Comparative Analyses Between Lolium/Festuca Introgression Lines and Rice Reveal the Major Fraction of Functionally Annotated Gene Models Is Located in Recombination-Poor/Very Recombination-Poor Regions of the Genome

Abstract

Publication of the rice genome sequence has allowed an in-depth analysis of genome organization in a model monocot plant species. This has provided a powerful tool for genome analysis in large-genome unsequenced agriculturally important monocot species such as wheat, barley, rye, Lolium, etc. Previous data have indicated that the majority of genes in large-genome monocots are located toward the ends of chromosomes in gene-rich regions that undergo high frequencies of recombination. Here we demonstrate that a substantial component of the coding sequences in monocots is localized proximally in regions of very low and even negligible recombination frequencies. The implications of our findings are that during domestication of monocot plant species selection has concentrated on genes located in the terminal regions of chromosomes within areas of high recombination frequency. Thus a large proportion of the genetic variation available for selection of superior plant genotypes has not been exploited. In addition our findings raise the possibility of the evolutionary development of large supergene complexes that confer a selective advantage to the individual.

THE rice genome, which has been physically mapped, sequenced, and annotated, can be used to predict gene content and order among monocots. Thus rice provides a powerful tool for genomic research and map-based cloning of target genes in large-genome crop species (the increased genome size of these species results from the presence of large amounts of additional repetitive DNA). Comparative analysis between the rice genome and other monocot plant species has demonstrated that gene order in different monocot species has, to a significant degree, been maintained during evolution (Bennetzen and Freeling 1993; Keller and Feuillet 2000). However, there are many examples where the microsyntenic relationship between rice and other monocot species breaks down as a result of gene duplications, deletions, translocations, and inversions (Bennetzen and Ramakrishna 2002; Song et al. 2002; Sorrells et al. 2003; La Rota and Sorrells 2004; Miftahudin et al. 2005).

Irrespective of the <100% syntenic relationship between the monocots, the rice genome provides a very powerful tool for predicting the gene content and order in species with large and hence difficult to sequence genomes, such as grass (e.g., Lolium and Festuca species), wheat, barley, and oats, at the macrosyntenic level.

Previous comparative mapping analyses in large monocot genomes have been undertaken via genetic mapping or bin mapping of orthologous markers since full-genome sequencing of these species is not presently viable. Both methods have been used successfully (Van Deynze et al. 1995, 1998; Sorrells et al. 2003; Peng et al. 2004; Qi et al. 2004). The major limitations with genetic mapping are: (1) it is difficult and time consuming to map large numbers (thousands) of orthologous markers in sufficiently large mapping populations to enable high-resolution comparative analyses; (2) genetic mapping gives no indication of the physical distances between genes, which is of key importance for positional cloning; and (3) the requirement for polymorphism limits or prevents the mapping of some markers. Bin mapping allows the relatively fast mapping of large numbers of markers as has been demonstrated in wheat (Lazo et al. 2004; Qi et al. 2004). This system relies on large chromosomal deletions that divide the chromosomes into physically demarcated intervals or bins into which markers can be placed. A limitation of wheat bin mapping is the low resolution of the system; i.e., the number of bins into which each chromosome is divided is small; e.g., chromosome 3A is composed of only six bins (Munkvold et al. 2004). In addition, multiple chromosomal deletions frequently complicate deletion-bin mapping (Qi et al. 2003).

The data obtained from genetic mapping and deletion-bin mapping have indicated that large numbers of coding sequences in the monocots studied are located in gene-rich regions that undergo high frequencies of recombination. In wheat, for example, the deletion-bin mapping of 16,000 expressed sequence tag loci (ESTs) found that of 23 bins with higher gene densities, all but 1 were located in the distal 40% of the chromosome arm (Qi et al. 2004). Relatively few genes are thought to be located in proximal recombination-poor regions of the chromosomes (e.g., Erayman et al. 2004; Qi et al. 2004).

The Lolium/Festuca introgression system provides an alternative method of comparative mapping, similar to wheat deletion mapping but with potentially far greater resolution (King et al. 2002a,b). The Lolium/Festuca introgression system is based on a series of seven monosomic substitution lines. In each of these seven lines one of the chromosomes of Lolium perenne (2n = 2x = 14; DNA content = 2034 Mbp; Bennett and Smith 1976) has been replaced by its homeologous equivalent from Festuca pratensis (2n = 2x = 14; DNA content = 2181 Mbp; Bennett et al. 1982). The F. pratensis chromosome is then broken into different-sized segments in a BC₂ mapping population through recombination with the homeologous L. perenne chromosome. The ability to distinguish the chromosomes of Lolium and Festuca using genomic in situ hybridization (GISH), coupled with the high frequency of recombination between the chromosomes of Lolium and Festuca (King et al. 1998, 1999), enables the identification and characterization of individuals that carry different-size Festuca chromosome segments. For example, an introgression map of L. perenne/F. pratensis chromosome 3, homeologous to rice chromosome 1, has been generated. The introgression map is composed of 16 individuals, each of which carry different-size Festuca chromosome segments. Alignment of overlapping Festuca chromosome segments effectively divides L. perenne/F. pratensis chromosome 3 up into 18 physically demarcated bins. Screening the individuals that make up the introgression map of Lolium/Festuca chromosome 3 for the presence or the absence of Festuca polymorphisms allows genetic markers, e.g., restriction fragment length polymorphisms (RFLPs), amplified fragment length polymorphisms (AFLPs), simple sequence repeats (SSRs), and single-nucleotide polymorphisms (SNPs), to be assigned to one of the 18 introgression bins (King et al. 2002a,b).

A critical factor, in addition to comparative analysis, for map-based cloning and also for the development of efficient plant breeding strategies is knowledge of how recombination frequency and distribution relate to gene distribution throughout the genome. For example, the development of new plant varieties depends on the ability to assemble specific allelic combinations of genes that give rise to superior genotypes. This process requires the hybridization of specific parental material carrying target alleles. Recombination during meiosis leads, via reassortment, to the generation of individuals carrying the different combinations of alleles upon which the breeder can make selections. The ability to assemble specific allelic combinations is, however, completely dependent on the frequency and distribution of recombination. For example, it is relatively easy to select for specific allelic combinations and select against linked deleterious alleles located in regions of the genome that show high frequencies of recombination. In contrast, it is extremely difficult, if not impossible, to select for specific allelic combinations and to break linkages with deleterious alleles present in regions of the genome that show little or no recombination. Similarly a sufficiently high frequency of recombination is required for the isolation of genes via map-based cloning strategies such as chromosome landing; i.e., genes located in an area of high recombination will be relatively easy to isolate while genes located in regions of the genome with low recombination will be difficult if not impossible to isolate.

Here we describe the exploitation of the published rice genome sequence to bin map sequences from functionally annotated gene models on approximately every 5th to 10th BAC/PAC clone from rice chromosome 1 to the Lolium/Festuca chromosome 3 introgression map to (1) further elucidate the syntenic relationship between rice and the large-genome monocots, (2) determine the physical location of rice chromosome 1 genes in large-genome monocots, and (3) determine the relationship between gene distribution and recombination in large-genome crop species.

MATERIALS AND METHODS

Plant material:

A 14-chromosome L. perenne/F. pratensis monosomic substitution plant (13 Lolium chromosomes and a single group 3 Festuca chromosome; King et al. 2002a,b) was isolated from the progeny of a cross between a triploid Lolium/Lolium/Festuca hybrid (male parent) and diploid L. perenne, c.v. Liprio (female parent) (King et al. 1998). The monosomic substitution plant was backcrossed as the male parent to the same diploid L. perenne genotype (female parent) to produce a BC₂ mapping population (King et al. 2002a,b). The BC₂ population was genetically mapped using AFLPs (King et al. 2002a). Sixteen BC₂ plants carrying a recombinant chromosome were analyzed using GISH. The Festuca segments in these 16 genotypes were measured and used to produce an introgression map of Festuca chromosome 3 in which the chromosome was physically divided into 18 bins (King et al. 2002b).

DNA extraction, PCR, and clean up:

Genomic DNA was extracted using an AutoGen 740 (AutoGen, Holliston, MA).

All PCR amplifications carried out in this work were performed using Faststart Taq DNA polymerase and the manufacturer's buffer systems (Roche, Lewes, UK) in ABI9700 thermocyclers (Applied Biosystems, Warrington, UK). Thermal cycling was performed beginning with 5 min at 96°; followed by 10 cycles of 1 min at 95°, 1 min at 60° (with the temperature reduced by 0.5° per cycle), 3 min at 72°; followed by 30 cycles of 1 min at 95°, 1 min at 55°, and 3 min at 72°; and a final extension step of 7 min at 72°. For a small number of primers the initial annealing temperature was 55°, dropping to 50° over the course of the first 10 cycles.

DNA was cleaned either directly from the PCR amplification using the QIAquick PCR Purification Kit (QIAGEN, Crawley, UK) or from an agarose gel using the QIAquick Gel Extraction Kit (QIAGEN).

Development and introgression mapping of SNPs:

Strategy:

1. Develop primers, on the basis of conserved orthologous sequence alignments, from approximately every 5th to 10th BAC clone from rice linkage group 1.

2. Identify primers that generate SNPs between Lolium and Festuca group 3 chromosomes.

3. Assign each of the SNP markers located on Lolium/Festuca linkage group 3 to one of the 18 bins making up the introgression map of this chromosome.

Primer design:

Primers were developed from sequence from approximately every 5th to 10th BAC/PAC clone from rice linkage group 1 on the basis of the following three strategies:

1. Sequence from a putative gene on a selected rice BAC/PAC clone was put through a blastn search against other monocot ESTs and cDNAs in the public domain. Primers were designed from conserved regions of sequence in two different exons from the rice gene on either side of an intron.

2. Sequence from a putative gene on a selected rice BAC/PAC clone was put through a blastn search against other monocot ESTs and cDNAs in the public domain. Primers were designed from conserved regions of sequence from within one exon of the rice gene.

3. Access to a GeneThresher database (Gill et al. 2006) in which 25,000 sequences have already been aligned to the rice genome allowed the selection of a sequence from a putative gene on a BAC/PAC clone of interest and primers were then designed from within that sequence. [GeneThresher sequences used had all previously been aligned to a BAC/PAC clone on rice linkage group 1. As an additional check all sequences used were searched against the rice database using the The Institute for Genomic Research (TIGR) blastn function for coding sequences (CDSs).]

Rice genome sequence was obtained from The Institute for Genomic Research (http://www.tigr.org/tdb/e2k1/osa1/pseudomolecules/info.shtml).

Blastn searches against all monocot ESTs and cDNAs were carried out via Plant GDB blastn search (http://www.plantgdb.org/PlantGDB-cgi/blast/PlantGDBblast).

Alignment of the GeneThresher database with the rice genome sequence was carried out through GRAMENE (http://www.gramene.org/Oryza_sativa/).

SNP discovery (detection of polymorphism between Lolium and Festuca group 3 chromosomes):

All primers were initially tested by the PCR amplification of genomic DNA from the plants involved in the production of the chromosome 3 monosomic substitution [F. pratensis, L. perenne (diploid), and L. perenne (tetraploid)] (King et al. 1998), the L. perenne/L. perenne/F. pratensis triploid hybrid, and the group 3 substitution line itself with the amplification products run on a 1% agarose gel.

Primers that produced no amplification products in F. pratensis or F. pratensis amplification products that did not appear in either the triploid hybrid or chromosome 3 substitution when run on the agarose gel were discarded.

Primers producing a single amplification product in each of the five genotypes listed above and that showed no differences in running speed on agarose gels were re-PCRed and the products cleaned and sequenced.

Primers producing more than one band when run on the agarose gel had the correct size band (estimated from the rice sequence) cut from the gel. The DNA was then cleaned and sequenced.

Sequencing was carried out on an ABI 3100 (Applied Biosystems). The sequences obtained for the five genotypes were aligned using Genedoc (http://www.psc.edu/biomed/genedoc). The aligned sequences were then checked for the presence of single-nucleotide polymorphisms between the parental F. pratensis genotype and the two L. perenne genotypes. The presence of SNP markers located in the Lolium and Festuca group 3 chromosomes was confirmed by their presence in the sequences of the triploid hybrid and group 3 substitution.

Introgression mapping of group 3 SNP markers:

Having established the presence of a SNP between F. pratensis and L. perenne group 3 chromosomes the primers were then used to amplify genomic DNA of the 16 genotypes composing the Festuca introgression map. Screening for the presence or the absence of the F. pratensis and L. perenne bases in each of the 16 genotypes carrying different-sized and overlapping Festuca chromosome segments allowed each SNP, derived from rice linkage group 1 BAC/PAC clones, to be assigned to a bin on the Festuca introgression map. The sequence data from F. pratensis was blastn searched back against the rice genome to confirm that the products amplified in the Lolium and Festuca genotypes were orthologous to the original sequence obtained from rice.

In addition to the above procedure a modified procedure was applied for some primers designed across introns. These primers occasionally produced amplification products in the five genotypes [F. pratensis, L. perenne (diploid and tetraploid), L. perenne/L. perenne/F. pratensis triploid hybrid, and the group 3 substitution] that showed differences in running speed on agarose gels. DNA containing a heteroduplex of F. pratensis and L. perenne DNA, i.e., the triploid hybrid and the substitution line, showed retardation in the running speed compared to DNA from only one species. The lag in running speed was presumably due to imperfectly paired DNA, i.e., the presence of loops, etc., within the heteroduplex. The retardation was also observed when heteroduplex DNA was amplified in one or more of the 16 genotypes composing the Lolium/Festuca introgression map. It was therefore possible to physically locate the polymorphism (and hence the rice BAC/PAC clone on which it was located) to its correct bin on the Lolium/Festuca map. The amplified F. pratensis product was sequenced and the sequences blastn searched back against the rice genome to confirm amplification of the correct product.

RESULTS

Primer design:

Three strategies, as outlined in materials and methods, were followed for primer design and hence SNP discovery. Ninety-one percent of primers designed within two exons on either side of an intron produced an amplification product in both Festuca and Lolium. Of these, 37% were successful in physically mapping a BAC/PAC clone; i.e., the amplification products showed polymorphism between Festuca and Lolium. A number of primers produced in this way produced amplification products that failed to produce sequence data of sufficient quality (without cloning) for reliable SNP discovery. Ninety-six percent of primers designed entirely within an exon produced amplification products in Festuca and Lolium and of these 40% were successful in physically mapping a BAC/PAC clone. While sequence data generated from the amplification products of these primers were much clearer and more reliable, the sequences in general showed much lower levels of polymorphism between Festuca and Lolium. Ninety-five percent of the primers designed using the GeneThresher database produced amplification products in Festuca and Lolium, 30% of which were successful in the physical mapping of a BAC/PAC clone. This final strategy, however, was employed only on a very small number of BAC/PAC clones and these numbers do not therefore provide an accurate estimate of the success of this third strategy.

Syntenic relationship between rice and the grasses:

We bin mapped sequences from 69 BAC/PAC clones from rice chromosome 1 onto the Lolium/Festuca chromosome 3 introgression map, 67 of which were mapped using primers designed from sequence within functionally annotated gene models and 2 within putative transposon sequence (Figure 1). The BAC/PAC clones on rice linkage group 1 were numbered 1–393 according to the minimum tiling path from The Institute of Genomic Research as of September 2006. This numbering is used for Figure 1. The GenBank accession number for each BAC/PAC clone used is given in Table 1. An extended version of Table 1 is available in supplemental material at http://www.genetics.org/supplemental/. This table gives full primer details along with the locus identifier of the functionally annotated gene model used.

Figure 1.—

Introgression map of Lolium/Festuca chromosome 3. Physical and genetic distances are shown on the left. Horizontal black arrows indicate sites of recombination between Festuca and Lolium (alternating blue and yellow colors are used only to aid discrimination between different recombinant lines carrying different-sized segments). The positions of the centromere (blue arrow) and the nucleolar organizer region (NOR) (red arrows) are also shown on the left (King et al. 2002b). The NOR is also shown as a red overlay on the diagram of the chromosome. Bin allocation of the BAC/PAC clones from rice linkage group 1 is shown on the right (see Table 1 for GenBank accession numbers for each clone). BAC/PAC clones shown in pink map out of order as compared to rice.

TABLE 1.

Introgression-mapped rice linkage group 1 BAC/PAC clones

BAC/PAC clone no. as used in Figure 1	GenBank accession no.	BAC/PAC clone no. as used in Figure 1	GenBank accession no.
1	AP003727	203	AP003334
3	AP002863	207	AP003310
5	AP002882	213	AP003300
7	AP003219	215	AP003048
8	AP002867	221	AP003022
14	AP003233	227	AP002897
15	AP002860	230	AP002819
21	AP002483	233	AP003220
25	AP002522	245	AP003710
29	AP003209	258	AP003264
34	AP002523	264	AP003455
39	AP002484	270	AP003245
44	AP002746	296	AP003218
47	AP003244	299	AP003371
52	AP003052	303	AP003344
58	AP002745	305	AP003368
64	AP003434	308	AP003280
69	AP002855	310	AP003299
75	AP001278	313	AP003232
81	AP001366	318	AP003411
84	AP000969	321	AP003302
91	AP003103	325	AP003734
101	AP003206	335	AP004072
107	AP003934	340	AP003794
112	AP002953	348	AP003261
117	AP003143	352	AP003346
121	AP003793	355	AP003349
131	AP002870	366	AP003437
137	AP003021	368	AP006167
141	AP006530	371	AP004332
147	AP002968	376	AP003269
181	AP003445	380	AP003259
187	AP004233	389	AP003277
188	AP004243	391	AP003627
197	AP003924	—

Shown is numbering used for introgression mapping of BAC/PAC clones in Figure 1 together with the GenBank accession number used for each BAC/PAC clone (GenBank accession number obtained from The Institute of Genomic Research; http://www.tigr.org/tigr-scripts/e2k1/irgsp_orderedBAC.spl?db=osa1&chr=1).

The data obtained allowed us to infer the distribution of rice linkage group 1 on Lolium/Festuca chromosome 3 and the syntenic relationship between rice and Lolium/Festuca at the macro level. The linear order of BAC/PAC clones was virtually the same in Lolium/Festuca as in rice (Figure 1). We observed only two disruptions to macrocolinearity between the two species. BAC/PAC clones 1, 3, and 5 map to the second bin on the arm that carries a nucleolar organizer region (NOR) (between physical distances of 82.4 and 86.6%) instead of, as expected, to the end bin (between physical distances of 86.6 and 100%). BAC/PAC clone 6, however, was found to be physically located in the end bin, i.e., between 86.6 and 100%. The second disruption involved BAC/PAC clones 299, 303, and 305 that mapped to the bin on the non-NOR arm between physical distances of 19.7 and 28.9%. However, BAC/PAC clones 286 and 308 map in the neighboring bin between physical distances of 28.9 and 38.0% as expected. The first disruption to macrocolinearity of the rice BAC/PAC clones was therefore the smaller of the two consisting of a maximum of 5 BAC/PAC clones, respectively. The second disruption has the potential to be somewhat larger, consisting of a maximum of 21 BAC/PAC clones. However, introgression mapping of the rice BAC/PAC clones between 286 and 299 is required to assess the actual extent of this disruption to macrocolinearity.

The work described is not testing the microcolinearity between Lolium/Festuca and rice. Also, by the nature of the primer design we concentrated on generally conserved sequences. Therefore it was not possible to develop SNPs for grass and rice sequences that have diverged.

DISCUSSION

Recombination distribution and gene density:

We have previously shown that recombination frequency varies along the length of Lolium/Festuca chromosome 3 (King et al. 2002b). Two major peaks in the frequency of recombination are located at 12% of the distance along the chromosome from the telomere of the non-NOR arm and at 18% of the distance from the telomere of the NOR arm. In the non-NOR arm the distal 15.5% (which just covers the peak in recombination) contains 9% of the BAC/PAC clones from rice chromosome 1 (Figure 2). In the NOR arm the distal 20% (which again just covers the peak in recombination) also contains 9% of the BAC/PAC clones from rice chromosome 1. Therefore 82% of the BAC/PAC clones from rice chromosome 1 are physically located in the remaining 64.5% of the chromosome, much of which has a very low level of recombination (Figure 2). (The peaks in recombination frequency are actually fairly narrow with the most distal regions of both arms also having low recombination frequencies.) The estimate above is therefore an overestimate of the number of BAC/PAC clones within the regions of high recombination frequency. If the BAC/PAC clones in the end bin of the NOR arm and the end two bins of the non-NOR arm (Figure 2) are also considered as low recombination areas, then 90% of the BAC/PAC clones from rice linkage group 1 are located within regions of low recombination frequencies (which equates to 85% of the physical distance on Lolium/Festuca chromosome 3).

Figure 2.—

Introgression map of Lolium/Festuca chromosome 3 showing the percentage of distribution of BAC/PAC clones, functionally annotated gene models (referred to as genes in the figure) and transposons, retrotransposons, and hypothetical proteins from rice linkage group 1 as identified by The Institute of Genome Research. Shown on the right of the diagram are the distributions within the terminal bins covering the peaks in recombination frequency and the remainder of the chromosome with a much lower recombination frequency. On the left of the diagram are the distributions specifically within the most distal bins (before the peaks in recombination frequency) and the region covering the centromere and the NOR where recombination frequency is extremely low.

The distribution of BAC/PAC clones (together with the functionally annotated gene models on each clone as listed by The Institute of Genome Research) from rice allows an estimate of the distribution of gene density from rice linkage group 1 to be plotted on Lolium/Festuca chromosome 3 (Figure 2). The terminal 15.5% of the non-NOR arm (which includes the peak in recombination frequency for that arm) contains 12.6% of the functionally annotated gene models (calculated by counting all the loci minus transposons, retrotransposons, and hypothetical proteins, listed by The Institute of Genome Research for the BAC/PAC clones bin mapped within the terminal 15.5% of the non-NOR arm) found in rice linkage group 1 while the terminal 20% of the NOR arm (which again includes the peak in frequency of recombination for that arm) contains 11.8% of the genes. The remaining 64.5% of the chromosome therefore contains 75.6% of the functionally annotated gene models from rice linkage group 1. Thus, there is a fairly even distribution of genes across the physical distance of the chromosome with the greatest proportion of genes on rice linkage group 1 located in regions of low or very low recombination. (The number rises to 81.4% if the distal regions of the chromosome arms are also included as regions of low recombination frequency.) In rice, the density of expressed genes is greater on the distal portions of the chromosome arms compared with the regions around the centromeres (International Rice Genome Sequencing Project 2005). Therefore, the areas of low recombination do contain, as expected, a higher proportion of transposons, retrotransposons, and hypothetical proteins compared to those regions of higher recombination frequency; i.e., 6% of the transposons, retrotransposons, and hyopothetical proteins are located in the distal 15.5% of the non-NOR arm and 7% are located in the distal 20% of the NOR arm with 86% located in the remaining 64.5% of the chromosome.

We undertook a specific examination of the classes of genes (functionally annotated gene models as compared to transposons, retrotransposons, and hypothetical proteins) in the regions of the chromosome exhibiting very low frequencies of recombination, i.e., from 38 to 73% of the distance along the chromosome (Figure 1). This region covers 35% of the physical distance of Lolium/Festuca chromosome 3 and contains both the centromere and the NOR. Recombination in this region of the chromosome is extremely low throughout, almost touching zero in places (King et al. 2002b). Despite this low frequency of recombination, 50% of BAC/PAC clones from rice group 1 containing 36% of the functionally annotated gene models are physically located within this 35% region of Lolium/Festuca chromosome 3 (Figure 2).

Recombination and gene distribution in other monocots:

Localization of chiasma to the distal ends of the chromosomes is not specific to Lolium and Festuca. For example, Snape et al. (1985) demonstrated little or no crossing over between the centromere and the NOR in chromosome 1B of wheat. More recently wheat deletion mapping has demonstrated chiasma localization; i.e., it has been shown that there is a gradient in recombination increasing from the centromere to the telomere (Lukaszewski and Curtis 1993; Akhunov et al. 2003a). Erayman et al. (2004) showed that one-fourth of the wheat genome present around the centromere accounted for <1% of the total recombination. Chiasma localization has also been demonstrated in sorghum (Klein et al. 2003; Kim et al. 2005) and barley (Künzel et al. 2000). The average rate of recombination across the heterochromatic portion (concentrated in proximal regions) of sorghum was ∼34-fold lower than that of recombination in euchromatic regions (concentrated in more distal regions of the chromosomes). In barley chromosome 3H, a region spanning the centromere and covering ∼45% of the total physical length of the chromosome was reported to show no recombination (Künzel et al. 2000).

The gene loci distribution we describe differs from the general conclusions drawn in wheat over the last decade, where the large majority of genes have been described as being located toward the distal regions of the chromosome arms (leaving the pericentromeric/proximal regions of the chromosome largely gene poor). Erayman et al. (2004) located 94% of the genes in wheat to 48 gene-rich regions (29% of the wheat genome). No gene-rich regions were located in the centromeric regions of any of the wheat chromosomes and only one small gene-rich region containing 4% of the arm's genes was observed in the proximal 20% of any chromosome. A very similar pattern of gene distribution has been described for sorghum where higher gene densities were reported for the distal regions of chromosome 3 compared to the pericentromeric region (Klein et al. 2003; Kim et al. 2005). Likewise in barley, recombination hotspots making up 4.9% of the total barley genome were found to contain 47.3% of the 429 markers (most thought to represent genes) mapped to the physical map (Künzel et al. 2000). These regions of high recombination were located mainly in the distal regions of the chromosomes but also in some interstitial positions specific to each of the individual chromosome arms.

Although much of the published work has indicated the presence of distally located gene-rich regions, it has recently been speculated that there is actually a more even distribution of genes in the wheat genome (Devos et al. 2005) in agreement with results from the early stages of cereal comparative mapping. From this early work it was apparent that the gene content of whole rice chromosomes (particularly rice linkage groups 8, 10, 11, and 12) might map into pericentromeric/proximal regions in wheat and barley (Moore 1995; Moore et al. 1995; Van Deynze et al. 1995). As a rice chromosome is likely to contain thousands of genes, the proximal regions are also likely therefore to contain a similar number of genes. Studies exploiting barley deletions also inferred this physically (Korzun and Künzel 1996) by showing that RFLPs localized on the linkage maps of rice chromosomes 5 and 10 were allocated to cytologically defined regions of barley chromosome 5 (1H). The rice map of linkage group 5 was divided into two parts relating to the distal portions of both the short and the long arms of the barley chromosome. The map of rice chromosome 10, however, was related to an interstitial segment on the long arm of barley chromosome 5 (1H), a region highly suppressed in recombination activity. This distribution of genes is also substantiated by the National Science Foundation project in bin mapping wheat unigenes. First, for example, the consensus chromosome for wheat group 3 shows the most proximal bin in the long arm to have a higher number of mapped ESTs than some bins in more distal positions. In addition, in both long and short arms a central bin had very high numbers of mapped ESTs (Munkvold et al. 2004). Second, synteny between rice chromosome 1 and the wheat group 3 consensus chromosome would suggest that a large number of the BAC/PAC clones from rice linkage group 1 are actually located in the proximal regions on wheat group 3. Approximately BAC/PAC clone 90 (AP000815) to BAC/PAC clone 253 (AP003760) were located in the bins immediately surrounding the centromere; i.e., ∼163 BAC/PAC clones (42% of the total from rice linkage group 1) are located in these proximal regions (Munkvold et al. 2004). If the most proximal half of each chromosome arm is looked at then ∼250 BAC/PAC clones (64% of the total) from rice linkage group 1 are located within this region. It is possible that microsynteny may be greatly disrupted within these proximal regions but this would be in contradiction to the general evolutionary trends previously described (Akhunov et al. 2003b). These studies showed that new loci, as well as fixed deletions, are more frequent in high-recombination regions at the distal ends of the wheat chromosomes. As a result of these trends the distal regions of wheat chromosome arms have been evolving faster than the proximal regions, resulting in more frequent exceptions to the colinearity and microcolinearity between rice and wheat at the distal chromosome regions.

Likewise, alignment of the sorghum 3 physical map to rice chromosome 1 shows the location of 157 BAC/PAC clones from rice chromosome 1 occurring within the region from 62.0 to 105.2 cM (43.2 cM) that spans the sorghum centromere.

The results from the Lolium/Festuca introgression mapping strongly support these latter findings, i.e., large numbers of BAC/PAC clones and hence gene loci (albeit with a relatively high ratio of transposons, retrotransposons, and hypothetical proteins) mapping to regions of the chromosomes showing very low frequencies of recombination.

The more recent work described above suggests that physical mapping of ESTs in both wheat and sorghum may not be detecting proximally located genes that are orthologous to sequences in rice. Both the deletion maps of wheat and the in situ hybridization maps of sorghum are built on the mapping of wheat and sorghum ESTs followed by comparisons to rice made via their homology to rice genes. In contrast, in the introgression mapping work described here, orthologous grass markers have been developed directly from rice coding sequences. It is these orthologous markers that have then been bin mapped in Lolium/Festuca. The reasons for the absence of genes in the proximal regions of wheat and sorghum could therefore possibly result from (1) only partially representative EST libraries or (2) an inability to detect ESTs as a result of low sequence polymorphism (Sorrells et al. 2003 and Randhawa et al. 2004 reported between 24 and 45% of loci as monomorphic and therefore that could not be mapped to either nulli-tetrasomic or deletion lines).

Synteny:

Macrocolinearity between Lolium/Festuca chromosome 3 and rice chromosome 1 appears to be largely intact with only the two disruptions, neither of which is as extensive as the segmental inversion in the short arm of chromosome 3 of sorghum compared to rice as reported by Klein et al. (2003), which involved ∼63 of the rice linkage group 1 BAC/PAC clones.

Microcolinearity is probably less intact although to answer this properly would require the introgression mapping of several sequences from functionally annotated gene models from the same BAC/PAC clone. Disruption to microcolinearity would be consistent with work in wheat where some bins on the deletion maps have been shown to have a greater synteny to rice compared to others. However, even in the conserved bins, an average of 35% of the putative single-copy genes still mapped to different and syntenically unexpected rice linkage groups (Sorrells et al. 2003; La Rota and Sorrells 2004). Similar disruption to synteny at the individual gene level has also been described in other cereals such as maize (Bennetzen and Ramakrishna 2002).

Physical alignment of Lolium/Festuca chromosome 3 and wheat homeologous group 3 chromosomes:

It is possible to make initial comparisons between Lolium/Festuca and other monocot species through rice. Figure 3 shows the BAC/PAC clones from rice linkage group 1 that have been physically mapped on Lolium/Festuca chromosome 3 compared to their physical locations in wheat. To avoid possible complications with microsynteny only those BAC/PAC clones that have been mapped in both Lolium/Festuca and wheat using the same functionally annotated gene model in rice are shown. For example, BAC/PAC clone 371 (AP004332) has been physically mapped in both Lolium/Festuca and wheat, using sequence from Locus_OS01g70170 (transaldolase 2, putative, expressed).

Figure 3.—

Alignment of Lolium/Festuca chromosome 3 introgression map with wheat consensus group 3 deletion map via rice linkage group 1 BAC/PAC clones (the same functionally annotated gene model has been used to physically map the BAC/PAC clone in both Lolium/Festuca and wheat). Black lines are drawn between BAC/PAC clones mapping to a particular bin on both maps where synteny is maintained. Pink lines are drawn between BAC/PAC clones mapping to a particular bin on both maps where synteny is not maintained. Blue lines are drawn between BAC/PAC clones mapping to a particular bin on the Lolium/Festuca map but that while mapped to wheat group 3 chromosomes cannot be mapped to a particular bin on the wheat consensus map. Red lines are drawn for BAC/PAC clones mapped to two different positions on the wheat consensus group 3 chromosome. BAC/PAC clones 75 and 107 map to Lolium/Festuca chromosome 3 but map to a different wheat group as shown.

Many of the 21 gene loci used to compare the Lolium/Festuca chromosome 3 introgression map with the wheat group 3 consensus map can presently be mapped to a more specific bin in Lolium/Festuca. Despite the complicated overall picture there does appear to be a fairly good level of synteny between Lolium/Festuca 3 and wheat consensus group 3. This does not extend to all loci, however, and with only 21 putative gene loci mapped in common a great deal of care must be taken in making generalizations. Figure 3 does show the very clear benefits that accrue from pooling work from different species. For example, where loci have been mapped to a narrower region in one species, the physical map may be used to increase the resolution of the physical map in the other species.

The consequences of low recombination on allelic reassortment:

Our research strongly suggests that modern and past breeding strategies may have only skimmed the surface of the genetic variation available for the development of improved varieties; i.e., the majority of selections for new and novel allelic combinations have, by default, been centered on relatively few genes located in regions of high recombination near the ends of the chromosomes. Linked genes in areas of very low recombination will assort only very rarely, if at all. In addition, the probability of detecting a rare recombinant event in a region of low recombination frequency in breeding programs without prior knowledge and the use of markers flanking these target regions of the genome is very small. From these data we predict that, in grasses, for example, 47% of the sequences that are orthologous to rice linkage group 1 coding sequence will remain linked and thus will be resistant to selection in breeding programs and isolation via map-based cloning. The Lolium/Festuca introgression system will provide a means of exploiting these low recombination regions. Lolium/Festuca group 3 monosomic substitution is fully fertile and therefore it is possible to generate many thousands of BC₂ plants. These plants can then be screened for flanking markers either side of a region of interest to break the region into much smaller physical bins. With the rice BAC/PAC clones physically mapped it will then be possible to say within fairly narrow limits which region(s) of the rice genome is syntenic to this targeted region of Lolium/Festuca. This procedure has presently been exploited in grass and led to the isolation of a gene that confers delayed senescence and that underlies Mendel's I locus that controls color in pea (Armstead et al. 2006, 2007).

In addition, the presence of large numbers of coding sequences located in regions of the genomes of grass, wheat, and sorghum, etc., that undergo very low levels of recombination also has a significant evolutionary significance; i.e., specific allelic combinations will remain linked over many generations; e.g., it has been speculated that such regions of the genome may carry coadapted gene complexes, supergenes, or homozygous housekeeping genes.

The work we have described is enabling the transfer of information between rice and the large-genome monocot crop species and is being used to inform breeding and basic research programs. A genomewide high-resolution comparison of gene order and distribution in rice, Lolium, wheat, barley, etc., is presently being undertaken.