The transposition pathway for members of the IS200
IS family occurs using ssDNA substrates and intermediates. In vitro
IS excision requires both transposon ends to be single stranded, and insertion of the excised single-stranded circular transposon intermediate requires access to a ssDNA target (Guynet et al., 2009
). We now demonstrate that excision and insertion of two family members, IS608
, occur preferentially in the lagging strand DNA template in vivo
Excision is dependent on replication direction through the IS: it is high when the active IS strand is on the lagging strand template but substantially lower when part of the leading strand template. We also examined insertions of a plasmid-localized IS608 into normally replicating E. coli chromosomes and spontaneous insertions of ISDra2 into D. radiodurans chromosome I. They were largely directed to the lagging strand template resulting in a skew of strand-specific insertion on either side of the replication origin in these bidirectionally replicating chromosomes. For a unidirectionally replicating E. coli plasmid, insertions occurred into only one strand. Importantly, the orientation effect for ISDra2 insertion was abolished when transposition was triggered by γ-irradiation, accompanied by an increase in transposition frequency, consistent with the observation that γ-irradiation induces a repair pathway resulting in massive amounts of ssDNA with no obvious strand bias.
insertions into the E. coli
chromosome were fairly evenly distributed but a significant number occurred in the highly transcribed rrn
genes (which are oriented in the sense of replication) suggesting that high transcription levels might also provide accessible ssDNA for IS608
insertion, e.g. by generating R-loops or by affecting replication fork passage. Replication in E.coli
has been estimated to be approximately 20-fold faster than is transcription (800 nt/s versus 20 to 50 nt/s; Kornberg and Baker, 1992
). Replication forks may stall at tRNA and other highly expressed genes, possibly because transcription complexes collide “head-on” with the forks. Replication forks also stall upon co-directional encounters with RNA polymerase (Elias-Arnanz and Salas, 1997
) (Mirkin et al., 2006
) possibly due to a trapped RNA polymerase not readily displaced from DNA by fork progression.
That ssDNA at the replication fork facilitates IS608
transposition is reinforced by studies which perturb the fork using temperature sensitive DnaG (primase) or DnaB (helicase) mutants. DnaG inactivation prevents initiation of Okazaki fragment synthesis, increasing the average length of ssDNA upstream of the first complete Okazaki fragment on the lagging strand template. (Louarn, 1974
; Fouser and Bird, 1983
). DnaB inactivation results in accumulation of large amounts of ssDNA mostly likely arising from degradation of both the nascent DNA and leading strand template or from uncoupled leading-strand synthesis (Belle et al., 2007
). We observed a significant stimulation of IS608
excision after transitory inactivation of both dnaGts
If excision occurs at the replication fork and requires ssDNA, the probability that both ends are within the single stranded region of the lagging strand template should decrease with increasing IS length and thus the size of the IS should influence excision frequency exactly as we observe. Moreover, the efficiency of excision was generally higher in the dnaGts mutant even at the permissive growth temperature of 30°C and the slope of the curve was less steep. At the sublethal temperature of 33°C, excision was even higher and the length dependence even less marked. This is consistent with an increase in ssDNA length of on the lagging strand template resulting from a lower Okazaki fragment initiation frequency in the dnaGts mutant even at temperatures permissive for growth and suggests that the slope is a function of the ssDNA length available on the lagging strand template.
We also investigated the effect of DnaG overproduction. Expression of a cloned wildtype dnaG
gene not only suppressed the dnaGts
phenotype but resulted in an even more pronounced length dependence in the wildtype background suggesting that, as observed in vitro
(Zechner et al., 1992
; Sanders et al., 2010
), DnaG concentration controls the frequency of initiation and Okazaki fragment size. More importantly, this result would suggest that DnaG is not saturating at the normal replication fork in vivo
While the slopes of the curves presumably reflect the length of ssDNA on the lagging strand template, the explanation for the apparent inflection of the curves for the longer transposons is less clear. It is possible, for example, that we are observing effects of two phenomena: an initial probability that both ends are in a single-stranded form and also the probability of both ends finding each other. Further analysis is required to determine the exact reason behind this behaviour.
Although, for simplicity, we describe the lagging strand template as single stranded at the fork, it is important to note that it is not naked but protected by proteins such as single strand binding protein (Ssb; Shereda et al., 2008
). This implies that the transposition machinery can access the DNA through the protecting protein and raises the question of whether TnpA can recognise Ssb or other components of the replisome. Experiments to investigate this are in progress.
We also used the natural Tus/Ter
replication fork termination system (Neylon et al., 2005
; Kaplan and Bastia, 2009
) to examine whether blocked forks might also attract IS608
insertions. The Tus–Ter
complex forms a transient barrier to the replicative helicase, DnaB, when a fork arrives in the non-permissive direction (Neylon et al., 2005
). When provided with appropriate target tetranucleotides, Tus-dependent IS608
insertions readily occurred close (26–77 nt) to the Ter
site on the lagging strand template, consistent with nucleotide resolution mapping of the terminated nascent DNA in vitro
and in vivo
showing that the final lagging-strand primer sites are arrested 50–70 nucleotides upstream of Ter
(Hill and Marians, 1990
; Mohanty et al., 1998
While our data are consistent with the idea that stalled forks favor IS608
insertion, we do not yet know whether the ssDNA substrates are directly available at blocked forks or are generated during their processing, e.g. during replication restart or repair of ds breaks caused by replication arrest (Michel et al., 1997
; Bierne and Michel, 1994
To determine whether other IS200/IS605 family members might locate suitable ssDNA substrates for transposition, we annotated several complete bacterial genomes for ISs and identified several that harbour multiple copies of these family members. In the majority, these ISs showed a similar orientation bias to IS608 and ISDra2 relative to replication direction. Thus, targeting the ssDNA available on the lagging strand template appears to be a general theme among IS200/IS605 family members.
Replication direction and therefore identification of the lagging strand is generally implied from GC skew, the preference for G over C on the leading strand thought to be the result of differential repair (Lobry, 1996
; Grigoriev, 1998
). More strictly speaking, we observed that IS orientation was correlated with the GC skew of the region into which they were inserted rather than with replication direction per se
suggesting that the insertions predated the genome rearrangements whose scars are revealed by changes in GC skew. This has two important implications: either that transposition is infrequent or that transposition events become genetically fixed in the population.
family members are not alone in showing asymmetric strand preference in insertion. Other transposable elements such as Tn7
also appear to do so (Peters and Craig, 2001
) (Hu and Derbyshire, 1998
is targeted to replication forks during conjugative plasmid transfer and inserts in a specific orientation. The transposon-encoded TnsE protein is instrumental in targeting the transpososome to the junction between single and double stranded DNA by interaction with the β-clamp component of the replication apparatus (Parks et al., 2009
; Chandler, 2009
). Insertion of Tn7
into a replication fork likely occurs within the dsDNA covered by an Okazaki fragment. IS903
insertion bias in the conjugative F plasmid might also reflect targeting to the conjugative replication fork. It is worthwhile noting that IS10
have also exploited host replication: both are activated by transient formation of hemimethylated DNA following fork passage (Roberts et al., 1985
; Yin et al., 1988
; Dodson and Berg, 1989
However, there are major mechanistic differences between Tn7, IS903 and members of the IS200/IS605 family, suggesting that different pathways are at play. Perhaps most importantly, Tn7 transposes using a dsDNA intermediate in contrast to the ssDNA species used by IS608 and ISDra2. For Tn7 and many other transposons with dsDNA intermediates, strand transfer into a dsDNA target occurs using the 3’OH groups on each complementary strand at each end. Insertion of the first strand into a single strand target would leave a fatal break. Thus, whereas an overarching theme in DNA transposition may be the exploitation of replication forks, different elements do so in different ways.
The unique excision and insertion properties of the IS200/IS605 family may make them useful tools for probing ssDNA structures in vivo. The relationship between excision and IS length might be used to determine the effect of various factors on the state of the replication fork in vivo. For example, treatments leading to reduced fork velocity, Okazaki fragment synthesis, uncoupling of lagging from leading strand synthesis or simply forks blocked by different factors could all be probed using an excision assay as an experimental readout. We are aware that the topology of the replication fork of small, multicopy plasmids may differ in some respects from that of the chromosome, and this system may also be used to explore these possible differences. We are at present testing these possibilities experimentally.