Introduction

With the advent of large-scale genetics and an increased consciousness of polyploidization in plant diversification and evolution, attention in genome size variation in eukaryotes has increased considerably over the past period (Oliver et al. 2007). Furthermore, predominantly in plants, a great deal of research has been concentrated for gene repetition and whole genome duplication events (De Storme and Mason 2014). Nonetheless, information regarding the mechanisms involved in genome augmentation, and to extent implementation for breeding, is scarce.

Polyploids have a prominent breeding significance as they can lead to greater vegetative harvests, higher quality, superior tolerance against biotic and abiotic stresses and greater geographic dispersion (Dewitte et al. 2012; Younis et al. 2014). The two central modes of polyploidization are asexual polyploidization via somatic cell chromosome doubling and sexual polyploidization through the formation of functional 2n gametes. In this respect, 2n gametes are invaluable to the development of polyploids. As a consequence, 2n gamete exploitation is a significant method to polyploidy as they produce ‘in one step’ novel (autopolyploids or allopolyploids) species (Brownfield and Köhler 2011; Younis et al. 2014). Furthermore, sexually (meiotic) formed polyploids, versus somatic (mitotic) polyploids, can be of immense significance in breeding programs for they can associate genetic attribute of polyploidy with sexual hybridization and meiotic crossovers (Dewitte et al. 2012; Younis et al. 2014). As a consequence 2n gametes are invaluable in terms of enrichment of genetic variability and heterosis.

The formation of 2n gametes is usually due to genetic or environmental influences that lead to meiotic declinations during micro- and megasporogenesis (Brownfield and Köhler 2011). Consequently, polyploids are formed after the union of two gametes (2n + 2n, bilateral pathway), or the uneven ploidy level combination (2n + n, unilateral pathway), and backcrossing to a diploid to form higher ploidy levels (den Nijs and Peloquin 1977; Husband 2004). Sources of 2n gametes ought to be properly identified when utilized in breeding programs. Most of the detection procedures concentrate on pollen, since it is far-off less elaborate to screen than egg cells (Dewitte et al. 2012). Therefore, detection of 2n pollen can be accomplished using four distinctive procedures: (i) pollen grain size assessment, (ii) direct estimation of DNA content in gametes via flow cytometry, (iii) cytogenetic investigation of the microsporogenesis, and (iv) ploidy-level analysis of the offspring (Bretagnolle and Thompson 1995). Unfortunately, even though there are ample reports of 2n gametes in individual plants (ornamental, cultivars and hybrids), still assessments of the occurrence and diversity of 2n gametes in natural populations are limited (Kreiner et al. 2017).

Avena ventricosa Bal. ex Coss. is an endemic C-genome diploid species (Cyprus is the only recorded region in Europe) having plausible role in the evolution of Avena spp. polyploids via interspecific hybridization (Nikoloudakis and Katsiotis 2008; Liu et al. 2017). Nevertheless, its populations remain largely uncharacterized, eventhough it encompass a vast breeding potential due to the large genetic affinity to the cultivated hexaploid oats (Katsiotis et al. 1995) and due to its biotic stress tolerance (Loskutov 2001).

The main objective of the current study was to identify genotypes of A. ventricosa populations, which have the capacity of producing 2n gametes as a first step for utilization in breeding schemes. For that reason, we screened individuals for pollen grain size variability and used flow cytometric analysis to detect 2n gametes. After identifying 2n gamete producers, pollen mother cells (PMCs) were analyzed to possibly detect the underlined mechanism(s) resulting in their development.

Materials and methods

Plant material

Avena ventricosa seeds from four different Cypriot populations (ARI00-845, ARI00-853, ARI00-856 and ARI00-876; Table 1) were acquired from the Agricultural Research Institute of Cyprus (ARI). All seeds were germinated in 10-in. pots containing peat, in a growth room under a 16/8 light/dark cycle (200 μeinstein/m−2/s−1) at 25 °C. Plants were transferred in a glasshouse and regularly watered till anthesis.

Table 1 Avena ventricosa genotypes analyzed in the present study, coordinates of populations and rates of unstrained pollen measured
Full size table

Pollen size and viability

Pollen grains were sampled from up to ten plants per accession. Mature (but undehisced) anthers were excised from the primary floret of sampled spikelets from each plant, and pollen grains were released from the anther with a needle. Pollen stainability, as a measure of pollen viability, was assessed by staining mature pollen in a 1% aceto-carmine solution. The AxioCam MRc Zeiss system (Zeiss, Jena, Germany) was used for imaging. Regularly shaped and darkly stained pollen grains were considered to be viable. Image J (https://imagej.nih.gov/ij/) was employed to examine pollen grains’ number and size. More than 500 pollen grains were analyzed from each genotype.

Flow cytometry

The following procedures were completed at 4 °C or on ice. A two-step procedure using Otto I [0.1 M citric acid monohydrate, 0.5% (v/v) Tween 20 (Sigma-Aldrich, St. Louis, MO, USA)] and Otto II buffers [0.4 M Na2HPO4.12H2O, 0.05 mg/mL Propidium Iodide (Sigma-Aldrich, St. Louis, MO, USA), 0.05 mg/mL RNase (Takara, Otsu Shiga Japan)] was used for both pollen and leaf tissues. In order to isolate pollen grains, we placed mature anthers in Otto I buffer and vortexed them vigorously for 10 s. The homogenate was then passed through a 100 μm pre-filter to remove large contaminants, while allowing pollen grains to pass. The flow through, with the collected pollen, was loaded to a bursting filter (30 μm) placed on a 1.5 mL Eppendorf tube and pollen grains were moderately pressed against the filter for a few sec, using a rounded end rod, as previously described (Kron and Husband 2012). Nuclei were rinsed with 0.5 mL Otto I buffer.

Approximately one cm2 of excised leaf tissue was placed in a plastic 60-mm petri dish settled on ice with 0.5 mL of pre-chilled Otto I buffer. Tissues were chopped using a sterilized hand razor blade for 2–3 sec. The homogenate was further filtered through a 30 μm filter to retain debris. Homogenates (0.2 mL) were added to a labeled tube containing 1 mL of Otto II buffer, and incubated in dark for 20 min prior to analysis.

Ploidy levels were estimated using an Accuri C6 flow cytometer (Accuri Cytometers, Inc., Ann Arbor, MI, USA), following a standard protocol (Galbraith 2009). Analysis was based on light-scatter and fluorescence signals emitted from a 20-mW laser illumination at 488 nm. Precision of the instrument was validated using 6-peak fluorescent bead mixtures (Spherotech), as suggested by the manufacturer (CFlow User Guide, Accuri). Threshold levels were set (80,000 on FSC-H and 1000 FL-2) to eradicate unrelated debris from being detected. Fluidics were set on slow and measurements were collected to a total count of 10,000 nuclei. In order to discriminate among 2n gametes and aggregates (two or more adhering nuclei), 2C events with a markedly lower FL-H/FL-A comparative to that of single 2C nuclei were removed (Kreiner et al. 2017). The regions of nuclei were located diagonally using a FL3-A/FL2A plot and values representing peak positions and variances were linearly estimated across the lowest peaks in order to export. The flow histograms showed sharp peaks, with CVs typically < 5%, for both the pollen and leaf nuclei. Replicates were highly reproducible with little systematic errors.

Pollen mother cells analysis

Pollen mother cells (PMCs) were examined from florets that were excised from panicles collected at the booting stage (the flag leaf was 3–7 cm above the base of the last true leaf). Immature panicles were fixed for 48 h in Carnoy’s buffer (6:3:1, alcohol:chloroform:acetic acid v/v/v) and subsequently stored in 70% (v/v) ethanol at 4 °C. Anthers were isolated and squashed in one drop of 1% aceto-carmine solution. Chromosome configurations at meiosis were studied from individual plants (population) that contained large pollen grains.

Results and discussion

Pollen grain size variation and 2n gametes

In the present study, pollen from individuals belonging to four different A. ventricosa populations (Table 1) was analyzed in order to calculate the percentage of viability and reveal variances in grain size (possibly identify genotypes producing 2n gametes). In two of the populations (ARI100-856 and ARI100-876), the fraction of unviable pollen was rather low (about 1–2%); indicative of normal meiosis and production of functional gametes. However, higher frequencies (about 5%) of unstained pollen grains were detected in the rest. In an attempt to correlate pollen sterility to possible meiotic irregularities, the size of pollen grains from the two populations (ARI100-845 and ARI100-853) was estimated. After examination using the ImageJ software, genotypes from the ARI100-845 population were found to contain a higher than normal amount of unstained grains, while large pollen grains were additionally detected at a low frequency (Fig. 1).

Fig. 1
figure 1

a Frequency distribution of pollen size in A. ventricosa ARI100-845 population producing normal and large pollen grains. b Differences in size between normal size and large (arrow) pollen grains in A. ventricosa

Full size image

Presence of large pollen has been associated to 2n gametes in many species. This correlation is attributed to the positive relationship between DNA content and cell size or shape, which in turn affects pollen diameter and axes lengths (Katsiotis and Forsberg 1995; Dewitte et al. 2012). Usually, the presence of large pollen grains is portrayed as a bimodal distribution, opposed to a normal size distribution. However, deviation was noticed in the case of A. ventricosa ARI100-845 pollen grains (Fig. 1a), possibly due to the low frequency of 2n gametes and because size distribution of normal and large pollen may overlap. In some genera such as grasses, a broad overlap in size distribution among small and large pollen has been reported, resulting to an uncertain frequency determination of 2n pollen (Jansen and Den Nijs 1993). On the other hand, a bimodal size distribution, can be more evident in relation to pollen grains as was observed in Hibiscus (Laere et al. 2009). Still, a cut-off size value of the pollen grain can be set in order to indicate individuals that produce 2n gametes (Crespel et al. 2006). Pollen grain shape (spherical as an alternative to ellipsoidal) has also been related to ploidy levels (Ramanna et al. 2003; Akutsu et al. 2007; Dewitte et al. 2009). Indeed, large A. ventricosa pollen were found to deviate from the elliptic shape and were spherically shaped (Fig. 1b).

The occurrence of 2n-gamete rates mostly relies on the genotype of plant species, but can greatly differ within a plant species and has been found to be significantly uneven even within flowers of the same plant (Ramsey and Schemske 1998; Dewitte et al. 2009). Generally, 0.1–2.0% of gametes can be expected to be 2n in a non-hybrid plant population (Ramsey 2007). Analysis of pollen grain size in the A. ventricosa populations in the present study (Fig. 1) suggested that a small proportion of pollen remained unreduced (1.21%). Previous studies (Katsiotis and Forsberg 1995) screening Avena tetraploid species [the AABB-genome A. vaviloviana (Malz.) Mordv. PI412767] detected large pollen grains at approximately 1.0% frequency, which is in agreement with the current work. Low percentage of 2n gametes is expected in wild populations since they are not evolutionary favored. Moreover, diploid individuals that produce a great amount of 2n gametes are well decreasing their contribution to the diploid population gene pool, due to genetic irregularities and therefore are not evolutionary favored. As a result, 2n gametes are fairly infrequent in wild populations, and could be triggered only when biotic or environmental factors promote genetic instability (Kreiner et al. 2017).

Unfortunately, the detection of larger than usual pollen grains only circumstantially points to the occurrence of 2n gametes and does not definitely relates to duplicated DNA content (Dewitte et al. 2012). In order to correlate the discovery of large pollen to the presence of 2n A. ventricosa gametes, flow cytometry analysis was employed; since this is a more stringent and reliable method identifying ploidy directly by the quantification of nuclear pollen DNA content, and in parallel is a high-throughput substitute to cytological analysis (Ramsey 2007; Kreiner et al. 2017). Somatic cells have a 2C, and at a lesser extend, a 4C DNA content, while nuclei in pollen grains should only be 1C. At the present study, flow cytometry associates the DNA content of a small fraction of pollen nuclei of the A. ventricosa genotype (Fig. 2) to the DNA content of somatic leaf tissue, revealing pollen grains containing 2C nuclei. This is also observed for other grass species like Triticum aestivum L. (Pan et al. 2004).

Fig. 2
figure 2

Flow cytometric histograms of A. ventricosa ARI100-845 showing differences in holoploid genome size. Pollen grain (grey) and leaves (black) nuclei were isolated, stained with propidium iodide, and run on the flow cytometer

Full size image

Flow cytometric analyses yielded high-resolution histograms with relatively little background noise and low coefficients of variation (Fig. 2). Mean CVs were 3.68% (1C) and 3.77% (2C) for pollen nuclei, and 2.85% (2C) and 3.01% (4C) for A. ventricosa somatic nuclei samples, suggesting that the analyses were not negatively affected by secondary metabolites. Higher frequency of 2n gametes (6.53%) was detected with flow cytometry comparing to pollen grain size analysis. However, this is rather anticipated since 2n gametes are screened using pollen size as an indicator. Based on a rather limited quantity of pollen grains, 2n ‘producers’ are recognized by the presence of large pollen, whereas absence indicates non-producers (Kreiner et al. 2017). Nevertheless, there is no clear cut-off value suggestive of a ploidy threshold and a binomial grain size distribution is not always observed. Furthermore, since nuclei sample size increases (in the current study 10,000 nuclei were detected with flow cytometry), detection of rare events and their quantification is achieved in a more precise manner. Therefore, an underestimation of 2n gametes is expected when compared to flow cytometric analysis. For instance using pollen size as an indicator, Bretagnolle (2001) and Ramsey (2007) observed that 10 and 44%, respectively, of individuals studied formed 2n gametes. By contrast, Kreiner and colleagues, when they employed flow cytometry and somatic correction, found that almost 100% of the equivalent individuals had detectable 2n gametes (Kreiner et al. 2017).

Chromosome pairing and chiasmata

Trying to identify possible mechanisms for the production of 2n gametes, A. ventricosa ARI00-845 PMCs were analyzed. The result of chromosome counting showed diploid level with 2n = 14. A large number of meiotic abnormalities were revealed (Fig. 3), without however unequivocally identifying the 2n gamete cytological mechanism involved. In general, meiotic abnormalities at different stages could be summed up as follows: chromosomal inversions causing several chromosomal bridges at anaphase I, the occurrence of laggard chromosomes in anaphase I and telophase I, non-oriented bivalents at the equatorial plate, formation of micronuclei in tetrad cells and chromosomal stickiness. Surprisingly, synapsis among homologous chromosomes was low, considering the fact that A. ventricosa is a self-pollinated diploid. Furthermore, an extensive number of inversions and intercalary chiasmata among rod shaped bivalents were detected. As a result, several chromosomal fragments and chromosomal bridges were recorded. Among the bivalents formed, almost half of them were rod shaped with simply one chiasma, whereas the rest were typical ring-shaped bivalents (having two chiasmata). Larger configurations were not recorded in general, even though in some cases, chromosomal stickiness tangled all chromosomes, forming a single chromatin cluster. On average, 0.28 univalents (ranging from 0 to 2), 3.61 rod bivalents (ranging from 0 to 7) and 3.21 ring bivalents (ranging from 0 to 7) were recorded in 25 PMCs (Table 2).

Fig. 3
figure 3

PMCs of A. ventricosa ARI100-845 population. a Three chromosomal bridges among chromatids moving to opposite poles (left cell). Black arrows designate laggards and grey arrows designate chromosomal fragments. b Strand of chromatin specified by the arrow. c Unaligned chromosomes in a cell with an early constriction (arrow). d PMC in metaphase I. Chromosomes with six inversions loops and rod bivalents as preferred configuration (3II ring + 4II rod). e Intercellular chromatin migration indicated by arrow

Full size image
Table 2 Mean chromosome pairing at metaphase I of A. ventricosa ARI100-845 (range in parentheses)
Full size table

In one case, a PMC was detected baring the twofold number of chromosomes (28) at metaphase I (Fig. 4). This alone is indicative of a pre-meiotic doubling that could hence lead to the production of 2n gametes. The presence of 2n gametes in plants can occur due to meiotic deviations including abnormalities in chromosome pairing, omission of a meiotic division phase, pre-meiotic doubling, disoriented spindle formation or premature cytokinesis (Douches and Quiros 1988; Taschetto and Pagliarini 2003; Dewitte et al. 2010). From a genetic viewpoint the two main mechanisms involve either the First Division Restitution (FDR), where homologous chromosomes never synapse leading to univalent formation during meiosis-I, or the Second Division Restitution (SDR), where the sister chromatids will end-up at the same pole (Hermsen 1984). Less commonly described are indeterminate meiotic restitution (IMR) and post meiotic restitution (PMR) reported in lilly and potato, respectively (Bastiaanssen et al. 1998; Dewitte et al. 2012). Furthermore, the molecular procedures in addition to gene mutations implicated in abnormal orientation of spindles and the subsequent production of functional 2n gametes has been elucidated in Arabidopsis (D’Erfurth et al. 2008, 2009, 2010). In the case of Avena however, reports of 2n production correspond to pre-meiotic doubling and cytomixis. A specific type of cytomixis in prophase-I and metaphase-I stages leading to the movement of the entire chromosome complement has been observed in A. eriantha Dur. (Sheidai et al. 2003). Furthermore, cytomixis in PMCs of A. canariensis Baum, Rajhathy & Sampson, was also reported in wild populations (Morikawa and Leggett 1996). Finally, migration of chromosomes via intercellular channels has also been detected in Brazilian hexaploid cultivars (Baptista-giacomelli et al. 2000). Such a procedure might lead to the formation of mixoploid cells, as well as, meiocytes with higher ploidy level (2n-meiocytes).

Fig. 4
figure 4

Pollen mother cells of A. ventricosa ARI100-845 population with 28 chromosomes (a) and seven bivalents (b), during metaphase I

Full size image

In some instances, PMCs of A. ventricosa had cell wall constrictions aligned with the metaphase plate at metaphase I (Fig. 3). As a consequence, the presence of a nearly cell wall constriction at metaphase I could indicate development of 2n gametes. Such deviations in meiotic cell plate development either involve precocious initiation of cytokinesis or loss of meiotic cell plate establishment. In both scenarios, physical separation of nuclei is affected, eventually resulting to diploid or polyploid cells, and thus forming the sexual grounds for the entire genome doubling (De Storme and Mason 2014). On the other hand, cytological studies in more than a few species, have confirmed that abnormalities in one of these processes (such as deficiencies in spindle body replication and loss of essential kinetochore constituents) causes broad miscarriage of mitotic karyodivision, yielding polyploid endomitotic cells (Dikovskaya et al. 2007).

Finally, the presence of 2n gametes in A. ventricosa could be the outcome of somatic ploidy/endoreduplication. It has been shown in several plant species, that meiotic endomitosis and polyploid meiocyte establishment appears to be associated with adverse climatic circumstances. Consequently, somatic ploidy modification could act as a stress-induced mechanism, assembling adaptive chromosomal modification or polyploidization to overcome stress stimuli (Klindworth and Williams 2001; Malallah and Attia 2003). In conclusion, A. ventricosa ARI100-845 constitutes a valuable genetic resource that could facilitate the incorporation of desirable traits to cultivated oats, via interspecific hybridization and translocations.