Genetics, Vol. 148, 1507-1524, April 1998, Copyright © 1998
Factors Affecting Inverted Repeat Stimulation of Recombination and Deletion in Saccharomyces cerevisiae
Kirill S. Lobacheva,b,
Boris M. Shor1,b,
Hiep T. Trana,
Wendy Taylor2,a,
J. Dianne Keen3,a,
Michael A. Resnicka, and
Dmitry A. Gordenina,b
a Laboratory of Molecular Genetics, National Institute of Environmental Health Sciences, Research Triangle Park, North Carolina 27709
b Department of Genetics, St. Petersburg State University, St. Petersburg, 199034 Russia
Corresponding author:
Dmitry A. Gordenin, Mail Drop D3-01, Laboratory of Molecular Genetics, National Institute of Environmental Health Sciences, 111 TW Alexander Dr., P.O. Box 12233, Research Triangle Park, NC 27709, gordenin{at}niehs.nih.gov (E-mail).
 | ABSTRACT |
|---|
Inverted DNA repeats are an at-risk motif for genetic instability that can induce both deletions and recombination in yeast. We investigated the role of the length of inverted repeats and size of the DNA separating the repeats for deletion and recombination. Stimulation of both deletion and recombination was directly related to the size of inverted repeats and inversely related to the size of intervening spacers. A perfect palindrome, formed by two 1.0-kb URA3-inverted repeats, increased intra- and interchromosomal recombination in the adjacent region 2,400-fold and 17,000-fold, respectively. The presence of a strong origin of replication in the spacer reduced both rates of deletion and recombination. These results support a model in which the stimulation of deletion and recombination by inverted repeats is initiated by a secondary structure formed between single-stranded DNA of inverted repeats during replication.
MANY genomes are known to contain hotspots for spontaneous and induced genetic changes. Inverted DNA repeats (IRs) were the first example of a motif, rather than a specific sequence, having a profound effect on genetic stability. It was discovered that long (485 bp and 1515 bp) palindromic sequences (perfect head-to-head IRs) are deleted at extremely high rates in Escherichia coli (COLLINS 1981 
; COLLINS et al. 1982 ; see also review in LEACH 1994 ). While short (less than 20 bp) palindromes and IRs separated by a unique spacer region are usually much more stable, they can be associated with the breakpoints of deletion mutations such as observed in E. coli (ALBERTINI et al. 1982 ; GLICKMAN and RIPLEY 1984 ) and in mammalian cells (see review in MEUTH 1989 ). Distantly separated long (more than 1 kb) IRs, as found in bacterial transposons Tn10 and Tn5, are also prone to deletion in bacteria (ROSS et al. 1979 ; BERG et al. 1981 ; EGNER and BERG 1981 ; FOSTER et al. 1981 ).
The mechanism of IR-stimulated deletion formation is generally acknowledged to involve an interaction between IRs (for reviews see ERLICH 1989 ; LEACH 1994 ). The initial step involves the formation of a stem-like (or a hairpin-like) structure in a single-strand DNA (ssDNA) of the IRs, possibly during replication. Deletions could then arise by various pathways. The hairpin could be cleaved by a structure-specific nuclease and the surrounding DNA could be end-joined, resulting in a deletion (DASGUPTA et al. 1987 ; ERLICH 1989 ; LEACH 1994 ; AKGUN et al. 1997 ). Alternatively, the hairpin could block DNA synthesis, leading to the slippage of DNA polymerase between fortuitous short repeats that might be present near the base of the stem (BERG et al. 1981 ; COLLINS 1981 ; EGNER and BERG 1981 ; FOSTER et al. 1981 ; Figure 1). In support of the latter mechanism, a stem structure formed in vitro blocks DNA synthesis (HUANG and HEARST 1980 ; HUANG et al. 1981 ; KAGUNI and CLAYTON 1982 ; LADUCA et al. 1983 ) and causes its own bypass through replication slippage (CANCEILL and EHRLICH 1996 ).
View larger version (11K):
In this window
In a new window
Download PPT slide
|
Figure 1.
—Replication model for IR-generated genomic rearrangements (based on GORDENIN et al. 1993 ). Presented is a strand of the LYS2 gene with an IR-insert undergoing replication. During replication of the IR sequence (long arrows), single-stranded regions of DNA could give rise to a stem-like secondary structure, thereby inhibiting extension of the 3'-end. This could lead to deletion or recombination. Deletion pathway: replication slippage between short direct repeats (arrows inside black boxes) at the base of secondary structure leads to deletion of the IR (precise excision). Recombination pathway: the stalled replication generates recombinagenic single-stranded ends which can initiate recombination with homologous LYS2 sequence. If recombination involves identical sequences in sister chromatids it will not result in rearrangements. If a repeat in the same or another chromosome is involved, recombination could lead to deletions and translocations.
|
|
Several results in bacteria support the view that IR-stimulated deletions arise by an interaction between IRs. First, the rate of IR-deletion is directly related to the repeat size (DASGUPTA et al. 1987 ; WILLIAMS and MULLER 1987 ; PEETERS et al. 1988 ; WESTON-HAFER and BERG 1989 ; SINDEN et al. 1991 ) and inversely related to the distance (spacer) between the repeats (SINDEN et al. 1991 ; CHALKER et al. 1993 ). Second, the IR-deletion rate is increased when there is a high likelihood of unwinding and forming a stable secondary structure due to the nucleotide composition of the IR and/or spacer (SINDEN et al. 1991 ; CHALKER et al. 1993 ). Third, cruciform structures formed by inverted repeats were detected in vivo using a psoralen crosslinking assay (ZHENG and SINDEN 1988 ; ZHENG et al. 1991 ; SINDEN et al. 1991 ). Support for interaction of IRs in ssDNA leading to deletions comes from in vivo evidence of extreme instability of IRs in ssDNA. This was obtained with the single-stranded fd bacteriophage (HERRMANN et al. 1978 ) or M13-based replicons ( JANNIERE and EHRLICH 1987 ; D'ALENCON et al. 1994 ) and with bacterial conjugation, during which the transferred DNA is single stranded (BERG et al. 1983 ; SYVANEN et al. 1986 ). Dependence of the palindrome deletion rate on orientation relative to replication origin was found to be in agreement with IR-stimulated replication slippage occurring predominantly in the lagging strand (TRINH and SINDEN 1991 ; ROSCHE et al. 1995 ), which is likely to have many single-stranded regions. Inverted repeats in bacteria also stimulate more complex rearrangements, such as formation of inverted (BI and LIU 1996 ; LIN et al. 1997 ) and direct (PINDER et al. 1997 ) dimers of plasmids. These can be explained by a template switching mechanism similar to replication slippage.
Deletions of IRs also occur in eukaryotes and they have been examined extensively in the yeast Saccharomyces cerevisiae. Replication is involved in IR-stimulated deletions, since mutations in DNA polymerases (DNA-Pol) 
and 
increase deletions of the 1.5-kb Tn5 IRs separated by a long (2.7 kb) spacer and deletions of palindromes or quasipalindromes (IRs separated by very short distance; GORDENIN et al. 1992 , GORDENIN et al. 1993 ; RUSKIN and FINK 1993 ). The strongest effect was observed for deletions of the Tn5 long IRs. The deletion rate was elevated as much as 1000-fold by mutations in DNA-Pol (GORDENIN et al. 1992 , GORDENIN et al. 1993 ), demonstrating that even relatively stable IRs have a large potential to undergo rearrangement. Genomes of higher eukaryotes contain many imperfect and nearly perfect IRs (WILSON and THOMAS 1974 ; CECH and HEARST 1975 ; DEININGER and SCHMID 1976 ; BELL and HARDMAN 1977 ; DEUMLING 1978 ; HARDMAN et al. 1979 ; HOUCK et al. 1979 ; KRAYEV et al. 1980 ; BIEZUNSKI 1981A , BIEZUNSKI 1981B ; RUSSELL and MANN 1986 ; SCHROTH and HO 1995 ) which could be a source of IR-stimulated genome instability. Frequent deletions of very long perfect palindromes were recently described in mice (COLLICK et al. 1996 ; AKGUN et al. 1997 ). Based on the location of the breakpoints, many of the deletions can be explained by a replication slippage mechanism. However, an alternative mechanism is needed to account for the class of deletions that do not remove the center of a palindrome. Such a class was observed for one of the constructs studied (AKGUN et al. 1997 ).
In yeast, IRs not only stimulate deletions but also increase homologous recombination in adjacent regions. We found that IRs elevate both interchromosomal (allelic) recombination (GORDENIN et al. 1993 ) and intrachromosomal recombination between homologous or diverged direct repeats (TRAN et al. 1997 ). This recombinagenic effect appears to be a general feature of IR motifs, since in yeast the terminal repeats of bacterial transposon Tn5 and inverted URA3 genes increase recombination in an adjacent sequence (GORDENIN et al. 1993 ; TRAN et al. 1997 ). Elevated intrachromosomal recombination of the sequences within the IRs was recently demonstrated for the IRs of a long (15 kb) palindromic transgene in mice (AKGUN et al. 1997 ).
We proposed (GORDENIN et al. 1993 ; TRAN et al. 1997 ) that arrest of the replication complex at the basal part of the stem formed by IRs in ss-DNA could initiate IR-associated recombination as well as deletions. Homologous recombination could be an alternative to replication slippage when DNA elongation is blocked by a noncanonical DNA structure (Figure 1). Arrest of the replication complex can lead to recombination in several different ways: (i) It was suggested (BIERNE and MICHEL 1994 ; KUZMINOV 1995 ) and recently demonstrated in E. coli (MICHEL et al. 1997 ) that impaired DNA strand elongation can lead to formation of a double-strand break (DSB). DSBs can also occur as a result of cleavage by structure-specific endonucleases as proposed by LEACH 1994 . (ii) The stalled 3' end of the nascent strand could separate from the template, forming a 3'-tail that could invade homologous double-stranded DNA (ds-DNA). This is reminiscent of the model suggested for bypass of a DNA lesion by means of recombination-dependent replication in bacteriophage T4 (FORMOSA and ALBERTS 1986 ). (iii) A single-stranded region that can occur near the secondary structure formed by IRs in the template strand could initiate a search for homology as proposed in TRAN et al. 1997 .
Homologous recombination stimulated by IRs can lead to a much larger variety of genome rearrangements than simple deletion of IRs, especially in genomes containing many repeats. Therefore, it is important to know the types of changes that can be stimulated by IRs and what types of IRs are prone to deletion and recombination. In this paper we investigate the mechanisms of IR-stimulated deletion and recombination and the potential for various IR motifs in yeast to cause such rearrangements. Among the factors examined are the location of an IR within a replicon, the size of an IR and the distance between IRs. In support of the replication model (Figure 1) we found that both deletions and recombination are more frequent as the length of repeat is increased and/or when the spacer is decreased. By varying these factors a hyperrecombinagenic IR-structure that increases recombination as much as 17,000-fold was identified. Even relatively small IRs have the potential to be at-risk motifs for recombination when DNA replication is altered, as shown with a DNA polymerase (DNA-Pol) mutant pol3-t.
| MATERIALS AND METHODS |
|---|
Strains and plasmids:
All strains used in this study (Table 1) are isogenic and are derived from pol3-t-DM-MAT lys2-Tn5-13 ura3-x leu2-2 trp1-
1 (GORDENIN et al. 1992 ). POL and pol3-t strains containing insertions of bacterial transposon Tn5, quasipalindrome InsH and nonpalindromic insertion InsE in the BamHI-XhoI region of LYS2 have been described previously (GORDENIN et al. 1993 ). The plasmids pACYC184::Tn5 and pLL12::Tn5-13 (GORDENIN et al. 1988 , GORDENIN et al. 1993 ) with insertion of Tn5 were used to change the length of IRs or spacer between them. (Nucleotide sequences and detailed descriptions of all plasmids constructed in this study are available upon request.) The length of the Tn5 IRs (1534 bp) was gradually decreased. The URA3 fragment flanked by inverted polylinkers from the pUC4K-IXX plasmid (BARANY 1985 ) was cloned between either HpaI, EheI, XhoI, Eco47 III, HindIII and Bgl II sites located symmetrically at a distance of 185, 323, 485, 705, 1115, or 1515 bp, respectively, from the external ends of the Tn5 (Figure 2 and Table 1). (Note: the size of the URA3 fragment varied between 1112 and 1206 bp, depending on the polylinker sites used for construction.) Those inserts were used to study the effect of IR size on deletion and recombination. The insert with 323-bp IRs and 1.2-kb URA3 spacer, used as a control in set experiments, was designated InsL. A set of plasmids with constant length of 1515-bp inverted repeats of Tn5 and varying spacer was constructed as follows. At the first step, the central Bgl II-Bgl II fragment of Tn5 was replaced by the 7233-bp BamHI-BamHI fragment from bacteriophage lambda DNA. Subsequently, the URA3 gene (1.2 kb) was cloned into the EcoRI site of lambda DNA resulting in 1515-bp inverted repeats separated by 8.5-kb spacer construction. The 8.5-kb spacer was shortened in two subsequent steps. First, the Bgl II-Bgl II 3103-bp fragment with the flanking 550-bp was deleted (the size of flanking deletion was confirmed by sequencing) which generated the 4.8-kb spacer. Second, the BspMII-PmaCI (1483 bp) deletion in the sequence of lambda DNA produced the 3.3-kb spacer. To estimate the effect of a replication origin on IR stability, the 1.5-kb fragment containing the TRP1 gene and the strong origin of replication ARS1 was cloned between 1515-bp and 185-bp inverted repeats using Bgl II and HpaI sites inside IS50 repeats. As a control, 1515- and 185-bp inverted repeats, separated by the same fragment of TRP1 gene but with an inactivated origin of replication, were constructed. For this purpose we used the mutated ARS1 from the plasmid Yrp12S9 (CELNIKER et al. 1984 ) which contains a 20-bp deletion in the A-domain completely destroying an ARS function. U-strains and OR-strains (Table 1) with URA3 or TRP1-marked derivatives of the Tn5 were created by one-step replacement of the Tn5-13 insert in the pol3-t-DM strain. Strains F5 and F60, containing 323-bp IRs with a 58-bp spacer, were created as follows. An XcaI-HindIII deletion inside the URA3 gene separating the 323-bp inverted repeats derived from Tn5 led to shrinkage of the distance between inverted repeats from 1.1 kb to 58 bp. The resulting quasipalindrome (named as InsQ) was recloned in centromere plasmid pFL39 (BONNEAUD et al. 1991 ) and subsequently, the generated plasmid was introduced into the U8 and U28 strains. Transformants were replica plated on media containing 5-fluoro-orotic acid (5-FOA; BOEKE et al. 1984 ), and ura3 mutants that resulted from a gene conversion transfer of the 323-bp quasipalindrome from the centromere plasmid into the chromosomal LYS2 gene were selected. In several cases the lys2 allele with the Tn5 insert was modified by replacement transformation in the strains carrying a lys2 duplication (see below). All replacements were confirmed by Southern analysis.
View larger version (22K):
In this window
In a new window
Download PPT slide
|
Figure 2.
—Tn5-based inserts in the yeast LYS2 gene. All inserts are located at the same site in the BamHI-XhoI region of LYS2 (GORDENIN et al. 1993 ). The position of the restriction sites and distances between them in LYS2 are shown, along with the orientation of LYS2 on chromosome II (arrowhead shows LYS2 transcription direction). The inverted repeats (gray) and flanking short direct repeats (black) are indicated as arrows; the spacer between the IRs is shown as a thick black line. (A) IRs with a long spacer. The Tn5-derivatives are flanked by the duplicated 9-bp sequence of LYS2. The size of the long inverted repeats in different constructs varied from 185–1535 bp. The inverted repeats of the Tn5 are separated by an internal spacer containing the KmR gene (2.7 kb; the initial Tn5 sequence), the S. cerevisiae TRP1 gene (1.5 kb), the S. cerevisiae URA3 gene (1.1 kb) or the URA3 gene containing fragments of bacteriophage lambda DNA of various sizes (up to 7.3 kb). Three series of Tn5-based inverted repeats were constructed in this study: (i) different size IRs (185, 323, 485, 705, 1115, 1515 bp) with a constant URA3 spacer (1.2 kb) (The 323-bp IR with a URA3 spacer was referred to as InsL.); (ii) constant size IRs (1515 bp) separated by different spacers (3.3, 4.8, 8.5 kb) that contained the URA3 gene and fragments of bacteriophage lambda DNA; and (iii) IRs (1515 and 185 bp) separated by the TRP1 gene (1.5 kb) containing a strong origin of replication ARS1 or a mutated ARS1. (B) IRs with a short spacer. The IRs are InsH, with a 69-bp IR and a 9-bp spacer, and InsQ, with a 323-bp IR and a 58-bp spacer. Both inserts are flanked by 9-bp direct repeats. (C) Nonpalindromic insert InsE. This insert resulted from an imprecise deletion of Tn5(URA3), leaving a small piece of the external end of the left Tn5-IR (GORDENIN et al. 1993 ). InsE is 61-bp long and contains 6-bp flanking direct repeats.
|
|
The URA3-based inserts in the LYS2 gene were developed from constructs described in TRAN et al. 1997 (Figure 3). The ADE2 gene (2241 bp) from plasmid pASZ11 (STOTZ and LINDER 1990 ) was inserted between ura3 repeats (containing a NcoI frameshift mutation) located in inverted (LIR) or direct orientation (DIR) or behind a single ura3-NcoI gene (SOLO). These constructs were placed in both orientations into the XhoI site of LYS2 on the plasmid and then transferred into the chromosomal LYS2 gene by a gene disruption procedure (ROTHSTEIN 1983 ), either into the Lys+ derivative of POL-DM strain, yielding T105, IR28, IS27 strains, or into the POL and pol3-t strains already carrying a lys2 duplication (see below). All strains with URA3 based inserts carried the ade2 EcoRI-StuI deletion allele and complete deletion of the URA3 gene (GORDENIN et al. 1993 ). The P1 strain carrying an insertion of the URA3 -palindrome was made by transformation of the POL-DM strain with a ligation mix that contained two fragments: (i) 5' LYS2+URA3 fragment—the 5' region (HpaI to XhoI sites) of LYS2 fused (XhoI/Nsi I) to the Nsi I-Bgl II region of URA3, and (ii) URA3+3' LYS2 fragment—the same region of URA3 fused (XhoI/Nsi I) to the XhoI site of the 3' region (XhoI to BstEII sites) of LYS2. The P30 strain was created by replacement of the ura3-NcoIADE2 insertion in strain IR28 with a single copy of the URA3 gene. The structure of all URA3-based insertions in the XhoI site of LYS2 was confirmed by Southern analysis.
View larger version (21K):
In this window
In a new window
Download PPT slide
|
Figure 3.
—URA3-based inserts in LYS2. PAL, a perfect palindrome consisting of URA3 inverted repeats (gray lines); LIR, long inverted repeats of the URA3 sequence (gray lines) separated by the ADE2 gene; DIR, the ADE2 gene flanked by direct repeats of URA3; SOLO, a single copy of both the URA3 and the ADE2 genes or just a single copy of URA3 (not shown). The DIR and SOLO constructs were inserted in both orientations. The NcoI site in URA3 that was mutated in some constructs is indicated. Arrowheads show direction of transcription for the URA3, ADE2, LYS2 genes.
|
|
Three mutant lys2 alleles cloned either into pFL26 or pFL34 integrative plasmids (BONNEAUD et al. 1991 ) were used in this study to create intrachromosomal and interchromosomal lys2 duplications (Figure 4). The lys2-8 allele derived from the pLL12-lys2-8 plasmid (NOSKOV et al. 1990 ) is an EcoRI-HindIII fragment (4795 bp) of the LYS2 gene containing a missense mutation Ser880 to Tyr880 due to a C
A base pair change located at position 2639 of the reading frame. The lys2-3' and lys2-5' alleles are 3'- and 5'-truncated copies of the LYS2 gene, corresponding to a 3486-bp EcoRI-BamHI fragment and a 1691-bp XhoI-HindIII fragment, respectively. These mutant alleles were cloned into pFL26 and pFL34 plasmids taking into account the orientation of LEU2, URA3 and LYS2 in the chromosome relative to the centromere, so that crossing-over would not create dicentrics incapable of propagation. Only this orientation enables detection of interchromosomal crossing-over. Uncut plasmids with lys2 mutant alleles were introduced into the strains containing different insertions in the LYS2 gene. Transformants with lys2 mutant alleles integrated into the LYS2, LEU2 or URA3 locus were selected and analyzed using genetic and Southern analysis. Yeast strains carrying lys2 heteroalleles are listed in Table 1.
Genetic and molecular procedures:
Genetic and molecular procedures were described previously (GORDENIN et al. 1991 –1993). Rates of recombination and deletion (precise excision) and their 95% confidence intervals were determined in fluctuation tests as previously described using 12–30 independent cultures (TRAN et al. 1995 ). Two rates were considered to be significantly different if the 95% confidence intervals for those rates did not overlap. Since some of the strains carried the temperature-sensitive mutation, pol3-t, all tests were performed at 25°. Temperature sensitive strains were maintained at 20°.
In order to determine the association of intrachromosomal and interchromosomal conversion with exchange, about 100 independent Lys+ recombinants were isolated and genetically characterized. For studies of intrachromosomal recombination, the isolated Lys+ recombinants were replica plated on complete and selective media lacking uracil or leucine, depending on which gene, URA3 or LEU2, was placed between the lys2 repeats.
Intrachromosomal conversion not associated with crossing-over leads only to changes inside lys2 repeats leaving the sequence between repeats intact (Leu+ or Ura+ phenotype of the Lys+ revertants). Crossing-over or single-strand annealing (HABER 1995 ) between direct lys2 repeats in the same chromosome results in deletion of one of the lys2 sequences together with plasmid DNA, which can be identified by the loss of the central LEU2 or URA3 marker (Leu- or Ura- phenotype of the Lys+ revertants; Figure 5A and Figure 6C). In the case of interchromosomal recombination, Lys+ recombinants were characterized according to their ability to papillate on media containing 5-FOA, as described in HARRIS et al. 1993 . Lys+ recombinants which arose by conversion not associated with crossing-over yielded many papillae due to frequent intrachromosomal recombination between the ura3-x and URA3 genes flanking one of the lys2 sequences located on chromosome V (see Figure 5). In contrast, Lys+ recombinants carrying reciprocal chromosome II:V translocations rarely papillate on 5-FOA media because the URA3 genes placed on different chromosomes recombine very rarely. A limited number of recombinants were checked by Southern blot analysis to prove the genetic changes corresponded to actual physical changes.
View larger version (16K):
In this window
In a new window
Download PPT slide
|
Figure 5.
—Ectopic location of lys2 repeats. (A) Interchromosomal recombination. Normal location of LYS2 is on chromosome II. Various inserts were created at this position. To create a lys2 duplication, plasmids carrying lys2 mutant alleles were integrated into the URA3 locus of chromosome V or into the LEU2 locus of chromosome III. The second lys2 allele is flanked by URA3 or LEU2 locus of chromosome III. The second lys2 allele is flanked by URA3 or LEU2 sequences (wild-type and mutant alleles). (B) Intrachromosomal recombination. Integration of plasmids into the LYS2 locus of chromosome II creates intrachromosomal lys2 repeats separated by either the URA3 or LEU2 genes. The lys2-x corresponds to a point mutation lys2-8 (analogous to point mutation shown in the figure), the lys2-D3' truncation, or the lys2-D3' truncation (see materials and methods and Figure 4). (It should be noted that, of the two possible ways of lys2-8 integration, centromere distal and centromere proximal, we used the centromere distal lys2-8 shown in the figure.) The lys2::Ins represents the lys2 allele containing an IR or a control insert (indicated as a triangle). pFL26 (5558 bp) and pFL (3799 bp) sequences are shown as a diagonally striped line.
|
|
View larger version (32K):
In this window
In a new window
Download PPT slide
|
Figure 6.
—Selective system to study the effect of IRs on deletion and recombination. One of the Tn5-based (Figure 2) or URA3-based (Figure 3) inserts was placed into the LYS2 gene. (A) Selection for deletion. Deletion of the insert and of one flanking 9 bp direct repeat (precise excision) restores the LYS2 reading frame on chromosome II. (B) Selection for interchromosomal recombination. The second mutant lys2 allele was inserted into chromosome III (LEU2 locus) or chromosome V (URA3 locus; see materials and methods and Figure 4 and Figure 5). Lys+ recombinants could occur by gene conversion of the insert with or without crossing-over, or infrequently by reciprocal recombination in the region of the LYS2 gene (at the "S" in the figure). The exchange between nonhomologous chromosomes leads to translocations which, for the exchanges involving chromosomes II and V, could be identified genetically (HARRIS et al. 1993 ), as described in materials and methods. (C) Selection for intrachromosomal recombination. Recombination between lys2 repeats in the same chromosome occurs by gene conversion either without or with crossing-over (identified by the loss of the central LEU2 or URA3 marker as described in materials and methods).
|
|
| RESULTS |
|---|
Experimental systems:
A series of isogenic haploid strains was developed to study deletions of IRs and recombination stimulated by IRs (Figure 6). The strains used to study deletions (Figure 6A) were Lys- and carried a lys2 allele with either the Tn5 IR-insert or various sizes of inserts that were derived from Tn5 (Figure 2) and retained the Tn5 external ends. Each insert was flanked by short (9 bp) direct repeats of the LYS2 sequence present at the ends of Tn5 (GORDENIN et al. 1992 ). Deletions removing the insert and one copy of the repeat can lead to Lys+ reversion. Since deletions of IRs separated by a long spacer were infrequent in wild type strains, these deletions were studied in strains carrying the pol3-t mutation which increases the deletion rate as much as 1000-fold (GORDENIN et al. 1992 , GORDENIN et al. 1993 ).
The Lys- haploid strains used to study recombination (Figure 6B and Figure C) contained two lys2 alleles in ectopic positions. One lys2 allele at the normal chromosome II position carried either an IR or a non-IR insert as a control. In addition to the Tn5-based IRs (Figure 2) we developed a series of IRs based on the yeast URA3 gene (Figure 3). This allowed us to address the generality of recombinagenic effects of IRs as well as to construct the long palindrome URA3-PAL (Figure 3 and MATERIALS AND METHODS). The second lys2 allele (Figure 4) carried a homologous sequence overlapping the site containing the insert in the first allele. In one group of strains, the second lys2 allele was placed as a direct repeat in the same chromosome and the lys2 repeats were separated by vector sequence carrying either a URA3 or a LEU2 marker (Figure 5B and Figure 6C). In the other group of strains, the second lys2 allele was placed in another chromosome, either chromosome V near the URA3 locus, or chromosome III near the LEU2 locus (Figure 5A and Figure 6B). The rate of appearance of Lys+ was 10–1000-fold more with the constructs allowing recombination (Figure 6B and Figure C) than with the constructs that could revert only by deletions (Figure 6A). (Two exceptions are described in the footnotes for Table 7.)
View this table:
In this window
In a new window
|
Table 4.
Influence of an ARS sequence on intrachromosomal recombination and deletion stimulated by the Tn5 LIR
|
|
To estimate the effect of IRs on recombination, the relative and absolute increases in recombination rate were determined when the allele in chromosome II contained or lacked an IR. As a non-IR control for Tn5-based IRs, we used the 61-bp InsE insertion derived from the Tn5 allele (Figure 2) and located at the same site as Tn5-13 (GORDENIN et al. 1992 ). As a non-IR control for the URA3-based IRs with the ADE2 spacer, we used two different orientations of direct repeats of URA3 separated by the ADE2 gene and two different orientations of unique URA3 and ADE2 sequences (DIR and SOLO in Figure 3). A single URA3 sequence served as a non-IR control for the URA3-based perfect palindrome URA3-PAL.
Some of the Lys+ recombinants were due to conversion associated with crossing-over (Figure 6B and Figure C) or due to rare events of reciprocal recombination in the region denoted by an "S" in Figure 6. Exchanges between nonhomologous chromosomes leading to translocations, or between repeats in the same chromosome leading to deletions, were identified both genetically and by Southern analysis, as described in MATERIALS AND METHODS. We established earlier the synergistic interaction between the recombinagenic effects of IRs and a defect in DNA-Pol, pol3-t (GORDENIN et al. 1993 ; TRAN et al. 1997 ). [The pol3-t mutation was later identified as a missense mutation Gly641 (GAC) to Ala641 (AAC) (TRAN et al. 1997 ) near the conserved region VI putative nucleotide binding and active site domains of the enzyme (WANG 1996 ).] Therefore studies of recombination were performed in isogenic POL and pol3-t strains.
IRs can stimulate inter- and intrachromosomal recombination:
We determined the effect of two types of IRs, Tn5 and URA3-ADE2, on both inter- and intrachromosomal recombination (Table 2 and Table 3). There was as much as a 50-fold stimulation in recombination by IRs, that for interchromosomal recombination was independent of the chromosomal location of ectopic sequence, and for intrachromosomal recombination was independent of the marker between intrachromosomal repeats. The absolute increase in recombination rate caused by IRs was much greater in pol3-t than in POL strains carrying wild-type DNA Pol genes. The synergistic interaction between IRs and the pol3-t mutation was previously shown for allelic recombination (GORDENIN et al. 1993 ) and for intrachromosomal recombination between homologous and homeologous ade2 repeats (TRAN et al. 1997 ).
Recombination events that are associated with exchange (crossing-over or single-strand annealing) will lead to rearrangements (deletions and translocations). The overall incidence of exchange associated with IR-stimulated recombination was the same as for recombination where IRs were absent (22–85% for intrachromosomal recombination and 5–13% for interchromosomal recombination). For the intrachromosomal recombination, the variation of the exchange fraction was statistically significant. This could be due to the differences in particular constructs and/or the presence of the pol3-t mutation. Based on the presented results we cannot draw a conclusion about the origin of such variation. Nevertheless, the higher level of associated exchange for all cases of intrachromosomal recombination compared to interchromosomal recombination is consistent with the results of JINKS-ROBERTSON et al. 1993 using other recombination systems.
Effects of an internal replication origin on IR-stimulated recombination and deletion:
According to the replication model described in Figure 1, the likelihood of deletion and recombination is increased when an IR is encountered during nascent strand elongation (see DISCUSSION). We, therefore, investigated the consequences of placing a bi-directional replication origin between IRs. The TRP1 fragment containing an ARS1 sequence was inserted between the 1515-bp IRs of Tn5 (Figure 2). The ARS1 functions as a strong replication origin when placed at various chromosomal locations (FERGUSON and FANGMAN 1992 ; BREWER and FANGMAN 1993 ). There was an approximately five-fold decrease in the deletion rate in the pol3-t strain caused by ARS1 (TRP1 ARS1+; Table 4) as compared to the ARS1- control in which the same TRP1 fragment had a 20-bp deletion that inactivates ARS1 (TRP1 ARS1-; see MATERIALS AND METHODS). The TRP1 ARS1+ spacer also caused a statistically significant decrease in the rate of intrachromosomal recombination in both POL3 and pol3-t strains. There was no effect of ARS1 on recombination when either TRP1 ARS1+ or TRP1 ARS1- was placed between the short (185 bp) versions of the Tn5 IRs. These repeats lost their ability to induce recombination (see next section). The lack of an ARS1+ effect when placed between non-recombinagenic IRs supports the conclusion that the ARS1+ suppression of recombination is specific to recombination events induced by IRs.
Dependence of IR-stimulated recombination and deletions on the size of IRs and the distance between IRs:
Based on the model in Figure 1, increasing the length of IRs or decreasing the spacer between IRs should provide more opportunity to form a secondary structure, and thus increase the rate of IR stimulated deletion or recombination. Therefore, several Tn5 derivatives were constructed in which the size of the IR or the distance between IRs was varied (Table 1; Figure 2). There was a gradual increase of both deletion and intrachromosomal recombination rates either with increase in IR size or with decrease in the distance between IRs (Table 5 and Table 6). Interactions between IRs may still be possible for the smallest (185 bp) IRs and for the largest spacer (8457 bp) based on the incidence of deletions in these constructs. However, since recombination rates when these inserts were present did not exceed the rates for the non-IR insert InsE (compare with the data in Table 2), most of these recombination events were not initiated by IR interaction.
Based on results described in the previous section, closely spaced IRs should be very recombinagenic. We, therefore, investigated the effect of Tn5- and URA3-based closely spaced IRs on inter- and intrachromosomal recombination (Table 7). Reducing the distance between IRs to only 58 bp results in relatively short IRs (323 bp) becoming recombinagenic (compare InsQ to InsL and InsE). The strongest recombinagenic effect of the closely spaced 323-bp IRs (InsQ) was observed in the pol3-t background. Reduction of the spacer from 1.1 kb (InsL) to 58 bp (InsQ) caused a 15-fold increase in the interchromosomal, and a 24-fold increase in the intrachromosomal, recombination in the pol3-t background. This correlated with an approximately 50-fold higher deletion rate of InsQ compared with InsL in the pol3-t strain (data not shown). Even short (69 bp) closely spaced IRs of the quasipalindrome InsH (compare with InsE) stimulated intrachromosomal recombination (6-fold in the pol3-t background). Among the intrachromosomal recombinants induced by quasipalindrome InsH, 20% were associated with exchange resulting in deletions of the LEU2 marker between the lys2 repeats. No conclusion could be drawn about the effect of InsH on interchromosomal recombination, since reversion rates via deletions were very high (see footnotes b, c for Table 7).
Since the recombinagenic effect of IRs was greatly increased when the distance between IRs was decreased, we constructed a long perfect palindrome URA3-PAL (Figure 3). The final step in construction of URA3-PAL involved transformation of the ligated DNA directly into yeast (see MATERIALS AND METHODS), since long perfect palindromes cannot be propagated in bacteria (for review, see LEACH 1994 ). In order to confirm that the perfect palindrome structure did not undergo rearrangement in yeast, we checked by Southern analysis the size of the insert and the retention of the Bgl II site at the junction between the inverted repeats and the NdeI sites located close (144 bp) to the internal ends of repeats (data not shown). In the strains used to measure deletion (Figure 6A), reversions of the lys2::URA3-PAL allele were infrequent (<10-8). This was probably due to the IRs being flanked by only small (4 bp) direct repeats. We observed the frequency of at least 10-6 of Ura- 5-FOA-resistant mutants. This is several orders of magnitude higher than the expected frequency of two independent mutations inactivating both URA3 genes in the URA3-PAL. If these events represent imprecise excision (deletions) of the URA3-PAL, the rate of such deletions is 100-fold more than the rate of imprecise deletions of Tn5 in a POL strain, where 1.5-kb IRs were separated by a 2.7-kb spacer (GORDENIN et al. 1993 ). The rate of deletions for URA3-PAL could be even higher, since many deletions may alter only the center of the palindrome, leaving one or both URA3 genes functional. Deletions of URA3-PAL were not investigated further in this study. URA3-PAL was extremely recombinagenic, both for interchromosomal and intrachromosomal recombination (Table 7). The recombination rates were 17,000-fold higher for interchromosomal recombination, and 2400-fold higher for intrachromosomal recombination, than when the insert contained a single URA3. The recombinagenic effect of URA3-PAL was 50–340 times stronger than the recombinagenic effect of URA3-LIR separated by the 2.5-kb ADE2 spacer (compare with data in Table 3). The long palindrome was efficient in stimulating rearrangements, since 22% of the intrachromosomal recombinants (43 out of 198 Lys+ recombinants) had lost the DNA between the lys2 direct repeats.
| DISCUSSION |
|---|
DNA sequence motifs that are at-risk for genetic change in wild-type or mutation prone cells have been identified in various organisms. Among them are the following commonly occurring at-risk motifs: microsatellites, minisatellites, triplet repeats, short separated repeats, mirror repeats, and inverted repeats. IRs are an important class of at-risk motifs, prone to deletions, that have been extensively studied in bacteria. Relatively little is known about IR-stimulated deletions in eukaryotes and there have been only a few studies on recombination stimulated by IRs. We developed several genetic systems to understand the role of DNA structure and genetic factors in IR-stimulated deletion and recombination in yeast and to assess the potential for the instability caused by this at-risk motif.
Our results suggest common events in IR-induced deletion and various kinds of recombination, as well as a high potential for IRs to stimulate these events. We shall discuss our data in the framework of the replication model for IR deletions and IR-stimulated recombination.
Mechanism of IR-stimulated recombination and deletion in yeast:
In the replication model for IR-stimulated deletions and recombination (Figure 1), ssDNA regions are developed within the IRs that lead to the formation of duplexes and stem structures. We have suggested that single-strand regions can occur during replication (especially lagging strand) and they may be more extensive in mutants such as pol3-t (GORDENIN et al. 1992 ). [By way of example, we show the IRs that form a stem structure in Figure 7A (bottom part) as completely single-stranded, although the ssDNA could be interrupted by double-strand DNA as long as complementary ssDNA regions are sufficient to support stable association of IRs.] We suggest that the likelihood of long complementary ssDNAs occurring within both repeats would be greater if replication across the IRs were accomplished by a single replication complex (Figure 7A) than by two independent replication complexes moving in opposite directions (Figure 7B). Even if single-strand regions form in the lagging strand templates of both replication forks (not shown on Figure 7) they would not lead to intrastrand stem-like structures. In support of this, we have demonstrated that the presence of a strong replication origin between repeats suppresses IR deletions and IR-stimulated intrachromosomal recombination (Table 4). The suppressing effect is clearly due to the presence of the functional ARS1, since suppression of intrachromosomal recombination was observed only for recombination events stimulated by IRs. However, while there is suppression of both deletions and recombination by a functional ARS1, it is incomplete. This could be due to the ARS1 not always serving as an origin. The efficiency of the ARS1 between IRs as a replication origin can vary in different chromosomal contexts (lys2 duplication vs. single lys2) and between genotypes (POL3 vs. pol3-t). In this case both IRs would be replicated from the origin located upstream to the LYS2 gene (Figure 4 and Figure 7A), as determined by FREUDENREICH et al. 1997 and by TRAN et al. 1995 . Another possibility is that not all IR-stimulated rearrangements are sensitive to the replication origin between the IRs. A related observation concerning position of the replication origin has been made in E. coli. When the spacer between IRs of bacterial transposon Tn5 was replaced by a large fragment of pUC19 containing a unidirectional replication origin, the rate of IR deletions decreased more than 100-fold (KIL' et al. 1994 ).
View larger version (12K):
In this window
In a new window
Download PPT slide
|
Figure 7.
—The effect of position of replication origin on opportunity for IRs to form secondary structure. Inverted repeats are shown as thick black arrows. Open triangular arrowheads show 3' nascent strands. The direction of movement of the replication fork (dashed arrows) is indicated. (A) Initiation of replication from a distant origin (ori). (B) Initiation of replication from an origin located between IRs.
|
|
The observation that stimulation of deletion, as well as recombination, increased with size of IR (Table 5 Table 6 Table 7), excludes models where only a small region of an IR is genetically active. The increase in the rate of IR deletions with an increase in IR size provides further evidence for the IR motif, not just sequence, being able to stimulate deletions via interaction between repeats.
We also found that increasing the distance between IRs led to decreased rates of both deletion and recombination. In the framework of the replication model (Figure 1), this suggests that concerted formation of ssDNA in IRs becomes less likely with increased distance between the repeats.
IR-stimulation of rearrangements via homologous recombination is a general phenomenon:
IRs appeared capable of stimulating both intrachromosomal recombination leading to deletions and interchromosomal recombination leading to translocations. The lower incidence of associated exchange for the interchromosomal as compared to the intrachromosomal recombination is consistent with the results of JINKS-ROBERTSON et al. 1993 obtained in different genetic systems and backgrounds. The actual association with exchange for interchromosomal recombination may be higher, since half of the crossovers occurring in G2 would produce unbalanced translocations resulting in lethality and would thus not be recoverable in haploids (discussed in JINKS-ROBERTSON et al. 1993 ).
Based on our previous (GORDENIN et al. 1993 ; TRAN et al. 1997 ) and current results we conclude that regardless of the sequence (Tn5 terminal repeats and URA3 repeats) the IR is a recombinagenic at-risk motif. IR-stimulated recombination was found for nearly all sizes of IRs studied. Shorter IRs appeared to be highly recombinagenic only when separated by a relatively short spacer and when present in the pol3-t mutants.
IRs have a high potential for stimulating both deletions and recombination:
Since palindromes and quasipalindromes have the highest capability to form a secondary structure, these IRs are expected to be the most efficient in causing deletions and (or) recombination. This agrees with their being the most deletion-prone IRs in several yeast systems (HENDERSON and PETES 1993 ; RUSKIN and FINK 1993 ). Deletions of long palindromes in mice also occur frequently (COLLICK et al. 1996 ; AKGUN et al. 1997 ). It is noteworthy, that long (up to 15 kb) palindromes are maintained in mice although they are very unstable, while replicons with >150–200 bp palindromes in bacteria cannot be propagated (reviewed in ERLICH 1989 ; LEACH 1994 ). The long (1020 bp) perfect palindrome URA3-PAL in yeast also appeared capable of propagation [overall rate of deletions measured by loss of URA3 function was 10-6–10-5 (data not shown)]. It remains to be determined which factors are responsible for maintaining longer palindromes in eukaryotes.
While IRs can stimulate recombination and deletion, correlation between the two events is not complete. The quasipalindrome InsH was deleted about 1000-fold more frequently than Tn5 in POL and about 50 times more in the pol3-t background (GORDENIN et al. 1993 ; TRAN et al. 1995 ), while there was more stimulation of interchromosomal recombination by Tn5 as compared to InsH in the POL strains (Table 7). However, the InsH was more recombinagenic in the pol3-t background. These results appear, at first, to be inconsistent with our replication model since IR-stimulated genetic changes would be expected to increase with the likelihood of forming a secondary structure. We suggest that additional factors responsible for completing the genetic events are likely to influence the outcomes. For example, there is also the possibility of recombination between sister chromatids which would be undetected in the present systems (discussed by SARGENT et al. 1997 , and by ZOU and ROTHSTEIN 1997 ). The recombination pathway could depend on the type of IR. Similarly, the choice of deletion vs. recombination (Figure 1A and Figure B) could also depend on the IR structure. The systems described in this study should be useful in identifying genetic factors, such as pol3-t, that can influence the impact of the IR at-risk motif.
The fact that weakly recombinagenic IRs, i.e., InsQ and InsH, are strongly recombinagenic in pol3-t mutants, demonstrates that there are genetic backgrounds where relatively stable IRs (and possibly other at-risk motifs) can become unstable. There are motifs in the human genome that are analogous to InsQ and InsH IRs. The 323-bp IRs of InsQ (323 bp) are close in size to Alu repeats. These Alu repeats are abundant in the human genome and can be associated with rearrangements that lead to disease (CALABRETTA et al. 1982 ; LEHRMAN et al. 1985 , LEHRMAN et al. 1986 , LEHRMAN et al. 1987 ; WEINER et al. 1986 ; JALANKO et al. 1995 ; MACINA et al. 1995 ). Based on Alu-PCR (NELSON et al. 1989 ) and indirect physical methods (DEININGER and SCHMID 1976 ; DEININGER 1989 ), many Alu-repeats form closely spaced IRs although they are nonidentical. Sequencing of the human genome revealed examples of closely spaced Alu-IRs (MACINA et al. 1995 ; PAULY et al. 1995 ; UNFRIED et al. 1995 ; KIYOSAWA and CHANCE 1996 ; KODA et al. 1997 ) including some that are nearly palindromic (HANKE et al. 1995 ). Genetic factors that enhance recombination stimulated by InsQ in yeast could be homologous to human genetic factors that might increase rearrangements associated with Alu-IRs. These factors would be important in genomic stability and disease.
It is important to note that Alu-repeats, as well as other repeats in the human genome, are highly diverged. Thus, sequence divergence and genetic factors that might affect interactions between diverged sequences could affect the potential for IR-induced genome instability. In addition to diverged IRs formed by long repeats, some of the small repeats, microsatellites and minisatellites are analogous to diverged long IRs. The unstable triplet repeats CTG (CAG) or CGG (CCG) (WARREN 1996 ; WELLS 1996 ; MANDEL 1997 ) and minisatellites rich in dA-T dinucleotide (FEARON et al. 1990 ; YU et al. 1997 ) are capable of expansion and deletion, and are associated with chromosomal fragile sites and disease. These sequences can form imperfect hairpins analogous to the hairpins that can be adopted by a ssDNA of diverged inverted repeats (MITAS 1997 ). While IRs induce recombination between diverged repeats (TRAN et al. 1997 ), the influence of IR divergence within repeats on IR-induced instability has not been examined. Since the long unstable palindrome URA3-PAL causes up to 17,000-fold induction of recombination, this motif provides opportunities for investigating the consequences of DNA divergence.
The hyperrecombination effect of the URA3-PAL motif demonstrates the high potential of IR-processing mechanisms in stimulating rearrangements. This structure enhanced recombination up to 17,000-fold, creating the strongest recombinational hotspot identified. The level of recombination induced by URA3-PAL may indicate a highly reactive intermediate such as a double-strand break (DSB). DSBs are efficient at causing homologous recombination in mitotically dividing yeast cells (HABER 1995 ; RESNICK et al. 1995 ) and they are initiators of meiotic recombination (LICHTEN and GOLDMAN 1995 ). Based on indirect data it was proposed (BIERNE and MICHEL 1994 ; KUZMINOV 1995 ) that stalled elongation could result in the collapse of replication forks and DSB in E. coli. Later it was shown that arrested replication leads to DSBs in E. coli (MICHEL et al. 1997 ). A DSB in the vicinity of the 140-bp perfect palindrome was recently demonstrated in yeast during meiosis, although a role for replication remains to be established (NAG and KURST 1997 ). If a DSB arises at the secondary structure of an IR during replication, this lesion could irreversibly direct subsequent events toward a recombination pathway. If a DSB is not the cause of IR-stimulated recombination, another highly recombinagenic intermediate should exist to account for high recombination rates.
The replication model of IR-stimulated recombination (Figure 1) may be applicable to other kinds of DNA sequences capable of forming secondary structures interfering with replication, for example sequences capable of forming triplex DNA and Z-DNA. These motifs have been shown to be recombinagenic. We propose that there are mutations, such as the pol3-t defect in DNA-Pol described in this study, that can act synergistically with these motifs. Further studies in model systems such as yeast will lead to the identification of additional at-risk motifs and factors that can act synergistically to cause genome instability.
| FOOTNOTES |
|---|
1 Present address: Department of Microbiology and Immunology, State University of New York, Health Science Center at Brooklyn, Brooklyn, NY 11203. 
2 Present address: Duke University Medical School, Box 27550, Durham, NC 27708.
3 Present address: Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139.
| ACKNOWLEDGMENTS |
|---|
We are grateful to E. GOLOVANOVA for excellent and enthusiastic technical support of experiments performed in the D.A.G lab, to A. KATENEVA for assistance in the experiments, to Dr. J. L. CAMPBELL for the plasmid Yrp12S9, and to Dr. J. KIRCHNER, Dr. R. KOKOSKA and Dr. J. MASON for helpful comments on the manuscript. This work was supported by an International Research Grant from the Howard Hughes Medical Institute #75195-545401 (to D.A.G and M.A.R) and an Interagency Agreement from the U. S. Department of Energy DE-A105-94ER61940 (to M.A.R).
| LITERATURE CITED |
|---|
AKGUN, E., J. ZAHN, S. BAUMES, G. BROWN, and F. LIANG et al., 1997 Palindrome resolution and recombination in the mammalian germ line. Mol. Cell. Biol. 17:5559-5570[Abstract].
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].
BARANY, F., 1985 Single-stranded hexameric linkers: a system for in-phase insertion mutagenesis and protein engineering. Gene 37:111-123[Medline].
BELL, A. J. and N. HARDMAN, 1977 Characterization of foldback sequences in hamster DNA using electron microscopy. Nucleic Acids Res. 4:247-268[Abstract/Free Full Text].
BERG, D. E., C. EGNER, B. J. HIRSCHEL, J. HOWARD, L. JOHNSRUD et al., 1981 Insertion, excision, and inversion of Tn5. Cold Spring Harbor Symp. Quant. Biol. 45 (Pt. 1): 115–123.
BERG, D. E., C. EGNER, and J. B. LOWE, 1983 Mechanism of F factor-enhanced excision of transposon Tn5. Gene 22:1-7[Medline].
BI, X. and L. F. LIU, 1996 DNA rearrangement mediated by inverted repeats. Proc. Natl. Acad. Sci. USA 93:819-823[Abstract/Free Full Text].
BIERNE, H. and B. MICHEL, 1994 When replication folk stops. Mol. Microbiol. 13:17-23[Medline].
BIEZUNSKI, N., 1981a Structure and distribution of inverted repeats (palindromes). I. Analysis of DNA of Drosophila melanogaster. Chromosoma 84:87-109[Medline].
TOP
ABSTRACT
MATERIALS AND METHODS