INTRODUCTION

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

The grass family Poaceae includes major crop plants such as wheat (Triticum aestivum), barley (Hordeum vulgare), oat (Avena sativa), rice (Oryza sativa), and maize (Zea mays). One of its tribes, Triticeae, contains more than 15 genera and 300 species including wheat and barley. Among the important crop plants of the family, rice has the smallest genome (415 Mb), which is only about three times larger than the model plant Arabidopsis (Bennett and Smith, 1976; Arumuganathan and Earle, 1991; Table I). The barley genome is about 12 times larger than that of rice, and maize is about six times. Wheat and barley genomes are of about the same size except that the bread wheat is hexaploid and thus is about three times the size of barley.

The gene size and number in most of the higher plants, especially within the grass family, are expected to be the same. The gene-containing fraction of the Arabidopsis genome is 0.85 (Barakat et al., 1998). Thus, the corresponding fractions for rice, maize, barley, and wheat are expected to be 0.28, 0.05, 0.025, and <0.01, respectively. Therefore, it is imperative to localize and mark the gene-containing regions to understand and efficiently manipulate the crop plant genomes. Second, as shown by the comparisons of the genetic linkage maps, the gene order and synteny are conserved among various Poaceae species (Hart, 1987; Ahn and Tanksley, 1993; Ahn et al., 1993; Devos et al., 1994; Van Deynze et al., 1995a), despite the fact that wheat, rice, and maize diverged more than 50 million years ago (Bennetzen and Freeling, 1993). Based on the genetic linkage map comparisons, the rice genome was divided into conserved blocks, which were proposed to have assembled in different combinations in various Poaceae chromosomes (Moore et al., 1995). The above observations and other phylogenetic and molecular studies strongly suggested that cereal genomes originated from a common ancestor (Clayton and Renvoize, 1986; Kellogg, 1998; Watson and Dallwitz, 1992). Monophyletic origin of the grasses, resulting in as much as 35-fold genome size difference without altering gene number, synteny, and colinearity, is another enigma. In this article, we will focus on the current understanding of the structure, distribution, and evolution of the gene-containing regions of the grass genomes.

	GENE DISTRIBUTION IN WHEAT

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

Studies on the genome organization of wheat suggested that more than 85% of the wheat genes are present in less than 10% of the chromosomal regions (Gill et al., 1996a, 1996b; Sandhu, 2000; Sandhu et al., 2001). A typical pattern of gene and recombination distribution on the wheat chromosomes is shown in Figure 1. The gene-containing regions of wheat were identified, marked, and bracketed by physical mapping of the genes and the markers on an array of single-break overlapping deletion lines (Endo and Gill, 1996; Gill et al., 1996a, 1996b; Faris et al., 2000; Sandhu et al., 2001). On each wheat chromosome, there are about six to eight gene-rich regions physically spanning about 10% of the region. The gene-rich regions are interspersed by gene-poor regions that are mainly comprised of retrotransposon-like repetitive sequences (SanMiguel et al., 1996; Barakat et al., 1997; Llaca and Messing, 1998; Feuillet and Keller, 1999). Physical location, structural organization, and gene densities of the gene-rich regions were similar among the three genomes of bread wheat (Gill et al., 1996a, 1996b; Sandhu, 2000; Sandhu et al., 2001). The gene-rich regions vary in the number of genes and the encompassed physical region. For example, most of the wheat homoeologous group 1 genes are present in eight distinct gene-rich regions. These gene-rich regions physically encompass 1% ("1S0.8 region") to 5% ("1L0.7 region") of the group 1 chromosomal region. Among the gene-rich regions, a minimum number of genes was present in the "1L1.0 region," which contained only 3% of the group 1 genes compared with maximum of 32% in the "1S0.8 region" (Fig. 1). In general, gene density in the distal gene-rich regions is higher as compared with the proximal.

View larger version (28K): [in this window] [in a new window]   Figure 1.   Distribution of genes and recombination on wheat homoeologous group 1 chromosomes. The consensus chromosome and the location and size of the gene-rich regions were drawn to scale based on the average size of the three homoeologous chromosomes 1A, 1B, and 1D. Names of the gene-rich regions are given on the left side of the consensus chromosome. In the nomenclature of the gene-rich regions (e.g. 1S0.8), the first digit represents wheat homoeologous group followed by the short arm (S) or long arm (L) letter. The last two numeral numbers represent fraction length of the gene-rich region location. Actual physical size (black) and the ratio of physical to genetic distance (red) for a region are given on the right-hand side of the consensus chromosome. Sizes (in Mb) of gene-rich regions were calculated based on the cytological measurements (measurements of the region bracketed by the flanking deletion line breakpoints in comparison with the total genome size in microns), and were drawn to scale. Percentage of genes in a gene-rich region with respect to the total group 1 genes was calculated from the overlapping deletion line mapping results for 147 cDNAs, and PstI genome clones from 26 different libraries from seven different species of family Poaceae. Recombination in a chromosomal region was calculated by comparing a deletion line-based physical map with the consensus genetic linkage map of Triticeae (Van Deynze et al., 1995a; Sandhu, 2000).

Analysis of the gene-containing regions and sequence data generated in different members of grass family suggest that the gene-containing regions are partitioned into the gene-rich and poor compartments (see later sections). The gene-rich regions appear to vary in the number and size of the "mini" gene-rich regions, and the size of the interspersed gene-poor compartments. For example, the "1S0.8 region" seems to have smaller number/proportion of gene-poor compartments as compared with the "1L0.7 region" because of the 12-times difference in gene density (Fig. 1). The gene density among the mini-gene-rich regions may further vary as much as 10-fold (see later sections).

The precision of localizing the gene-rich regions is a function of the number of deletion lines. The actual physical size of the gene-rich regions is probably much smaller because imprecise bracketing due to fewer number of deletion lines will result in overestimate of the spanning region. This is evident from the studies, where the individual maps of chromosomes 1A, 1B, and 1D localized the genes to 50% of the chromosomal region. With a 3-fold increase in the number of deletion line on the consensus map, genes for the same chromosome were localized to 10% of the chromosomal region (Gill et al., 1996a, 1996b; Sandhu, 2000). These results suggest that with availability of additional deletion breakpoints, the gene-rich regions may further be localized to much smaller physical regions than the proposed 10%.

The proposed gene distribution model of wheat is based on a fairly random sample of about 1% to 5% of the total genes (Gill et al., 1996a, 1996b; Sandhu et al., 2001). The genes/markers included in the gene distribution studies were: random clones from 26 different cDNA libraries of seven different Triticeae species, morphological markers, agronomically important genes, and PstI genomic clones. A possible exception is the multicopy gene families, which are probably not well represented in the study because of the technical mapping difficulty. It would be interesting to study the distribution of multicopy gene families in comparison to the single/few copy genes.

	GENE-POOR REGIONS

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

DNA reassociation kinetics studies showed that non-transcribing repeat (NTR)-DNA is an integral part of most plant genomes and its amount is proportional to the genome size (Flavell et al., 1974). Most plant genomes are large and complex and NTR-DNA is primarily composed of retrotransposons (Bennetzen et al., 1998; Shirasu et al., 2000; Wicker et al., 2001). The NTR-DNA of higher plants can be grouped into a few distinct classes based on the sequence comparisons (for review, see Bennetzen, 2000). Thus, the current composition of plant NTR-DNA seems to be a result of multiple invasions by different retrotransposons. Replication of retrotransposons then occurred followed by their inactivation by transpositioning and/or heterochromatinization. The NTR-DNA is unevenly distributed in the plant genomes. Paucity of genes observed from physical maps (Sandhu et al., 2001) and an abundance of heterochromatin visualized as C-bands (Dvorak and Chen, 1984; Curtis and Lukaszewski, 1991; Gill et al., 1991; Jiang et al., 1996) allegorize that repetitive DNA is abundant around the centromeric regions (Copenhaver and Preuss, 1999; Puechberty et al., 1999). The same seems to be true in the smaller plant genomes, as shown by the sequencing information in Arabidopsis and some other organisms (Miller et al., 1998; Presting et al., 1998; Copenhaver et al., 1999).

In addition to the retrotransposons, pseudogenes seem to be an important part of the non-transcribed chromosomal regions (Watterson, 1983; Galili and Feldman, 1984; Zhu et al., 1994; Wendel, 2000). Resistance gene analogs are the best studied example. DNA fragments with structural similarities to the known disease and pest resistance genes have been isolated in many plant species such as soybeans (Glycine max), potato (Solanum tuberosum), rice, barley, wheat, beans (Phaseolus vulgaris), etc. (Kanazin et al., 1996; Leister et al., 1996, 1998; Yu et al., 1996; Feuillet et al., 1997; Zhou et al., 2001). Very few of the resistance gene analogs have been shown to be transcribing (Collins, 1999). Pseudogenes are expected to be particularly abundant in polyploids (for review, see Stephens, 1951; Wendel, 2000), such as bread wheat, where most genes have three structural copies. However, a significant proportion of the wheat genes follow a single-factor Mendalian inheritance (3:1 ratio in F₂), suggesting that only one of the three copies is functional. The other two copies are either nonfunctional or have acquired a different function. An example of such a case was observed in wheat, where the homoeologs of two cDNA clones were observed in the functional centromeric region (Sandhu et al., 2001). Most centromeres are highly heterochromatic and are expected to be inactive for gene expression. In the other two homoeologous chromosomes, these two cDNAs are not present in the centromeric region (Sandhu et al., 2001). Although not confirmed yet, expression of the centromeric copies of the genes is highly unlikely. These observations suggest that perhaps there are genes present in the highly heterochromatic, gene-poor blocks of the genome that are probably not expressing because of the surrounding chromatin structure. A similar observation, at a much finer scale, however, was made in yeast (Schizosaccharomyces pombe), where a stretch of approximately 11 kb of DNA was inactive for gene expression as well as recombination (Grewal et al., 1998). Histone deacetylase-mediated chromatin remodeling in the region, however, initiated both gene expression and recombination.

Centromeric regions are not the only places that are abundant in repeated DNA. Regions present between two gene-rich regions are also composed of NTR-DNA (Fig. 1). Some of the regions present even near to the tip of chromosome arms are deficient in genes.

	GENE DISTRIBUTION IN TRITICEAE

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

Gene synteny and colinearity is conserved among Triticeae genomes. The indirect evidence came from Sears (1954), where he showed that loss of any Triticeae chromosome can be at least partially compensated for by one of its homoeologous chromosomes. The term "homoeologous" was coined to represent loss-reparatory, non-homologous chromosomes. The first direct evidence came from the isozyme marker analysis in the early 1980s, where it was shown that the marker location and the relative order were conserved among homoeologs (Hart, 1987). Comparisons of the high-density genetic linkage maps have shown that the gene order and relative recombination are so conserved among Triticeae species that it is possible to construct an accurate consensus map (Van Deynze et al., 1995a). These observations strongly suggest that the structural and functional organization of Triticeae genomes is very similar. Confirmed in barley, the distribution of genes among other Triticeae species, therefore, is similar to that of wheat (Kunzel et al., 2000). Translocation breakpoint-based physical maps clearly partitioned the barley chromosomes into gene-rich and gene-poor compartments (Kunzel et al., 2000). The location and the relative size of gene-rich regions in barley were very similar to that of wheat.

	DISTRIBUTION OF GENES IN POACEAE

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

Now, it is becoming evident that the gene order is conserved among the living organisms and its extent depends upon the evolutionary distance. Comparisons of different genetic maps exemplified this statement and showed that similarity is much greater among the Triticeae genomes as compared with other Poaceae species (Moore et al., 1995). The order of about 62% markers is conserved between rice and maize (Ahn and Tanksley, 1993) compared with 94% between wheat and oat (Van Deynze et al., 1995b). The similar estimates for the wheat and barley comparisons were even higher than 94%. Translocation breakpoints based barley physical maps showed that the gene order, and the extent and physical distribution of recombination were very similar to that of the three wheat genomes (Kunzel et al., 2000). Sequencing information near the adh1-F region in maize showed that intergenic regions were mainly composed of retrotransposons inserted within each other (SanMiguel et al., 1998). Wheat and maize diverged about 60 million years ago, but still conserved chromosomal segments are observed. The long arm of maize chromosome 9 and most of wheat chromosome 7 have originated from a common ancestral chromosome (Devos et al., 1994).

Various Poaceae genomes can be aligned by reshuffling the conserved linkage blocks of individual chromosomal segments (Moore et al., 1995). This alignment was primarily based on DNA and morphological markers that were used as anchors. Even in smaller genomes, like rice, gene-containing regions account for about 24% of the genome (Barakat et al., 1997). As observed in wheat, barley, maize, and rice, partitioning of genomes into gene-rich and gene-poor compartments; therefore, it most likely occurs in all Poaceae genomes. Gene distribution has been studied in animal systems and it seems that all animal genomes and chromosomes, to some degree, are divided into gene-rich and gene-poor compartments (Clay and Bernardi, 2001; for review, see Sumner et al., 1993).

	RELATIONSHIP BETWEEN DISTRIBUTION OF RECOMBINATION AND GENES

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

Comparisons of recombination among wheat and rye (Secale cereale) C-bands revealed that recombination is uneven along the Triticeae chromosomes (Dvorak and Chen, 1984; Curtis and Lukaszewski, 1991; Stein et al., 2000). Similar observations have been made in other plant and animal systems (Rick, 1971; Bollag et al., 1989; Ganal et al., 1989). Comparisons of deletion line-based physical maps with the genetic linkage maps elegantly showed that most of recombination occurs in the gene-rich regions of the wheat genome (Gill et al., 1996a, 1996b; Sandhu et al., 2001). Similar comparison in barley using translocation breakpoint-based physical maps confirmed its similarity in recombination distribution to that of wheat (Kunzel et al., 2000). In wheat, comparisons of the low-density wheat maps suggested that the recombination only occurs in the gene-containing regions (Werner et al., 1992; Gill et al., 1993; Kota et al., 1993; Delaney et al., 1995a, 1995b; Mickelson-Young et al., 1995). The high-density map comparisons confirmed that essentially no recombination occurs in the gene-poor regions (Gill et al., 1996a, 1996b; Weng et al., 2000; Sandhu et al., 2001). However, it became apparent that gene-rich regions differ in the extent of recombination (Sandhu, 2000). The gene-rich region "1S0.8" has 30-fold higher recombination as compared with the "1S0.4 region" (Fig. 1). Centromeres are known to suppress recombination in the vicinity in most of the higher eukaryotes (Puechberty et al., 1999). Essentially no recombination is observed in the proximal 30% of the wheat chromosomes, despite the presence of the gene-rich regions (Fig. 1).

Besides the centromere, other factors also seem to affect the extent of recombination. A proximal gene-rich region of wheat chromosome 6 ("6L0.4 region") has about 8-fold higher recombination than a distal gene-rich region ("6L0.7 region") of the same chromosome (D. Sandhu and K.S. Gill, unpublished data). The extent of recombination varies greatly even within the same gene-rich region. In barley, segments within a 1-Mb gene-rich region may vary as much as 10-fold for the extent of recombination (Wei et al., 1999).

	STRUCTURE OF THE GENE-RICH REGIONS OF POACEAE

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

Despite the conservation of gene synteny and colinearity, Poaceae genome size may vary as much as 35 times among species. Given that all the Poaceae genomes seem to be partitioned into gene-rich and -poor compartments, it is imperative to understand if the amplification is uniform within genomes or confined only to the gene-poor regions. The first line of evidence proposing a non-proportional amplification of the gene-poor regions of the larger genomes comes from the size estimates of the gene-rich regions. In the case of proportional amplification, the gene-containing fraction is expected to be the same in all Poaceae species. The best estimates for the gene-containing regions of wheat, barley, maize, and rice are 7%, 12%, 17%, and 24%, respectively (Carels et al., 1995; Barakat et al., 1997; Sandhu et al., 2001). Further, the gene density of the gene-containing regions should differ, proportional to the genome size. In case of uniform amplification, the gene density of wheat should be 35 times less than that of rice. However, recent studies suggest that the average gene density within the Triticeae gene-rich regions is 10 to 20 genes per 100 kb, compared with 15 to 25 in rice.

Gene density in gene-rich regions in grasses is comparable with the average gene density in Arabidopsis, which is one gene per 4 to 5 kb (Quigley et al., 1996). The higher gene density example is in Lrk10 region of wheat, where an average distance between genes is 4.6 kb (Feuillet and Keller, 1999; Feuillet et al., 2001). The lower gene density example is for a gene-rich region around the Mlo locus of barley, where the same estimate is 20 kb (Panstruga et al., 1998). In this study, it is interesting to note that all three genes were present within 25 kb of the total 60-kb contig. In a 16-kb region containing the starch-branching enzyme I of Ae. tauschii, the gene density was 19 genes per 100 kb (Rahman et al., 1997). A 340-kb rice region around Adh1-Adh2 contains 33 genes, with an average of one gene per 10.3 kb (Tarchini et al., 2000).

Direct comparisons of the similar regions between larger and smaller genome species suggest that the gene-containing regions are about two times larger irrespective of the difference in the genome size. Difference in genome size between maize and sorghum is 3.3 times compared with 35 times between wheat and rice. However, the difference in gene density is about the same. The distances among genes in the receptor-like kinase gene cluster in wheat were only 2 to 3 times more than the homoeologous region in rice (Feuillet and Keller, 1999). The sorghum Adh region of 78.2 kb corresponded to 225 kb in maize (Tikhonov et al., 1999). The distance between Adh1 and u22 was 2.4 times higher in maize as compared with sorghum (SanMiguel et al., 1996). These observations suggest that the amplification of a genome in gene-containing regions is much less than in gene-poor regions. Occasionally, localized amplification may occur in a region because of retrotransposon invasion irrespective of the species or the size difference. The genes Sh2 and A1 are 21 kb apart in rice, 22 kb in sorghum, and 140 kb in maize (Civardi et al., 1994; Bennetzen et al., 1998). For most other regions compared so far, average amplification in maize was only about 2 to 3 times that of sorghum and rice.

At a lower resolution, genes in cereal genomes appear to be clustered in small chromosomal regions separated by large blocks of repeat retrotransposon-like sequences (SanMiguel et al., 1996; Barakat et al., 1997; Llaca and Messing, 1998; Feuillet and Keller, 1999). At a higher resolution, gene-rich regions appear to be further consisted of mini-gene-rich regions interspersed by NTRs (Fig. 2). At sequence level, genes present in mini-gene-rich regions are further separated by intergenic NTR sequences consisting of retrotransposons (Fig. 2). Currently available data suggest that size and distribution of the mini-gene-rich region or the intergenic region do not follow any pattern.

View larger version (35K): [in this window] [in a new window]   Figure 2.   A wheat chromosome representing genome organization and gene distribution in cereal genomes. A gene-rich region and a gene-poor region are further amplified to show structural organization within these regions.

	EVOLUTION OF GENE-CONTAINING REGIONS IN POACEAE

TOP ABSTRACT INTRODUCTION GENE DISTRIBUTION IN WHEAT GENE-POOR REGIONS GENE DISTRIBUTION IN TRITICEAE DISTRIBUTION OF GENES IN... RELATIONSHIP BETWEEN... STRUCTURE OF THE GENE-RICH... EVOLUTION OF GENE-CONTAINING... LITERATURE CITED

It is now clear that retrotransposon invasion is the main factor contributing to the differential amplification of the Poaceae and other genomes. Retrotransposon insertions have doubled the size of the maize genome in the last 3 million years (SanMiguel et al., 1998). Presence of species- (Kumar et al., 1990; Pestova et al., 1998; Grutzner et al., 1999; Linares et al., 1999), genera- (SanMiguel et al., 1998; Schmidt et al., 1998; Staginnus et al., 1999; Pearce et al., 2000; Shi and Endo, 2000), and family- (Zhang et al., 1995) specific repeats strongly suggest a repeated invasion by different retrotransposons over time. Differences in the insertion dates for retrotransposons near the Adh1 locus were observed between maize and sorghum (SanMiguel et al., 1998). Five repeat sequences were found common between maize and wheat, of which only one was common between oat and maize (Zhang et al., 1995). The repeats common between wheat and maize were probably in the progenitor genome before divergence. The same argument can be extended to tribe- and species-specific repeats (Fig. 3). In addition to the new invasions, the proportion of the existing repeats fluctuates and perhaps some repeats get eliminated. Absence of four repeats, common between maize and wheat, in oat supports this hypothesis (Zhang et al., 1995).

View larger version (31K): [in this window] [in a new window]   Figure 3.   A hypothetical model for evolution of gene-containing regions in cereal genomes.

Upon invasion, two processes will restrict retrotransposon amplification to adjacent gene-poor regions, resulting into tandem blocks of NTRs. First, retrotransposons preferentially transpose to the nearby chromosomal location (Van Schaik and Brink, 1959; Greenblatt and Brink, 1962). Second, a gene-rich region containing genes flanked by genes important for fertility and viability will be resistant to retrotransposon invasion. However, transpositions in the intergenic regions of gene clusters will also be tolerated. Retrotransposon-like repeat sequences have been observed in almost all gene-containing regions sequenced, although their number and distribution vary (SanMiguel et al., 1996; Bennetzen et al., 1998; Llaca and Messing, 1998; Panstruga et al., 1998; Feuillet and Keller, 1999; Tarchini et al., 2000; Dubcovsky et al., 2001; Feuillet et al., 2001). The number of repeats in the intergenic regions should be a good indicator of the importance of the surrounding genes for the viability of the plant and its relative location within the gene-rich region. Sequence information around zein gene cluster in maize (Llaca and Messing, 1998) and Mlo locus of barley (Panstruga et al., 1998) suggested that very few retrotransposons were found in these regions. Because of the least selection pressure, amplification of the gene-poor regions will be most favored, resulting in the blocks of NTRs separating the gene-containing regions. Junctions of the gene-rich and -poor blocks are expected to have more intergenic repeats and thus lower gene density.

For the survival of any species, large-scale amplification of retro-elements needs to be stopped at some point. Insertional inactivation by adjoining retro-elements seems to have played a major role in restricting retro-element proliferation. The maize region containing Adh1-F and u22 genes was mainly composed of nested clusters of retro-elements inserted within each other (SanMiguel et al., 1996). In the gene-poor regions, heterochromatinization because of long stretches of inactive DNA will silence any active retro-element or gene. Open reading frames have been observed in the centromeric regions of wheat, humans, and Arabidopsis, which are known to be transcriptionally inactive and highly hetrochromatic (Copenhaver et al., 1999; Puechberty et al., 1999; Sandhu et al., 2001).

The reason for the differential amplification of the genomes within the same family is not known; however, it seems to depend upon the type of the invading retro-element. For example, the long terminal repeat type of retrotransposons seems to transpose preferentially into the gene-poor region, whereas the miniature inverted-repeat transposable elements type prefer gene-rich regions (for review, see Bennetzen, 2000). Because of the selection pressure, the extent of amplification for the retro-elements preferring gene-poor regions would be higher than the elements preferring gene-rich regions. Majority of the NTRs in larger genomes are composed of elements preferring gene-poor regions for transposition.