In order to establish a transposition assay for mPing in yeast, we modified successful assays previously used for the maize Ac/Ds elements (the hAT superfamily) 2425 and the rice Osmar elements (the Mariner superfamily) 9. The most significant modification involved the construction of separate expression vectors containing fusions of the inducible yeast GAL1 promoter to either the Ping ORF1 or the Ping TPase coding regions (Figure 1). The third plasmid vector in this assay was a selectable marker composed of the yeast ADE2 gene with its coding region interrupted by mPing [including the TSD (TAA)] (Figure 1). Growth on medium lacking adenine requires the excision of mPing and the in-frame repair of ADE2. Expression of the Ping proteins (induced by galactose) and selection for excision was performed concurrently on solid media lacking adenine. In this assay, each colony represents a unique excision event, thus eliminating an artifactually high frequency due to founder effects from early transposition events. This strict selection is sensitive to slight changes in the health of the culture, due to factors such as age, density and growth phase. In recognition of this source of variation, we pooled the excision frequencies from two or more separate experiments, each with multiple replicates (see Additional File 1).

Co-expression of Ping TPase and ORF1 proteins in a single yeast cell resulted in a very low excision frequency of mPing from ADE2 (0.03 ± 0.02) (Figure 2, Additional File 1). We hypothesized that the low frequency could be due to the activity of a potential NES located in the C-terminal region of the TPase (Figure 2, Additional File 2) 26. In order to determine the impact of this putative NES on activity, we mutated two leucine residues that are conserved in most predicted NESs (L384A, L386A)26. Alteration of this sequence resulted in significantly higher excision frequencies (5.9 ± 3) (Figure 2A, Additional File 1), suggesting that the NES was inhibiting TPase function in yeast. No ADE2 revertant colonies were observed when Ping ORF1 or TPase -NES proteins were expressed in isolation (data not shown).

The increased activity of the NES mutant provided the experimental resolution necessary to test the impact of site-directed mutations in conserved residues of domains predicted to be necessary for transposition. To test the suspected DNA binding domain of ORF1 protein, we disrupted the Myb-like sequence with a mutation in the helix-turn-helix fold (L225P). Co-expression of ORF1 (L225P) and the TPase -NES proteins resulted in no detectible excision. Similarly, mutation of the catalytic DDE domain of the TPase protein (D295A) led to a reduction in excision frequency of ~150-fold (0.04 ± 0.02) (Additional File 1). These results are consistent with the hypothesis that the DNA binding and catalytic functions are divided between the ORF1 and TPase proteins respectively.

In rice and transgenic Arabidopsis, both Ping and Pong products can mobilize mPing even though their ORF1 and TPase proteins share only 62% and 77% amino acid identity, respectively 920. To test whether Pong proteins can catalyze mPing transposition in yeast, we replaced Ping ORF1 and TPase with their Pong counterparts and combined these in yeast strains with the mPing reporter (see Methods). The results (Figure 2, Additional File 1) indicate that: (i) Pong proteins catalyze low frequency excision of mPing; (ii) introduction of the NES mutation (L418A, L420A) into the Pong TPase significantly increases the excision frequency; and (iii) Pong ORF1 functions with Ping TPase -NES and Ping ORF1 functions with Pong TPase -NES to catalyze mPing excision.

Another way to address the question of cross-mobilization is to ask whether Ping proteins can mobilize a deletion derivative of Pong. To this end, we generated a 500 bp nonautonomous Pong (mPong) and inserted it into the exact same position in ADE2 as mPing (see Methods). As shown in Figure 2B, while mPong is mobilized by the Ping and Pong proteins, the frequency is much lower than mPing (note the different scales on the Y axes). The possible reasons underlying this difference is revisited in the discussion.

The high frequency of transposition catalyzed by the NES mutation permitted the isolation of many excision and reinsertion sites. Excision of mPing in both rice and transgenic Arabidopsis is unusual for a plant element in that it is usually precise 927. In other words, there is no alteration (footprint) at the excision site because mPing and one copy of the 3 bp TSD (usually TAA) are almost always removed. In our yeast assay, precise excision could be easily detected because it restores the HpaI restriction site used for cloning mPing into ADE2 (see Methods). We isolated yeast ADE2 revertants [mPing mobilized by Ping ORF1 and Ping TPase (L384/386A)] and analysed the excision sites by polymerase chain reaction (PCR) and HpaI digestion (Figure 3A). Almost all of the excision events (109/110) were precise. The one imperfect event was due to a 3 bp deletion that maintained the reading frame of ADE2 (data not shown).

Another unusual feature of all characterized members of the PIF/Harbinger superfamily is an insertion sequence preference that extends beyond the 3 bp TSD. This was first observed for the maize PIF element, with a 9 bp extended insertion site preference (the TSD and 3 bp on each side) 4. mPing was similarly shown to have a 9 bp insertion preference both in rice 12 and in transgenic Arabidopsis 9. To investigate whether a preference was manifest in the yeast assay, we used inverse PCR and transposon display to isolate mPing reinsertion sites in the yeast genome (see Methods). A consensus sequence generated from the sequences of 22 independent events revealed a 9 bp insertion site preference that was similar to the mPing insertion preference in rice and transgenic Arabidopsis (Figure 3B, Additional File 3). The insertion of mPong elements was also verified by flanking sequence analysis for four events mobilized by Ping proteins and five events catalyzed by Pong proteins (Additional File 3).

Our finding that the TPase NES had significant effects on activity in yeast suggested that protein localization is critical for mPing transposition. Many other TPase proteins harbor nuclear localization signals (NLS) that facilitates their access to the DNA substrate 2829. Analysis of Ping proteins with the nuclear prediction software NucPred indicated that the ORF1 protein has a potential NLS and, as such, is likely to be nuclear-localized (score = 0.9) 30. In contrast, the TPase has a much lower NucPred score (0.37), indicating it may not utilize a NLS to gain access to the DNA.

We used cerulean florescent protein (CFP) fusions to test the localization of the Ping proteins in yeast. ORF1 protein was found to be primarily nuclear and co-localized with the nuclear marker protein, NLS-EYFP (Figure 4) 31. In contrast, wild type TPase showed a diffuse staining pattern. Figure 4 shows that the TPase (L384/386A) NES mutant shows increased CFP fluorescence in the nucleus, confirming that the NES plays a role in nuclear export of the TPase. Co-expression of the Ping ORF1 and TPase proteins together did not significantly alter the localization pattern of either protein (data not shown).

Our findings indicate that the NES encoded by wild type TPase inhibits mPing transposition in yeast and that mutation of this motif increases nuclear localization and transposition activity. In order to determine if the NES also reduced transposition in plants, we modified the previously developed Arabidopsis transposition assay to test the effect of the TPase (L384/386) NES mutation. The reporter construct for this assay was mPing inserted between the CaMV 35S promoter and a gfp reporter gene (Figure 5A). We transformed the CaMV 35S:Ping cDNA with and without the TPase NES mutation (L384/386A) because the wild type construct was shown previously to express both ORF1 and TPase proteins from a single construct in Arabidopsis (see Methods) 9. A comparison of the number of transgenic plants that showed mPing excision (determined by green fluorescent protein (GFP) visualization and PCR analysis) indicated that the alteration of the NES resulted in significantly higher excision activity (40% versus 15%, P < 0.005). In addition, the NES mutation resulted in more GFP fluorescence in both the cotyledons and mature leaves (Figure 5B). Finally, PCR analysis using primers that flank the mPing insertion site showed an increase in the amount of the ~400 bp (excision) product in seedlings containing the mutant TPase.

Excision of mPing in both rice and transgenic Arabidopsis is usually perfect as it involves the removal of both mPing and one copy of the TSD. This stands in marked contrast to all other plant class 2 elements where one or more nucleotides usually remain at the excision site as a so-called transposon footprint 11. While all but one of the excision events of mPing in yeast were perfect (109/110) (Figure 3), one could argue that this result is an artifact of selection for ADE2 reversion. This would be the case if only the wild type sequence at this site was able to restore ADE2 function. However, in a parallel study of a rice Stowaway element, inserted at exactly the same site in ADE2 as mPing, all ADE2 revertants had footprints with insertions of 6, 9 or-12 bp 32.

Thus, excision site repair in yeast is influenced by currently unknown differences in the catalytic properties of the transposase source. Plant transposable element excision sites are thought to be repaired primarily by the non-homologous end-joining pathway (NHEJ), which is well characterized in yeast 33. In a prior study, Yu et al. 25 analysed the excision of the maize Ds element by the Ac transposase (a member of the hAT superfamily) in yeast strains that were mutant for various steps in the NHEJ pathway. They found that mutations in a subset of the genes tested influenced both the quality and frequency of excision site repair 25. In this regard, we have begun to analyse excision of mPing and the rice mariner element OsStow35 32 in the same yeast mutant backgrounds to address why mPing but not Ds or OsStow is repaired perfectly.

Another distinguishing feature of the transposition reaction of characterized PIF/Harbinger elements is an extended insertion sequence preference of 9-15 bp 456. That the target preference is conserved (to the extent that our yeast sample size can differentiate, Figure 3) indicates that one or both of the element-encoded proteins is responsible for this specificity. As such, the yeast assay should be a valuable tool that could allow the rapid analysis of mutant ORF1 and/or TPase proteins in order to determine if they contribute to insertion preference.

Perhaps the most unusual feature of the PIF/Harbinger superfamily is that transposition requires two element-encoded proteins, ORF1 and TPase 910. Prior analysis of the reconstructed Harbinger3N_DR element from zebrafish suggested that ORF1 protein binds to the TIR and is required in human cells for nuclear localization of TPase (which lacks a nuclear localization sequence). In this study, the use of the wild type ORF1 and TPase proteins from either Ping or Pong resulted in negligible excision of mPing (Figure 2). This result was unexpected as these proteins catalyzed robust excision of mPing from the gfp reporter in transgenic Arabidopsis 9. Mutagenesis of a predicted NES in the C-terminal region of the TPase from both Ping and Pong (Additional File 2) led to significant increases in the excision frequency of mPing (Figure 2A).

Increased nuclear localization of the mutant versus the wild type Ping TPase (Figure 4) suggests that the low excision frequency catalyzed by wild type TPase is a direct result of its export from the nucleus. However, if this is the case, why does wild type Ping and Pong TPase catalyze mPing excision in both rice and Arabidopsis 920? One reason may be that yeast, like other simple eukaryotes, undergoes a 'closed' mitosis where the nuclear envelope does not break down 34. Unlike the situation in plants, TPase exported to the cytoplasm in yeast cannot get back into the nucleus during mitosis. Thus, mutation of the NES prevents or slows nuclear export of TPase and keeps it in the nucleus.

An alternative model to explain the difference in activity of the wild type TPase in yeast and plants is that the NES may not function in rice and Arabidopsis. To address this question, we generated transgenic Arabidopsis plants with wild type and mutant TPase (Figure 5A). Although excision frequency is difficult to quantify precisely, we found that both the percentage of plants showing excision events and PCR products resulting from mPing excision were significantly increased in plants harbouring the TPase with the mutant NES (Figure 5). The identification of a hyperactive mutation in yeast, and its subsequent validation in plants, indicates the potential value of the yeast assay as a rapid and facile screen for ORF1 and TPase mutations that reduce or increase the frequency of transposition or alter the quality of excision or insertion events.

In addition to its potential utility at dissecting the catalytic features of the PIF/Harbinger superfamily, the yeast assay may be of value in determining the reason why mPing, and not other nonautonomous elements, become MITEs. As is true for many relatively recent discoveries, the definition of just what is a MITE is continuously being refined as an increasing number of genome sequences become available. Stated simply, a MITE is a class 2 nonautonomous element that can attain very high copy numbers (hundreds or thousands) through the rapid amplification of one or a few identical or nearly identical founder elements 3536. MITEs are usually short (~100-500 bp) and are either of unknown origin or, like mPing, are deletion derivatives of a larger autonomous element.

One of the outstanding questions about MITEs is whether their ability to attain high copy numbers is an inherent feature of the element or a fortuitous circumstance. In other words, is there something special about the structure of MITEs that, for example, makes them better substrates for transposase? Alternatively, are MITEs just average nonautonomous elements that happen to be in the genome when, for example, an unrelated autonomous element is producing transposase? The results of this study favour the former model where MITEs, such as mPing, have distinctive structural features (compared to other low copy nonautonomous elements) that facilitate their rapid amplification and spread in the genome. Although mPing and mPong are both deletion derivatives of larger autonomous elements, the excision frequency of the former is much higher than the latter (Figure 2A, B). In fact, the frequency of mPing excision is ~40- to 50-fold higher than that of mPong when either Ping or Pong proteins catalyze transposition. The relative success of mPing in yeast mirrors what occurs in rice; there are less than five deletion derivatives of mPong in all of the sequenced rice genomes, while most characterized strains of japonica rice have 25-50 mPing copies and a few have over 1000 12. As such, the yeast assay can be used to determine the excision frequencies of a variety of constructs, including additional deletion derivatives of Pong and Ping, chimeric elements where regions of mPing and mPong are swapped and site-directed mutations throughout mPing. In this way we should be able to determine the cis-features that distinguish MITEs from low copy number nonautonomous elements.

Saccharomyces cerevisiae strain DG2523 and the pWL89a vector were described previously 2437. The following primers were used to prepare the elements for ligation into the HpaI site of ADE2: mPing F 5'-AAGGCCAGTCACAATGGGG-3', mPing R 5'-AGGCCAGTCACAATGGCTAG-3'; mPong F 5'-AAGGCCAGTCACAATGGGG-3'; and mPong R 5'-AGGCCATTCACAATGCAGT-3'. The mPong element was constructed in order to contain 251 bp from the 5' end of Pong and 245 bp from the 3' end of Pong (sequence available upon request). Genes from cDNA clones [GenBank: AK068363.1, GenBank: AK068654.1] or genomic DNA were cloned into the Gateway cloning vectors, pENTR or pDONR Zeo, and then transferred to destination vectors with LR Clonase reactions (Invitrogen, California, USA). Sequence alterations were made using the Quikchange II Mutagenesis Kit (Stratagene, California, USA).

Ping and Pong ORFs (wild type and mutant) were cloned into pAG413GAL-ccdb (ORF1) and pAG415GAL-ccdb (TPase) vectors 38. Transformed yeast were grown to saturation (36-48 h) in 5 ml of CSM -his-leu-ura with dextrose, washed with 5 ml sterile water, resuspended in 0.5 ml water and plated onto CSM -ade -his -leu -ura with galactose as the sole carbon source. Colonies were counted after incubation at 30°C for 15 days. Viable counts were made by plating 50 μl or 100 μl of a 4 × 10⁴ dilution on yeast extract peptone dextrose plates.

ADE2 revertant yeast was grown in selective media and genomic DNA isolations were performed using an E.Z.N.A. Yeast DNA Kit (Omega Bio-tek, Georgia, USA). PCR [PCR Master Mix (Promega, Wisconsin, USA)] using primers flanking the mPing excision site were digested with HpaI (New England Biolabs, Massachusetts, USA) and then analysed by gel electrophoresis. PCR products were cloned for sequencing using Topo-pCR2.1 (Invitrogen). Inverse PCR and transposon display was performed as described 9. The insertion site analysis figure was made using the program Pictogram http://genes.mit.edu/pictogram.html.

The yeast nuclear marker vector NLS-EYFP (pPS1888) was described previously 31. N-terminal cerulean fusion protein constructs were constructed using pAG415GAL-cerulean-ccdb 38. Yeast were grown on selective media with galactose, visualized with a Zeiss Axio Imager M1 microscope equipped with a 63× oil objective and SlideBook 4.0 software (Intelligent Imaging Innovations, Denver, USA). GFP expression was detected in seedlings with a Zeiss Discovery V12 fluorescence stereoscope.

The coding regions of the Ping cDNA [GenBank: AK068363.1] were cloned into the pENTR vector for mutagenesis and sequencing before transfer to the pEarleyGate 204 vector 39. Wild type and NES mutant versions of TPase were transformed into Arabidopsis thaliana (Columbia ecotype) containing a previously transformed single copy of the mPing reporter 9 using a simple Agrobacterium tumefaciens (GV3103) floral dip method 40. Seedlings were germinated on 0.2% phytagel plates (0.5 Murashige and Skoog, 1% sucrose) containing the appropriate selection reagents (7.5 μg/ml PPT, 50 μg/ml Kan, 150 μg/ml Timentin) in a Percival growth chamber (12 hrs light, 21°C) and analysed by fluorescent microscopy. DNA from 2-week-old seedlings was extracted by the CTAB method with PCR performed as described to confirm transformation and detect mPing excision 9.