Transcription of AAT•ATT Triplet Repeats in Escherichia coli Is Silenced by H-NS and IS1E Transposition

Background The trinucleotide repeats AAT•ATT are simple DNA sequences that potentially form different types of non-B DNA secondary structures and cause genomic instabilities in vivo. Methodology and Principal Findings The molecular mechanism underlying the maintenance of a 24-triplet AAT•ATT repeat was examined in E.coli by cloning the repeats into the EcoRI site in plasmid pUC18 and into the attB site on the E.coli genome. Either the AAT or the ATT strand acted as lagging strand template in a replication fork. Propagations of the repeats in either orientation on plasmids did not affect colony morphology when triplet repeat transcription using the lacZ promoter was repressed either by supplementing LacIQ in trans or by adding glucose into the medium. In contrast, transparent colonies were formed by inducing transcription of the repeats, suggesting that transcription of AAT•ATT repeats was toxic to cell growth. Meanwhile, significant IS1E transposition events were observed both into the triplet repeats region proximal to the promoter side, the promoter region of the lacZ gene, and into the AAT•ATT region itself. Transposition reversed the transparent colony phenotype back into healthy, convex colonies. In contrast, transcription of an 8-triplet AAT•ATT repeat in either orientation on plasmids did not produce significant changes in cell morphology and did not promote IS1E transposition events. We further found that a role of IS1E transposition into plasmids was to inhibit transcription through the repeats, which was influenced by the presence of the H-NS protein, but not of its paralogue StpA. Conclusions and Significance Our findings thus suggest that the longer AAT•ATT triplet repeats in E.coli become vulnerable after transcription. H-NS and its facilitated IS1E transposition can silence long triplet repeats transcription and preserve cell growth and survival.


Introduction
AATNTTA triplet repeats are types of trinucleotide repeats that are highly cumulative in human genome [1,2], in contrast, they are much less abundant in many prokaryotic genomes, such as in E.coli, showing a biased distribution towards eukaryotic genomes [3,4,5,6,7]. Further, AATNTTA repeats are also found to be overrepresented in intronic and intergenic regions, but underrepresented in exons and UTRs, showing a nonrandom nature of distribution in an individual genome [1,2].
AATNTTA repeats have been characterized in vitro to form various types of non-B DNA secondary structures, including hairpin, triplex, non-H DNA [8,9,10,11], which potentiallychallenge the stable maintenance of the repeats in genomes [8].
Consisting with this, it has been found in prokaryotes that certain proteins known as chromosome structural proteins bind preferentially to AT-rich DNA that is normally seen in promoter region, and which may make the DNA segments inactive to avoid forming non-B secondary structure during DNA transcription [12]. H-NS and its paralogue StpA are two of the main chromosome structural proteins in E.coli, which bind AT-rich DNA sequences with overlapped specificity; H-NS binds to more than 1000 genes when repressing the transcription of the target DNA region [12].
The non-B structures formed by AATNTTA triplet repeats were found to be similar to those of non-B secondary structures formed by disease causing trinucleotide repeats such as CAG?CTG, CGG?CCG and GAA?TTC, which are associated with more than 40 human genetic diseases, including Huntington's disease or fragile X syndrome [8,13]. Consistent with this, a typical expansion feature of AATNTTA triplet repeats was found in an in vitro amplification assay [9,10,13,14,15,16,17,18]. Moreover, in human populations similar instabilities of AATNTTA triplet repeats were recently found to be associated with the high IgE blood syndrome in Chinese children [19], with schizophrenia [20,21], cocaine addiction [22], and with the high prevalence of depression in adult Parkinson's disease patients [23]. More strikingly, a similar triplet repeat expansion was also found to be associated with the propagation of a 33-bp AT-rich repeat, which displayed chromosomal fragilities in humans, such as FRA3B, FRA16B, and FRA10B [24,25].
However, the maintenance and biased distribution of AATNTTA repeats in different chromosomal regions in a genome or in different genomes are still poorly understood. To gain some understandings of the maintenance of AATNTTA triplet repeats in vivo and its biased distribution within a genome or in genomes, we performed experiments using a 24-triplet AATNTTA repeats in the E.coli model system. We found that propagation of the repeats on plasmids was overall normal, but that transcription of the repeats in either orientation with either AAT or ATT serving as transcribing template induced significant changes in colony morphology, leading to formation of convex colonies from normal colonies, which then progressively transformed into transparent colonies. Interestingly, we found that IS1E transposition from the chromosome to either the proximal site of the promoter of lacZ in the plasmids or into the AATNTTA triplet repeats themselves were capable of reversing colony changes by repressing transcription of the repeats. And we further found that the AT-rich repeats binding protein H-NS, a chromosomal structuring protein, but not its paralogoue StpA, played dual roles in triplet repeats binding and in facilitating IS1E transposition into the repeats containing plasmid, which ultimately prevented transcription of the triplet repeats.

Results
Transcription of a 24-triplet AATNTTA repeat on a plasmid induces morphological changes of bacterial colonies from convex to transparent Plasmids carrying a 24-triplet AATNTTA repeat with either the AAT or the ATT strand as lagging strand template for DNA replication were initially constructed as described in the materials and methods. They were named as pAAT 24 and pTTA 24 , respectively. Plasmid pAAT 8 was a deletion product occasionally obtained when propagating pAAT 24 in a recA mutant, and the pTTA 8 plasmid was obtained by reversing the AAT 8 into the opposite orientation.
When the pAAT 24 and pTTA 24 plasmids were propagated in the E.coli JM83 wildtype strain, normal convex colonies on LB plates containing ampicillin were formed, but the colonies started to become flat when plates were left at 4uC overnight ( Figure 1). Almost all flat colonies became concave, and then progressively turned into transparent colonies. In contrast, a similar phenomenon was not observed with JM83 cells when they propagated pAAT 8 , pTTA 8 , and pUC18 plasmids growing under the same conditions (data not shown). This indicated that only colonies carrying longer AATNTTA triplet repeats suffered morphological changes when grown overnight.
Several processes may be responsible for the observed alterations in colony morphology, including triplet repeats transcription from the lacZ promoter ( Figure 1A) [26], DNA double strand breaks caused by formation of non-B secondary structures, such as non-H structure, which may be targeted by certain DNA structure specific nucleases [27,28], or activation of the cryptic phageW 80 in the JM83 chromosome, leading to cell lysis.
To test these possibilities, plasmids pAAT 24 and pTTA 24 were also transformed into JM83 mutants defective in homologous recombination, such as DrecA, DrecB, recF::Tn10 Kan R , and DruvABC. We wanted to test if the colony morphology changes may be hinted at AATNTTA triplet repeats breaks in plasmids, which requires homologous recombination repair, and which relies on the rec gene functions. It was observed that DNA double strand breaks in E.coli genome or plasmid affected cell viability and also the colony morphology, leading to plasmid loss or cell death.
However, we found that compared to wild type cells, propagations of pAAT 24 and pTTA 24 plasmids in all rec mutants tested did not significantly enhance morphological changes (data not shown). Therefore, we excluded double strand break formation as the dominant cause for the observed changes in colony morphology.
To test whether the genetic background of the E. coli host made a difference, the growth and colony morphology of AB1157recF::Tn10 and W3110DstpA::cat cells containing plasmids pAAT 24 and pTTA 24 were compared to that of JM83. Although their genetic backgrounds were significantly different from JM83 and no similar cryptic phages were found, AB1157recF::Tn10 and W3110DstpA::cat colonies suffered the same morphological changes as JM83 cells (Figure1 B). These suggested that the observed morphological changes as associated with AATNTTA triplet repeats only depended on the presence of the plasmids, but not on the genetic background or cryptic phage of the host. Therefore, all subsequent work was done using the JM83 wild type and its derivative strains.
As shown in Figure 1A, AATNATT triplet repeats were subcloned at the EcoRI site of pUC18, which is located downstream from the lacZ promoter ( Figure 1A). Therefore, we wondered whether transcription of the AATNATT triplet repeats from the lacZ promoter occurred in the stationary phase, thus causing directly or indirectly colony morphology changes.
To test whether transcription was responsible for the observed changes in colony morphology, we constructed JM83Dhns and JM83DstpA::cat mutants defective in chromosomal structuring proteins H-NS and StpA, respectively. We therefore correlated the possible effects of transcription of AATNATT triplet repeats in pAAT 24 and pTTA 24 with morphological changes in the two strains by supplementing with IPTG, or by co-expressing LacI Q from a plasmid. Interestingly, compared to JM83 and its rec derivatives, propagation of pAAT 24 and pTTA 24 in the JM83Dhns mutant did not provoke changes in colony morphologies, but propagation in JM83DstpA::cat did, indicating that H-NS, but not StpA, was required for this process ( Figure 1B).
H-NS binds to AT-rich DNA, which may have an effect on AATNTTA triplet repeats transcription and somehow stimulate morphological changes of the colonies. To test additional effects, pAAT 24 and pTTA 24 were transformed into the DhnsrecB double mutant. No significant differences in morphological changes between the single Dhns and the DhnsrecB double mutant were observed (data not shown). Therefore, transcription of AATNTTA triplet repeats did not cause double strand breaks in the presence or absence of the H-NS protein.
Transcription of AATNATT triplet repeats stimulates transposition of IS1E from the chromosome to plasmids During prolonged incubation of morphologically changed colonies, secondary colonies can be slowly regenerated from the transparent cells containing pAAT 24 and pTTA 24 plasmids that eventually form healthy colonies in situ ( Figure 1). The plasmid DNA was isolated from these regenerated colonies, and analyzed for variations on an agarose gel ( Figure 2A). Plasmids recovered from the regenerated colonies were bigger in size, running between the monomeric and the dimeric plasmids ( Figure 2A).
Restriction enzyme digestions showed that extra PstI cutting sites were introduced to the repeats, because pUC18 has only one PstI restriction site. We sequenced the recovered plasmids and performed blast searches. This analysis indicated that the additional PstI cutting sites were the result of an IS1E insertion into the plasmid.
The IS1E insertions were often on the 59 side of the flanking sequence of the AATNATT triplet repeats, that is, within the promoter region of the lacZ gene ( Figure 2B). For example, IS1E insertions occurred with a higher frequency into the ''GGAATTGTG'' site, where the first adenine base was the starting site for LacZ mRNA transcription. Another high frequency site was ''GATTACGAA'', containing the ''GAA'' of the EcoRI restriction site 59 to the repeats ( Figure 2B). These findings suggested that transcription of the AATNATT triplet repeats was responsible for the negative effects on cell growth.   24 ), which behaves similarly when a lacIQ plasmid was coexpressed, or 5% glucose was added in the LB plate (For clarity the data were not shown); (5) JM83 DstpA::cat (pAAT 24 ); (6) JM83DstpA::cat(pAAT 24 ) with 5% glucose; (7) JM83DstpA::cat(pATT 24 ); (8) JM83DstpA::cat (pATT 24 ) with 5% glucose; (9) W3110DstpA::cat (pAAT 24 ); (10) W3110DstpA::cat(pAAT 24 ) with 5% glucose; (11) W3110DstpA::cat(pATT 24 ); (12) W3110DstpA::cat(pATT 24 ) with 5% glucose; (13) AB1157 recF (pAAT 24 ); (14) AB1157 recF (pAAT 24 ) with 5% glucose; (15) AB1157 recF (pATT 24 ); and (16) AB1157 recF (pATT 24 ) with 5% glucose. doi:10.1371/journal.pone.0014271.g001 AATNATT triplet repeats serve as targets for IS1E transposition in repeats transcription Besides recovering IS1E insertions upstream of the AATNATT repeats, we also isolated plasmids containing IS1E insertions in the AATNATT repeating array itself ( Figure 2B and Figure 3). EcoRI fragments of ,1 kb in size containing at least one IS1E element within the AATNATT repeats were recovered from both pAAT 24 and pTTA 24 plasmids ( Figure 3A). These findings are of particular interest, as this is the first demonstration of AATNATT repeats acting as potential hotspots for transposition in vivo. In this case, we have further found that the insertion of IS1E into AATNATT repeats rely on the repeats orientations in transcription. IS1E appeared to favour inserting itself into the AATNATT repeats with its unique PstI restriction site distal to the PstI restriction site in the vector when AAT orientation is transcribed, generating larger DNA fragment after PstI digestion ( Figure 3A lane 1 and 3, and 3B), while insertion of IS1E in the repeats array will use the opposite direction if the ATT orientation is transcribed, generating small DNA fragment by PstI digestion ( Figure 3A lane 5 and 7, and 3B). These suggested that ISIE transposition into AATNATT repeats occurred in repeats transcription. Transcription of AAT repeats and ATT repeats may produce distinct repeats DNA conformations that may differently be recognized by H-NS, and therefore differently affect the IS1E transpositions. Based upon the fact that IS1E transposition into AATNATT repeats expands the repeats by 3 triplets of either AAT or ATT repeats (9 nts) at both the 59 and 39 ends flanking the IS1E insertion, which may imply a potential way of making repeats expansion by certain transposons in trinucleotide repeats [13,29,30,31,32].

Transcription of AATNATT triplet repeats is inactivated by IS1E transposition
To understand the biological significance of the IS1E transposition during the transcription of AATNATT triplet repeats, we have further analyzed the effect of IS1E transposition on lacZ transcription in the plasmids. To this end, we eliminated the AATNATT triplet repeats in pAAT 24 IS1E-A using EcoRI digestion (as marked in Figure 2B), and subsequently re-ligated into the repeats-free pAAT 24 IS1E-A vector at the EcoRI restriction site.
As determined by DNA sequencing, insertion of an IS1E into pAAT 24 IS1E-A did not alter the lacZ gene promoter significantly ( Figure 2), however, the IS1E element could potentially use its promoter to transcribe both the transposase gene (encoded by it) and the AATNATT triplet repeats downstream the IS1E. We conducted a conventional a-complementation analysis by using AATNATT triplet repeats free pAAT 24 IS1E-A and a positive control plasmid, pUC18 by transforming these plasmids into JM83 cells. The transformants were grown on LB plates containing ampicillin, IPTG, and X-gal. JM83 cells carrying the repeats free pAAT 24 IS1E-A plasmid formed white colonies while the pUC18 plasmid formed blue colonies ( Figure 4). Therefore, transposition of IS1E into the upstream region of the AATNATT triplet repeats in the lacZ promoter repressed transcription of the repeats. IS1E elements have two partially overlapping open reading frames, InsA and InsB', which have a relative reading frame of 0 and -1, respectively [29,30,31,32]. InsA protein regulates transcription of the gene producing the IS1E transposease, while InsAB' binding to both the left and right IS1E terminals represses IS1E transposase transcription from a promoter found partly in IRL and simultaneously inhibits transposition [29,30,31,32]. As shown here, such a regulatory mechanism may contribute to the repression of AATNATT triplet repeats transcription.
This has been further confirmed that repression of AATNATT triplet repeats transcription by IS1E insertions helped plasmid stability; we co-propagated a LacI Q producing plasmid with either the pAAT 24 or the pTTA 24 plasmids. Co-propagation of the LacI Q plasmid and pAAT 24 or pTTA 24 improved plasmid recovery, which was also consistent with the result of adding 5% glucose to the medium to repress lacZ gene transcription (data not shown). These results further indicated that transcription of AATNATT triplet repeats was responsible for the morphological changes of the E.coli cells, and that inactivation of transcription by IS1E insertions into the lacZ promoter or the AATNATT triplet repeats proximal to the promoter helped cells recover from the transcriptional stress.

H-NS silences transcription by binding to AATNATT triplet repeats and promotes IS1E transposition
H-NS selectively silences bacterial genes associated with pathogenicity, and the gene's responses to environmental insults [33,34,35,36]. H-NS binds preferentially to the AT-rich motifs displaying planar curvature, which is likely to appear in bacterial promoters [33,36,37]. In addition, H-NS also influences transposition and recombination [38,39,40]. In some transposition processes, strains with hns mutations show a low-level of transposase production. In IS1 transposition, reduction of InsAB' production to barely detectable levels has been reported [38,39,40]. Because we observed here that H-NS was required for IS1E transposition (Figure 1), while it was also implicated in the colony morphological changes in conjunction with the transcription of the 24 AATNATT triplet repeats, we determined whether the hns gene product itself may have a role in triplet repeats transcription by measuring the effects of H-NS on plasmid copy number variations in the wild type and the Dhns mutant of JM83. The results indicated that transcription of the two orientations of AATNATT triplet repeats was differently affected by H-NS. However, the plasmid copy number of pAAT 24 was not significantly affected by the presence or absence of H-NS, or by the presence or absence of IPTG induction (Table 1). Only 86% of the plasmid copy number of JM83 was detected in the JM83Dhns mutant when induced by IPTG; 69% of the plasmid copy number of JM83 with pATT 24 in the JM83Dhns was detected without IPTG induction; and 61.7% of the plasmid copy number of JM83 was detected in JM83Dhns with IPTG induction (Table 1). To rule out that multiple copies of pAAT 24 and pTTA 24 mitigated H-NS effects in the JM83 wild type, we integrated a single copy of a 24triplet AATNATT repeat in two orientations into the attB site located on the E.coli JM83 chromosome (see Materials and Methods and Figure 1). Propagations of the repeats carrying strains were monitored by analyzing changes of bacterial cell morphology using confocal microscopy. Filamentous growth of the cells was noted when propagating the AATNATT repeats in the chromosome of the JM83Dhns mutant, but not in the JM83 wildtype ( Figure 5), nor in JM83Dhns and JM83 strains carrying either the pAAT 24 or pATT 24 plasmid (data not shown). These suggested that H-NS affected the AATNATT triplet repeats on the chromosome in a dosage dependent manner.

Discussion
In human beings, the expansion and contraction of trinucleotide repeats CAGNCTG, CGGNCCG and GAANTTC are associated with more than 40 human genetic diseases and cancers, including Fragile X, Huntington's disease, SCA1-12, MD, and Fredericha 's ataxia [8,13]. Similarly, as one of the most abundant and most polymorphic trinucleotide repeats in the human genome, the AATNATT triplet repeats also show instabilities, which have recently implicated in some human health symptoms, for example, high IgE blood symptom in Chinese children [19], schizophrenia [20,21], cocaine addiction [22], and prevalence of depression in Parkinson's disease patients [23]. However, understanding the transactions of AATNATT triplet repeats in vivo has so far been largely elusive. In this work, we found that a 24-triplet AATNTTA repeat can be propagated stably in E.coli when it was not transcribed. Transcription of the triplet repeats invoked cell toxicity, and therefore had to be silenced by using H-NS or by H-NS facilitated IS1E transposition. In contrast, propagation and transcription of a short AATNTTA repeat of 8 triplets under otherwise similar conditions did not show cell toxicity, nor H-NS and H-NS facilitated IS1E transposition, suggesting that transcription mediated cell toxicity is dependent on the length of the AATNTTA triplet repeats, and that similar sized AATNTTA repeat may also possibly be intrinsically vulnerable for transcription in all eukaryotes and prokaryotes. Consisting with this, we found that AATNTTA repeats longer than 16 triplets do not exist in the databases of human genomic plus transcript, the mouse genomic plus transcript and the others as set by NCBI. Although we realized that very long AATNTTA repeats have been found in many species, including human being and Drosophila etc., this may further implicate that transcription acted as selective pressure against long AATNTTA repeats (.17 triplets) to be distributed in the coding region of genes.
The mechanism underlying the induction of cell toxicity and the cellular morphological changes by transcription of AATNATT triplet repeats is complex. In this work, we have ruled out the effects of generation of DNA double strand breaks and the cryptic phage induced cell lysis on colony morphological alterations. While our work directly implicated an effect of transcription of longer AATNATT triplet repeats on cellular morphological changes, we reasoned that transcription of longer AATNATT triplet repeats may facilitate the repeats to form certain types of non-B DNA secondary structures, such as non-H structure, which recruits binding of histone-like protein such as H-NS etc ( Figure 6) [9], causing depletion of nuclear structure associated proteins in the chromosome in cells. The cellular morphological changes could be induced due to the depletion of the histone-like proteins, which may also includes HU and IHF etc in some situations [35,40,41,42]. In support of this idea, it was found that propagation of AATNTTA triplet repeats containing different triplets formed non-H structure in vivo [9], and also simultaneous depletions of H-NS, HU and IHF in Escherichia coli K-12 are lethal [41,42]. Cells under the situation of depletion of histone-like  proteins decrease the negative superhelicity of their chromosomal DNA, show increased lethality [41,42]. Similarly, we think that might also happen during the transcription of AATNATT triplet repeats in plasmids that may facilitate the repeats to form non-B secondary structures, and that depletes the H-NS and affects the cell growth. This fits with the observation of the involvement of H-NS, but not its paralogue StpA, in causing the transcription mediated problems. The H-NS was found to affect DNA compaction and transcription regulation through physical interactions with ATrich DNA motifs with curvature [33,34,36,37]. However, H-NS has also been implicated with helping targeting DNA molecules in transposition or cleaving the ends of the transposing elements ( Figure 6) [30,31,32,38,39,40]. Related to all these processes, our work suggested that H-NS affected IS1E transposition in a DNA sequence dependent manner, promoting distinct repression of AAT and ATT triplet repeats transcription and different transposition rates.
AATNATT triplet repeats may behave like some disease causing trinucleotide repeats, showing expansion and contraction instabilities. It was demonstrated in vitro assay that AATNATT triplet repeats produced expansion and contraction instabilities during replication [16,17,18]. However, similar triplet repeats instabilities were not detected in our in vivo assay. We have examined a large number of isogenic mutants defective in homologous recombination, methyl-directed mismatch repair, and nucleotide excision repair to search for the triplet repeats instabilities. However, under the condition of triplet repeats transcription, we failed to see any significant expansion and contraction instabilities, except for repeats instability caused by IS1E insertions (data not shown). Our work has therefore shown that longer AATNATT triplet repeats are vulnerable after transcription in E.coli cells that have therefore developed a set of mechanisms to inhibit its transcription by using both H-NS silencing and H-NS facilitated transposition inactivation, which somehow stabilize the repeats from causing cell morphological alteration and probably also repeats expansion and contraction. While bearing the findings of IS1E transposition into both the promoter region and the repeats array, we are unable to distinguish if the AATNATT triplet repeats serve as usual hotspots of transpositions, or does it do so only in the repeats transcription? Our findings of the IS1E transposition depended on the repeats orientation in transcription could suggest that the transposition is linked to the transcription of AATNATT triplet repeats. Interestingly, a similar transposition of Tn5 into GAANTTC repeats during the repeats forming triplex has also recently been demonstrated in vitro [43].

Conclusions
We found that transcriptions of longer AATNATT triplet repeats located on plasmid and in the E.coli genome were responsible for the formation of abnormal cell colonies. The bacterial colonies were sick in morphology, which was accompanied by significant IS1E transposition and filamentous cell growth. H-NS was found to be a key protein for these processes; presumably inactivating triplet repeats transcription by differently binding to the repeat arrays and then promoting IS1E transposition ( Figure 6). These findings suggest that RNA transcription of AATNATT triplet repeats may serve as a selective pressure for biased distribution of AATNATT repeats in different chromosomal regions.

Plasmids
The plasmids used in this study were pKOV, pUC18 and its AATNATT repeats carrying derivatives, pAAT 24 and pTTA 24 [45,46]. Plasmid pAAT 24 was constructed by cloning the (AAT) 24 repeats in the EcoRI site of pUC18 plasmid, which was on the lagging-strand template of the replication fork (a gift from C. Abbott, University of Edinburgh). Inversion of the trinucleotide repeat array of (AATNATT) 24 to generate plasmid pTTA 24 was performed by using EcoRI cleavage, and followed by religation using T4 DNA ligase. The plasmids pAAT 24 and pTTA 24 were all confirmed by DNA sequencing. Plasmids pAAT 24 IS1E-A , pAA-T 24 IS1E-B, pAAT 24 IS1E-C, were plasmids of pAAT 24 carrying Figure 6. A model illustrating the response of E.coli cell to the transcription of AATNATT repeats in plasmid and genome. RNA transcription by using lacZ promoter opened the double stranded AATNATT repeats that could promote the repeats to form DNA secondary structures that may recruit histone-like proteins, H-NS, to the repeats DNA, causing depletion of this protein in the chromosome, which further resulting in DNA instability and/or cell death. Repression by H-NS deactivated the transcription while also facilitating IS1E transposion, the repeats transcription is therefore further silenced by the transposition, over time, the repeats might be able to be converted into mixed repeats by IS1E and eventually form ''Junk DNA''. doi:10.1371/journal.pone.0014271.g006 IS1E elements in the promoter region of the lacZ (pAAT 24 IS1E-A , pAAT 24 IS1E-B) or in the AAT repeat array (pAAT 24 IS1E-C), respectively. Plasmids pTTA 8 and pTTA 8 were the deletion products of plasmid pAAT 24 when propagated in a JM83recA::cat R mutant. DNA sequencing primer used in this work is 59-AT-CCACATTGCCCTCCATC-39, which was synthesized by Huada Co. Ltd (Beijing).

Media and bacterial cultivation
Luria-Bertani broth (L-broth) was utilized for the cultivation of bacteria at 30uC and 43uC when constructing the strains of JM83-AAT 24 , JM83-ATT 24 by using pKOV integration [44,47]; while all other types of cultivations were carried out at 37uC. Ampicillin was applied by a concentration of 100 mg/ml when it was required. Transformation was performed by using a CaCl 2 method [47].

Plasmid DNA isolation and agarose gel electrophoresis
Plasmid DNA was prepared using a kit purchased from Qiagen after the propagations of the plasmid carrying strains for a period of time, normally overnight cultivation was applied. Agarose gel electrophoresis was conducted according the reference [47] on 0.8% gels (Flowgen).

Examination for the repeat tract instability
Plasmids population was examined following the method [28]: Briefly, monomeric plasmid DNA was used to transform E. coli strains of interest and plasmid DNA prepared from a population of transformants. Cells from roughly 4 primary transformants were harvested in 5 ml L-broth and 50 ml of this suspension was diluted into 5 ml L-broth and grown for 24 hours. This corresponds to 30 generations of cell growth. Plasmid DNA was isolated and cleaved with EcoRI. The fragments were end-labelled with S 35 -dATP using DNA polymerase I Klenow fragment and resolved on 8% native polyacrylamide gels. Bands were visualised either using X-ray film or on a Molecular Dynamics phosphor Imager [28].
DNA sequencing and determination of the transposon DNA cycle-sequencing was performed using a kit purchased from PE Applied Biosystems. And DNA sequence was extracted from DNA sequencing has been used for searching the homology of the inserted DNA against NCBI database by using the Blast search engine on the NCBI website, IS1E transposition was therefore determined based on the Blast search.

Determination of the plasmid copy number
Plasmid copy numbers in the strains of JM83 wildtype and JM83Dhns mutant carrying pAAT 24 and pATT 24 were determined as the following: strains carrying pAAT 24 and pATT 24 were initially propagated overnight in LB medium. And such overnight cultures were then made into aliquots by 1 mL into two test tubes of 1.5 mL, respectively. Induction of the repeats transcription was applied to only one sample by using IPTG by a final concentration of 0.4 mmol/L, and the two test tubes derived from the overnight culture were further cultivated for 4 hours at 37uC. Total DNA including both chromosomal DNA and plasmid DNA were prepared by following the method described in the reference [48]: Briefly, the cultures were centrifuged at 10,000 rpm at 4uC for 5 minutes. And the pellets were resuspended by adding 50 uL 10 mmol/L EDTA and 50 uL freshly prepared solution containing 0.2 moL/L, NaOH, and 0.5% sucrose, and were further incubated at 70uC for 5 min. 1.5 uL of 4 Mol/L KCl and 0.5 uL of 0.4% bromophenol blue were added when the mixture was cooled down to the room temperature, and 20 uL was analyzed by running a 0.5% agarose gel. The ratio of the total plasmid DNA and the chromosomal DNA (mainly one bulky band) was obtained by measuring their band area appeared on the agarose gel by using software (Quantity One V.4.6.2); three independent measurements were performed.
Integration of the AATNTTA repeats into chromosome attB site Plasmid pKOV was used as an integrative tool for the construction of JM83-AAT 24 and JM83-ATT 24 [45]. Two DNA primers were designed and utilized as follows: upper strand primer: 59 GTGTTCAGCGGCCGCTCCGGGCTATGAAA-TAGAAAAATGAATCCGTTGCCTGCGTTATC39, and lower strand primer: 59CAGGATGGCGGCCGCCCATCTGGTAT-CACTTAAAGGTATTAAAAACCCCACAGATGCG39, which were synthesized by Shanghai Sangon Co. Ltd. They bear a NotI restriction site (underlined sequence), respectively, and also contain part of attB sequence, and the flanking sequence of lacZ open reading frame of plasmids pAAT 24 and pTTA 24 . Integrative plasmids pKOV-AAT and pKOV-ATT were constructed by inserting the PCR products at the NotI sites.
Strains of JM83-AAT 24 and JM83-ATT 24 were selected against the following criteria [45]. For using antibiotic, 20 mg/ml of chromamphenicol was utilized; for selection against sacB, sucrose was added into the LB medium to a final concentration of 5% (w/v). PCR amplification using the same DNA primers as abovementioned was further performed for the confirmation of the JM83-AAT 24 , JM83-ATT 24 .

Observation of the cell morphological changes during cultivation
Morphological alterations of the E.coli vegetative cells grown in LB broth were monitored by using a light microscope with oil immersion objective. Cells were stained by safranin before observation [49].