- © 2019 American Society of Plant Biologists. All rights reserved.
Abstract
Spatial organization of chromatin contributes to gene regulation of many cellular processes and includes a connection of chromatin with the nuclear lamina (NL). The NL is a protein mesh that resides underneath the inner nuclear membrane and consists of lamins and lamina-associated proteins. Chromatin regions associated with lamins in animals are characterized mostly by constitutive heterochromatin, but association with facultative heterochromatin mediated by Polycomb-group (PcG) proteins has been reported as well. In contrast with animals, plant NL components are largely not conserved and NL association with chromatin is poorly explored. Here, we present the connection between the lamin-like protein, CROWDED NUCLEI1 (CRWN1), and the chromatin- and PcG-associated component, PROLINE-TRYPTOPHANE-TRYPTOPHANE-PROLINE INTERACTOR OF POLYCOMBS1, in Arabidopsis (Arabidopsis thaliana). We show that PWO1 and CRWN1 proteins associate physically with each other, act in the same pathway to maintain nuclear morphology, and control expression of a similar set of target genes. Moreover, we demonstrate that transiently expressed PWO1 proteins form foci located partially at the subnuclear periphery. Ultimately, as CRWN1 and PWO1 are plant-specific, our results argue that plants might have developed an equivalent, rather than homologous, mechanism of linking chromatin repression and NL.
INTRODUCTION
Spatial organization of chromatin in the nucleus is nonrandom and contributes to gene regulation for many cellular processes (Van Bortle and Corces, 2012). Spatial organization can be categorized into several layers (Gibcus and Dekker, 2013). For instance, early cytological studies in plants showed that a subset of chromatin (heterochromatin) is densely packed and constitutes distinguishable structures called “chromocenters” (Fransz et al., 2002). Heterochromatin can be either constitutive or facultative, thus either be very stable or forming in a cell-type–dependent or cue-dependent manner. Further cytological developments revealed that interphase chromosomes occupy topological spaces in the nucleus and form distinct entities known as “chromosome territories” (Lichter et al., 1988). Moreover, chromatin conformation capture methods proved the abundance of long- and short-range genome loops, which have essential roles in gene transcription and coregulation (Kadauke and Blobel, 2009). Another aspect is the distribution of the chromatin in the interior and periphery of the nucleus (Gordon et al., 2015).
The nuclear periphery is a subnuclear space in the vicinity to the nuclear envelope (NE), a double lipid bilayer forming inner and outer nuclear membrane, and nuclear lamina (NL), a nuclear protein mesh physically associated with the inner nuclear membrane. In mammals, the NL comprises lamin-associated membrane proteins and lamins themselves, the latter being categorized as A- and B-type, depending on the structural properties and expression pattern (Dechat et al., 2010). Interestingly, lamin proteins bind specific, nonrandom regions of the chromatin and form lamina-associated domains (LADs) in animals, that are present either constitutively or form in specific cell-types (facultative LADs; Meuleman et al., 2013). LADs are generally gene-poor, associate with constitutive heterochromatin, and genes residing in LADs show no or low transcriptional activity. LADs themselves possess several features of heterochromatin—they overlap with regions that replicate late in s-phase and are enriched in repressive histone marks of histone H3 Lys-9 di/trimethylation (H3K9me2/3) and, at LAD boundaries, in H3K27me3 (van Steensel and Belmont, 2017). H3K27me3 may not overlap with LADs, but be associated with highly structured regions between LADs (Vieux-Rochas et al., 2015) and particularly assemble with lamin A, which is also present in the nuclear interior (Cesarini et al., 2015).
H3K27me3 is established by a complex of the highly-conserved Polycomb-group (PcG) proteins known as Polycomb Repressive Complex2 (PRC2; Margueron and Reinberg, 2011; Derkacheva and Hennig, 2014). A canonical form of PRC2 was initially discovered in Drosophila melanogaster and the core catalytic component is the methyltransferase Enhancer of zeste. PRC2 is present also in Arabidopsis (Arabidopsis thaliana), albeit with several components being multiplicated and partially diversified in expression pattern and set of targets. Enhancer of zeste homologs in Arabidopsis PRC2 are CURLY LEAF (CLF), MEDEA, and SWINGER (SWN).
LADs have not yet been identified in plants. A recent analysis identified the association of a plant nuclear pore component, with DNA/chromatin (Bi et al., 2017) but the extent of LADs in this data set is unclear and will require mapping of core-lamina–like associated proteins. Plants were long believed not to contain lamin homologs. Interestingly, the plant NL contains lamin-like proteins (Ciska and Moreno Díaz de la Espina, 2014), which, similarly to their animal counterparts, possess a coiled-coil protein motif and show general localization to the nuclear periphery (Gardiner et al., 2011; Ciska and Moreno Díaz de la Espina, 2014). Lamin-like genes in Arabidopsis form the plant-specific CROWDED NUCLEI (CRWN) family containing four members (CRWN1, CRWN2, CRWN3, and CRWN4). CRWN proteins are orthologs of NUCLEAR MATRIX CONSTITUENT PROTEIN1, which was initially discovered in carrot (Daucus carota) as matrix-associated protein localizing to the nuclear periphery (Masuda et al., 1997) that is land-plant–specific (Ciska et al., 2013). CRWN proteins were also biochemically identified from Arabidopsis nuclei: a crude NL fraction that had mesh-like structures composed of fibrous materials was isolated by nuclei isolation and removal of membranes. The subsequent mass spectrometry analyses resulted in the identification of 660 proteins including CRWN1 and CRWN4 (Sakamoto and Takagi, 2013). The most prominent phenotypes of CRWN family loss-of-function mutants are reduced nuclear size, increased nuclear DNA density, and abnormal nuclear shape, with the most pronounced effect seen in certain combinations of higher order mutants (Dittmer et al., 2007; Sakamoto and Takagi, 2013; Wang et al., 2013). Furthermore, CRWN proteins interact with other plant NE/NL components such as SAD1/UNC-84 HOMOLOGY1 (Graumann, 2014) and KAKU4 (Goto et al., 2014). Interestingly, CRWN genes also affect constitutive heterochromatin organization and 3D chromosome arrangement—different CRWN mutant combinations show altered chromocenter integrity (Dittmer et al., 2007; Wang et al., 2013) and single mutants of crwn1 or crwn4 exhibit elevated trans-chromosomal interactions, but unchanged contact frequency in cis (Grob et al., 2014). However, the exact mechanism of CRWN protein function in chromatin organization remains unknown.
In our previous work (Hohenstatt et al., 2018), we identified the Arabidopsis protein PROLINE-TRYPTOPHANE-TRYPTOPHANE-PROLINE (PWWP) INTERACTOR OF POLYCOMBS1 (PWO1) as a novel plant-specific factor associated with chromatin and PRC2. We showed that PWO1 binds histones and a wide range of histone modifications. PWO1 interacts physically with PRC2 methyltransferases (MEDEA, SWN, CLF), displays epistatic genetic interaction with CLF, controls expression of several PRC2-dependent target genes and is needed for full H3 and H3K27me3 occupancy at several PcG target genes. Importantly, we demonstrated that PWO1 changed nuclear localization of CLF in a transient system and recruited CLF to subnuclear speckles.
Here, we present a connection between PWO1 and the Arabidopsis NL-like structure, possibly linking H3K27me3-marked chromatin and the nuclear periphery in plants. We demonstrate that PWO1 and PWO1-CLF speckles are positioned predominantly close to the nuclear periphery in Nicotiana benthamiana and that PWO1 localizes to subnuclear structures in stable Arabidopsis transgenic lines. We show that PWO1 interactors are enriched in the NL-like fraction and PWO1 associates with CRWN1 physically, genetically, and functionally, as they both control expression of a similar set of genes. Consequently, we uncover that PWO1 regulates nuclear size, a phenotypic trait related to NL-like structure. Taken together, our results uncover a novel function of PWO1 and provide a novel link between chromatin regulation and the subnuclear periphery in plants.
RESULTS
Detailed Analyses of PWO1 Subnuclear Localization
We previously demonstrated that PWO1 forms speckle-like foci upon inducible, transient expression in N. benthamiana (Hohenstatt et al., 2018). We also showed that upon coexpression in this system, CLF forms colocalizing speckles with PWO1, in contrast with the uniform nucleoplasmic signal of CLF when expressed alone (Hohenstatt et al., 2018). To further characterize the PWO1-CLF speckles, we coexpressed PWO1-GFP (green fluorescent protein) and CLF-mCherry fusion proteins in N. benthamiana leaves using a β-estradiol–inducible 35S promoter (i35S), and revealed recruitment of CLF to PWO1-speckles and colocalization of both proteins in speckles and the nucleoplasm (Figure 1A), as observed in Hohenstatt et al. (2018). The usage of an inducible promoter minimized the formation of overaccumulated protein aggregates. Of note, we previously demonstrated that PWO1 lacking the PWWP1 domain does not form speckles under the same conditions, indicating that the observed protein localization is not an effect of protein overexpression (Hohenstatt et al., 2018).
PWO1 Localization to Peripheral Speckles in N. benthamiana.
(A) A representative nucleus of N. benthamiana leaves coinfiltrated with i35S:CLF-GFP and i35S:PWO1-mCherry. Leaves of N. benthamiana were infiltrated with estradiol-inducible constructs. The images of epidermis nuclei were acquired 16 to 20 h postinduction. Scale bar = 10 µm.
(B) FRET-APB measurements for nuclei exemplified in (A). FRET was measured for speckles (spec, red arrow) and diffuse localization (blue arrow). An average of efficiencies for n = 15 to 22 nuclei is shown. As positive control, a PWO1-GFP-mCherry and as negative controls, CLF-GFP and PWO1-GFP were measured. Significance level was measured using Student’s t test and is represented by the asterisks: *P < 0.05; **P < 0.01, ***P < 0.001. Error bars = sd.
(C) A nucleus of N. benthamiana transiently expressing PWO1-GFP. The images were acquired and processed like in (A). Scale bar = 10 µm.
(D) Number of speckles and their localization per individual nucleus.
As a next step, we employed Förster resonance energy transfer with acceptor photobleaching (FRET-APB) to further characterize the PWO1-CLF interaction and demonstrated that FRET efficiencies for coexpressed samples are significantly higher than the negative controls (PWO1-GFP and CLF-GFP expressed without donor mCherry construct; Figure 1B). Interestingly, the FRET-APB donor signal intensity was higher in speckles than in the nucleoplasm, suggesting that PWO1 and CLF might form particularly tight interactions in the speckles (Figure 1B).
More detailed examination of PWO1 localization revealed a nonrandom pattern of subnuclear localization of the speckles as they are predominantly found close to the nuclear periphery in all nuclei (n = 13) examined (Figures 1C and 1D; Supplemental Figure 1). In addition, a small number of speckles was located in the nucleus interior or in vicinity to the nucleolus (Figure 1D).
Next, we sought to investigate PWO1 localization in transgenic Arabidopsis lines expressing PWO1-GFP from the PWO1 promoter (PWO1:PWO1-GFP). When conventional confocal microscopy did not result in sufficient detail and resolution, we performed super-resolution microscopy (using AiryScan technology; Zeiss) in different root layers. PWO1 localized to the nucleoplasm and subnuclear structures in cortex and epidermis tissues of the root meristematic zone (Figure 2). However, the nature and detailed position of the structures remain unclear, but is clearly not entirely at the nuclear periphery. Interestingly, imaging of nuclei from the epidermis and cortex of root elongation zone revealed more uniform nucleoplasmic distribution, implying a tissue-specific pattern of PWO1 localization.
PWO1 Localization in Arabidopsis.
Super-resolution microscopy images (LSM880 AiryScan; Zeiss) from roots of PWO1:PWO1-GFP. The images represent nuclei from the meristematic zone epidermis in an overview (A) and or zoomed in (B), elongation zone epidermis (C), meristematic zone cortex (D), and elongation zone cortex (E). “Black holes” are nucleoli. Cells were counterstained with Propidiumiodide (red). The most prominent subnuclear structures are marked with arrows. Scale bars = 5 µM.
Overall, we showed that PWO1 localization, in both the transient expression system and stable Arabidopsis line, is not uniform and involves formation of foci in N. benthamiana, which are predominantly localized at the nuclear periphery. In our transient system, we also demonstrated that PWO1 strongly recruits CLF to peripheral speckles in the nucleus, possibly indicating that PWO1 might serve as a bridging link between the plant PRC2 and the NE.
PWO1 Physically Associates with NL/NE Proteins
Given the partially peripheral localization of PWO1, we sought to decipher the connection between PWO1 and the NE by searching for PWO1 protein interactors. We conducted coimmunoprecipitation (CoIP) experiments coupled with MS using the PWO1:PWO1-GFP Arabidopsis transgenic line. The protein abundance was scored via label-free quantification analysis, and a comparison to background sample (Col-0, wild type) was used to identify significantly enriched proteins in the PWO1:PWO1-GFP line. As a result, we obtained a list of 99 putative PWO1 interactors.
Consistently with the chromatin- and PcG-related role of PWO1 (Hohenstatt et al., 2018), we detected other proteins with similar functions (Figure 3A; Supplemental Table 1). Those included TELOMERE REPEAT BINDING1 (TRB1), belonging to the TRB family that recruits core PcG proteins to DNA via recognition of telobox cis-regulatory motifs (Zhou et al., 2016); YING YANG1, a transcription factor corepressing plant disease susceptibility genes (Lai et al., 2014) and Arabidopsis putative ortholog of the animal PRC2 recruiter YING YANG1/Pleiohomeotic (Atchison et al., 2003); AT1G79020, an ortholog of Enhancer of Polycomb-like proteins and an interactor of NuA4 and SWR1-C complexes responsible for histone acetylation and histone variant exchange (Bieluszewski et al., 2015); PWWP domain protein3, which is found in a complex with PcG proteins (Zhou et al., 2018); and MSI1, an Arabidopsis PRC2 component (Hennig and Derkacheva, 2009). PWO1 did not precipitate other PRC2 members, suggesting that the physical interaction of PWO1 with several PRC2 components may be transient or not very stable. Moreover, PWO1 immunoprecipitated various components of chromatin remodeling complexes, such as CHROMATIN REMODELING12 and 17, ACTIN RELATED PROTEIN4, AtBAF60/AtSWP73B/CHC1, and SWITCH/SUCROSE NONFERMENTING3A. These results are in agreement with PWO1’s putative role in controlling nucleosome occupancy and decreased histone H3 enrichment in pwo1 at several PcG targets (Hohenstatt et al., 2018). In addition, PWO1 coprecipitated various histone proteins, confirming the previously shown in vitro interactions with histone H3. Lastly, we discovered that PWO1 is associated with members of the PEAT-complex (PWWP domain‐containing proteins-Enhancer of Polycomb‐Related [EPL]-AT‐rich interaction domain‐containing proteins [ARID]-telomere repeat binding proteins [TRB] complex), which was recently identified to silence transposable elements (Tan et al., 2018).
Identification of PWO1 Interactors.
(A) MS results for bait protein and selected PWO1 interactors. NL fraction represents proteins detected in crude NL extract.
(B) Overlap between Arabidopsis NL fraction and PWO1 interactors. PWO1 interactors correspond to proteins showing significant (P < 0.05) enrichment over the wild type in CoIP-MS/MS experiment. Significance level of the overlap was calculated using the Student’s t test. For the full list of 65 overlapping proteins, see Supplemental Table 1.
Interestingly, we discovered that ∼56% of PWO1 putative interactors (55 of 98 proteins) overlap with the components of a crude plant NL-like fraction (Figure 3B; Sakamoto and Takagi, 2013), including constitutive NL/NE members, such as CRWN1, CRWN4, and SAD1/UNC-84 DOMAIN PROTEIN 1. The overlap contained also chromatin-associated proteins, like SKI-INTERACTING PROTEIN1, CHROMATIN REMODELING17, ACTIN RELATED PROTEIN4, AtBAF60/CHC1, and MSI1 (Figure 3B; Supplemental Table 1). However, PEAT complex members, which include PWO1, PWO2, and PWO3, are not identified in the NL fraction (Tan et al., 2018). Overall, the high number of PWO1 interactors overlapping with the NL fraction is in agreement with the presence of PWO1 speckles at the nuclear periphery, which were revealed in the transient expression experiments.
We focused further analyses on the putative interaction with CRWN1, a nuclear lamin-like protein with a prominent role in nuclear morphology (Wang et al., 2013), chromocenter (Dittmer et al., 2007), and interchromosomal contact frequency regulation (Grob et al., 2014). To detect a direct PWO1-CRWN1 interaction, we performed Yeast-Two-Hybrid (Y2H) and FRET-APB in N. benthamiana. Of note, in our hands, CRWN1 protein transiently expressed in N. benthamiana leaves showed not only NE-localization, but was also present uniformly in the nucleoplasm and caused nuclear membrane deformations, as reported in Goto et al. (2014) and Graumann (2014). Y2H and FRET-APB experiments revealed no or very low interaction between CRWN1 and full-length PWO1, but much stronger signal over negative controls when only a C-terminal PWO1 fragment (634 to 769 amino acids) was used instead (Figure 4; Supplemental Figures 2 and 3). This result might indicate steric constrains of overall PWO1 protein folding caused by introduction of fusion tags (Gal4 [GALactose metabolism4]-domains or FRET fluorophores), which may mask the CRWN1 binding site in full-length PWO1 construct versions. By contrast, the use of a truncated PWO1 protein containing the CRWN1-interacting region might resolve potential steric effects. The interaction with CRWN1 was shown to be specific to the C-terminal part of PWO1 and not a simple outcome of random binding of any truncated PWO1 versions, as the other PWO1 fragments (frag.1 and frag.2) still show a background-level interaction (Figure 4B).
Analysis of PWO1-CRWN1 Interaction.
(A) Scheme of PWO1 protein with different coding sequence fragments used in (B) and (C). NLS, predicted nuclear localization signal. Scale bar on the top of the scheme = 100 amino acids.
(B) Y2H analysis on PWO1-CRWN1 interaction. Overall yeast growth and protein interaction among CRWN1, PWO1, and PWO1 fragments was assessed on transformation medium (-LW, -Leucine Tryptophane) and selection medium (-LWH, -Leucine Tryptophane Histidine). AD, activation domain; BD, DNA-binding domain.
(C) FRET-APB-based interaction between PWO1 and CRWN1. FRET efficiency was calculated for nuclei from the positive control (n = 10), negative controls (n = 15 to 17), and interaction samples (n = 15 to 27). Error bars = sd values. Significance between interaction samples and their respective negative controls was inferred from Student’s t test. Significance threshold: *P < 0.05, **P < 0.01.
Interestingly, the CRWN1-interacting region in the PWO1 sequence does not contain any annotated protein domain (Figure 4A). However, an alignment to the closest PWO1 plant homologs revealed a high level of conservation for the PWO1 C-terminal sequence in Brassicaceae species other than Arabidopsis (Supplemental Figure 3), highlighting a functional importance of this fragment. By contrast, the PWO1 C-terminal fragment is largely absent from PWO2 and PWO3, suggesting partially separate functions between PWO family proteins in Arabidopsis.
PWO1 Interacts Genetically with CRWN1
Given the physical association of PWO1 and CRWN1, we sought to investigate their functional connection by studying pwo1, crwn1, and pwo1 crwn1 mutants at the phenotypic level. As PWO1 was shown to interact with PRC2 components (Hohenstatt et al., 2018), we included also swn and swn crwn1 lines in our analysis to characterize the link between the NL-like structure and canonical PRC2 members.
A prominent feature of mutants of NE or NL components is the reduction in nuclear size and change of nuclear shape. To characterize alterations in nuclear morphology, we isolated nuclei from whole seedlings, stained them with 4′,6-diamidino-2-phenylindole (DAPI), and measured their nuclear area and circularity (Figure 5). Consistently with other studies (Wang et al., 2013), we observed a reduction in nuclear area and increase in circularity index in crwn1, compared with the wild-type control (Col-0). Interestingly, lower average nuclear area was also observed in pwo1, albeit with no significant changes in nuclear shape. Importantly, we detected no substantial alterations in ploidy levels in pwo1 in comparison with the wild type (similar to results obtained for crwn1 mutants; Dittmer et al., 2007), suggesting that changes in endoreduplication do not underlie the observed reduction in nuclear size (Supplemental Figure 4).
Genetic Interaction of pwo1, crwn1, and swn.
(A) DAPI-stained nuclei in different genotypes. Scale bars = 20 µm.
(B) and (C) Nuclear area (B) and circularity (C) measurements. Calculations were performed on Z plane showing the largest area of particular nucleus in the Z-stack. A quantity of 72 to 92 nuclei per genotype were measured. Circularity index was calculated as 4π (area/perimeter^2), with the maximal value of 1.0 corresponding to the perfect circle. Error bars are derived from standard error (SE). Significance was calculated using one-way analysis of variance followed by Student’s t test (P < 0.001).
(D) The relationship between circularity index and nuclear area in nuclei from genotypes used in the study. Blue diamonds represent individual nuclei; black line displays the trendline.
The analysis of pwo1 crwn1 phenotype revealed that both genes act in the same pathway to control nuclear size as the double mutant phenocopies either of the single mutants (Figure 5B). Interestingly, we also observed a genetic interaction for the nuclear shape phenotype. As pwo1 crwn1 shows a circularity index average similar to pwo1, the CRWN1-dependent nuclear shape alteration was suppressed by the lack of functional PWO1 (Figure 5C). Consistently, we found that the nuclei from pwo1 crwn1 are characterized by a broad distribution of circularity index levels, similarly to the pwo1 single mutant. By contrast, the nuclei from crwn1 are narrowly distributed and enriched at the high circularity index levels (Figure 5D). This result suggests that PWO1 is required for CRWN1-mediated control of nuclear shape.
Furthermore, we observed no significant changes in nuclear morphology in swn. The double mutant crwn1 swn showed an increased average circularity and reduced average nuclear area, hence resembling crwn1. Consistent with this, a distribution of individual nuclei in crwn1 swn is more similar to the distribution observed in crwn1, rather than swn. The phenocopy of nuclear morphology changes between swn crwn1 and crwn1, yet lack of such a phenotype in swn, suggests that SWN does not influence nuclear architecture.
Overall, our results suggest that PWO1 and CRWN1 determine nuclear size and shape within the same genetic pathway. By contrast, a canonical PRC2 component, SWN, does not influence morphology of the nucleus, irrespective of CRWN1 function.
Loss of PWO1 and CRWN1 Does Not Drastically Affect Reciprocal Localization
Given the physical and genetic interaction between PWO1 and CRWN1, we asked whether their patterns of subnuclear localization are dependent on each other. We showed that PWO1 forms foci in the meristematic zone of the Arabidopsis root. In turn, CRWN1 is known to localize exclusively to the nuclear periphery, forming a thin ring delineating the border of the nucleus (Supplemental Figure 5; Dittmer et al., 2007). However, our confocal microscopy analyses of the lines PWO1:PWO1-GFP crwn1 and CRWN1:CRWN1-GFP pwo1 pwo3 revealed no obvious changes in subnuclear localization in comparison to the control lines (Supplemental Figures 5 and 6). Therefore, our results suggest that the subnuclear localizations of PWO1 and CRWN1 are either independent of each other or that only mild alterations occur, which are difficult to detect by conventional confocal microscopy.
Loss of PWO1 and CRWN1 CRWN2 Affect Expression of a Similar Set of Genes
To further understand the connection between PWO1 and CRWN family genes, we investigated the global transcriptomic changes in seedlings of pwo1 and crwn1 crwn2 mutants using RNA sequencing (RNA-Seq). Given the phenotypes of crwn1 and crwn1 crwn2 (Dittmer et al., 2007), we decided to use the stronger double crwn1 crwn2 mutant to depict a larger effect of CRWN family-dependent transcriptomic changes. Differential expression analysis (see “Methods”) identified 179 significantly upregulated and 242 significantly downregulated differentially expressed genes (DEGs) in pwo1, as compared with the wild-type control (Col-0). In turn, similar analysis on crwn1 crwn2 revealed 602 significantly upregulated and 214 significantly downregulated DEGs. Cross-comparison of the data sets from both genotypes revealed a significant overlap of 71 upregulated and 73 downregulated DEGs (Figures 6A and 6B). These results imply that PWO1 and CRWN1/CRWN2 regulate a partially similar set of target genes.
DEG Analysis—Overlap Between pwo1 And crwn1/2.
(A) Venn diagram comparing upregulated-DEGs in pwo1 (pwo1-upregulated) and crwn1/2 (crwn1/2-upregulated). Significance level of the overlap was calculated in the software “R” using a hypergeometric test.
(B) Venn diagram comparing downregulated-DEGs in pwo1 (pwo1-downregulated) and crwn1/2 (crwn1/2-downregulated). Significance level of the overlap was calculated in the software “R” using a hypergeometric test.
(C) GO-term enrichment analysis on commonly upregulated-DEGs in pwo1 and crwn1/2 as compared with all TAIR10 protein-coding genes (“total”). The percentage corresponds to the number of genes enriched in significantly abundant (FDR < 0.005) GO terms in the overlap shown in (A) or the whole protein-coding gene number based on TAIR10 annotation.
Next, we asked whether common DEGs affected by both pwo1 and crwn1/2 are associated with specific functional categories. Our gene ontology (GO) analysis (see “Materials and Methods”) on the overlap of upregulated DEGs showed a significant enrichment of the functions related to response to stress (biotic and abiotic), iron homeostasis, and transport of amino acids/nitrate (Figure 6C). In turn, DEGs from the overlap of downregulated targets are involved specifically in the response to auxin.
Given PWO1’s involvement in the PRC2 pathway (Hohenstatt et al., 2018), we hypothesized that upregulated DEGs in pwo1 are covered by H3K27me3 in wild-type plants. Therefore, we compared our list of upregulated-DEGs in pwo1 and a published data set of H3K27me3 targets (Oh et al., 2008). Indeed, we detected a significant overlap of 105 genes present in both lists (Figure 7A), showing that PWO1 is required for the repression of a subset of PRC2 targets. As PWO1 associates with CRWN1, we asked whether mutants in the CRWN gene family affect expression of H3K27me3 targets as well. Interestingly, the comparison between upregulated DEGs in crwn1/2 and H3K27me3-covered genes revealed a significant number of genes shared between two data sets (Figure 7A). Finally, cross-comparison of upregulated DEGs in pwo1 and crwn1/2, and H3K27me3 targets gave 36 genes shared by all three data sets (Figure 7A). Their function is related to stress response and amino acid/nitrate transport, but not to iron metabolism (Figure 7B), which implicates a role of PWO1 and CRWN1 in iron homeostasis facilitated in a PRC2-independent manner. We noted also a substantial number of H3K27me3 targets whose expression is affected only by either loss of PWO1 or CRWN1 CRWN2. This result suggests that PWO1 and CRWN1 mediate repression (directly or indirectly) of PRC2 targets also independent of each other.
DEG Analysis—Cross-Comparison Between Mutants And H3K27me3 Targets.
(A) Venn diagram comparing: upregulated-DEGs in pwo1 (pwo1-upregulated), in crwn1/2 (crwn1/2-upregulated), and H3K27me3 targets in Col-0. The significance level of overlapping gene number is as follows: 4.56e-06 (H3K27me3 targets versus pwo1-upregulated), 2.20e-06 (H3K27me3 targets versus crwn1/2-upregulated), and 5.64e-28 (H3K27me3/pwo1-upregulated versus H3K27me3/crwn1/2-upregulated). P values were calculated in the software “R” by a hypergeometric test.
(B) GO-term enrichment analysis on commonly upregulated DEGs in pwo1 and crwn1/2, and H3K27me3 targets in Col-0, as compared with all TAIR10 protein-coding genes (“total”). The percentage corresponds to the number of genes enriched in significantly abundant (FDR < 0.0005) GO terms in the common target genes shared by all three data sets in (A) or the whole protein-coding gene number based on TAIR10 annotation.
DISCUSSION
Here, we demonstrate that the Arabidopsis chromatin- and PRC2-associated protein PWO1 interacts physically with numerous NL-like components, including CRWN1—a coiled-coil analog of lamin proteins in Arabidopsis. We report that PWO1 and CRWN1 are epistatic in controlling nuclear size and affect expression of a similar set of PRC2 target genes. Moreover, we show that PWO1 in N. benthamiana localizes to subnuclear foci, which are located mostly within the nuclear periphery. PWO1 in Arabidopsis show also a nonuniform localization in meristematic tissues; however, the features and precise subnuclear position of observed structures require further exploration. Future studies should also elucidate whether PWO1 structures in Arabidopsis are separate entities or form a large interconnected subnuclear network.
Our findings suggest a conservation of the association between a repressive chromatin environment and the nuclear periphery across biological kingdoms, including plants. Strikingly, the exact mechanism and linking factors involved in this phenomenon are frequently species-specific, as has been reported for i.e. Caenorhabditis elegans-specific 4 (Gonzalez-Sandoval et al., 2015) or Silent Information Regular4, present in Saccharomycetaceae only (Taddei et al., 2004; Hickman et al., 2011). Identification of PWO1, a novel plant-specific bridging factor, adds another evolutionary distinct element in an otherwise conserved machinery linking chromatin repression with NL-like structures.
The pathways linking repressive and specifically PRC2-regulated chromatin and the NL vary not only between species, but also within the same organism. Many protein effectors associated with the NL in fungi or animals regulate only a specific subset of targets and play roles also outside of the nuclear periphery (Harr et al., 2016). Consequently, our results on PWO1 subnuclear localization, transcriptomic analyses in pwo1 and crwn1/2, and predominantly nucleoplasmic distribution of H3K27me3 in plants (Mathieu et al., 2005) collectively suggest a NL- and PWO1-dependent repression of only a specific set of PRC2 targets. We speculate that the other, PWO1-independent, pathways bridging the NL and PRC2 also exist, as suggested by the PWO1-independent CRWN1/2-mediated repression of PRC2 target genes. Certainly, an identification of such pathways will be the essential step for the future. The feature of altered nuclear size with unchanged endoreduplication in mutants of stable NL components (like crwn1) and its associated members (like pwo1) provides a measurable phenotype for unbiased mutagenic screens (Goto et al., 2014).
Target-specificity of NL-chromatin bridging pathways is demonstrated also by the phenotypes of respective mutants. In animals, lack of bridging factors is frequently represented by a normal viability and only mild aberrations in standard conditions, with a substantial effect seen only at selected developmental stages or environments (Ozawa et al., 2006; Towbin et al., 2012; Gonzalez-Sandoval et al., 2015). Similarly, the pwo1 mutant shows a mild phenotype, with obvious changes known so far for one specific trait only (flowering time; Hohenstatt et al., 2018). As disruption of all PWO family members in pwo1/2/3 displays severe effect (seedling lethality; Hohenstatt et al., 2018), we cannot exclude that the mild phenotype in pwo1 single mutant is caused by the overlapping function of PWO family proteins. However, a nuclear area reduction seen already in pwo1 single mutant and the absence of a conserved CRWN1-interacting motif in PWO2 and PWO3 suggests that PWO1 has a NL-connected role separate from the other PWO family members. PWO1/2/3 are part of the recently discovered PEAT complex (Tan et al., 2018), and disruption of either the ARID, the TRB, or the EPL component results in early seedlings lethality, similar to pwo1/2/3 triple mutants (Hohenstatt et al., 2018; Tan et al., 2018). Thus, the fraction of PWO1, together with PWO2 and PWO3, present in the PEAT complex (which is not purified in the crude lamina extraction) is probably required for proper seedling growth whereas PWO1 association with NL components regulates nuclear size.
Furthermore, our results on PWO1 subnuclear localization showed formation of predominantly peripheral speckles in N. benthamiana and nonuniform subnuclear localization in Arabidopsis. Formation of distinguishable nucleoplasmic foci is a prominent feature of PcG proteins in animals, specifically PRC1 (Pirrotta and Li, 2012) and was reported to be dependent on PRC2 activity (Hernández-Muñoz et al., 2005). Moreover, animal H3K27me3 is found in large aggregates at the nuclear periphery when associated with late-replicating chromatin (Hernández-Muñoz et al., 2005). Similarly to animal models, PRC1 members in Arabidopsis were found to localize into multiple speckles as well (Libault et al., 2005; Calonje et al., 2008; Berry et al., 2017), albeit without any preference toward the nuclear periphery.
Our transcriptomic analyses, presented together with the nuclear size phenotype, highlight another important point. As reduced nuclear area causes elevated frequency of interchromosomal contacts (Grob et al., 2014), changes in the gene expression in pwo1 and crwn1/2 might reflect indirect effects coming from altered spatial organization of the chromatin. Thus, a generation and cross-comparison of transcriptomic data from the other mutants affected in nuclear size should be performed to uncouple effects of direct target regulation from the indirect effects.
In short, we provide a link between NL-like structures and chromatin repression in plants, highlighting an importance of nuclear architecture proteins in the control of gene expression across biological kingdoms. Deepening our understanding of PWO1 serves as a fascinating opportunity to decipher the mechanistic link between gene silencing and the nuclear periphery.
METHODS
Plant Material
Arabidopsis (Arabidopsis thaliana) seedlings were grown on 1/2-strength Murashige and Skoog plates in 16-h light/8-h dark conditions at 22°C (light intensity was 120 µmol/m2/s using Planox Eco 451178.009 LED bulbs [RZB]; long-day [LD] conditions) and transferred to soil, when later stages of development were needed. For the knock-out mutations, the following transgenic lines were used: pwo1-1 (SAIL_342_C09), crwn11 (SALK_025347), crwn1-1 crwn2-1 (SALK_025347, SALK_076653), and swn-7 (SALK_109121). For transient assays, Nicotiana benthamiana plants were grown in soil in LD conditions (22°C) up to the 4th week and subsequently used for infiltration. The sequences of oligonucleotides used for genotyping are as described in Wang et al. (2013) and .
Cloning
Regarding the vectors used for FRET-APB, pMDC7 derivatives containing GFP, mCherry, and GFP+mCherry coding sequences were created as described in Bleckmann et al. (2010). Next, genomic sequence of CRWN1 (AT1G67230), coding region of full PWO1 sequence (AT3G03140) or coding region of PWO1 C-terminal fragment were ligated into pCR8-GW-TOPO using the TA Cloning Kit (Thermo Fisher Scientific) to create entry vectors. To construct translational fusion plasmids, insert sequences from the entry vectors were introduced into pMDC7 derivatives via Gateway reaction from the Gateway cloning protocol (Thermo Fisher Scientific). For Y2H experiments, the coding sequences of CRWN1, PWO1, and PWO1 fragments were used to create entry vectors via TA Cloning and subsequently introduced into pGADT7 and pGBKT7 Y2H vectors (Clontech) using LR reaction (Thermo Fisher Scientific).
CoIP-MS/MS
Nuclear proteins were isolated from 3 g of whole seedlings from PWO1:PWO1-GFP or Col-0 (negative control) lines grown in LD conditions. Four biological replicates per genotype were used. Nuclei were extracted without prior fixation of the tissue as described in Smaczniak et al. (2012) and the further procedure was performed as described by Smaczniak et al. (2012). For immunoprecipitation, a µMACS GFP Isolation Kit (Miltenyi Biotec) was used. Peptide spectra were obtained with a model no. LTQ Orbitrap XL Mass Spectrometer (Thermo Fisher Scientific). Protein identification was performed by database search with the Andromeda search engine (Cox et al., 2011). Raw data processing was done with the software MaxQuant (v1.1.1.36; Cox and Mann, 2008). To distinguish PWO1-interacting proteins from the background, Student’s t test with a false discovery rate (FDR) adjustment was performed on label-free quantification values. For overlap between NL and PWO1 interactors, hits corresponding to chloroplastic genes were excluded from the data sets.
Subnuclear Localization and FRET-APB
For FRET-APB and examination of subnuclear localization in N. benthamiana, estradiol-inducible pMDC7 vector derivatives were transformed into Agrobacterium tumefaciens (GV3101 PMP90 strain with p19 silencing suppressor plasmid) and grown onYeast Extract Broth (YEB) medium plates for 2 d. Bacterial lawn was scraped into infiltration medium (5% [w/v] Suc, 0.1% [v/v] Silwet-L77, 450 µM of acetosyringone) to OD = 0.8 and kept on ice for 1 h. Bacteria were infiltrated into 4-week–old N. benthamiana leaves using 1-mL syringes without needles. After 48 h, an induction was performed by painting the abaxial side of the leaves with 20 to 50 μM of Beta-estradiol solution in 0.1% (v/v) TWEEN-20. At 16- to 20-h postinduction, small pieces of leaves around infiltration spots were cut, mounted on the microscopic slides, and subjected to confocal microscopy on a LSM780 (Zeiss) or SP8 (Leica). High-quality images were acquired with 3- to 4-line averaging and their brightness was adjusted in the software FIJI/ImageJ (National Institutes of Health).
The vicinity of speckles to the nuclear periphery or nucleolus was inferred from the distance between border of the speckle/the border of GFP nucleoplasmic signal and a nuclear/nucleolar rim in the brightfield channel. The vicinity was defined as the distance below the threshold of 0.5 μm, similarly to Poulet et al. (2017). The distance was measured using software FIJI/ImageJ. For FRET-APB, enhanced GFP (eGFP) was excited at 488 nm with an argon laser and mCherry at 561 nm with a helium laser. Photobleaching was performed using a 561-nm laser in five frames with 100% power on a model no. SP8 Confocal Microscope (Leica). FRET efficiency was calculated as described by Bleckmann et al. (2010). Subnuclear localization of candidate proteins in Arabidopsis was analyzed on 2-week–old seedlings of stable transgenic lines grown vertically on 1/2-strength Murashige and Skoog plates for conventional confocal microscopy and 6-d–old plants for AiryScan microscopy (Zeiss). Seedlings were transferred to the microscopic slides and cut around the hypocotyl, leaving the root on the slide. eGFP signals in roots were analysed using excitation at 488 nm with an argon laser. Images were acquired with 3- to 4-line averaging.
For superresolution imaging, the confocal laser scanning microscope model no. LSM880 (Zeiss) with AiryScan was used with a C-Apochromat 40×/1.2 Korr FCS M27 water immersion objective (Zeiss). The excitation wavelengths of 488 nm for eGFP and 561 nm for propidium iodide were used and emission was detected from 495 to 550 nm and 570 to 620 nm, respectively. The super-resolution mode of the AiryScan detector was used for acquisition at a pixel size of 0.04 μm and the processing of the acquired raw images was done using the standard 2D AiryScan filtering (Wiener filter associated with deconvolution at “auto”) in the software ZEN blue (Zeiss). Images from both channels were merged and only their brightness adjusted using the software ZEN blue. All samples from different tissues were treated the same during the processing.
Y2H
Yeast cultures were grown at 28°C on Yeast Peptone Dextrose (YPD) or on selective Synthetic Dextrose (SD) media. The AH109 strain was used for transformation, following protocol described in the Yeast Protocols Handbook (version no. PR973283 21; Clontech), with both Gal4-BD and Gal4-AD constructs added. Transformants were selected on SD medium lacking Trp and Leu (SD-LW). Protein interaction was assessed by transformant growth on selective, but low stringency SD media additionally lacking His (SD-LWH) permitting the identification of weak interactions (in Y2H).
Nuclear Morphology Analysis
For analysis, 0.5 g of 2-week-old seedlings were fixed in 10 mL of freshly prepared ice-cold 4 (w/v) % formaldehyde in PBS buffer for 20 min under vacuum. The seedlings were washed three times for 5 min in PBS buffer. After removal of PBS, the material was chopped on ice with a razor blade in 50 μL of Nuclear Isolation Buffer (NIB: 500 mM of Suc, 100 mM of KCl, 10 mM of Tris-HCl at pH 9.5, 10 mM of EDTA, 4 mM of spermidine, 1 mM of spermine, and 0.1% [v/v] Beta-mercaptoethanol). An additional 450 μL of NIB was added and cellular solution was filtered through 50-μm cell strainers (cat. no. 04-0042-2317; Partec). The filtrate was centrifuged at 500g, for 3 min, 4°C. The pellet was resuspended in 40 μL of NIB and 2 to 3 μL was spread on a microscopic slide, left to dry, and mixed subsequently with 4 µg/mL of DAPI solution (cat. no. 6335; Roth) in Vectashield mounting medium (cat. no. H-1000; Vector Laboratories). DAPI-stained nuclei were visualized in Z-stacks done on a model no. SP8 Confocal Microscope (Leica) using a 405-nm diode. Acquired Z-stacks were manually thresholded and used for nuclear shape and size measurements in the FIJI implementation of the software ImageJ2 (National Institutes of Health; Schindelin et al., 2012). Nuclear area and circularity index measurements were performed on Z plane showing the largest area of particular nucleus in the Z-stack. Circularity index was calculated as: 4π (area/perimeter^2), with a maximal value of 1.0 corresponding to the perfect circle. The measurements were done on 72 to 92 nuclei per genotype. Significance was calculated using one-way analysis of variance and Student’s t test.
Ploidy Analyses
Ploidy analyses were performed with 10-d–old whole seedlings (eight seedlings per sample) and cotyledons (12 per sample). Tissue was chopped in nuclei isolation buffer (15 mM of Tris; 2 mM of Na2-EDTA; 0.5 mM of Spermine; 80 mM of KCL; 20 mM of NaCl; 15 mM of β-Mercaptoethanol; 0.1% Triton X-100, pH-adjusted to 7.5), filtered through a 50-μm mesh, stained with Propidium iodide (0.005 mg/mL) and the DNA content was measured using a Fluorescence Activated Cell Sorter (Aria; BD Bioscience). Cycle value, which represents the mean number of endoreduplication cycles per nucleus, was calculated as in Barow and Meister (2003).
Transcriptomics
RNA was extracted from 2-week–old seedlings using the RNeasy Plant Mini Kit (cat. No. 74904; Qiagen), according to the manufacturer’s manual. RNA was treated with DNase and its quality was assessed on a Nanodrop 1000 (Thermo Fisher Scientific), by Qubit RNA BR Assay (cat. no. Q12210; Thermo Fisher Scientific) or using the RNA 6000 Nano Kit (cat. no. 5067-1511; Agilent) on a Bioanalyzer (cat. no. 2100; Agilent). First-strand cDNA was synthesized using the RevertAid First Strand cDNA Synthesis Kit (cat. no. K1622; Thermo Fisher Scientific) with oligo-d(T) primers, according to the manufacturer’s manual. cDNA was used for library preparation and sequencing on a HiSeq 2000 (Illumina). Alternatively, cDNA was used for RT-qPCR with the KAPA SYBR FAST Master Mix (cat. no. KK4600; Kapa Biosystems) on an iQ5 Detection System (Bio-Rad). Expression levels were calculated by application of ΔΔCt method. Sequences of oligonucleotides are shown in Supplemental Table 2. All the transcriptomic analyses were done in three or five independent biological replicates for RNA-Seq or RT-qPCR samples, respectively.
RNA-Seq Analysis
2 × 90-bp paired-end raw sequencing reads were scored by quality, then cleaned and trimmed from 5′ and 3′ ends in the softwares Trimmomatic (v0.35; http://www.usadellab.org/cms/?page=trimmomatic) and FastQC (v0.10; https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Clean reads were aligned to The Arabidopsis Information Resource 10 (TAIR10) reference genome by the software TopHat (v2.0.12; https://ccb.jhu.edu/software/tophat/index.shtml) with mate inner distance (−r) = 20 bp, segment length = 30 bp, and minimal intron length (−i) = 20 bp, according to Trapnell et al. (2009). Alignment files were processed in two different pipelines: Cufflinks (Tuxedo protocol; Trapnell et al., 2013) and EdgeR (Robinson et al., 2010). Fragments-per-kilobase of transcript-per-million mapped reads values were calculated in Cufflinks v2.2.1 with enabled: reference annotation (−g), multi read correction (−u), fragment bias correction (−b), and minimal intron length set to 20 bp. Cufflinks output was used to create common transcript reference file in the software Cuffmerge (v2.2.1.0; part of the Cufflinks suite) and perform DEG analysis in the software Cuffdiff (v2.2.1; part of the Cufflinks suite) with multi-read correction and fragment bias correction. For EdgeR, mapped reads were transformed into counts using HTSeq (v0.6.1; https://htseq.readthedocs.io/en/release_0.11.1/) and used as input for EdgeR (v3.3; https://bioconductor.org/packages/release/bioc/html/edgeR.html). Features with <1 count per million reads were discarded. For remaining features, differential expression was computed with adjusted P values < 0.05. DEG analysis in either of the pipelines concerned comparison between wild-type and mutant (pwo1 or crwn1/2) samples. To ensure the stringency of the analysis, only those genes that were present in DEG lists from both pipelines were taken for subsequent steps. The top DEGs in pwo1 were validated by RT-qPCR using five independent biological replicates (Supplemental Figure 7). The oligonucleotides used for the validation are depicted in Supplemental Table 2.Hierarchical clustering of the replicates (Supplemental Figure 8) was done based on the matrix containing “counts per million” values calculated in the software HTSeq. Features with <1 count per million reads were discarded. Heatmap with correlation between replicates was obtained in the software Heatmap3 (v1.1.1; https://www.rdocumentation.org/packages/heatmap3/versions/1.1.1) using default methods for distance computing and dendrogram reordering.
Secondary Bioinformatic Analyses
Gene ontology of DEGs was inferred from Singular Enrichment Analysis on the AgriGO server (Du et al., 2010). Statistical significance was calculated using the Fisher test with the Yekutieli adjustment method and the threshold of FDR < 0.01 was applied. Arabidopsis TAIR10 genes were used as a reference.
For cross-comparison of DEGs in different mutant backgrounds, Venn diagrams were created in the software VENNY v2.1 (Oliveros, 2007). Statistical significance of overlapping gene number was calculated in the software “R” using the hypergeometric test. The total number of genes was set according to TAIR10 annotation (protein coding genes exclusively). The total number of proteins was set according to TAIR8 annotation, as the same reference was used for peptide annotation in MS analysis.
Accession Numbers
Sequence data from this article can be found in EMBL/GenBank data libraries under the accession numbers at3g03140 (PWO1), at2g23380 (CLF), at4g02020 (SWN), at1g67230 (CRWN1), at1g13220 (CRWN2), at5g65770 (CRWN4), at5g04990 (SUN1). RNA-Seq data (raw reads and a processed file) were deposited in the Gene Omnibus under series number GSE106615.
Supplemental Data
Supplemental Figure 1. Z-stack montage of PWO1-GFP localization to speckles in N. benthamiana.
Supplemental Figure 2. Subcellular localization of different constructs’ combinations used for FRET-APB.
Supplemental Figure 3. PWO1 multiple alignment.
Supplemental Figure 4. Ploidy levels in pwo1.
Supplemental Figure 5. CRWN1 localization in pwo1/3.
Supplemental Figure 6. PWO1 localization in crwn1.
Supplemental Figure 7. RT-qPCR validation of RNA-Seq data for selected DEGs.
Supplemental Figure 8. Hierarchical clustering between samples in RNA-seq experiment.
Supplemental Table 1. List of PWO1 putative interactors.
Supplemental Table 2. Oligonucleotides for qPCR used in the study.
Dive Curated Terms
The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:
Acknowledgments
We thank Dr. Eric Richards (Boyce Thompson Institute, Ithaca, NY) for sharing crwn mutants and CRWN1-GFP lines, Dr. Ruediger Simon (Heinrich Heine University Düsseldorf, Germany) for providing FRET vectors, and Keygene N.V. (Wageningen, Netherlands) for the training in RNA-Seq analysis. AiryScan imaging (Zeiss) was performed at the Center of Advanced Imaging, Heinrich Heine University Düsseldorf. We thank Dr. Julia Kleinmanns, Dorota Komar, and Dr. Suraj Jamge for critical revision of the article. This work was supported by the German Academic Exchange Service (Deutsche Akademische Austauschdienst Dienst; grant to Kalyanikrishna), the European Commission 7th FP Project (ITN EpiTRAITS), the Boehringer Ingelheim Foundation (PLUS3 Program), the Deutsches Forschungsgemeinschaft (CRC973), and European Cooperation in Science & Technology COST Action (INDEPTH CA16212).
AUTHOR CONTRIBUTIONS
P.M., S.F., M.H., and D.S. designed the research; D.S., K.K., G.A., and M.H. designed the CoIP-MS/MS experiment; M.H. performed the CoIP-MS/MS experiment; C.S. analyzed MS spectral data; P.M., M.H., and K. performed nuclear morphology measurements; P.M. and S.F. performed microscopy analyses on PWO1 subnuclear localization in N. benthamiana and FRET analyses; P.M. did Y2H experiments, genetic interaction studies, subnuclear localization in Arabidopsis analyses, and transcriptomic experiments; Y.S. carried out superresolution imaging; P.M. wrote the article; D.S. and S.F. revised the article.
Footnotes
↵1 These authors contributed equally to this article.
↵2 Present address: Department of Cell and Developmental Biology, John Innes Centre, Norwich NR4 7UH, United Kingdom.
↵3 Present address: School of Natural Sciences, National University of Ireland, Galway H91 TK33, Ireland.
↵[OPEN] Articles can be viewed without a subscription.
- Received September 6, 2018.
- Revised February 27, 2019.
- Accepted March 22, 2019.
- Published March 26, 2019.