Non-Rabl chromosome organization in endoreduplicated nuclei of barley embryo and endosperm tissues

Abstract

Rabl organization is a type of interphase chromosome arrangement with centromeres and telomeres clustering at opposite nuclear poles. Here, we analyzed nuclear morphology and chromosome organization in cycling and endoreduplicated nuclei isolated from embryo and endosperm tissues of developing barley seeds. We show that endoreduplicated nuclei have an irregular shape, less sister chromatid cohesion at 5S rDNA loci, and a reduced amount of centromeric histone CENH3. While the chromosomes of the embryo and endosperm nuclei are initially organized in Rabl configuration, the centromeres and telomeres are intermingled within the nuclear space in the endoreduplicated nuclei with an increasing endoreduplication level. Such a loss of chromosome organization suggests that Rabl configuration is introduced and further reinforced by mitotic divisions in barley cell nuclei in a tissue- and seed age-dependent manner.

Barley, CENH3, CEREBA repeat, embryo, endoreduplication, endosperm, Hordeum vulgare, nuclear organization, Rabl configuration

Introduction

Chromosome structure and organization play important roles in replication, transcription, and genome repair (Misteli, 2020). Their organization includes the formation of nucleosomes, as the basic unit of chromatin, and their assembly into higher order domains. These domains represent different chromatin states characterized by specific histone and/or DNA modifications, and vary in their transcription, replication, or DNA repair patterns (Roudier et al., 2011; Sequeira-Mendes et al., 2014). Individual interphase chromosomes occupy a specific nuclear space known as chromosome territories (CTs) (reviewed in, for example, Schubert and Shaw, 2011; Grob, 2020).

In Arabidopsis thaliana (Arabidopsis) with a small and repeat-poor genome, the peri(centromeric) regions form ­heterochromatic chromocenters from which euchromatic chromosome arms emanate and the telomeric regions often surround the nucleolus (Fransz et al., 2002, 2003). The CTs of 45S rDNA-bearing chromosomes cluster preferentially around the nucleolus, but the CTs of the remaining chromosomes are positioned randomly in roots and leaves (Pecinka et al., 2004) or show pairwise CT associations in the seed endosperm (Baroux et al., 2017). In contrast, plants with large and repeat-rich genomes often harbor Rabl chromosome organization, with the centromeres and telomeres clustering at the opposite nuclear poles (Santos and Shaw, 2004), first described in 1885 by the cytologist and anatomist Carl Rabl based on nuclei of Caudata (Rabl, 1885). Because Rabl organization is widespread in cereals with large genomes such as bread wheat (17 Gbp/1C) or barley (5.1 Gbp/1C), and diminishes with a decreasing genome size in, for example, maize (2.4 Gbp/1C) or rice (0.43 Gbp/1C) (Fujimoto et al., 2005a), it has been hypothesized that it is determined by the nuclear genome size (reviewed in, for example, Santos and Shaw, 2004). Although this correlation holds true over distantly related phylogenetic groups such as Poaceae and Brassicaceae (Němečková et al., 2020; Shan et al., 2021), it is not universal. For example, the majority of root nuclei (but not leaf nuclei) show Rabl in the small genome grass Brachypodium distachyon (0.35 Gbp/1C), while some Liliaceae species with giant genomes (~35–50 Gbp/1C) lack Rabl organization (Fujimoto et al., 2005a; Idziak et al., 2015). Therefore, other hypotheses suggest that Rabl might be a preserved organization of mitotic chromosomes. However, why this configuration is maintained in some but not other species remains unclear.

To investigate the relationships between occurrence of the Rabl configuration, endoreduplication, and tissue age, we made use of the endosperm and to a smaller extent embryo tissues of barley grains (Nowicka et al., 2021a). In cereals, the endosperm progresses through several stages of development (reviewed in, for example, Olsen, 2001; Sabelli and Larkins, 2009), connected with major changes in transcriptional regulation (Sreenivasulu et al., 2004; Pfeifer et al., 2014). The endosperm forms most of the seed mass and serves as the main energy storage tissue for the embryo during germination. In contrast to vegetative tissues, cereal endosperm shows endoreduplication, a process during which the genome duplicates without mitosis (Sabelli and Larkins, 2009).

Endoreduplication in cereal grains does not contribute to the seed size (Nowicka et al., 2021a), but a study in Arabidopsis suggests that it might be linked with high metabolic activity of certain cells (Baroux et al., 2004). Endoreduplicated nuclei allow the study of whether chromosome organization patterns are correlated with ploidy. In Arabidopsis, endoreduplicated nuclei from leaves showed loss of positional sister chromatid (SC) cohesion and had generally reduced heterochromatin compaction (Schubert et al., 2006). In rice, endoreduplication forced Rabl chromosome organization in xylem vessel cells (Prieto et al., 2004). Finally, conserved Rabl organization was observed in endoreduplicated nuclei of bread wheat embryo and endosperm tissues (Wegel and Shaw, 2005).

Here, we studied chromosome organization in mitotically cycling and endoreduplicated nuclei isolated from embryo and endosperm tissues of developing barley grains. Our data suggest that while the chromosomes of embryo nuclei are organized mostly according to a Rabl pattern, the nuclei of endosperm tissues adopt a non-Rabl organization with increasing C-value and number of days after pollination. Collectively, our study provides comprehensive characteristics of embryo and endosperm nuclear processes during the key stages of barley grain development.

Materials and methods

Plant materials and growth conditions

The spring cultivar (cv.) Compana (PI 539111, NSGC of the USDA-ARS, USA) of cultivated barley (Hordeum vulgare subsp. vulgare) was used in this study. For germination, grains were evenly spread in a Petri dish on filter paper soaked with distilled water, stratified at 4 °C in the dark for 48 h, and germinated in the dark at 25 °C for 3 d. Sprouting seedlings were planted into 5 × 5×5 cm peat pots with a mixture of soil and sand (2:1, v/v) and grown in a phytochamber under a long-day regime (16 h daylight with an intensity of 200 μmol m⁻² s⁻¹ and temperature 20 °C; 8 h night with 16 °C; 60% humidity). After 2 weeks, seedlings were transferred into 15 × 15 × 15 cm pots and grown under the same conditions until flowering. The day of pollination was monitored using the morphology of the stigma and anthers according to the Waddington scale (W10) (Waddington et al., 1983) as we established previously (Kovacik et al., 2020; Nowicka et al., 2021b). Seeds were collected from the center of the spike at 4, 8, 16, 24, and 32 days after pollination (DAP).

Isolation of nuclei and flow sorting

Nuclei were isolated from root apical meristem (RAM), embryo, and endosperm tissues. For isolation of root nuclei: 70 roots of seedlings at 2 d after germination were cut ~1 cm from the apex and collected in a drop of distilled water. Next, roots were drained on a cellulose tissue paper, rinsed in 10 mM Tris buffer pH 7.0, fixed with 2% (v/v) formaldehyde/Tris buffer for 20 min on ice, and washed three times for 5 min each with Tris buffer also on ice. Root apices were excised ~1 mm from the tip and homogenized in 500 µl of LB01 buffer (15 mM Tris–HCl pH 7.5, 2 mM NaEDTA, 0.5 mM spermine, 80 mM KCl, 20 mM NaCl, and 0.1% Triton X-100) for 13 s at 15 000 rpm using a Polytron PT1300D homogenizer (Kinematica AG).

For isolation of embryo nuclei, ~100 embryos were manually dissected from seeds at 8 DAP, using an SZX16 binocular microscope (Olympus). First, seeds were peeled (manual removal of hulls) and placed on a Petri dish in a drop of 1× phosphate-buffered saline (PBS) (Kovacik et al., 2020). Dissected embryos were collected into a 1.5 ml Eppendorf tube containing 200 µl of 1× PBS and kept on ice until the sampling was finished. Tubes were low-speed centrifuged (1 min, 2000 g), PBS was removed, embryos were rinsed in Tris buffer, fixed with 2% (v/v) formaldehyde/Tris buffer for 20 min on ice, and washed three times for 5 min with Tris buffer also on ice. Embryos were homogenized with a pellet pestle in the same Eppendorf tubes with 500 µl of LB01 buffer. For 16 DAP and older seeds, at least 50 embryos were manually dissected using a binocular microscope. Embryos were collected into a small beaker containing PBS and kept on ice. Subsequently, the material was rinsed in Tris buffer, pre-fixed with 4% (v/v) formaldehyde/Tris buffer for 30 min by vacuum infiltration on ice, followed by fixation with the same solution for 30–40 min without a vacuum also on ice, and washed in Tris buffer. Samples were chopped in 2 ml of LB01 buffer with a razor blade on a Petri dish.

For isolation of endosperm nuclei, ~80 of the whole peeled seeds from 4 and 8 DAP endosperm samples were gathered into a small beaker kept on ice. Seeds were rinsed in Tris buffer, pre-fixed with 4% (v/v) formaldehyde/Tris buffer for 20 min on ice, fixed without vacuum for 40 min, and washed. Samples were chopped with a razor blade in 2–3 ml of LB01 buffer on a Petri dish. For endosperm samples at ≥16 DAP, ~60 whole peeled seeds with embryos removed were collected into a small beaker, cut with the razor blade into 1 mm thick transversal slides, pre-fixed for 30 min, then fixed for 40 min. Samples were chopped with a razor blade in 3–4 ml of LB01 buffer on a Petri dish.

For flow sorting of nuclei, the crude homogenates of all samples were double-filtered first through a 50 µm and then a 20 µm pore size mesh. Nuclei suspensions were stained with 2 µg ml⁻¹ DAPI. Approximately 500 nuclei for each C-value were flow-sorted into a 2 µl drop of sorting buffer (100 mM Tris–HCl pH 7.5, 50 mM KCl, 2 mM MgCl₂, 0.05% Tween-20, and 5% sucrose) on poly-lysine (Menzel Gläser, J2800AMNZ) or superfrost Plus (Menzel Gläser, J1810AMNZ) microscopic slides using a FACSAria II SORP flow cytometer and sorter (BD Biosciences, Santa Clara, CA, USA). Slides were air-dried for 1 h at room temperature and stored at –20 °C until use.

Mitotic chromosome preparations

Metaphase chromosomes were prepared from synchronized root tips (Lysák et al., 1999). Briefly, germinated seedlings were transferred into Hoagland’s nutrient solution containing 2 mM hydroxyurea for 18 h. Then the roots were washed in distilled water and cultured in hydroxyurea-free Hoagland’s solution for 5.5 h. To accumulate cells at metaphase, the roots were treated for 2 h with Hoagland’s solution containing 2.5 μM amiprophos-methyl (A0185, Duchefa Biochemie). Subsequently, the root tips were fixed in ice-cold 90% acetic acid for 10 min followed by three washes in 70% ethanol and stored in 70% ethanol at −20 °C. Chromosomes were prepared using the drop technique (Danilova et al., 2012). In brief, maceration of root tips was performed at 37 °C for 57 min using an enzyme mixture consisting of 4% (w/v) cellulase Onozuka R-10 (Yakult Pharmaceutical Industry, 150422-01) and 1% (w/v) pectolyase Y23 (Duchefa, 9033-35-6) in KCl buffer (75 mM KCl, 7.5 mM EDTA, pH 4). The quality of chromosome spreads was evaluated using a phase-contrast microscope (Primo Star, Zeiss), and the slides with at least five metaphases were used for fluorescence in situ hybridization (FISH). Before a hybridization step, the slides were pre-treated with pepsin (10 µg ml^–1 in 10 mM HCl) at 37 °C for 10 min, then rinsed in 2× SSC followed by RNase A treatment (described below).

Fluorescence in situ hybridization

Three combinations of the following probes were used in the double- or triple-color FISH experiments. For detecting barley centromeres, a synthetic 28-mer oligonucleotide (5ʹ-AGGGAGA-3ʹ)₄ probe labeled at the 5ʹ end with Cy3 or Cy5 (Eurofins) was used. The probe targets a centromeric retroelement-like element CEREBA conserved among cereal centromeres (Hudakova et al., 2001). For detecting the telomeres, we used a synthetic 28-mer oligonucleotide probe (5ʹ-CCCTAAA-3ʹ)₄ corresponding to the Arabidopsis-type telomeric repeat and labeled at the 5ʹ end with Cy3 or Cy5 (Eurofins). The 5S rDNA probe was amplified from cv. Compana genomic DNA with the primers 5ʹ-GGATGCGATCATACCAGCAC-3ʹ and 5ʹ-GACATGCAACTATCTATTTGT-3ʹ using biotin-dUTP or digoxigenin-11-dUTP (both Roche, 11093070910 and 11093088910) during PCR. The 45S rDNA probe was labeled with biotin-dUTP or digoxigenin-11-dUTP from the pTa71 plasmid containing a 9.1 kb fragment of rDNA sequence from bread wheat (Gerlach and Bedbrook, 1979) using nick translation kits (both Roche, 11745824910 and 11745816910) according to the manufacturer’s instructions.

FISH was performed as described (Karafiátová et al., 2013; Nowicka et al., 2020), with the following modifications. Preparations were air-dried at room temperature, rinsed in 2× SSC, treated with RNase A (50 µg ml^–1 in 2× SSC; Thermo Fisher, EN0601) for 30 min at 37 °C, and washed with 2× SSC and 1× PBS. Subsequently, slides were post-fixed with 4% formaldehyde/1 ×PBS for 15 min and washed with 1× PBS. A hybridization mixture contained a cocktail of two or three probes, each with a final concentration of 400 ng µl^–1, and 1 µg of sheared salmon sperm DNA (Invitrogen, AM9680), 50% (v/v) deionized formamide, 10% (v/v) dextran sulfate, and 2× SSC. For biotin-dUTP- and digoxigenin-11-dUTP-labeled probes, the hybridization mixture was heated for 4 min at 95 °C, cooled on ice, and denatured again together with target DNA on slides for 4 min at 80 °C. For oligo-probes, the step of hybridization mixture pre-denaturation was skipped. Biotin-dUTP was detected either by (i) streptavidin-Cy3 (1:500, Molecular Probes, SA1010) or (ii) goat anti-avidin conjugated with biotin (1:100, Vector Laboratories, NC9256157) followed by avidin conjugated with Texas Red (1:1000, Vector Laboratories, NC9172942). Digoxigenin-dUTP was detected either with (i) an anti-digoxigenin antibody conjugated with fluorescein isothiocyanate (FITC, 1:200 Roche, 11207741910) or (ii) a mouse anti-digoxigenin antibody (1:250, Roche, 11333062910) followed by application of a goat anti-mouse antibody conjugated with Alexa Fluor 488 (1:200, Molecular Probes, A32723). The preparations were counterstained with DAPI in Vectashield (Vector Laboratories, H-1200-10).

ImmunoFISH

Slides were air-dried at room temperature, post-fixed with 4% formaldehyde/1× PBS for 15 min, and washed with 1× PBS. Immunostaining was carried out as described (Jasencakova et al., 2000) with minor modifications. In brief, preparations were incubated with the rabbit anti-barley-αCENH3-specific primary antibody (1:200; Sanei et al., 2011) at 4 °C overnight and the secondary goat anti-rabbit-Alexa Fluor 488 (1:300, Molecular Probes, A11008) at 37 °C for 90 min. Before FISH, slides were fixed in 3:1 ethanol/acetic acid for 10 min, followed by 10 min fixation with 3.7% formaldehyde/1× PBS. Slides were washed with 1× PBS. FISH steps were performed as described above, excluding pepsin and RNase A treatment.

Microscopy

The images were taken with an AxioImager Z2 upright microscope (Zeiss, Oberkochen, Germany) equipped with a pE-4000 LED illuminator light source (CoolLED, Andover, UK), motorized four-position excitation filter wheel, laser-free confocal spinning disk device (DSD2, Andor, Belfast, UK), and a ×100/1.4 NA Oil M27 Plan-Apochromat (Zeiss) objective. Image stacks of 40–80 slides depending on the C-value of the nucleus, on average, with a 0.2 µm z-step were acquired separately for each fluorochrome using the appropriate excitation [DAPI λ=390/40 nm, green fluorescent protein (GFP) λ=482/18 nm, red fluorescent protein (RFP) λ=561/14 nm, Cy5=640/14 nm] and emission (DAPI λ=452/45 nm, GFP λ=525/45 nm, RFP λ=609/54 nm, Cy5=676/29 nm) filters. For fluorescence detection, the 4.2 megapixel sCMOS camera (Zyla 4.2) was used and the iQ 3.6.1 acquisition software (both Andor) was used to drive the microscope.

Image analysis

The images were converted into .ims format with Imaris File Converter 9.2.1 (Bitplane, Zurich, Switzerland) and exported as maximum intensity projection (mip) tif files with Imaris 9.7 software (Bitplane). For ­visualization of the surface and shape of the nuclei, the Imaris 9.7 function ‘Surface’ was used for rendering the DAPI-stained nucleus surface and to obtain 3D nucleus images. Then, functions ‘Slide viewer’ and ‘Section view’ were applied to visualize inside the nucleus. For determination of the nucleus area, perimeter, and circularity, the nucleus area (NA) and perimeter (NP) of the X–Y middle slide view tif images were measured in FIJI (ImageJ2; https://imagej.net/software/fiji/) calibrated with an internal size control. The nucleus circularity index (NCI) was calculated according to the formula: NCI=4π×NA/(NP)² (Ankam et al., 2018).

To construct the karyotype, chromosomes were paired based on the chromosomal position of rDNAs and CEREBA. The karyotype was prepared in Adobe Photoshop 6.0 (Adobe Systems Corporation, San Jose, CA, USA). Individual chromosomes were classified according to Fukui et al. (1994) and Kapusi et al. (2012). For FISH signal scoring, the number of FISH signals per nucleus was quantified in FIJI with the ‘Multipoint’ tool using mip tif images. Quantitative analysis of CENH3 co-localization with CEREBA was performed in FIJI calibrated with an internal size control using fluorescent intensity ‘Plot Profile’ for both correlated signals.

Total fluorescence intensity measurements of all CENH3 and CEREBA per nucleus were done in FIJI using mip tif images. For each nucleus, 2–10 regions of interest (ROIs) defined manually with a constant size of 3.5 × 3.5 µm were evaluated. For green and red channels, the same ROIs were analyzed, and for each of them the fluorescence intensity ratio of CENH3/CEREBA was calculated. For the DAPI channel (3.5 × 3.5 µm), ROIs located in the middle of the nucleus were evaluated.

For Rabl configuration analysis, Imaris applications ‘Surface’ and ‘Spot detection’ were used for rendering the nucleus surface and modeling the centromere/telomere arrangements, respectively. The space in the nucleus occupied by centromeres and telomeres was measured using Imaris ‘Measurement point’ in polygon mode. The detailed Imaris-based image analysis workflow is described separately (Randall et al., 2022). The Imaris statistic output files reporting on the distance between centromeres and telomeres were exported for each nucleus separately.

Image data normalization and statistical analysis

Scored numbers of FISH signals were normalized to the number of signals per nucleus at the G₁ phase (Supplementary Fig. S1). Hence, for the number of 5S major rDNA loci, scoring values were divided by two or by three for embryo and endosperm data, respectively. In the case of 5S minor and 45S rDNA loci, raw data were divided by four and six for embryo and endosperm, respectively. For CEREBA, data were normalized to 14 and 21 for embryo and endosperm, respectively. Distances between centromeres and telomeres were expressed as a ratio to the nucleus diameter (ND).

All scoring and measurement, raw and normalized data were tested for Gaussian distribution. To return to Gaussian distribution, data expressed as percentages were first arcsine transformed. Next, relevant comparisons were carried out by two-way ANOVA (factor 1=C-value, factor 2=DAP) and post-hoc Duncan’s multiple ranges (P≤0.05) test. To evaluate the statistical differences between embryo and RAM samples, one-way ANOVA (factor 1 tissue) was applied. Pearson’s correlation coefficient analysis was used to visualize relationships among the measured and evaluated parameters (NCI, NA, NP, and co-localization between CENH3/CEREBA). Statistical analyses were performed in Statistica v. 12 (Statsoft Inc.) or Minitab v. 18 (Minitab). Boxplots were drawn using the ggplot GUI online tool (https://shiny.gmw.rug.nl/ggplotgui/).

Results

Endoreduplication affects the morphology of barley embryo and endosperm nuclei

We isolated G₁ (2C/3C), G₂ (4C/6C), and endoreduplicated (8C/≥12C) nuclei from barley embryo and endosperm tissues (embryo/endosperm C- values, respectively) at 4, 8, 16, 24, and 32 DAP as described (Nowicka et al., 2021a), and analyzed their morphology. Nuclei from highly dividing RAM tissues were used as somatic control. Based on our initial assessment, we noted differences in nuclear shape and therefore calculated the nucleus circularity index (NCI) (Ankam et al., 2018) using nucleus area (NA) and nucleus perimeter (NP) values (Materials and methods; Fig. 1A, B; Supplementary Figs S2, S3).

Fig. 1.

Endoreduplicated barley grain nuclei have altered shape. (A) Representative DAPI-stained endosperm nuclei of different C-values collected at 8 and 24 days after pollination (DAP). The upper panels show 3D maximum intensity projections (mip) and the lower panels their 2D X–Y middle slide view (msv). Scale bars=10 µm. (B) Boxplots showing the nucleus circularity index (NCI) for nuclei of different tissues, C-values, and DAP. Root apical meristem (RAM) nuclei were used as the vegetative tissue control. NCI was calculated using the following formula NCI=4π ×NA/(NP)² (Ankam et al., 2018). Original data for NA and NP are depicted in Supplementary Fig. S2. The lower and upper hinges of the boxplots correspond to the first and third quartiles of the data, and the black lines within the boxes mark the median. Whiskers mark 10% and 90% intervals. A total of 75 nuclei were measured in three microscopic slides. Black squares represent outliers. Different letters indicate significant differences (P≤0.05, two-way ANOVA, factor 1=C-value, factor 2=DAP, followed by Duncan post-hoc test). The summary of ANOVA is presented in Supplementary Fig. S2D. Statistical significance between embryo and RAM samples was evaluated with one-way ANOVA, ns, not significant (P_2C=0.453, P_4C=0.101). (C) Example images of 24 DAP 3C and 24C DAPI-stained endosperm nuclei presented in 3D mip (left panel) and their surface reconstructions using Imaris software (right panels). Insets display in more detail the absence (3C) and presence (24C) of nuclear grooves. Scale bars=10 µm (main) and 2 µm (insets). Additional images are presented in Supplementary Fig. S4.

The 2C and 4C embryo nuclei had a nearly ideal circular shape (NCI ≥0.91) during the whole of seed development, but the NCI of endoreduplicated (8C) nuclei from 24 and 32 DAP was significantly reduced to ~0.75 (two-way ANOVA; Fig. 1B; Supplementary Fig. S2D). Hence, the circularity of the embryo nuclei depended on the degree of endoreduplication but not the number of DAP. In contrast, NCI of endosperm nuclei was influenced not only by the C-value but also by DAP. The NCI of 3C and 6C nuclei was ~0.89 from 4 to 16 DAP, then reached its maximum of ~0.95 at 24 DAP and decreased to 0.82–0.88 at 32 DAP. Endoreduplicated endosperm nuclei exhibited an ellipsoid shape with NCIs between 0.75 (12C at 8 DAP) and 0.67 (24C at 24 DAP), and the ellipticity increased during seed maturation and desiccation (Fig. 1A, B; Supplementary Fig. S2D). Interestingly, rendering of the surface of endoreduplicated endosperm nuclei revealed grooves of variable dimensions that were not observed in the nuclei with lower C-values (Fig. 1C; Supplementary Fig. S4).

Loss of sister chromatid cohesion at 5S rDNA loci of endoreduplicated nuclei

The altered morphology of barley endoreduplicated seed nuclei stimulated us to also explore the chromosome organization. As detection of single-copy sequences in barley interphase nuclei is not possible using current cytology tools, we focused on the arrangement of major tandem repetitive regions 5S rDNA, 45S rDNA, and CEREBA centromeric repeats using FISH (Fig. 2A, B).

Fig. 2.

Sister chromatid cohesion at 5S and 45S rDNA loci in barley seed nuclei. (A) Karyotype of cv. Compana showing localization of 5S rDNA, 45S rDNA, and CEREBA centromeric repeats on metaphase chromosomes by FISH. DNA was counterstained with DAPI. Arrows indicate 5S minor rDNA loci. Scale bar=10 µm. (B) Ideogram of cv. Compana (based on A). (C) Representative endosperm nuclei with different C-values collected at 8 and 24 DAP after FISH with 5S (orange) and 45S (violet) rDNA probes. The larger and brighter 5S rDNA signals correspond to the 5S major loci. DNA was stained with DAPI (gray). Scale bars=10 µm. (D, E) Boxplots showing the number of (D) 5S major and (E) 45S rDNA FISH signals per nucleus for different tissues, C-values, and DAP. Root apical meristem (RAM) nuclei were used as the vegetative tissue control. The lower and upper hinges of the boxplots correspond to the first and third quartiles of the data, and the black lines within the boxes mark the median. Whiskers mark 10% and 90% intervals. At least 75 nuclei from three microscopic slides were scored. Black squares represent outliers beyond the whiskers. Different letters indicate significant differences (P≤0.05, two-way ANOVA, factor 1=C-value, factor 2=DAP, followed by Duncan test). No significant differences were found for the 45S rDNA (P>0.05). The summary of ANOVA is presented in Supplementary Figs S5B and S7A. Statistical significance between the embryo and RAM samples was evaluated with one-way ANOVA, *** significant at P≤0.001, ns, not significant (5S P_2C=0.771). Normalized data (Supplementary Fig. S1) for 5S major and 45S rDNA signals are provided in Supplementary Figs S5C and S7B, respectively.

The genome of cv. Compana contains a cytologically distinguishable single large 5S rDNA cluster at the bottom arm of chromosome 2H (hereafter 5S major) and two smaller clusters at the bottom arms of chromosomes 3H and 7H (hereafter 5S minor; Fig. 2A–C). In G₁ (2C) and G₂ (4C) nuclei of the embryo samples up to 16 DAP and root samples, we observed two separate 5S major rDNA signals, which suggests SC cohesion during the G₂ phase and separation of the two homologous chromosomes in both G₁ and G₂. However, in embryo nuclei from older (24 and 32 DAP) seeds, 4C nuclei contained mostly three to four 5S major FISH signals and 8C nuclei contained five to seven such signals, indicating a reduced SC cohesion at 5S major (Fig. 2D; Supplementary Fig. S5A). G₁ (3C) endosperm nuclei showed mostly three 5S rDNA major FISH signals, but G₂ (6C) and once-endoreduplicated (12C) nuclei displayed SC separation progressing in a DAP-dependent manner. Finally, the majority of twice-endoreduplicated ­endosperm (24C) ­nuclei had 21 FISH signals at 24 DAP, corresponding to full SC separation at 5S rDNA major (Fig. 2C, D; Supplementary Fig. S5). Statistical analysis by ANOVA showed that the 5S rDNA major SC separation in embryo and endosperm significantly increased with rising C-values, developmental progression (DAP), and as a result of an interaction between both factors (Supplementary Fig. S5). The same trend was observed for 5S rDNA minor loci (Supplementary Fig. S6).

Analogous analysis of the 45S rDNA, located on barley chromosomes 5H and 6H (Kapusi et al., 2012), revealed a lower than expected number of FISH signals. This suggests a persistent SC cohesion and tendency toward 45S rDNA clustering (Fig. 2C, E; Supplementary Figs S5A, S7). Neither C-value nor DAP affected this pattern (two-way ANOVA, P>0.05, no significant differences for single factors and their interaction, Supplementary Fig. S7). Hence, the organization of 45S rDNA loci remained relatively intact, suggesting a locus-specific control of SC alignment in endoreduplicated barley seed nuclei.

Decondensation of CEREBA and reduction of CENH3 in endoreduplicated nuclei

After FISH with CEREBA repeats, we observed on average 12 signals at 8 DAP and 11 signals at 32 DAP in 2C and 4C embryo nuclei (Fig. 3; Supplementary Fig. S8A). The average number of signals increased significantly to ~14–17 after endoreduplication (8C). In 3C and 6C endosperm nuclei, the average number of CEREBA foci ranged from nine to 17 at different DAPs. In 12C and 24C nuclei, the CEREBA signals appeared less compact, often splitting into several smaller foci, and this trend was more pronounced with increasing DAP (Fig. 3B, C). Two-factor ANOVA revealed an additive effect of DAP in combination with C-value on the number of CEREBA foci in both embryo and endosperm nuclei (Supplementary Fig. S8B, C). The high number of CEREBA signals most probably indicates relaxation of centromeric repeats or a larger distance between individual sister chromatids at the centromeric region.

Fig. 3.

CEREBA organization in barley seed nuclei. (A) Representative endosperm nuclei of different C-values collected at 8 and 24 DAP after FISH with CEREBA centromeric repeat (red). DNA was stained with DAPI (gray). Insets are enlarged in (B). Scale bars=10 µm. (B) Insets (1–7) marked in (A) show the variable size of CEREBA FISH signals in the 6C–24C endosperm nuclei. Scale bar=2 µm. (C) Boxplots showing the number of CEREBA-FISH signals per nucleus for different tissues, C-values, and DAP. Root apical meristem (RAM) was used as the vegetative tissue control. The lower and upper hinges of the boxplots correspond to the first and third quartiles of the data, and the black lines within the boxes mark the median. Whiskers mark 10% and 90% intervals. At least 75 nuclei from three microscopic slides were evaluated. Black squares represent outliers beyond the whiskers. Different letters indicate significant differences (P≤0.05, two-way ANOVA, factor 1=C-value, factor 2=DAP, followed by Duncan test). The summary of ANOVA is presented in Supplementary Fig. S8B. Statistical significance between embryo and RAM samples was evaluated with one-way ANOVA, * significant at P≤0.05, ns, not significant (P_4C=0.147). Normalized data (Supplementary Fig. S1) are provided in Supplementary Fig. S8C.

To understand whether a reduced compaction of the centromeric regions would prevent centromere maturation, we set out to measure the levels of CENH3 in different types of nuclei. For this, we performed CEREBA and barley αCENH3 immunoFISH in 8 and 24 DAP embryo and endosperm nuclei, measured the fluorescence signal intensities over an intersecting line, and calculated their Pearson correlation (Fig. 4A, B; Supplementary Fig. S9). This indicated that all CEREBA foci also contain a CENH3 signal, but the latter appeared to be weaker with increasing C-value in 24 DAP endosperm nuclei (Pearson correlation 0.95 in 3C and 0.75 in 24C). To quantify this interesting observation, we measured signal intensities within ROIs of fixed size and calculated the αCENH3/CEREBA ratio (Fig. 4C; Supplementary Fig. S10). Since it was previously shown that the αCENH3 immunosignal reflects the amount of CENH3 (Lermontova et al., 2006), we measured both FISH and immunosignals in ROIs and calculated the αCENH3/CEREBA ratio. In both embryo and endosperm tissues, there was a significant reduction of αCENH3 relative to CEREBA repeats in endoreduplicated nuclei at 24 DAP (Fig. 4C). Interestingly, this was an effect not only of the C-value but also of DAP as, for example, 6C and 12C nuclei had significantly more αCENH3 at 8 DAP than at 24 DAP (Fig. 4C; Supplementary Fig. S10C). Thus, the decondensation of centromeres occurring in endosperm nuclei correlates with a lesser loading of αCENH3.

Fig. 4.

Loss of HvCENH3 signals in endoreduplicated nuclei of seed tissues. (A) Representative endosperm nuclei of different C-values collected at 8 and 24 DAP after immunostaining followed by FISH (ImFISH) for barley αCENH3 (green) and CEREBA (red). DNA was stained with DAPI (gray). Scale bars=10 µm. (B) Fluorescence intensity plot profiles (y-axis; arbitrary units, A.U.) showing the quantified co-localization of HvCENH3 and CEREBA signals. Intersects used for quantification are highlighted by a pink line in the ImFISH images in (A). Rr displays Pearson’s co-localization coefficient. Data for embryos are presented in Supplementary Fig. S9. (C) αCENH3/CEREBA fluorescence intensity signal ratio (based on Supplementary Fig. S10A) measured for the same-size squared regions of interest (ROIs). Values are means (±SD) from three biological replicates (microscopic slides) marked as black spots, each with at least 100 measured ROIs. Different letters indicate significant differences (P≤0.05, two-way ANOVA, factor 1=C-value, factor 2=DAP, followed by Duncan test). The summary of ANOVA is shown in Supplementary Fig. S10B. DAPI fluorescence measurements are presented in Supplementary Fig. S10C.

Loss of Rabl chromosome configuration in seed endoreduplicated nuclei

In the light of the massive changes in chromosome organization, we asked whether barley seed nuclei retain a Rabl configuration. For this, we imaged the 3D distribution of FISH signals targeting the CEREBA and telomeric repeats in 8 and 24 DAP embryo and endosperm nuclei (Figs 5, 6). Besides nuclei with a typical Rabl configuration, we observed several types of nuclei with dispersed and non-polar centromeric and telomeric signals. To quantify the degree of signal dispersion versus clustering, we measured the shortest distance of each centromere signal to the next telomere signal and expressed it relative to the diameter of the nucleus (Fig 5A, B; image processing workflow as described in Randall et al., 2022). In the 2C and 4C embryo nuclei with a typical Rabl organization, the average, relative distance between centromere and telomere clusters was ~30–40% of ND. In 8C nuclei with more dispersed signals, the relative distance was only ~12% of ND. Assessing the distance distribution among all samples, ANOVA revealed that the C-value (but not DAP) affected the relative positioning of centromeres and telomeres in embryo nuclei (Supplementary Fig. S11A). Similar patterns were found in endosperm nuclei, but with an effect of C-value, DAP, and a combination thereof (Fig. 5B; Supplementary Fig. S11A).

Fig. 5.

Endoreduplication disrupts the Rabl chromosome organization in barley nuclei. (A) Schematic overview of the processing of the raw images and quantified parameters applied for centromere and telomere positioning in the interphase nucleus. (B) Boxplots showing the shortest distance between centromeres and telomeres normalized to nucleus diameter (ND). The lower and upper hinges of the boxplots correspond to the first and third quartiles of the data, and the black lines within the boxes mark the median. Whiskers mark 10% and 90% intervals. Black squares represent outliers beyond the whiskers. The measurements were performed in Imaris software after FISH signal segmentation and nucleus surface rendering. Ten randomly selected nuclei of each C-value/time point were used for the analysis. Different letters indicate significant differences (P≤0.05, two-way ANOVA, factor 1=C-value, factor 2=DAP, followed by Duncan test). The summary of ANOVA is presented in Supplementary Fig. S11A. (C) Boxplots showing the shortest distance of centromeres and telomeres to the nucleus surface. Data acquisition, plot organization, and statistics were performed as described in (B). The summary of ANOVA is presented in Supplementary Fig. S11B. (D) Venn diagrams show the percentage of nuclear space occupied by centromeres and telomeres. The measurements were performed in Imaris software after FISH signal segmentation, nucleus surface rendering, and manual measuring of the nucleus territories occupied by centromeres and telomeres. Ten randomly selected nuclei of each C-value/time point were used for the analysis. The summary of ANOVA is presented in Supplementary Fig. S11C.

Fig. 6.

Three phenotypes of chromosome organization at interphase. (A) Raw images show representative embryo and endosperm nuclei of different C-values revealing Rabl, intermediate, and non-Rabl chromosome organization as determined based on FISH with CEREBA (red) and telomeric (blue) probes. DNA was stained with DAPI (gray). The 3D image segmentation pictures of the surface of the nucleus and FISH signals allow visualization of the spatial distribution of the centromeres and telomeres within the nucleus. Clipping planes (c.p.) represent sections through the 3D modeled nuclei. Scale bars=10 µm. (B) Recognition of the three chromosome organization phenotypes based on Fig. 5B. Interm.=intermediate. (C) Percentage of nuclei with Rabl, intermediate, and non-Rabl chromosome organization. Values are means (±SD) from three biological replicates (microscopic slides), each with at least 25 evaluated nuclei and indicated as a black spot. The same letters indicate samples that do not show significant differences (P≤0.05, two-way ANOVA, factor 1=C-value, factor 2=DAP, followed by Duncan test). There were no significant differences between the embryo samples. The summary of ANOVA is shown in Supplementary Fig. S11D.

In addition, we noted that centromeric and telomeric signals are occasionally located away from the nuclear periphery, which might be another indication of altered chromosome organization. To quantify this reorganization, we measured the shortest distance of centromeric and telomeric FISH signals to the nuclear periphery, defined by the boundary of DAPI staining (Randall et al., 2022). We confirmed that centromeres and telomeres became more dispersed towards the interior of the nucleus in endoreplicated embryo and endosperm nuclei, with a gradual relocation depending on the C-value and DAP (Fig. 5A, C; Supplementary Fig. S11B). Furthermore, to quantify the dispersion of telomeres and centromeres, we calculated the volume occupied by connected centromere signals and the same for telomeres. We expressed these values relative to the volume of the nucleus to provide an estimate of the spatial dispersion (Fig. 5A, D). In the 2C and 4C embryo nuclei, these domains occupied 18–26% of the nuclear volume but this increased to 44–45% in 8C nuclei. Statistical analysis showed that the C-value (for centromeres and telomeres) and DAP (only for telomeres) influenced the expansion of the signals (Supplementary Fig. S11C). In endosperm, centromeres and telomeres covered 27–35% of the nuclear volume of 3C and 6C nuclei at 8 DAP. With increasing C-value and DAP, they dispersed over the nuclear volume even more and reached 68% and 70% of nuclear space, respectively, with 27% overlap in 24C endosperm nuclei at 24 DAP. In endosperm nuclei, the dispersion increased with the C-value, DAP, and their combination (Supplementary Fig. S11C).

Based on the above observations, we defined three arbitrary categories of nuclear organization: (i) Rabl; (ii) intermediate; and (iii) non-Rabl, with a median shortest centromere to telomere distance of 42, 23, and 7%, respectively (Fig. 6A, B). We quantified frequencies of these categories over the experimental points (Fig. 6C). The Rabl configuration was present in ≥85% (n=21 of 25) of 2C and 4C embryo nuclei and in 77.3% (n=19 of 25) of 8C nuclei. The Rabl-type nuclei were substituted by the intermediate type and the proportion of the non-Rabl type remained very low (20%; n=5 of 25). In endosperm, the Rabl configuration appeared in the majority (64%; n=16 of 25) of 3C and 6C nuclei at 8 DAP. With increasing C-value and DAP, the proportion of nuclei with an intermediate and non-Rabl organization became dominant. For instance, 12C endosperm nuclei showed an almost 3-fold reduction in Rabl nuclei from 8 DAP (42%, n=10 of 25) to 24 DAP (15%, n=4 of 25). In parallel, non-Rabl organization increased from 8% (n=2 of 25) at 8 DAP to 31% (n=8 of 25) at 24 DAP. In the extreme case of 24 DAP 24C endosperm, there were nuclei only with non-Rabl (49%, n=13 of 25) and intermediate (51%, n=14 of 24) organization. Statistical analysis confirmed that the Rabl organization was lost with increasing C-value, DAP, and their combination (Supplementary Fig. S11D).

Discussion

Here, we revealed a remarkable plasticity in the morphology of nuclei and arrangements of interphase chromosomes in nuclei from developing barley embryo and endosperm tissues (Fig. 7). Our study shows that the tissue type, level of endoreduplication, and the age after pollination are the major determinants of the observed differences.

Fig. 7.

Graphical summary of the major findings. Dividing nuclei have a round shape; adherent sister chromatids keep an equal amount of CENH3 histone and organize chromosomes according to a Rabl pattern. Endoreduplication alters the nuclear shape and causes positional loss of sister chromatid cohesion and loss of histone CENH3. In addition, they show a non-Rabl chromosome organization.

Chromosome organization has been explored in actively dividing meristematic and somatic tissues of barley, which contain mostly spherical nuclei with a smooth surface (Němečková et al., 2020). These nuclei are from cells that are either mitotically cycling or resting in the G₀/G₁ phase (Jasencakova et al., 2000). Surprisingly, endoreduplicated endosperm nuclei adopted a very irregular shape, with channels lacking DNA staining, suggesting invagination of the nuclear membrane (Fig. 1). Similar shapes have been reported for endoreduplicated nuclei of several distantly related plants including Allium cepa, Narcissus, Pisum sativum, or Solanum lycopersicum (Collings et al., 2000, and references therein; Bourdon et al., 2011, 2012). This suggests that the effect of endoreduplication on the shape of the nucleus and particularly the regularity of its boundary is potentially widespread, consistent with the proposal that complex surface structures may be typical for nuclei of high ploidies (Pirrello et al., 2014). It is assumed that the grooves and invaginations may keep the necessary nucleus to cytoplasm surface ratio (Bourdon et al., 2012). Our work shows that invaginations are not a pure effect of endoreduplication in barley and that the seed developmental stage, tightly linked to its physiological state, plays a role. Some nuclear surface irregularities could be a result of metabolic activity in embryo and endosperm cells, possibly due to filling of cells with active and/or storage compounds or due to the altered cytoskeleton impacting the integrity of the nuclear envelope. This phenotype could also be linked to programmed cell death that is typical for the endosperm of most cereals (Young et al., 2000).

The hallmark of the Rabl configuration is the centromere and telomere clustering at opposite nuclear poles. However, >130 years after its discovery (Rabl, 1885), the principles of this organization remain a matter of debate (Santos and Shaw, 2004). In some plants, the Rabl configuration was long thought to be the only type of genome organization. However, an increasing number of studies suggest tissue-specific variation in chromosomal organization (Fujimoto et al., 2005a; Idziak et al., 2015; Němečková et al., 2020; Shan et al., 2021). Some cells lose the Rabl pattern soon after entering the interphase, whereas others retain the organization throughout the interphase and until the next mitosis (Cowan et al., 2001). Nuclear genome size and chromosome length were postulated as two possible factors conferring the Rabl configuration (Santos and Shaw, 2004). While this holds for many species, some plants with giant genomes lack Rabl organization (Fujimoto et al., 2005a). Also, there are striking differences in genome organization between some closely related species. A well-described example are Brachypodium species, where B. distachyon shows a Rabl configuration in root nuclei while its relative B. stacei, with a similar genome size but twice as many chromosomes, does not (Idziak et al., 2015). This suggests that genome size alone cannot serve as a universal rule defining the Rabl organization. Recently, it was proposed that the Condensin II complex plays a major role in 3D interphase genome organization, and that an incomplete set of its subunits favors a Rabl-like pattern across the tree of life (Hoencamp et al., 2021). Applicability of this classification for the organization of plant chromosomes still requires investigations because all plants sequenced to date contain a full set of Condensin II subunits, in spite of their diverse chromosome organization (Fujimoto et al., 2005b; Schubert, 2009).

So far, barley has been considered as a species with a strict Rabl chromosome organization. We showed that there is variability in the chromosome configuration in barley seed tissues that is affected by the tissue type, seed developmental stage (days after pollination), and strongly by endoreduplication (C-value) and combination of the latter two factors. In contrast, Rabl organization was maintained in the embryo and endosperm nuclei of bread wheat (Wegel and Shaw, 2005), which could be due to analysis of younger tissues that contained only a small portion of endoreduplicated nuclei or, less likely, due to genuine species-specific differences. Importantly, correlation between loss of Rabl organization and degree of endoreduplication favors models suggesting that Rabl configuation is established and reinforced during mitotic cell divisions (Santos and Shaw, 2004). Our data show that the amount of nuclei with Rabl decreases not only with the number of DNA replications (that are not followed by mitosis), but also with the time since the last replication (Fig. 6). However, which molecular factors ensure relatively stable clustering of centromeres and telomeres in between divisions remains currently unknown.

The other observed changes in chromosome organization add to the little-explored organization of endoreduplicated nuclei of cereal seeds (Wegel and Shaw, 2005; Wegel et al., 2005; Bauer and Birchler, 2006). Here, repositioning of centromeres and telomeres from the nuclear envelope more into the nuclear space may contribute to loss of Rabl configuration (Fig. 5) (Santos and Shaw, 2004). Another observed alteration in chromosome organization was related to absence of SC cohesion at 5S rDNA loci (Fig. 2). Although we cannot draw any conclusions about the organization of singly-copy sequences, our data suggest that SCs are absent at least in some parts of the endoreduplicated barley chromosomes. This is reminiscent of the loss of SC cohesion along chromosome arms in Arabidopsis nuclei with a C-value of 4C or more (Schubert et al., 2006). At centromeric regions, we found an increasing number of CEREBA foci in endoreduplicated nuclei (Fig. 3), which is similar to Arabidopsis (Schubert et al., 2006; Baroux et al., 2017). This suggests a relaxed control of heterochromatin compaction at centromeres upon endoreduplication in barley, which is in contrast to the situation in maize (Bauer et al., 2006). We also observed reduction in centromeric histone CENH3 in endoreduplicated nuclei (Fig. 4). This is to be expected because CENH3 loading occurs in G₂ phase, that is skipped in the endoreduplication cycle (Lermontova et al., 2007). Furthermore, data from Arabidopsis show that CENH3 is not produced during endoreduplicative S-phase (Lermontova et al., 2011). Interestingly, we found a significant replication-independent loss of CENH3 in nuclei of the same C-value in later versus earlier seed developmental stages.

What the significance of the manifold changes in endoreduplicated barley nuclei is remains currently unknown. Speculatively, it could be linked with transcriptional reprogramming and a boost in synthesis of specific storage compounds in endosperm (Sabelli and Larkins, 2009). Furthermore, it could be related to a loss of mitotic activity and onset of the cellular trajectory towards programmed cell death that occurs in large parts of cereal endosperm (Nowicka et al., 2021a).

In conclusion, our study highlights previously underappreciated dynamics in chromosome organization of barley embryo and endosperm nuclei upon endoreduplication. The most notable change is the progressive loss of polar chromosome organization and the disruption of centromere and telomere clusters. This shows that the Rabl chromosome arrangement is not a general rule for barley, and that mitosis may function as a mechanism reinforcing this organization. In general, these data help in understanding the principles and dynamics of genome organization during the course of plant development.

Supplementary data

The following supplementary data are available at JXB online.

Fig. S1. Schematic drawing of G₁ diploid and triploid cells.

Fig. S2. Geometrical characteristics of flow-sorted embryo and endosperm nuclei.

Fig. S3. Pearson correlation coefficient analysis between nuclear geometry parameters.

Fig. S4. The complex shape of endoreduplicated nuclei.

Fig. S5. Sister chromatid organization for the 5S major rDNA locus in barley seed nuclei.

Fig. S6. Sister chromatid organization for the 5S minor rDNA loci in barley seed nuclei.

Fig. S7. Sister chromatid organization for the 45S rDNA loci in barley seed nuclei.

Fig. S8. CEREBA organization in barley seed nuclei of different C-values and DAP.

Fig. S9. Co-localization of CENH3 and CEREBA signals in barley embryo nuclei.

Fig. S10. Fluorescence measurements of αCEN3H3, CEREBA, and DAPI signals.

Fig. S11. Statistical analysis supporting the Rabl chromosome configuration study.

Abbreviations:

Abbreviations:
 
• CENH3

  centromere-specific histone H3

 
• CEREBA

  CENTROMERIC RETROELEMENT OF BARLEY

 
• CT

  chromosome territory

 
• DAP

  days after pollination

 
• FISH

  fluorescence in situ hybridization

 
• FITC

  fluorescein isothiocyanate

 
• mip

  maximum intensity projection

 
• msv

  middle slide view

 
• NA

  nucleus area

 
• NCI

  nucleus circularity index

 
• ND

  nucleus diameter

 
• NP

  nucleus perimeter

 
• ROI

  region of interest

 
• SC

  sister chromatid

Acknowledgements

We thank, Dr A. Doležalová for sharing the immunoFISH protocol and anti-αCENH3 antibody, E. Jahnová for technical assistance, and Z. Bursová for plant care. The authors thank COST Action no. CA 16212 ‘INDEPTH’ for training in Imaris software.

Author contributions

AP and AN: design; AN, LF, MK, and JZ: performing experiments; MS: flow sorting of the nuclei; AN: data analysis and figure preparation; CB: providing expertise in microscopic image processing analysis by Imaris; AP and AN: writing, with contributions from all authors. All authors approved the final version of this article.

Conflict of interest

The authors declare that they have no conflict of interest in relation to this work.

Funding

This research was funded by the Czech Science Foundation grants 18-12197S and 21-02929S (to AP), Purkyně Fellowship from the Czech Academy of Sciences to AP, and the European Regional Development Fund project ‘Plants as a tool for sustainable global development’ (no. CZ.02.1.01/0.0/0.0/16_019/0000827).

Data availability

All data supporting the findings of this study are available within the paper and within its supplementary data published online.