Tandem-repetitive noncoding DNA: forms and forces

A model of sequence-dependent, unequal crossing-over and gene amplification (slippage replication) has been stimulated in order to account for various structural features of tandemly repeated DNA sequences. It is shown that DNA whose sequence is not maintained by natural selection will exhibit repetitive patterns over a wide range of recombination rates as a result of the interaction of unequal crossing-over and slippage replication, processes that depend on sequence similarity. At high crossing-over frequencies, the nucleotide patterns generated in the simulations are simple and highly regular, with short, nearly identical sequences repeated in tandem. Decreasing recombination rates increase the tendency to longer and more-complex repeat units. Periodicities have been observed down to very low recombination rates (one or more orders of magnitude lower than mutation rate). At such low rates, most of the sequences contain repeats which have an extensive substructure and a high degree of heterogeneity among each other; often higher-order structures are superimposed on a tandem array. These results are compared with various structural properties of tandemly repeated DNAs known from eukaryotes, the spectrum ranging from simple-sequence DNAs, particularly the hypervariable mini-satellites, to the classical satellite DNAs, located in chromosomal regions of low recombination, e.g., heterochromatin.     

On the mechanism of amplification of satellite II DNA sequences of the domestic goat Capra hircus

A restriction enzyme analysis of the satellite II DNAs of the domestic goat Capra hircus, sheep Ovis aries and ox Bos taurus (p = 1.720, 1.723 and 1.722 g/cm3, respectively) has been carried out and shows that, although all three are composed of repeat units of 700 base-pairs, goat satellite II is present in the genome primarily in the form of 2100 base-pair trimers. Unequal crossing-over between repeat units has produced an oligomer series, whose oligomers gradually decrease in copy number the further they are in molecular weight from the trimer. The trimer population is much more uniform than the monomer population, as most trimers have similar restriction patterns, whereas their component monomers differ considerably in their restriction properties. This heterogeneity was confirmed by cross-hybridization of the different monomers of a cloned trimer. Here, heterologous hybrids were much less stable than the homologous hybrids. Attempts were made to simulate such an oligomer series by computer, using a longitudinal (saltatory), and two horizontal (unequal crossover and drive) models. Simulations of both the saltatory and drive mechanisms could produce the oligomer series in approximately the observed ratios, but only the former could simultaneously produce other restriction properties of this sequence family. This was because the other restriction sites were in a different (monomer) register, and it is difficult for a drive model promoting traits in only one register to fix properties in different registers. The unequal crossover model proposed by Smith (1973, 1976) generally produced homogeneous arrays from heterogeneous arrays, but higher-order repeat structures could be produced when the efficiency of crossing-over between different monomer types was reduced. In most of these arrays, the dimer was the predominant oligomer, but in approximately 10% the trimer was predominant. Since the unequal crossover model produces dimeric arrays with high frequency given appropriate conditions, it is an attractive model for explaining the production of satellite DNAs whose structure has evolved through a series of doublings, such as mouse major satellite DNA and the "alphoid" satellite sequences of primates. Other factors necessary for this model to work are generally considered to be natural components of the speciation process. It is therefore suggested that, although the saltatory model conforms most closely to the observed structure of goat satellite II, this particular satellite DNA may represent one of the few cases when the unequal crossover mechanism does not give rise to a dimeric structure.     

Persistence of Tandem Arrays: Implications for Satellite and Simple-Sequence Dnas

Recombination processes acting on tandem arrays are suggested here to have probable intrinsic biases, producing an expected net decrease in array size following each event, in contrast to previous models which assume no net change in array size. We examine the implications of this by modeling copy number dynamics in a tandem array under the joint interactions of sister-strand unequal crossing over (rate gamma per generation per copy) and intrastrand recombination resulting in deletion (rate epsilon per generation per copy). Assuming no gene amplification or selection, the expected mean persistence time of an array starting with z excess copies (i.e., array size z + 1) is z(1 + gamma/epsilon) recombinational events. Nontrivial equilibrium distributions of array sizes exist when gene amplification or certain forms of selection are considered. We characterize the equilibrium distribution for both a simple model of gene amplification and under the assumption that selection imposes a minimal array size, n. For the latter case, n + 1/alpha is an upper bound for mean array size under fairly general conditions, where alpha(= 2 epsilon/gamma) is the scaled deletion rate. Further, the distribution of excess copies over n is bounded above by a geometric distribution with parameter alpha/(1 + alpha). Tandem arrays are unlikely to be greatly expanded by unequal crossing over unless alpha much less than 1, implying that other mechanisms, such as gene amplification, are likely important in the evolution of large arrays. Thus unequal crossing over, by itself, is likely insufficient to account for satellite DNA.     

Mechanisms of human minisatellite mutation in yeast

Minisatellites are tandem repeat loci, with repeat units ranging in size from 5 bp to 100 bp. The total lengths of repeat arrays vary from about 0.5 kb to 30 kb, and excessive variability in allele length at human minisatellite loci is the result of germline-specific complex recombination events generating new length alleles. Minisatellite alleles also mutate to new lengths in somatic cells, but this occurs at a much lower rate than in the germline. Since recombination is involved in minisatellite mutation, the yeast Saccharomyces cerevisiae is a suitable model organism that has been employed to further dissect the molecular basis of mutation events at human minisatellites. These studies have shown that the mutational behaviour of a minisatellite in meiosis is not determined by the intrinsic properties of the repeat array, but are highly dependent on the position of the minisatellite in the genome. The processes for minisatellite mutation in yeast and humans are identical in the sense that mutation is indeed driven by meiotic recombination, but differ with regard to the types of structural changes that are generated by the recombination events. Tetrad analyses showed that inter-allelic transfers of repeats occur by conversion and not crossing over, and that several chromatids can be involved in successive recombination events in one meiosis, resulting in mutant alleles in several spores. It has been demonstrated that the genes SPO11 and RAD50, involved in the initiation of recombination events, are required for human minisatellite mutation in yeast meiosis. Intrinsic properties of the repeat array appear to determine the stability of human minisatellites in yeast mitosis, since mitotic mutation rates in yeast are highly variable between minisatellites. The repair genes RAD27 and DNA2 stabilise human minisatellites in yeast mitosis, while RAD5 has no effect on mitotic stability. MSH2 depresses human minisatellite frequency in meiotic cells of yeast.     

Evolution of satellite DNAs from the genus Palorus--experimental evidence for the "library" hypothesis

Satellite DNA profiles have been characterized in the congeneric species Palorus ratzeburgii, Palorus subdepressus, Palorus genalis, and Palorus ficicola (Coleoptera, Insecta), each of which contains a single, A + T-rich satellite DNA comprising a considerable portion of the genome (20%-40%). These satellites exhibit insignificant mutual sequence similarity. Using PCR assay, it has been shown that all four sequences are present in each of the tested Palorus species: one of them is amplified into a high copy number or a major satellite, while the three others are in the form of low-copy-number repeats estimated to make up approximately 0.05% of the genome. Each of the four satellites is interspecifically high conserved concerning the sequence, monomer length, and tandem repeat organization. Major, as well as low- copy-number, satellites are colocalized in the regions of pericentromeric heterochromatin on all chromosomes of the complement. The low-copy-number satellites are dispersed between the large arrays of the major satellite over the whole heterochromatic block. Our results explain satellite DNA evolution, confirming the hypothesis that related species share a "library" of conserved satellite sequences, some of which could be amplified into a major satellite. Due to the evolutionary dynamics of satellite DNAs, the content of the "library" is variable; the elimination of some sequences parallels the creation of the new ones. Quantitative changes in satellite DNAs, induced by occasional amplification of satellite repeat from the "library", could possibly occur in the course of the speciation process, thus forming a species-specific profile of satellite DNAs.      

On the mode of evolution of alpha satellite DNA in human populations

Summary The hypothesis that highly reiterated satellite DNAs in present-day populations evolve by molecular mechanisms that create, by saltatory amplification steps, new long arrays of satellite DNA, and that such long arrays are used for homogenization purposes, has been tested both in mouse and in humans. In mouse, the data obtained are consistent with this hypothesis. This was tested in more detail on chromosomes 13 and 21 of the human genome. A Centre d'Etudes du Polymorphisme Humain family, which in some individuals exhibits strong supplementary DNA bands following TaqI restriction endonuclease digestion and conventional gel electrophoresis, was analyzed by pulse field gel electrophoresis following restriction by BamHI. The supplementary bands on chromosome 13 (18 times the basic alpha satellite DNA repeat) and on chromosome 21 (a 9.5-mer) segregated with centromeric alpha satellite DNA blocks of 5 and 5.3 megabases, respectively. These are by far the largest alpha satellite block lengths seen in all chromosome 13 and chromosome 21 centrometric sequences so far analyzed in this manner. The possibility that these supplementary alpha satellite sequences were created in single individuals by saltatory amplification steps is discussed in light of our own data and that published by others. It is proposed that deletion events and unequal cross-overs, which both occur in large satellite DNA arrays, contribute to the homogenization of size and sequence of the alpha satellite DNA on most chromosomes of humans.     

Subunit structure of chromatin and the organization of eukaryotic highly repetitive DNA: recurrent periodicities and models for the evolutionary origins of repetitive DNA.

A restriction enzyme analysis of the repeat structure of mouse satellite, sheep satellite II, human highly repetitive fractions, calf satellite I, and a repetitive fraction of the rat indicates that those DNAs share repeat periodicites in common with one another and with the highly repetitive component α DNA of the African green monkey. The basic repeat periodicity of component α is 176 ± 4 nucleotide base-pairs: the repeat periodicities of the various highly repetitive fractions described here also seem based on this fundamental unit, but it is disguised by a superimposed, higher order repeat organization in each case. The higher orders of organization are based on integral multiples of the basic unit which may reflect the nucleosome spacing of constitutive heterochromatin. With the exception of component α DNA, which shows a repeat structure based on a monomer of 176 ± 4 nucleotide base-pairs, all of the highly repetitive DNAs examined showed a preference for even-numbered or geometric multiples of the basic unit in their higher order sequence organization. It is suggested that such organization is a relatively recent development in the hierarchical evolution of the sequences.Several models are discussed which may account for the higher order organization and expansion of these highly repetitive DNAs. Either a modified unequal crossover model (Smith, 1973) or a modified replicative loop model (Keyl, 1965a) seems consistent with many of the properties of highly repetitive DNAs. The models may have implications for the number, distribution and intranuclear rearrangements of transcribed sequences associated with such DNAs.     

Ribosomal DNA Evolution at the Population Level: Nucleotide Variation in Intergenic Spacer Arrays of Daphnia Pulex

Nucleotide variation was surveyed in 21 subrepeat arrays from the ribosomal DNA intergenic spacer of three Daphnia pulex populations. Eighteen of these arrays contained four subrepeats. Contrary to expectations, each of the four positions within the array had a different consensus sequence. However, gene conversion, involving sequences less than the length of a subrepeat, had occurred between subrepeats in different positions. Three arrays had more than four subrepeats and were undoubtedly generated by unequal crossing over between standard-length arrays. The data strongly suggested that most unequal exchanges between arrays are intrachromosomal and that they occur much less frequently than unequal exchanges at the level of the entire rDNA repeat. Strong associations among variants at different positions allowed the recognition of five groups of arrays, two of which were found in more than one population. Five of the seven individuals surveyed had arrays from more than one group. Analysis of the distribution of nucleotide variation suggested that the populations were quite divergent, a result that is concordant with previous surveys of allozyme and mitochondrial DNA variation. It was suggested that some of the subrepeat array types are quite old, at least predating the recolonization of pond habitats in the midwestern United States after the last glaciation.     

High-resolution mapping of repetitive DNA by in situ hybridization: molecular and chromosomal features of prominent dispersed and discretely localized DNA families from the wild beet species Beta procumbens

Members of three prominent DNA families of Beta procumbens have been isolated as Sau3A repeats. Two families consisting of repeats of about 158 bp and 312 bp are organized as satellite DNAs (Sau3A satellites I and II), whereas the third family with a repeat length of 202 bp is interspersed throughout the genome. Multi-colour fluorescence in situ hybridization was used for physical mapping of the DNA families, and has shown that these tandemly organized families occur in large heterochromatic and DAPI positive blocks. The Sau3A satellite I hybridized exclusively around or near the centromeres of 10, 11 or 12 chromosomes. The Sau3A satellite family I showed high intraspecific variability and high-resolution physical mapping was performed on pachytene chromosomes using differentially labelled repeats. The physical order of satellite subfamily arrays along a chromosome was visualized and provided evidence that large arrays of plant satellite repeats are not contiguous and consist of distinct subfamily domains. Re-hybridization of a heterologous rRNA probe to mitotic metaphase chromosomes revealed that the 18S-5.8S-25S rRNA genes are located at subterminal position on one chromosome pair missing repeat clusters of the Sau3A satellite family I. It is known that arrays of Sau3A satellite I repeats are tightly linked to a nematode (Heterodera schachtii) resistance gene and our results show that the gene might be located close to the centromere. Large arrays of the Sau3A satellite II were found in centromeric regions of 16 chromosomes and, in addition, a considerable interspersion of repeats over all chromosomes was observed. The family of interspersed 202 bp repeats is uniformly distributed over all chromosomes and largely excluded from the rRNA gene cluster but shows local amplification in some regions. Southern hybridization has shown that all three families are specific for genomes of the section Procumbentes of the genus Beta.     

Linkage disequilibria and the site frequency spectra in the su(s) and su(w(a)) regions of the Drosophila melanogaster X chromosome

Over the last decade, surveys of DNA sequence variation in natural populations of several Drosophila species and other taxa have established that polymorphism is reduced in genomic regions characterized by low rates of crossing over per physical length. Parallel studies have also established that divergence between species is not reduced in these same genomic regions, thus eliminating explanations that rely on a correlation between the rates of mutation and crossing over. Several theoretical models (directional hitchhiking, background selection, and random environment) have been proposed as population genetic explanations. In this study samples from an African population (n = 50) and a European population (n = 51) were surveyed at the su(s) (1955 bp) and su(w(a)) (3213 bp) loci for DNA sequence polymorphism, utilizing a stratified SSCP/DNA sequencing protocol. These loci are located near the telomere of the X chromosome, in a region of reduced crossing over per physical length, and exhibit a significant reduction in DNA sequence polymorphism. Unlike most previously surveyed, these loci reveal substantial skews toward rare site frequencies, consistent with the predictions of directional hitchhiking and random environment models and inconsistent with the general predictions of the background selection model (or neutral theory). No evidence for excess geographic differentiation at these loci is observed. Although linkage disequilibrium is observed between closely linked sites within these loci, many recombination events in the genealogy of the sampled alleles can be inferred and the genomic scale of linkage disequilibrium, measured in base pairs between sites, is the same as that observed for loci in regions of normal crossing over. We conclude that gene conversion must be high in these regions of low crossing over.     

Molecular analysis of a polymorphic domain of alpha satellite from the human X chromosome.

Alpha satellite DNA, a diverse family of tandemly repeated DNA sequences located at the centromeric region of each human chromosome, is organized in a highly chromosome-specific manner and is characterized by a high frequency of restriction-fragment-length polymorphism. To examine events underlying the formation and spread of these polymorphisms within a tandem array, we have cloned and sequenced a representative copy of a polymorphic array from the X chromosome and compared this polymorphic copy with the predominant higher-order repeat form of X-linked alpha satellite. Sequence data indicate that the polymorphism arose by a single base mutation that created a new restriction site (for HindIII) in the sequence of the predominant repeat unit. This variant repeat unit, marked by the new HindIII site, was subsequently amplified in copy number to create a polymorphic domain consisting of approximately 500 copies of the variant repeat unit within the X-linked array of alpha satellite. We propose that a series of intrachromosomal recombination events between misaligned tandem arrays, involving multiple rounds of either unequal crossing-over or sequence conversion, facilitated the spread and fixation of this variant HindIII repeat unit.     

Characterization of repeat arrays in ultra‐long nanopore reads reveals frequent origin of satellite DNA from retrotransposon‐derived tandem repeats

Amplification of monomer sequences into long contiguous arrays is the main feature distinguishing satellite DNA from other tandem repeats, yet it is also the main obstacle in its investigation because these arrays are in principle difficult to assemble. Here we explore an alternative, assembly‐free approach that utilizes ultra‐long Oxford Nanopore reads to infer the length distribution of satellite repeat arrays, their association with other repeats and the prevailing sequence periodicities. Using the satellite DNA‐rich legume plant Lathyrus sativus as a model, we demonstrated this approach by analyzing 11 major satellite repeats using a set of nanopore reads ranging from 30 to over 200 kb in length and representing 0.73× genome coverage. We found surprising differences between the analyzed repeats because only two of them were predominantly organized in long arrays typical for satellite DNA. The remaining nine satellites were found to be derived from short tandem arrays located within LTR‐retrotransposons that occasionally expanded in length. While the corresponding LTR‐retrotransposons were dispersed across the genome, this array expansion occurred mainly in the primary constrictions of the L. sativus chromosomes, which suggests that these genome regions are favourable for satellite DNA accumulation.     

The Evolution of Restricted Recombination and the Accumulation of Repeated DNA Sequences

We suggest hypotheses to account for two major features of chromosomal organization in higher eukaryotes. The first of these is the general restriction of crossing over in the neighborhood of centromeres and telomeres. We propose that this is a consequence of selection for reduced rates of unequal exchange between repeated DNA sequences for which the copy number is subject to stabilizing selection: microtubule binding sites, in the case of centromeres, and the short repeated sequences needed for terminal replication of a linear DNA molecule, in the case of telomeres. An association between proximal crossing over and nondisjunction would also favor the restriction of crossing over near the centromere. The second feature is the association between highly repeated DNA sequences of no obvious functional significance and regions of restricted crossing over. We show that highly repeated sequences are likely to persist longest (over evolutionary time) when crossing over is infrequent. This is because unequal exchange among repeated sequences generates single copy sequences, and a population that becomes fixed for a single copy sequence by drift remains in this state indefinitely (in the absence of gene amplification processes). Increased rates of exchange thus speed up the process of stochastic loss of repeated sequences.     

Quantitative genetic variation and developmental clocks

It is well-known that most genetic variation affects quantitative traits, and natural or artificial selection can act to change quantitative features of organisms more rapidly than qualitative ones. Surprisingly, variability is not confined to outbred species, but also occurs in inbred mice at a much higher rate than expected from known mutation rates. The size and shape of organisms and their constituent parts are, at least in part, controlled by the number of cell divisions, and there is published evidence for the existence of developmental clocks, which may count cell divisions. A molecular model for a developmental clock was previously proposed. It depends on the DNA methylation of repeated sequences of DNA, where the methylation of each additional sequence is tied to DNA synthesis and therefore cell division. The number of repeats specifies the number of divisions which will occur before a signal is produced which can activate or inactivate one or more genes. It is known that crossing over occurs between sister chromatids, and where tandemly repeated sequences occur unequal exchange can generate a larger or smaller number of repeats. An example of this is seen in the well-known variability of "minisatellite" sequences in human DNA. Unequal sister chromatid exchange can occur in mitotic and meiotic cells in the germ line, and in the case of developmental clock sequences could generate variation in clock length which in turn would directly affect quantitative traits. These events can be regarded as a special case of molecular drive during evolution.     

The role of background selection in shaping patterns of molecular evolution and variation: evidence from variability on the Drosophila X chromosome

In the putatively ancestral population of Drosophila melanogaster, the ratio of silent DNA sequence diversity for X-linked loci to that for autosomal loci is approximately one, instead of the expected "null" value of 3/4. One possible explanation is that background selection (the hitchhiking effect of deleterious mutations) is more effective on the autosomes than on the X chromosome, because of the lack of crossing over in male Drosophila. The expected effects of background selection on neutral variability at sites in the middle of an X chromosome or an autosomal arm were calculated for different models of chromosome organization and methods of approximation, using current estimates of the deleterious mutation rate and distributions of the fitness effects of deleterious mutations. The robustness of the results to different distributions of fitness effects, dominance coefficients, mutation rates, mapping functions, and chromosome size was investigated. The predicted ratio of X-linked to autosomal variability is relatively insensitive to these variables, except for the mutation rate and map length. Provided that the deleterious mutation rate per genome is sufficiently large, it seems likely that background selection can account for the observed X to autosome ratio of variability in the ancestral population of D. melanogaster. The fact that this ratio is much less than one in D. pseudoobscura is also consistent with the model's predictions, since this species has a high rate of crossing over. The results suggest that background selection may play a major role in shaping patterns of molecular evolution and variation.     

Homogenization of Tandemly Repeated Nucleotide Sequences by Distance-Dependent Nucleotide Sequence Conversion

Previous work revealed that recurrent mutations (= mutation occurring more than once) in the tandemly repeated arrays present in nontranscribed spacers (NTS) of ribosomal RNA genes (rDNA) are clustered, i.e., they most frequently occur in repeats with adjacent or alternate distribution. A possible explanation is that the likelihood of heteroduplex formation, a prerequisite of gene conversion, decreases with the distance between repeats. To test this possibility, evolution of an array of 11 initially homogeneous repeats was computer simulated using three models, two assuming that the likelihood of heteroduplex formation decreases with increasing distance between the repeats and one assuming that it is constant. Patterns of mutation distribution obtained in computer simulations were compared with the distribution of mutations found in the repeated arrays in the NTS of seven rDNA clones. The patterns of mutations generated by the models assuming that the likelihood of heteroduplex formation decreases as distance between the repeats increases agreed with the patterns observed in rDNA; the patterns generated by the model assuming that the likelihood is independent of distance between repeats disagreed with the patterns observed in the rDNA clones. The topology of the heteroduplex formed between DNA in adjacent repeats predicts that the most frequently occurring conversions in the NTS repeated arrays will be shorter than the length of the repeat. The topology of the heteroduplex also predicts that if the heteroduplex leads to crossing over a circular repeat is excised. It is speculated that the circle can transpose or can be amplified via rolling circle replication and subsequently transpose. It is also shown that homogenization of the NTS repeated arrays proceeds at different rates in different species.     

Repeat instability at human minisatellites arising from meiotic recombination.

Little is known about the role of meiotic recombination processes such as unequal crossover in driving instability at tandem repeat DNA. Methods have therefore been developed to detect meiotic crossovers within two different GC-rich minisatellite repeat arrays in humans, both in families and in sperm DNA. Both loci normally mutate in the germline by complex conversion-like transfer of repeats between alleles. Analysis shows that inter-allelic unequal crossovers also occur at both loci, although at low frequency, to yield simple recombinant repeat arrays with exchange of flanking markers. Equal crossovers between aligned alleles, resulting in recombinant alleles but without change in repeat copy number, also occur in sperm at a similar frequency to unequal crossovers. Both crossover and conversion show polarity in the repeat array and are co-suppressed in an allele showing unusual germline stability. This provides evidence that minisatellite conversion and crossover arise by a common mechanism, perhaps by alternative processing of a meiotic recombination initiation complex, and implies that minisatellite instability is a by-product of meiotic recombination in repeat DNA. While minisatellite recombination is infrequent, crossover rates indicate that the unstable end of a human minisatellite can act as a recombination warm-spot, even between sequence-heterologous alleles.