Viewpoints Origin-Dependent Inverted-Repeat Amplification: A Replication-Based Model for Generating Palindromic Amplicons

Bonita J. Brewer*, Celia Payen, M. K. Raghuraman, Maitreya J. Dunham Department of Genome Sciences, University of Washington, Seattle, Washington, United States of America

Introduction (dsDNA) break and implicate DNA fusions composed of alleles from both homologs (either homologous or non-homologous), of one of the parents in a 2:1 ratio, Models proposed to explain the gener- Break-Induced Replication (BIR), Micro- consistent with the hypothesis that the ation of palindromic (or quasipalindromic) homology/Microsatellite-Induced Repli- event occurred in a meiotic or a pre- structures during segmental amplification cation (MMIR), and/or inverted or direct- meiotic division [8,14,15]. In all cases of almost invariably begin with double- ly repeated sequences that adopt unusual de novo triplications with an inverted stranded DNA breaks that are repaired secondary structures for their repair [4]. central copy, the distal portion of the in different ways—for example, by end-to- From the molecular analysis [7] of a yeast chromosome was retained. This finding end fusion at short inverted repeats, by strain that contains amplified copies of the appears to eliminate models such as the non-allelic homologous recombination at gene for the high affinity sulfur transporter, Breakage-Fusion-Bridge model of McClin- low copy repeats, or by break-induced SUL1 [3], we derived a new general model tock [17], at least in their simplest forms, replication at regions of microhomology. that explains the generation of interstitial as models that invoke a dsDNA break However, the specific class of amplicons tandem inverted repeat arrays of chromo- cannot easily explain the retention of distal that consists of interstitial inverted tripli- some segments in yeast and in human sequences. cations has no completely satisfactory cancers, and of de novo congenital inverted explanation. By examining the molecular triplications and other chromosomal rear- Saccha- The SUL1 Amplicon: A Yeast structure of a specific amplicon in rangements. We propose that cells commit romyces cerevisiae, we derived a model that Model for Human Inverted a singular error in replication: the ligation does not require an initiating double- Triplications of the nascent leading strand to the nascent stranded break but rather invokes an lagging strand at the replication fork. This Subjecting wild-type yeast to long-term underappreciated potential error in rep- model can potentially explain the origin of growth in medium that is limiting for lication to explain the generation of an many palindromic rearrangements and sulfur selects for cells that have amplified initial hairpin-cappedlinearintermedi- their structural, enzymatic, and genetic the SUL1 (high affinity sulfur transporter) ate. The model furthermore can ex- requirements. locus along with variable amounts of plain the final structure and the pathway for forming this and other types of flanking DNA [3]. To understand the amplicons in both yeast and humans. A Unique Class of Genomic chromosomal structure of one particular Because the model requires the presence Rearrangements amplification event, Araya et al. [7] of both an origin of replication and sequenced the genome of a strain selected short, closely spaced, flanking inverted In the past dozen years, examples of under sulfur limitation. The novel junction repeats, we call this model Origin- patients with various types of developmen- sequences they identified—occurring at 7- Dependent Inverted-Repeat Amplifica- tal and physical abnormalities have been bp, closely spaced, inverted repeats tion (ODIRA). found to harbor de novo triplications with (Figure 1A) in the nearby genes CTP1 an inverted central copy for regions of and PCA1 (Figure 1B)—allowed them to Background chromosomes 2, 3, 5, 7, 9, 10, 12, 13, and deduce the structure of the amplicon: a 56 15 [8–16]. In three of the cases, where the tandem array of alternating head-to-head/ Exposure to environmental stress often parent of origin could be determined, the tail-to-tail copies of an approximately 11- selects for cells that have amplified genes inverted triplication was found to be kb region that contains SUL1 and the involved in the amelioration of that stress. Sometimes the connection between the Citation: Brewer BJ, Payen C, Raghuraman MK, Dunham MJ (2011) Origin-Dependent Inverted-Repeat gene amplification and stress makes intu- Amplification: A Replication-Based Model for Generating Palindromic Amplicons. PLoS Genet 7(3): e1002016. itive sense, such as the amplification of the doi:10.1371/journal.pgen.1002016 gene for dihydrofolate reductase when Editor: Susan M. Rosenberg, Baylor College of Medicine, United States of America yeast or mammalian cells are treated with Published March 17, 2011 methotrexate [1,2]; in other cases, the link Copyright: ß 2011 Brewer et al. This is an open-access article distributed under the terms of the Creative is less obvious [3]. To explain the Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, mechanisms involved in the localized provided the original author and source are credited. amplification of specific genomic loci, Funding: BJB, MKR, and MJD are supported by a grant from the NIH (GM018926); MJD is a Rita Allen Scholar. models have been proposed based on the The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the DNA structures of the end products of manuscript. amplification [4–6]. Many of the current Competing Interests: The authors have declared that no competing interests exist. models begin with a double strand DNA * E-mail: [email protected]

PLoS Genetics | www.plosgenetics.org 1 March 2011 | Volume 7 | Issue 3 | e1002016 adjacent origin of replication, ARS228 replication mechanisms that invoke an alternating head-to-head and tail-to-tail (Figure 1C). Southern blot analysis con- initial dsDNA break or the 39 end of repeats and a terminal deletion would be firmed the inverted nature of the 11-kb single-stranded DNA (ssDNA) [4]. The recovered. However, comparing this pro- tandem repeats of the SUL1 locus [7], and presence of short inverted repeats flanking posed structure to the chromosome actu- array comparative genome hybridization the rearrangement breakpoints of the ally recovered in the haploid strain used (CGH) confirmed the retention of distal SUL1 amplicon might suggest mechanisms for the sulfur-limited selection, it is clear chromosome II sequences [3]. This SUL1 that involve the formation of hairpins that BFB cycles cannot readily explain this amplicon therefore contains several im- through intrastrand annealing in the particular SUL1 amplification event as the portant features of the human inverted exposed ssDNA in one of the parental distal sequences were retained. In addi- triplication syndromes and provides a strands at a replication fork or the tion, BFB cannot easily explain the human model for understanding the formation of extrusion of a cruciform in duplex DNA, triplications with an inverted center copy this type of amplification event. leading ultimately to a hairpin-capped as BFB is inherently an intrachromosomal dsDNA break. Replication of the hair- event and the three triplications where the Applying Existing Models to pin-capped linear would generate an iso- parent of origin was studied clearly Explain the SUL1 Amplicon dicentric chromosome (with the hairpin at included DNA from both homologs of its center) that would then be subject to one of the parents. We explored all of the Many models have been proposed to Breakage-Fusion-Bridge (BFB) cycles [17]. existing models in a similar way, but were explain genomic rearrangements. Such Upon capture of a telomere at the unable to explain simultaneously the models include recombination, repair or resulting break, a stable chromosome with generation of an uneven number of copies

Figure 1. The chromosomal context for SUL1 amplification. (A) The inverted repeat sequences in CTP1 and PCA1 that define the breakpoints of a specific SUL1 amplification event [7]. (B) The structure of the wild type SUL1 locus that includes the nearby origin of replication, ARS228. (C) The inferred structure of the head-to-head/tail-to-tail 56SUL1 amplification product recovered after selective growth of a haploid yeast strain in medium limiting for sulfur. doi:10.1371/journal.pgen.1002016.g001

PLoS Genetics | www.plosgenetics.org 2 March 2011 | Volume 7 | Issue 3 | e1002016 of amplified genes in an alternating head- the annealed circular molecule (Fig- dimeric plasmid integrates into a chromo- to-head/tail-to-tail tandem configuration, ure 2Aiv and Av), a linear duplex with some by homologous recombination. The the perfect reuse of both proximal and hairpins at both ends (a ‘‘dog bone’’; fact that the creation and integration of distal break sites, the retention of distal Figure 2Avi) is released. Because the the circular intermediate do not need to chromosomal segments, and the creation displaced fragment contains the origin occur in the same cell cycle greatly of genetically mixed amplicons through ARS228, the ‘‘dog bone’’ could be con- increases the chances of recovering the any of the existing models that used a verted to a dimeric circular molecule by final chromosomal amplicon. The high DNA break as the initiating event. replication in the next S-phase density of potential origins in the yeast and (Figure 2Avii). human genomes (yeast, one every ,20 kb, An Origin-Dependent Inverted- Up to this point there are three OriDB at http://www.oridb.org/index. Repeat Amplification Model interesting features of the model: first, php; human, one every ,68 kb; [24]) Explains the SUL1 Amplicon within the dimeric circle are the inverted and the frequency of closely spaced SUL1 genes and the rearrangement break- (#65 bp) inverted 7 bp repeats (one every Intrigued by the inclusion of a potential points that satisfy the sequencing results of ,250 bp; BJB, unpublished; based on origin of replication (ARS228) in the SUL1 Araya et al. [7]; second, there were no random scans of 100 kb segments of the amplicon (Figure 1), we wondered whether dsDNA breaks, hairpin cleavages, or DNA yeast genome using the ‘‘Palindrome’’ the presence of bidirectional forks might repair processes required; and third, the program at http://mobyle.pasteur.fr/cgi- play a role in the amplification process presence of the dimeric circle confers a bin/portal.py?form=palindrome) suggest beyond the proposed mechanisms of selective advantage on the cell because that amplification by this mechanism need break-induced replication (BIR), microho- that cell now has three copies of SUL1. not be limited to specific loci. mology-microsatellite-induced replication Missegregation of the dimeric circle can (MMIR), fork stalling and template switch- cause multiple copies of the circle to Extension to Other ing (FoSTeS), or serial replication slippage accumulate, providing a further selective Amplification and Genome (SRS) [4,18–20]. The short inverted advantage. At some later time, the ampli- Rearrangement Events in Yeast repeats and their close spacing (i.e., within fication event can be stabilized by the and Humans the size of eukaryotic Okazaki fragments) integration of one of the dimeric plasmids could permit an aberrant replication back into the chromosomal SUL1 locus by The simplicity of our model makes it intermediate to form if one or both of conventional homologous recombination appealing, but is there existing evidence to the replication forks regressed by just a few (Figure 2C). To achieve the five copies of support it? We have characterized a base pairs. In this scenario, the 39 end of SUL1 [7], two independent integrations of second independent SUL1 amplicon that the leading strand of a replication fork the inverted-repeat, dimeric circular mol- mimics the features of the sequenced SUL1 initiated at the origin ARS228 (Figure 2Ai ecule would be required. Other ways to amplicon [7], but with different potential and Aii) becomes detached from the generate the 56 copies include extrachro- inverted repeats at the junctions (C. Payen leading strand template after synthesis of mosomal concatemerization of the dimeric and M. J. Dunham, unpublished results). the second copy of the short inverted molecule—by the rolling circle model A search of the literature failed to uncover repeat has occurred. The detached end proposed by Futcher for the yeast 2- any model that includes all of the features then anneals to its complement in the micron plasmid [23]—before integration we have described; however, strand single-stranded portion of the lagging into the chromosome or by unequal sister switching from leading to lagging at a strand template (Figure 2Bi, Bii, and Biii). chromatid recombination after the initial replication fork has been proposed to In this new location, the 39 end primes integration of the dimeric circle. In the occur in bacteria [25,26]. We found synthesis on the lagging template case of the human triplication disorders, reports that transformation of both yeast (Figure 2Biv) and becomes ligated to the the ‘‘dog bone’’ intermediate generated and mammalian cells with hairpin capped adjacent Okazaki fragment of the lagging from one homolog in a division prior to linear molecules (‘‘dog bones’’) resulted in strand, creating a continuous DNA strand meiosis would replicate to generate the the expected dimeric inverted circular between the two nascent strands at the dimeric inverted circle and then integrate plasmids after replication in vivo [27,28]. fork—a ‘‘closed’’ fork (Figure 2Bv). by homologous recombination into the We have independently confirmed that We have illustrated the aberrant event other homolog during meiosis to generate ARS fragments capped with hairpins are occurring at the oppositely oriented forks the observed 2:1 allele ratio in the inverted converted to palindromic dimers in yeast on both sides of ARS228 (Figure 2Aiii), triplication chromosome. (M. M. Walker, M. K. Raghuraman, and thereby generating a self-complementary, In comparison to existing models, this B. J. Brewer, unpublished results). This circular DNA intermediate that is an- new model is relatively simple, requiring dimerization preserves the terminal se- nealed to the two parental strands but with only a single type of error in replication to quences of the capped ARS fragments, ‘‘closed’’ forks that are unable to progress generate the extrachromosomal interme- distinct from the dimerization of uncapped into the adjacent chromosomal regions. diate in amplification. The model de- linear fragments reported by Kunes et al. To complete replication of the chromo- mands (1) that the amplicon contain an [29]. We also searched for more examples some and permit segregation of the two origin of replication, both to generate the of gene amplification that could be parental chromosome strands, an ap- self-complementary single-stranded circu- explained by our model and found several proaching fork from a nearby origin on lar intermediate and to convert it to an instances in both yeast and mammalian either or both sides of the closed loop extrachromosomal dimeric, inverted-re- cells where the end products or the would facilitate the branch migration or peat plasmid; (2) that pairs of inverted intermediates are consistent with such a fork reversal at the closed forks through a repeats flank the origin in close enough ligation of leading to lagging strands at a combination of both topological and proximity to each other that each pair replication fork. enzymatic forces [21,22]. As the advanc- could lie within a single Okazaki-sized The first example involves amplification ing forks replicate through the region of single-stranded gap; and (3) that the of the gene for dihydrofolate reductase

PLoS Genetics | www.plosgenetics.org 3 March 2011 | Volume 7 | Issue 3 | e1002016 Figure 2. The Origin-Dependent Inverted-Repeat (ODIRA) model for Amplification of chromosomal segments. (A) An overview of the release of a closed circular, self-complementary intermediate that arises from aberrant replication. (B) Details of the mechanism that leads to ligation of the leading and lagging strands at short, closely spaced, inverted repeats (IRs; labeled as a a9 and b b9). (C) Replication and reinsertion of the inverted dimeric amplicon into the genome by homologous recombination. See text for detailed explanations. doi:10.1371/journal.pgen.1002016.g002

PLoS Genetics | www.plosgenetics.org 4 March 2011 | Volume 7 | Issue 3 | e1002016 (DFR1 in yeast and DHFR in Chinese of this finding is that during the genera- strands at the fork. Subsequent resolution hamster ovary cells). A subset of indepen- tion of the CUP1/SFA1 isochromosome, of the closed fork by replication/branch dent DHFR amplification events in Chi- chromosome V did not suffer a double- migration and addition of a telomere to nese hamster ovary cells contains chromo- stranded break. the broken end would generate the same somally integrated repeats of alternating The generation of the extrachromosom- structure that is usually attributed to orientations that include one or more al DFR1, ADH4, and CUP1/SFA1 ampli- breakage of isodicentric chromosomes by replication origins in each repeat [2], a cons can be easily explained by our model, BFB. It is interesting that ‘‘inv del dup’’ pattern very similar to the chromosomally as each contains one or more origins of chromosomes were the predominant class amplified SUL1 locus of yeast. The replication and appropriately placed short of rearranged chromosomes found by structures of yeast chromosomally ampli- inverted repeats. Amplification of DFR1 Narayanan et al. [32] when selecting for fied DFR1 amplicons were not deter- would require ‘‘closure’’ at both of the the loss of a marker that was distal to the mined; however, one methotrexate-resis- diverging forks (Figure 3-I), similar to what Alu inverted repeats on yeast chromosome tant survivor maintained the amplified we have proposed for SUL1, while ampli- V. copies of DFR1 as extrachromosomal 11- fication of ADH4 or CUP1/SFA1 would kb circular molecules composed of an require only a single event in the fork Replication Delays and Gene inverted dimer of the DFR1 gene and the moving toward the centromere (Figure 3- Amplification adjacent origin of replication ARS1524 [1]. II). In all four of these cases an acentric, While the authors did not sequence the extrachromosomal, nearly perfect palin- In our model for inverted amplicons, a junctions, several examples of short invert- dromic DNA molecule, either circular or closed fork can form when both copies of a ed repeats occur in the genome at the linear, is the result. Similar examples can short inverted repeat lie within a single- margins of the amplified region. In all be found in mammalian cells: hairpin- stranded gap on the lagging strand of a respects, this circular inverted dimer of capped linear fragments are maintained as replication fork. Therefore, as long as the DFR1 exactly conforms to the expelled palindromic extrachromosomal tiny epi- space between the inverted repeats is not and replicated, extrachromosomal mole- somes (ETEs; [28]); double-minute chro- greater than the Okazaki gap on the cule predicted by our model (Figure 2Avii mosomes [33], isochromosomes [34], and lagging strand, both the length of the and Figure 3-I). homogeneously staining regions (HSRs; inverted repeat and the amount of time A second example from the yeast [35,36]) with inverted repeat architecture the repeats persist in a single-stranded literature is the amplification of the have all been recovered from tumor cell form would influence the probability of ADH4 gene in adh1 cells that had been lines; and well-characterized isochromo- forming a closed fork at that particular treated with antimycin A [30,31]. In this somes in humans occur at regions that position. The 320-bp inverted Alu se- case, the amplicon was most frequently contain large inverted repeats [37–39]. quences that were placed centromere found as an acentric isochromosome While many of the extrachromosomal proximal to the SFA1 and CUP1 genes (Figure 3-II) of approximately 40 kb molecules described so far lack centro- on the left arm of yeast chromosome V encompassing the terminal ,20 kb of the meres, we wish to point out that our model increased the formation of extrachromo- left arm of chromosome VII. In this could also provide the starting point for somal amplicons between 250- and terminal segment of chromosome VII are the Breakage-Fusion-Bridge cycle de- 11,000-fold, depending on the percent two potential origins of replication (likely scribed by Barbara McClintock [17] if identity between the repeats, and a ARSs at 8 and 17 kb on chromosome VII; just the fork moving away from a centro- 25,000-fold increase in the formation of OriDB, http://www.oridb.org/) that lie mere were to experience ligation of the ‘‘inv dup del’’ chromosomes when the on either side of the ADH4 gene. The leading strand to the lagging strand repeats were perfect matches [32]. The junctions of independent isolates were (Figure 3-III). Subsequent expulsion and degree of identity between the repeated mapped by restriction digestion and replication of this hairpin would create an Alus would certainly influence the proba- Southern blotting and found to lie in a isodicentric chromosome that would form bility of cross-fork annealing, but the roughly 2-kb region [30] that contains a bridge at anaphase, be broken at presence of the inverted repeat has also more than ten pairs of interrupted short cytokinesis, and undergo repair by fusion been shown to slow progression of the inverted repeats. in the next cycle. The chromosome replication fork as much as 6-fold [42], A third example of an extrachromo- becomes stable when the broken end providing more time for the fork to somal inverted amplicon similar to the acquires telomeric sequences either by de ‘‘close.’’ case of ADH4 just described was generat- novo telomere addition or by recombina- Other methods of slowing fork progres- ed from an artificial construct on the left tion with another chromosome [40]. sion would also be expected to increase the arm of chromosome V in haploid strains There are many examples of such chro- formation of closed forks. Two recent of yeast [32]. The CUP1 and SFA1 genes, mosomes in the clinical literature of de yeast papers describe dicentric, palin- along with inverted human Alu sequences novo chromosomal abnormalities [41]— dromic chromosomes that were created separated by a 12-bp spacer, were insert- chromosomes that end in an inverted by interfering with replication fork pro- ed near the CAN1 gene, and clones duplication but are missing the terminal gression in yeast. Mizuno et al. [43] placed resistant to copper and formaldehyde portion of the original chromosome (also inverted replication termination sequences were selected. The most common ampli- known as an ‘‘inv dup del’’ chromosome). at the ura4+ locus of fission yeast and con recovered consisted of an ,80-kb Our model also provides a second way followed the fate of the chromosome over inverted dimeric linear (Figure 3-II) with to generate ‘‘inv dup del’’ chromosomes time after induction of the fork-blocking the Alu sequences at the center and a copy without cycles of BFB (Figure 3-IV). After factor Rts1. Both acentric and dicentric of the native origin, ARS504,neareach generating a closed fork at the telomere- palindromic chromosomes were generated telomere. It should be noted that in all proximal fork, the fork proceeding toward at high frequency at inverted repeats cases, the cells also retained a full-length the centromere could suffer a single- within their artificial construct, but, to copy of chromosome V. The implication stranded break in one of the parental their surprise, no double-stranded breaks

PLoS Genetics | www.plosgenetics.org 5 March 2011 | Volume 7 | Issue 3 | e1002016 Figure 3. Palindrome formation and resolution by ODIRA generates a range of amplification products. In each of the examples I–IV, initiation of replication from an origin near the sequence labeled E generates bidirectional forks that have progressed through the flanking sequences D and F. In the four scenarios depicted, either one or both of the replication forks becomes ‘‘closed’’ by ligation of the leading strand to the lagging strand. Open arrows indicate the direction that flanking replication forks move to expel the ‘‘closed’’ fork intermediate by branch migration. (I) In this example, both forks ‘‘close.’’ Replication forks approaching from either direction (open arrows) will release the hairpin-capped linear that contains genes D, E, and F. Subsequent replication of this ‘‘dog bone’’ molecule generates a dimeric, palindromic circular molecule that can reintegrate through homologous recombination into the original chromosome, or into a homolog if one is present, or by random integration elsewhere in the genome. These steps are described in detail in Figure 2. (II) In this example, the fork closest to the centromere ‘‘closes’’ while the telomere proximal fork remains active, fusing with distal replicons to allow replication of segments G, H, and the right telomere. The fork moving outward from the centromere (open arrow) will dislodge a fragment that has a hairpin at one end and a telomere at the other. Subsequent replication results in an isochromosome containing genes D–H. These acentric chromosomes are relatively stable in yeast [30,32], and in human cells stability can be improved by the acquisition of a neocentromere [52] or by chromosome tethering [53,54]. (III) In this example, only the distal fork closes. Completion of replication by a fork from the telomere proximal side of the closed fork releases a similar intermediate as in (II). Replication of the fragment with a hairpin near gene F generates an isochromosome that contains the centromere and genes A–F. Cycles of BFB will occur until the break is healed by the addition of a telomere. (IV) This example begins as the one in (III); however, the fork near gene D suffers a ssDNA break. Repair of this break by ligation of the broken strand to the nascent strand generates a branched intermediate that can be resolved by a fork moving in from the region of genes G and H. Two aberrant chromosome fragments are generated by this resolution, both requiring telomere addition to become resistant to nucleases. The fragment containing the centromere is an ‘‘inv dup del’’ chromosome indistinguishable from a chromosome generated by BFB (III). The fate of the acentric fragment is expected to be similar to that of an acentric isochromosome. doi:10.1371/journal.pgen.1002016.g003 were detected as a precursor. Paek et al. mation of the isodicentric chromosome in form of aberrant template switching was [44] obtained an isodicentric palindromic various mutant strains, they ruled out any involved, although the models they pro- version of the budding yeast’s chromo- involvement of double-stranded break posed were topologically complex and/or some VII at naturally occurring inverted repair pathways, post-replication repair, lacked specific details. Our model of repeats by disrupting replication in a and break-induced replication. The au- origin-dependent, inverted-repeat amplifi- checkpoint mutant. By studying the for- thors of both papers suggested that some cation could be the common mechanism

PLoS Genetics | www.plosgenetics.org 6 March 2011 | Volume 7 | Issue 3 | e1002016 that explains both of these chromosomal simple way to generate a specific class of and isochromosomes; Figure 3). Second, rearrangements. inverted amplicons. To determine just the causative event is not a double- Applying replication stress to mamma- how frequent amplification might occur stranded break but is an error in replica- lian cells in culture, in the form of by our proposed mechanism requires a tion—transfer of the 39 end of the leading carcinogens and/or mutagens, has been better cataloging of the structure of the strand to the lagging strand template at the found to cause the release of circular amplified DNA. Array CGH and deep same fork. Third, the 39 end of the leading inverted-repeat intermediates from the sequencing can pinpoint regions of the strand does not need to cover much genome. For example, Cohen et al. [45] genome that are amplified with respect to territory in search of homology as the found that the origin region of an a reference genome, but they do not complementary lagging strand template is integrated SV40 genome in a Chinese distinguish between extrachromosomal just angstroms away. Fourth, there is no hamster cell line is expelled from the and integrated copies—nor do they deter- obvious need to relocate polymerases or chromosome as a circular inverted repeat mine the chromosomal location or orien- helicases at the fork to restart replication amplicon. They also observed, under the tation of the additional, integrated copies from the 39 end of the displaced strand, as same experimental treatments, extrachro- unless the novel junctions are specifically the lagging strand machinery would be mosomal circular molecules of genomic looked for among the non-aligning se- available for this purpose. Fifth, the DNA [46] that we predict would also quencing reads. More complete analysis of ‘‘closed’’ fork should be displaceable from contain an origin of replication. Cohen et both yeast and human amplicons is the parental strands by branch migration al. [45] proposed a mechanism called ‘‘U- definitely needed. In the clinical literature, brought about by the combination of the turn’’ replication in which the leading there is a recurring comment that tripli- enzymatic activities of the helicases and strand folds back on itself and primes a cations with inverted central copies are topoisomerases that travel with the fork second strand using the nascent leading vastly underreported and therefore under- that approaches from the neighboring strand as template. Although they did not appreciated (for example, [8]). While there origin and the positive supercoils that specifically predict the importance of an are striking structural similarities between accumulate ahead of it. Sixth, the dis- origin of replication or the short, closely events in yeast and human cells, the scale placed circle is an autonomously replicat- spaced inverted repeats, or suggest how of the chromosomal rearrangements is ing entity, so the creation of the interme- the open end of the hairpin would be vastly different. For example, the sizes of diate and its reintegration into the repaired, their U-turn model proposed the regions in the triplication disorders can chromosome need not be temporally that the hairpin essentially replicates itself be several megabases. Is it reasonable to coupled. Seventh, the model supplies an out of the chromosomal context. The open expect the displacing forks to be able to alternate method to McClintock’s BFB for end is then sealed by some unspecified travel such long distances? Clearly the mechanism and the hairpin capped linear generating ‘‘inv dup del.’’ And finally, answer is not known, but as a point of molecule is subsequently replicated to many of the steps in the model we have comparison, when BIR was first described create the dimeric, inverted, circular presented are experimentally testable— in yeast it seemed amazing that a single molecule. perhaps most easily in yeast where it is fork could traverse the entire length of a More recent studies of human cancers possible to select directly for desired yeast chromosome arm [49,50]. A second by genome-wide analysis of palindrome amplification events and where the start- way in which yeast and mammalian cells formation revealed a widespread increase ing constructs and genetic backgrounds differ is in their propensity to undergo in the frequency of palindromic sequences can be manipulated, but also during clonal homologous recombination in response to (detected by their ability to ‘‘snap-back’’ expansion of transformed mammalian after denaturation) that are sometimes double-stranded breaks: yeast is extremely cells in culture. associated with the ends of amplified proficient at mitotic homologous recombi- regions as inferred from array CGH nation, while in human cells non-homol- Acknowledgments [47,48]. While the authors did not distin- ogous events predominate [51]. To gen- We would like to apologize in advance to guish between chromosomal and extra- erate the inverted triplication disorders, the dimeric circle must undergo homolo- anyone whose work is germane but not found chromosomal palindromes, it is possible by us in our PubMed searches: we would gous recombination with the chromosome. that they were detecting the same type of welcome hearing from you. We are grateful for circular inverted dimeric molecules stud- However, because the human inverted the thoughtful and helpful comments of three ied by Lavi and colleagues [45,46]. triplication disorders occur in meiosis, anonymous reviewers who pointed out other homologous recombination might in fact examples in the literature for us to consider. We Conclusions and Future be favored. thank members of the Brewer/Raghuraman and Dunham labs for useful discussions and Directions Our pathway, in addition to being rather simple, is novel and noteworthy critical comments on the manuscript. In addi- There are many pathways—including for several reasons. First, it suggests a tion, we appreciate the feedback and insights of Carlos Araya, Stan Fields, Julia Sidorova, Ray repair, recombination, and replication— unifying mechanism for a diverse set of Monnat, Santhosh Girirajan, and Jim (‘‘dog that contribute to genome rearrange- gene amplification outcomes (tandem in- bone’’) Thomas. ments. Our model of Origin-Dependent verted repeats, inverted double minutes, Inverted-Repeat Amplification provides a terminal inverted duplication/deletions,

References 1. Huang T, Campbell JL (1995) Amplification of a 2. Ma C, Looney JE, Leu TH, Hamlin JL (1988) 3. Gresham D, Desai MM, Tucker CM, Jenq HT, circular episome carrying an inverted repeat of Organization and genesis of dihydrofolate reduc- Pai DA, et al. (2008) The repertoire and dynamics the DFR1 locus and adjacent autonomously tase amplicons in the genome of a methotrexate- of evolutionary adaptations to controlled nutrient- replicating sequence element of Saccharomyces resistant Chinese hamster ovary cell line. Mol limited environments in yeast. PLoS Genet 4: cerevisiae. J Biol Chem 270: 9607–9614. Cell Biol 8: 2316–2327. e1000303.

PLoS Genetics | www.plosgenetics.org 7 March 2011 | Volume 7 | Issue 3 | e1002016 4. Hastings PJ, Lupski JR, Rosenberg SM, Ira G translocate along a duplex. J Biol Chem 278: homologous recombination that maintains palin- (2009) Mechanisms of change in gene copy 49171–49182. dromes. Cell 138: 855–869. number. Nat Rev Genet 10: 551–564. 22. Olavarrieta L, Martinez-Robles ML, Sogo JM, 39. Scott SA, Cohen N, Brandt T, Warburton PE, 5. Vialard F, Mignon-Ravix C, Parain D, Stasiak A, Hernandez P, et al. (2002) Supercoil- Edelmann L (2010) Large inverted repeats within Depetris D, Portnoi MF, et al. (2003) Mechanism ing, knotting and replication fork reversal in Xp11.2 are present at the breakpoints of isodi- of intrachromosomal triplications 15q11-q13: a partially replicated plasmids. Nucleic Acids Res centric X chromosomes in Turner syndrome. new clinical report. Am J Med Genet A 118A: 30: 656–666. Hum Mol Genet 19: 3383–3393. 229–234. 23. Futcher AB (1986) Copy number amplification of 40. Yu S, Graf WD (2010) Telomere capture as a 6. Wang J, Reddy KS, Wang E, Halderman L, the 2 micron circle plasmid of Saccharomyces frequent mechanism for stabilization of the Morgan BL, et al. (1999) Intrachromosomal cerevisiae. J Theor Biol 119: 197–204. terminal chromosomal deletion associated with triplication of 2q11.2-q21 in a severely malformed 24. Cadoret JC, Meisch F, Hassan-Zadeh V, Luyten I, inverted duplication. Cytogenet Genome Res infant: case report and review of triplications and Guillet C, et al. (2008) Genome-wide studies 129: 265–274. their possible mechanism. Am J Med Genet 82: highlight indirect links between human replica- 41. Zuffardi O, Bonaglia M, Ciccone R, Giorda R 312–317. tion origins and gene regulation. Proc Natl Acad (2009) Inverted duplications deletions: underdi- 7. Araya CL, Payen C, Dunham MJ, Fields S (2010) Sci U S A 105: 15837–15842. agnosed rearrangements?? Clin Genet 75: Whole-genome sequencing of a laboratory- 25. Ahmed A, Podemski L (1998) Observations on 505–513. evolved yeast strain. BMC Genomics 11: 88. template switching during DNA replication 42. Voineagu I, Narayanan V, Lobachev KS, 8.DevriendtK,MatthijsG,HolvoetM, through long inverted repeats. Gene 223: Mirkin SM (2008) Replication stalling at unstable Schoenmakers E, Fryns JP (1999) Triplication 187–194. inverted repeats: interplay between DNA hairpins of distal chromosome 10q. J Med Genet 36: 26. Kugelberg E, Kofoid E, Andersson DI, Lu Y, and fork stabilizing proteins. Proc Natl Acad 242–245. Mellor J, et al. (2010) The tandem inversion Sci U S A 105: 9936–9941. 9. Eckel H, Wimmer R, Volleth M, Jakubiczka S, duplication in Salmonella enterica: selection drives 43. Mizuno K, Lambert S, Baldacci G, Murray JM, Muschke P, et al. (2006) Intrachromosomal unstable precursors to final mutation types. Carr AM (2009) Nearby inverted repeats fuse to triplication 12p11.22-p12.3 and gonadal mosai- Genetics 185: 65–80. generate acentric and dicentric palindromic cism of partial tetrasomy 12p. Am J Med Genet A 27. Cote AG, Lewis SM (2008) Mus81-dependent chromosomes by a replication template exchange 140: 1219–1222. double-strand DNA breaks at in vivo-generated mechanism. Genes Dev 23: 2876–2886. 10. Ounap K, Ilus T, Bartsch O (2005) A girl with cruciform structures in S. cerevisiae. Mol Cell 31: 44. Paek AL, Kaochar S, Jones H, Elezaby A, inverted triplication of chromosome 3q25.3Rq29 800–812. Shanks L, et al. (2009) Fusion of nearby inverted and multiple congenital anomalies consistent with 28. Harada S, Uchida M, Shimizu N (2009) Episomal repeats by a replication-based mechanism leads to 3q duplication syndrome. Am J Med Genet A high copy number maintenance of hairpin- formation of dicentric and acentric chromosomes 134: 434–438. capped DNA bearing a replication initiation that cause genome instability in budding yeast. 11. Reddy KS, Logan JJ (2000) Intrachromosomal region in human cells. J Biol Chem 284: Genes Dev 23: 2861–2875. triplications: molecular cytogenetic and clinical 24320–24327. 45. Cohen S, Hassin D, Karby S, Lavi S (1994) studies. Clin Genet 58: 134–141. 29. Kunes S, Botstein D, Fox MS (1990) Synapsis- Hairpin structures are the primary amplification 12. Rivera H, Bobadilla L, Rolon A, Kunz J, mediated fusion of free DNA ends forms inverted products: a novel mechanism for generation of Crolla JA (1998) Intrachromosomal triplication dimer plasmids in yeast. Genetics 124: 67–80. inverted repeats during gene amplification. Mol of distal 7p. J Med Genet 35: 78–80. 30. Dorsey M, Peterson C, Bray K, Paquin CE (1992) Cell Biol 14: 7782–7791. 13. Verheij JB, Bouman K, van Lingen RA, van Spontaneous amplification of the ADH4 gene in 46. Cohen S, Lavi S (1996) Induction of circles of Lookeren Campagne JG, Leegte B, et al. (1999) Saccharomyces cerevisiae. Genetics 132: 943–950. heterogeneous sizes in carcinogen-treated cells: Tetrasomy 9p due to an intrachromosomal 31.WaltonJD,PaquinCE,KanekoK, two-dimensional gel analysis of circular DNA triplication of 9p13-p22. Am J Med Genet 86: Williamson VM (1986) Resistance to antimycin molecules. Mol Cell Biol 16: 2002–2014. 168–173. A in yeast by amplification of ADH4 on a linear, 47. Tanaka H, Bergstrom DA, Yao MC, Tapscott SJ 14. Ungaro P, Christian SL, Fantes JA, Mutirangura A, 42 kb palindromic plasmid. Cell 46: 857–863. (2005) Widespread and nonrandom distribution Black S, et al. (2001) Molecular characterisation of 32. Narayanan V, Mieczkowski PA, Kim HM, of DNA palindromes in cancer cells provides a four cases of intrachromosomal triplication of Petes TD, Lobachev KS (2006) The pattern of structural platform for subsequent gene amplifi- chromosome 15q11-q14. J Med Genet 38: 26–34. gene amplification is determined by the chromo- cation. Nat Genet 37: 320–327. 15. Mercer CL, Browne CE, Barber JC, Maloney VK, somal location of hairpin-capped breaks. Cell 48. Tanaka H, Cao Y, Bergstrom DA, Kooperberg C, Huang S, et al. (2009) A complex medical 125: 1283–1296. Tapscott SJ, et al. (2007) Intrastrand annealing phenotype in a patient with triplication of 2q12.3 33. Wahl GM (1989) The importance of circular leads to the formation of a large DNA palindrome to 2q13 characterized with oligonucleotide array DNA in mammalian gene amplification. Cancer and determines the boundaries of genomic CGH. Cytogenet Genome Res 124: 179–186. Res 49: 1333–1340. amplification in human cancer. Mol Cell Biol 16. Harrison KJ, Teshima IE, Silver MM, Jay V, 34. Italiano A, Attias R, Aurias A, Perot G, Burel- 27: 1993–2002. Unger S, et al. (1998) Partial tetrasomy with Vandenbos F, et al. (2006) Molecular cytogenetic 49. Llorente B, Smith CE, Symington LS (2008) triplication of chromosome (5) (p14-p15.33) in a characterization of a metastatic lung sarcomatoid Break-induced replication: what is it and what is it patient with severe multiple congenital anomalies. carcinoma: 9p23 neocentromere and 9p23-p24 for? Cell Cycle 7: 859–864. Am J Med Genet 79: 103–107. amplification including JAK2 and JMJD2C. 50. Malkova A, Ivanov EL, Haber JE (1996) Double- 17. McClintock B (1941) The Stability of Broken Cancer Genet Cytogenet 167: 122–130. strand break repair in the absence of RAD51 in Ends of Chromosomes in Zea mays. Genetics 26: 35. Nonet GH, Carroll SM, DeRose ML, Wahl GM yeast: a possible role for break-induced DNA 234–282. (1993) Molecular dissection of an extrachromo- replication. Proc Natl Acad Sci U S A 93: 18. Payen C, Koszul R, Dujon B, Fischer G (2008) somal amplicon reveals a circular structure 7131–7136. Segmental duplications arise from Pol32-depen- consisting of an imperfect inverted duplication. 51. Sonoda E, Hochegger H, Saberi A, Taniguchi Y, dent repair of broken forks through two alterna- Genomics 15: 543–558. Takeda S (2006) Differential usage of non- tive replication-based mechanisms. PLoS Genet 36. Schreiner B, Greten FR, Baur DM, Fingerle AA, homologous end-joining and homologous recom- 4: e1000175. Zechner U, et al. (2003) Murine pancreatic tumor bination in double strand break repair. DNA 19. Chauvin A, Chen JM, Quemener S, Masson E, cell line TD2 bears the characteristic pattern of Repair (Amst) 5: 1021–1029. Kehrer-Sawatzki H, et al. (2009) Elucidation of genetic changes with two independently amplified 52. Murmann AE, Conrad DF, Mashek H, the complex structure and origin of the human gene loci. Oncogene 22: 6802–6809. Curtis CA, Nicolae RI, et al. (2009) Inverted trypsinogen locus triplication. Hum Mol Genet 37. Barbouti A, Stankiewicz P, Nusbaum C, duplications on acentric markers: mechanism of 18: 3605–3614. Cuomo C, Cook A, et al. (2004) The breakpoint formation. Hum Mol Genet 18: 2241–2256. 20. Chen JM, Chuzhanova N, Stenson PD, Ferec C, region of the most common isochromosome, 53. Kanda T, Otter M, Wahl GM (2001) Mitotic Cooper DN (2005) Intrachromosomal serial i(17q), in human neoplasia is characterized by a segregation of viral and cellular acentric extra- replication slippage in trans gives rise to diverse complex genomic architecture with large, palin- chromosomal molecules by chromosome tether- genomic rearrangements involving inversions. dromic, low-copy repeats. Am J Hum Genet 74: ing. J Cell Sci 114: 49–58. Hum Mutat 26: 362–373. 1–10. 54. Kanda T, Wahl GM (2000) The dynamics of 21. Kaplan DL, Davey MJ, O’Donnell M (2003) 38. Lange J, Skaletsky H, van Daalen SK, Embry SL, acentric chromosomes in cancer cells revealed by Mcm4,6,7 uses a ‘‘pump in ring’’ mechanism to Korver CM, et al. (2009) Isodicentric Y chromo- GFP-based chromosome labeling strategies. J Cell unwind DNA by steric exclusion and actively somes and sex disorders as byproducts of Biochem Suppl Suppl 35: 107–114.

PLoS Genetics | www.plosgenetics.org 8 March 2011 | Volume 7 | Issue 3 | e1002016