be compared with the one shown in Fig. 6B. The graph shows % of DNA converted to "intermolecular" product as a function of [Mu fragment]. weak but significant resemblance to a Mu DNA end (27).
The Cleavage Site Is Optional on the Mu Fragment-Recall that the Mu transposon ends include two types of specific DNA sequences: recognition sites and cleavage sites. The unjoinable and R1-R1 fragments discussed above are, respectively, partially or entirely lacking a cleavage site sequence (Fig. 2B). We found that donor fragments with functional, uncleaved or precleaved cleavage sites also activated non-Mu transposition (Fig. 5, lanes 2-7). These fragments produced the figure-eight intermolecular product described above plus a novel intermolecular product. This novel product was probably due to transposition of a donor fragment together with the non-Mu DNA (see figure legend for details).
Precise, quantitative comparisons of the activities of different fragments was difficult, due to differences in the types of products each produced. But as a rough estimate, the uncleaved and precleaved fragments seemed to be less active in non-Mu transposition assays than were either the unjoinable or the R1-R1 fragment. Of all the fragments tested, the R1-R1 fragment seemed the most potent activator of non-Mu transposition (Fig. 5, lanes 11-13). This suggests that a cleavage site on the Mu DNA is at best neutral, and possibly inhibitory, to non-Mu transposition.
Fragments Containing a Single Recognition Site Support Non-Mu Transposition-We next asked whether the fragments needed to contain two recognition sites (positioned as the natural R1 and R2 sites) to activate non-Mu transposition. We expected that the two sites contribute to transpososome stability by correctly positioning two of the four MuA subunits with respect to each other. To determine whether this contribution is an essential one, we constructed fragments containing a single R1 site (see Fig. 2B). These R1 fragments were 22 base pairs long, containing half the sequence of the 46-base pair R1-R1 fragment. This minimal substrate was sufficient to permit non-Mu transposition (Fig. 6A). To observe activity with this fragment, we did require longer incubation times and higher fragment concentrations than with the fragment with two sites (data not shown and Fig. 6B). Nonetheless, the inter-molecular X174 transposition product was clearly visible and present only in reactions containing the Mu fragment (Fig. 6A,  see figure legend). We also confirmed that a nonspecific sequence could not substitute, by varying the concentration of the R1 fragment with a fragment of identical length but unrelated sequence (Fig. 6B). These results point to a powerful role of the recognition sequence in activating non-Mu transposition. They suggest a specific stimulatory effect of the recognition sequence on MuA transposase, independent of the physical positioning of subunits provided by two linked recognition sites.
Hybrid Complexes Are Stable to Competitor DNA-Transpososomes that synapse two native Mu DNA ends are extremely stable (22). Several lines of evidence suggest that both the recognition sequences and the cleavage sequence contribute to this stability through interactions with MuA (11,28). However, it is unclear to what extent stability correlates with functionality, and we wondered whether hybrid transpososomes are long-lived or transient. Our interest in this question was part of two larger questions. (i) Toward understanding the physical basis of transpososome stability, do non-Mu sequences permit assembly of long-lived transpososomes? (ii) Toward understanding the energetics of transposition, does a transpososome need to be long-lived to be functional?
Both DNA mobility-shift assays and transposition assays were used to probe the stability of hybrid transpososomes. DNA mobility-shift (or band-shift) assays were necessary because some of the relevant complexes are inactive for transposition. Activity assays were necessary to distinguish active from inactive complexes and to assess the longevity of complexes that were not stable to electrophoresis.
To do these experiments, we first established a transposition assay using non-Mu donor fragments. In place of normal MuA recognition sites, these fragments contained two copies of a sequence derived from X174 (Fig. 2B). The chosen X174 sequence has some resemblance to a MuA recognition sequence and was selected through a functional assay for its ability to be transposed by MuA (27). MuA transposed these fragments, albeit very poorly and only when MuA was present at 5-10 times our standard concentration of 50 nM (data not shown). Transposition of this non-Mu fragment was stimulated by a fragment containing bona fide MuA recognition sites (data not shown). Depending on which Mu fragment was used, we saw three types of products. 1) One product was the figure-eight intermolecular product described above. 2) Another product, indicated in the figure with a tailed-circle, contained (data not shown) a X174 dimer (as determined by two dimensional gel electrophoresis) and a Mu fragment (as determined by radiolabeling the Mu fragment). This tailedcircle product had some linear character, as defined by its relative migration in gels with different running buffers and by its sensitivity to exonuclease V (data not shown). This product probably results from pairwise transfer of a fragment and X174, and/or from reuse of a figure-eight product as a target in a second round of transposition. Band-shift assays allow us to distinguish three types of MuA⅐DNA complexes (Fig. 7B) (21). (i) One MuA monomer bound to one DNA fragment. These complexes are unstable to competitor DNA or heparin. (ii) MuA bound to DNA in a 2:1 ratio, also unstable to competitor DNA or heparin. These complexes could be either MuA tetramers in a synaptic complex or dimeric MuA bound to a single fragment. (iii) MuA bound to DNA in a 2:1 ratio but stable to competitor DNA and/or heparin. This third class is likely to be synaptic complexes.
The band-shifts revealed a difference in the stability of complexes formed on an unjoinable Mu fragment versus an R1-R1 fragment (Fig. 7B). Recall that the R1-R1 fragment lacks the entire cleavage site sequence, while the unjoinable fragment only lacks a single A from the cleavage site. There are also minor differences in the recognition sequences of these two fragments (see Fig. 2B). Complexes formed on the unjoinable fragment were stable either to competitor DNA (lane 6) or to heparin (lane 5), whereas those formed on the R1-R1 fragment were stable to competitor DNA (lane 12), but not to heparin (lane 11). The R1-R1 results are the first evidence we have seen that heparin directly destabilizes MuA complexes, as opposed to providing a "sink" for protein that has dissociated from DNA. These results also suggest that the cleavage site DNA helps stabilize synaptic complexes.
The non-Mu (X174-derived) fragment did not by itself support the formation of competitor-stable complexes (Fig. 7B,  lane 3). Yet in the presence of the unjoinable Mu fragment, complexes with the non-Mu fragment were stable to either heparin or competitor DNA (lanes 8 and 9). Thus, the unjoinable fragment has a stabilizing influence on these hybrid complexes. The R1-R1 fragment was not sufficient to stabilize complexes on the non-Mu fragment, as assessed by band-shift assays (lanes 14 and 15). However, the next experiment shows that some R1-R1⅐non-Mu hybrid complexes were stable to competitor DNA, though not sufficiently stable to be detected by gel electrophoresis.
We next used transposition activity to probe complex stability (Fig. 7C). In this experiment, detectable transposition of non-Mu fragments depended on complementation with Mu fragments (data not shown). Complexes were assembled with a mixture of unlabeled Mu fragments and labeled non-Mu fragments, in the absence of magnesium (to prevent transposition). In a second, 20-min incubation, we added an excess of competitor DNA to challenge the preformed complexes. Finally, in a third incubation, we added target DNA and magnesium to initiate transposition of the non-Mu fragments.
Hybrid transpososomes survived the challenge with competitor DNA, whether the unlabeled Mu partner was the R1-R1 or the unjoinable fragment. Control experiments confirmed that the competitor DNA was sufficient to abolish transposition if complexes were not pre-assembled in advance of adding the competitor (data not shown). If complexes were pre-assembled, the transposition efficiency was essentially the same irrespective of whether the competitor was added before or together with the magnesium and the target (i.e. second or third incubation). Thus, a functional assay reveals that transpososomes containing a non-Mu donor DNA are long-lived, with a half-life greater than 20 min.
Finally, we used a similar strategy to assess the longevity of hybrid complexes with full-sized X174 donor (Fig. 7D). This time, the DNA present in the preassembly step was X174 RFI and a Mu fragment. Preformed complexes were challenged with excess R1-R1 fragment, and then transposition was initiated with magnesium. Substantial intermolecular product appeared, despite the added challenge fragment and even if the original Mu fragment contained only a single R1 site (Fig. 7D,   lane 8). Therefore, active complexes formed between X174 and a fragment with a single R1 site are long-lived. These results show that despite all the deficiencies of their component parts, these hybrid complexes are indeed stable transpososomes.
Mini-Mu Plasmids Can Activate Non-Mu Transposition-Armed with the knowledge that non-Mu DNA can transpose, we revisited an unexplained family of products seen in previous studies and discovered the products to be the result of intramolecular transposition of X174. These products were observed in reactions containing mixtures of wild-type MuA and MuA with active site mutations (specifically, the D269N and E392Q substitutions, MuADE/NQ). This mutant protein cannot perform donor cleavage or strand transfer, but it is efficiently incorporated into stable synaptic complexes. The activity of transpososomes containing both wild-type and mutant subunits depends on the placement of subunits. For example, some mixed complexes are fully active, some are not active at all, and some are able to complete cleavage and strand transfer of only one of the two Mu DNA ends (23,29,30).
In transposition reactions containing a mini-Mu donor plasmid, X174 RFI DNA, MuB, and a 1:1 mixture of MuA and MuADE/NQ, the most abundant products were the result of mini-Mu donor transposing into X174 target (labeled Mu-X interST in Fig. 8). But an additional family of products appeared that ran between relaxed mini-Mu and relaxed X174 (Fig. 8, lanes 3 and 4, labeled X-intraST and marked with a bracket). These unexplained products did not contain sequences from the mini-Mu plasmid, as determined by Southern blot analysis (data not shown) and by their insensitivity to BglI, a restriction enzyme that cleaves the Mu plasmid but not X174 (Fig. 8, lane 4). The products did contain X174 DNA, as again determined by Southern blotting (data not shown) and restriction analysis (Fig. 8, lane 5). Although the mini-Mu was not covalently joined to the final products, it was an essential component of the reaction (Fig. 8, lane 6), reminiscent of the role of the Mu donor fragments in the reactions described above.
These X-only products were most likely generated by hybrid complexes having the following structure: the two donor sites were provided by a mini-Mu plasmid and a X174 molecule, respectively, the target site was another site on the same X174 molecule, and a MuADE/NQ subunit was incorporated in a position that blocked joining of the mini-Mu molecule to X174 (Fig. 8B). The resulting products were intramolecular transposition products of X174, similar to those characterized in Fig. 3.
Although the products of these reactions were intramolecular, the reactions required the MuB protein (data not shown). This suggests a role for MuB in bringing together two large DNA molecules in one transpososome. Ordinarily, the two large DNAs are a donor and a target respectively. In this special case, both DNAs participated as donors, and one (the non-Mu) was additionally a target.
The mutant version of MuA was needed to see intramolecular X174 products (Fig. 8A, lane 2). However, it is unclear whether the mutant protein stimulated non-Mu transposition, or simply permitted detection of a reaction whose products are normally obscured by other products. If all of the MuA subunits have functional active sites, products of the analogous reaction should include a mini-Mu molecule. The products would have a mobility similar, though not identical, to standard transposition products. Thus, it is possible that non-Mu transposition occurs as a side reaction in most transposition experiments. Regardless of which explanation is correct, these experiments revealed non-Mu transposition under reaction conditions that FIG. 7. Hybrid transpososomes are stable to competitor DNA. A, summary of experimental design. Transpososome complexes were assembled, challenged with an excess competitor DNA fragment, and assayed for stability. B, band shift assays directly address complex stability. In a first incubation, complexes were formed by incubating 50 nM labeled fragment and 50 nM unlabeled fragment with 150 nM MuA for 40 min. In some cases the labeled and unlabeled were the same fragment; both were included to keep the DNA concentration constant between experiments. During a second 60-min incubation, preformed complexes were challenged with 0.2 g of heparin (h: lanes 2, 5, 8, 11, and 14) or 800 nM cold R1-R1 fragment (d: lanes 3, 6, 9, 12, and 15). Complex stability was then confirmed by gel electrophoresis and autoradiogram. MuB, target DNA, and MgCl 2 were not included at any point in this experiment. C, transposition assays reveal that hybrid transpososomes are long-lived. In this autoradiogram, the only visible product results from strand transfer of a non-Mu fragment into a X174 target. Complexes were assembled by incubating 200 nM MuA with 50 nM labeled non-Mu fragment and 50 nM unlabeled Mu fragment for 4 h. The absence of divalent metal at this step prevented transposition (assembly on precleaved fragments does not require divalent metal). In a second 20-min incubation, excess competitor DNA (500 nM unlabeled, unjoinable fragment) was added to the reactions shown in lanes 1 or 2, to challenge the stability of preformed complexes. In a third incubation, target DNA and MgCl 2 were added, as well as competitor DNA to reactions shown in lanes 3 and 4. Numbers above the gel refer to the incubation step for which competitor DNA was added (see Fig. 7A). Lanes 1 and 3, during the first incubation the unlabeled fragment was unjoinable; lanes 2 and 4, it was R1-R1 fragment. D, transposition assays reveal stability of transpososomes formed on X174 RFI DNA. In a first incubation, 150 nM MuA was incubated for 1 h with 570 nM MuB, X174, and Mu fragment in 10 mM CaCl 2 to permit assembly but prevent transposition. Either the R1-R1 fragment was present at 8 nM (lanes 1-4), or the R1 fragment was present at 1.4 M (lanes 5-8). In a second incubation, excess competitor DNA (500 nM R1-R1 fragment) was added to the reactions shown in lanes 4 or 8 to challenge the stability of preformed complexes. In a third incubation, MgCl 2 was added, along with 300 nM additional MuB, to initiate transposition. Numbers above the gel refer to the incubation step for which excess competitor DNA was added (see Fig. 7A). Lanes 1 and 5 confirm that no transposition occurred in the absence of magnesium. Lanes 3 and 7 confirm that the competitor DNA competed successfully. are quite different from the fragment assays described above. Thus, multiple reaction conditions reveal the power of Mu DNA to stimulate transposition. DISCUSSION We find that Mu DNA can activate MuA to transpose non-Mu DNA. Similar promiscuous activity has been observed previously in vivo, with a donor plasmid containing a single transposon end sequence. This "single-end" transposon improvises a second "end" from other sequences on the same plasmid and transposes with a frequency 100-fold above background. Though this single-end transposon is ϳ1000-fold less active than a transposon with two proper ends, a plasmid without any end sequence is indistinguishable from transposase-free controls (31). The transposons Tn3, Tn1721, and Tn21 have also been shown to perform single-ended transposition in vivo (32)(33)(34), and in vivo use of cryptic recombination signal sequences is well documented for Rag-1/2 (35).
Of course, for all these transposable elements the dominant pathway requires pairing of two bona fide transposon ends. Promiscuous transposition may have evolved as a default pathway for the rare times when a transpososome begins to assemble on a single end sequence. If a single end were to transpose without first pairing with some other DNA sequence, it would cause chromosomal rearrangements deleterious to the host. Promiscuous transposition, in which a nearby site is synapsed with a transposon end and the two transpose together, would be less harmful. For transposons that remain integrated at one location for many host generations, preserving the host genome is almost as important as preserving their own.
Alternatively, promiscuous transposition may be irrelevant in the wild, as the presence of one transposon end usually means another end is nearby. Nonetheless, the promiscuous activities exposed in artificial settings reflect an important aspect of the natural activity of transposases. For example, the Mu synaptic complex is extremely stable, which helps ensure that transposition does not abort before both ends have transposed (22). Some of the protein-protein and protein-DNA interactions that stabilize native synaptic complexes presumably contribute to synapsing promiscuous sites. The transposase is designed to tenaciously bind two pieces of DNA, and promiscuous activities are likely a consequence of this tenacious binding.
DNA Recognition Sites Help Organize MuA Subunits-A DNA fragment containing two MuA recognition sites strongly activates non-Mu transposition. A shorter fragment with only one recognition site is far less effective, even if protein and DNA concentrations are well above the K d for simple binding. These results suggest that one important function of the MuA recognition sites, both for standard Mu transposition and for non-Mu transposition, is to spatially constrain two MuA subunits. By bringing two monomers into close proximity to each other, the DNA sites can contribute greatly to driving transpososome assembly.
Other studies also highlight the importance of Mu DNA in spatially organizing the MuA monomers to form an active tetramer. First, under stringent reaction conditions tetramer assembly requires at least two recognition sites at each transposon end (11,28). Transpososome assembly is also sensitive to the spacing between these two required recognition sites (28). Second, again under stringent conditions, the sites at the two ends must be inverted relative to each other. This orientation requirement suggests an additional level of organization, in which the two ends together establish the orientation of all four MuA subunits (11,36). Third, the accompanying paper suggests that the recognition sites help position MuA subunits relative to the cleavage sites (27).
On the full Mu transposon, the R1 and R2 sites bring together and position two MuA subunits. The L1 and L2 sites probably function similarly. Although L2 is 80 base pairs from L1, the DNA-bending protein HU brings L1 and L2 together for assembly (37).
MuA Recognition Sites Are Allosteric Activators of Transposition-A fragment with just one MuA recognition site is sufficient to activate transposition of non-Mu DNA. This fragment can pair with X174 RFI DNA to form a MuA transpososome, which is active for transposition even after challenge with a competitor DNA. These results are consistent with and extend a previous study, which showed that under permissive reaction conditions a plasmid with a single MuA recognition site could support assembly of MuA tetramers detectable by protein cross-linking (11).
A single recognition site, absent the nearby cleavage site, is unlikely to constrain the position of more than one MuA subunit. Chemical and nuclease protection patterns strongly suggest that a MuA monomer contacts the length of an entire recognition site (38). Multiple subunits could make overlapping contacts, but an extensive overlap seems unlikely. Thus our single-site R1 fragments are unlikely to have provided much energy toward organizing the subunits of the transpososome. The ability of these R1 fragments to activate transposition points to an additional, more specific role of the individual recognition sites. The simplest explanation is that interactions with the recognition sequence induce a conformational change in MuA necessary for transposition. Most likely, this conformational change allosterically activates transpososome assembly. Although a single recognition site may directly affect only one MuA molecule, allosteric changes in that molecule can cooperatively recruit other MuA molecules to the complex, some bound to non-Mu DNA. It is also possible that the recognition sites induce conformational changes that are important for a post-assembly step, such as DNA cleavage or strand transfer.
The Role of the Cleavage Site-Previous studies of MuA indicated a role for the cleavage site in assembling stable synaptic complexes (28,39). Results of this study support that conclusion (Fig. 7B). Why, then, is the cleavage site not an important feature of activator fragments for transposition of non-Mu DNA? We can offer two thoughts on this topic. (i) Participation in super-stable complexes may not be ideal for an activator fragment, as complexes between two fragments compete with formation of hybrid complexes. (ii) The adjoining paper describes 18 of the sites from X174 that transposed as part of hybrid transpososomes (27). The cleavage site sequence is a strong feature among these sites. Thus, the two halves of the complexes in our studies complement each other. The Mu DNA fragment provides the recognition sites to align two subunits and promote allosteric changes in MuA. The non-Mu DNA provides the cleavage site sequence, which serves as the actual site of transposition and may also provide stabilizing contacts to the complex.
Sequence-specific Activation of Other Proteins-The MuA recognition sites perform a complex set of functions. They provide a structural buttress for the transpososome, arranging multiple protein subunits with precision. They also individually activate MuA subunits, probably by inducing conformational changes that are transmitted from the protein's DNA binding domain to the other domains involved in forming a functional transpososome.
There are many other contexts in which DNA sequences provide an assembly board for protein-protein interactions. For example, DnaA sites scaffold the initiation machinery at the OriC origin of replication (40). Many bacterial repressors multimerize through cooperative binding to neighboring recognition sites, and eukaryotic transcription usually involves an even more elaborate set of protein complexes that fully assemble only when constrained by DNA (41). In some of these cases, DNA sequences direct the function of a protein complex simply by providing spatial organization for that complex without additionally inducing allostery. For example, the commonly used yeast two-hybrid assay relies on the fact that the Gal-4 transcription factor uses DNA simply to position itself at a gene promoter. Likewise, the repressor has an experimentally alterable linker connecting its DNA binding domain to its cooperativity (protein-protein interaction) domain, making it unlikely that DNA allosterically induces cooperative binding of repressor (42).
Nonetheless, DNA-promoted allostery may be wide-spread as a means of directing a protein activity to a specific DNA sequence. For example, some restriction enzymes cleave the DNA at a fixed distance from a recognition site. The specificity of these enzymes probably relies on allosteric changes transmitted from the DNA-binding domain to the catalytic domain (43). Some transcriptional regulators, for example Ets-1 and the glucocorticoid receptor, are also effected by DNA-promoted allostery (44). Finally, DNA-promoted allostery is probably common among transposases. Binding of Tn5 transposase to its recognition sequence promotes protein dimerization by altering a helix connected to a C-terminal protein interaction domain (45). In addition, upon forming a synaptic complex Tn5 transposase makes significant changes in its DNA contacts, both within the recognition site and around the cleavage site, suggesting changes in the protein structure (46).
Transposases are a particularly interesting example of DNApromoted allostery, as potentially are site-specific recombinases, DNA repair complexes such as MutSHL, and certain restriction enzymes, like FokI, that may multimerize only on DNA (47)(48)(49). These proteins all function in catalytic pathways, but their activities require pre-assembly into multimeric protein-DNA complexes. Thus, DNA-promoted conformational changes could function at multiple points along the pathway, including the assembly step and subsequent reaction steps. It will be interesting to learn whether the allosteric changes induced by MuA recognition sites directly affect the cleavage and strand transfer reactions or are primarily required for assembly of a functional complex.