Alternative Mechanisms for Tn5 Transposition

  • Asad Ahmed mail

    Affiliation: Department of Biological Sciences, University of Alberta, Edmonton, Alberta, Canada

  • Published: August 28, 2009
  • DOI: 10.1371/journal.pgen.1000619

Reader Comments (1)

Post a new comment on this article

Alternative Explanations

Posted by telnet on 26 Mar 2010 at 08:37 GMT

Chalmers and Kleckner addressed many of the issues raised here in their 1996 EMBO J. paper. Dimer donors and synapsis of elements of located on partially replicated sister chromosomes were the most favored explanation of cointegrates.

The synapsis of Tn5 ends does not depend on supercoiling and will therefore suffer even fewer constraints than those of Tn10, perhaps explaining the increased level of off-pathway products with this element.

Regarding the dimer donor: if one transforms a purified monomer into a recA- cell and immediately performs a standard plasmid prep, such has the CsCl or Qiagen protocol, the material is always contaminated by a small, but significant, quantity of dimer.

Finally, as stated in this article, it is always true that "the results of in vitro studies should be extrapolated to biological phenomena with care." However, in this case there are so many confounding factors operating in the 'black box', which is the cell, that in vitro observations are perhaps more reliable.

No competing interests declared.

RE: Alternative Explanations

asadahmed1 replied to telnet on 31 Mar 2010 at 15:45 GMT

The commentator has raised three points. I answer these points first, and then provide the reasons for these answers.

The first point is that 'Chalmers and Kleckner (EMBO J.,1996) raised the same issues and concluded that synapsis of elements located on partially replicated chromosomes, and dimer donors, were the most favored explanations for the formation of cointegrates'. The present work shows that, while dimer donors do contribute to the formation of cointegrate-like structures in both Tn10 and Tn5, synapsis of transposon ends present on sister chromatids probably does not. The likely reason is that these authors did not have access to the findings presented in this paper.

The second point is that 'after transformation of CsCl-purified monomers into recA- cells, the material is always contaminated by a small, but significant, quantity of dimers'. This is not true since we verified that both monomeric and dimeric plasmids were pure before and after transformation and there is no question about their purity. Of course, one can never eliminate this possibilty entirely, but other evidence presented below indicates that the monomeric preparations were not contaminated with dimers.

The third point is that in view of many 'confounding factors, in vitro observations are more reliable than the results of in vivo studies'. In fact, there are no confounding factors; the behavior of Tn5 is completely normal for a transposon that utilizes both the replicative and conservative mechanisms concurrently. The reason that most previous reports state that the mechanism is exclusively conservative is because evidence to the contrary (i.e., for replicative transposition) had been suppressed.

There are three possible mechanisms for the formation of cointegrates. The first is the replicative mechanism (Shapiro, PNAS 1979) which begins with nicking of the transposon termini to expose the 3' ends that are joined to 5' ends from the nicked target site. The Shapiro intermediate thus formed is replicated to produce a cointegrate. The model is supported by extensive biochemical studies of Mizuuchi, Chaconas and others. It also provides elegant explanations for the formation of adjacent deletions and replicative inversions from both inside and outside ends of the transposon. The second mechanism (Berg, PNAS 1983) suggests that cointegrates arise by a conservative, cut-and-paste mechanism from dimeric donors. One copy of the donor plasmid flanked by a pair of transposons could be cut out from the dimeric donor and inserted between a pair of staggered nicks at the target site. This model is supported in the present study by the behavior of purified dimers. The third mechanism (Chalmers and Kleckner, EMBO J. 1996) is also conservative, and produces various rearrangements from monomeric donors. Monomers can generate products characteristic of replicative transposition (cointegrates, adjacent deletions, and duplicative inversions) if synapsis occurs between the appropriate pairs of ends on replicating sister chromatids. I show below that this model is not compatible with our observations on Tn10.

In our experiments, we used CsCl-purified monomeric and dimeric plasmids carrying Tn5 or Tn10 (at similar locations) as donors, and pOX38 as the recipient plasmid, in standard mating-out assays. The rates of cointegrate formation under these conditions were:

Tn10 monomeric donors: 0.9 x 10-9 cointegrates/ donor cell per division
Tn10 dimeric donors: 5.6 x 10-9 cointegrates/ donor cell per division

The 'cointegrates' produced by the monomers were never found to be true cointegrates. Southern analysis revealed that they involved fusions of donor and recipient plasmids at random sites unrelated to the transposon and the element was never duplicated.These fusion structures were stable in a recA+ host. Thus, these 'cointegrates' seem to have been formed by illegitimate recombination. On the other hand, the cointegrates formed (at a higher frequency) by dimeric donors included both fusion structures as well as true cointegrates. The latter had duplicate copies of the transposon at each plasmid junction and were unstable in a recA+ host. It was therefore concluded that (i) Tn10 monomeric plasmids can not produce true cointegrates and (ii) Tn10 dimeric plasmids can produce cointegartes presumably by the dimer donor model (cut-and-paste from a dimer). Hence the few reports in literature of the formation of cointegrates by Tn10 are probably due the presence of dimers in those plasmid preparations. Had our monomeric preparations been contaminated with dimers, at least some of the fusion structures would have displayed the structure of cointegrates, but none was found.

Now, turning to Tn5 plasmids, the results were as follows:

Tn5 monomeric donors: 0.1 x 10-7 cointegrates/ donor cell per division
Tn5 dimeric donors: 0.7 x 10-7 cointegrates/ donor cell per division

In this case, cointegrates from monomeric donors had the structure expected of true cointegrates, always involved duplication of Tn5 or one of its constituent IS50 elements, and were extremely unstable when transferred into a recA+ host. The same was true for cointegrates produced by dimeric donors except that they arose at a higher frequency. All of these results were confirmed by Southern analysis. Hence, it was concluded that (i) Tn5 monomers can produce genuine cointegrates, and (ii) the frequency is higher with dimeric donors probably because of the operation of two pathways - replicative transposition and conservative transposition (cut-and paste from dimeric donors). It is probably not due to the synapsis of transposon ends present on sister chromatids since that model did not work with Tn10, an element which undoubtedly transposes by the conservative mechanism (Haniford, Mobile DNAII, 2002).

Now turning to intramolecular cointegrates (adjacent deletions and inversions), the replicative model explains all of these rearrangements satisfactorily. The dimer-donor model does not even address the problem of intramolecular events. The model of Chalmers and Kleckner makes a bold attempt to explain these rearrangements - normally associated with replicative transposition - by a conservative mechanism from monomers. If synapsis occurs between appropriate pairs of ends on replicating sister chromatids, then intramolecular events could produce either a 'duplicative' inversion or an adjacent deletion depending on the orientation of the target (see Figure 6 in EMBO J. 1996 paper). This model does explain the observations with Tn5 , but not with Tn10. For example, it does not account for the non-occurrence of duplicative inversions from Tn10. It also does not explain why there are no adjacent deletions arising from the outside ends of Tn10. (The absence of true cointegrates from monomeric Tn10-containing plasmids, predicted by this model, has already been mentioned.) Hence, we are left only with the replicative model (and to a limited extent, the dimer-donor model) to explain all of the Tn5-promoted rearrangements observed in vivo.

With regard to the comment about the primacy or reliability of in vitro observations over in vivo observations, I quote from the report of an anonymous reviewer:

"For Dr. Ahmed's proposal to be tenable a target DNA molecule would need to be assimilated by the Tn5 synaptic complex before release of the donor backbone. In Tn10 this cannot occur. Since a cocrystal structure for Tn5 transposase complexed with DNA is available I decided to have a look at the structure. Interestingly, and in support of Dr Ahmed's proposal, the active site residues of the transposase and the 3' ends of the transposon appear to be accessible by a target molecule without requiring displacement of the donor backbone. Dr. Ahmed may want to consider adding a figure to that effect - perhaps a local crystallographer could help and provide a more rigorous assessment of the situation than this reviewer can."

This comment shows that not much credence can be placed in this in vitro work. A fine example of approaching a biological problem can be seen in the work on Tn10. First, the phenomenon was studied in detail, then extensive in vivo studies were undertaken to develop a working model, and only then in vitro work was done to test the model. This the right way of studying a new biological phenomenon. This was unfortunately not done in the case of Tn5. A model was developed first; the rest was done afterwards. It is better to put the horse before the cart - not the cart before the horse.

No competing interests declared.