Chromosomal Characterization of the Three Subgenomes in the Polyploids of Hordeum murinum L.: New Insight into the Evolution of This Complex

Hordeum murinum L. is a species complex composed of related taxa, including the subspecies glaucum, murinum and leporinum. However, the phylogenetic relationships between the different taxa and their cytotypes, and the origin of the polyploid forms, remain points of controversy. The present work reports a comparative karyotype analysis of seven accessions of the H. murinum complex representing all subspecies and cytotypes. The karyotypes were determined by examining the distribution of the repetitive Triticeae DNA sequences pTa71, pTa794, pSc119.2, pAs1 and pHch950, the simple sequence repeats (SSRs) (AG)10, (AAC)5, (AAG)5, (ACT)5, (ATC)5, and (CCCTAAA)3 via in situ hybridization. The chromosomes of the three subgenomes involved in the polyploids were identified. All tetraploids of all subspecies shared the same two subgenomes (thus suggesting them to in fact belong to the same taxon), the result of hybridization between two diploid ancestors. One of the subgenomes present in all tetraploids of all subspecies was found to be very similar (though not identical) to the chromosome complement of the diploid glaucum. The hexaploid form of leporinum came about through a cross between a tetraploid and a third diploid form. Exclusively bivalent associations among homologous chromosomes were observed when analyzing pollen mother cells of tetraploid taxa. In conclusion, the present results identify all the individual chromosomes within the H. murinum complex, reveal its genome structure and phylogeny, and explain the appearance of the different cytotypes. Three cryptic species are proposed according to ploidy level that may deserve full taxonomic recognition.

For many years it remained unclear whether the polyploid forms of H. murinum were auto-or allopolyploids [11][12][13][14][15][16]. Recently, however, allopolyploidy has been strongly supported by molecular phylogenetic analyses that differentiate 2x, 4x and 6x forms [17][18][19]. Indeed Jakob and Blattner [20] indicate that glaucum was involved in the formation of the tetraploids together with a now likely extinct taxon belonging to the same Hordeum Xu genome group, and that a third, closely related taxon contributed to the formation of the hexaploid leporinum. Nevertheless, additional cytogenetic analyses are needed to reveal the true genomic constitution of the different H. murinum taxa and cytotypes.
The aim of the present work was to examine the karyotypes of a representative sample of H. murimum accessions covering all subspecies and cytotypes. If the diploid glaucum was involved in the origin of the polyploid cytotypes, its chromosomes should be present in them. The same is true if tetraploid forms were involved in the origin of the hexaploid forms.

Materials and Methods
Material representing all three subspecies and cytotypes of the H. murinum complex was obtained from the IPK Germplasm Bank (Gatersleben, Germany). Table 1 provides information on the accession numbers and places of origin of the material used.

Chromosome preparation
Root tips were obtained from seedlings and exceptionally from plants grown in pots in a greenhouse. Meiotic divisions were observed in pollen mother cells. Chromosome preparations were obtained as previously described [21].

Fluorescence microscopy and imaging
Slides were examined using a Zeiss Axiophot epifluorescence microscope. Biotin/Cy3, digoxigenin/FITC and DAPI stained images were recorded with each filter set using a cooled CCD camera (Nikon DS). The localization of the signals relative to the DAPI staining pattern was resolved by merging images using Adobe Photoshop, employing only those functions that applied equally to all pixels.

Ploidy level of H. murinum taxa
In agreement with previous chromosome counts for H. murinum taxa, diploid (2n = 2x = 14), tetraploid (2n = 4x = 28) and hexaploid (2n = 6x = 42) cytotypes were detected among the examined accessions (Table 1). However, accession GRA2735supposedly glaucum -was found to have 2n = 28 instead of the expected 2n = 14. Further, accession GRA1183 -supposedly murinum -was found to have 2n = 42 instead of the expected 2n = 28. Individuals of these accessions were therefore grown in the field and identified morphologically following the key provided by Bothmer et al. [8]. Accession GRA2735 showed features diagnostic of glaucum, including pedicelate central spikelets longer than the lateral spikelets, and anthers with purple dots (Fig. S1). This is the first report of a tetraploid form of glaucum. Accession GRA1183, however, showed the diagnostic characteristics of leporinum (pedicelate central spikelets shorter than the lateral spikelets). Figure 1 shows the distinctive hybridization patterns obtained with a number of repetitive probes in metaphase chromosomes of the H. murinum diploid accession BCC2002.

Karyotype analysis of diploid H. murinum
FISH analysis using probe pTa71 revealed four signals at the secondary constriction of the two satellited chromosome pairs (Fig. 1a). One of these chromosome pairs also carried a pTa794 signal in a more proximal position (Fig. 1d). Another pair of chromosomes carrying a pTa794 signal in a distal position on the short arm was detected ( Fig. 1d and j). Thus, the combination of pTa71 and pTa794 easily distinguishes three chromosome pairs (Fig. 1k, top row). The pAs1 probe returned similar banding patterns for different chromosomes, making clear identifications very difficult (Fig. 1f). No signals were observed with probe pSc119.2.
The most clear, intense and rich pattern of signals was obtained with probe (AAG) 5 , which clearly identified the seven homologues chromosome pairs (Fig. 1e). (AAC) 5 produced similar though slightly less diagnostic patterns (Fig. 1h). Most of the AAC sites colocalized with AAG in the pericentromeric regions. Well defined and intense signals were observed with (AG) 10 near the centromeres of four chromosome pairs (Fig. 1c). In contrast (ATC) 5 ( Fig. 1b) and (ACT) 5 (Fig. 1g) only revealed signals of weak intensity that were little suitable as diagnostic markers. Finally, the expected signals for the telomeric probe were seen at the ends of all chromosome arms. In addition, interstitial telomeric repeats were found in the satellited chromosome pair carrying both pTa71 and pTa794 signals ( Fig. 1i-j).
Once the hybridisation pattern of each probe was analysed, multiple target in situ experiments using two-by-two combinations of the probes (reprobing the same chromosome preparations) allowed the characterization of the seven chromosome pairs (e.g., Fig. 1a-f). Following the classical nomenclature used for Triticeae chromosomes with unknown homoeology, karyotypes were constructed arranging chromosomes A to G in order of decreasing length with the satellited chromosomes at the end (Fig. 1k).
Chromosome identification and karyotype diversity within tetraploid H. murinum   In all tetraploid accessions, probe pHch950 hybridized in a dispersed fashion with seven chromosome pairs ( Fig. 2a and f). Six intense pTa71 signals were detected at the secondary constriction of the three satellited chromosome pairs. Two further minor pTa71 signals were seen on the long arms of a small metacentric chromosome pair (Fig. 2g). Six pTa794 signals were detected, the four strongest in two satellited chromosome pairs (Fig. 2g). Besides those expected at both chromosome ends, telomeric interstitial signals were observed on one of the two satellited chromosome pairs carrying (as in diploids) pTa71 and pTa794 signals ( Fig. 2g-h).
pSc119.2 revealed subtelomeric signals of different intensity ranging in number from four on two chromosome pairs in accessions GRA1021 and BCC2007, to ten on four chromosome In situ hybridization in accessions GRA1144 (a-b), GRA2735 (c-d) and BCC2007 (e-h) illustrating the patterns obtained with probes pHch950, pTa71, pTa794, pSc119.2, pAs1, (AAG) 5 , (AAC) 5 , (AG) 10 and the telomeric (Tel.) probe (CCCTAAAG) 3 in tetraploid H. murinum taxa. In a, the arrows point to (AG) 10 signals in five pairs of chromosomes. One pair carrying weak signals is shown in the insets. In b and e, the arrows point to pSc119.2 signals. The arrowheads in f point to the 14 chromosomes labeled with pHch950. In g, the arrows and arrowheads point to minor pTa794 and pTa71 sites respectively. In h, the arrows and arrowheads point to interstitial telomeric repeats and stronger, derived pTa71 signals (see g) respectively. The satellited chromosome pair belonging to subgenome Xu carrying pTa71, pTa794 and interstitial telomeric sequences is amplified in g9 and h9. Scale bar = 10 mm. doi:10.1371/journal.pone.0081385.g002 pairs in accession GRA2894 (compare Fig. 2b and e). This indicates diversity among the 4x accessions for the presence/ absence of this repetitive sequence. Similarly, (AG) 10 showed well defined signals of different intensity in different chromosomal regions (from the centromere to a subtelomeric position) in five or six chromosome pairs, depending on the accession analyzed (Fig. 2a).
The pAs1 probe revealed a rich pattern of signals on all chromosome arms, making individual chromosome identification very difficult (Fig. 2d and f). However, all 14 chromosomes pairs were easily identified on the basis of the chromosomal location of (AAG) 5 (Fig. 2c-e and Fig. 3a). The (AAC) 5 probe mostly colocalized with (AAG) 5 but produced a less diagnostic pattern (Fig. 2b). Finally, as in the diploids, no reliable (ATC) 5 or (ACT) 5 signals were observed; no attempt was therefore made to characterize their in situ patterns in detail.
The results obtained with the pHch950 probe were particularly important for identifying the chromosomes of the two subgenomes involved in the tetraploids (Fig. 2a and f). Firstly, the seven chromosome pairs revealed with the pHch950 probe were analyzed in detail with the diagnostic pTa71, pTa794 and (AAG) 5 probes. With the exception of the non-satellited chromo-  Fig. 2c and e show those of GRA2735 and BCC2007. Note that the in situ patterns for each accession -even those belonging to different H. murinum subspecies -are very similar to one another. (b) Karyotypes of accession GRA2894, providing a representative sample of the chromosomal distribution of different probes in tetraploids of H. murinum. Note that signals from probes pTa71, pTa794 and pSc119.2 were drawn over the DAPI-stained chromosomes. Asterisks indicate the polymorphic sites (presence/absence) observed among the tetraploid accessions (Table 2). doi:10.1371/journal.pone.0081385.g003 some pair carrying the minor 45S rDNA locus, the chromosomes showed sizes, morphologies and hybridization patterns similar to those seen for the chromosomes of diploid H. murinum (compare Fig. 1k and Fig. 3a). The above seven pairs were identified as belonging to subgenome Xu and were arranged in the karyotypes according to the nomenclature system previously employed for the diploids. The pair with no counterpart in the diploid was identified as G Although other minor differences were found between the Xu subgenome and the genome of the diploids (for example, FISH with the probe pTa794 returned no signal on chromosome E of the tetraploids), the physical map returned by the probes revealed great similarities between the diploid glaucum and the subgenome Xu of the tetraploids (compare Fig. 1k and Fig. 3b). For example, the satellited chromosome pair of this subgenome carries both 45S rDNA and 5S rDNA loci and interstitial telomeric sequences, just like the satellited chromosome F of diploids (compare Figs. 1j and  2g-h).
Once the identity of the Xu chromosomes was established, the seven chromosome pairs without pHch950 signals were characterized. These chromosomes should not share the same Xu genome, and the provisional designation of subgenome Xv is here proposed. Among the Xu and Xv chromosomes, homoeology relationships could be established for the two pairs of satellited chromosomes: chromosomes F (which, like chromosome F of diploids and subgenome Xu of tetraploids, carries both rDNA sequences) and G (which, like chromosome G of the diploids has a large satellite on the long arm) (Fig. 2g). The remaining chromosomes of subgenome Xv (A to E) were arranged in decreasing order of the length; however, no homoeologies can be established between these chromosomes and their counterpart in the subgenome Xu without further genetics analysis.
Finally, the easy identification of all 14 chromosome pairs after ND-FISH with (AAG) 5 allowed chromosome-by-chromosome analysis of the hybridization patterns of the remaining probes ( Fig. 2 e-f), and the assessment of cytogenetic diversity by comparing the homologous chromosomes of the five tetraploid accessions (Fig. 3a). Accession GRA2894, which showed the largest number of pSc119.2 and (AG) 10 signals among the tetraploids, was chosen to represent the karyotypes based on probes pTa71, pTa794, pAs1, pSc119.2 and (AG) 10 (Fig. 3b). The results are summarized in Table 2.
Identification of the three subgenomes presents in hexaploid H. murinum Figure 4 shows the distinctive hybridization patterns obtained with seven probes in metaphase chromosomes of the hexaploid accession GRA1183. Karyotypes for six probes were constructed (Fig. 5).
As in the tetraploids, seven chromosomes pairs were distinguished using pHch950 (Fig. 4a). The pTa71 probe revealed 10 strong signals at the secondary constriction of the five satellited chromosome pairs. Two minor pTa71 signals were also seen on a small metacentric chromosome pair (Fig. 4c). Eight pTa794 signals were detected, the six strongest in three satellited chromosomes pairs (Fig. 4c). As in the diploid and tetraploid taxa, one satellited chromosome pair (carrying both ribosomal loci) showed interstitial telomeric signals (Fig. 4d). The chromosomal distribution revealed by pAs1 was coincident with that seen for the diploids and tetraploids, although in the hexaploids (with more chromosomes of similar morphology and in situ patterns) it was very hard to identify individual chromosomes (Fig. 5). Ten pSc119.2 signals of different intensity were localized in subtelomeric regions of four chromosome pairs (Fig. 4g).
Once again, among the SSR probes investigated, the richest pattern of signals was obtained with probe (AAG) 5 (Fig. 4b and f). (AAC) 5 signals mostly co-localized with clusters of AAG repeats, although some chromosome were very enriched in (AAG) 5 signals and showed no accompanying (AAC) 5 signal (compare Fig. 4e-f). As in the diploid and tetraploid accessions, no very reliable signals were observed with (ACT) 5 and (ATC) 5 . Finally, discrete (AG) 10 signals were observed in different locations in 12 chromosome pairs (Fig. 5).
The same strategy used for the characterization of the subgenomes in the tetraploids was followed in the hexaploids. Firstly, the seven chromosome pairs revealed with pHch950 were analyzed in detail with the diagnostic probes pTa71, pTa794 and (AAG) 5 . This set of chromosomes has the same morphology and in situ patterns as the chromosomes of the Xu subgenome of the tetraploids. The 14 chromosome pairs that showed no hybridization with probe pHch950 were easily separated into two groups of seven chromosome pairs. One group showed the same morphology and in situ patterns as the chromosomes of the subgenome Xv observed in the tetraploids, and was easily identified (Fig. 4b). The other group, whose members were very enriched in (AAG) 5 signals, belongs to a related genome here named subgenome Xw (Fig. 5). Probes pTa794 and pTa71 diagnostically identified two chromosomes belonging to subgenome Xw. Like F chromosomes belonging to subgenomes Xu and Xv, the pair of satellited chromosomes carrying proximal 5S rDNA sites was identified as chromosome pair F. The other pair of Xw satellited chromosomes carrying only 45S rDNA was identified as chromosome pair G (Fig. 4c). The remaining chromosomes belonging to the Xw subgenome (A to E) were arranged in the karyotype on the basis of chromosome size. Once again, no homoeologies could be established between these chromosomes and the respective Xu and Xv chromosomes.
Finally, the identification of chromosomes with (AG) 10 and pSc119.2 signals was performed after reprobing with (AAG) 5 (Fig. 4g). (AG) 10 was found on chromosomes belonging to all three subgenomes (the positions on Xu and Xv were the same as seen on the chromosomes of the tetraploids). It is noticeable that, as in the diploids (genome Xu), no signals were observed with the probe pSc119.2 on chromosomes belonging to subgenome Xw (Fig. 5, Table 2)

Meiotic behaviour of chromosomes
The meiotic behaviour of the chromosomes was analyzed in pollen mother cells at metaphase I in tetraploids BCC2007 and GRA2735. Only bivalents were formed (mostly ring-type); no univalents, multivalents or other chromosome associations were observed. Sequential experiments using two-by-two combinations of the probes unequivocally identified the seven bivalents. In both accessions, the bivalents only involved homologous chromosomes (Xu-Xu and Xv-Xv) (Fig. 6).

Identification of H. murinum chromosomes
The first step in understanding the genome structure and evolution of a species is the unambiguous discrimination of its chromosomes. The identification of each H. murinum chromosome pair, including its hexaploids, is here reported for the first time. As indicated by other authors, the five Triticeae probes used (pTa71, pTa794, pAs1, pHch959 and pSc119.2) were insufficient for the reliable identification of most chromosomes [15,18,24]. However, the pattern of distribution of several SSRs was shown very useful in chromosome identification. Indeed, the use of (AAG) 5 in combination with the morphology of DAPI-stained chromosomes was enough to easily distinguish all individual chromosomes. The use of a single probe for chromosome identification facilitates the co-localization of other probes carrying different fluorochromes in two-colour in situ experiments. With the exception of (ACT) 5 and (ATC) 5 (which were ineffective as chromosome markers), the probes used in the present work provided a saturated physical map of H. murinun with a rich set of cytogenetic landmarks distributed throughout all chromosome arms.
It is well documented that several members of the tribe Triticeae commonly show 5S rDNA loci in homoeologous group 5 (e.g., wheat, rye and H. bulbosum), while 45S rDNA loci are commonly present in homoeologous groups 5 and 6 (e.g., barley and H. chilense) [25][26][27]. On the basis of the location of rDNA loci, chromosomes F and G of (sub)genomes Xu, Xv and Xw should be assigned respectively to the Triticeae homoeologous groups 5 and 6. However, homoeologies can only be fully supported by demonstrating conserved synteny in further analysis. Since H. murinum chromosomes can be identified when in bivalent configuration (Fig. 6), the analysis of meiotic pairing in the hybrids of H. murinum with H. vulgare should allow homoeologous relationships to be established between the species, as reported for H. bulbosum when analyzing its hybrids with H. vulgare [27].

The origin of polyploids in the H. murinum complex
The confluence of distinguishable groups of seven chromosome pairs in the tetraploids and hexaploids contributes to our knowledge on the origin of the polyploid forms of H. murinum. The present results show the morphology and in situ patterns of the chromosomes in the diploid glaucum to be very similar to a set of 14 chromosomes in the tetraploid and hexaploid cytotypes. This supports the idea that glaucum was the diploid donor of the subgenome Xu present in H. murinum polyploids, and agrees with the assumption of Rajhathy and Morrison [11] (based on studies of meiosis) that one of the genomes of the murinum tetraploid cytotypes is that of diploid glaucum. Nearly nothing is known about the identity of the non-glaucum diploid progenitors. Recent molecular phylogenies have shown that ancient H. murinum diploid progenitors involved in the origin of polyploids should belong, along with glaucum, to the same Xu genome group of Hordeum species that became isolated after the separation of their sister group, i.e., the H genome lineage of H. vulgare and H. bulbosum [17,19,28]. The present results seem to reflect the existence of substantial differences in the amount and distribution of certain repeat DNA sequences among the subgenomes present in polyploids. Thus, even assuming that the Xu, Xv and Xw genomes have a monophyletic origin, their separation from a common ancestor must have occurred at a relatively early stage. A detailed chromosome analysis of other diploid forms of Hordeum and closely related Triticeae species might provide new clues regarding the identity of the Xv and Xw donor progenitors, or support the hypothesis suggested by Jacob and Blattner [20] that these genomes belong to extinct species.
Irrespective of the identity of the non-glaucum parents, the donor species of the subgenome Xv must have been common to all tetraploids of all subspecies. Indeed, only minor differences were seen when comparing karyotypes for the analyzed probes of the five tetraploid accessions examined. The in situ patterns of the Xu and Xv subgenome chromosomes of the hexaploids are almost identical to the corresponding chromosomes present in tetraploids, indicating a common origin and low divergence after the addition of the third Xw subgenome at the hexaploid level.

Genome remodelling and polyploidy
Allopolyploidy has been an important mechanism in the rapid genomic evolution of the members of Triticeae [29][30][31]. The present work contributes to our understanding of the coevolution of the Xu, Xv and Xw subgenomes following the polyploidization process. In general, the chromosomes of the same (sub)genome at different ploidy levels are very similar, suggesting that no great karyotypic alteration occurred after polyploidization. Only chromosome G of subgenome Xu present in the polyploids appears to have no definitive counterpart in diploid glaucum. This agrees with the fact that diploid glaucum has two pairs of satellited chromosomes (F and G), as revealed by the presence of secondary constrictions and major 45S rDNA signals (Fig. 1k, [24]), while only one such pair (chromosome F) is present in the Xu subgenome of the polyploids [15,18]. Genome remodelling after polyploidization could have led to the replacement of the satellited chromosome G observed in the diploids (with major 45S rDNA sites) by the small metacentric chromosome with only minor 45SrDNA signals found in the tetraploid and hexaploid cytotypes. Deletion of the 45S rDNA genes is likely the main cause of this karyotypic difference. In fact, deletion or inactivations of ribosomal genes are common in hybrid and polyploid members of Triticeae. Intergenomic translocations are also common structural changes detected in allopolyploids. An example of this is the well characterized cyclic translocation of chromosomes 4A, 5A and 7B in tetraploid and hexaploid wheats [32]. No large translocations were detected in the present study. However unlike diploid glaucum, which shows no pSc119.2 signals [24,33], pSc119.2 signals were observed in subtelomeric positions on one Xu chromosome arm of three tetraploids and the hexaploid accessions analyzed in the present work. This suggests the presence of a terminal translocation or perhaps simply the jumping of transposable elements after allopolyploidization. The presence of the pSc119.2 signals in the same location on chromosome FXu as seen in two leporinum accessions (from Italy and France) might indicate a common origin for these samples. In contrast, the interstitial telomeric repeats on chromosome FXu, invariably found in the seven accessions here analyzed, suggest that this chromosomal rearrangement must be an ancient characteristic of the Xu genome. This may have occurred before the diversification of the H. murinum species but after the diversification of genomes Xu and H since no interstitial telomeric repeats are seen in barley [22].

Taxonomy of the H. murinum complex
The H. murinum complex is probably one of the best studied groups in the genus Hordeum. However, the taxonomic treatment of H. murinum taxa has always been controversial. Although it is well accepted that H. murinum can be split into three subspeciesglaucum, murinum and leporinum [8] -their geographical and ecological distinctiveness has led some authors to contemplate three separate species: H. glaucum Steudel, H. murinum sensu stricto and H. leporinum Link [34,35]. Another problem of the complex is the taxonomic status of the different cytological forms of leporinum (which are not recognized as distinct taxa since they show no distinctive morphological traits). An unexpected result of the present work was the finding of 28 chromosomes in accession GRA2735, classified as glaucum by the seed bank delivery information and confirmed as such by the information in Fig.  S1. This is the first report of a tetraploid form of glaucum, which further reflects the difficulties encountered in dealing with the complex.
The present cytogenetic results showed no greater similarity to exist between the three leporinum tetraploid accessions than between tetraploids of the different subspecies. This agrees with many phylogenetic studies that have been unable to separate  Fig. 4b). Note that the pTa71, pTa794 and pSc119.2 signals were drawn over the DAPIstained chromosomes. The orange signals in the pAs1 karyotype are the stronger red signals observed in the (AG) 10 karyotype. doi:10.1371/journal.pone.0081385.g005 tetraploids of leporinum and murinum. Indeed, populational analyses of several quantitative and qualitative features, and indeed of different molecular markers, have revealed no significant differentiation between tetraploid populations morphologically classified as different taxa [9,[35][36][37]. Some authors suggest that the different subspecies form a continuous morphological cline strongly correlated with bioclimatic variables [18]. Others suggest the existence of hybrids of murinum and leporinum [37], while yet others indicate the separation of these taxa must have occurred relatively recently [20]. The present results suggest that all murinum tetraploids should be included within the same species.
In conclusion, this study reveals the origin of the polyploids of the H. murinum subspecies and highlights their phylogenetic relationships. It is here suggested that the members of the complex be divided into three subgroups according to their ploidy level: 2xmurinum, 4x-murinum and 6x-murinum. 2x-murinum with the haploid genome formula Xu, should include the diploid form of glaucum; 4x-murinum, with the haploid genome formula XuXv, should include the currently recognized murinum and the tetraploid cytotypes of glaucum and leporinum. Finally, 6x-murinum with the haploid genome formula XuXvXw, should include the hexaploid cytotypes of leporinum. Certainly, the taxonomic treatment of H. murinum based exclusively on morphological criteria would appear to be questionable.