Development of Hyperactive Sleeping Beauty Transposon Vectors by Mutational Analysis

Development of Hyperactive Sleeping Beauty Transposon Vectors by Mutational Analysis

ARTICLE doi:10.1016/j.ymthe.2003.11.024 Development of Hyperactive Sleeping Beauty Transposon Vectors by Mutational Analysis Hatem Zayed,1 Zsuzsanna...

725KB Sizes 0 Downloads 24 Views

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

Development of Hyperactive Sleeping Beauty Transposon Vectors by Mutational Analysis Hatem Zayed,1 Zsuzsanna Izsva´k,1,2 Oliver Walisko,1 and Zolta´n Ivics1,* 2

1 Max Delbru¨ck Center for Molecular Medicine, Robert Ro¨ssle Strasse 10, D-13092 Berlin, Germany Institute of Biochemistry, Biological Research Center of the Hungarian Academy of Sciences, Szeged, Hungary

*To whom correspondence and reprint requests should be addressed. Fax: +(49) 30 9406-2547. E-mail: [email protected]

The Sleeping Beauty (SB) transposable element is a promising vector for transgenesis in vertebrates and is being developed as a novel, nonviral system for gene therapeutic purposes. A mutagenesis approach was undertaken to improve various aspects of the transposon, including safety and overall efficiency of gene transfer in human cells. Deletional analysis of transposon sequences within first-generation SB vectors showed that the inverted repeats of the element are necessary and sufficient to mediate high-efficiency transposition. We constructed a ‘‘sandwich’’ transposon, in which the DNA to be mobilized is flanked by two complete SB elements arranged in an inverted orientation. The sandwich element has superior ability to transpose >10-kb transgenes, thereby extending the cloning capacity of SB-based vectors. We derived hyperactive versions of the SB transposase by single-amino-acid substitutions. These mutations act synergistically and result in an almost fourfold enhancement of activity compared to the wildtype transposase. When combined with hyperactive transposons and transiently overexpressed HMGB1, a cellular cofactor of SB transposition, hyperactive transposases elevate transposition by almost an order of magnitude compared to the first-generation transposon system. The improved vector system should prove useful for efficient gene transfer in vertebrates.

INTRODUCTION Considerable effort has been devoted to the development of gene delivery strategies for the treatment of inherited and acquired disorders in humans [1]. These methods can be broadly classified as viral and nonviral technologies, and all have advantages and limitations. Viral vectors, where available, are efficient at introducing and expressing genes in cells. However, adapting viruses for gene transfer restricts genetic design to the constraints of the virus in terms of size, structure, and regulation of expression. Nonviral methods, including DNA-condensing agents, liposomes, microinjection, and ‘‘gene guns,’’ might be easier and safer to use than viruses, but are not equipped to promote integration into chromosomes. As a result, stable gene transfer frequencies using nonviral systems have been very low. A relatively new addition to the gene therapist’s toolbox is transposable element-based gene vectors [2,3]. Transposons are genetic elements that have the distinctive ability to move in genomes. Especially useful for genetic analyses are members of a class of transposable elements that move via a ‘‘cut-and-paste’’ mechanism: the transposase catalyzes excision of the transposon from

292

its original location and promotes its reintegration elsewhere in the genome [4]. The simplest DNA transposons are framed by terminal inverted repeats (IR) and contain a single gene encoding a transposase. The transposase can trans-mobilize elements as long as they retain the IRs, which forms the basis of powerful experimental manipulation. The P transposable element has revolutionized Drosophila genetics and is widely used as a vector for germ-line transgenesis and insertional mutagenesis in flies [5]. Until very recently, transposon vectors were not available for genetic analyses in vertebrates. This is because the vast majority of elements currently residing in vertebrate genomes are transpositionally inactive [6 – 9]. To address this problem, two Tc1/mariner-like elements called Sleeping Beauty (SB) [2] and Frog Prince [10] were reactivated from ancient transposon fossils recovered from fish and frog genomes. SB shows efficient transposition in a variety of vertebrate (including human) cell lines in tissue culture [3,11] and in the mouse in vivo, both in somatic tissues [12] and in the germ line [13 – 17]. Recent experiments from several laboratories have demonstrated some advantages of SB over the currently

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy 1525-0016/$30.00

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

used viral and nonviral vectors, including stable, singlecopy integration [12,18], use of simple plasmid vectors [3,18], and long-term expression of integrated transgenes at therapeutic levels [12,19 – 22]. However, in vivo transformation rates with naked SB plasmids administered through the tail vein into mice were only around 5% in the liver [12]. Two immediately obvious areas in which the efficiency of SB-mediated gene transfer could potentially be improved are the efficiency of vector delivery and the intrinsic transpositional activity of the element itself. In this work we aimed at a refinement of the SB system for higher transpositional activity. Previous improvements have been made by manipulating either the transposon IRs or the transposase protein. For example, hyperactive SB vectors have been generated by reducing the length of vector DNA outside the transposon in donor plasmids [3] or by the introduction of sitespecific mutations into the IRs [23]. However, even the improved vectors are subject to size restrictions: transposition frequency of SB decreases with increasing length of the transposon [3,18,24]. Large (>10 kb) pieces of genomic DNA flanked by two identical copies of Paris elements have been mobilized in Drosophila virilis [25]. Paris is a Tc1/mariner-type transposon, which suggests that mimicking such a naturally occurring arrangement could possibly extend the capacity of SB vectors to transpose large DNAs. Mutations in the SB transposase have been shown to result in hyperactivity, yielding transposase versions with an approximately threefold higher activity than the wild-type SB transposase [18]. There are several mechanisms of hyperactivity in transposases. For example, hyperactive phenotypes of the bacterial element Tn5 are due to the reduction of the self-inhibitory activity of intact Tn5 transposase [26], a reduced affinity of an inhibitor protein to the transposase [27], or an increase in the binding affinity of the transposase to its binding sites within the transposon IRs [28]. The combination of these three hyperactive mutants yields a synergistic effect, leading to an extraordinarily active transposase [29]. Interestingly, amino acid replacements that change glutamic acid (E) residues to lysine (K) led to hyperactive transposase versions in three different transposon systems, Tn5 [26,28], Tn10 [30], and Himar1 [31]. Introducing a proline residue, a secondary structure breaker, at a defined site in the Tn5 transposase also resulted in a hyperactive mutant [27]. In this work we set out to improve SB’s transpositional efficiency by modifying the structure of the transposon IRs and by rationally designed site-directed mutagenesis of the transposase. We constructed a ‘‘sandwich’’ transposon, in which two complete SB elements are arranged in a head-to-head orientation. The sandwich element has superior ability to transpose large transgenes in tissue culture transposition assays in hu-

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

man cells. We identified three single-amino-acid substitutions that result in hyperactivity of the SB transposase. When combined, these mutations have a synergistic effect and result in an almost fourfold enhancement of activity compared to the wild-type transposase. Together with improved transposon vectors and transiently overexpressed HMGB1, hyperactive transposases boost transpositional activity to about eightfold compared to the wild-type components of the transposon system. The modified vector system represents a significant advance in vector development for safe and efficient gene delivery in vertebrates.

RESULTS Defining the Minimal cis Requirements for Sleeping Beauty Transposition SB has a pair of transposase-binding sites at the ends of its IRs, a structure termed IR/DR [32] (Fig. 1A). In nature, the two functional components of the SB transposable element are physically linked: the transposase gene is located between the two IR/DRs (Fig. 1A). However, the two components are separable; a gene of interest can be cloned between the IR/DRs, and the transposase can be supplied in trans, expressed from the transposase gene located either on the same [33] or on a second plasmid [2] (Fig. 1A). First-generation SB transposon vectors contain DNA sequences of about 120 bp between the left IR/DR and the transposase gene as well as a significant portion (f200 bp) of the transposase coding region [2] (Fig. 1A). This presents a possible safety issue for the application of the SB transposon system for gene therapy, because these sequences can potentially provide sufficient homology for recombination with the transposase gene in cells, thereby creating an autonomous, and therefore uncontrollable, transposable element. In an attempt to define the minimal cis requirements for SB transposition, we subjected the pT/neo transposon donor construct (SB transposon marked with a neo transgene) [2] to an analysis in which parts of the transposable element between the IR/DR and the neo transgene were deleted (Fig. 1B). We cotransfected the donor constructs with a plasmid expressing the transposase (pCMV-SB) into HeLa cells. We placed the cells under antibiotic (G418) selection and compared the numbers of antibiotic-resistant colonies as a measure of transpositional efficiency. Construction of the deletion constructs required the introduction of an SpeI restriction site, which did not lead to major changes in transposition efficiency compared to the pT/neo control (Experiments 1 and 2 in Fig. 1B). Further elimination of sequences of up to 112 bp still allowed efficient transposition (Experiments 3 to 5), whereas complete elimination of intervening sequences between the internal transposase binding site and the transgene drastically reduced transposition efficiency (Experiment

293

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

FIG. 1. Definition of the minimal cis requirements for Sleeping Beauty transposition by deletional analysis. (A) Schematic representation of the structural organization of the two functional components of SB elements. In nature, the gene encoding the trans-acting transposase is physically linked to the cis-acting terminal inverted repeats (black arrows at the ends of the transposon) that contain the binding sites for the transposase (small white arrows). In the laboratory, these two components are separated. A gene of interest (such as the neo antibiotic resistance gene) is cloned between the IR/DR repeats, and the transposase is expressed from the transposase gene maintained on a separate plasmid. Black arrowheads indicate enhancer/promoter elements driving the expression of the transgenes. In addition to the IR/DR repeats, some sequences upstream of the transposase gene as well as transposase coding sequences are retained in firstgeneration SB vectors. (B) Deletional analysis of DNA sequences between the left IR/DR and the SV40 promoter-driven neo transgene. On the left, schematic drawings show the IR/DR repeats with the transposase binding sites, the neo transgene driven by the SV40 enhancer/promoter, and sequences between the IR/ DR and the transgene that were subject to deletion. The dotted lines represent the deleted sequences in the different constructs. On the right, the transpositional activities of the different deletion derivatives are shown relative to that of the wild-type transposon vector pT/neo, which was set to 100%.

6). These results suggest that most of the transposon sequences between the IR/DR and the transgene are not required for efficient transposition, but a minimum of 8 bp downstream of the internal transposase binding site must be retained. We wondered whether this requirement is sequence-specific; thus, we replaced the 8 bp in the transposon with an unrelated sequence. Incorporation of unrelated sequences nearly restored wild-type levels of transposition (Experiment 7), suggesting a requirement for a spacer region between the transposase binding site and the promoter driving the expression of the neo gene. These results collectively establish that the IR/DRs are sufficient to mediate high level transposition. A Sandwich Arrangement of Two Complete Sleeping Beauty Transposons Can Mobilize Large Transgenes Several improved versions of the SB transposon have been engineered [3,23], but all suffer from limitations in insert size. The Tc1/mariner element Paris has been shown to form mobile, composite elements by flanking long DNAs in an inverted orientation [25]. We reasoned that by mimicking the structure of such composite elements,

294

the capacity of SB transposase to mobilize long transgenes might be enhanced. The sandwich element is a new transposable entity in which DNA flanked by two copies of SB is mobilized. A requirement for such a transposon to work is the inability of the individual SB units to transpose on their own. The terminal nucleotides of the Tc1 element in Caenorhabditis elegans have been previously shown to be required for element excision [34]. Therefore, we mutated the terminal 5V-CA bases of the right IR of pT/neo to 5V-GC (Fig. 2A). To test the effects of these mutations on transposition, we compared the activity of the pT*/neo (with mutant right IR) transposon donor construct to that of pT/neo (wild-type control), using the transposition assay described above. In the negative control, a plasmid expressing h-galactosidase (pCMVh) replaced the transposase. The results showed that the CA ! GC mutations completely abolish transposition of T*/neo (Fig. 2B), indicating that transposition of individual SB elements from sandwich constructs can be efficiently inhibited. All four binding sites within the IR/DR structure (Fig. 2A) are required for SB transposition [3]. Therefore, a further (suspected) requirement for the sandwich trans-

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

doi:10.1016/j.ymthe.2003.11.024

FIG. 2. Mutations in the right inverted repeat of Sleeping Beauty interfere with transposition, but not with the binding capacity of the transposase. (A) Schematic representation of the SB transposon. The transposase gene is flanked by IR/DR-type inverted repeats (black arrows), which contain the binding sites for the transposase (white arrows at the ends of the IRs). Two base-pair changes were introduced at the terminus of the right IR. The C in the sequence 5V-CAGTTGAAG. . . is the first base of the transposon. (B) A transposon with mutant right IR cannot transpose. Efficiency of transposition is assessed as an increase in G418-resistant colony numbers in the presence (SB) versus in the absence (hgal) of transposase. Numbers are per 3  104 transfected HeLa cells. The graph shows that a transposon that has the mutant IR (pT*/neo) cannot be mobilized by the transposase. (C) Transposase can bind the mutant IR. Electrophoretic mobility shift assay using 32P-radiolabeled wild-type (wt) or mutated (IR*) IR fragments as probes and N-123, a derivative of SB transposase containing the specific DNA-binding domain of the SB transposase within the N-terminal 123 amino acids.

poson to work is that the transposase should be able to bind to all of its binding sites within the composite element. We radiolabeled both the wild-type and the mutant IRs and examined their ability to be bound by the transposase in a mobility shift experiment, using N123 (the DNA-binding domain of SB transposase) (Fig. 2C). The results showed no difference between the wild-

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

ARTICLE

type and the mutant IR fragments in terms of binding to N123 (Fig. 2C, lanes 2 and 4). These results therefore demonstrate that the induced mutations interfere only with the catalytic steps of transposition and not with SB transposase binding. Next, marker transgenes were cloned between two T* transposons (containing no transgene) in an inverted orientation, thereby forming a sandwich-like arrangement, from which the two individual SB elements cannot excise but together define a new composite element (Fig. 3A). The structure of the sandwich transposon is therefore as follows: (intact left IR) – body of SB element – (disabled right IR) – insert with a selection marker – (disabled right IR) – body of SB element – (intact left IR). Earlier results showed that, although SB was able to transpose transgenes of up to f10 kb, the efficiency of transposition significantly dropped as the elements got longer than 4 kb in length [3]. Therefore, we subcloned a 4.7-kb piece of DNA into the sandwich vector to yield a total transposon length of about 7.7 kb (construct pT/ SA7.7 in Fig. 3B). We subcloned an additional 4.5-kb piece of DNA containing the lacZ gene into pT/SA7.7 to yield a total transposon length of 12.2 kb (construct pT/ SA12.2 in Fig. 3B). We tested the efficiency of transposition of the sandwich constructs using the in vivo transposition assay and wild-type transposon constructs as controls for comparison. The sandwich transposon T/SA7.7 jumped about 3fold more efficiently than the similar-size, wild-type marker transposon T7.5 and about 2.2-fold more efficiently than T6.2, a wild-type transposon that contains the same transgene insert as T/SA7.7 (Fig. 3B). This result indicates that the sandwich vector is indeed more efficient in transposing relatively long DNA fragments than wild-type SB. Transposition of the sandwich element T/ SA12.2 was still more efficient than that of a 10.3-kb-long wild-type transposon (Fig. 3B). However, the sandwich transposon apparently abides by the same rule as wildtype SB, namely, that transposition rates are inversely proportional to the length of the transposon [3] (Fig. 3B). Our results suggest that increasing the numbers of binding sites for the transposase can improve transposition of large-size transposable elements and establish the sandwich element as a useful transposon vector for stable integration of large transgenes. Mutagenesis of the Sleeping Beauty Transposase at a Linker Region between the DNA-Binding and the Catalytic Domains Transposons and their hosts have coevolved and developed strategies that reduce the negative effects on the host but ensure proliferation of the element [9]. Thus, those elements that were apparently very successful in propagating themselves within a genome and in colonizing new genomes through horizontal transmission, such as Sleeping Beauty [6,7], are unlikely to represent

295

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

FIG. 3. The sandwich Sleeping Beauty vector shows enhanced capacity to transpose long transgenes. (A) Outline of wild-type and sandwich SB transposons. In the sandwich vector, two complete SB elements flank a transgene to be mobilized in an inverted orientation. The individual SB units cannot transpose due to the mutations in their right IRs (asterisks). Only the full, composite element can transpose. The small, white arrows are the binding sites for the transposase within the left (large black arrows) and right (large white arrows) IR/DR repeats. (B) Comparison of the respective transpositional efficiencies of wild-type and sandwich transposon vectors. Transfections were done in human HeLa cells, and efficiency of transposition is expressed as a ratio of colony numbers in the presence versus in the absence of transposase. The graph shows that the sandwich transposon vector (black columns) has superior ability over wild-type transposons to integrate transgenes longer than 7 kb.

their most active forms. This predicts that hyperactive versions of transposases can be generated by mutational analysis, which is the case for several transposases, including Tn5 [26 – 28], Tn10 [30], Himar1 [31], and SB [18]. Mutations into genes can be introduced in either a random or a site-directed fashion. We explored three different approaches to site-directed mutagenesis of the SB transposase: (1) mutagenesis of a linker region between the DNA-binding and the catalytic domains, (2) replacement of acidic amino acids with basic amino acids, and (3) incorporation of naturally occurring sequence variants into the transposase. The basis of the first approach is that, in the Tn5 transposase, there is interference between the C-terminal region and the N-terminus during interaction with the

296

transposon DNA [35]. Introduction of a proline residue relieves this interference by facilitating a conformational change, thereby leading to a hyperactive transposase [27,35]. SB transposase consists of an N-terminal DNAbinding domain that mediates interaction with the transposon IRs and a catalytic domain responsible for the DNA-cleavage and -joining reactions [2,6,8]. A region immediately following the DNA-binding domain has a negative impact on transposase binding to the transposon inverted repeats (Fig. 4A), supporting the hypothesis that it promotes an unfavorable conformation of the transposase. This region is predicted to assume a helical conformation and is conserved in the Tc1 family (Fig. 4B). Based on the Tn5 observations, we reasoned that introduction of proline residues in the predicted helix

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

doi:10.1016/j.ymthe.2003.11.024

ARTICLE

FIG. 4. A region between the DNA-binding and the catalytic domains of the Sleeping Beauty transposase interferes with efficient substrate binding. (A) Mobility shift experiment using N-123 and N-161 derivatives of the SB transposase and a radiolabeled IR/DR fragment as a probe. N-123 contains the N-terminal 123 amino acids of the transposase, encompassing the DNA-binding domain responsible for binding to two sites within each of the IR/DR repeats. N-161 contains the N-terminal 161 amino acids of the transposase, encompassing the DNA-binding domain plus a predicted helix. The two retarded bands (arrowheads) in the N-123 sample represent complexes in which one (faster migrating complex) or both (slower migrating complex) transposase binding sites within the transposon IR/DR probe are bound by the transposase. The gel shows severe reduction of complex formation by N-161. In the control reaction (C), no protein is added to the probe. (B) A conserved, predicted helix in Tc1-like transposases between the DNA-binding and the catalytic domains. An amino acid alignment of Tc1-like transposase segments is shown. The exact amino acid positions at which N-123 and N-161 terminate are indicated. The predicted helical region is boxed, and small black arrowheads mark those amino acids within the helix in the SB transposase that were replaced by proline residues. The position of the first D residue of the DDE catalytic triad is indicated. Amino acid sequence alignment, sequence conservation, and secondary structure prediction were done using ClustalW, BOXSHADE 3.21, and PredictProtein softwares, respectively.

motif between the DNA-binding and the catalytic domains could lead to a change in transposase structure, thereby allowing better access of the transposase to its binding sites. Toward that end, we introduced the mutations L132P, F134P, T136P, and D140P (only one at a time) into the transposase (Fig. 4B). Next, we tested the transpositional activities of the mutant transposases relative to the wild-type (SB10) transposase in the transposition assay. The first three mutations essentially abolished transposition, whereas D140P showed about 17% of the wild-type activity (Fig. 5A). These results indicate that these amino acid replacements are detrimental to the activity of the SB transposase, possibly because changing the spatial arrangement of

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

the transposase domains leads to a nonfunctional structure. Nevertheless, the introduction of less drastic changes possibly by random mutagenesis of this region of the transposase is a promising strategy for future work. Amino Acid Replacements That Result in Hypertransposing Mutants of the Sleeping Beauty Transposase Based on findings that some of the Tn5, Tn10, and Himar1 hyperactive mutations are acidic to basic amino acid replacements, we hypothesized that similar mutations also have the potential to increase transpositional activity of the SB transposase. There are altogether 28 aspartic acid (D) and glutamic acid (E) residues in the SB

297

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

FIG. 5. Effects of amino acid substitutions on the efficiency of Sleeping Beauty transposition. (A) Effects of single-amino-acid replacements. Shown are the transpositional efficiencies of 22 single-amino-acid mutants of the transposase relative to the wild-type SB10 transposase (white column). From these, 15 are E to K/R mutations, 5 are proline mutations in the linker region between the DNA-binding and the catalytic domains (Fig. 4B), and 2 are naturally occurring sequence variants. The three individual hyperactive mutants identified in this screen are shown as black columns. (B) Effects of hyperactive mutations in combinations. Transposition was assayed in human HeLa cells, and the activity of wild-type transposase (SB10, white column) is taken as a reference and set to 100%. (C) Effects of combinations of hyperactive transposases, hyperactive transposons, and HMGB1. Transposition was assayed in human HeLa cells, and the activity of the firstgeneration SB system (SB10 plus T/zeo, white column) is taken as a reference and set to 100%.

transposase, which are listed in Table 1. We categorized these amino acids with respect to their conservation in the Tc1 family (Table 1). Conserved amino acids likely play crucial roles in transposase activity. For example, the DDE residues of the catalytic domain are absolutely conserved in the transposases and are required for trans-

298

position [8,36]. Therefore, we did not subject conserved D or E residues to mutagenesis. The remaining 15 D or E amino acids were replaced by either a lysine (K) or an arginine (R) residue (Table 1). Mutations E6K, D10K, D17K, D68K, D86K, E92K, E93K, E158K, D164K, E174K, E216K, and E321R reduced

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

TABLE 1: List of aspartic acid and glutamic acid residues in the Sleeping Beauty transposase Amino acid residue in SB transposase E6 D10 D17 D68 E69 D86 E92 E93 D140 D142 D153 E154 E158 D164 E174 D210 E216 D220 D244 D246 D260 E267 D274 E279 E284 E306 E307 E321

Status in other Tc1/mariner transposases

Change made

Nonconserved Nonconserved Nonconserved Nonconserved Conserved Nonconserved Nonconserved Nonconserved Nonconserved Nonconserved Conserved Conserved Nonconserved Nonconserved Nonconserved Conserved Nonconserved Conserved Conserved Conserved Nonconserved Conserved Conserved Conserved Conserved Conserved Conserved Nonconserved

K K K K — K K K K K — — K K K — K — — — K — — — — — — R

Whether these D and E amino acids are conserved in other Tc1-like transposases and the amino acid replacements that were introduced into the positions of the nonconserved residues are indicated. The DDE residues of the catalytic triad are shown in bold.

transposition frequency to barely measurable levels, whereas D140K and D142K reduced transposition to about 70 and 50%, respectively (Fig. 5A), indicating that these amino acids play critical roles in SB transposase activity. Importantly, however, transposition activity of D260K was about 40% higher than that of wild-type SB10 (Fig. 5A), demonstrating that an acidic-to-basic change in this position improves the function of the transposase. Our third approach to mutagenesis of the SB transposase was to introduce amino aids that naturally occur in SB or related transposases. Such an approach has been shown to be useful for the generation of hyperactive versions of SB [18]. We evaluated the effects of two amino acid changes in the transposase: R115H and R143C. The R115H substitution was made based on a comparison between SB and the Tdr1 transposase in zebrafish [32]. These two transposable elements represent closely related subfamilies of Tc1-like transposons in fish genomes and show about 80% identity in transposase sequence. Therefore, these two sequences probably represent variants of a transposase that had been selected for activity in nature. The amino acid residue

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

in position 115 in the Tdr1 transposase is a histidine, which is expected to preserve the positive charge in this position of the transposase polypeptide. The second mutant that we tested, the R143C substitution, is a naturally occurring mutation in the SB transposase, possibly generated at a mutable CpG site in the transposase gene. The R143C mutant is also called SB9 and represents a particular intermediate version of the transposase that we obtained during the reconstruction process of the SB transposase gene [2]. Both the R115H and the R143C mutants showed hyperactivity: R115H by about 60% and R143C by about 25% compared to the wild-type SB10 transposase (Fig. 5A). Next, we asked the question whether combinations of the D260K, R115H, and R143C hyperactive mutations would result in an additive or a synergistic effect. Toward this end, we engineered the three possible double mutants and a triple mutant. R115H/D260K showed a 3.7-fold, R115H/R143C a 3.2-fold, R143C/D260K a 2.6fold, and the R115H/D260K/R143C combination a 2.3fold increase in transposition activity compared to the wild-type transposase (Fig. 5B). These results indicate that the R115H mutation acts synergistically with both D260K and R143C. We sought to determine whether incorporation of the previously described T136R/M243Q/ VVA253HVR hyperactive mutations (collectively referred to as SB11) [18] would further increase transposition activity of our hypertransposing mutants. SB11 showed a 2.3-fold higher activity than wild-type transposase, and this level of activity remained unchanged when SB11 was combined with the R115H/D260K mutations (Fig. 5B). Altogether, our results demonstrate that a mutagenesis approach to the development of hyperactive transposases is viable and that the R115H/D260K mutant (hereafter referred to as SB12) is the most active SB version described to date. Combination of Hyperactive Components Results in a Further Enhancement of Sleeping Beauty Transposition Activity A reasonable expectation is that combinations of hyperactive transposases with hyperactive transposons result in a further (either additive or synergistic) increase in overall transposition frequencies. However, this was not the case in a previous study in which the SB11 hyperactive mutant was used in combination with an improved transposon vector, but was found to show no further enhancement [18]. We evaluated our SB12 double mutant in combination with two hyperactive transposon vectors. The first, pT/ zeo322, is a vector that has only a short segment of DNA outside the transposon and was previously found to show an approximately twofold increase in the efficiency of transposition [3]. This effect can probably be attributed to the improved ability of the two ends of the transposon to pair during transposition. The second is based on the T2

299

ARTICLE

element that has base pair changes in the IRs [23] and has been previously found to show an approximately threefold enhancement of transposition compared to the wildtype transposon [18]. We modified this vector to have a

doi:10.1016/j.ymthe.2003.11.024

short outer distance between the IRs and named it pT2/ zeo322. As first-generation components for reference, we used the SB10 transposase and the pT/zeo vector [3]. As found before for the SB11 hyperactive transposase [18],

FIG. 6. Effects of different transposase/transposon ratios on the efficiency of Sleeping Beauty transposition. (A) Optimal transposase to transposon ratio is dependent on the concentration of transfected DNA. 50 ng transposon plasmid (pT/neo) DNA was mixed together with increasing amounts of transposaseexpressing plasmid (pCMV/SB) DNA (50 ng, 1:1; 100 ng, 2:1; 500 ng, 10:1; 1 Ag, 20:1; 1.5 Ag, 30:1), the plasmid mixtures were cotransfected into human HeLa cells, and the transposition efficiencies were calculated (black columns). Transposition was also measured with the same series of DNA mixtures diluted 5- (gray columns) or 10-fold (striped columns) before transfection. Transposase vs transposon ratios represent ratios of amounts of plasmid DNAs. 50 ng, 10 ng, and 5 ng denote the amounts of transposon plasmid DNA per transfection. Transpositional activity measured at the 1:1 ratio for the 50 ng transfection is taken as reference and set to 100%. (B) Comparison of the transpositional efficiencies of the SB10 (wild-type) and SB12 (hyperactive) transposases at different transposase/transposon ratios. 500 ng of transposon plasmid DNA was mixed together with either an equal amount (1:1) or 2-, 10-, 24-, or 50-fold diluted transposase-expressing plasmid DNA (0.5:1, 0.1:1, 0.04:1, 0.02:1, respectively), before transfection into HeLa cells. pCMV-h was used as a filler DNA to keep the total amount of the transfected DNA constant. Transpositional activity of SB10 measured at the 1:1 ratio is taken as reference and set to 100%.

300

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

our SB12 mutant did not act in either an additive or a synergistic way together with the T2-based transposon: the overall efficiency of transposition with this combination was about twofold higher than that of the wild-type (Fig. 5C). However, a combination of the SB12 double mutant transposase with the pT/zeo322 transposon vector showed an additive effect on transposition and displayed an about fivefold enhancement in transposition compared to wild-type components (Fig. 5C). We next evaluated the effect of transiently overexpressed HMGB1 on transposition rates with the hyperactive transposon components. HMGB1 was previously found to increase SB transposition in wild-type mammalian cells about threefold [37]. In combination, HMGB1, the SB12 transposase, and the pT/zeo322 transposon showed an over eightfold increase in transposition compared to the wildtype SB system (Fig. 5C). This result demonstrates that the SB12 hyperactive transposase mutant acts in an additive manner with a hyperactive transposon and with a rate-limiting cellular host factor in mediating highefficiency transposition in human cells. Effects of Absolute and Relative Cellular Concentrations of Components of the Sleeping Beauty Transposon System on the Efficiency of Transposition Because one transposon molecule contains four transposase binding sites, the ideal molar ratio of transposase to transposon is expected to be 4:1. Indeed, cellular concentration of the transposase can be a limiting factor of transposition under certain conditions, i.e., increasing expression of the transposase results in increasing numbers of transposition events [3,18]. However, exceeding transposase expression beyond a certain threshold level can have a negative impact on transposition, an effect termed overproduction inhibition [38]. Consistent with an overproduction inhibition effect, Yant et al. [12] have found an optimal ratio of 1:25 of SB transposase-expressing plasmid to transposon donor plasmid in the mouse liver in vivo. We addressed the issue of molar ratios of the two components of the SB transposon system by measuring transpositional efficiencies in cells transfected with the transposase expression plasmid and a transposon donor plasmid in a ratio range of 1:1 to 30:1, but in three different plasmid concentrations. Fig. 6A shows that the more-transposase-more-transposition rule applies at low concentrations, but as the transposase concentration increases, an inhibitory effect is observed. No inhibitory effect was seen at the 30:1 ratio when 5 ng of transposon plasmid DNA was transfected. However, at 10 and 50 ng transposon DNA, a plateau in transpositional efficiency was found at the 20:1 and 2:1 ratios, respectively (Fig. 6A). Nevertheless, at these concentrations, transfection of a higher amount of transposase-expressing plasmid than transposon plasmid was necessary for optimal transposition (Fig. 6A). However, increasing the amount of trans-

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

fected transposon DNA to 500 ng resulted in a shift in optimal plasmid ratios (Fig. 6B), indicating that high transposon concentrations cannot be matched by high transposase concentrations, possibly due to overproduction inhibition. Under these conditions, dilution of the transposase plasmid up to 10-fold resulted in more transposition events, whereas the transposase became limiting upon further dilution (Fig. 6B). Consistent with earlier findings [18], our observations imply that the optimal plasmid ratios are primarily dependent on how much DNA is introduced into the cell. Next, we compared the SB12 hyperactive mutant to wild-type SB10 in terms of sensitivity to overproduction inhibition (Fig. 6B). Similar to SB10, transposition by SB12 plateaued at a 1:10 ratio of transposase plasmid to transposon plasmid (Fig. 6B). This result indicates that the hyperactive phenotype of SB12 is unlikely to be due to reduced overproduction inhibition. Importantly, SB12 displayed hyperactivity at all of the plasmid ratios tested (Fig. 6B). These data suggest that regulation of the transposase by overproduction inhibition sets a limit to transposition in any given cell and that transposase-to-transposon plasmid ratios need to be carefully optimized for any given tissue, either in vitro or in vivo.

DISCUSSION In this paper, we report a significant enhancement of the transpositional activity of the Sleeping Beauty transposable element, a promising vector for nonviral gene transfer in vertebrate species. The transposon has two functional components: the DNA sequences within the terminal inverted repeats of the element and the transposase protein. Improvements in the overall activity of the transposon can be made by manipulating either or both of these components. Accordingly, we took two experimental approaches: (1) modify the structure of the substrate transposon by flanking large transgenes with complete SB elements, thereby mimicking naturally occurring composite elements; (2) generate hypertransposing mutants of the transposase by replacing some of the nonconserved amino acids. The Sandwich Sleeping Beauty Element Improves Transposition of Long Transgenes We have shown that the sandwich element has an enhanced capacity to transpose long transgenes (Fig. 3B). The structure of the sandwich transposon is somewhat similar to that of the bacterial transposons Tn5 and Tn10. These elements might have been fortuitously generated by transposition of two insertion sequence elements on both sides of an immobile segment containing antibiotic resistance genes [39,40]. This situation can also arise, probably by chance, in other transposition systems, resulting in new, composite, mobile elements. Indeed, a

301

ARTICLE

pair of Paris elements that flank a nonrepetitive sequence of more than 10 kb in an inverted orientation was shown to be able to transpose in D. virilis [25]. Why does the sandwich vector transpose long transgenes better than the wild-type SB transposon? We have shown earlier that (1) long elements tend to transpose less efficiently than short ones, likely because the ends of long elements cannot pair easily during synaptic complex formation [3], and (2) the DNA-bending protein HMGB1 plays an important role in SB transposition likely by aiding the pairing of transposon ends [37]. Thus, we suggest that an increase in the number of transposase binding sites (from four to eight) in the sandwich transposon can partially rescue synaptic complex formation of long elements, presumably due to the more pronounced action of transposase – transposase interactions and HMGB1 at the transposon inverted repeats. Artificially made, sandwich-like Mos1 mariner elements similar to the ones described here for SB have been found to have increased mobility in Drosophila [41], suggesting common underlying mechanisms in composite transposon mobilization in the Tc1/mariner family. Mutational Analysis of the Sleeping Beauty Transposase and Isolation of Hyperactive Mutants By using a limited site-directed mutagenesis screen, we identified hyperactive versions of the SB transposase (Fig. 5A). Three different approaches were undertaken for the choice of induced mutations: (1) modification of a linker region that separates the DNA-binding and the catalytic domains, (2) systematic replacement of acidic residues with basic amino acids, and (3) substitution for amino acids that had been selected in nature. The predicted helix spanning the region between the N-terminal DNA-binding domain and the catalytic domain is conserved in the Tc1 family (Fig. 4B) and is not part of the DNA-binding domain [2]. When present in recombinant transposase preparations, this region interferes with binding to the transposon inverted repeats (Fig. 4A). We reasoned that disruption of the local structure at this region might result in a conformational change of the transposase that is more favorable for DNA binding. Proline is a secondary structure breaker in proteins and it has been widely used to modify the conformation of proteins. A hyperactive Tn5 transposase mutant was generated by introducing a proline residue that is thought to interrupt the interference between the N-terminal and the C-terminal regions of the transposase [35]. Unfortunately, the mutants that we generated all showed severely impaired transposition (Fig. 5A), indicating the functional importance of this helix in SB transposition. Future work should be directed to random mutagenesis of the helix in the hope that less drastic changes could result in hyperactive phenotypes.

302

doi:10.1016/j.ymthe.2003.11.024

The rationale behind the change of all nonconserved acidic amino acid residues to basic amino acids is that in several transposition systems, including Tn5 [26,28], Tn10 [30]. and Himar1 [31], some hyperactive mutations fall into this class. Acidic-to-basic amino acid changes might eliminate (or at least reduce) the unfavorable charge – charge interaction between the acidic amino acid residues and the negatively charged phosphate backbone of the transposon (or target) DNA [28] or might overcome the self-inhibiting properties of transposase [26]. Most of the mutations that we introduced into the SB transposase resulted in a decrease in the efficiency of transposition (Fig. 5A), suggesting very little functional redundancy in the transposase sequence. A marked sensitivity of transposase to mutations was previously noted for the Mos1 mariner element [42]. Nevertheless, one of the substitutions, the D260K mutation, produced a hyperactive phenotype (Fig. 5A). The aspartic acid in position 260 is either lysine or arginine in other Tc1-like transposases (Fig. 7), suggesting that lysine and arginine can better function in that sequence context. It is possible that a particular version of fish Tc1-like transposases did contain K or R at position 260, but this amino acid got replaced at some point during transposase evolution, because it is functionally nonessential for the transposase. It is important to note that four hyperactive mutations reported from Tc1/mariner elements are located within the same 8-amino-acid segment in the catalytic domains of these transposases (Fig. 7). The D260K mutation (this work) and the V253H and A255R mutations [18] in Sleeping Beauty, and the H267R mutation in Himar1 [31], all map to the same region just preceding the E/D residue of the DDE (DDD in mariner elements) catalytic triad (Fig. 7). It is therefore possible that these mutations result in a slight conformational change that is more favorable for catalysis. Because three of these four mutations are K and R replacements (Fig. 7), the local shift to positive charge might enhance target DNA capture, a function likely encoded in the catalytic domain [43,44]. Further biochemical work will be required to substantiate either of these hypotheses. The D260K mutation acts synergistically with two other, naturally occurring mutations, R115H and R143C (Fig. 5B). The R115H/D260K and R115H/R143C double mutants exhibited about 3.7- and 3.2-fold increase in transposition activity over wild-type transposase, respectively (Fig. 5B). It seems that the R115H mutation is important in the double mutants since the R143C/D260K mutant showed only about 2.6-fold increase in activity (Fig. 5B). We found that hyperactivity of the R115H/D260K mutant (referred to as SB12 by keeping the convention of naming versions of the SB transposase) cannot be attributed to a reduction in overproduction inhibition (Fig. 6B). Importantly, SB12 displayed additive effects with a hyperactive transposon vector (Fig. 5C). Transposition of this hyperactive system

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

ARTICLE

doi:10.1016/j.ymthe.2003.11.024

FIG. 7. Locations of hyperactive mutations. Amino acid alignment of transposase segments of Tc1-like transposons and that of the Himar1 mariner element highlights a region where several hyperactive mutations are located. The alignment shows that D260 of SB is not conserved: several Tc1-like transposases contain lysine or arginine in this position. The E residue (D in Himar1) of the DDE catalytic triad is indicated.

is further enhanced by transient overexpression of HMGB1, a cellular cofactor of SB transposition [37] (Fig. 5C). The collective effect of these components is an approximately eightfold increase in transposition, compared to the first-generation SB system (Fig. 5C). In addition to relative efficiencies of transposition, it is useful to calculate the absolute numbers of transformant cells that can be generated with transposition in a given transfected cell population. For this, we cotransfected a GFP expression construct together with the transposon components into human HeLa cells and normalized transposition rates with transfection efficiencies, based on GFP fluorescence (data not shown). We found that about 2% of cells that had taken up transposon DNA will undergo a transposition event using the first-generation SB system, whereas stable transgenesis rates are about 10 – 15% using combinations of the hyperactive transposon/transposase components. Transposition can be further optimized by systematic adjustment of transposase and transposon concentrations in transfected cells, because different ratios of the transposase expression and transposon donor plasmids can greatly influence transposition efficiencies (Fig. 6). Although Sleeping Beauty is an element reconstructed from transposon fossils, it most likely represents a transposon that was once active in fish genomes. The fact that hyperactive versions of the SB transposase can be generated implies that this transposase has not been selected for the highest possible activity in nature. This is because transposition can potentially endanger the survival of the host organism and, consequently, that of the transposable element. The overall activity of SB for in vivo gene transfer can be improved by matching the transposon with highly efficient DNA delivery technologies and by increasing its activity in the cell. In this paper we provid-

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy

ed evidence that the latter strategy is clearly promising for the development of more efficient transposon-based gene vectors.

MATERIALS AND METHODS Plasmids. The CA ! GC change in the right IR was induced in pT and pT/neo plasmids using the primers FTC-3 [2] and SB-SAND (5VTTGCATCTAGAGCGGCCGCGTTGAAGTCGGAAGTTTACATACACCTTTAGCC-3V) for pT and PR-Neo (5V-CCTTGCGCAGCTGTGCTCGACG3V) and SBSAND for pT/neo. The SB-SAND primer introduces XbaI (bold italic) and NotI (bold) restriction sites. The mutated rIR was subcloned in pT and pT/neo to yield pT* and pT*/neo. pEGFPC1 (Clontech) was linearized with AseI, Klenow-filled, and ligated with NotI-digested, Klenow-filled pT*, yielding the cloning intermediate pT*pEGFPC1. The same fragment was also cloned into the BlpI site in pT to yield pT6.2. The complete T* transposon was moved in an EcoRI/XbaI fragment into pUC18 and subsequently cloned in an XbaI/ScaI fragment into pT*pEGFPC1 to yield the sandwich transposon pT/SA7.7, which has the 4.7kb EGFPC1 fragment between two T* transposons in an inverted orientation. pT/SA7.7 was linearized with BglII, Klenowfilled, and ligated with a 4.5kb PstI fragment of pCMVh (Clontech) containing a CMVpromoterdriven lacZ gene, to yield pT/SA12.2. Site-directed mutagenesis of the transposase gene was done by PCR using pCMV-SB as a template. Primer sequences are available from the authors upon request. All mutations were confirmed by sequencing. pT/neo deletion derivatives were generated by PCR using primers that generate an SpeI site. Primers for pT/neo-SpeI were 5V-AGTCCTTGAAATACATCCACAGGTACAG-3Vand 5V-AGTCTGTGGAATGTGTGTCAGTTAGG3V. Primers for pT/neo-SpeI/D112 were 5V-AGTCTGTGG-AATGTGTGTCAGTTAGG-3V and 5V-AGTTTAAAGGCACAGTCAACTTAGTG-3V. Primers for pT/neo-SpeI/D46 were 5V-AGTTTAAAGGCACAGTCAACTTAGTG-3Vand 5V-AGTCTGATA-GACTTATTGACATCATTTGAG-3V. Primers for pT/neoSpeI/D66 were 5V-AGTCTGTGGAATGTGTGTCAGTTAGG-3V and 5V-AGTAAGCTTCTAAAGCCATGACATCATTTTC-3V. Primers for pT/neoSpeI/D112-mut were 5V-AGTTTAAAGGCACAGTCAACTTAGTG-3V and 5V-AGTCACGTTCATGAGTCAACTTAGTGTATGTAAAC-3V. The PCR products were circularized after T4-kinase treatment by ligation and transformed in Escherichia coli (DH5a). Clones were verified by SpeI digestion. Bacterial expression vectors were made in pET21a (Novagen), encoding a C-terminal histidine tag. pET21a/N-123 was described earlier [2]. pET21a/

303

ARTICLE

N-161 was made by cutting a pET21a-derived plasmid containing the fulllength transposase gene with MscI/NotI and recircularizing the plasmid. Cell culture and transfections. HeLa cells were maintained in DMEM containing 10% fetal bovine serum. Transposition assays were done as described [2]. If not stated otherwise, 105 cells were transfected with 90 ng each of the transposon donor plasmid and the transposase-expressing helper plasmid using Fugene6 transfection reagent (Roche). Two days posttransfection, the cells were replated and selected in 1.4 mg/ml G418. After 2 weeks of selection, the resistant colonies were stained and counted. Recombinant protein expression and purification. Induction of Histagged, recombinant protein expression was in E. coli strain BL21(DE3) (Novagen) by the addition of 0.4 mM IPTG at 0.5 OD at 600 nm and continued for 2.5 h at 30jC. Cells were sonicated in H-buffer [25 mM Hepes (pH 7.5), 15% glycerol, 0.25% Tween 20] containing 2 mM hmercaptoethanol, 1 M NaCl, and 1 Complete (Boehringer Mannheim), and 20 mM imidazole (pH 8.0) was added to the soluble fraction before it was mixed with Ni2+ – NTA resin (Qiagen) according to the recommendations of the manufacturer. The resin was washed with sonication buffer containing 30% glycerol and 50 mM imidazole, and bound proteins were eluted with sonication buffer containing 300 mM imidazole and dialyzed overnight at 4jC against sonication buffer without imidazole. Electrophoretic mobility shift assay. The probes containing either the left or the right IR of SB were end-labeled using [a-32P]dATP and Klenow. Nucleoprotein complexes were formed in 20 mM Hepes (pH 7.5), 0.1 mM EDTA, 0.1 mg/ml BSA, 150 mM NaCl, 1 mM DTT in a total volume of 10 Al. Reactions contained labeled probe, 1 Ag poly(dI – dC), 100 pg labeled fragment, and 1 pmol protein. After 15 min incubation on ice, 5 Al of loading dye containing 50% glycerol and bromophenol blue was added, and the samples were loaded onto a 4% polyacrylamide gel.

ACKNOWLEDGMENTS We thank E. Stu¨we and A. Katzer for their technical assistance and members of the Transposition Group at the MDC for critically reading the manuscript. This work was supported by EU Grant QLG2-CT-2000-00821. RECEIVED FOR PUBLICATION SEPTEMBER 23, 2003; ACCEPTED NOVEMBER 29, 2003.

REFERENCES 1. Verma, I. M., and Somia, N. (1997). Gene therapy—promises, problems and prospects. Nature 389: 239 – 242. 2. Ivics, Z., Hackett, P. B., Plasterk, R. H., and Izsva´k, Z. (1997). Molecular reconstruction of Sleeping Beauty, a Tc1-like transposon from fish, and its transposition in human cells. Cell 91: 501 – 510. 3. Izsva´k, Z., Ivics, Z., and Plasterk, R. H. (2000). Sleeping Beauty, a wide host-range transposon vector for genetic transformation in vertebrates. J. Mol. Biol. 302: 93 – 102. 4. van Luenen, H. G., Colloms, S. D., and Plasterk, R. H. (1994). The mechanism of transposition of Tc3 in C. elegans. Cell 79: 293 – 301. 5. Cooley, L., Kelley, R., and Spradling, A. (1988). Insertional mutagenesis of the Drosophila genome with single P elements. Science 239: 1121 – 1128. 6. Ivics, Z., Izsva´k, Z., Minter, A., and Hackett, P. B. (1996). Identification of functional domains and evolution of Tc1-like transposable elements. Proc. Natl. Acad. Sci. USA 93: 5008 – 5013. 7. Lohe, A. R., Moriyama, E. N., Lidholm, D. A., and Hartl, D. L. (1995). Horizontal transmission, vertical inactivation, and stochastic loss of mariner-like transposable elements. Mol. Biol. Evol. 12: 62 – 72. 8. Plasterk, R. H., Izsva´k, Z., and Ivics, Z. (1999). Resident aliens: the Tc1/mariner superfamily of transposable elements. Trends Genet. 15: 326 – 332. 9. Hartl, D. L., Lozovskaya, E. R., Nurminsky, D. I., and Lohe, A. R. (1997). What restricts the activity of mariner-like transposable elements? Trends Genet 13: 197 – 201. 10. Miskey, C., Izsvk, Z., Plasterk, R. H., and Ivics, Z. (2003). The Frog Prince: a reconstructed transposon from Rana pipiens with high transpositional activity in vertebrate cells. Nucleic Acids Res. 31: 6873 – 6881. 11. Luo, G., Ivics, Z., Izsva´k, Z., and Bradley, A. (1998). Chromosomal transposition of a Tc1/mariner-like element in mouse embryonic stem cells. Proc. Natl. Acad. Sci. USA 95: 10769 – 10773. 12. Yant, S. R., et al. (2000). Somatic integration and long-term transgene expression in normal and haemophilic mice using a DNA transposon system. Nat. Genet. 25: 35 – 41.

304

doi:10.1016/j.ymthe.2003.11.024

13. Fischer, S. E., Wienholds, E., and Plasterk, R. H. (2001). Regulated transposition of a fish transposon in the mouse germ line. Proc. Natl. Acad. Sci. USA 98: 6759 – 6764. 14. Dupuy, A. J., Fritz, S., and Largaespada, D. A. (2001). Transposition and gene disruption in the male germline of the mouse. Genesis 30: 82 – 88. 15. Horie, K., et al. (2001). Efficient chromosomal transposition of a Tc1/mariner-like transposon Sleeping Beauty in mice. Proc. Natl. Acad. Sci. USA 98: 9191 – 9196. 16. Carlson, C. M., et al. (2003). Transposon mutagenesis of the mouse germline. Genetics 165: 243 – 256. 17. Horie, K., et al. (2003). Characterization of Sleeping Beauty transposition and its application to genetic screening in mice. Mol. Cell. Biol. 23: 9189 – 9207. 18. Geurts, A. M., et al. (2003). Gene transfer into genomes of human cells by the Sleeping Beauty transposon system. Mol. Ther. 8: 108 – 117. 19. Yant, S. R., et al. (2002). Transposition from a gutless adeno-transposon vector stabilizes transgene expression in vivo. Nat. Biotechnol. 20: 999 – 1005. 20. Montini, E., et al. (2002). In vivo correction of murine tyrosinemia type I by DNAmediated transposition. Mol. Ther. 6: 759 – 769. 21. Ortiz-Urda, S., et al. (2003). Sustainable correction of junctional epidermolysis bullosa via transposon-mediated nonviral gene transfer. Gene Ther. 10: 1099 – 1104. 22. Belur, L. R., et al. (2003). Gene insertion and long-term expression in lung mediated by the Sleeping Beauty transposon system. Mol. Ther. 8: 501 – 507. 23. Cui, Z., Geurts, A. M., Liu, G., Kaufman, C. D., and Hackett, P. B. (2002). Structure – function analysis of the inverted terminal repeats of the Sleeping Beauty transposon. J. Mol. Biol. 318: 1221 – 1235. 24. Karsi, A., Moav, B., Hackett, P., and Liu, Z. (2001). Effects of insert size on transposition efficiency of the Sleeping Beauty transposon in mouse cells. Mar. Biotechnol. 3: 241 – 245. 25. Petrov, D. A., Schutzman, J. L., Hartl, D. L., and Lozovskaya, E. R. (1995). Diverse transposable elements are mobilized in hybrid dysgenesis in Drosophila virilis. Proc. Natl. Acad. Sci. USA 92: 8050 – 8054. 26. Wiegand, T. W., and Reznikoff, W. S. (1992). Characterization of two hypertransposing Tn5 mutants. J. Bacteriol. 174: 1229 – 1239. 27. Weinreich, M. D., Gasch, A., and Reznikoff, W. S. (1994). Evidence that the cis preference of the Tn5 transposase is caused by nonproductive multimerization. Genes Dev. 8: 2363 – 2374. 28. Zhou, M., and Reznikoff, W. S. (1997). Tn5 transposase mutants that alter DNA binding specificity. J. Mol. Biol. 271: 362 – 373. 29. Goryshin, I. Y., and Reznikoff, W. S. (1998). Tn5 in vitro transposition. J. Biol. Chem. 273: 7367 – 7374. 30. Sakai, J., and Kleckner, N. (1996). Two classes of Tn10 transposase mutants that suppress mutations in the Tn10 terminal inverted repeat. Genetics 144: 861 – 870. 31. Lampe, D. J., Akerley, B. J., Rubin, E. J., Mekalanos, J. J., and Robertson, H. M. (1999). Hyperactive transposase mutants of the Himar1 mariner transposon. Proc. Natl. Acad. Sci. USA 96: 11428 – 11433. 32. Izsva´k, Z., Ivics, Z., and Hackett, P. B. (1995). Characterization of a Tc1-like transposable element in zebrafish (Danio rerio). Mol. Gen. Genet. 247: 312 – 322. 33. Mikkelsen, J. G., et al. (2003). Helper-independent Sleeping Beauty transposon – transposase vectors for efficient nonviral gene delivery and persistent gene expression in vivo. Mol. Ther. 8: 654 – 665. 34. Vos, J. C., De Baere, I., and Plasterk, R. H. (1996). Transposase is the only nematode protein required for in vitro transposition of Tc1. Genes Dev. 10: 755 – 761. 35. Davies, D. R., Goryshin, I. Y., Reznikoff, W. S., and Rayment, I. (2000). Three-dimensional structure of the Tn5 synaptic complex transposition intermediate. Science 289: 77 – 85. 36. Doak, T. G., Doerder, F. P., Jahn, C. L., and Herrick, G. (1994). A proposed superfamily of transposase genes: transposon-like elements in ciliated protozoa and a common ‘‘D35E’’ motif. Proc. Natl. Acad. Sci. USA 91: 942 – 946. 37. Zayed, H., Izsva´k, Z., Khare, D., Heinemann, U., and Ivics, Z. (2003). The DNA-bending protein HMGB1 is a cellular cofactor of Sleeping Beauty transposition. Nucleic Acids Res. 31: 2313 – 2322. 38. Lohe, A. R., and Hartl, D. L. (1996). Autoregulation of mariner transposase activity by overproduction and dominant-negative complementation. Mol. Biol. Evol. 13: 549 – 555. 39. Berg, D. E., Berg, C. M., Sasakawa, C., Zhou, M., and Reznikoff, W. S. (1984). Bacterial transposon Tn5: evolutionary inferences. Mol. Biol. Evol. 1: 411 – 422. 40. Kleckner, N., Chalmers, R. M., Kwon, D., Sakai, J., and Bolland, S. (1996). Tn10 and IS10 transposition and chromosome rearrangements: mechanism and regulation in vivo and in vitro. Curr. Top. Microbiol. Immunol. 204: 49 – 82. 41. Lozovsky, E. R., Nurminsky, D., Wimmer, E. A., and Hartl, D. L. (2002). Unexpected stability of mariner transgenes in Drosophila. Genetics 160: 527 – 535. 42. Lohe, A. R., De Aguiar, D., and Hartl, D. L. (1997). Mutations in the mariner transposase: the D,D(35)E consensus sequence is nonfunctional. Proc. Natl. Acad. Sci. USA 94: 1293 – 1297. 43. Katzman, M., and Sudol, M. (1995). Mapping domains of retroviral integrase responsible for viral DNA specificity and target site selection by analysis of chimeras between human immunodeficiency virus type 1 and visna virus integrases. J. Virol. 69: 5687 – 5696. 44. Junop, M. S., and Haniford, D. B. (1997). Factors responsible for target site selection in Tn10 transposition: a role for the DDE motif in target DNA capture. EMBO J. 16: 2646 – 2655.

MOLECULAR THERAPY Vol. 9, No. 2, February 2004 Copyright D The American Society of Gene Therapy