Synergy of Two Reference Genomes for the Grass Family

If one considers the Gordon Conference in Plant Genetics in 1997 as the origin of the international effort to sequence the rice (Oryza sativa) genome and the publication of the data in 2005 as the completion of production and analysis, it took about 8 years and 14 major laboratories from nine countries (International Rice Genome Sequencing Project, 2005). In the case of sorghum (Sorghum bicolor) with nearly twice the size of the rice genome, its origin dates back to the Plant Animal Genome Meeting in 2005 and it has a completion date in 2008, with only three production laboratories and a smaller group of analysts than for rice (Paterson et al., 2009). This comparison clearly demonstrates the acceleration and efficiency in genome analysis. While sequencing cDNAs of any particular species has been and will be important for our understanding of gene expression and the annotation of genes, it has been insufficient to gain insight in the organization of genes in their chromosomal context. Protein and mRNA accumulation does not distinguish the contribution of highly homologous gene copies. Furthermore, we know now that major traits are manifested by sequences not transcribed. For instance, the domestication of maize (Zea mays) is linked to a sequence 60 to 90 kb upstream of the coding region of the Tb1 gene (Clark et al., 2004), indicating that coding regions are a poor predictor of the sizes of genes and that transposable elements in introns and nontranscribed regions are filtered by expression. Therefore, the foundation for the genetic and evolutionary analysis of any organism will be ultimately its genomic sequence. One of the arguments has been that the coding portion of complex genomes represents only a small fraction of the entire genome, and this subset, rather than the rest, should be sequenced because of cost and time (Whitelaw et al., 2003). Based on such arguments, the sorghum genome has been sequenced by applying methylation filtration technology with fewer than 550,000 reads (Bedell et al., 2005). However, distribution of presumably genetically inert sequences appears to be uneven, preventing simple fractionation methods to separate these types of DNA material for selective sequencing (Haberer et al., 2005). Such attempts also would fail to provide us with an understanding of spacing and linear order of sequences. Moreover, increasing knowledge of repetitive DNA sequences, their role, and divergence not only shines new light on chromosome evolution and gene regulation, but also improves the resolution of genetic maps and the algorithms to assemble contiguous sequences from shotgun reads. For instance, not only SSR markers, but also nested retrotransposons, have become invaluable genetic markers (Devos et al., 2005). Critical to the sequencing of large chromosomes has been the DNA shotgun sequencing method and the use of single and paired synthetic universal primers (Messing et al., 1981; Vieira and Messing, 1982). The method is based on fragmenting DNA into small sizes, purifying them by cloning, and defining the start of sequencing with a short oligonucleotide. Because fragmentation produces overlapping fragments, sequences can be concatenated by overlapping sequence information (Larson and Messing, 1982), thereby reconstructing contiguous sequences (contigs), which was first exemplified by the complete structure of a plant DNA virus (Gardner et al., 1981). In principle, there is no limit to the reiteration of contig building and, indeed, large chromosomes have been assembled (International Human Genome Sequencing Consortium, 2004). There are two valuable experimental validations to properly assemble contigs. One is restriction mapping because the sizes of restriction fragments have to match those predicted from the assembled sequence, which was already used in the case of the plant DNA virus (Gardner et al., 1981). The second is recombination maps because the genetic map has to be collinear with the sequence. These types of validations have also resulted in strategies to sequence genomes in increments and by large consortia, as it was the case for the rice genome (International Rice Genome Sequencing Project, 2005). Restriction fragment length patterns can be used to construct clone maps from bacterial artificial chromosomes (BACs) containing genomic DNA of 100 to 150 kb. Because genomic libraries are formed from partially digested genomic DNA, complete digests produce common-sized restriction fragments from overlapping clones, facilitating their concatenation into BAC contigs, called finger-printed contigs. Moreover, the process can be adapted to high-throughput robotics so that even large genomes can be mapped in relatively short periods of time (Nelson et al., 2005). Taking advantage of genetically mapped DNA sequences, these finger-printed contigs can be ordered according to their chromosomal position. In turn, sequences linked to individual BAC clones also become mapped in silico. Such physical map resources are critical for any genome sequencing approach, whether it is a concerted effort sequencing a selection of ordered overlapping clones or whole-genome shotgun (WGS) sequencing. Furthermore, if one genome is sequenced, alignment of BAC-linked sequences of a second closely related genome can be used to order clone libraries of the second genome (Gregory et al., 2002). Indeed, both the sequencing of sorghum and maize are based on physical maps that have been constructed based on the rice sequence (Wei et al., 2007; Paterson et al., 2009). Another major difference of sequencing sorghum by WGS compared to the human genome by WGS (Venter et al., 2001) was the generation of shotgun libraries with different insert sizes of 2.44, 6.4, 6.88, 8.6, 34.7, 91, and 108 kb and sequencing its ends at different depths (Paterson et al., 2009). By careful selection of sheared DNA fragments within a narrow size range and long paired high-quality reads of over 700 bp, sequence assembly algorithms can rely on the distance of two paired reads (Jaffe et al., 2003), resulting in better concatenation and fewer supercontigs. Recent advances in using solid support as opposed to capillary electrophoresis (CE) to increase throughput of sequence reads have introduced a new level of massive parallelization of sequencing, also referred to as nextgen (next generation) sequencing, although the concept of single-nucleotide extension by polymerase represents the oldest sequencing method (Wu and Kaiser, 1968). While these advances have permitted deep sequencing of RNA mixtures and single-nucleotide polymorphism analysis of sequenced genomes (Brenner et al., 2000), it has been less clear whether they would also be a new leap in allelic sequencing or even sequencing entire genomes. The main force would be cost and speed. Recently, we tested a SOLiD (sequence by oligonucleotide ligation and detection) nextgen system and it appears to be feasible to achieve 1 Gb/d for $1,000. The technical challenge, however, is the read length of the nextgen systems. It may be conceivable to apply a hybrid approach and reduce redundancy of CE sequencing from 8 to 10 times to a much lower level by mixing them with short paired reads from nextgen systems. Still, in addition to read length, accuracy of the nextgen sequencing technology has to go into the equation as well. Although 454 sequencing achieves longer sequencing reads than other nextgen sequencing methods, it also appears to have higher error rates up to 10% depending on sequence composition (Brockman et al., 2008). It also has a lower throughput than the Solexa system, but Solexa has very short reads of about 35 nucleotides and an error rate of 2% to 3%, which is compensated by the higher redundancy of sequences (Dohm et al., 2008). The new SOLiD system differs from the 454 and Solexa systems because it uses DNA ligase for querying the order of nucleotides and claims to have an error rate of only 0.1% (http://appliedbiosystems.cnpg.com/lsca/webinar/rhodes/chemistry/20070618), but it also has very short reads like Solexa, although ABI claims that read length of more than 100 could be achieved with improved dyes and emulsion PCR. Nevertheless, a hybrid approach of CE and nextgen sequencing, particularly if combined with paired reads, should provide accelerated access to genomic sequences and be a catalyst for comparative genomics. If hybrid sequencing approaches are applied in the future for additional genomes of the Poaceae, the use of reference genomes of high quality will have an even greater impact. As demonstrated by various controls, the physical map of sorghum, the synteny with rice, and its new sequencing strategy contributed to a high-quality sorghum genome sequence in record time and great efficiency. Therefore, one can consider both the rice and sorghum genome sequence as a synergistic reference. Indeed, alignments of genes from rice and sorghum with maize has led to the identification of genes conserved through ancestry, also called orthologous gene copies, whereas genes in either of the three genomes that deviate from a common order are inserted in chromosomal regions by illegitimate recombination similar to transposable elements after speciation and are referred to as paralogous gene copies (Messing and Bennetzen, 2008). In this article, we can see how orthologous and paralogous sequences have been identified in rice, sorghum, and maize. Sequence alignments have played a critical role in defining related DNA sequences (Maizel and Lenk, 1981). In most complex genomes, like those of plants, amplification of small genomic sequences and dispersal of sequence copies by illegitimate recombination have been the most common mode to generate a large proportion of repeat sequence families. Because of the transition through RNA intermediates (class I transposable elements), the increase in copy number of specific sequences is very high compared to DNA transposition (class II transposable elements), which occurs either through single copies or without replication by straight transposition. Therefore, it is not surprising that the major force in genome size variation in plants is the differential burst of retrotransposition (Messing and Bennetzen, 2008). Because at the time of insertion the retrotranscript forms long terminal repeats (LTRs), the insert is flanked by identical sequences that can undergo recombination before meiosis so that either a single retroelement is lost in the next generation or even unrelated sequences between two neighboring elements. This activity can then reverse chromosome expansion and the remaining LTR is called a solo LTR. However, detection of retroelements in the genome based on paired flanking LTRs is hampered by layered insertions and mechanisms that disrupt the intact structure of retroelements like chromosome breaks. Such fractured elements may simulate solo LTRs, but do not reduce chromosome expansion as rapidly as unequal crossovers of LTRs do. Expansion and contraction of chromosomes in progenitors of rice, sorghum, and maize. The assumption is made that expansion and contraction of chromosomes occurred in recent times long after speciation, perhaps mostly in the last 1 million years (Du et al., 2006). In rice, (1) expansion in pericentromeric regions was followed by (2) contraction and low retrotransposition throughout the chromosome (Ma and Bennetzen, 2004). Sorghum experienced mainly expansion in pericentromeric regions (Paterson et al., 2009). Maize underwent WGD of two diverged progenitors, then early on chromosome breakage and fusion, resulting in a mosaic of syntenous chromosome blocks (shown in different colors), which was followed by uneven expansion and contraction of those blocks et al., 2006). Therefore, distribution of retroelement is more even than in The size of maize is or (Wei et al., of sorghum (Paterson et al., and of rice (International Rice Genome Sequencing Project, 2005). forms of have been a the is DNA and Because the a these elements have been called Although are the detection of with sequences has been in to transposable the terminal sequences different species et al., et al., et al., 2008). analysis of these elements can be it to allelic sequences from the species to individual of sequences for Still, the to which they chromosomal expansion is very small because in sorghum they to about and in maize of genomic DNA (Paterson et al., 2009). the other they presumably to of genes and gene fragments because they can copy orthologous gene copies and Messing, 2006). a similar can be contributed to transposable elements in rice et al., and elements in sorghum (Paterson et al., 2009). Therefore, transposable elements to and in as for paralogous gene copies. a genome of II elements only a small of genomic sequences. One could be that they are more because in to insert into genes and have even been called a for genes et al., 2002). Another is that they are not and undergo is that they are relatively small in size and can be with to In genes transposable elements without on gene and are as as retroelements in rice, but in sorghum only and in maize of the repetitive DNA (Haberer et al., 2005). This distribution is with the gene in these genomes. Furthermore, in sorghum, retroelements are in pericentromeric and genes in are nearly collinear with the regions (Paterson et al., 2009). of repeat elements for genome the detection of genes because can be with repeat sequences. Because repeat sequences are also it gene from the with genomic sequences (Messing et al., 2004). important of gene can from collinear of If repeat sequences very after speciation, the that the sequence inserted into the order in one species compared to is Therefore, one can the assumption that predicted genes conserved in the order in chromosomal regions do not to either of transposable elements and are Still, within the genome, orthologous genes can be and have a as a paralogous gene mechanisms may to gene copies most by unequal Another or is of a short genomic sequence containing a single gene inserted The is of a gene by a transposable to rice and sorghum, maize chromosomal regions match them indicating that maize underwent a WGD et al., 2004). after genes more than of the time the second copy was lost (Messing et al., 2004). It has been that unequal over between LTRs led to of the second copy because the of the other copy it and of chromosomal regions greater to meiosis 2009). In addition to unequal over of LTRs, II elements could have played an even greater role in chromosome contraction because of the of to on the ends of two elements by long of sequences and and 2008). The was by meiosis to of Still, the remaining copies alignment of chromosomal regions of maize with sorghum and Because sorghum is collinear with rice million years of nearly the of the we can that the rice genome the genome perhaps for the of and chromosome of species that deviate from chromosomal with rice and sorghum presumably formed by and of chromosomal Therefore, alignment of rice chromosomes with the gene map of maize can be used to the chromosome of the two progenitors of which to have 10 chromosomes (Wei et al., Indeed, the of sorghum also 10 chromosomes and from the two progenitors of maize about the million years et al., 2004). While the sorghum not undergo the two other progenitors to maize. Therefore, maize first but selection meiosis massive a of 10 and the of 10 chromosomes of smaller This was with and Furthermore, insertions of retroelements contributed to the divergence of regions and a second of genome size of the maize genome after a genome is to a can of chromosomal (Maizel and Lenk, 1981). Rice has such major et al., that range in size from 1 to with a of and about of the Therefore, there are genes by these one could that the of the has also by However, based on rates of gene copies, it is to the time that this occurred because of the are over a time before the speciation of rice et al., for the one of regions of rice chromosome and which could have occurred as as million years and Sequencing Consortium, 2005). could from sequencing a genome that to rice as to Sequencing of the organism a genetic that over time by the second copy of genes in the genome et al., 2004). of an back to million years before the progenitors of rice, sorghum, and maize million years 2001) and on rice chromosomes and that it the of the maize this chromosomal orthologous gene copies in sorghum can be and Messing, Rice is orthologous to sorghum whereas rice is orthologous to sorghum with the that gene copies already in a of these Therefore, the rice and sorghum genomes collinear (Paterson et al., 2009). Furthermore, because maize from of two progenitors, rice is orthologous to maize and maize and rice is orthologous to maize and maize 1 1 in and Messing, The gene on maize chromosome to a gene copy on sorghum and rice While the of the gene on maize chromosome 1 is on maize was by on maize but the was then the other on maize 1 after and on sorghum chromosome also has two copies. of the maize gene indicating that the has diverged so that the gene copies to This is with the assumption in most gene in et al., 2003). Indeed, the alignment of these orthologous chromosomal that insertion and of genes occurred before and after the of rice, sorghum, and maize so that the orthologous gene between any of two species and has now of genes and Messing, analysis also that chromosome have occurred these within the to gene Such a also the concerted evolution of gene the other one can to such a because of selection of homologous chromosome In this of between the be insertions and would be with such The on rice is than on rice Rice has more genes than rice in the the is even more in maize than in rice and maize and to rice and are the maize and are compared to rice and This difference in size of maize also a that one has lost a gene As already the could be by contraction through unequal over of neighboring LTRs or of a on a of DNA transposable elements. In the alignments also and the entire of sequence of plant The mechanisms however, do not the and sizes of gene families. The most common for the amplification of gene copies is unequal over resulting in Still, layered single gene insertions or into can disrupt or the size of such the most gene RNA genes (Messing et al., However, the first plant genomes been sequenced, it that about of the genes in and of the genes in rice have been Genome Rice Genome Sequencing Project, 2005). The number appears to be lower with only of the genes in sorghum as (Paterson et al., in genome the of genes and of alignments so that comparison is In any whole-genome analysis a large of gene amplification in plant genomes. of genes are not and to variation in the size of gene in size from two to in the rice genome (International Rice Genome Sequencing Project, 2005). The gene in sorghum appears to with a the of sorghum (Paterson et al., 2009). The most rapidly gene are For instance, the in maize unequal within single and Bennetzen, However, this of amplification does not the insertion of to transposable elements and as that occurred in the of species do not the DNA with these types of of orthologous regions of rice, sorghum, and maize containing the genes that the genes after the divergence of the to million years and Messing, The gene was and inserted into a different chromosomal to million years which then was Therefore, the progenitors of maize and sorghum have both but rice does It is that one of three progenitors for maize and sorghum about million years the three progenitors different in gene As a this gene copy is in sorghum and one of the progenitors of maize. this gene copy is in the chromosomal not because of gene but because it was Another occurred million years in which is also in the orthologous region in sorghum and the region in that a similar without the use of transposable elements also after In with one in sorghum and maize underwent similar of gene insertion and without the use of transposable elements has also been in rice et al., 2004). This analysis that gene copies at different of divergence of within the of species and for maize contributed to of straight and Messing, While is about gene it is clear that gene copies are not and The are the of gene copies from WGD For instance, a gene accumulation in the of maize of gene copy to different of these resulted in the of the and genes in maize that a with the but different and Because accumulation is not for plant other of diverged of the have been in that is a common for gene et al., 2003). In the case of the genes it is to of copies of genes in only to be in sorghum, of genes, are a lower level of gene than in maize and Messing, One of gene appears to be the of and Messing, It appears that those genes can be but their at a very level and and Messing, 2003). It has been that of accelerated of mRNA and Furthermore, are less as and their Another is gene resulting in the as However, in the gene of intact genes could gene and Messing, one is This is similar to the of transposable elements and et al., et al., of the of a gene that are from in which can with the of DNA by For instance, it has been that the in to the has in the region that in the gene expression should et al., Furthermore, the is in the but not in the One of the in the expression of paralogous gene copies is highly conserved coding sequences. and sequence alignments are and can only be by individual and genomic gene copies. By of a gene by genomic one can a polymorphism to the analysis of sequences and Messing, allelic variation of gene copies is lower than paralogous copies even if they are Therefore, polymorphism can even be used to gene expression in different of the of expression of allelic gene copies also led to the of variation in gene copy number within the which can be based on additional sequence amplification or of sequence repeats and Messing, 2003). Furthermore, there to be in the expression of allelic and gene copies between different maize In the case of the most gene copies, one could even a divergence in regulation, which could the genetic on et al., have even been in plant are into different The of gene in allelic regions of the species is manifested in of a that have an important on the of Indeed, maize with different of gene in both gene expression of allelic and gene copies not an and Messing, 2003). the there was an on the increase and of gene indicating that could produce their expression the unequal expression of of a gene one about the role of less or The is in the annotation of predicted In which to the group of and are to be recent in and have that are not only but also in small RNA sequences, which could a role and 2008). Such a even a role in the differential expression of the maize in hybrid as Because of the of gene sequences within the plant sequence analysis of entire genomes of species have alignments of large chromosomal exemplified by the with the rice and sorghum genome as synergistic The of alignments has been in clone maps the genetic map and the in gene analysis. In the of genomes, diverged the of the us with a organization of of other genomes with such a organization has the of chromosomes from genomes and analysis of gene If we have from one plant of species is plant chromosomes an of sequence it was that sequences of genes are because of transposable elements. have now through alignments of orthologous chromosomal regions of species within the that are more as Moreover, the mechanisms such to be and are and