The following species were analyzed: Arabidopsis thaliana (L.) Heynh. (2n = 10), Secale cereale L. (rye, 2n = 14), Triticum aestivum L. (wheat, 2n = 42), Aegilops speltoides ssp. aucheri (Boiss.) Chennav. (2n = 14+supernumerary B chromosomes) and Hordeum vulgare L. (barley, 2n = 14).

Flower buds and/or root tips of wheat, rye and barley were fixed for 45 min in ice-cold 4% (w/v) paraformaldehyde in 1xMTSB buffer (50 mM PIPES, 5 mM MgSO4, and 5 mM EGTA, pH 7.2). After washing in 1xMTSB, chromosome spreads were prepared by squashing. Young Ae. speltoides spikes were pretreated in ice-cold water for 24 h and then fixed in ethanol:acetic acid (3:1) for at least 4 days. Afterwards, the spikes were stained with aceto-carmine and chromosomes were prepared by squashing in 45% acetic acid.

Tissue sections of Ae. speltoides were prepared according to Steedman (1957) and Braszewska-Zalewska et al. (2013). Briefly, developing seeds were excised from spikelets and fixed in freshly prepared 3% (w/v) paraformaldehyde with 0.05% Triton X100 in 1 × PBS buffer on ice for 5 h. Then, dehydration was performed in an ethanol series (ethanol/1 × PBS buffer) from 30 to 96% for 30 min in each at room temperature and in 96% for 30 min at 37°C. Afterwards, the tissues were infiltrated with PEG1500-wax and embedded in a small casting mold (1–3 seeds per block). The blocks were cut into 10 μm slices using a Leica microtome (RM2265; knifes 35N from Feather company). The slices were transferred onto poly-L-lysine-coated slides using forceps and a brush and were stretched by adding a drop of 1 μl water over each slice. The rest of PEG-wax was removed from the dry slides by washing them in 90% ethanol.

Differentiated 2–16C leaf nuclei of A. thaliana were isolated and flow sorted according to their DNA content from differentiated rosette leaves after formaldehyde fixation using a FACS Aria (BD Biosciences) as described (Pecinka et al., 2004).

To evaluate the substructure of CENH3 containing chromatin, immunostaining was performed according to Jasencakova et al. (2000). CENH3 was detected with rabbit anti-grass CENH3 primary antibodies (Sanei et al., 2011) and goat anti-rabbit rhodamine (1:300; Jackson Immuno Research Laboratories) or goat anti-rabbit Alexa488 secondary antibodies (1:200; Molecular Probes). Spindle fibers were labeled with monoclonal mouse anti α-tubulin (1:200; clone DM 1A, Sigma) and anti-mouse Alexa488 (1:400; Molecular Probes) antibodies.

For FISH the 180-bp centromeric repeat sequence pAL of A. thaliana (Martinez-Zapater et al., 1986) was generated by PCR as described (Kawabe and Nasuda, 2005). The centromeric retrotransposon CRW2 of wheat was generated by PCR as described by Li et al. (2013). These probes as well as the subtelomeric repeat HvT01 (Schubert et al., 1998) and BAC7 containing centromere-specific repeats of barley (Hudakova et al., 2001; Houben et al., 2007) were directly labeled by nick translation with TexasRed-dUTP, Alexa488-dUTP and Cy3-dUTP according to Ward (2002). FISH was performed according to Schubert et al. (2001).

For the colocalization of CENH3 immunosignals with centromeric FISH signals, immunostaining was performed first. The slides were treated with 10 mM citrate buffer (pH 6) in a microwave at 800 Watt for 60 s according to Chelysheva et al. (2010). Then the primary antibodies were applied and immunostaining was performed as described (Jasencakova et al., 2000). Prior FISH the slides were treated with ethanol:acetic acid (3:1) fixative for 10 min and freshly prepared 4% formaldehyde in 1 × PBS for 10 min, followed by three times washing for 5 min in 1 × PBS. These steps are important to stabilize the immunosignals during the following FISH procedure, which was performed as described (Ma et al., 2010). Nuclei and chromosomes were counterstained with DAPI (1 μg/ml) in Vectashield (Vector Laboratories).

To analyse the substructure of chromatin beyond the classical Abbe/Raleigh limit (super-resolution) spatial Structured Illumination Microscopy (3D-SIM) was applied using a C-Apo 63 × /1.2 W Korr objective of an Elyra PS.1 microscope system and the software ZEN (Carl Zeiss GmbH). Images were captured using 405, 488, and 561 nm laser lines for excitation (42, 34, and 28 μm grids for 561, 488, and 405 nm excitations; 5 rotations) and the appropriate emission filters. 3D-SIM stacks with a step size of 110 nm were acquired consecutively for each fluorophore starting with the highest wavelength dye to minimize bleaching. SIM image stacks were used to produce 3D movies by the Imaris 8.0 (Bitplane) and ZEN 2012 software.

To test whether there is a varying centromeric chromatin ring formation in interphase nuclei of different endopolyploidy, flow sorted 2–16C differentiated leaf nuclei of A. thaliana were analyzed using the centromere-specific DNA repeat pAL as FISH probe. In addition to globular structures, also ring and half-ring structures were identified at all ploidy levels. By SIM it became obvious that these centromere structures are composed of a network of chromatin loops (Figure 1). Ring and half-ring formation appears around chromocenters and especially after centromere association, meaning that centromeres tend to fuse in interphase nuclei. In 16C nuclei due to decreased cohesion aligned centromeric chromatid regions start to separate (Schubert et al., 2006). Then, more than 10 centromere signals (corresponding to the chromosome number in A. thaliana) may appear. These centromere regions keep the globular substructure, but no further ring formation was observed (Figure 1).

To investigate the ultrastructure of active centromeres during cell division, rye centromeres were labeled with CENH3-specific antibodies and analyzed by SIM. During mitosis and meiosis as well as in interphase nuclei of roots and anthers CENH3 containing chromatin domains form ring-like structures measuring ~0.5–1.0 μm (Figure 2; Supplementary Movies 1–12). When these rings belonging to single or paired homologs, comprising the centromeres of either two or four sister chromatids, associate or align (e.g., in interphase and prophase I, respectively), they fuse and form bigger rings. During late somatic and meiotic prophase these rings split again and compose smaller ones clearly visible at somatic metaphase chromosomes and at bivalents. Already in metaphase I some of the rings of the sister centromeres split again and become then distantly separated in interkinesis. Similarly, the formation of ring-like structures by CENH3 containing nucleosomes was observed during meiosis of hexaploid wheat (Figure 3).

Anti-CENH3 and the centromere-specific repeat CRW2 (Liu et al., 2008) were used to analyse the centromere substructures of Ae. speltoides. In most embryonic interphase nuclei the CENH3-labeling shows clearly a ring formation by looped chromatin fibers (Figures 4A_{1, 2}) but not at metaphase centromeres (Figure 4A₃). Irrespective of the presence of supernumerary B chromosomes all chromosomes showed the same centromere chromatin substructure, although we could not identify the Bs based on B-specific probes (Figure 4A₃; 2n = 14 + 2B).

In spike meristems both CENH3 and CRW2 form spherical reticulate substructures during the somatic cell cycle which intermingle among each other. But no ring formation was observed (Figures 4B_1-4). The differently labeled, but with CENH3 chromatin intermingled CRW2 repeat fibers indicate that most of CRW2 is not associated with CENH3 (Figures 4A_1, B_1-4). The supernumerary chromosomes in pro-metaphase (Figure 4B₂; 2n = 14+1B) and metaphase (Figure 4B₃; 2n = 14+2B) obviously do not show a deviating centromere structure.

Compared to embryo nuclei, endosperm nuclei exhibit a more compact CENH3 chromatin organization. Thus, a ring formation is less often visible (Figure 4C).

We conclude that the centromere chromatin organization may differ between tissues of individual Ae. speltoides plants.

CENH3 antibodies and the centromere-specific repeat containing BAC7 probe of barley (Hudakova et al., 2001; Houben et al., 2007) were applied to label the centromere substructures of barley. For comparison, also the subtelomeric repeat HvT01 (Schubert et al., 1998) was applied as a FISH probe.

In interphase nuclei both CENH3-positive chromatin and centromeric repeats may establish ring structures (Figures 5A_1-4). CENH3 chromatin is embedded in centromeric repeats as a condensed globular structure or CENH3-ring-in-centromere repeat-ring configurations may appear. During the somatic cell cycle the centromeric repeats compose reticulate substructure. This is true also for the subtelomeric repeat HvT01, but it may also compose ring chromatin structures (Figures 5B_1-3). The observation that in identical nuclei differently shaped chromatin substructures occur, excludes that preparation artifacts inducing the ring structure formation arose.

In short, we conclude that the formation of chromatin rings is not a specific feature of centromeres.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The reviewer ML and handling Editor declared their shared affiliation, and the handling Editor states that the process nevertheless met the standards of a fair and objective review.