A Bacterial Genetic Screen Identifies Functional Coding Sequences of the Insect mariner Transposable Element Famar1 Amplified From the Genome of the Earwig, Forficula auricularia

Corresponding author: David J. Lampe, Duquesne University, 600 Forbes Ave., Pittsburgh, PA 15282., lampe{at}duq.edu (E-mail)

Communicating editor: M. J. SIMMONS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Transposons of the mariner family are widespread in animal genomes and have apparently infected them by horizontal transfer. Most species carry only old defective copies of particular mariner transposons that have diverged greatly from their active horizontally transferred ancestor, while a few contain young, very similar, and active copies. We report here the use of a whole-genome screen in bacteria to isolate somewhat diverged Famar1 copies from the European earwig, Forficula auricularia, that encode functional transposases. Functional and nonfunctional coding sequences of Famar1 and nonfunctional copies of Ammar1 from the European honey bee, Apis mellifera, were sequenced to examine their molecular evolution. No selection for sequence conservation was detected in any clade of a tree derived from these sequences, not even on branches leading to functional copies. This agrees with the current model for mariner transposon evolution that expects neutral evolution within particular hosts, with selection for function occurring only upon horizontal transfer to a new host. Our results further suggest that mariners are not finely tuned genetic entities and that a greater amount of sequence diversification than had previously been appreciated can occur in functional copies in a single host lineage. Finally, this method of isolating active copies can be used to isolate other novel active transposons without resorting to reconstruction of ancestral sequences.

MARINER transposable elements are a large and diverse family of small eukaryotic transposons. These elements are 1.3 kb in length and encode a single protein, the mariner transposase, that allows them to mobilize their DNA in virtually any eukaryotic genetic background through a cut-and-paste mechanism. When expressed with the appropriate promoters, these elements are active in Bacteria and Archaea as well (RUBIN et al. 1999 ; ZHANG et al. 2000 ).

In Eukarya, organisms as diverse as Hydra and humans contain mariners and often more than one type (where "type" refers to all copies descended from a single horizontally transferred ancestral element in a population of host organisms). The human genome, for example, contains two distinct types of mariner elements (ROBERTSON and MARTOS 1997 ; ROBERTSON and ZUMPANO 1997 ) while that of Caenorhabditis elegans contains nine (WITHERSPOON and ROBERTSON 2003 ). When these and similar elements are isolated through PCR, screens of genomic libraries, or genomic sequencing projects, their transposase genes are almost always found to be defective in some way, due to the presence of either numerous indels or in-frame stop codons that would inactivate them. Thus, although we know of many hundreds of different inactive mariners through their sequences, only two have been demonstrated to be functional. The first of these is Mos1, a particularly active copy of the original mariner element isolated from Drosophila mauritiana (MEDHORA et al. 1991 ). The second is Himar1, reconstructed as a consensus from many copies of a type of mariner element found in the horn fly, Haematobia irritans (LAMPE et al. 1996 ). Both of these elements have found wide use as genetic tools in a broad variety of species in all domains of life (LOHE and HARTL 1996B ; GUEIROS-FILHO and BEVERLEY 1997 ; COATES et al. 1998 , COATES et al. 2000 ; RUBIN et al. 1999 ; ZHANG et al. 2000 ).

Despite our knowledge of many different mariners, we do not yet fully understand how they evolve or persist as functional genetic entities. Any model of their evolution must explain their activity, distribution, and diversity. A general model for mariner evolution has been developed and modified by several research groups (ROBERTSON and LAMPE 1995 ; HARTL et al. 1997C ; LAMPE et al. 2003 ). This model posits that a single copy of a mariner transposon invades the germline of an organism by an unknown mechanism, perhaps by hitchhiking on viruses (HARTL et al. 1997A ). This founder element increases in copy number through transposition and spreads through the population via sexual reproduction. As copy number increases, the rate of transposition may slow, as intrinsic (e.g., "overproduction inhibition"), emergent (e.g., dominant-negative inhibition by defective transposases; HARTL et al. 1997C ), or host-mediated (e.g., transcriptional repression of repeated sequences) regulatory mechanisms take effect or evolve de novo. Random mutations accumulate in all element lineages, and since the transposase produced by functional elements acts at random on any recognizable copies in the same genome, there is no effective selection against nonfunctional yet recognizable copies. Eventually, mutations inactivate all copies, and it is this final deteriorated state in which we find most mariner sequences.

Only by undergoing repeated horizontal transfer can some mariners escape this fate and stay ahead of mutation, drift, and emergent regulatory mechanisms. These horizontal transfers are selection events, since only active elements will be able to establish new lineages in new populations (ROBERTSON and LAMPE 1995 ; WITHERSPOON 1999 ; SILVA and KIDWELL 2000 ; LAMPE et al. 2003 ). Nonsense or missense mutations in typical genes are immediately tested by natural selection, but several such mutations may accumulate with no consequence in the transposase gene of a mariner lineage before being tested by the next horizontal transfer event. The divergence between successive founder elements could be significant, so this process of drift interspersed with selection events may accelerate the generation of mariner diversity (LAMPE et al. 2001 ).

Some organisms contain mariners that, although diverged in sequence from their putative ancestral element, may still be active because at least some copies appear intact (GOMULSKI et al. 1997 ; LAMPE et al. 2003 ; WITHERSPOON and ROBERTSON 2003 ). The European earwig, Forficula auricularia, a common household and garden insect in North America, is one such organism (LAMPE et al. 2003 ). F. auricularia contains many copies of Famar1, an element in the mellifera subfamily of mariners and very closely related (98% amino acid identity for their ancestral sequences) to the Ammar1 of the honeybee, Apis mellifera. Individual copies of Famar1 differ from each other on average by 2.3% at the amino acid level (LAMPE et al. 2003 ). Although some of these copies appear intact, it is not obvious which are active and which contain inactivating point mutations. This situation is ideal for studying the later steps of the evolution of a mariner transposon in a host population, because it represents neither a very recent colonization where all of the copies are nearly identical (e.g., CAPY et al. 1992 ; GARCIA-FERNANDEZ et al. 1995 ) nor a terminal stage in which all copies are inactive.

We report here the application of bacterial genetic methods to discriminate between functional and nonfunctional coding sequences of Famar1 obtained via genomic PCR. By sequence analyses we find that the transposase genes of Famar1, including those still encoding a functional transposase, evolve under no selection for the conservation of their function (i.e., neutrally) within F. auricularia.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Media and antibiotics:
Bacteria were grown at the temperatures indicated in Luria broth (LB) or on agar plates as described (SAMBROOK et al. 1989 ). Papillation assays were performed on thick (50 ml) MacConkey lactose agar plates. Antibiotic concentrations were ampicillin (Amp), 100 µg; gentamicin (Gen), 10 µg; tetracycline (Tet), 15 µg; naladixic acid (Nal), 20 µg; chloramphenicol (Cam), 34 µg; streptomycin (Str), 75 µg; and apramycin (Apr), 80 µg/ml, respectively, unless otherwise noted.

Plasmids and bacterial strains:
Plasmids and bacterial strains used in this study are listed and described briefly in Table 1. A Famar1 "minimariner" with an open reading frame (ORF) through one inverted terminal repeat (ITR) was constructed by amplifying pminiAm (LAMPE et al. 2001 ) with the primers T7 (5'-TAATACGACTCACTATAGGG) and AmORF-R2 (5'-CAGATCTGAGTGAAATCCGCAATTAC) in one PCR and the primers SP6 (5'-CGATTTAGGTGACACTATAG) and Am1212f (5'-AAATGGGCAACACATTACAGAA) in another PCR. The PCR products were kinased and ligated together as described previously (LAMPE et al. 2001 ). The ligation reaction was used in another PCR using the primers T7 and SP6 to recover a product that had two ITRs that were distinct. This product was cut with the restriction enzymes XhoI and HindIII and cloned into the identical sites of pCDNAII, resulting in the plasmid pFaORF.

To make a Famar1 plasmid capable of mediating papillation in Escherichia coli, pFaORF was cut with BglII and the BamHI/BglII fragment of pRZ1495 containing the lacZYA and Tet^R genes ligated to it. This ligation creates a fusion of the open reading frame in one ITR of FaORF and lacZ from pRZ1495 at amino acid coding position 8 of the lacZ gene. This fusion created pFaLacTet.

A single-copy plasmid containing the lacZ/FaORF fusion was created in two steps. First, pFaLacTet was digested with NotI, the overhanging ends made blunt with the Klenow fragment of E. coli DNA polymerase I, and the DNA cut again with BamHI. This fragment was ligated to the plasmid pACMarKan that had been digested with BstEII, made blunt as above, and digested again with BglII. The resulting plasmid, pACFaLacTet, contains the FaLacTet papillation element inside the ITRs of the Himar1 transposon on a p15a origin of replication. Himar1 carrying FaLacTet was mobilized onto a matable F plasmid by cotransformation of RZ212 E. coli with pACFaLacTet and pBADC9. The cells were selected with Amp/Tet/Gen. The transposase encoded by pBADC9 can mobilize the Himar1 carrying FaLacTet off pACFaLacTet and onto the F plasmid, which is matable. These cells were mated to DH5 E. coli and recipient cells containing an insertion of Himar1 on the F selected with Nal/Tet. The resulting strain, DL13, was used in a papillation assay to detect functional and nonfunctional coding sequences of Famar1 and Ammar1.

Mating-out strains of E. coli for both Famar1 and Himar1 were constructed as follows. The Apr^R cassette from pOJ427 was inserted as a BamHI/XbaI blunt fragment into the BglII site of pFaORF to make pEB17. The BamHI restriction fragment from pEB17 containing the ITRs and Apr^R was moved into pACYC184 at the BamHI site to make pEB19. EB8 E. coli cells conferring Gen^R, Cam^R, and Apr^R were produced by transforming pEB19 into electrocompetent RZ212 E. coli containing pOX38-Gen (F) (JOHNSON and REZNIKOFF 1984 ).

Identical versions of the above plasmids were constructed for Himar1. The Apr^R cassette was inserted into a blunt BglII site of pmmOrf (LAMPE et al. 1999 ) as above to produce pEB18, and the ITRs and Apr^R were moved onto pACYC184 by inserting a BamHI fragment of pEB18 into the BamHI site to make pEB20. The E. coli strain EB9 was produced by transforming pEB20 into electrocompetent RZ212 cells containing pOX38-Gen (F') as described above.

An E. coli genetic screen for Famar1 sequences encoding functional transposase:
The genome of F. auricularia was screened for the presence of functional and nonfunctional coding sequences of Famar1 using the PCR and a Famar1-specific E. coli papillation screen similar to the one developed for Himar1 (LAMPE et al. 1999 ). Copies of Famar1 were amplified from genomic F. auricularia DNA using the primers Famar7F-2 (5'-TTGAACCATGGAAAATCAAAAGGAAC) and Famarstop (5'-TTCTGCAGTCATGGAACTAAATAACTTTA) based on the consensus sequence of Famar1 (LAMPE et al. 2003 ). Copies of Ammar1 were amplified from the A. mellifera genome using the primers Am24cod-f (5'-GTTGAACCATGGAAAATCAAAAGGAACATTA) and Am349R(Pst) (5'-TTCTGCAGAATAACTTTATTCTGTAATGTG). The forward primers extend into the coding sequences by seven and eight codons, respectively, due to the necessity of introducing a NcoI restriction site at the start codon, so we were unable to determine if these first few amino acids varied between copies. Amplified coding sequences were cloned into the NcoI and PstI sites of pBAD24 (GUZMAN et al. 1995 ). This ligation was transformed into electrocompetent DL13 E. coli cells, plated onto MacConkey lactose agar plates containing Amp and Tet, and grown at 32° for no more than 4 days. Papillation normally occurs after 52 hr under these conditions for Himar1 (LAMPE et al. 1999 ). The total number of colonies was recorded as were the number of colonies that were papillating. Papillating colonies were deemed to contain functional copies of the Famar1 coding sequence. Plasmids from papillating and nonpapillating colonies were sequenced. An identical procedure was used to screen the genome of A. mellifera to search for functional and nonfunctional coding sequences of the closely related Ammar1 transposon (ROBERTSON and MACLEOD 1993 ).

Quantification of relative transpositional activity of copies of Famar1:
The relative activity of each functional Famar1 coding sequence was measured by using a Famar1-specific mating-out assay, similar to one developed for Himar1 transposase (LAMPE et al. 1999 ). Each individual functional Famar1 coding sequence in pBAD24 was transformed into the Apr^R/Gen^R/Str^R E. coli strain EB8 (Table 1), and for comparison HimarI was transformed into the Apr^R/Gen^R/Str^R E. coli strain EB9. The cells were plated on Gen, Amp, and Apr LB-agar plates. Three colonies from each plate were then grown in culture overnight using the same antibiotics. The next day, 10 µl of these cells was mixed with 30 µl of an overnight culture of RZ221 E. coli cells in 1 ml of LB broth for mating. The cells were grown at 37° with slow rotation on a roller drum for 6 hr. Suitable dilutions of matings were plated on Nal-Gen plates to select total exconjugates and Apr-Gen to select exconjugate transposition products.

The proportion of all exconjugant bacteria (Gen^R colonies, see above) whose F plasmids carried transposition products was used to estimate the activity of each mariner copy. This "Apr^R/Gen^R" ratio was measured at least three times for each Famar1 assayed. To control for experimental variations, the Himar1 transposase was also assayed in triplicate alongside each set of Famar1 experiments. The ratios were log₁₀ transformed to normalize their distribution (the variance of the ratios increases with their magnitude; not shown). The average of the appropriate Himar1 control measurements was subtracted from the Famar1 measurements to yield activity values relative to Himar1. The average of the transformed, then normalized measurements was used as the estimated relative activity for each Famar1 copy.

The activity of some Famar1 copies was measured in more than one experiment, and some separately cloned copies (Fa6, Fa7, and Fa8) were identical in sequence. To simplify analysis and avoid incorrectly inflating the number of independent measures, only one set of measurements (the first) was used for each unique (by sequence) Famar1. The choice of sets does not significantly affect any results (not shown.)

Statistical analysis to determine whether Famar1 transposase activity is affected as amino acid changes accumulate is problematic. Due to phylogenetic structure, some Famar1 copies have accumulated nearly identical sets of mutations and so cannot be treated as independent trials. For each such set of nearly identical Famar1, only one member of the set was used in the analysis. This eliminates much of the nonindependence, although some phylogenetic correlations and homoplasies remain. A Spearman rank regression test (ZAR 1999 ) was then applied to the reduced data set to detect any correlation of activity with amino acid divergence from the ancestral sequence. The test was applied to all possible reduced data sets; set choice does not affect the results.

To estimate the statistical power of this method, new activity values were generated for each observed divergence value by either (a) resampling the observed Famar1 activities or (b) generating random activity values with the observed variances. Linear trends of varying size were then added to the simulated data sets and the above testing method was applied. This procedure was repeated over all possible combinations of eliminated Famar1 copies (as above). To produce a conservative estimate of power, results from the activity-generating method and Famar1 subset that resulted in the least power are reported.

Construction of the Famar1 "ancestral" sequence:
To compare the activities of clones isolated from the genomic screen to those of the element that invaded the genome of F. auricularia, an "ancestral" Famar1 element was constructed by a PCR-ligation-PCR technique (ALI and STEINKASSERER 1995 ). The ancestral sequence was inferred by maximum likelihood and agreed with that inferred previously when genomic copies of this element were isolated (LAMPE et al. 2003 ). Clones isolated from the activity screen were examined to determine which parts overlapped the ancestral coding sequence, and these were used to "stitch" together a sequence encoding the ancestral transposase. Clones pBADFa5 and pBADFa4 cover the ancestral sequence up to amino acid 270 when spliced together at amino acid 77 (Fa5, amino acids 1–77; Fa4, amino acids 78–269). pBADFa6 covers the ancestral sequence from amino acid 270 on. The final product of this PCR-ligation-PCR is an ancestral sequence, which was cloned into pBAD24 as a NcoI/PstI restriction fragment to give pDL129.

Phylogenetic and statistical analyses:
Sequences were aligned using ClustalX 1.8 with default settings (JEANMOUGIN et al. 1998 ). Minor modifications were made by hand near small insertions in some sequences. Phylogenetic analyses were done with PAUP*4.0b8 (SWOFFORD 2001 ). The equilibrium base frequencies and the transition/transversion ratio (equivalently, ) were estimated by maximum likelihood (ML) using an arbitrarily selected, most-parsimonious phylogeny resulting from an initial heuristic search (starting tree obtained by stepwise addition, followed by tree-bisection and -reconnection branch swapping). With these parameters fixed, 10 heuristic ML searches were performed (with the above search settings) to find the most likely phylogeny. This phylogeny was used in all subsequent analyses.

To detect evidence of selection, the CODEML program of PAML 3.0a (YANG 1997 ) was used in conjunction with likelihood-ratio tests (LRT; EDWARDS 1992 ) as described (LAMPE et al. 2003 ). Since the Famar1 clade of mariners is nested within the Ammar1 clade, the sequence at the root node of the Famar1 clade represents the Famar1 ancestral sequence. The node nearest the root inferred for the entire data set (by midpoint rooting with a molecular clock assumed) was used as the ancestral Ammar1.

Estimating the genome size of F. auricularia and the copy number of Famar1:
The genome size of F. auricularia was kindly determined by Spencer Johnson at Texas A&M University. Tissue samples from F. auricularia were run on an Epic (New York) Elite cytometer with excitation from an argon laser tuned to 514 nm at 0.4 W. The samples were stained with propidium iodide (50 ppm). The head and thorax of a D. melanogaster Canton-S strain was coprepared and costained with each head from the earwig to provide a reference value for a genome of known size. Red fluorescence from isolated stained nuclei was read using a 610-nm high pass filter to exclude reflected laser light at 514 nm. All samples were stained 40–60 min prior to analysis and 6000–8000 nuclei were scored for each sample.

The copy number of Famar1 in F. auricularia was determined essentially as described in ROBERTSON and LAMPE 1995 . The copy number of Famar1 in F. auricularia was determined by using a quantitative genomic slot blot and the radiolabeled full-length coding sequence of pBadFa2. The probe was radiolabeled with dATP (50 µCi, 3000 Ci/mmol) using 5x labeling buffer (Promega, Madison, WI) and the Klenow fragment of E. coli DNA polymerase. Genomic F. auricularia DNA prepared using a QIAGEN (Chatsworth, CA) genomic DNA purification kit was blotted onto nitrocellulose at concentrations of 1000, 100, and 10 ng and plasmid DNA from pBadFa2 was blotted at concentrations of 10, 3, 1, and 0.3 ng. DNA to be blotted was added to 150 µl of 2x SSPE (300 mM NaCl, 20 mM NaH₂PO₄, pH 7.4, 2 mM EDTA) and 100 ng/ml of polydI/dC, boiled for 5 min, cooled on ice, and blotted onto nitrocellulose. The samples were hybridized to the probe in an overnight reaction in 5 ml of hybridization solution containing 5x SSPE, 2x Denhardt's solution, 150 µg/ml herring sperm DNA, 0.25% SDS, and 50% formamide at 45°. The samples were washed at 65° for 30 min (each wash) in solutions with 1% SDS and declining concentrations of SSPE from 2x to 1x to 0.5x. A standard curve of hybridization intensity vs. mass was constructed on the basis of cloned Famar1 pBADFa2 using Molecular Imager software (Bio-Rad, Richmond, CA). Hybridization intensities of known masses of Famar1 genomic DNAs were compared to the standard curve and the copy number was estimated from this value and from the genome size determination.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

A large fraction of the F. auricularia genome consists of copies of the Famar1 transposon:
The genome size of F. auricularia was estimated by flow cytometry of propidium-iodide-stained nuclei compared to similarly treated D. melanogaster. This method showed that F. auricularia has a relatively large diploid genome of 3.37 ± 0.13 pg (2C), which is 7.65 times as large as the genome of D. melanogaster. Using 180 Mb for the D. melanogaster genome size (ADAMS et al. 2000 ), the haploid genome size of F. auricularia is thus estimated to be 1.377 Gb. Slot-blot analysis of Famar1 in the genome of F. auricularia indicates that copies of it account for 4% of the total genome, or 57 Mb (Fig 1). Since Famar1 is 1287 bp, the copy number of this element is thus 44,000. This value is even larger than that for Himar1, which exists in 17,000 copies in H. irritans. It seems likely that the large genome size of the earwig is due to the presence of these and other different kinds of transposable elements, each present in many copies, as is the case with many organisms, including humans (LANDER et al. 2001 ).

View larger version (72K): In this window In a new window Download PPT slide   Figure 1. Slot-blot analysis F. auricularia genomic DNA to determine the copy number of Famar1. A slot blot of F. auricularia genomic DNA probed with a full-length coding sequence of Famar1. Genomic DNA is in the left column. Control Famar1 plasmid DNA is in the right column. Densities were compared by phosphorimaging, a standard curve was constructed, and a value for percentage of the genome occupied by Famar1 copies was calculated as described in the RESULTS. We estimate that 44,000 copies of Famar1 are in the genome of F. auricularia and that this element accounts for 4% of the genome of this insect.

F. auricularia contains copies of Famar1 that encode a functional transposase:
Some copies of Famar1 were known to be intact from a previous analysis of a F. auricularia genomic library (LAMPE et al. 2003 ). We reasoned that some of these copies might encode functional transposase, although we did not know which ones, since none of them differed from the reconstructed ancestral amino acid sequence by fewer than seven changes. We screened >1900 PCR-amplified genomic copies of Famar1 coding sequences in the E. coli papillation screen to find functional ones. Our PCR procedure may have excluded a minority of copies whose ITRs had diverged somewhat because we used exact primers; however, there is no reason to believe that the excluded copies differ from those that amplified in any significant respect.

The papillation assay, used extensively to study prokaryotic transposons, relies on the ability of a specially engineered Famar1 transposon to convert lac(-) E. coli to a lac(+) phenotype via transposition (HUISMAN and KLECKNER 1987 ; KREBS and REZNIKOFF 1988 ; LAMPE et al. 1999 ). This transposon carries a lacZ gene deleted for the first eight amino acids. If this transposon can be mobilized off its plasmid and into an E. coli gene in frame, a fusion protein that has ß-galactosidase activity may result. These events can be visualized easily by the appearance of red bumps, or papillae, on an otherwise colorless colony when plated on MacConkey lactose agar (for details, see Fig 2). The number of papillae present after a certain period of time is a rough measure of the transpositional activity of the element. A typical colony of cells carrying a functional Famar1 developed hundreds of papillae.

View larger version (32K): In this window In a new window Download PPT slide   Figure 2. An E. coli papillation screen to isolate functional and nonfunctional copies of Famar1 and Ammar1. A papillation screen in E. coli to distinguish potentially active vs. inactive copies of Famar1 and Ammar1 by PCR amplication of genomic copies of coding sequences from these elements. A strain of lac(-) E. coli is constructed that contains a modified F' plasmid. This F' contains a promoterless lacZ gene that also lacks a translational start site. This lacZ gene is fused to one ITR of Famar1/Ammar1 (these elements have identical ITRs) via an open reading frame in the ITR. E. coli cells containing this construct are still lac(-). Genomic copies of the coding sequences of Famar1 and Ammar1 are amplified and cloned into the expression vector pBAD24. Ligation reactions with these products are transformed into the papillation strain and plated on MacConkey agar. After 50 hr at 32°, small red bumps (papillae) form on colonies containing potentially active copies of the transposase sequences. These papillae are due to the conversion of some of the cells in the colony from lac(-) to lac(+) via transposition of the Famar1-lacZ transposon off F' and into a nonessential E. coli gene, creating a gene fusion that provides ß-galactosidase activity. Cells undergoing this conversion can utilize the lactose in the medium and thus grow faster than their lac(-) neighbors. Cells not showing papillae after 4 days were deemed to contain inactive copies of transposase.

In this screen we identified 45 copies of Famar1 that encoded a functional transposase (2.4% of the total copies examined). Twenty of these were sequenced. We also sequenced multiple nonfunctional copies, of both Famar1 and the related Ammar1 from the honey bee (LAMPE et al. 2003 ). It is important to note that the copies that we identified as functional are only potentially active in F. auricularia itself. There is no screen for the activity of Famar1 in vivo. Mutated, transpositionally inactive mariners can repress transposition through dominant-negative effects (LOHE et al. 1996 ; HARTL et al. 1997B ), so transposition in this insect may be repressed by some of the large fraction (>97%) of apparently nonfunctional coding sequences. Of the 2000 copies of Ammar1 that we screened from the genome of the European honey bee in a similar fashion, none appeared to encode functional transposase. This result is in keeping with our observation that all of the Ammar1 copies examined, in both this screen and sequenced genomic clones isolated from a bacteriophage library (LAMPE et al. 2003 ), contain obvious inactivating mutations of one sort or another (i.e., indels or nonsense mutations) and are more diverged from their presumed ancestor than are those of Famar1. The sequences of these coding regions from both species can be found in GenBank (accession nos. AY226463, AY226464, AY226465, AY226466, AY226467, AY226468, AY226469, AY226470, AY226471, AY226472, AY226473, AY226474, AY226475, AY226476, AY226477, AY226478, AY226479, AY226480, AY226481, AY226482, AY226483, AY226484, AY226485, AY226486, AY226487, AY226488, AY226489, AY226490, AY226491, AY226492, AY226493, AY226494, AY226495, AY226496, AY226497, AY226498, AY226499, AY226500, AY226501, AY226502, AY226503, AY226504, AY226505, AY226506, AY226507).

Sequence analysis of Famar1 and Ammar1 coding sequences:
A phylogenetic tree based on maximum likelihood of all of the clones examined in this study is shown in Fig 3. The tree is consistent with data produced by sequencing entire genomic clones of Ammar1 and Famar1, in that Ammar1 and Famar1 are closely related and copies of Ammar1 are much more diverged from their presumed ancestor compared with those of Famar1 (LAMPE et al. 2003 ). As many as nine amino acid changes separate functional Famar1 from the ancestral Famar1 (e.g., in Fa3). The amino acid distance between each Famar1 and the ancestor is the number of changes inferred on branches connecting the two. Conceptual translations of all copies can be obtained from D. Lampe upon request.

View larger version (26K): In this window In a new window Download PPT slide   Figure 3. Phylogenetic analysis of the Famar1 and Ammar1 coding sequences isolated in this study. The maximum-likelihood phylogenetic tree is based on the coding sequences of all elements isolated in this study. This tree was produced with PAUP* and is midpoint rooted. Functional copies of the Famar1 transposase coding sequence are shown in boldface italic type. Branches are scaled in units of expected nucleotide changes per codon, according to the scale bar, and were derived from the distances calculated by CODEML (YANG 1997 ). The branches of the horizontal transfer lineage are thicker than the other branches.

Current models of mariner transposable element evolution predict that all copies, whether active or not, should evolve neutrally in a host population; that is, there should be no significant lack of nonsynonymous codon mutations compared to synonymous, presumably neutral mutations. However, we do expect to find evidence of selection on branches in a phylogeny that contain one or more horizontal transfer events (WITHERSPOON 1999 ; LAMPE et al. 2003 ). Therefore, we looked for evidence of selection in three sets of branches in the phylogeny, chosen a priori: branches within the Famar1 clade, the five branches connecting the ancestors of the Famar1 and Ammar1 clades (thicker branches in Fig 3), and the remaining branches, which reflect evolution within A. mellifera. For each set of branches, a separate value of the parameter (essentially equal to d_N/d_S, which is the rate of nonsynonymous evolution relative to synonymous evolution) was estimated by maximum likelihood (CODEML, PAML 3.0a; YANG 1997 ). Since a mutational bias against transversions can be incorrectly interpreted as evidence of selection, (the strength of the mutational bias against transversions) was estimated simultaneously to account for that effect. This model yields = 4.82, _A (Ammar1) = 0.91, _T (transfer) = 0.55, _F (Famar1) = 0.97. The smaller the value of , the more stringent selection is against nonsynonymous changes. Values near 1 imply no selection.

The likelihood of this model was compared to three separate submodels in which one of the three parameters, _F, _A, or _T, was fixed at one and all other parameters (base frequencies, branch lengths, other values, and ) were reestimated. These are submodels in which no selection is allowed in the respective set of branches. The likelihoods of these submodels were then compared to the likelihood of the model with three freely estimated values by likelihood ratio tests (1 d.f.; EDWARDS 1992 ) to determine whether differed significantly from one in that set of branches. The appropriate significance cutoff criterion for three tests is = 0.05/3 = 0.017 (Bonferroni correction, SOKAL and ROHLF 1981 ). No significant evidence of selection was obtained in any of the three tests (P = 0.87, 0.47, and 0.25 for Famar1, Ammar1, and transfer branches, respectively.) Thus mariners within A. mellifera and F. auricularia evolve neutrally.

There is slight evidence of selection in the horizontal transfer lineage (represented by the thick branches connecting the ancestral Ammar1 with the ancestral Famar1 in Fig 3), where at least one horizontal transfer event must have occurred. Relatively little evolution has occurred on these branches (only 18 mutations with a ratio of 7:11 synonymous:nonsynonymous, instead of the expected 4.7:13.3; reconstructed by CODEML with one free per branch), so the statistical test has little power for detecting selection. With only 18 changes, _T would have to be 0.28 to yield a significant result, instead of the observed 0.55. If such lax selection is typical of horizontal transfer events, it will be necessary to examine numerous such events simultaneously to rigorously determine whether or not selection is acting there.

Just as the ancestral, founding Famar1 element was selected for its functionality, the Famar1 copies identified as functional by the papillation screen are a nonrandom subset of the mostly nonfunctional Famar1 family. We therefore examined the evolution on branches of the phylogeny (Fig 3) connecting the ancestor to functional Famar1 copies by specifying two parameters within the Famar1 clade: one for branches leading to at least one functional element and the other for branches leading only to nonfunctional elements. As expected, there is slight evidence for selection in branches leading to functional Famar1 copies, but it is not significant (LRT, P > 0.05). If the within-population evolution of functional copies does differ from that of nonfunctional copies, it will require a larger data set to establish that fact.

Activity of individual Famar1 copies:
We quantified the transpositional activity of each active Famar1 copy in a bacterial mating-out assay. An outline of this assay is shown in Fig 4A and the results of the assay are shown in Fig 4B. The variances of the repeated activity measurements are not homogeneous (Bartlett's test, d.f. = 17, P < 0.05; SOKAL and ROHLF 1981 ), so nonparametric statistics are used in further analyses. The estimated activities of Famar1 copies differ significantly (Kruskal-Wallis test of location, d.f. = 18, P < 0.005; SOKAL and ROHLF 1981 ). Activities span a 10-fold range, from roughly one-fifth as active as Himar1 (e.g., Fa15) to approximately twice as active (e.g., Fa6).

View larger version (20K): In this window In a new window Download PPT slide   Figure 4. Quantification of individual functional Famar1 copies. (A) An overview of the mating-out assay to quantify the activity of individual active Famar1 copies. The mating-out assay involves the mobilization of a Famar1 transposon carrying an apramycin-resistance gene from a donor plasmid to a target F' plasmid. F' plasmids are subsequently transferred to a different strain of E. coli by mating. A determination of the total number of F' plasmids mated is made by plating the mating mixture on naladixic acid (Nal) and gentamycin (Gen), which selects for all F' plasmids in the recipient strain. A determination of what fraction of F' plasmids contain a transposon insertion is made by plating the mating mixture on Nal and apramycin (Apr), which selects only those F' plamids that contain a transposon insertion in the recipient strain. Activity is expressed as the ratio of AprR colonies/GenR colonies (x105). (B) Activity measurements (transposition rates relative to Himar1; see MATERIALS AND METHODS) for unique Famar1 copies that encode a functional transposase. Individual measurements are shown as circles; crosses mark the estimated activity of each Famar1.

Since Famar1 copies do differ in their activities, we looked for any tendency of accumulating amino acid changes to decrease or increase activity. None is apparent in Fig 5. Activity decreases slightly with amino acid divergence, but the trend is not significant (Spearman rank regression test, P > 0.05, d.f. = 15; ZAR 1999 ). For this analysis, one member of each of the pairs (Fa4, Fa13), (Fa2, Fa18), and (Fa1, Fa11) was removed from the data, since they have largely redundant phylogenetic histories (Fig 5). According to a conservative estimate of the power of this test, a linear trend of ±0.13 relative log₁₀ activity units would have been detected with at least 90% probability. This corresponds to a 35% change in activity per amino acid change or an 15-fold change expected over the nine-change distance observed in the data.

View larger version (13K): In this window In a new window Download PPT slide   Figure 5. Activities of Famar1 copies encoding functional transposase as a function of amino acid divergence from the ancestral sequence. Activity vs. number of amino acid differences from the ancestral Famar1. No correlation is evident (see text). The vertical axis is log10 scaled to allow easier comparisons of relative rates. Measurements for Fa7, Fa8, and Fa20 are not shown or used, since these copies are identical in sequence to others in the data set.

Some amino acid changes undoubtedly have dramatic effects on element activity (e.g., LOHE et al. 1997 ), but no such changes could be unambiguously identified in our data. All the inactive elements sequenced here contained stop codons and frameshifting indels that would prevent expression of most of the transposase ORF. There are 55 variable sites among the active Famar1, and the changes we observed at these sites do not have dramatic effects on activity. Thus, among Famar1 copies with measurable activity, our data suggest that most amino acid changes have modest effects, if any, and that they do not predominantly decrease nor increase activity.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Copies of Famar1 encoding functional transposase are detected by a bacterial genetic screen:
This study reports the construction and use of a bacterial genetic screen to isolate coding sequences of functional copies of the Famar1 mariner transposase from the earwig, F. auricularia. Famar1, a member of the mellifera subfamily of mariner transposons, is only the third known active mariner element, in addition to Mos1 and Himar1 (JACOBSON et al. 1986 ; LAMPE et al. 1996 ). The fraction of copies remaining in the F. auricularia genome that appear to encode a functional transposase is small (2.4%), yet this translates into a total of 1000 copies encoding a transposase capable of mediating transposition. This begs the question of whether Famar1 is actually still transposing in this earwig species. It is known that some mutations of transposase can be dominant negative in their phenotype and thus can repress transposition (LOHE et al. 1997 ). Moreover, mariner transposons seem to exhibit a gene dosage effect, which leads to "overexpression inhibition," a phenomenon that can lead to wild-type transposase repressing transposition as well (LOHE and HARTL 1996A ; LAMPE et al. 1998 ). Finally, hosts contain endogenous systems to silence transposons and viruses, such as RNAi (WATERHOUSE et al. 2001 ). The complex interplay of these factors is still poorly understood and remains an area in need of further research.

Sequence analysis of Famar1 shows that even remaining functional coding sequences have evolved neutrally:
The current model of mariner evolution requires that functional elements evolve neutrally within a population (reviewed by EICKBUSH and MALIK 2002 ). Clearly, elements that are initially nonfunctional cannot evolve under selection. Neutral evolution is the expected result in that case, so it is interesting only if the elements in question can be shown to be, or to have been, functional. Most previous work (e.g., ROBERTSON and LAMPE 1995 ; LAMPE et al. 2003 ) has used genomic clones or sequence to demonstrate neutral evolution and relies on indirect arguments to establish the functionality of at least the founding mariner copy. It is argued that the ancestor and its immediate descendants must have been functional; otherwise, they could not have created the large number of descendant copies we now observe. This is compelling, but several mariner subfamilies in C. elegans appear to thrive even though their ancestral (and modern) transposase genes contain multiple ORF-interrupting stop codons and indels (WITHERSPOON and ROBERTSON 2003 ). The functionality of numerous mariner copies in Drosophila was demonstrated by CAPY et al. 1992 , but these copies were too similar to each other to allow statistical tests to distinguish between neutral and selected evolution.

Our data are unique in that they directly demonstrate the functionality of transposases from certain mariner copies that are also shown to have diverged neutrally, thus confirming a basic presumption of the current model of mariner evolution. Our results are in keeping with those of previous studies (ROBERTSON and LAMPE 1995; LAMPE et al. 2003 ; WITHERSPOON and ROBERTSON 2003 ) and those of most other DNA-mediated transposons so far examined (WITHERSPOON 1999 ; SILVA and KIDWELL 2000 ). No statistically significant evidence of selection was detected in any part of the sequence data set, even for functional coding sequences. A small degree of selection was observed between the ancestors of Famar1 and Ammar1, in which at least one horizontal transfer event must have happened. Depending on the typical stringency of selection at horizontal transfers and the number of mutations that accumulate between them, the predicted selection may become statistically significant only after several transfers have occurred in a lineage (WITHERSPOON 1999 ).

No clustering of functional transposase sequences in the tree was detected, with one caveat. In two cases, identical sequences were isolated: Fa6, Fa7, and Fa8 are identical, as are Fa17 and Fa20. This might indicate a greater abundance of these particular sequence types. A clustering of functional sequences in one part of the tree has been predicted to indicate that a subset of the elements might be evolving into a new mariner subtype (LAMPE et al. 2001 ). These data are not extensive enough to test that hypothesis, nor can they provide the critical predicted link between ITR and coding sequence coevolution.

Sequence divergence between horizontal transfers—effects on mariner diversity and implications for mode of selection:
It has been suggested that mariners may diverge by very slowly accumulating mutations during repeated horizontal transfers, each involving few, if any, changes (HARTL et al. 1997A ; ROBERTSON 2002 ). However, in the F. auricularia genome we find copies of Famar1 that have diverged substantially from their founder under no apparent selection yet still encode a functional transposase (e.g., Fa3, which differs from the Famar1 ancestor by nine amino acid changes, or 2.6%). This suggests that mariners may diversify by leaps of a half-dozen or more amino acid changes per horizontal transfer and that fewer horizontal transfer events than previously thought are needed to explain mariner sequence diversity.

This combination of infrequent selection and divergence by sizable doses of random mutations is apparently enough to maintain minimal mariner functionality: the ability to transpose at a moderate rate, that is, frequently enough to allow their proliferation, but not so often as to severely reduce their hosts' fitness. However, it may not be enough to create and maintain more precise or complex adaptations such as optimal transposition rates or self-regulatory mechanisms. Thus mariners would be coarsely adapted entities at best, whose sequence variability would be matched by functional variability in transposition rate and its regulation. This in turn might help explain the wide variation in mariner copy number across host species.

Practical considerations:
Several practical applications of our results are apparent. First, as the third known functional mariner element, Famar1 thus expands the choices of those using mariners as genetic tools.

Second, it has been shown that mariners from different subfamilies do not interact (LAMPE et al. 2001 ). Thus a mariner from one subfamily can be used as a genetic tool in the presence of another, considerably broadening their utility as a group. We used Himar1 in just this way in this study to create the E. coli papillation strain DL13 for Famar1, in which the Famar1 indicator transposon was carried between the ITRs of Himar1 for the purpose of mobilizing it onto an F plasmid.

Finally, our genomic screen eliminates the need to reconstruct consensus or ancestral sequences to obtain a functional element. Previously, we reconstructed the Himar1 transposon by in vitro mutagenesis to ensure its activity (LAMPE et al. 1996 ) and the Sleeping Beauty Tc1-like element was reconstructed by heroic means (IVICS et al. 1997 ). These approaches are no longer necessary if one screens a genome that contains at least some elements that, although divergent from the consensus, still appear to be intact. Finding other novel functional mariners (or any DNA-mediated transposon that is functional in E. coli) should now be facile.

	FOOTNOTES

Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AY226463, AY226464, AY226465, AY226466, AY226467, AY226468, AY226469, AY226470, AY226471, AY226472, AY226473, AY226474, AY226475, AY226476, AY226477, AY226478, AY226479, AY226480, AY226481, AY226482, AY226483, AY226484, AY226485, AY226486, AY226487, AY226488, AY226489, AY226490, AY226491, AY226492, AY226493, AY226494, AY226495, AY226496, AY226497, AY226498, AY226499, AY226500, AY226501, AY226502, AY226503, AY226504, AY226505, AY226506, AY226507.
¹ These authors contributed equally to this work.

	ACKNOWLEDGMENTS

We thank M. Walden for F. auricularia specimens, K. Walden for assistance in cloning pFaLacTet, J. McCormick and W. Reznikoff for the gift of strains and plasmids, S. Johnson for determining the genome size of F. auricularia, and H. Robertson for critically reading the manuscript. We acknowledge the support of the Duquesne University Automated Sequencing Center. This study was supported by National Institutes of Health grant 2R01AI33586 and National Science Foundation grant MCB-0091044.

Manuscript received September 11, 2003; Accepted for publication November 13, 2003.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ADAMS, M. D., S. E. CELNIKER, R. A. HOLT, C. A. EVANS, and J. D. GOCAYNE et al., 2000  The genome sequence of Drosophila melanogaster. Science 287:2185-2195.[Abstract/Free Full Text]

ALI, S. A. and A. STEINKASSERER, 1995  PCR-ligation-PCR mutagenesis: a protocol for creating gene fusions and mutations. Biotechniques 18:746-750.[Medline]

CAPY, P., J. R. DAVID, and D. L. HARTL, 1992  Evolution of the transposable element mariner in the Drosophila melanogaster species group. Genetica 86:37-46.[CrossRef][Medline]

CHANG, A. C. and S. N. COHEN, 1978  Construction and characterization of amplifiable multicopy DNA cloning vehicles derived from the P15A cryptic miniplasmid. J. Bacteriol. 134:1141-1156.[Abstract/Free Full Text]

COATES, C. J., N. JASINSKIENE, L. MIYASHIRO, and A. A. JAMES, 1998  Mariner transposition and transformation of the yellow fever mosquito, Aedes aegypti. Proc. Natl. Acad. Sci. USA 95:3748-3751.[Abstract/Free Full Text]

COATES, C. J., N. JASINSKIENE, D. MORGAN, L. R. TOSI, and S. M. BEVERLEY et al., 2000  Purified mariner (Mos1) transposase catalyzes the integration of marked elements into the germ-line of the yellow fever mosquito, Aedes aegypti. Insect Biochem. Mol. Biol. 30:1003-1008.[CrossRef][Medline]

EDWARDS, A. W. F., 1992 Likelihood. Johns Hopkins University Press, Baltimore.

EICKBUSH, T. H., and H. S. MALIK, 2002 Origin and Evolution of Retrotransposons. ASM Press, Washington, DC.

GARCIA-FERNANDEZ, J., J. R. BAYASCAS-RAMIREZ, G. MARFANY, A. M. MUNOZ-MARMOL, and A. CASALI et al., 1995  High copy number of highly similar mariner-like transposons in planarian (Platyhelminthe): evidence for a trans-phyla horizontal transfer. Mol. Biol. Evol. 12:421-431.[Abstract]

GOMULSKI, L. M., C. TORTI, A. R. MALACRIDA, and G. GASPERI, 1997  Ccmar1, a full-length mariner element from the Mediterranean fruit fly, Ceratitis capitata. Insect Mol. Biol. 6:241-253.[CrossRef][Medline]

GUEIROS-FILHO, F. J. and S. M. BEVERLEY, 1997  Trans-kingdom transposition of the Drosophila element mariner within the protozoan Leishmania. Science 276:1716-1719. (erratum: Science 277: 753).[Abstract/Free Full Text]

GUZMAN, L. M., D. BELIN, M. J. CARSON, and J. BECKWITH, 1995  Tight regulation, modulation, and high-level expression by vectors containing the arabinose PBAD promoter. J. Bacteriol. 177:4121-4130.[Abstract/Free Full Text]

HANAHAN, D., 1983  Studies on transformation of Escherichia coli with plasmids. J. Mol. Biol. 166:557-580.[Medline]

HARTL, D. L., A. R. LOHE, and E. R. LOZOVSKAYA, 1997a  Modern thoughts on an ancyent marinere: function, evolution, regulation. Annu. Rev. Genet. 31:337-358.[CrossRef][Medline]

HARTL, D. L., A. R. LOHE, and E. R. LOZOVSKAYA, 1997b  Regulation of the transposable element mariner. Genetica 100:177-184.[CrossRef][Medline]

HARTL, D. L., E. R. LOZOVSKAYA, D. I. NURMINSKY, and A. R. LOHE, 1997c  What restricts the activity of mariner-like transposable elements. Trends Genet. 13:197-201.[CrossRef][Medline]

HUISMAN, O. and N. KLECKNER, 1987  A new generalizable test for detection of mutations affecting Tn10 transposition. Genetics 116:185-189.[Abstract/Free Full Text]

IVICS, Z., P. B. HACKETT, R. H. PLASTERK, and Z. IZSVAK, 1997  Molecular reconstruction of Sleeping Beauty, a Tc1-like transposon from fish, and its transposition in human cells. Cell 91:501-510.[CrossRef][Medline]

JACOBSON, J. W., M. M. MEDHORA, and D. L. HARTL, 1986  Molecular structure of a somatically unstable transposable element in Drosophila. Proc. Natl. Acad. Sci. USA 83:8684-8688.[Abstract/Free Full Text]

JEANMOUGIN, F., J. D. THOMPSON, M. GOUY, D. G. HIGGINS, and T. J. GIBSON, 1998  Multiple sequence alignment with Clustal X. Trends Biochem. Sci. 23:403-405.[CrossRef][Medline]

JOHNSON, R. C. and W. S. REZNIKOFF, 1984  Copy number control of Tn5 transposition. Genetics 107:9-18.[Abstract/Free Full Text]

JOHNSON, R. C., J. C. YIN, and W. S. REZNIKOFF, 1982  Control of Tn5 transposition in Escherichia coli is mediated by protein from the right repeat. Cell 30:873-882.[CrossRef][Medline]

KIESER, T., M. J. BIBB, M. J. BUTTNER, K. F. CHATER and D. A. HOPWOOD, 2000 Practical Streptomyces Genetics. John Innes Centre, Norwich, United Kingdom.

KREBS, M. P. and W. S. REZNIKOFF, 1988  Use of a Tn5 derivative that creates lacZ translational fusions to obtain a transposition mutant. Gene 63:277-285.[CrossRef][Medline]

LAMPE, D. J., M. E. CHURCHILL, and H. M. ROBERTSON, 1996  A purified mariner transposase is sufficient to mediate transposition in vitro. EMBO J. 15:5470-5479. (erratum: EMBO J. 16: 4153).[Medline]

LAMPE, D. J., T. E. GRANT, and H. M. ROBERTSON, 1998  Factors affecting transposition of the Himar1 mariner transposon in vitro. Genetics 149:179-187.[Abstract/Free Full Text]

LAMPE, D. J., B. J. AKERLEY, E. J. RUBIN, J. J. MEKALANOS, and H. M. ROBERTSON, 1999  Hyperactive transposase mutants of the Himar1 mariner transposon. Proc. Natl. Acad. Sci. USA 96:11428-11433.[Abstract/Free Full Text]

LAMPE, D. J., K. K. O. WALDEN, and H. M. ROBERTSON, 2001  Loss of transposase-DNA interaction may underlie the divergence of mariner-family transposable elements and the ability of more than one mariner to occupy the same genome. Mol. Biol. Evol. 18:954-961.[Abstract/Free Full Text]

LAMPE, D. J., D. J. WITHERSPOON, F. N. SOTO-ADAMES, and H. M. ROBERTSON, 2003  Recent horizontal transfer of mellifera subfamily mariner transposons into insect lineages representing four different orders shows that selection acts only during horizontal transfer. Mol. Biol. Evol. 20:554-562.[Abstract/Free Full Text]

LANDER, E. S., L. M. LINTON, B. BIRREN, C. NUSBAUM, and M. C. ZODY et al., 2001  Initial sequencing and analysis of the human genome. Nature 409:860-921.[CrossRef][Medline]

LOHE, A. R. and D. L. HARTL, 1996a  Autoregulation of mariner transposase activity by overproduction and dominant-negative complementation. Mol. Biol. Evol. 13:549-555.[Abstract]

LOHE, A. R. and D. L. HARTL, 1996b  Germline transformation of Drosophila virilis with the transposable element mariner. Genetics 143:365-374.[Abstract]

LOHE, A. R., D. T. SULLIVAN, and D. L. HARTL, 1996  Subunit interactions in the mariner transposase. Genetics 144:1087-1095.[Abstract]

LOHE, A. R., D. DE AGUIAR, and D. L. HARTL, 1997  Mutations in the mariner transposase: the D,D(35)E consensus sequence is nonfunctional. Proc. Natl. Acad. Sci. USA 94:1293-1297.[Abstract/Free Full Text]

MAKRIS, J. C., P. L. NORDMANN, and W. S. REZNIKOFF, 1988  Mutational analysis of insertion sequence 50 (IS50) and transposon 5 (Tn5) ends. Proc. Natl. Acad. Sci. USA 85:2224-2228.[Abstract/Free Full Text]

MEDHORA, M., K. MARUYAMA, and D. L. HARTL, 1991  Molecular and functional analysis of the mariner mutator element Mos1 in Drosophila. Genetics 128:311-318.[Abstract]

ROBERTSON, H. M., 2002 Evolution of DNA Transposons in Eukaryotes. ASM Press, Washington, DC.

ROBERTSON, H. M. and D. J. LAMPE, 1995  Recent horizontal transfer of a mariner transposable element among and between Diptera and Neuroptera. Mol. Biol. Evol. 12:850-862.[Abstract]

ROBERTSON, H. M. and E. G. MACLEOD, 1993  Five major subfamilies of mariner transposable elements in insects, including the Mediterranean fruit fly, and related arthropods. Insect Mol. Biol. 2:125-139.[Medline]

ROBERTSON, H. M. and R. MARTOS, 1997  Molecular evolution of the second ancient human mariner transposon, Hsmar2, illustrates patterns of neutral evolution in the human genome lineage. Gene 205:219-228.[CrossRef][Medline]

ROBERTSON, H. M. and K. L. ZUMPANO, 1997  Molecular evolution of an ancient mariner transposon, Hsmar1, in the human genome. Gene 205:203-217.[CrossRef][Medline]

RUBIN, E. J., B. J. AKERLEY, V. N. NOVIK, D. J. LAMPE, and R. N. HUSSON et al., 1999  In vivo transposition of mariner-based elements in enteric bacteria and mycobacteria. Proc. Natl. Acad. Sci. USA 96:1645-1650.[Abstract/Free Full Text]

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

SILVA, J. C. and M. G. KIDWELL, 2000  Horizontal transfer and selection in the evolution of P elements. Mol. Biol. Evol. 17:1542-1557.[Abstract/Free Full Text]

SOKAL, R. R., and F. J. ROHLF, 1981 Biometry. W. H. Freeman, San Francisco.

SWOFFORD, D. L., 2001 PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods). Sinauer Associates, Sunderland, MA.

WATERHOUSE, P. M., M. B. WANG, and T. LOUGH, 2001  Gene silencing as an adaptive defence against viruses. Nature 411:834-842.[CrossRef][Medline]

WITHERSPOON, D. J., 1999  Selective constraints on P-element evolution. Mol. Biol. Evol. 16:472-478.[Abstract]

WITHERSPOON, D. J. and H. M. ROBERTSON, 2003  Neutral evolution of ten types of mariner transposons in the genomes of Caenorhabditis elegans and C. briggsae. J. Mol. Evol. 56:751-769.[CrossRef][Medline]

YANG, Z., 1997  PAML: a program package for phylogenetic analysis by maximum likelihood. Comput. Appl. Biosci. 13:555-556.[Free Full Text]

ZAR, J. H., 1999 Biostatistical Analysis. Prentice-Hall, Upper Saddle River, NJ.

ZHANG, J. K., M. A. PRITCHETT, D. J. LAMPE, H. M. ROBERTSON, and W. W. METCALF, 2000  In vivo transposon mutagenesis of the methanogenic archaeon Methanosarcina acetivorans C2A using a modified version of the insect mariner-family transposable element Himar1. Proc. Natl. Acad. Sci. USA 97:9665-9670.[Abstract/Free Full Text]

This article has been cited by other articles:

				K. J. Clark, D. F. Carlson, M. J. Leaver, L. K. Foster, and S. C. Fahrenkrug Passport, a native Tc1 transposon from flatfish, is functionally active in vertebrate cells Nucleic Acids Res., March 1, 2009; 37(4): 1239 - 1247. [Abstract] [Full Text] [PDF]

				C. Miskey, B. Papp, L. Mates, L. Sinzelle, H. Keller, Z. Izsvak, and Z. Ivics The Ancient mariner Sails Again: Transposition of the Human Hsmar1 Element by a Reconstructed Transposase and Activities of the SETMAR Protein on Transposon Ends Mol. Cell. Biol., June 15, 2007; 27(12): 4589 - 4600. [Abstract] [Full Text] [PDF]

				C. Feschotte The piggyBac transposon holds promise for human gene therapy PNAS, October 10, 2006; 103(41): 14981 - 14982. [Full Text] [PDF]

				J. C. Silva, F. Bastida, S. L. Bidwell, P. J. Johnson, and J. M. Carlton A Potentially Functional Mariner Transposable Element in the Protist Trichomonas vaginalis Mol. Biol. Evol., January 1, 2005; 22(1): 126 - 134. [Abstract] [Full Text] [PDF]

				K. Lipkow, N. Buisine, and R. Chalmers Promiscuous Target Interactions in the mariner Transposon Himar1 J. Biol. Chem., November 19, 2004; 279(47): 48569 - 48575. [Abstract] [Full Text] [PDF]

				K. Lipkow, N. Buisine, D. J. Lampe, and R. Chalmers Early Intermediates of mariner Transposition: Catalysis without Synapsis of the Transposon Ends Suggests a Novel Architecture of the Synaptic Complex Mol. Cell. Biol., September 15, 2004; 24(18): 8301 - 8311. [Abstract] [Full Text] [PDF]