Six accessions belonging to A. mongolicum, A. cristatum, A. desertorum, A. fragile, and A. michnoi were used in this study (Table 1). Seeds of each accession were randomly selected for cytogenetic investigation.

The seeds were germinated on moist filter paper in Petri dishes at room temperature. Root tips of of 1–2 cm length were excised and pretreated with nitrous oxide at a pressure of 8 atm for 2 h at room temperature following the method of Kato (1999). Subsequently, the root tips were fixed in ethanol–glacial acetic acid (3:1, v/v) for at least 30 min at room temperature. Each root tip was squashed in a drop of 45% acetic acid.

All potential dinucleotide and trinucleotide SSRs and four satellite DNA sequences, namely pAs1 (Rayburn and Gill 1986), pSc119.2 (Bedbrook et al. 1980), 5S and 45S rDNA, were used to generate DNA probes. The procedure for labeling the above sequences was performed in accordance with the method of Dou et al. (2016).

The FISH experiments were conducted following the method of Dou et al. (2009; 2016). The images were captured with a Photometrics CoolSNAP CCD camera (Roper Scientific, Trenton, NJ, USA) using a fluorescence microscope (model DM R HC, Leica Microsystems, Wetzlar, Germany) and processed with the Meta imaging system (Universal Imaging Corporation, West Chester, PA, USA). Finally, the images were adjusted with Adobe Photoshop 6.0 for contrast and background optimization.

A chromosome number of 2n = 14 was detected in A. mongolicum (Fig. 1a–c) and A. cristatum (Fig. 1d–h), whereas a chromosome number of 2n = 28 was detected in most individuals of the other Agropyron species (Figs 1i–r, 2a–r). Both diploid and tetraploid species were detected. However, monosomic chromosomes were identified in many cases in tetraploid species (Figs 1o, r 2k, r). The repetitive sequence pAs1 showed multiple subtelomeric, intercalary, or pericentromeric hybridization sites on all chromosomes. The hybridization pattern of pAs1 was much more useful for distinguishing each chromosome of the P genome in all species compared with the other repetitive sequences. The tentative karyotypes of all individuals analyzed were determined using pAs1 as a land marker and the chromosome arm ratio and relative length as references (Fig. 3). Four pairs of metacentric chromosomes and three pairs of submetacentric chromosomes were identified in both A. cristatum and A. mongolicum. In addition, eight pairs of metacentric chromosomes and six pairs of submetacentric chromosomes were identified in the other tetraploid species. Although a high level of intra- and inter-species chromosomal polymorphisms were detected, homoeologous chromosomes among species were tentatively identified and designated 1-7 (1^{^}-7^{^}) (Fig. 3). In this designation, the numerals do not correspond with those of the homoeologous groups used in the nomenclature system of the chromosomes of wheat and barley. Seven homoeologous chromosomes were identified:

• Chromosome 1 is a metacentric chromosome with pAs1 hybridization signals in pericentromeric regions in most cases.

• Chromosome 2 is a metacentric chromosome lacking pAs1 hybridization signals in the subtelomeric region of the short arm in most cases.

• Chromosome 3 is a metacentric chromosome with pAs1 hybridization signals distributed from the intercalary to subtelomeric regions of both arms.

• Chromosome 4 is the smallest metacentric chromosome.

• Chromosome 5 is the largest submetacentric chromosome.

• Chromosome 6 is a submetacentric chromosome with the smallest arm ratio (short arm to long arm).

• Chromosome 7 is a submetacentric chromosome lacking pAs1 hybridization signals in the subtelomeric region of the short arm.

Figure 1.

FISH patterns of mitotic metaphase chromosomes detected using pAs1 probes (red) combined with 5S rDNA, 45S rDNA, and pSc119.2 probes in A. mongolicum (a–c); 5S rDNA, 45S rDNA, pSc119.2, AG, and AC in A. cristatum (2x) (d–h); 5S rDNA, 45S rDNA, pSc119.2, AAG, ACG, AGG, CAG, CAT, AG, and AC in A. cristatum (4x) (i–r). The pAs1 signals (red) were removed artificially in all images except in (a, d, i). Arrows indicate the target signals (green). Bar = 10μm.

Figure 2.

FISH patterns of mitotic metaphase chromosomes detected using pAs1 probes (red) combined with 5S rDNA, 45S rDNA, pSc119.2, AG, and AC probes in A. desertorum (a–e); 5S rDNA, 45S rDNA, pSc119.2, AAG, ACT, AG, and AC in A. fragile (f–l); 5S rDNA, 45S rDNA, pSc119.2, CAT, AC, and AG in A. michnoi (m–r). The pAs1 signals (red) were removed artificially in all pictures except in (a, f, m). Arrows indicate the target signals (green). Scale bar = 10 μm.

The hybridization sites of 5S rDNA and 45S rDNA were stably detected in all analyzed species. The 5S rDNA hybridization signals were detected in one pair of chromosomes in the diploid species A. mongolicum and A. cristatum (Fig. 1a, d) and in two pairs of chromosomes in the other tetraploid species (Figs 1i, 2a, f, m). The physical position of the 5S rDNA sites were the most conserved, which were stably detected in the intercalary regions of the short arms of homoeologous chromosome 6 (6 and 6^{^} chromosomes). Notably, additional 5S rDNA sites in the pericentromeric regions of the long arms of chromosome 6^{^} and additional 5S rDNA sites in the intercalary regions of the short arms of chromosome 6^{^} were detected in A. desertorum and A. fragile, respectively (Fig. 3).

Four highly conserved 45S rDNA sites were stably revealed in subtelomeric regions in both diploid species (Fig. 1b and e). However, variable numbers of 45S rDNA sites (from four to six) were identified in tetraploid species: six for A. cristatum (Fig. 1j), five for A. desertorum (Fig. 2b), four for A. fragile (Fig. 2j) and five for A. michnoi (Fig. 2n). Most of the 45S rDNA sites were physically mapped to the subtelomeric regions of homoeologous chromosomes 2, 5, 6, and 7 among the tetraploid species. Occasionally, 45S rDNA sites were detected on chromosome 1 and in the subtelomeric regions of the long arms of chromosomes 4 and 6 in A. desertorum (Fig. 3). The 45S rDNA sites on homoeologous chromosome 6 were the most conserved, which were detected on both chromosomes 6 and 6^{^} in all tetraploid species. In addition, the 45 S rDNA sites on homoeologous chromosome 7 were only detected on one set of the homoeologous chromosomes (7 or 7^{^}) (Fig. 3, Table 2).

A highly variable number of pSc119.2 hybridization sites was identified at either or both ends of the chromosomes in all diploid and tetraploid species, except A. mongolicum. The tetraploid species A. cristatum (Fig. 1k) and A. desertorum (Fig. 2c) harbored a high number of pSc119.2 hybridization sites (5–8; Table 2), whereas the other species carried as few as two sites. No conserved pSc119.2 sites were identified among the species.

Among the four potential dinucleotide SSR probes, (AC)₁₅, (AG)₁₅, (AT)₁₅, and (GC)_15, only (AC)₁₅ and (AG)₁₅ produced detectable hybridization signals, which were observed in subtelomeric regions, in all species analyzed, except A. mongolicum (Figs 1 and 2, Table 2). The number of hybridization signals for both (AC)₁₅ and (AG)₁₅ was highly among the deferent species. Agropyron desertorum (Fig. 2d) and A. fragile (Fig. 2l) carried a high number of (AC)₁₅ hybridization signals (8–10), whereas the other species carried a low number of (AC)₁₅ hybridization signals (1–3) (Table 2). Although the (AC)₁₅ hybridization signals were frequently detected in homoeologous 1 chromosomes in most species, the conserved distribution of the (AC)₁₅ signal among all investigated species was not observed. Agropyron fragile (Fig. 2k) and A. michnoi (Fig. 2r) carried a high number of (AG)₁₅ hybridization signals (6–8), whereas the others carried low number of (AG)₁₅ hybridization signals (3–4; Table 2).

Numbers followed by ” or ’ are the number of hybridization sites of the respective sequence. ” Indicates a pair of homologous chromosomes; ’ indicates one of the homologous chromosomes. Numbers in parentheses indicate the chromosome designation shown in Fig. 3

All ten potential trinucleotide SSRs probes, (AAG)₁₀, (AAC)₁₀, (AAT)₁₀, (ACG)₁₀, (ACT)₁₀, (AGG)₁₀, (CAC)₁₀, (CAG)₁₀, (CAT)₁₀, and (GGC)₁₀, were used to characterize the chromosomes of all species. No hybridization signals for any trinucleotide SSR probes were detected in the diploid species A. mongolicum and A. cristatum ‘Fairway’ or in the tetraploid species A. desertorum. A small number of SSR hybridization signals were identified in A. cristatum (4x), A. fragile, and A. michnoi, although these signals were not commonly shared among the species. Agropyron cristatum (4x) harbored the greatest number of trinucleotide SSR hybridization sites. One hybridization site of (AAG)₁₀, (ACG)₁₀, (AGG)₁₀, (CAG)₁₀ and (CAT)₁₀ was well identified in the individuals of A. cristatum (4x) (Fig. 1l–p). All (AAG)₁₀, (ACG)₁₀, (CAG)₁₀ and (CAT)₁₀ sites were physically mapped to the pericentromeric regions of the long arms of chromosome 2. This finding implies the co-localized distribution of (AAG)₁₀, (ACG)₁₀, (CAG)₁₀, and (CAT)₁₀ in A. cristatum (4x). However, (AGG)₁₀ was physically mapped to the pericentromeric region of the short arm of chromosome 5 (Table 2). Two trinucleotide SSRs (ACT)₁₀ and (AGG)₁₀, gave rise to hybridization signals in A. fragile (Fig. 2i and j). The SSR (ACT)₁₀, which was not detected in the other species analyzed, was localized in the intercalary regions of the short arms of chromosome 4 in A. fragile (Fig. 2j).The (AGG)₁₀ was hybridization site was physically mapped to subtelomeric regions in A. fragile (Fig. 2i) rather than to pericentromeric regions as in A. cristatum (4x) (Fig. 1n). Only one trinucleotide SSR, (CAT)₁₀, was detected in A. michnoi. The (CAT)₁₀ was hybridizatin site was localized to intercalary regions in A. michnoi (Fig. 2p) rather than to pericentromeric regions as in A. cristatum (4x) (Fig. 1p).

B chromosomes were identified in a few individuals of A. mongolicum (Fig. 1a) and A. desertorum (Fig. 2b) but not in the other Agropyron species analyzed. The B chromosomes in A. mongolicum and A. desertorum were intensely stained by the pAs1 probe. A distinct primary constriction was identified in the B chromosomes of both species. On the basis of the position of the primary constriction, the B chromosome in A. mongolicum was submetacentric, whereas that of A. desertorum was of metacentric type (Fig. 4).

Figure 3.

Molecular karyotypes of six Agropyron species probed using pAs1 (red) combined with 5S rDNA (green), 45S rDNA (green) or pSc119.2 (green) sequences. The numbers 1 to 7 designate seven different homologous chromosomes, whereas the same number followed by “^” indicates the respective homoeologous chromosomes in each genome of tetraploid species.

Figure 4.

FISH patterns of B chromosomes detected by pAs1 (red). a submetacentric B chromosomes in A. mongolicum b metacentric B chromosomes in A. desertorum.

In this study, all potential dinucleotide and trinucleotide SSRs and four satellite DNA repeats were used to characterize the P genome of five Agropyron species. Unlike other genomes such as the H, I, A, B, and D genomes in Triticeae (Cuadrado et al. 2007; 2008; Dou et al. 2016), which include a number of SSR hybridization signals based on FISH, the P genome in Agropyron harbored only the dinucleotide SSRs AG and AC in all species, except A. mongolicum, and fewer trinucleotide SSRs in the different species. The distribution of AAG corresponded to the distribution of N bands, which are composed of heterochromatin in Triticeae (Pedersen et al. 1996). In the P genome, only one AAG site was detected in the intercalary regions in one accession, A. cristatum (4x). The pAs1 in the P genome was detected more frequently in subtelomeric or pericentromeric regions than in intercalary regions, and the repeatedly detected 45S rDNA, pSc119.2, AC, and AG repeats were localized to subtelomeric regions. Forty-eight P-genome-specific sequences have been identified from the DOP-PCR product of the 6P chromosome of A. cristatum, and 14 of 48 sequences have been physically mapped to chromosomes, centromeres, pericentromeric regions, or subtelomeric regions throughout the P genome (Han et al. 2017). Taken together, the results of the present study suggest that the tandem-repetitive sequences in the P genome are much more intensively localized to the pericentromeric and subtelomeric regions rather than the intercalary regions.

The lack of tandem-repetitive sequences in intercalary regions makes the accurate discrimination of each chromosome of the P genome more difficult. In this study, 5S rDNA, 45S rDNA, and pAs1 repeats were more stable and informative. Four to five pairs or homoeologous chromosomes could be accurately identified in each diploid and tetraploid species based on the above-mentioned chromosomal markers, the chromosome length and the arm ratio.

The P genome of A. mongolicum, which includes no AG, AC or pSc119.2 repeats in the subtelomeric regions of any chromosome, is distinct from the P genome of the other Agropyron species analyzed. Structural rearrangements of some chromosomes were revealed between the two diploid species A. cristatum and A. mongolicum (Hsiao et al. 1986, 1989). In the present study, genome rearrangements were difficult to identify accurately, reflecting the low resolution of the chromosomal markers. However, discrepancies in the localization of the 45S rDNA sites were revealed on chromosomes 5 and 3 in A. mongolicum and A. cristatum, respectively. This observation strongly suggested that the chromosome structure arrangements involved these two chromosomes. The differentiation of both diploid species revealed in the present study suggests that the P genome of A. cristatum should be designated P^c, and the P genome of A. mongolicum should be designated P^m, consistent with the suggestions of Mellish et al. (2002). Agropyron mongolicum differs from other Agropyron species in producing a narrow, linear, glabrous spike (Dewey 1981). Data from AFLP markers discriminated A. mongolicum from the other Agropyron species (Mellish et al. 2002). Taken together, these results are strongly suggestive of a distant relationship between A. mongolicum and the other Agropyron species.

The tetraploid A. cristatum is considered to be an autopolyploid originating from the diploid species A. cristatum. However, the molecular karyotype of this tetraploid was not consistent with a doubled diploid karyotype. High variation in the number of hybridization sites and localization of 45S rDNA, AC, AG, and pSc119.2 repeats was revealed in the present study. Notably, a few trinucleotide SSRs were detected in tetraploid species rather than in diploid species. These observations indicate that high genomic variation, including DNA sequence deletion, amplification and genomic rearrangement, might have occurred in the transition from diploid to tetraploid species.

Agropyron desertorum is considered to be an allotetraploid between diploid A. mongolicum and A. cristatum (Asay et al. 1992, Mellish et al. 2002). The localization of a numer of 45S rDNA sites in A. desertorum was highly variable compared with the other species analyzed. Notably, one homoeologous set of chromosome 6 (6^{^}) contained additional 5S rDNA sites in the pericentromeric regions of the long arm. The homoeologous chromosome 6 stably carried 5S rDNA sites in the short arms of all Agropyron species studied. The distinct differences in the FISH patterns between chromosomes 6 and 6^{^} strongly support the allopolyploid origin of A. desertorum. However, the FISH pattern of chromosome 6^{^} in A. desertorum was not similar to either A. mongolicum or A. cristatum.

Agropyron fragile has been suggested to be an autotetraploid derivative of A. mongolicum (Asay et al. 1992). The chromosomes of A. fragile include a few AG, AC, and pSc119.2 hybridization sites. The genomic characters of A. fragile are different from those of A. mongolicum. The present results did not support the suggestion that A. fragile is an autotetraploid derivative of A. mongolicum. Moreover, additional 5S rDNA sites were revealed in the intercalary regions of the short arms of one homoeologous set of chromosome 6 (6^{^}). Similar to A. desertorum, the unique FISH pattern revealed in chromosome 6^{^} is supportive of a potential allopolyploid origin for A. fragile.

Agropyron michnoi exhibited a FISH pattern more similar to that of A. cristatum (4x). Specifically, the chromosomal co-linearity of the detected 45S rDNA sites was well retained between the two species, suggesting that A. michnoi might be more closely related to A. cristatum (4x) than to other Agropyron species studied.

The repeats, such as AC, AG, and pSc119.2, that are localized in subtelomeric regions, showed a highly varying number of hybridizations among species with the P genome and within populations, particularly in tetraploid species. The presence or absence of hybridization signals might reflect the deletion or duplication of the repetitive sequences. Unequal crossing over is a type of gene duplication or deletion event (Graur et al. 2000). Highly variable repeats of the above-mentioned sequences in tetraploid species, which include two sets of the basic P genome, suggests that variability of repeats may be strongly driven by unequal crossing over during the interfered meiosis process in tetraploid species with two similar genomes. The small number of trinucleotide SSRs hybridization sites and aneuploids frequently detected in tetraploid species might further confirm interference of meiosis in tetraploid species.

Specific FISH patterns of trinucleotide SSRs were detected in three different Agropyron species. Extensive intra- or inter-population genetic variation has been detected in Agropyron species (Mellish et al. 2002, Che et al. 2011, Chen et al. 2013). However, limited materials of each species were analyzed in the present study. Whether the FISH patterns of trinucleotide SSRs also exhibit high intraspecific variation needs further investigation.