Mechanisms of Dinucleotide Repeat Instability in Escherichia coli

Corresponding author: Marc Bichara, Cancérogénèse et Mutagénèse Moléculaire et Structurale, UPR 9003, CNRS, Pôle API, Boulevard Sébastien Brant, 67400 Strasbourg-Illkirch, France., bichara{at}esbs.u-strasbg.fr (E-mail)

Communicating editor: R. MAURER

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The high level of polymorphism of microsatellites has been used for a variety of purposes such as positional cloning of genes associated with diseases, forensic medicine, and phylogenetic studies. The discovery that microsatellites are associated with human diseases, not only as markers of risk but also directly in disease pathogenesis, has triggered a renewed interest in understanding the mechanism of their instability. In this work we have investigated the role of DNA replication, long patch mismatch repair, and transcription on the genetic instability of all possible combinations of dinucleotide repeats in Escherichia coli. We show that the (GpC) and (ApT) self-complementary sequence repeats are the most unstable and that the mode of replication plays an important role in their instability. We also found that long patch mismatch repair is involved in avoiding both short deletion and expansion events and also in instabilities resulting from the processing of bulges of 6 to 8 bp for the (GpT/ApC)- and (ApG/CpT)- containing repeats. For each dinucleotide sequence repeat, we propose models for instability that involve the possible participation of unusual secondary structures.

MOST genomes contain many repetitive nucleotide sequences that vary in complexity from complete genes to simple sequences of one or a few base pairs. The physical organization of repeat sequences can differ from widely dispersed copies of a relatively long, complex sequence to tandem arrays of simple sequence composition. Microsatellites, defined as regions in which 1–10 bp are tandemly repeated eight or more times, have assumed an increasingly important role as markers in the genome and have been applied in fields as disparate as tumor biology, forensic medicine, and population genetic analysis. Microsatellite repetitive sequences are widely distributed in all the genomes studied to date. They comprise unstable regions that undergo mutational changes (generally additions or deletions of integral numbers of repeats) at rates much greater than that observed for nonrepetitive sequences (for a review see DEBRAUWERE et al. 1997 ; SIA et al. 1997 ). Alterations in the length of repetitive sequences within or near genes can alter their expression or function. In humans a number of genetic diseases, such as myotonic dystrophy, Huntington's disease, etc., result from expansion of repetitive tracts (reviewed in GORDENIN et al. 1997 ; WARTER and TRANCHANT 1998 ; WELLS et al. 1998 ). In addition, an increase in tract instability is diagnostic of some human cancers (JACKSON et al. 1998 ), particularly those associated with defective DNA mismatch repair (WOOSTER et al. 1994 ; JIRICNY 1998 ; LOEB 1998 ). As a consequence, every aspect of the biology of simple repeats is now the subject of intensive studies.

The two types of mechanisms usually invoked to explain simple repeat instability are unequal recombination and DNA polymerase slippage (for recent reviews see WELLS 1996 ; WELLS et al. 1998 ). There is as yet no direct experimental evidence showing that unequal recombination can generate repeat stretches of different lengths. However, tandem repeat sequences are known to be intrinsically susceptible to slipped-strand mispairing during replication, resulting in short deletions and/or additions (STREISINGER et al. 1966 ; LEVINSON and GUTMAN 1987 ; KUNKEL 1990 ; YANG and MASKER 1996 ). In this mechanism, slippage between template and primer strands occurs during replication, producing a bulge composed of an unpaired repeat unit. If the bulge is formed on the template strand, the resulting mutation is a deletion; a bulge on the primer strand yields an addition. In Escherichia coli, the mutagenic intermediates expected to give rise to these mutations (extrahelical bulges of 2 bp in the case of dinucleotide repeats) are known to be good substrates for the long patch mismatch repair system (LPMR; ALBERTINI et al. 1982 ; CARRAWAY and MARINUS 1993 ; FRIEDBERG et al. 1995 ). This replicative model is in agreement with the increase in short additions/deletions that has been observed within (GpT) repetitive sequences in strains deficient in mismatch repair in both E. coli and yeast (LEVINSON and GUTMAN 1987 ; STRAND et al. 1993 , STRAND et al. 1995 ; JOHNSON et al. 1996 ).

When the nucleotide sequence of the repeats supports the formation of secondary structures (cruciform, hairpins, Z-DNA), the replication of the repeat can lead to larger deletions or expansions. The usual replicative model to explain such large events involves the formation of such secondary structures (bulges, palindromes) that, once replicated, will lead to expansions if present on the nascent strand or to deletions, if present on the template strand. In E. coli, for example, plasmids containing (GpC) dinucleotide repeats were shown to exhibit unusual DNA structures in vitro, which correlated with an increased frequency of long deletion events (LDEs; FREUND et al. 1989 ; BICHARA et al. 1995 ).

In this work, we have examined the role of major biological processes (replication, long patch mismatch repair, and transcription) in the instability of all the possible combinations of dinucleotide repeats. For this purpose, we established the instability spectra of (ApT) (GpC) (GpT/ApC), and (ApG/CpT) repeats using a plasmid-based assay.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Bacterial strains and medium:
The E. coli strain XL1 blue from Stratagene [La Jolla, CA; recA1 endA1 gyrA96 thi-1 hsdr17 supE44 relA1 lac (F' proAB lacI^q ZM15 Tn10)] was used for large-scale preparations of parental plasmids (see below). Wild-type strain JM103 (F'traD36 lacI^q(lacZ)M15 proA⁺B⁺/endA1 supE sbcBC thi-1 rpsL (Str^r) (lac-pro) (P1) and its LPMR-deficient derivative, JM103 mutS, were used for comparisons between LPMR-proficient and -deficient backgrounds, respectively.

Construction of plasmids:
Constructions were made starting from plasmid pUCL+, a pUC8 derivative in which an AflIII site was introduced at position 1601 by oligonucleotide-directed mutagenesis (SCHUMACHER et al. 1998 ). pCUL- is a pUCL+ derivative in which both the origin of replication and the lacZ' gene have been inverted to ensure that the direction of plasmid replication and of transcription of the lacZ' gene is the same in both plasmids ([see Figure 1 and SCHUMACHER et al. 1998 for a more precise description of the starting plasmid]). These two plasmids were used to generate the different derivatives containing the sequences (GpC)₁₅, (ApT)₂₄, (GpT)₂₄, (ApC)₂₄, (ApG)₂₄, and (CpT)₂₄ by insertion of oligonucleotides in the SalI site of the polylinker. Cloning of the non-self-complementary sequences (GpT/ApC) and (ApG/CpT) results in two possible orientations of the inserted sequences. The nomenclature used for one pair of plasmids, for example, pUCL + (GT)₂₄, and pCUL-(GT)₂₄ is the following: the sequence as written corresponds to the sequence of the leading template strand in the 5' to 3' orientation. The insert of the plasmid pair containing the repeat sequence in this particular orientation is referred to as in the (GpT) orientation. On the other hand, insertion of the oligonucleotides in the other orientation will lead to plasmids pUCL + (AC)₂₄ and pCUL - (AC)₂₄, which will be referred to as in the ApC orientation.

View larger version (12K): In this window In a new window Download PPT slide   Figure 1. Description of pUCL+ and pCUL- plasmids. In these plasmids the dinucleotide sequence was inserted at the SalI site of the polylinker. In pCUL-, the HaeII fragment containing the lacZ gene and the AflIII fragment containing the origin of replication have been reversed. In both constructs, the replication of the plasmids and the transcription of the lacZ gene are codirectional.

Insert length analysis (see Figure 2):
The plasmid DNA was digested with HindIII, 5'-³²P-labeled, and subsequently cut with EcoRI. The resulting digests were loaded on a 10% sequencing gel and run for 2 hr at 50°. The bands corresponding to expansions and deletions within the repeat sequence were detected and quantified using a Molecular Dynamics (Sunnyvale, CA) Phosphorimager. Quantitation of bands representing 0.1% of mutations were usually reproducibly achieved.

View larger version (51K): In this window In a new window Download PPT slide   Figure 2. Expansion and deletion patterns observed after insert length analysis (see MATERIALS AND METHODS) run on a 10% acrylamide gel for 2 hr at constant temperature (50°). Insert length-analysis of (GpC)15, (ApT)24, (GpT)24, (ApC)24, (CpT)24, and (ApG)24-containing plasmids are presented from left to right. For each insert, the analysis of the plasmid grown in a wild-type strain (W-T), in the mutS derivative without IPTG and of the insert cloned in pCUL- plasmid and grown in mutS strain without IPTG are shown from left to right. The main band corresponds to the undeleted plasmid. Note that the migrations of the (GpT/ApC) and the (ApG/CpT) inserts are not the same depending on the orientation of the insert. Indeed, in the (GpT) and the (ApG) orientation, the GpT- and the ApG-containing strand are labeled while the complementary strand is labeled in the opposite orientation. Plasmids grown in the mutS derivative in presence of IPTG and the parental plasmids are not shown. The "visual" difference between the gels and the instability spectra (Figure 3 and Figure 4) results from the subtraction of the spectrum of the parental plasmid from the spectrum deduced from the gels (see MATERIALS AND METHODS).

Determination of the mutation spectra:
For each dinucleotide repeat two independent experiments were conducted and, as no significant differences were observed, the one corresponding to the gel shown was selected for this study. Each plasmid was first prepared in strain XL1 blue to yield what is referred to as the parental plasmid preparation. These preparations inevitably contain mutations accumulated during the estimated 40 generations of growth of the corresponding cultures. The instability spectrum of each parental plasmid preparation was determined using the insert length analysis described above. The parental plasmid preparations were then used to transform strain JM103 and its mutS derivative (JM103 mutS). Cells were plated to yield between 10,000 and 100,000 transformants per ampicillin plate. All the transformants from one plate were pooled and diluted to yield 10⁷ cells per 250 ml of LB-ampicillin (50 µg/ml). Cultures were subsequently incubated for 11 hr at 37°. Under these conditions, the number of generations in liquid cultures was measured to be 15 ± 1 and the overall number of generations giving rise to these plasmids was estimated to be ~38 in both wild-type and mutS strains. As both the parental plasmids and the plasmids for which analysis is presented in Figure 2 result from an equivalent number of generations, they usually exhibit approximately the same level of instability. When necessary, transcription from the lacZ promotor was induced by the addition of isopropyl thiogalactoside (IPTG) (2 mM) to the plates and to the liquid media. Large-scale plasmid preparations performed on these cultures gave comparable yields of plasmidic DNA (~400 µg/250 ml) whatever the insert present on the plasmid. The length of the insert in the resulting molecules was then analyzed biochemically. The mutation spectrum for a given plasmid in a given strain was obtained by subtracting the mutation spectrum of the parental plasmid from the observed spectrum. As the number of generations giving rise to plasmid DNA in each case was equivalent in both wild-type and mutS strains and in the presence or absence of IPTG, a direct comparison of sequence instability was possible.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Rationale of the work:
To study the genetic instability of dinucleotide repeats without going through a genetic screen, we used a plasmid-based assay in which the different inserts were sufficiently unstable to allow direct visualization of polymorphisms on an acrylamide gel (see MATERIALS AND METHODS, Insert length analysis, and Figure 2).

The peculiar mode of replication of ColE1-derived plasmids was used as a tool to investigate the dependence of dinucleotide repeat sequence instability on replication mechanisms. In such plasmids, DNA polymerase I (PolI) mediates leading strand synthesis for approximately the first 400 nucleotides at which point it is replaced by DNA polymerase III (PolIII). In contrast, lagging strand synthesis, which is initiated at a primosome assembly site (PAS) is carried out exclusively by PolIII. Further away from the origin of replication, concurrent synthesis of leading and lagging strands is performed by a dimer of the polymerase III holoenzyme (reviewed in BACKMAN et al. 1978 ; KORNBERG and BAKER 1992 ). In pUCL+, the repeat sequences are located close to the origin of replication (400 bp) and are therefore replicated in a "PolI/PolIII" mode whereas in pCUL-, they are placed farther away from this origin (1600 bp; Figure 1), and, as a consequence, are replicated in a "PolIII/PolIII" mode. Cloning of the repeat sequences into the polylinker of two different pUC8 derivatives (Figure 1) gave rise to two sets (pUCL+ and pCUL-) of six plasmids, containing the sequences (GpC)₁₅, (ApT)₂₄, (GpT)₂₄, (ApC)₂₄, (ApG)₂₄, and (CpT)₂₄, considering the two possible orientations of the inserts (GpT/ApC) and (ApG/CpT). This enabled the investigation of the potential role of the two replication modes in dinucleotide repeat instability. For the purpose of simplicity the role of the replication mode in this instability was restricted to an analysis in absence of transcription and of long patch mismatch repair. For the same reasons, the contribution of the long patch mismatch repair system to the instability of these repeat sequences was investigated by performing the experiments in a mismatch-proficient and a mismatch-deficient (mutS) strain without induction of transcription from the lacZ promotor. To analyze the influence of transcription on the instability of these repeat sequences in the absence of LPMR, pUCL+ plasmids containing the dinucleotide tracts were used to transform the JM103 mutS strain under uninduced conditions (i.e., in absence of the inducer IPTG, where transcription from the lacZ' promoter is repressed by the LacI repressor) or under induced conditions (in the presence of IPTG).

Both expansions and deletions have been categorized into short (±1 dinucleotide repeat) and long (> ±1 dinucleotide repeats) mutation events. Abbreviations are SDEs and SEEs for short deletion events and short expansion events, respectively, and LDEs and LEEs for long deletion events and long expansion events, respectively.

Instability of the two self-complementary sequences (GpC)₁₅ and (ApT)₂₄:
Among the different dinucleotide repeats studied, only (GpC)₁₅ and (ApT)₂₄ exhibit self-complementary sequences, allowing the formation of hairpin structures when single stranded.

Instability of the (GpC) alternating sequence has been determined for a repeat length of only 15 repeats (vs. 24 for all the other inserts) in view of its extreme instability. Indeed, attempts to clone inserts containing 24 GC repeats result in a majority of deleted molecules, thus preventing the study of its instability. One key observation concerning this dinucleotide repeat is the complete absence of both short and long expansions (Figure 2 and Figure 3). In the wild-type strain, long deletion events occur at a frequency of 4%. In the mutS background SDEs are considerably increased to reach a frequency of 7.4%, while the overall deletion frequency is >12%. When this insert is present within the polylinker of the pCUL- plasmid, there is a dramatic decrease (50 times) of the LDEs, while the SDEs remain essentially unaffected (see Figure 2 and Figure 3). On the other hand, induction of transcription does not alter significantly the stability of the (GpC) repeats (Figure 3; Table 2).

View larger version (27K): In this window In a new window Download PPT slide   Figure 3. Instability spectra of plasmids containing the (GpC)15 (top) and (ApT)24 (bottom) inserts. The abscissa represents the mutations in base pairs as identified by their migration pattern following biochemical analysis (see MATERIALS AND METHODS). The height of the vertical bars gives the percentage mutation frequency, determined by quantification of the radioactive bands on acrylamide gels using a Molecular Dynamics Phosphorimager (see MATERIALS AND METHODS). White bars: pUCL+ derivative in wild-type strain (without induction of transcription); light gray bars, pUCL+ derivative in mutS strains (without induction of transcription); dark gray bars, pUCL+ derivative in mutS strain [with induction of transcription by IPTG (2 mM)]; black bars, pCUL- derivative in mutS strain (without induction of transcription).

As opposed to the (GpC) alternating sequence, instability within the (ApT) repeated sequence does include both expansions and deletions, with expansions representing 25% of the mutational events in the wild-type strain (see Figure 2 and Figure 3). In this strain, SDEs represent 13% of the deletions and SEEs 50% of all the expansion events. In a mismatch repair-deficient strain, the proportion of both the SDEs and the SEEs are increased to reach a level that corresponds to 50 and 75% of the total of the deletions and expansions, respectively. As observed for the (GpC) sequence, although less dramatically, (ApT) dinucleotide repeat instability is dependent on the replication mode; in the pCUL- plasmid, both LEEs and LDEs are decreased by a factor of 2 and 3, respectively, as compared to pUCL+ derivative. As observed for the (GpC) repeats, instability within (ApT) repeats does not significantly vary upon induction of transcription by IPTG (see Figure 3, Table 2).

Instability within the non-self-complementary sequences (GpT/ApC) and (ApG/CpT):
For the (GpT/ApC) and (ApG/CpT) sequences, which are not self-complementary, the two possible orientations of each insert have been considered.

In a wild-type strain, both the (GpT) and the (ApC) orientation are quite stable, showing only 1% LDEs and no detectable expansions (Figure 2 and Figure 4). Deletions of up to -24 base pairs can be scored when the (ApC) sequence is on the leading strand [(ApC) orientation], while only up to -18 bp deletions can be scored when the (GpT) repeat is situated in the leading strand. Although longer deletion events can be detected in the (ApC) orientation, no significant quantitative difference in the frequency of instability can be observed between the two orientations (see Figure 4).

View larger version (35K): In this window In a new window Download PPT slide   Figure 4. Instability spectra of plasmids containing the (GpT/ApC)24 insert [top; the (GpT) orientation is on the left side and the (ApC) orientation is on the right] and the (ApG/CpT)24 insert (bottom; the ApG orientation is on the left side and the CpT orientation is on the right). The abscissa represents the mutations in base pairs as identified by their migration pattern following biochemical analysis (see MATERIALS AND METHODS). The height of the vertical bars gives the percentage mutation frequency, determined by quantification of the radioactive bands on acrylamide gels using a Molecular Dynamics Phosphorimager (see MATERIALS AND METHODS). White bars, pUCL+ derivative in wild-type strain (without induction of transcription); light gray bars, pUCL+ derivative in mutS strains (without induction of transcription); dark gray bars, pUCL+ derivative in mutS strain [with induction of transcription by IPTG (2 mM)]; black bars, pCUL- derivative in mutS strain (without induction of transcription).

As expected, inactivation of the LPMR system (mutS strain) leads to an increase of the short deletion and expansion events. More surprising is the fact that LEEs and LDEs are also increased, with mutations involving deletions and expansions of up to three to four dinucleotide repeats (6–8 bp) being mutS dependent (see gel on Figure 2 and Figure 4). Indeed, LEEs, which are absent in the wild-type strain, reach 1.2 and 3.4% in the LPMR-deficient strain, for the (GpT) and (ApC) orientations, respectively. Although the overall increase of the LDEs in the mutS strain is less dramatic than expansions, deletions involving up to four dinucleotides are increased by a factor of two- and fivefold for (GpT) and (ApC) orientations, respectively (see Figure 4).

When the GpT and ApC inserts are present in the pCUL- vector, there is almost no effect of the replication mode on the frequency of deletion and expansion since the instability spectra are almost identical to when they are present on the pUCL+ plasmid. In the same way, the induction of transcription in the presence of IPTG does not seem to significantly alter the spectra for both orientations of the insert.

Plasmids containing the (ApG/CpT) insert allowed evaluation of instability in the ApG and CpT orientations. In a wild-type strain the (CpT) orientation gives rise to higher levels of instability than the (ApG) orientation; there are three times more expansions in the (CpT) orientation as compared to the (ApG) orientation (2.4% vs. 0.8%, respectively) and almost five times more deletions (2.9% vs. 0.6%, respectively). When the (ApG) sequence is on the leading strand, no LEEs occurred; however, LEEs do represent almost 40% of the expansions in the (CpT) orientation (Figure 4).

Both (CpT) and (ApG) repeats produced expected increases of the SDEs and SEEs in the mutS as compared to the wild-type strains. In addition, in the mutS strain, there is a surprisingly large increase in the LDEs in the (ApG) orientation (eightfold), while LDEs are increased by a factor of only 2.2 in the (CpT) orientation. In the mutS strain and in the (ApG) orientation, SDEs occur at a higher frequency (7.8%) as compared to SEEs (2.7%). This bias between short deletions and additions also exists in the (CpT) orientation, but to a lesser extent (Figure 4). When the insert is located farther away from the origin of replication (pCUL- plasmid) the levels of neither the LEEs nor the LDEs are altered as compared to the corresponding pUCL+ plasmid; however, the short expansions now reach the level of the short deletions (between 8 and 9%). Increased transcription from the lacZ promoter has no influence on the level of instability of the (ApG/CpT) insert whatever the orientation of the insert relative to the origin of replication is.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The genome of most organisms thus far examined contains many tracts of repetitive DNA called microsatellites. The discovery that a number of human diseases are the direct consequence of mutations within such repeats has triggered considerable interest in the mechanisms that change the number of copies of repeated DNA sequences (WARTER and TRANCHANT 1998 ; WELLS et al. 1998 ). Although major differences between mammalian and bacterial cells exist in terms of DNA replication enzymology and chromatin assembly, we feel that common mechanisms may govern such instability. As dinucleotide repeats are among the simplest and most common tandem repeats, we analyzed the role of the dinucleotide sequence and of major biological processes (namely, replication, mismatch repair, and transcription) on their instability in the genetically well-defined organism E. coli.

Our results show that the overall level of instability of dinucleotide repeats is directly related to the self-complementarity of a sequence (Table 1). Indeed, in a wild-type strain, both (GpC) and (ApT) self-complementary sequences exhibit a much higher degree of instability than the non-complementary sequences. In a wild-type strain, deletions are the predominant events in all the cases except for the (ApG/CpT) insert in both the (ApG) and (CpT) orientation where expansions and deletions occur with a quite similar frequency. The complete absence of expansions observed with the (GpC), containing insert (Table 1) will be discussed below under Mechanisms of instability.

Influence of the mode of replication:
Data obtained from plasmids pCUL- and pUCL+ show that LDEs are clearly more abundant when the (GpC) and (ApT) repeat sequences are located within the region of the plasmid where replication takes place using the PolI/PolIII mode of replication (pUCL+ plasmids) compared to the situation in which the insert is replicated using the PolIII/PolIII mode of replication (pCUL- derivative). This effect is more dramatic for the (GpC) insert (50-fold decrease in instability in pCUL- as compared to pUCL+ derivatives) than for the (ApT)-containing plasmids (only a 3-fold decrease between the pUCL+ and the pCUL- derivatives; Table 2). This suggests that a large majority of the long deletions observed within these two self-complementary sequences results from replication of the leading strand of the plasmid during the PolI/PolIII mode. This finding also seems valid for the LEEs observed with the (ApT) insert, although to a lesser extent (reduction of 2-fold in pCUL- as compared to pUCL+). In contrast, the instability spectra of the (GpT/ApC) and (ApG/CpT) dinucleotide repeats are independent of the modes of replication of the plasmids. This absence of influence of the replication modes can formally give rise to at least three interpretations: (i) the mechanism giving rise to these events is replication independent; (ii) mutations occur at a similar frequency during both leading and lagging strand synthesis, in both the PolI/PolIII and PolIII/PolIII replication modes, respectively; and (iii) instability occurs exclusively during PolIII-mediated lagging strand synthesis. These possibilities will be further discussed in the paragraph below under Mechanism of instability.

Influence of the LPMR system:
Genetic and in vitro studies have provided evidence supporting the correction of small heterologies by the E. coli mismatch repair system (reviewed in FRIEDBERG et al. 1995 ). It is clearly established that bulges containing up to three bases can be repaired as efficiently as certain base substitution mismatches. Nevertheless, larger bulges that have been shown to be resistant to such a repair process were not located in the context of repetitive sequence (KRAMER et al. 1984 ; CARRAWAY and MARINUS 1993 ). In our plasmid-based system, the effect of the LPMR system is to strongly stabilize the dinucleotide repeats: in its absence, the overall increase of instability increases from 2- to 3-fold for the self-complementary (GpC)₁₅ and (ApT)₂₄ sequences to >10-fold for the (GpT/ApC)-containing plasmids (see Table 1). The observed stabilization is effective on both the expansions and the deletions (Table 1).

In both of the self-complementary sequences, (GpC)₁₅ and (ApT)₂₄, only short events (SDEs and SEEs) are increased in the mutS strain, confirming that the perfect hairpin structures that are believed to give rise to longer expansion or deletion events in that sequence context are not processed by the mismatch repair system (FISHEL et al. 1986 ).

Suprisingly, for both (GpT/ApC) and (ApG/CpT) dinucleotide repeats, we found that long events are also increased in the mutS derivative (see Figure 2). Thus, for the (GpT/ApC) sequences, bulges of at least 6 bp are repaired by the LPMR system. Moreover, the mismatch repair-dependent correction seems more efficient when these bulges are present on the nascent strand (see Table 2). Indeed, long expansion events of up to three to four dinucleotide repeats observed in the mutS strain are completely absent when the LPMR system is active. In addition, analysis of the frequency of LDEs resulting from the processing of bulges of the same size on the template strand shows that, in the wild-type strain, they are diminished by factors of only 2 and 5 for the (GpT) and (ApC) orientation, respectively, as compared to the mutS strain (Figure 2 and Figure 4).

While analysis of the role of the LPMR system on the (ApG/CpT) insert demonstrates that LPMR also corrects bulges or structures that can lead to long deletion events, no bias of correction between template and nascent strands is observed for this insert. Nevertheless, a higher efficiency of correction of structures present on the template strand is observed in the (ApG) orientation, since the frequency of LDEs is reduced eightfold by the action of the LPMR system in the (ApG) orientation as compared to a twofold reduction in the (CpT) orientation (Table 2). This observation can be interpreted simply by considering that the bulges or the structures involved in the formation of LDEs are more efficiently recognized by the mutS protein in the (ApG) orientation than in the (CpT) orientation.

Influence of transcription:
In theory, DNA transcription can interfere with instability of repeat sequences in different ways. DNA replication and transcription can occur simultaneously on the same DNA molecule, leading to eventual head-on or codirectional collisions of the RNA polymerase with the replication fork. Such collisions can lead to an increase of illegitimate recombination resulting in deletions within plasmid vectors in E. coli (JAWORSKI et al. 1989 ; VILETTE et al. 1995 , VILETTE et al. 1996 ), and transcription has been shown to moderately destabilize simple repetitive sequences in yeast (WIERDL et al. 1996 ). On the other hand, transcription can modulate the formation of secondary structures that are thought to play a key role in the mechanisms of instability (PEARSON and SINDEN 1996 ). As an example, it has recently been shown that active transcription increases the frequency of deletion within stretches of CTG/CAG repeat sequences (BOWATER et al. 1997 ). To analyze the influence of transcription on the instability of these dinucleotide repeat sequences, plasmids pUCL+ containing the dinucleotide tracts were used to transform the mutS derivative of JM103 strain under conditions where transcription from the lacZ' promoter was either repressed by the LacI repressor or induced by IPTG (see MATERIALS AND METHODS). Our results suggest that the level of instability is not significantly affected in any of the dinucleotide repeats by induction of transcription (Figure 3 and Figure 4; Table 2). It should be noted that in the same experimental system and under the same induction conditions, the effect of induction of transcription on the instability of CTG/CAG triplet sequences was much more dramatic (S. SCHUMACHER, I. PINET, R. P. P. FUCHS and M. BICHARA, unpublished results).

Mechanisms of instability:
The current model of DNA repeat instability involves DNA polymerase slippage. This model predicts that the secondary structure that can be adopted by the repeat sequences either as single-stranded or as double-stranded DNA should play a key role in the instability of such sequence, first by slowing down the rate of progression of the ongoing polymerase (WEAVER 1984 ) and subsequently by influencing the conformation or the stability of the bulge formed by the unpaired repeats, thus altering the frequency of occurrence of the deletion or expansion events.

PolI-dependent mechanism of instability within the self-complementary sequences (GpC)₁₅ and (ApT)₂₄: Among the dinucleotide repeats, only the two self-complementary repeats (GpC)₁₅ and (ApT)₂₄ can adopt hairpin structures when single stranded. It is striking to observe that the instability within the only two sequences that can adopt perfect hairpin structures is more dependent upon the PolI/PolIII mode of replication. As PolI is known to promote foldback mechanisms upon encountering such structures (BACKMAN et al. 1978 ; PAPANICOLAOU and RIPLEY 1989 , PAPANICOLAOU and RIPLEY 1991 ; RIPLEY 1990 ), it is tempting to postulate that most of the deletions observed within these two inserts result from foldback of the nascent strand, followed by a misalignment mediated by a palindromic structure during replication of the leading strand DNA template. Indeed, in a similar system, we have been able to demonstrate that the only LDEs occurring within a (GpC) tract whose mechanism has been unambiguously determined result from such a process (BICHARA et al. 1995 ). The specificity of this mechanism toward self-complementary sequences, together with the potentially extreme thermal stability of GpC-containing duplexes, may explain the very high instability of these sequence repeats.

Deletion within (GpT/ApC) and (ApG/CpT) sequences is likely to occur during lagging strand synthesis: Although the absence of dependence of the instability of the other inserts [(GpT/ApC) and (ApG/CpT)] on the PolI/PolIII mode of replication can be formally interpreted in different ways (see Influence of the mode of replication, above), it seems likely that the observed deletions are occurring during the replication of the lagging strand. Indeed, as deletions between direct repeats separated by palindromic sequences have been shown to occur preferentially during lagging strand synthesis in E. coli (TRINH and SINDEN 1991 , TRINH and SINDEN 1993 ), it has been suggested that secondary structures giving rise to long mutation events are formed primarily during the replication of the lagging strand template (MCMURRAY 1995 ; WELLS 1996 ; WELLS et al. 1998 ).

Instability within (ApG/CpT) sequences: a role for H-DNA structure? The difference in the long expansion events observed between the (ApG) and the (CpT) orientation may argue in favor of a specific mechanism in the (ApG) orientation. Due to the polypurine-polypyrimidine nature of the (ApG/CpT) tract, such a mechanism could involve the transient formation of an H-DNA structure that is composed of triple-stranded and single-stranded regions (HANVEY et al. 1988 ; HTUN and DAHLBERG 1989 ). In this secondary structure, which has been demonstrated to exist for (ApG/CpT) sequence repeats on plasmids (HTUN and DAHLBERG 1989 ; PANYUTIN and WELLS 1992 ), the DNA duplex is disrupted and the polypyrimidine strand is folded back, enabling it to progress down in the major groove of the other half of the repeat. According to this model, the donated third strand is always the pyrimidine-containing strand, which can associate with purines in the acceptor region, thus forming H-DNA. In the context of DNA replication, the foldback reaction of the pyrimidine-containing strand would lead to a structure favored by the presence of a single-stranded stretch of polypyrimidine in the (ApG) orientation during lagging strand synthesis. This could in turn explain the difference between the frequency of long expansion events as measured for the (ApG) and the (CpT) orientation.

Conclusion:
Microsatellite sequences are highly polymorphic regions of the genome. It is generally accepted that the degree of instability of such repeats is directly related to the length of the perfect repeat. In this work, we show that for a given length, dinucleotide microsatellite instability is highly dependent on the sequence of the dinucleotide. We propose that polymerase slippage, which is the mechanism of choice to explain such instability, is influenced by the polymerase involved (DNA polI vs. DNA polIII in our experiments) and by the possibility for each dinucleotide repeat to form sequence-specific secondary structures. Our results also suggest that the molecular basis for the recognition of bulges by the LPMR system may be different, depending on whether the bulge is formed within a repeat or within a random sequence.

	ACKNOWLEDGMENTS

We thank I. B. Lambert for helpful critical reading of the manuscript. S. Schumacher was financially supported in part by a grant from the Association pour la Recherche sur le Cancer.

Manuscript received July 13, 1999; Accepted for publication November 2, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ALBERTINI, A. M., M. HOFER, M. P. CALOS, and J. H. MILLER, 1982  On the formation of spontaneous deletions: the importance of short sequence homologies in the generation of large deletions. Cell 29:319-328[Medline].

BACKMAN, K., M. BETLACH, H. W. BOYER, and S. YANOFSKI, 1978  Genetic and physical studies on the replication of ColE1-type plasmids. Cold Spring Harbor Symp. Quant. Biol. 43:69-76.

BICHARA, M., S. SCHUMACHER, and R. P. FUCHS, 1995  Genetic instability within monotonous runs of CpG sequences in Escherichia coli.. Genetics 140:897-907[Abstract].

BOWATER, R. P., A. JAWORSKI, J. E. LARSON, P. PARNIEWSKI, and R. D. WELLS, 1997  Transcription increases the deletion frequency of long CTG.CAG triplet repeats from plasmids in Escherichia coli. Nucleic Acids Res. 25:2861-2868[Abstract/Free Full Text].

CARRAWAY, M. and M. G. MARINUS, 1993  Repair of heteroduplex DNA molecules with multibase loops in Escherichia coli. J. Bacteriol. 175:3972-3980[Abstract/Free Full Text].

DEBRAUWERE, H., C. G. GENDREL, S. LECHAT, and M. DUTREIX, 1997  Differences and similarities between various tandem repeat sequences: minisatellites and microsatellites. Biochimie 79:577-586[Medline].

FISHEL, R. A., E. C. SIEGEL, and R. KOLODNER, 1986  Gene conversion in Escherichia coli. Resolution of heteroallelic mismatched nucleotides by co-repair. J. Mol. Biol. 188:147-157[Medline].

FREUND, A. M., M. BICHARA, and R. P. P. FUCHS, 1989  Z-DNA forming sequences are spontaneous deletion hot-spots. Proc. Natl. Acad. Sci. USA. 86:7465-7470[Abstract/Free Full Text].

FRIEDBERG, E. C., G. C. WALKER and W. SIEDE, 1995 Long-patch mismatch repair in prokaryotes, pp. 369–376 in DNA Repair and Mutagenesis, edited by E. C. FRIEDBERG, G. C. WALKER and W. SIEDE. ASM Press, Washington, DC.

GORDENIN, D. A., T. A. KUNKEL, and M. A. RESNICK, 1997  Repeat expansion—all in a flap? Nat. Genet. 16:116-118[Medline].

HANVEY, J. C., M. SHIMIZU, and R. D. WELLS, 1988  Intramolecular DNA triplexes in supercoiled plasmids. Proc. Natl. Acad. Sci. USA 85:6292-6296[Abstract/Free Full Text].

HTUN, H. and J. E. DAHLBERG, 1989  Topology and formation of triple-stranded H-DNA. Science 243:1571-1576[Abstract/Free Full Text].

JACKSON, A. L., T. G. NEWCOMB, and L. A. LOEB, 1998  Origin of multiple mutations in human cancers. Drug Metab. Rev. 30:285-304[Medline].

JAWORSKI, A., J. A. BLAHO, J. E. LARSON, M. SHIMIZU, and R. D. WELLS, 1989  Tetracycline promoter mutations decrease non-B DNA structural transitions, negative linking differences and deletions in recombinant plasmids in Escherichia coli. J. Mol. Biol. 207:513-526[Medline].

JIRICNY, J., 1998  Replication errors: cha(lle)nging the genome. EMBO J. 17:6427-6436[Medline].

JOHNSON, R. E., G. K. KOVVALI, S. N. GUZDER, N. S. AMIN, and C. HOLM et al., 1996  Evidence for involvement of yeast proliferating cell nuclear antigen in DNA mismatch repair. J. Biol. Chem. 271:27987-27990[Abstract/Free Full Text].

KORNBERG, A., and T. A. BAKER, 1992 Plasmids and organelles, pp. 637–680 in DNA Replication, edited by A. KERNBERG and T. BAKER. W. H. Freeman, New York.

KRAMER, B., W. KRAMER, and H. J. FRITZ, 1984  Different base/base mismatches are corrected with different efficiencies by the methyl-directed DNA mismatch–repair system of E. coli. Cell 38:879-887[Medline].

KUNKEL, T. A., 1990  Misalignment-mediated DNA synthesis errors. Biochemistry 29:8003-8010[Medline].

LEVINSON, G. and G. GUTMAN, 1987  High frequency of short frameshifts in poly-CA/GT tandem borne by bacteriophage M13 in Escherichia coli K-12. Nucleic Acids Res. 15:5323-5338[Abstract/Free Full Text].

LOEB, L. A., 1998  Cancer cells exhibit a mutator phenotype. Adv. Cancer Res. 72:25-56[Medline].

MCMURRAY, C. T., 1995  Mechanisms of DNA expansion. Chromosoma 104:2-13[Medline].

PANYUTIN, I. G. and R. D. WELLS, 1992  Nodule DNA in the (GA)37.(CT)37 insert in superhelical plasmids. J. Biol. Chem. 267:5495-5501[Abstract/Free Full Text].

PAPANICOLAOU, C. and L. S. RIPLEY, 1989  Polymerase-specific differences in the DNA intermediates of frameshift mutagenesis. In vitro synthesis errors of Escherichia coli DNA polymerase I and its large fragment derivative. J. Mol. Biol. 207:335-353[Medline].

PAPANICOLAOU, C. and L. S. RIPLEY, 1991  An in vitro approach to identifying specificity determinants of mutagenesis mediated by DNA misalignments. J. Mol. Biol. 221:805-821[Medline].

PEARSON, C. E. and R. R. SINDEN, 1996  Alternative structures in duplex DNA formed within the trinucleotide repeats of the myotonic dystrophy and fragile X loci. Biochemistry 35:5041-5053[Medline].

RIPLEY, L. S., 1990  Frameshift mutations: determinants of specificity. Annu. Rev. Genet. 24:189-213[Medline].

SCHUMACHER, S., R. P. P. FUCHS, and M. BICHARA, 1998  Expansion of CTG repeats from human disease genes is dependent upon replication mechanisms in Escherichia coli: the effect of Long Patch Mismatch Repair revisited. J. Mol. Biol. 279:1101-1110[Medline].

SIA, E. A., R. S. JINKS, and T. D. PETES, 1997  Genetic control of microsatellite stability. Mutat. Res. 383:61-70[Medline].

STRAND, M., T. A. PROLLA, R. M. LISKA, and T. D. PETES, 1993  Destabilization of tracts of simple repetitive DNA in yeast by mutations affecting DNA mismatch repair. Nature 365:274-276[Medline].

STRAND, M., M. C. EARLEY, G. F. CROUSE, and T. D. PETES, 1995  Mutations in the MSH3 gene preferentially lead to deletions within tracts of simple repetitive DNA in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 92:10418-10421[Abstract/Free Full Text].

STREISINGER, G., Y. OGADA, J. EMRICH, J. NEWTON, and A. TSUGITA et al., 1966  Frameshift mutations and the genetic code. Cold Spring Harbor Symp. Quant. Biol. 31:77-84[Abstract/Free Full Text].

TRINH, T. Q. and R. R. SINDEN, 1991  Preferential DNA secondary structure mutagenesis in the Lagging strand of replication in E. coli. Nature 352:544-547[Medline].

TRINH, T. Q. and R. R. SINDEN, 1993  The influence of primary and secondary DNA structure in deletion and duplication between direct repeats in Escherichia coli.. Genetics 134:409-422[Abstract].

VILETTE, D., S. D. EHRLICH, and B. MICHEL, 1995  Transcription-induced deletions in Escherichia coli plasmids. Mol. Microbiol. 17:493-504[Medline].

VILETTE, D., S. D. EHRLICH, and B. MICHEL, 1996  Transcription-induced deletions in plasmid vectors: M13 DNA replication as a source of instability. Mol. Gen. Genet. 252:398-403[Medline].

WARTER, J. M. and C. TRANCHANT, 1998  Hereditary neurological diseases caused by amplification of triplet repetitions. Presse Med. 27:376-381.

WEAVER, D. T., 1984  The role of palindromic and non-palindromic sequences in arresting DNA synthesis in vitro and in vivo. J. Mol. Biol. 180:961-986[Medline].

WELLS, R. D., 1996  Molecular basis of genetic instability of triplet repeats. J. Biol. Chem. 271:2875-2878[Free Full Text].

WELLS, R. D., A. BACOLLA, and R. P. BOWATER, 1998  Instabilities of triplet repeats: factors and mechanisms. Results Probl. Cell. Differ. 21:133-165[Medline].

WIERDL, M., C. N. GREENE, A. DATTA, S. JINKS-ROBERTSON, and T. D. PETES, 1996  Destabilization of simple repetitive DNA sequences by transcription in yeast. Genetics 143:713-721[Abstract].

WOOSTER, R., A.-M. CLETON-JANSEN, N. COLLINS, J. MANGION, and R. S. CORNELIS et al., 1994  Instability of short tandem repeats (microsatellites) in human cancers. Nat. Genet. 6:152-156[Medline].

YANG, Y. and W. MASKER, 1996  Instability of repeated dinucleotides in bacteriophage T7 genomes. Mutat. Res. 354:113-122[Medline].

This article has been cited by other articles:

				S. N. Shah and K. A. Eckert Human Postmeiotic Segregation 2 Exhibits Biased Repair at Tetranucleotide Microsatellite Sequences Cancer Res., February 1, 2009; 69(3): 1143 - 1149. [Abstract] [Full Text] [PDF]

				A. L. Seyfert, M. E. A. Cristescu, L. Frisse, S. Schaack, W. K. Thomas, and M. Lynch The Rate and Spectrum of Microsatellite Mutation in Caenorhabditis elegans and Daphnia pulex Genetics, April 1, 2008; 178(4): 2113 - 2121. [Abstract] [Full Text] [PDF]

				R. D. Wells, R. Dere, M. L. Hebert, M. Napierala, and L. S. Son Advances in mechanisms of genetic instability related to hereditary neurological diseases Nucleic Acids Res., July 8, 2005; 33(12): 3785 - 3798. [Abstract] [Full Text] [PDF]

				R. Dere, M. Napierala, L. P. W. Ranum, and R. D. Wells Hairpin Structure-forming Propensity of the (CCTG{middle dot}CAGG) Tetranucleotide Repeats Contributes to the Genetic Instability Associated with Myotonic Dystrophy Type 2 J. Biol. Chem., October 1, 2004; 279(40): 41715 - 41726. [Abstract] [Full Text] [PDF]

				D. Shinde, Y. Lai, F. Sun, and N. Arnheim Taq DNA polymerase slippage mutation rates measured by PCR and quasi-likelihood analysis: (CA/GT)n and (A/T)n microsatellites Nucleic Acids Res., February 1, 2003; 31(3): 974 - 980. [Abstract] [Full Text] [PDF]

				A. Pluciennik, R. R. Iyer, P. Parniewski, and R. D. Wells Tandem Duplication. A NOVEL TYPE OF TRIPLET REPEAT INSTABILITY J. Biol. Chem., September 8, 2000; 275(37): 28386 - 28397. [Abstract] [Full Text] [PDF]