Abstract: Many genes are members of large families that have arisen during evolution through gene duplication events. Our increasing understanding of gene organization at the scale of whole genomes is revealing further evidence for the extensive retention of genes that arise during duplication events of various types. Duplication is thought to be an important means of providing a substrate on which evolution can work. An understanding of gene duplication and its resolution is crucial for revealing mechanisms of genetic redundancy. Here, we consider both the theoretical framework and the experimental evidence to explain the preservation of duplicated genes.

Abstract: Genetic changes underlie tumor progression and may lead to cancer-specific expression of critical genes. Over 1100 publications have described the use of comparative genomic hybridization (CGH) to analyze the pattern of copy number alterations in cancer, but very few of the genes affected are known. Here, we performed high-resolution CGH analysis on cDNA microarrays in breast cancer and directly compared copy number and mRNA expression levels of 13,824 genes to quantitate the impact of genomic changes on gene expression. We identified and mapped the boundaries of 24 independent amplicons, ranging in size from 0.2 to 12 Mb. Throughout the genome, both high- and low-level copy number changes had a substantial impact on gene expression, with 44% of the highly amplified genes showing overexpression and 10.5% of the highly overexpressed genes being amplified. Statistical analysis with random permutation tests identified 270 genes whose expression levels across 14 samples were systematically attributable to gene amplification. These included most previously described amplified genes in breast cancer and many novel targets for genomic alterations, including the HOXB7 gene, the presence of which in a novel amplicon at 17q21.3 was validated in 10.2% of primary breast cancers and associated with poor patient prognosis. In conclusion, CGH on cDNA microarrays revealed hundreds of novel genes whose overexpression is attributable to gene amplification. These genes may provide insights to the clonal evolution and progression of breast cancer and highlight promising therapeutic targets.

Abstract: Opinions on the hypothesis that ancient genome duplications contributed to the vertebrate genome range from strong skepticism to strong credence. Previous studies concentrated on small numbers of gene families or chromosomal regions that might not have been representative of the whole genome, or used subjective methods to identify paralogous genes and regions. Here we report a systematic and objective analysis of the draft human genome sequence to identify paralogous chromosomal regions (paralogons) formed during chordate evolution and to estimate the ages of duplicate genes. We found that the human genome contains many more paralogons than would be expected by chance. Molecular clock analysis of all protein families in humans that have orthologs in the fly and nematode indicated that a burst of gene duplication activity took place in the period 350 650 Myr ago and that many of the duplicate genes formed at this time are located within paralogons. Our results support the contention that many of the gene families in vertebrates were formed or expanded by large-scale DNA duplications in an early chordate. Considering the incompleteness of the sequence data and the antiquity of the event, the results are compatible with at least one round of polyploidy.

Abstract: Recent genetic studies have established that the killer cell immunoglobulin-like receptor (KIR) genomic region displays extensive diversity through variation in gene content and allelic polymorphism within individual KIR genes. It is demonstrated by family segregation analysis, genomic sequencing, and gene order determination that genomic diversity by gene content alone gives rise to more than 20 different KIR haplotypes and at least 40-50 KIR genotypes. In the most reductionist format, KIR haplotypes can be accommodated within one of 10 different prototypes, each with multiple permutations. Our haplotype model considers the KIR haplotype as two separate halves: the centromeric half bordered upstream by KIR3DL3 and downstream by 2DL4, and the telomeric half bordered upstream by 2DL4 and downstream by 3DL2. There are rare KIR haplotypes that do not fit into this model. Recombination, gene duplication, and inversion can however, readily explain these haplotypes. Additional allelic polymorphism imposes extensive individual variability. Accordingly, this segment of the human genome displays a level of diversity similar to the one observed for the human major histocompatibility complex. Recent application of immunogenetic analysis of KIR genes in patient populations implicates these genes as important genetic disease susceptibility factors.

Abstract: Co-option occurs when natural selection finds new uses for existing traits, including genes, organs, and other body structures. Genes can be co-opted to generate developmental and physiological novelties by changing their patterns of regulation, by changing the functions of the proteins they encode, or both. This often involves gene duplication followed by specialization of the resulting paralogous genes into particular functions. A major role for gene co-option in the evolution of development has long been assumed, and many recent comparative developmental and genomic studies have lent support to this idea. Although there is relatively less known about the molecular basis of co-option events involving developmental pathways, much can be drawn from well-studied examples of the co-option of structural proteins. Here, we summarize several case studies of both structural gene and developmental genetic circuit co-option and discuss how co-option may underlie major episodes of adaptive change in multicellular organisms. We also examine the phenomenon of intraspecific variability in gene expression patterns, which we propose to be one form of material for the co-option process. We integrate this information with recent models of gene family evolution to provide a framework for understanding the origin of co-optive evolution and the mechanisms by which natural selection promotes evolutionary novelty by inventing new uses for the genetic toolkit.

Abstract: Initial human genome sequence analysis has revealed large segments of nearly identical sequence in particular chromosomal regions. The recent origin of these segments and their abundance (approximately 5%) has challenged investigators to elucidate their underlying mechanism and role in primate genome evolution. Although the precise fraction is unknown, some of these duplicated segments have recently been shown to be associated with rapid gene innovation and chromosomal rearrangement in the genomes of man and the great apes.

Abstract: The development of hepatocellular carcinoma (HCC) is a multistep process associated with changes in host gene expression, some of which correlate with the appearance and progression of tumor. Preneoplastic changes in gene expression result from altered DNA methylation, the actions of hepatitis B and C viruses, and point mutations or loss of heterozygosity (LOH) in selected cellular genes. Tumor progression is characterized by LOH involving tumor suppressor genes on many chromosomes and by gene amplification of selected oncogenes. The changes observed in different HCC nodules are often distinct, suggesting heterogeneity on the molecular level. These observations suggest that there are multiple, perhaps redundant negative growth regulatory pathways that protect cells against transformation. An understanding of the molecular pathogenesis of HCC may provide new markers for tumor staging, for assessment of the relative risk of tumor formation, and open new opportunities for therapeutic intervention.

Abstract: Background: The mitotic kinase-encoding gene STK15/BTAK/ AuroraA is associated with aneuploidy and transformation when overexpressed in mammalian cells. STK15 overexpression activates an unknown oncogenic pathway that involves centrosome amplification and results in missegregation of chromosomes. Because clinical prognosis and tumor aneuploidy are tightly linked in human bladder cancer, we examined whether increased STK15 copy number and protein levels are linked to aneuploidy in bladder cancers. Methods: STK15 protein was visualized by immunohistochemistry in 205 formalin-fixed, paraffin-embedded human bladder tumors. STK15 gene copy number was evaluated in 61 tumors by Southern blot hybridization and in 21 of these 61 tumors by fluorescence in situ hybridization (FISH). Copy numbers of chromosomes 3, 17, 20, and 21 were evaluated by FISH with chromosome-specific probes. STK15 expression levels were related to histologic grade, stage, and DNA ploidy of the tumors and to the patients’ follow-up data. The chisquare test for association was used to analyze the relationship between STK15 expression and pathologic features. All statistical tests were two-sided. Results: Tumors with low levels of STK15 amplification (3–4 copies) showed minimal deviation in their chromosome copy number and diploid or near-diploid total nuclear DNA content. Tumors with higher levels of STK15 amplification (>4 copies) had a major increase of chromosome copy number and of their total nuclear DNA content, i.e., exhibited pronounced aneuploidy. Elevated expression of STK15 was strongly associated with parameters of clinical aggressiveness including high histologic grade (P<.001), invasion (P<.001), increased rate of metastasis (P<.001), and decreased metastasis-free (P<.001) and overall (P<.001) survival of patients with bladder cancer. Conclusion: STK15 gene amplification and associated increased expression of the mitotic kinase it encodes are associated with aneuploidy and aggressive clinical behavior in human bladder cancer. [J Natl Cancer Inst 2002;94: 1320–9]

Abstract: T cell activation and immune function are regulated by costimulatory molecules of the B7 superfamily. Human B7-H3 is a recent addition to this family and has been shown to mediate T cell proliferation and IFN-γ production. In this work we describe the identification of the mouse B7-H3 homolog, which is ubiquitously expressed in a variety of tissues. Activated CD4 and CD8 T cells express a putative receptor that can be recognized by soluble mouse B7-H3-Ig molecules. While the mouse B7-H3 gene was found to contain a single copy, we discovered a novel isoform of human B7-H3 (named as B7-H3b hereafter) with four Ig-like domains that results from gene duplication and differential splicing. B7-H3b is the major isoform expressed in several tissues. This structural information suggests a genetic variation of the B7-H3 gene in mammalian species.

Abstract: We determined the complete mtDNA sequences of the millipedes Narceus annularus and Thyropygus sp. (Arthropoda: Diplopoda) and identified in both genomes all 37 genes typical for metazoan mtDNA. The arrangement of these genes is identical in the two millipedes, but differs from that inferred to be ancestral for arthropods by the location of four genes/gene clusters. This novel gene arrangement is unusual for animal mtDNA, in that genes with opposite transcriptional polarities are clustered in the genome and the two clusters are separated by two non-coding regions. The only exception to this pattern is the gene for cysteine tRNA, which is located in the part of the genome that otherwise contains all genes with the opposite transcriptional polarity. We suggest that a mechanism involving complete mtDNA duplication followed by the loss of genes, predetermined by their transcriptional polarity and location in the genome, could generate this gene arrangement from the one ancestral for arthropods. The proposed mechanism has important implications for phylogenetic inferences that are drawn on the basis of gene arrangement comparisons.

Abstract: Physical mapping of the breakpoints of a pericentric inversion of the X chromosome (46,X,inv[X][p21q27]) in a female patient with mild mental retardation revealed localization of the Xp breakpoint in the IL1RAPL gene at Xp21.3 and the Xq breakpoint near the SOX3 gene (SRY [sex determining region Y]-box 3) (GenBank accession number NM_005634) at Xq26.3. Because carrier females with microdeletion in the IL1RAPL gene do not present any abnormal phenotype, we focused on the Xq breakpoint. However, we were unable to confirm the involvement of SOX3 in the mental retardation in this female patient. To validate SOX3 as an X-linked mental retardation (XLMR) gene, we performed mutation analyses in families with XLMR whose causative gene mapped to Xq26-q27. We show here that the SOX3 gene is involved in a large family in which affected individuals have mental retardation and growth hormone deficiency. The mutation results in an in-frame duplication of 33 bp encoding for 11 alanines in a polyalanine tract of the SOX3 gene. The expression pattern during neural and pituitary development suggests that dysfunction of the SOX3 protein caused by the polyalanine expansion might disturb transcription pathways and the regulation of genes involved in cellular processes and functions required for cognitive and pituitary development.

Abstract: There is increasing evidence for the involvement of repetitive DNA sequences as facilitators of some of the recurrent chromosomal rearrangements observed in human tumors. The high densities of repetitive DNA, such as Alu elements, at some chromosomal translocation breakpoint regions has led to the suggestion that these sequences could provide hot spots for homologous recombination, and could mediate the translocation process and elevate the likelihood of other types of chromosomal rearrangements taking place. The Alu core sequence itself has been suggested to promote DNA strand exchange and genomic rearrangement, and it has striking sequence similarity to chi (which has been shown to stimulate recBCD-mediated recombination in Escherichia coli). Alu repeats have been shown to be involved in the generation of many constitutional gene mutations in meiotic cells, attributed to unequal homologous recombination and consequent deletions and/or duplication events. It has recently been demonstrated that similar deletion events can take place in neoplasia because several types of leukemia-associated chromosomal rearrangements frequently have submicroscopic deletions immediately adjacent to the translocation breakpoint regions. Significantly, these types of deletions appear to be more likely to take place when the regions subject to rearrangement contain a high density of Alu repeats. With the completion of the Human Genome Project, it will soon be possible to create more comprehensive maps of the distribution and densities of repetitive sequences, such as Alu, throughout the genome. Such maps will offer unique insights into the relative distribution of cancer translocation breakpoints and the localization of clusters of repetitive DNA.

Abstract: The duplication of short chromosomal segments in the genome may contribute to the formation of new species. In his Perspective, Lynch explores the potential implications of new work ( Bailey et al .) that calculates the number of segmental duplications in the human genome, as well as other recent comparative data on fine-scale changes in chromosomal architecture.

Abstract: Amplification of chromosomal band 11q13 is a common event in human cancer. It has been reported in about 45% of head and neck carcinomas and in other cancers including esophageal, breast, liver, lung, and bladder cancer. To understand the mechanism of 11q13 amplification and to identify the potential oncogene(s) driving it, we have fine-mapped the structure of the amplicon in oral squamous cell carcinoma cell lines and localized the proximal and distal breakpoints. A 5-Mb physical map of the region has been prepared from which sequence is available. We quantified copy number of sequence-tagged site markers at 42–550 kb intervals along the length of the amplicon and defined the amplicon core and breakpoints by using TaqMan-based quantitative microsatellite analysis. The core of the amplicon maps to a 1.5-Mb region. The proximal breakpoint localizes to two intervals between sequence-tagged site markers, 550 kb and 160 kb in size, and the distal breakpoint maps to a 250 kb interval. The cyclin D1 gene maps to the amplicon core, as do two new expressed sequence tag clusters. We have analyzed one of these expressed sequence tag clusters and now report that it contains a previously uncharacterized gene, TAOS1 (tumor amplified and overexpressed sequence 1), which is both amplified and overexpressed in oral cancer cells. The data suggest that TAOS1 may be an amplification-dependent candidate oncogene with a role in the development and/or progression of human tumors, including oral squamous cell carcinomas. The approach described here should be useful for characterizing amplified genomic regions in a wide variety of tumors.

Abstract: Recent studies have revealed a significant proportion of BRCA1 exon deletions or duplications in breast-ovarian cancer families with high probability of BRCA1- or BRCA2-linked predisposition, in which mutations of these genes have not been found. The difficulty of detecting such heterozygous rearrangements has stimulated the development of several new screening methods. Quantitative fluorescent multiplex PCR is based on simultaneous amplification of multiple target sequences under conditions that allow rapid and reliable quantitative comparison of the fluorescence of each amplicon in test samples and in controls. The modified method described here, named quantitative multiplex PCR of short fluorescent fragments (QMPSF), is particularly well suited for large genes. All BRCA1 coding exons were analyzed using four multiplexes in 52 families without point mutations in the exons or splice-sites of BRCA1 and BRCA2, and selected because of high probability of a BRCA1- or BRCA2-linked genetic predisposition. Five distinct BRCA1 rearrangements were detected: a deletion of exons 8–13, a duplication of exons 3–8, a duplication of exons 18–20, a deletion of exons 15–16, and a deletion of exons 1–22—which is the largest deletion found so far within the BRCA1 gene. The method described here lends itself to rapid, sensitive, and cost-effective search of BRCA1 rearrangements and may be included into the routine molecular analysis of breast-ovarian cancer predispositions. Hum Mutat 20:218–226, 2002. © 2002 Wiley-Liss, Inc.

Abstract: DNA copy number gains and amplifications at 17q are frequent in gastric cancer, yet systematic analyses of the 17q amplicon have not been performed. In this study, we carried out a comprehensive analysis of copy number and expression levels of 636 chromosome 17-specific genes in gastric cancer by using a custom-made chromosome 17-specific cDNA microarray. Analysis of DNA copy number changes by comparative genomic hybridization on cDNA microarray revealed increased copy numbers of 11 known genes (ERBB2, TOP2A, GRB7, ACLY, PIP5K2B, MPRL45, MKP-L, LHX1, MLN51, MLN64, and RPL27) and seven expressed sequence tags (ESTs) that mapped to 17q12-q21 region. To investigate the genes transcribed at the 17q, we performed gene expression analyses on an identical cDNA microarray. Our expression analysis showed overexpression of 8 genes (ERBB2, TOP2A, GRB2, AOC3, AP2B1, KRT14, JUP, and ITGA3) and two ESTs. Of the commonly amplified transcripts, an uncharacterized EST AA552509 and the TOP2A gene were most frequently overexpressed in 82% of the samples. Additional studies will be initiated to understand the possible biological and clinical significance of these genes in gastric cancer development and progression.

Abstract: It has been proposed that two events of duplication of the entire genome occurred early in vertebrate history (2R hypothesis). Several phylogenetic studies with a few gene families (mostly Hox genes and proteins from the MHC) have tried to confirm these polyploidization events. However, data from a single locus cannot explain the evolutionary history of a complete genome. To study this 2R hypothesis, we have taken advantage of the phylogenetic position of the lamprey to study the history of gene duplications in vertebrates. We selected most gene families that contain several paralogous genes in vertebrates and for which lamprey genes and an out-group are known in databases. In addition, we isolated members of the nuclear receptor superfamily in lamprey. Hagfish genes were also analyzed and found to confirm the lamprey gene analysis. Consistent with the 2R hypothesis, the phylogenetic analysis of 33 selected gene families, dispersed through the whole genome, revealed that one period of gene duplication arose before the lamprey-gnathostome split and this was followed by a second period of gene duplication after the lamprey-gnathostome split. Nevertheless, our analysis suggests that numerous gene losses and other gene-genome duplications occurred during the evolution of the vertebrate genomes. Thus, the complexity of all the paralogy groups present in vertebrates should be explained by the contribution of genome duplications (2R hypothesis), extra gene duplications, and gene losses.

Abstract: Gene and genome duplications have given rise to enormous variability among species in the number of genes within their genomes. Gene copies have in turn played important roles in adaptation, having been implicated in the evolution of the immune response, insecticide resistance, efficient protein synthesis, and vertebrate body plans. In this chapter, we discuss the life history of gene duplications, from their first appearance within a population, through the period during which they rise in frequency or disappear, to their long-term fate. At each phase, we discuss the evolutionary processes that have influenced the dynamics of gene duplications and shaped their ultimate roles within a population. We argue that there is no evidence that organisms have evolved strategies to promote gene duplication in order to permit adaptive evolution. In contrast, many mechanisms exist to silence or eliminate duplicated genes, suggesting that selection has acted largely to reduce the rate of gene duplication. We also argue that natural selection has functioned as an effective sieve, increasing the representation of beneficial gene duplicates among those that establish within a population and that play a long-term role in evolution. To refine our understanding of how selection acts on new gene duplications, we provide a model incorporating a single-copy gene, its gene duplicate, and selection either favoring heterozygotes or eliminating deleterious mutations. Although both forms of selection can increase the initial rate of spread of a gene duplicate, the efficacy with which they do so differs dramatically. Heterozygote advantage always increases the rate of spread and can have a large impact. In contrast, masking deleterious mutations never has a large effect on the rate of spread of the duplicate, and this minor effect can be negative as well as positive. In both cases, the degree of linkage between the two gene copies affects the rate of spread of the duplication. Finally, we discuss evolutionary processes that occur over longer periods after a gene duplication has become established within a population. These long-term processes include maintenance, inactivation, and diversification in function. Consideration of each of the short-term and long-term processes affecting duplicated genes illustrates the subtle ways in which selection has acted to shape genomic structure.

Abstract: The killer cell immunoglobulin-like receptors (KIR) form a family of highly homologous immune receptors that regulate the response of natural killer (NK) cells and some T cells. The genetics of the human KIR family is reviewed in this article. In human populations, diversity in KIR genotype arises from variations in gene content and allelic polymorphism. Comparisons of 81 human KIR sequences reveal past events ofgene duplication and recombination, and indicate that individual KIR genes have diversified from the influence of natural selection. Comparison and compilation of population studies reveal extensive KIR genotype variability within human populations and among them. Genomic analysis shows the KIR genes to be close to each other and separated by homologous sequences that promote haplotype diversification through assymetric recombination. In contrast, homologous recombination appears favored at a unique sequence in the center of the KIR locus, and much haplotypic diversity can be explained by recombination between a limited number of gene-content motifs in the centromeric and telomeric halves of the locus. The importance of NK cells for early defenses against infection suggests that human KIR genotype diversity is the accumulated consequence of a history of numerous and successive selective episodes by different pathogens on human NK-cell responses.

Abstract: Oncogene activation by gene amplification is a major pathogenetic mechanism in human cancer. Using comparative genomic hybridization, we determined that metastatic human colon cancers commonly acquire numerous extra copies of chromosome arms 7p, 8q, 13q, and 20q. We then examined the consequence of these amplifications on gene expression using DNA microarrays. Of 55,000 transcripts profiled, 2,146 were determined to map to one of the four common colon cancer amplicons and to also be expressed in normal or malignant colon tissues. Of these, only 81 transcripts (3.8%) demonstrated a 2-fold increase over normal expression among cancers bearing the corresponding chromosomal amplification. Chromosomal amplifications are common in colon cancer metastasis, but increased expression of genes within these amplicons is rare.

Abstract: Mitochondrial DNA depletion syndrome is a clinically heterogeneous group of disorders characterized by a reduction in mitochondrial DNA copy number. The recent discovery of mutations in the deoxyguanosine kinase (dGK) gene in patients with the hepatocerebral form of mitochondrial DNA depletion syndrome prompted us to screen 21 patients to determine the frequency of dGK mutations, further characterize the clinical spectrum, and correlate genotypes with phenotypes. We detected mutations in three patients (14%). One patient had a homozygous GATT duplication (nucleotides 763-766), and another had a homozygous GT deletion (nucleotides 609-610); both mutations lead to truncated proteins. The third patient was a compound heterozygote for two missense mutations (R142K and E227K) that affect critical residues of the protein. These mutations were associated with variable phenotypes, and their low frequencies suggests that dGK is not the only gene responsible for mitochondrial DNA depletion in liver. The patient with the missense mutations had isolated liver failure and responded well to liver transplantation, which may be a therapeutic option in selected cases.

Abstract: The wealth of comparative genomics data from yeast species allows the molecular evolution of these eukaryotes to be studied in great detail. We used “proximity plots” to visually compare chromosomal gene order information from 14 hemiascomycetes, including the recent Genolevures survey, to Saccharomyces cerevisiae. Contrary to the original reports, we find that the Genolevures data strongly support the hypothesis that S. cerevisiae is a degenerate polyploid. Using gene order information alone, 70% of the S. cerevisiae genome can be mapped into “sister” regions that tile together with almost no overlap. This map confirms and extends the map of sister regions that we constructed previously by using duplicated genes, an independent source of information. Combining gene order and gene duplication data assigns essentially the whole genome into sister regions, the largest gap being only 36 genes long. The 16 centromere regions of S. cerevisiae form eight pairs, indicating that an ancestor with eight chromosomes underwent complete doubling; alternatives such as segmental duplications can be ruled out. Gene arrangements in Kluyveromyces lactis and four other species agree quantitatively with what would be expected if they diverged from S. cerevisiae before its polyploidization. In contrast, Saccharomyces exiguus, Saccharomyces servazzii, and Candida glabrata show higher levels of gene adjacency conservation, and more cases of imperfect conservation, suggesting that they split from the S. cerevisiae lineage after polyploidization. This finding is confirmed by sequences around the C. glabrata TRP1 and IPP1 loci, which show that it contains sister regions derived from the same duplication event as that of S. cerevisiae.

Abstract: Gene amplification plays a critical role in tumor progression. Hence, understanding the factors triggering this process in human cancers is an important concern. Unfortunately, the structures formed at early stages are usually unavailable for study, hampering the identification of the initiating events in tumors. Here, we show that the region containing the PIP gene, which is overexpressed in 80% of primary and metastatic breast cancers, is duplicated in the breast carcinoma cell line T47D. The two copies are organized as a large palindrome, lying 'in loco' on one chromosome 7. Such features constitute the landmark of the breakage-fusion-bridge (BFB) cycle mechanism. In hamster cells selected in vitro to resist cytotoxic drugs, common fragile site (CFS) activation has been shown to trigger this mechanism. Here, we characterize FRA7I at the molecular level and demonstrate that it lies 2 Mb telomeric to the PIP gene and sets the distal end of the repeated sequence. Moreover, our results suggest that the BFB process was frozen within the first cycle by healing of the broken chromosome. T47D cells thus offer a unique opportunity to observe the earliest products of the BFB cycle mechanism. Our findings constitute the first evidence that this amplification mechanism can be initiated in vivo by fragile site activation.

Abstract: Most duplicate genes are eliminated from a genome shortly after duplication, but those that remain are an important source of biochemical diversity. Here, I present evidence from genome-scale protein-protein interaction data, microarray expression data, and large-scale gene knockout data that this diversification is often asymmetrical: one duplicate usually shows significantly more molecular or genetic interactions than the other. I propose a model that can explain this divergence pattern if asymmetrically diverging duplicate gene pairs show increased robustness to deleterious mutations.

Abstract: Although comparative genetic mapping studies show extensive genome conservation among grasses, recent data provide many exceptions to gene collinearity at the DNA sequence level. Rice, sorghum, and maize are closely related grass species, once sharing a common ancestor. Because they diverged at different times during evolution, they provide an excellent model to investigate sequence divergence. We isolated, sequenced, and compared orthologous regions from two rice subspecies, sorghum, and maize to investigate the nature of their sequence differences. This study represents the most extensive sequence comparison among grasses, including the largest contiguous genomic sequences from sorghum (425 kb) and maize (435 kb) to date. Our results reveal a mosaic organization of the orthologous regions, with conserved sequences interspersed with nonconserved sequences. Gene amplification, gene movement, and retrotransposition account for the majority of the nonconserved sequences. Our analysis also shows that gene amplification is frequently linked with gene movement. Analyzing an additional 2.9 Mb of genomic sequence from rice not only corroborates our observations, but also suggests that a significant portion of grass genomes may consist of paralogous sequences derived from gene amplification. We propose that sequence divergence started from hotspots along chromosomes and expanded by accumulating small-scale genomic changes during evolution.