Heterogeneous duplications in patients with Pelizaeus-Merzbacher disease suggest a mechanism of coupled homologous and nonhomologous recombination

We describe genomic structures of 59 X-chromosome segmental duplications that include the proteolipid protein 1 gene (PLP1) in patients with Pelizaeus-Merzbacher disease. We provide the first report of 13 junction sequences, which gives insight into underlying mechanisms. Although proximal breakpoints were highly variable, distal breakpoints tended to cluster around low-copy repeats (LCRs) (50% of distal breakpoints), and each duplication event appeared to be unique (100 kb to 4.6 Mb in size). Sequence analysis of the junctions revealed no large homologous regions between proximal and distal breakpoints. Most junctions had microhomology of 1-6 bases, and one had a 2-base insertion. Boundaries between single-copy and duplicated DNA were identical to the reference genomic sequence in all patients investigated. Taken together, these data suggest that the tandem duplications are formed by a coupled homologous and nonhomologous recombination mechanism. We suggest repair of a double-stranded break (DSB) by one-sided homologous strand invasion of a sister chromatid, followed by DNA synthesis and nonhomologous end joining with the other end of the break. This is in contrast to other genomic disorders that have recurrent rearrangements formed by nonallelic homologous recombination between LCRs. Interspersed repetitive elements (Alu elements, long interspersed nuclear elements, and long terminal repeats) were found at 18 of the 26 breakpoint sequences studied. No specific motif that may predispose to DSBs was revealed, but single or alternating tracts of purines and pyrimidines that may cause secondary structures were common. Analysis of the 2-Mb region susceptible to duplications identified proximal-specific repeats and distal LCRs in addition to the previously reported ones, suggesting that the unique genomic architecture may have a role in nonrecurrent rearrangements by promoting instability.     

A chromosomal rearrangement hotspot can be identified from population genetic variation and is coincident with a hotspot for allelic recombination

Insights into the origins of structural variation and the mutational mechanisms underlying genomic disorders would be greatly improved by a genomewide map of hotspots of nonallelic homologous recombination (NAHR). Moreover, our understanding of sequence variation within the duplicated sequences that are substrates for NAHR lags far behind that of sequence variation within the single-copy portion of the genome. Perhaps the best-characterized NAHR hotspot lies within the 24-kb-long Charcot-Marie-Tooth disease type 1A (CMT1A)-repeats (REPs) that sponsor deletions and duplications that cause peripheral neuropathies. We investigated structural and sequence diversity within the CMT1A-REPs, both within and between species. We discovered a high frequency of retroelement insertions, accelerated sequence evolution after duplication, extensive paralogous gene conversion, and a greater than twofold enrichment of SNPs in humans relative to the genome average. We identified an allelic recombination hotspot underlying the known NAHR hotspot, which suggests that the two processes are intimately related. Finally, we used our data to develop a novel method for inferring the location of an NAHR hotspot from sequence variation within segmental duplications and applied it to identify a putative NAHR hotspot within the LCR22 repeats that sponsor velocardiofacial syndrome deletions. We propose that a large-scale project to map sequence variation within segmental duplications would reveal a wealth of novel chromosomal-rearrangement hotspots.     

Mechanisms of tandem duplication in the Duchenne muscular dystrophy gene include both homologous and nonhomologous intrachromosomal recombination.

Three tandem duplications were previously identified in patients with Duchenne muscular dystrophy and were shown in each case to have a subset of dystrophin gene exons duplicated. The origin of these duplications was traced to the single X chromosome of the maternal grandfathers, suggesting that an intrachromosomal event (unequal sister chromatid exchange) was involved in the formation of these duplications. In the present study, a DNA segment containing the duplication junction and the normal DNA that corresponds to both ends of the duplicated region have been cloned. Subsequent mapping studies confirmed the tandem arrangement (head to tail) of these duplications and revealed their sizes to be 130 kb, approximately 300 kb, and 35-80 kb, respectively. Sequence analysis of the duplication junctions showed that one duplication was due to homologous recombination between two repetitive elements (Alu sequences) and the other two were due to recombination between unrelated nonhomologous sequences. In the latter cases, the preferred cleavage sites of the eukaryotic type I and II DNA topoisomerases were found at the junctions of these duplications, suggesting a possible role of these enzymes in the chromatid exchange events. This study provides the first insight into the molecular basis of gene duplications formed through unequal sister chromatid exchange in humans.     

Multiple cellular mechanisms prevent chromosomal rearrangements involving repetitive DNA

Repetitive DNA is present in the eukaryotic genome in the form of segmental duplications, tandem and interspersed repeats, and satellites. Repetitive sequences can be beneficial by serving specific cellular functions (e.g. centromeric and telomeric DNA) and by providing a rapid means for adaptive evolution. However, such elements are also substrates for deleterious chromosomal rearrangements that affect fitness and promote human disease. Recent studies analyzing the role of nuclear organization in DNA repair and factors that suppress non-allelic homologous recombination (NAHR) have provided insights into how genome stability is maintained in eukaryotes. In this review, we outline the types of repetitive sequences seen in eukaryotic genomes and how recombination mechanisms are regulated at the DNA sequence, cell organization, chromatin structure, and cell cycle control levels to prevent chromosomal rearrangements involving these sequences.     

Multiple cellular mechanisms prevent chromosomal rearrangements involving repetitive DNA

Repetitive DNA is present in the eukaryotic genome in the form of segmental duplications, tandem and interspersed repeats, and satellites. Repetitive sequences can be beneficial by serving specific cellular functions (e.g. centromeric and telomeric DNA) and by providing a rapid means for adaptive evolution. However, such elements are also substrates for deleterious chromosomal rearrangements that affect fitness and promote human disease. Recent studies analyzing the role of nuclear organization in DNA repair and factors that suppress non-allelic homologous recombination (NAHR) have provided insights into how genome stability is maintained in eukaryotes. In this review, we outline the types of repetitive sequences seen in eukaryotic genomes and how recombination mechanisms are regulated at the DNA sequence, cell organization, chromatin structure, and cell cycle control levels to prevent chromosomal rearrangements involving these sequences.     

Nebulin: A Study of Protein Repeat Evolution

Protein domain repeats are common in proteins that are central to the organization of a cell, in particular in eukaryotes. They are known to evolve through internal tandem duplications. However, the understanding of the underlying mechanisms is incomplete. To shed light on repeat expansion mechanisms, we have studied the evolution of the muscle protein Nebulin, a protein that contains a large number of actin-binding nebulin domains.Nebulin proteins have evolved from an invertebrate precursor containing two nebulin domains. Repeat regions have expanded through duplications of single domains, as well as duplications of a super repeat (SR) consisting of seven nebulins. We show that the SR has evolved independently into large regions in at least three instances: twice in the invertebrate Branchiostoma floridae and once in vertebrates.In-depth analysis reveals several recent tandem duplications in the Nebulin gene. The events involve both single-domain and multidomain SR units or several SR units. There are single events, but frequently the same unit is duplicated multiple times. For instance, an ancestor of human and chimpanzee underwent two tandem duplications. The duplication junction coincides with an Alu transposon, thus suggesting duplication through Alu-mediated homologous recombination.Duplications in the SR region consistently involve multiples of seven domains. However, the exact unit that is duplicated varies both between species and within species. Thus, multiple tandem duplications of the same motif did not create the large Nebulin protein.Finally, analysis of segmental duplications in the human genome reveals that duplications are more common in genes containing domain repeats than in those coding for nonrepeated proteins. In fact, segmental duplications are found three to six times more often in long repeated genes than expected by chance.     

Segmental duplications and copy-number variation in the human genome

The human genome contains numerous blocks of highly homologous duplicated sequence. This higher-order architecture provides a substrate for recombination and recurrent chromosomal rearrangement associated with genomic disease. However, an assessment of the role of segmental duplications in normal variation has not yet been made. On the basis of the duplication architecture of the human genome, we defined a set of 130 potential rearrangement hotspots and constructed a targeted bacterial artificial chromosome (BAC) microarray (with 2,194 BACs) to assess copy-number variation in these regions by array comparative genomic hybridization. Using our segmental duplication BAC microarray, we screened a panel of 47 normal individuals, who represented populations from four continents, and we identified 119 regions of copy-number polymorphism (CNP), 73 of which were previously unreported. We observed an equal frequency of duplications and deletions, as well as a 4-fold enrichment of CNPs within hotspot regions, compared with control BACs (P < .000001), which suggests that segmental duplications are a major catalyst of large-scale variation in the human genome. Importantly, segmental duplications themselves were also significantly enriched >4-fold within regions of CNP. Almost without exception, CNPs were not confined to a single population, suggesting that these either are recurrent events, having occurred independently in multiple founders, or were present in early human populations. Our study demonstrates that segmental duplications define hotspots of chromosomal rearrangement, likely acting as mediators of normal variation as well as genomic disease, and it suggests that the consideration of genomic architecture can significantly improve the ascertainment of large-scale rearrangements. Our specialized segmental duplication BAC microarray and associated database of structural polymorphisms will provide an important resource for the future characterization of human genomic disorders.     

The Contribution of Alu Elements to Mutagenic DNA Double-Strand Break Repair

Alu elements make up the largest family of human mobile elements, numbering 1.1 million copies and comprising 11% of the human genome. As a consequence of evolution and genetic drift, Alu elements of various sequence divergence exist throughout the human genome. Alu/Alu recombination has been shown to cause approximately 0.5% of new human genetic diseases and contribute to extensive genomic structural variation. To begin understanding the molecular mechanisms leading to these rearrangements in mammalian cells, we constructed Alu/Alu recombination reporter cell lines containing Alu elements ranging in sequence divergence from 0%-30% that allow detection of both Alu/Alu recombination and large non-homologous end joining (NHEJ) deletions that range from 1.0 to 1.9 kb in size. Introduction of as little as 0.7% sequence divergence between Alu elements resulted in a significant reduction in recombination, which indicates even small degrees of sequence divergence reduce the efficiency of homology-directed DNA double-strand break (DSB) repair. Further reduction in recombination was observed in a sequence divergence-dependent manner for diverged Alu/Alu recombination constructs with up to 10% sequence divergence. With greater levels of sequence divergence (15%-30%), we observed a significant increase in DSB repair due to a shift from Alu/Alu recombination to variable-length NHEJ which removes sequence between the two Alu elements. This increase in NHEJ deletions depends on the presence of Alu sequence homeology (similar but not identical sequences). Analysis of recombination products revealed that Alu/Alu recombination junctions occur more frequently in the first 100 bp of the Alu element within our reporter assay, just as they do in genomic Alu/Alu recombination events. This is the first extensive study characterizing the influence of Alu element sequence divergence on DNA repair, which will inform predictions regarding the effect of Alu element sequence divergence on both the rate and nature of DNA repair events.     

Junctions between genes in the haptoglobin gene cluster of primates.

To investigate the nature of the recombination that generated the haptoglobin three-gene cluster in Old World primates, we sequenced the region between the second gene (HPR) and the third gene (HPP) in chimpanzees (15 kb), as well as the region 3' to the cluster in humans (14 kb). Comparison to the previously sequenced human haptoglobin (HP) and HPR genes showed that the junction point between HP and HPR in humans (junction 1) was not identical to the junction point between the HPR and HPP genes of the chimpanzee (junction 2). An Alu sequence was found at each junction, but both Alu sequences lacked short direct repeats of the flanking genomic DNA. The lack of direct repeats implies that both junction Alu sequences are the products of recombination between different Alu elements. In addition, other insertion and deletion events are clustered in the regions near the junction Alu sequences. The observation that Alu sequences define the junctions between genes in the haptoglobin gene cluster emphasizes the importance of Alu sequences in the evolution of multigene families.     

Molecular structure and genetic stability of human hypoxanthine phosphoribosyltransferase (HPRT) gene duplications.

We have determined the genetic stability of three independent intragenic human HPRT gene duplications and the structure of each duplication at the nucleotide sequence level. Two of the duplications were isolated as spontaneous mutations from the HL60 human myeloid leukemia cell line, while the third was originally identified in a Lesch-Nyhan patient. All three duplications are genetically unstable and have a reversion rate approximately 100-fold higher than the rate of duplication formation. The molecular structures of these duplications are similar, with direct duplication of HPRT exons 2 and 3 and of 6.8 kb (HL60 duplications) or 13.7 kb (Lesch-Nyhan duplication) of surrounding HPRT sequence. Nucleotide sequence analyses of duplication junctions revealed that the HL60-derived duplications were generated by unequal homologous recombination between clusters of Alu repeats contained in HPRT introns 1 and 3, while the Lesch-Nyhan duplication was generated by the nonhomologous insertion of duplicated HPRT DNA into HPRT intron 1. These results suggest that duplication substrates of different lengths can be generated from the human HPRT exon 2-3 region and can undergo either homologous or nonhomologous recombination with the HPRT locus to form gene duplications.     

Genome-wide signatures of 'rearrangement hotspots' within segmental duplications in humans

The primary objective of this study was to create a genome-wide high resolution map (i.e., >100 bp) of 'rearrangement hotspots' which can facilitate the identification of regions capable of mediating de novo deletions or duplications in humans. A hierarchical method was employed to fragment segmental duplications (SDs) into multiple smaller SD units. Combining an end space free pairwise alignment algorithm with a 'seed and extend' approach, we have exhaustively searched 409 million alignments to detect complex structural rearrangements within the reference-guided assembly of the NA18507 human genome (18× coverage), including the previously identified novel 4.8 Mb sequence from de novo assembly within this genome. We have identified 1,963 rearrangement hotspots within SDs which encompass 166 genes and display an enrichment of duplicated gene nucleotide variants (DNVs). These regions are correlated with increased non-allelic homologous recombination (NAHR) event frequency which presumably represents the origin of copy number variations (CNVs) and pathogenic duplications/deletions. Analysis revealed that 20% of the detected hotspots are clustered within the proximal and distal SD breakpoints flanked by the pathogenic deletions/duplications that have been mapped for 24 NAHR-mediated genomic disorders. FISH Validation of selected complex regions revealed 94% concordance with in silico localization of the highly homologous derivatives. Other results from this study indicate that intra-chromosomal recombination is enhanced in genic compared with agenic duplicated regions, and that gene desert regions comprising SDs may represent reservoirs for creation of novel genes. The generation of genome-wide signatures of 'rearrangement hotspots', which likely serve as templates for NAHR, may provide a powerful approach towards understanding the underlying mutational mechanism(s) for development of constitutional and acquired diseases.     

Patterns of segmental duplication in the human genome

We analyzed the completed human genome for recent segmental duplications (size > or = 1 kb and sequence similarity > or = 90%). We found that approximately 4% of the genome is covered by duplications and that the extent of segmental duplication varies from 1% to 14% among the 24 chromosomes. Intrachromosomal duplication is more frequent than interchromosomal duplication in 15 chromosomes. The duplication frequencies in pericentromeric and subtelomeric regions are greater than the genome average by approximately threefold and fourfold. We examined factors that may affect the frequency of duplication in a region. Within individual chromosomes, the duplication frequency shows little correlation with local gene density, repeat density, recombination rate, and GC content, except chromosomes 7 and Y. For the entire genome, the duplication frequency is correlated with each of the above factors. Based on known genes and Ensembl genes, the proportion of duplications containing complete genes is 3.4% and 10.7%, respectively. The proportion of duplications containing genes is higher in intrachromosomal than in interchromosomal duplications, and duplications containing genes have a higher sequence similarity and tend to be longer than duplications containing no genes. Our simulation suggests that many duplications containing genes have been selectively maintained in the genome.     

Tandem and Interspersed Repeats Contribute to the Mosaic Structure of Segmental Duplications in the Human Genome

Intrachromosomal and interchromosomal segmental duplications account for more than 5% of the human genome. To analyze the processes resulting in the complex mosaic structure of duplicons, a draft human genome sequence was searched for duplicated segments of a genomic fragment of the pericentric region of the chromosome 21 short arm. The duplicons found consist of modules having paralogs in various genome regions. Module ends are flanked with various tandem or interspersed repeats, which are more unstable as compared with unique sequences. In most cases, the boundaries of duplicated segments exactly coincide with or are in close proximity to hot spots of various rearrangements within repeats or boundaries between repeats and unique sequences or between two different repeats. Homologous recombination between repetitive elements was assumed to be the major mechanism contributing to the mosaic structure of duplicons.     

Tandem and interspersed repeats contribute to the mosaic structure of segmented duplications in the human genome

Intrachromosomal and interchromosomal segmental duplications account for more than 5% of the human genome. To analyze the processes resulting in the complex mosaic structure of duplicons, a draft human genome sequence was searched for duplicated segments of a genomic fragment of the pericentric region of the chromosome 21 short arm. The duplicons found consist of modules having paralogs in various genome regions. Module ends are flanked with various tandem or interspersed repeats, which are more unstable as compared with unique sequences. In most cases, the boundaries of duplicated segments exactly coincide with or are in close proximity to hot spots of various rearrangements within repeats or boundaries between repeats and unique sequences or between two different repeats. Homologous recombination between repetitive elements was assumed to be the major mechanism contributing to the mosaic structure of duplicons.     

Unequal homologous recombination of human DNA on a yeast artificial chromosome.

We examined unequal homologous DNA recombination between human repetitive DNA elements located on a yeast artificial chromosome (YAC) and transforming plasmid molecules. A plasmid vector containing an Alu element, as well as a sequence identical to a unique site on a YAC, was introduced into yeast and double recombinant clones analyzed. Recombination occurs between vector and YAC Alu elements sharing as little as 74% identity. The physical proximity of an Alu element to the unique DNA segment appears to play a significant role in determining the frequency with which that element serves as a recombination substrate. In addition, cross-over points of the recombination reaction are largely confined to the ends of the repetitive element. Since a similar distribution of crossover sites occurs during unequal homologous recombination in human germ and somatic tissue, we propose that similar enzymatic processes may be responsible for the events observed in our system and in human cells. This suggests that further examination of the enzymology of unequal homologous recombination of human DNA within yeast may yield a greater understanding of the molecular events which control this process in higher eukaryotes.     

The Chinese hamster Alu-equivalent sequence: a conserved highly repetitious, interspersed deoxyribonucleic acid sequence in mammals has a structure suggestive of a transposable element.

A consensus sequence has been determined for a major interspersed deoxyribonucleic acid repeat in the genome of Chinese hamster ovary cells (CHO cells). This sequence is extensively homologous to (i) the human Alu sequence (P. L. Deininger et al., J. Mol. Biol., in press), (ii) the mouse B1 interspersed repetitious sequence (Krayev et al., Nucleic Acids Res. 8:1201-1215, 1980) (iii) an interspersed repetitious sequence from African green monkey deoxyribonucleic acid (Dhruva et al., Proc. Natl. Acad. Sci. U.S.A. 77:4514-4518, 1980) and (iv) the CHO and mouse 4.5S ribonucleic acid (this report; F. Harada and N. Kato, Nucleic Acids Res. 8:1273-1285, 1980). Because the CHO consensus sequence shows significant homology to the human Alu sequence it is termed the CHO Alu-equivalent sequence. A conserved structure surrounding CHO Alu-equivalent family members can be recognized. It is similar to that surrounding the human Alu and the mouse B1 sequences, and is represented as follows: direct repeat-CHO-Alu-A-rich sequence-direct repeat. A composite interspersed repetitious sequence has been identified. Its structure is represented as follows: direct repeat-residue 47 to 107 of CHO-Alu-non-Alu repetitious sequence-A-rich sequence-direct repeat. Because the Alu flanking sequences resemble those that flank known transposable elements, we think it likely that the Alu sequence dispersed throughout the mammalian genome by transposition.     

Human endogenous retroviral elements as indicators of ectopic recombination events in the primate genome

HERV elements make up a significant fraction of the human genome and, as interspersed repetitive elements, have the capacity to provide substrates for ectopic recombination and gene conversion events. To understand the extent to which these events occur and gain further insight into the complex evolutionary history of these elements in our genome, we undertook a phylogenetic study of the long terminal repeat sequences of 15 HERV-K(HML-2) elements in various primate species. This family of human endogenous retroviruses first entered the primate genome between 35 and 45 million years ago. Throughout primate evolution, these elements have undergone bursts of amplification. From this analysis, which is the largest-scale study of HERV sequence dynamics during primate evolution to date, we were able to detect intraelement gene conversion and recombination at five HERV-K loci. We also found evidence for replacement of an ancient element by another HERV-K provirus, apparently reflecting an occurrence of retroviral integration by homologous recombination. The high frequency of these events casts doubt on the accuracy of integration time estimates based only on divergence between retroelement LTRs.     

Analysis of the human Alu Ya-lineage

The Alu Ya-lineage is a group of related, short interspersed elements (SINEs) found in primates. This lineage includes subfamilies Ya1-Ya5, Ya5a2 and others. Some of these subfamilies are still actively mobilizing in the human genome. We have analyzed 2482 elements that reside in the human genome draft sequence and focused our analyses on the 2318 human autosomal Ya Alu elements. A total of 1470 autosomal loci were subjected to polymerase chain reaction (PCR)-based assays that allow analysis of individual Ya-lineage Alu elements. About 22% (313/1452) of the Ya-lineage Alu elements were polymorphic for the insertion presence on human autosomes. Less than 0.01% (5/1452) of the Ya-lineage loci analyzed displayed insertions in orthologous loci in non-human primate genomes. DNA sequence analysis of the orthologous inserts showed that the orthologous loci contained older pre-existing Y, Sc or Sq Alu subfamily elements that were the result of parallel forward insertions or involved in gene conversion events in the human lineage. This study is the largest analysis of a group of "young", evolutionarily related human subfamilies. The size, evolutionary age and variable allele insertion frequencies of several of these subfamilies makes members of the Ya-lineage useful tools for human population studies and primate phylogenetics.     

Small deletion variants have stable breakpoints commonly associated with alu elements

Copy number variants (CNVs) contribute significantly to human genomic variation, with over 5000 loci reported, covering more than 18% of the euchromatic human genome. Little is known, however, about the origin and stability of variants of different size and complexity. We investigated the breakpoints of 20 small, common deletions, representing a subset of those originally identified by array CGH, using Agilent microarrays, in 50 healthy French Caucasian subjects. By sequencing PCR products amplified using primers designed to span the deleted regions, we determined the exact size and genomic position of the deletions in all affected samples. For each deletion studied, all individuals carrying the deletion share identical upstream and downstream breakpoints at the sequence level, suggesting that the deletion event occurred just once and later became common in the population. This is supported by linkage disequilibrium (LD) analysis, which has revealed that most of the deletions studied are in moderate to strong LD with surrounding SNPs, and have conserved long-range haplotypes. Analysis of the sequences flanking the deletion breakpoints revealed an enrichment of microhomology at the breakpoint junctions. More significantly, we found an enrichment of Alu repeat elements, the overwhelming majority of which intersected deletion breakpoints at their poly-A tails. We found no enrichment of LINE elements or segmental duplications, in contrast to other reports. Sequence analysis revealed enrichment of a conserved motif in the sequences surrounding the deletion breakpoints, although whether this motif has any mechanistic role in the formation of some deletions has yet to be determined. Considered together with existing information on more complex inherited variant regions, and reports of de novo variants associated with autism, these data support the presence of different subgroups of CNV in the genome which may have originated through different mechanisms.