Molecular cytogenetic (FISH) and genome analysis of diploid wheatgrasses and their phylogenetic relationship

This paper reports detailed FISH-based karyotypes for three diploid wheatgrass species Agropyron cristatum (L.) Beauv., Thinopyrum bessarabicum (Savul.&Rayss) A. Löve, Pseudoroegneria spicata (Pursh) A. Löve, the supposed ancestors of hexaploid Thinopyrum intermedium (Host) Barkworth & D.R.Dewey, compiled using DNA repeats and comparative genome analysis based on COS markers. Fluorescence in situ hybridization (FISH) with repetitive DNA probes proved suitable for the identification of individual chromosomes in the diploid JJ, StSt and PP genomes. Of the seven microsatellite markers tested only the (GAA)n trinucleotide sequence was appropriate for use as a single chromosome marker for the P. spicata AS chromosome. Based on COS marker analysis, the phylogenetic relationship between diploid wheatgrasses and the hexaploid bread wheat genomes was established. These findings confirmed that the J and E genomes are in neighbouring clusters.


Introduction
The tribe Triticeae contains nearly 100 annual species including agronomically important domesticated crops such as wheat, which is globally the third most-produced cereal after maize and rice. Approximately 400 perennial grasses with diverse genetic architecture also belong to the Triticeae. Five major genomic types (St, P, Ns, J and E) can be discriminated in the diploid and polyploid species. The genus Agropyron (crested wheatgrass) includes 10-15 species, of which the diploids, tetraploids and hexaploids are all based on the P genome. The genus Pseudoroegneria (bluebunch wheatgrass) was built around the common genome St, which is one of the most important basic genomes of perennial Triticeae species. Within the genus Thinopyrum (tall and intermediate wheatgrass), about 20 diploid, allotetraploid, allohexaploid, octoploid and decaploid species possess the J, E and sometimes St genomes [1].
As a tertiary gene pool for hexaploid wheat, perennial diploid wheatgrass species are important genetic resources and attracted researchers' attention many decades ago because of their favourable and mostly untapped genetic background. Perennial grasses are highly tolerant PLOS ONE | DOI: 10 same, neighbouring or distantly separated clusters, and how these genomes are related to the St and P genomes and to the A/B/D genomes of wheat. One disadvantage of applying sequence-based markers for a single gene family is that only the variability within a small part of the genome is considered when comparing different species [37]. Genes conserved throughout evolution in terms of both sequence and copy number make it possible for large complex genomes to be compared at multiple loci. Conserved orthologous set (COS) markers were identified by the in silico comparison of rice, wheat and Brachypodium ESTs [38]. As they target to the exon-intron boundaries of genes conserved between the model and target species, COS markers are highly polymorphic. Orthologous regions on the chromosomes, represented by these markers, enable the genomes of rice, wheat, maize, sorghum, barley and Aegilops to be compared [27,39,40].
A set of COS markers covering the homoeologous group 1-7 chromosomes of wheat, combined with in situ hybridization using various repetitive probes provide a powerful set of tools for clarifying complex genomic relationships such as that between wheat and perennial grasses.
Detailed FISH-based karyotypes of each diploid wheatgrass species would make the rapid cytogenetic analysis of genetic materials possible. It is very difficult to follow gene transfer in different genetic lines carrying wheatgrass chromosomes or chromosomal segments in a wheat background because of the lack of molecular and cytogenetic markers. The aim of the present study was to set up detailed FISH karyotypes for three diploid wheatgrass species and to determine individual chromosomal markers. By comparing FISH and molecular marker data, a closer phylogenetic relationship between these genomes was highlighted.

McFISH analysis
Seed germination and root tip mitotic chromosome preparations were carried out according to Linc et al. [20].
Multicolor fluorescence in situ hybridization (mcFISH) was performed using standard repetitive DNA sequences, a telomere specific repeat (HT100.3) and seven microsatellite motifs, as listed in Table 2.
The probe pTa71(45S rDNA) contains a 9.05 kbp fragment, which is part of an rDNA repetitive unit consisting of one copy each of 18S rDNA, 5.8S rDNA and 25S rDNA, and an intergenic spacer from wheat cv. Chinese Spring [41]. The probe pAs1 contains a 1-kb DNA fragment isolated from Aegilops tauschii Coss. in the plasmid pUC8 [19]. The Afa-family repeats [42] were amplified from the genomic DNA of barley (Hordeum vulgare L). The DNA repetitive clone pSc119.2, inserted into the plasmid pBR322, contains a 120 bp repeat derived from an EcoRI relic DNA from rye cv. King II [43]. The HT100.3 telomere repeat (TTTAGGG)n sequences were originally isolated and amplified from Arabidopsis thaliana L. [44]. The Fat probe was labelled with Fluorescein by PCR amplification from the 3B_050_N05 BAC clone using primers GGGGAGCTTCTCACAACAAGCand TATTTACCACGGCATGTCGGG resulting in an approximately 460-bp fragment [45]. Microsatellite probes were amplified from wheat genomic DNA according to Vrana et al [46]. For 3-color FISH, the pSc119.2 and Afa-family DNA sequences were amplified and labelled by PCR either with biotin-11-dUTP (Roche) or with biotin-14-dATP (Invitrogen) and  In situ hybridization was carried out on mitotic chromosome spreads of each Agropyron species according to Linc et al. [23], with minor modifications. The hybridization solution per slide (20 μl) contained 50% formamide, 2×SSC, 10% dextran sulfate, 0.1% sodium dodecyl sulfate, 50 ng/μl carrier DNA, and a combination of two or three~300 pg/μl fluorochromelabeled probes. The slides were treated for 2 min at 80˚C followed by hybridization overnight at 37˚C. After documenting the hybridization patterns of the microsatellite probes, the slides were rinsed off (3×6 min in 50% formamide and 2×SSC at 42˚C) and re-hybridized with a combination of two different repetitive DNA sequences under the same conditions as described above. The chromosomes were counterstained with 2μg/ml DAPI (4'-6-diamino-2-phenylindole) and mounted in antifade solution (Vectashield, Vector Laboratories).
Fluorescent signals were visualized using a Zeiss AxioImager M2 epifluorescence microscope equipped with filter sets for detecting DAPI, FITC and Rhodamine signals. Images were captured with a Zeiss AxioCam MRm CCD camera and processed with Zeiss AxioVision 4.8.2. software. All the images presented in this paper were manipulated only to enhance contrast.
The chromosomes of diploid wheatgrass species were classified according to similarity of their FISH patterns and designated A-G with a genomic symbol given as superscript.
A total of 117 markers, specific for wheat homoeologous groups 1-7, representing both chromosome arms of each analyzed accessions were chosen from a publicly available COS  marker collection published by Quraishi et al. [38] and summarised in S1 Table. PCR reactions with primers for COS markers were performed in a 12 μL reaction volume as described by Molnár et al. [47] using the touchdown reaction profile: 94˚C (2 min), 10 cycles of (94˚C (0.5 min), Ta +5˚C (0.5 min) decreasing in 0.5˚C increments for every subsequent set of cycles, 72˚C (1 min), 30 cycles of (94˚C, 0.5 min), Ta˚C (0.5 min), 72˚C (1 min), hold at 72˚C (2 min). PCR amplicons have been separated by a Fragment Analyzer Automated CE System equipped with a 96-Capillary Array Cartridge (effective length 33 cm) (Advanced Analytical Technologies, Ames, USA) and analyzed using PROsize v2.0 software. The annealing temperature (Ta) for each COS marker, together with data for the PCR amplicons, are also included in S1 Table. Phylogenetic analysis The banding patterns of the PCR products were scored as present (1) or absent (0) for each marker. A dendrogram was constructed using the Cluster 3.0 program and visualized with Java TreeView 1.1.6r4. Genetic similarities between the 7 accessions/genotypes were measured using Jaccard's similarity coefficient based on the proportion of shared alleles. The Jaccard's similarity coefficients were calculated as follows: J(i 1 ,i 2 ) = a/(a+b+c) [48], where a is the number of bands shared by both individual samples (i 1, i 2 ), b is the number of bands where i 1 has a band, but i 2 does not, c is the number of bands where i 2 has a band, but i 1 does not.

FISH-based karyotype analysis
In the present study three diploid wheatgrass species were used for FISH-based karyotype analysis. Two populations (accessions) of the species A. cristatum and 3 populations each of P. spicata and T. bessarabicum with different geographical origin were examined (Table 1).
Detailed FISH karyotypes of the diploid species chromosomes were generated using repetitive DNA sequences after validating the chromosome morphology and arm ratios of at least 30 metaphase cells from each accession (Fig 2). Chromosome idiograms were also constructed after detailed studies of chromosome morphology (Fig 3). The simultaneous hybridization of the repetitive DNA probes Afa family (pAs1 for accession 362480); pSc119.2 (Fat probe for accession 362480) and pTa71 on mitotic metaphase cells made it possible to characterize single chromosomes of the three diploid wheatgrass species. Certain variability in the number and position of FISH signals was observed for some homologous chromosomes in different plants of each of the diploid genotypes. However, these polymorphisms did not hamper the precise characterization of all the individual chromosomes; the labelling patterns were species-and chromosome-specific.

Agropyron cristatum
All seven P genome chromosomes of both accessions carried specific pSc119.2 and Afa signals, so they could all be distinguished from each other. However, this species had the least characteristic chromosomes based on their FISH patterns, and many chromosome polymorphisms affecting all seven chromosome pairs were detected between the two accessions. Two of the seven chromosome pairs possessed NOR regions. The E P chromosome had a clearly distinguishable satellite in both accessions, while the satellite on A P was much narrower and was only visible in MvGB 1521 genotype. The Afa signals on E P appeared mainly in terminal and subterminal positions in both accessions. Instead of showing a pTa71 signal on A P S, MvGB1509 had a characteristic pSc119.2 site at the chromosome tip. Chromosome B P was found to be (sub)metacentric and carried Afa-family sites on both arm, mainly terminally (B P S) and subterminally (B P S). There was a characteristic Afa signal on B P S in the pericentromere region. C P , a large metacentric chromosome, had mainly Afa-family signal on both arms in subterminal and terminal positions. There was a faint pSc119.2 signal at the tip of C P S, but  only in the MvGB1521accession. The submetacentric chromosome D P carried Afa signals in both accessions, in subterminal-terminal positions. A strong pSc119.2 signal was located at the tip of D P S in accession MvGB1521. The submetacentric chromosome F P possessed Afa-family repeats on both arms, in terminal-subterminal positions. The G P submetacentric chromosome showed strong FISH-pattern polymorphism between the two chromosomes of the accessions analyzed. In MvGB1509, G P S carried a faint Afa-family signal at the pericentromere and a pSc119.2 signal at the tip of the chromosome arm.

Thinopyrum bessarabicum
All the chromosomes could be distinguished according to their FISH patterns. The number and position of signals was similar for all three accessions, so this species had the least chromosomal polymorphism among the three diploid genomes investigated. In contrast to the other genome types, the J-genome chromosomes carried either one or, more frequently, two distinct pSc119.2 signals in the telomeric regions of their short and/or long arms (Fig 2). The A J metacentric chromosome possessed a clear terminal pSc119.2 site on the long arm and diffused subterminal Afa signals on both arms. A distinct intercalary Afa signal in the proximal third of the long arm was characteristic of chromosome A J , which could be detected in all the accessions analyzed. The B J submetacentric chromosome showed a strong pSc119.2 signal on the short arm and faint, diffused subtelomeric Afa-family signals on both arms. Large terminal pSc119.2 signals were visible on both arms of the C J (sub)metacentric chromosome, and a strong subterminal Afa cluster was detected on the long arm. Terminal pSc119.2 signals were detected on both arms of the metacentric D J chromosome, which also carried a distinct subterminal Afa cluster on the short arm. Chromosome E J had a small satellite, which only showed a terminal pSc119.2 site in one accession. A large Afa cluster was observed in the perinucleolar region on the short arm and fuzzy signals were detected in a subterminal position on the long arm, with intensities varying among the accessions. Very weak, diffused signals of the Afa repeat were dispersed over the distal half of the E J long arm. The F J chromosome was characterized by its large satellite. Both the satellite and the long arm were terminated by strong pSc119.2 signals. A large, bright Afa cluster was located in a perinucleolar position and much weaker, diffused signals of this probe were found in the distal parts of both arms. Both satellite chromosomes, E J and F J , carried a 45S rDNA locus on their short arms (more fainted signals on MvGB1705 chromosomes). The G J submetacentric chromosome showed strong telomere pSc119.2 signals on both arms, and relatively faint, diffused Afa signals in the subterminal regions. The signals on the long arm were a little more intense than those on the short arm.

Pseudoroegneria spicata
The A S chromosome was submetacentric with a narrow satellite on the short arm. Strong, fuzzy Afa signals were found in the subterminal region of the long arm, extending toward the centromere, up to the middle part of the arm, though the intensity of these was much weaker. The Afa-family signal on the short arm was weaker and the pSc119.2 probe was totally absent from the satellite arm. Interestingly enough, only one of the two homologous A S chromosomes carried specific pSc119.2 signal in a subterminal position on the long arm. The B S chromosome had a secondary constriction, which failed to show a pTa71 signal, but possessed a strong DAPI-positive band. Large DAPI bands were also found at both termini of the chromosome. Bright, diffused Afa-family signals were observed on both arms, adjacent to the DAPI bands. A distinct DAPI band was present terminally on the short arm of the C S chromosome; diffused Afa-family signals were observed in the subterminal position on the short arm, adjacent to the DAPI band, and in the terminal position on the long arm. D S is a submetacentric chromosome with two DAPI-positive bands, a larger one on the short arm and a smaller one on the long arm). Afa-family signals were found subterminally on both arms but the signal intensity was greater on the long arm. E S had a secondary constriction in the proximal third of the short arm, which coincided with the location of the pTa71 signal. This chromosome showed the most complex FISH pattern in all the accessions investigated. Both arms carried clear terminal DAPI-positive bands. A strong pSc119.2 signal was found terminally on the satellite and a strong, fuzzy Afa-family signal on the distal half. This labelling pattern was detected on the E S chromosome of the accessions MvGB1615 and PI 362480, but the large proximal Afa cluster was absent from one of the homologous chromosomes of the MvGB1607 accession. A clearcut Afa cluster was present in the very proximal part of the long arm arm rather faint, fuzzy signals of this probe were found in a subterminal position, adjacent to the DAPI band. Two distinct DAPI bands were also detected in a subterminal position on the short and long arms of the F S and G S chromosomes. Bright, diffused Afa-family signals with approximately the same intensity were detected on both arms of F S , close to the DAPI bands. The G S chromosome showed a strong pSc119.2 signals on both arms (only on G S L of MvGB1607) but the intensity of both subterminal Afa-family signal was very low.
In addition to the Afa-family, the line PI 362480 was also subjected to FISH analysis with the Fat probe. Its Afa-karyotype was in good agreement with the chromosome patterns of the two other accessions. One chromosome pair was characterized by an abundance of Fat signals in the pericentromeric region, which is typical of group 4 chromosomes [45]. Based on the Afa pattern and the position of the large DAPI band, this chromosome corresponded to B S (current classification).

Microsatellite repeat analysis
Two-step fluorescence in situ hybridization was used to identify the hybridization patterns of seven SSR motifs on the chromosomes of three wheatgrass species. The first hybridization was carried out with the microsatellite probes (GAA)n, (ACT)n, (CAG)n, (AGG)n, (CAC)n, (AAC)n and (ACG)n. After the documentation of the FISH sites (if necessary), the slides were re-hybridized with the standard repetitive DNA probes pSc119.2, pTa71 and Afa-family in order to characterize individual chromosomes with the specific microsatellite location. Among the seven microsatellite repeats studied, only (GAA)n produced a distinct FISH signal on the A S short arm of the two accessions P. spicata (MvGB1607 and MvGB1605, not showed). Based on these results, the (GAA)n SSR motif can be used as an individual A S chromosome-specific cytogenetic marker. None of the other microsatellite repeats were suitable for use as molecular cytogenetic markers for the characterization of the J, St and P genomes.

COS marker analysis
COS markers specific for wheat homoeologous groups 1-7 were used for PCR analysis of the total genomic DNA of the perennial species in order to provide tools for detecting the genomes E, J, St and P in the wheat genetic background and to establish their genetic similarity relative to wheat. Among the 117 COS markers studied, 102 showed PCR products in the wheat control genotype (GK Ö thalom) or in at least one of the 6 wheatgrass lines, while 15 markers amplified no product. The 102 markers (wheat chromosome group 1: 12, group 2: 11, group 3: 18, group 4: 10, group 5: 12, group 6: 19, group 7: 20) resulted in 396 bands of various sizes (S1 Table).
In the case of P. spicata and A. cristatum two accessions of each species were investigated and only amplicons exhibiting similar size within the species but size polymorphism relative to wheat were considered to be suitable for pre-breeding purposes. In P. spicata genotypes MvGB1607 and MvGB1615, 123 and 140 loci, respectively, were assigned to the St genome, of which 62 loci (50 markers) showed significant size polymorphism. Finally, 42 loci (36 markers) were polymorphic among the 126 and 134 located on A. cristatum genotypes MvGB1521 and MvGB1509 (P-genome), respectively.
Genome-specific loci of the COS markers with significant (!5bp) length polymorphism between wheat genotype GK Ö thalom and the perennial grasses were considered to be suitable for the marker-assisted selection of new wheat-alien introgression lines in pre-breeding programs. In this study, 207 polymorphic loci of 76 markers were found to be potentially useful for the detection of the E, J, St or P genomes of perennial grasses.

Genetic similarity between perennial grasses and wheat
For phylogenetic analysis, the PCR amplicons were scored as present (1) or absent (0) for each marker and were used as character states. Amplicons of identical size were considered to be the same. Only bands shared between at least two genotypes were considered in the calculation of Jaccard's similarity coefficient. A total of 206 loci of 97 COS markers were scored in a global matrix of which 188 (91%) were polymorphic between the species. The genetic similarities ranged from 0.26 to 0.74. Differences in the degree of similarity between species were confirmed by statistical analysis. The dendrogram generated from the analysis of similarity (Fig 5) separated most of the populations and all the species, which split into three groups. The first group included the E-genome species T. elongatum, the J-genome species T. bessarabicum and T. aestivum. Both genotypes of the P-genome species A. cristatum fell in the second group, while the two genotypes of the St-genome species P. spicata formed a third, distantly separated clade.

Discussion
Perennial Triticeae species are important as tertiary gene pools for wheat breeding programs. However, detailed knowledge of the genetic and cytogenetic structure of these species is required for the targeted exploitation of their variability. The chromosomal positions of 45S and 5S rDNAs site have been determined for many perennial Triticeae species [1,49,50] and a detailed mcFISH karyotype has been constructed for the diploid E genome [23], but to date, no such FISH polymorphism experiments have been performed on other diploid Thinopyrum genomes. Far fewer species-specific cytogenetic and molecular markers are available than are needed for controlled, successful gene transfer between wheat and species belonging to the tertiary gene-pool of wheat.
The relationships between diploid perennials have always been an important subject of debate among scientists (taxonomists, cytogeneticists, etc.). Hsiao et al. [24] studied the karyotypes of 22 diploid perennial Triticeae species representing the P, St, J (E), H, I, Ns W and R genomes. Based on these earlier investigations, two satellite chromosomes of the J and E genomes were assigned to homologous groups 4 and 5, while Dvorak et al. [49] suggested that SAT chromosomes in the Triticeae might belong to groups 1, 5 and 6. Indeed, by using RFLP analysis of non-transcribed spacer of rRNA genes they assigned two satellite chromosomes of Elytrigia elongata (= Thinopyrum elongatum) to homoeologous groups 5 and 6, which was later confirmed by mcFISH of a set of ditelosomic and addition lines [23]. T. bessarabicum and T. elongatum genetic similarity based on previous studies is partially proved [36], therefore the satellite chromosomes of this species should also be placed to groups 5 and 6.
Position of rDNA signals on A. cristatum chromosomes however was different from what we have seen in T. bessarabicum: the signals of pTa71 probe were located in terminal regions of chromosome pairs A P and E P , the latter one been classified as 5P by Said et al. [51]. These authors also showed that the second satellite chromosome classified here as A P did not belong to group 1 and its relationship with wheat homoeologous chromosomes remained unknown. This chromosome did not possess 5S rDNA locus and could belong to group 6, although terminal position of rDNA locus was not typical for this group. Alternatively, the satellite chromosome A P of A. cristatum might has derived from group 6 chromosome as a result of translocation of transposition of rDNA locus.
The C-banding patterns of 10 diploid species were studied earlier by Endo and Gill [52], who drew the attention to the equivalence/similarity of the J and E genomes already 30 years ago. These two genomes cannot be distinguished on the basis of chloroplast sequence data, but chromosome pairing analysis in meiosis, karyotype differences, and data on 5S rDNA spacer and ITS sequences provide clear evidence that these two species belong to different genera [28,37]. The genetic similarity between the E and J genomes still represents one of the biggest controversies in perennial Triticeae studies. Thus, according to the review of Wang and Lu [36], approx. half of the publications based on different assays put the J and E genomes in the same cluster, while the other half in adjacent or even in distantly separated clusters.
In the present FISH karyotype analysis, the single chromosomal FISH pattern of T. bessarabicum showed some polymorphisms between the accessions, but in most cases they exhibited a consistent picture. The current FISH-based picture of the J genome is obviously quite different from that of the E genome presented in a previous paper [23]. The most striking difference was found for the E J and F J chromosomes. While the 5E chromosome of T. elongatum had a distinct NOR and 6E failed to exhibit the 45S rDNA site, both the satellited chromosomes of T. bessarabicum (E J and F J , exhibited clearly distinguishable 45S rDNA FISH signals on their short arms. This partially corresponds to the rDNA distribution in the Triticeae species [53]. Small 45S rDNA signals (NORs) are located in the subtelomeric regions of chromosomes A P and E P of A. cristatum and A S of P. spicata. The satellites on A. cristatum chromosomes are very small, so they can hardly be expected to possess distinct secondary constrictions without special pre-treatment, similarly to the chromosomes of diploid wheat species, which rarely express distinct satellites [54]. The present investigation may explain earlier findings, which failed to reveal satellite chromosomes in the A. cristatum karyotype [51,[54][55][56]. Li and Zhang [57] concluded that rDNA sites were unequal on homologous chromosomes of the J, St and E genomes, which partially correspond with the present data for the rDNA site on the P genome, where heteromorphism was observed on the A P chromosome.
The FISH analysis of P. spicata using repetitive DNA revealed highly specific labelling patterns for all the S genome chromosomes. The signal of the 45S rDNA probe on the A S chromosome was quite small and located in the subtelomeric region of the short arm in all the accessions. Murphy et al. [58] studied three accessions of P. spicata, and only in one did the A S chromosome contain a satellite. The other accessions only showed a visible secondary constriction on chromosome E S , in agreement with the present results.
The polymorphism of repetitive sequences due to ancient hybridization followed by the incomplete sorting of ancestral polymorphisms (in addition to mutations, chromosome recombination events, etc.) could be responsible for intraspecific variation in diploid species [35]. The present results showed similar intraspecific variation in all three diploid wheatgrass genomes studied. The greatest intraspecific FISH polymorphism was detected on the P genome chromosomes (A P , D P , G P ) and the lowest on the J genome chromosomes (F J ). When FISH polymorphisms between J-and E-genome chromosomes assigned to two neighbouring clusters were compared, the validation of the karyotypes of four T. elongatum (E) accessions with different geographical origin showed much more extensive variations in the probe hybridization patterns (2E, 3E, 4E, 5E, 6E, 7E) [23] than those detected on the J genome chromosomes in the present study.
With one exception, all the microsatellite repeats tested failed to show any hybridization signals on the St-, J-and P-genome chromosomes. Apart from a few unidentifiable FISH signals, no chromosome arm-specific, characteristic signals were detected. The lack of these trinucleotide sequences is not unusual for the Triticeae species. An earlier study on wheatgrass genomes suggested that reliable chromosome identification could not be achieved using the GAA sequences [59]. However, the distribution of GAA and ACG clusters can be used as additional cytogenetic markers to characterize tetra-and diploid Aegilops chromatin in the wheat genetic background [22]. Moreover, with the use of GAA and CAG motifs, the 6A m chromosome of T. monococcum can be clearly discriminated from other A genome chromosomes [60,61]. However, it has been proved that the number of hybridization sites may vary within a species (between individual accessions). It is probable that GAA sequences could only be useful as a molecular cytogenetic marker for perennial Triticeae species if a sufficient number of accessions per individual genome were tested. It was proved earlier [36,62] that the J/E and St genomes of perennial Triticeae species are evolutionary close to the ABD genomes of common wheat, and similar conclusions were drawn in the present FISH-based experiments and molecular marker analysis. This genome-based similarity explains the relative ease with which gene can be transferred from these genomes to the hexaploid wheat background [10][11][12]. This relationship could serve as a strong scientific background for successful gene transfers from wild Triticeae relatives carrying the J/E and St genomes for wheat improvement.
The results obtained on diploid wheatgrass species agreed well with previous studies and confirmed the high transferability of COS markers between species in the Triticeae [27,39,40,63,64]. The fact that 31-50% of the amplicons obtained on perennial wheatgrass species belonging to the tertiary gene pool were polymorphic relative to those obtained in hexaploid wheat suggested that a significant part of the genetic diversity in these wild species is due to the variability of intron regions [65]. Interestingly, using the same set of COS markers a similar ratio of polymorphic loci (46-53%) was obtained in diploid Aegilops species with the U, M, S or C genomes, which are more closely related to wheat [64].
In the present study, UPGMA cluster analysis based on 206 loci of 97 conserved orthologous nuclear genes showed that the E genome of T. elongatum and the J genome of T.
bessarabicum are more closely related to hexaploid wheat than the P genome of A. cristatum, while the St genome of P. spicata is the most distantly related genome. These results are consistent with those of Mahelka et al. [35], who found the same genetic relationships between these species based on one nuclear gene (GBSSI) sequence. More recently, Hu et al. [66] developed 258 EST and 46 PLUG (PCR-based Landmark Unique Gene) primer pairs and found 43 markers suitable for distinguishing T. elongatum chromatin from the wheat genomes. The comparison of molecular marker locations on the chromosomes indicated that the E genome is closely related to the D genome of wheat. After assigning COS markers that are polymorphic between the chromosomes of wheat and the perennial grasses they will also be suitable for identifying the chromatin of grass species introgressed into wheat, and for marker-assisted selection to facilitate the transfer of useful agronomic traits.

Conclusions
The comprehensive analysis of the genetic relationships between diploid wheatgrass species at the chromosomal-and DNA sequence level will facilitate the understanding of their evolutionary role within the Triticeae tribe. COS marker-based phylogenetic studies confirmed close genetic relationship between J and E genomes, while FISH using repetitive DNA probes proved to be suitable for characterization of individual J-, P-, and St genome chromosomes. Based on these detailed FISH-based molecular cytogenetic analysis and the selection of representative, polymorphic COS-markers of three diploid species will support the pre-breeding programmes of wheat aimed to utilize the genetic diversity of perennial wheatgrasses.
Supporting information S1 Table.