Exploiting repetitive sequences and BAC clones in Festuca pratensis karyotyping

The Festuca genus is thought to be the most numerous genus of the Poaceae family. One of the most agronomically important forage grasses, Festuca pratensis Huds. is treated as a model plant to study the molecular mechanisms associated with tolerance to winter stresses, including frost. However, the precise mapping of the genes governing stress tolerance in this species is difficult as its karyotype remains unrecognized. Only two F. pratensis chromosomes with 35S and 5S rDNA sequences can be easily identified, but its remaining chromosomes have not been distinguished to date. Here, two libraries derived from F. pratensis nuclear DNA with various contents of repetitive DNA sequences were used as sources of molecular probes for fluorescent in situ hybridisation (FISH), a BAC library and a library representing sequences most frequently present in the F. pratensis genome. Using FISH, six groups of DNA sequences were revealed in chromosomes on the basis of their signal position, including dispersed-like sequences, chromosome painting-like sequences, centromeric-like sequences, knob-like sequences, a group without hybridization signals, and single locus-like sequences. The last group was exploited to develop cytogenetic maps of diploid and tetraploid F. pratensis, which are presented here for the first time and provide a remarkable progress in karyotype characterization.


Introduction
Grasses (Poaceae) constitute a large family of monocotyledonous plants. This family encompasses cereals, turf and pasture grasses, and others such as bamboos or sugarcane. The area of grassland is estimated to be two times higher than the area of cropland. However, the genetic/ genomic resources for grassland species are still limited [1,2]. The most important forage grasses used in agriculture are species derived from the Festuca-Lolium complex, i.e. Festuca pratensis Huds. (meadow fescue), F. arundinacea Schreb. (tall fescue), Lolium multiflorum Lam. (Italian ryegrass), and L. perenne L. (perennial ryegrass). Festuca species are characterized by high tolerance to abiotic and biotic environmental stresses. On the contrary, Lolium species possess higher forage quality and productivity in favourable conditions but a significantly lower performance under stresses. The species of these two genera are closely related and their a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 desirable complementary agronomic traits can be combined in intergeneric hybrids and their introgression derivatives [3][4][5][6]. However, the mapping of the genes governing these important traits in particular species and in their intergeneric hybrids is difficult as the majority of chromosomes within Festuca and Lolium species cannot be unambiguously recognized.
Cytogenetics has proved to be an important tool to determine evolutionary and phylogenetic relationships between various plant groups, as well as to uncover the genome structure at the chromosomal level [7][8][9]. Fluorescence in situ hybridization (FISH) is one of the most important and useful methods that provide chromosome landmarks. FISH in combination with large-insert BAC-based clones restricted to single locus sequences have been efficiently used in karyotype analyses of such plant species as Brachypodium pinnatum [10], Daucus carota [11] and Phaseolus microcarpus [12]. Sequencing revealed that in plant genomes repetitive sequences constitute most of the DNA and represent up to 85% of the genome size [13][14][15][16]. This high content of repetitive sequences often causes trouble in the identification of single locus sequences that enable the recognition of individual chromosomes. However, this type of sequences, e.g. the Afa-family in the Hordeum vulgare karyotype, could be useful in chromosome identification [17].
The Festuca genus includes more than 500 species and is thought to be the most numerous and complex genus of Poaceae [18]. The ploidy level of Festuca species ranges from diploid (2n = 2x = 14) up to dodecaploid (2n = 12x = 84) [19]. The haploid genome size in Festuca ranges from 1.638 Gbp to 12.553 Gbp (per 1C DNA) [20]. The sequencing of the 4F chromosome of F. pratensis revealed that the most abundant repeats (about 30%) constitute Ty3gypsy-like elements [21].
Some chromosomes of Festuca can be recognized by 5S and 35S rDNA sequences [22][23][24]. In the F. pratensis karyotype the chromosome no. 2 (2F) with a secondary constriction (NOR) and the chromosome no. 3 (3F) with 5S rDNA loci are well-recognized (chromosome nomenclature according to Thomas et al. [22]). However, although this species is treated as a model plant within the group of Festuca-Lolium forage grasses to recognize the molecular basis of mechanisms involved in tolerance to low temperature, including frost [5,25], its remaining chromosomes have not been well-characterized to date [26].
In this paper, two libraries derived from F. pratensis nuclear DNA with various contents of repetitive DNA sequences, a BAC library and library representing sequences most frequently present in the F. pratensis genome, were used in a cytogenetic approach. We hypothesized that these two libraries could be an excellent source of molecular probes to be applied for the FISH procedure aiming at the identification of F. pratensis chromosomes and their particular arms. Thus, the main goal of the present study was to characterize the karyotype of diploid and tetraploid F. pratensis cultivars using BAC clones and probes enriched with sequence repeats, and finally to create the first detailed cytogenetic maps of this species.

Construction of a F. pratensis genomic DNA library
To study F. pratensis genome organization, a genomic library from F. pratensis nuclear DNA in a cloning vector was prepared. The DNA library, representing sequences most frequently present in the F. pratensis genome, was constructed according to the procedure of Nunome et al. [27]. Genomic DNA was isolated from ca. 100 mg of powdered leaves using the GeneJET Plant Genomic DNA Purification Kit (Thermo Scientific). The quality and quantity of DNA were analyzed in a 1% agarose gel and using a NanoDrop photometer (Thermo Scientific). Total genomic DNA was digested in a volume of 100 μl with 25 units of the restriction endonucleases HindIII (Promega) for 17 h at 37˚C. The enzyme was inactivated by heating for 10 min at 65˚C. Digested DNA was purified by extractions with phenol/chloroform and precipitation with 100% ethanol, and then subcloned into the HindIII cloning sites of the pBluescript KS(+) vector. To reduce the background of "empty" clones, digested plasmid DNA was dephosphorylated using alkaline phosphatase (calf intestinal; Promega). Ligation was performed in a total volume of 10 μl using 3 units of T4 DNA ligase (Promega) and the molar ratio of insert to vector was 3:1. Library construction and screening were performed directly by transformation of a ligate into competent cells of XL1Blue MRF' E. coli. Putative positive clones were screened on selective LB medium containing ampicillin at 100 μg.ml -1 , and subjected to isolation of plasmid DNA using GeneJet Plasmid Miniprep Kits (Thermo Scientific). The lengths of 469 clones were estimated by separation in 1% agarose gels.

Fluorescent in situ hybridization
Chromosome preparation. Metaphase accumulation and fixation procedures, as well as the preparation of chromosome slides for in situ hybridization, were carried out by the enzymatic squash method according to the protocol described by Książczyk et al. [24].
Probes. For the experiments presented here, BAC clones and a DNA library, representing sequences most frequently present in the F. pratensis genome, were used. From the BAC library moderate-and low-repetitive clones were selected on the basis of previously performed dot blot analysis. From the newly created library clones were chosen on the basis of their molecular size. In total, 131 BAC-based probes and 61 molecular probes enriched with sequence repeats were physically verified. To identify some chromosomes within the karyotype, rDNA sequences were used. The 5S rDNA probe was derived from Triticum aestivum clone pTa794 [28] and the 35S rDNA probe was generated from the coding region of 26S rDNA of Arabidopsis thaliana [29].
Hybridization. FISH experiments were done according to Książczyk et al. [31] with minor modifications. Briefly, the hybridization mixture contained 100 ng of each probe in the presence of 50% formamide, 2 × SSC and 10% dextran sulphate. All the clones were hybridized separately, one clone per one chromosome slide. Chromosomal DNA in hybridization mixture was denatured at 80˚C for 2 min and allowed to hybridize overnight at 37˚C. For the detection of the hybridization signals of probes labelled with digoxigenin, anti-digoxigenin conjugated with fluorescein isothiocyanate (FITC) was used.
Moreover, the location of selected clones was checked in relation to each other. The reprobing was conducted by first removing particular probes before the next round of hybridization with another probe, according to the procedure described by Heslop-Harrison [32]. The procedure involved washing in 4 × SSC with 0.2% Tween 20 (Sigma-Aldrich) three times per 10 min, and in 2 × SSC twice per 10 min. Then, the slides were dehydrated in an ethanol series at room temperature. The sets of probes were subsequently applied on each slide followed by the application of rDNA probes. In total three rounds of hybridization with the clones, and the fourth one with rDNA sequences, were performed.
All the steps in the genomic in situ hybridization (GISH) protocol were performed according to the Kosmala et al. [5] procedure with minor modifications. In GISH, the total genomic DNA of L. perenne was used as a probe and labelled with digoxigenin using a DIG-Nick Translation Kit according to the manufacturer's procedure (Sigma-Aldrich). Genomic DNA of F. pratensis was used as blocking DNA after being sheared by boiling for 30 min.
All the images were acquired using an Olympus XM10 CCD camera attached to an Olympus BX 61 automatic epifluorescence microscope. Image processing was carried out using Olympus Cell-F imaging software (ver. 3.1; Olympus Soft Imaging Solutions GmbH, Germany) and Micrographx Picture Publisher software (ver. 10; Corel Corporation, Canada). For the cytogenetic maps which were created, the distributions of hybridization signals in the chromosomes is shown in relative lengths determined as means of ten measurements.

Results
A DNA library representing sequences most frequently present in the F. pratensis genome DNA digested by HindIII was used for the construction of a genomic DNA library. After the isolation of inserted DNA, the clones were divided into groups according to their molecular size. A classification of clones is shown in Table 1. The most numerous class that was physically mapped in diploid F. pratensis chromosomes comprises sequences with a length of 10 000 bp and more. However, some clones from other groups were also verified to give signals on chromosomes. The clones with length smaller than 3 500 bp were not cytogenetically mapped due to the limited resolution of metaphase chromosomes in the FISH technique.

Chromosome nomenclature
For Festuca and Lolium species, two types of chromosome nomenclature exist according to the numbering systems of Thomas et al. [33] and of Triticeae [34]. Although in this work the nomenclature of Thomas et al. [33] was used, in the present study we propose a new nomenclature for the karyograms and cytogenetic maps which we created. This new nomenclature was developed due to the unambiguous identification of the first to seventh chromosome pairs identified by Thomas et al. [33]. We named chromosomes with letters from A to G. Chromosomes 2F and 3F (according to Thomas et al. [33]) correspond to our C and D chromosomes, respectively. Furthermore, a chromosome with an additional 5S rDNA locus was named chromosome E2. Chromosome E2 and chromosome E without rDNA sequences are presented separately on the karyograms. Table 1. Classification of selected molecular probes enriched with DNA sequence repeats according to their size. Clones which were not cytogenetically analysed are indicated by NA.

Number of clones in this group
In order to achieve a high genome coverage for the unmapped F. pratensis chromosomes, 131 BAC-based probes and 61 molecular probes enriched with sequence repeats were examined cytogenetically. Six groups of hybridization patterns were observed: (1) dispersed-like sequences, (2) chromosome painting-like sequences, (3) centromeric-like sequences, (4) knoblike sequences, (5) a group without hybridization signals, and (6) single locus-like sequences ( Table 2). The dispersed-like and chromosome painting-like sequences should be classified into the same group, because both represent highly repetitive sequences specific to species or genomes. However, in this paper they were divided into two separate groups on the basis of their distribution in chromosomes and potential usefulness for more detailed analysis of chromosome identification of Festuca-Lolium hybrids. The highest number of sequences within a group constituted of dispersed signals along both arms of the chromosomes (group no. 1; 85 clones) was found. Within the chromosome painting class (group no. 2; 21 clones) it was noticed that only one clone (393) mapped to the whole chromosomes, while the remaining chromosome painting-like sequences hybridized to the entirety of chromosomes excluding segments of the 35S rDNA loci (NOR) (Fig 1A), the NOR and centromeric regions (Fig 1B), as well as to chromosomes without centromeric regions ( Fig 1C). Moreover, hybridization of clone 183 to the genotype with additional B chromosomes revealed that this sequence is specific for the A chromosome set ( Fig 1B). The hybridization pattern of the remaining chromosome painting-like sequences resulted in signals that were present in supernumerary B chromosomes without pericentromeric (Fig 1A) or Table 2. Types and numbers of hybridization patterns after FISH with BAC-based probes and molecular probes enriched with sequence repeats in the F. pratensis genome.

F. pratensis BAC library (131 clones in total)
Dispersedlike sequences centromeric regions (Fig 1C). Within chromosome painting-like sequences, one species-specific clone (L16 Fp04) was identified; hybridization of this clone to chromosomes of a F. pratensis × L. perenne hybrid revealed signals restricted to the F. pratensis chromosomes. In Fig 2 the same metaphase plate after FISH with clone L16 Fp04 (Fig 2A) is compared with results obtained from GISH using total genomic DNA of L. perenne as a probe (Fig 2B). The group no. 3 encompasses sequences specific for centromeric regions of chromosomes. Within this group sixteen clones were identified, including clones with signals restricted to centromeric regions (Fig 3A), clones labelling (peri)centromeric regions, and some interstitial  Insight into Festuca pratensis karyotype positions (Fig 3B), as well as one clone (171) which was mapped in both (peri)centromeric and telomeric regions (Fig 3C). Hybridization patterns of the selected clones specific for centomeric regions with interstitial signals within chromosomes might be useful to facilitate chromosome identification (Fig 4). In Fig 4A, 4B and 4C the physical distribution of three sequences is shown (clone 595, 616 and G18 Fp01).
Among the clones analysed, knob-like sequences (group no. 4; 17 clones) were also identified. These sequences hybridized to specific and random blocks on chromosomes. The vast majority of them were mapped to segments located in all chromosomes. Clone 266 revealed a signal in telomeric and subtelomeric regions of all of the chromosomes. Similarly, several clones classified to this group labelled interstitial regions in both arms of all of the chromosomes (S1 Fig). Clones which revealed no hybridization signals were also identified in both libraries (group no. 5; 42 clones). The majority of unmapped clones was characterized by a low physical size of the inserted DNA in the library representing sequences most frequently present in the F. pratensis genome.
The last group of hybridization patterns consisted of clones apparently mapped to a single locus (group no. 6; 11 clones), of which the majority were derived from the BAC library. However, two single locus clones (clones 228 and 324) were also identified in the library representing sequences most frequently present in the F. pratensis genome. Some of these showed clear signals in an individual pair of chromosomes (Fig 5), but for some of them weak dispersed signals in the remaining chromosomes were also observed (data not shown). In Fig 5A the physical location of selected single locus clones is presented. The position of one of these clones (E5 Fp04) co-localized with 35S rDNA (Fig 5B). Clone K11 Fp04 was mapped to two pairs of chromosomes, including one unrecognized pair and one bearing 5S rDNA (chromosome pair D). However, in chromosome D this clone hybridized to arms without rDNA loci. The last clone (A21 Fp02) was mapped to five chromosomes: two C chromosomes in the arm without 35S rDNA loci, one pair without recognized landmarks, and only one D chromosome in the arm without 5S rDNA (Fig 5A).
The distribution of selected clones is presented in Table 3. The results of FISH experiments are summarized on the ideograms of F. pratensis chromosomes (Fig 6). To verify the position of clones in unrecognized chromosomes, experiments using different sets of probes on the same chromosome slides were performed, and rDNA-FISH was also carried out on these slides to discriminate rDNA-bearing chromosomes.
A pool of 17 clones was used to build a cytogenetic map of diploid F. pratensis and to examine similarities and differences between diploid and tetraploid form of this species with respect to their hybridization patterns (Table 3). In the tetraploid form, the clones 433 and 498 were mapped to almost the same positions, centromeric regions and the four largest metacentric A chromosomes. In the diploid form, signals of these clones were observed in centromeric regions and the two largest chromosomes. A similar pattern for clone 607 was also noticed, except for the telomeric-like sequences that were mapped on chromosomes of the diploid form only. Clone 228 was mapped in two and four unrecognized chromosome in diploid and tetraploid F. pratensis, respectively. Clones M21 Fp01 and E5 Fp04, specific for the 35S rDNA position in diploid F. pratensis, were also mapped to the 35S rDNA loci in the tetraploid form. Surprisingly, clone E15 Fp01, which was colocalized with 5S rDNA in the diploid form, was mapped in centromeric regions beyond the 5S rDNA loci in the tetraploid form. Sequence A21 Fp02 hybridized to C chromosomes, two unrecognized chromosomes, and to one D chromosome in diploids. In the tetraploid form this sequence mapped to C chromosomes and two unrecognized chromosomes. An intriguing position of clone 324 was observed in C chromosomes of tetraploids, where in one pair of chromosomes the sequence was visible in an arm without 35S rDNA loci, while in the second pair of chromosomes it was in an arm with 35S rDNA. Furthermore, in the diploid form this clone mapped to four unique positions, while in the tetraploid form four single locus signals plus additional signals in telomeric/subtelomeric regions were observed. The signals of M21 Fp02 in diploid F. pratensis were observed in D chromosomes and in four unrecognized chromosomes, but in the tetraploid form this sequence was visible in one pair of C chromosomes and four unrecognized chromosomes. Moreover, signals of clone N14 Fp03, which was mapped to selected chromosomes in the diploid form, were restricted to centromeric regions in the tetraploid form. Centromeric regions in the tetraploid form were also marked by two other clones (595 and 616). Additionally, for clone 595 four signals in unrecognized chromosomes were noticed, as well as telomeric-like signals in all chromosomes. For clone 616 two signals were visible in two C chromosomes, in the arms bearing 35S rDNA sequences. Three clones (A24 Fp02, O16 Fp03, K11 Fp04) revealed no signals in tetraploid F. pratensis. The schematic position of all mapped clones in tetraploid F. pratensis is shown in the ideogram (Fig 7).

Discussion
The BAC-FISH technique has not been widely used in Festuca and Lolium species. Until now no cytogenetic or physical map has been published for F. pratensis or for other species from the Festuca-Lolium complex. The available cytogenetic data about the karyotype structure of F. pratensis was restricted to the morphology of chromosomes and the identification of two pairs Table 3. Distributions of selected clones in diploid and tetraploid forms of F. pratensis. Positions of clones in rDNA-bearing chromosomes were assigned by reference to rDNA sequences; (M+) is the arm with an rDNA marker and (M-) the arm without an rDNA marker.

Name of clone Distribution in diploid F. pratensis Distribution in tetraploid F. pratensis
E15 Fp01 D chromosomes (M+)-colocalization with 5S rDNA sequence centromeric-like sequences, D chromosomes (M+), four unrecognized chromosomes G18 Fp01 diversified pattern across all chromosomes (see Fig 4C) centromeric-like sequences of chromosomes using rDNA-FISH [22,24]. In this study, our knowledge of genome organization of this species was extended. A novel source of genomic data, a library representing sequences most frequently present in the F. pratensis genome, has been developed and mapped in chromosomes. Moreover, clones from a BAC library containing a relatively high proportion of unique sequences were also used to analyze the chromosomes.
Although little is known about the structure of Festuca chromosomes, except for rDNA distribution [22,24], an effort has been invested in the last years to recognize these chromosomes. Kopecký et al. [26] showed the result of mapping BAC 1G18 in individual chromosomes of F. pratensis introgressed into L. multiflorum. This BAC clone probably contains a high amount of repetitive DNA, because signals were observed mostly in centromeric regions. Almost the same results were obtained in our experiments for C and D chromosomes (Fig 4C), but some discrepancies between both analyses were also detected including a higher number of signals  within all of the chromosomes in the present study, especially in the interstitial locations. The occurrence of minor differences could be related to different types of plant material used in both studies (a cultivar of F. pratensis and an introgression form of L. multiflorum with introgressed individual F. pratensis chromosomes) or to different conditions of post hybridization washes.
The availability of BAC-based chromosome-specific landmarks has enabled, for example, a more detailed analysis of the karyotype of Brachypodium species, including insight into the karyotype evolution [10,35]. In our study, a large pool of clones provided a good possibility to characterize the F. pratensis karyotype. The application of all of the identified single locus clones in combination with several repetitive sequences allowed to identify individual chromosomes and to develop cytogenetic maps of F. pratensis (Fig 6; Table 3) using a total of 28 out of 192 clones. The highest number of landmarks was found for one of the largest metacentric chromosome. Nevertheless, for one chromosome (named with letter B) single locus sequences were not found. In species with small genomes, such as Arabidopsis thaliana [36], Brachypodium distachyon [37] and Lotus japonicus [38], cytogenetic maps have been constructed using BAC clones as probes, enabling the efficacious integration of cytogenetic, physical and genetic maps. Integrated physical and genetic maps could be helpful in studies of genome organization and evolution, as well as for anchoring of whole genome shotgun sequences [39,40].
Sequences derived from genomic DNA libraries can be used for chromosome painting. Interestingly, we observed only one clone (393) which painted the entire karyotype. Several other clones produced a signal over all the chromosomes with the exception of 35S rDNA loci and/or centromeric regions. Centromeric regions are mainly composed of tandem repeats and retrotransposons [41,42], and centromere-specific sequences including crwydryn elements in rye [43]), cereba in barley [41], and CRM1 and CRM4 in maize [44] have been identified. Such elements may also exist in the F. pratensis genome, but were absent in the clones producing signals over all chromosomes except the centromeric regions.
Chromosome painting clones enable ancestral genome recognition in allopolyploid species. Zhang et al. [45] identified sequences that distinguish the three genomes of Triticum aestivum, and in Cucumis metuliferus a species-specific repetitive sequence (pMetSat) has been identified by Yagi et al. [46]. In our study, we observed a hybridization signal of the chromosome-like painting clone L16 Fp04 only in F. pratensis chromosomes in F. pratensis × L. perenne hybrids, indicating a species-specificity of this sequence. Moreover, we identified clone 183, which is specific for the A chromosome complement, because signals in additional B chromosomes did not occur. Houben et al. [47] found that supernumerary B chromosomes of rye incorporate large amounts of B-specific repeats as well as insertions from cytoplasmic organellar DNA. This might explain the lack of signals in additional B chromosomes in the F. pratensis karyotype (Fig 1B). On the other hand, most of the clones produced visible signals in both A and B sets (Fig 1A and 1C) supporting the hypothesis that supernumerary B chromosomes originated as a by-product of A chromosome evolution [48]. Thus, some sequences are probably common to both sets of chromosomes.
The physical mapping of selected clones in diploid and tetraploid F. pratensis revealed different hybridization patterns, e.g. clone N14 Fp03, which was mapped to selected chromosomes of the diploid form, resulted in centromeric-like signals in tetraploid forms (Fig 7,  Table 3). This is consistent with the hypothesis that a polyploid is not simply the sum of the diploid parents and that many changes including chromosome rearrangements appear after polyploidization [49]. For example, elimination of low-copy sequences derived from diploid progenitors has been observed in hexaploid bread wheat [50]. Probably, sequences A24 Fp02, O16 Fp03 and K11 Fp04, which gave single locus signals in the diploid form, were eliminated during the evolution of tetraploid forms (Fig 7, Table 3). Large scale re-organization including inversions, duplications and translocations has been evidenced in many plant polyploids [51,52]. Similarly, we found chromosome rearrangements in autopolyploid plants of F. pratensis; the clone 324 was mapped in two pairs of C chromosomes in tetraploids but the distribution of signal differs and one pair carries a signal in the arms with 35S rDNA loci, while in the second pair this sequence hybridized to the arms without 35S rDNA (Fig 7, Table 3). What is more, changes in the structure of D chromosome between both ploidy levels were identified; hybridization of clone A21 Fp02 resulted in signals in only one chromosome D and this clone did not give signals in this chromosome pair at the tetraploid level (Table 3). Similarly we observed hybridization signals for M21 Fp02 only in the D chromosomes in diploids but no signals in the tetraploid form. However, in the autotetraploid form this sequence was mapped in another rDNA-bearing chromosome (chromosome C) but in only one out of two chromosome pairs. Chromosomes C and D in the karyotype of F. pratensis bear rDNA loci, and it was previously reported that rDNA sequences can be highly polymorphic [24]. However, it is worthwhile to mention that most of the clones appeared in the expected numbers and positions in autotetraploid plants.
This research demonstrates the potential of using repetitive and unique types of sequences derived from BAC and genomic DNA libraries for the development of cytogenetic markers. These libraries allowed more detailed characterization of the chromosome complement of F. pratensis. A large group of clones constitutes sequences with a high amount of repetitive DNA, restricted to centromeric/pericentromeric and telomeric regions. Furthermore, the comparative mapping between Festuca and Lolium species also confirmed that these genomes consist of large number of repetitive sequences [21,53].
Our results could be a helpful reference and a good starting point for further research to identify the particular chromosomes of this species within the chromosomal set in intergeneric hybrids of F. pratensis with Lolium species. Moreover, this approach could be simultaneously associated with physical mapping of chromosomal regions carrying genes governing tolerance to environmental stresses, including low temperature, in F. pratensis and its intergeneric Festuca × Lolium hybrids.