The plant nuclear lamina disassembles to regulate genome folding in stress conditions

The nuclear lamina is a complex network of nuclear lamins and lamin-associated nuclear membrane proteins, which scaffold the nucleus to maintain structural integrity. In Arabidopsis thaliana, nuclear matrix constituent proteins (NMCPs) are essential components of the nuclear lamina and are required to maintain the structural integrity of the nucleus and specific perinuclear chromatin anchoring. At the nuclear periphery, suppressed chromatin overlapping with repetitive sequences and inactive protein-coding genes are enriched. At a chromosomal level, plant chromatin organization in interphase nuclei is flexible and responds to various developmental cues and environmental stimuli. On the basis of these observations in Arabidopsis, and given the role of NMCP genes (CRWN1 and CRWN4) in organizing chromatin positioning at the nuclear periphery, one can expect considerable changes in chromatin–nuclear lamina interactions when the global chromatin organization patterns are being altered in plants. Here we report the highly flexible nature of the plant nuclear lamina, which disassembles substantially under various stress conditions. Focusing on heat stress, we reveal that chromatin domains, initially tethered to the nuclear envelope, remain largely associated with CRWN1 and become scattered in the inner nuclear space. By investigating the three-dimensional chromatin contact network, we further reveal that CRWN1 proteins play a structural role in shaping the changes in genome folding under heat stress. Also, CRWN1 acts as a negative transcriptional coregulator to modulate the shift of the plant transcriptome profile in response to heat stress.

Article https://doi.org/10.1038/s41477-023-01457-2 initial experiment because of the availability of published data documenting changes in transcriptome, chromatin structure and genome organization during heat-stress acclimation 29,32 , which together implied possible dynamic chromatin positioning at the nuclear periphery. To address this question, we selected a few genomic loci belonging to plant nuclear lamina-associated domains (PLADs), and examined their localization in heat-stressed and control plants (Fig. 1a). In control plants (that is, without heat-stress treatment), fluorescent in situ hybridization (FISH) probes recognizing PLAD loci showed preferential localization near the nuclear envelope; on the contrary, such perinuclear positioning was lost in heat-stressed plants (Fig. 1b,c and Supplementary Table 1). In addition, our FISH experiment revealed a higher incidence of chromatin decondensation of the probed genomic regions in stress-treated plants, which was similar to the changes in centromeric repeats triggered by heat 29 (Supplementary Fig. 1). We also observed similar changes in chromatin localization when FISH was applied to a larger genomic region (~1.3 megabase pairs (Mb)), which mainly belonged to PLADs 20 . Of all the nuclei that we randomly sampled (n > 50), over 80% from mock-treated plants showed distinct FISH signals at the nuclear periphery; whereas over 80% from heat-stressed plants showed apparent chromatin decondensation along with considerable detachment from the nuclear envelope (Fig. 1d).
Overall, these observations agree with a recent study showing global rearrangement of chromatin organization induced by heat 32 , and further imply that it involves the dissociation of PLADs from the nuclear lamina. This idea prompted us to examine the potential dynamic interactions between PLAD loci and CRWN1 by using chromatin immunoprecipitation (ChIP). The CRWN1 ChIP was performed with a CRWN1:2HA tagging line, in which the fusion protein between CRWN1 and a tandem HA (haemagglutinin) tag could fully rescue crwn1 loss-of-function phenotypes 20 . We found enrichment of the selected PLAD loci by CRWN1:2HA in both control and heat-stressed plants. To our surprise, heat treatment did not weaken the interactions between CRWN1 and these loci (Fig. 1e). Since PLAD loci lost their perinuclear localization in heat-stressed plants, we proposed that the CRWN1 protein, and perhaps other components of the nuclear lamina, acquired alternative nuclear localization patterns.

Responses of the plant nuclear lamina to stress conditions
In Arabidopsis, two NMCP proteins, CRWN1 and CRWN4, as well as KAKU4, are known to be involved in establishing the plant nuclear lamina 8,12,14 . To address whether the nuclear lamina acquires an alternative organization under heat stress, we examined the localization of these proteins. In total, we analysed CRWN1:2HA, CRWN4:2HA and KAKU4:GFP 14 . We also examined proteins that are located at the nuclear periphery, but are not directly involved in assembling the nuclear lamina. The selected proteins were NUP1:GFP, which was a nuclear basket protein belonging to the nuclear pore complex, and SUN1:GFP, which was an integral protein located in the inner nuclear membrane 17,33,34 . We found that nuclear lamina proteins (that is, CRWN1, CRWN4 and KAKU4) lost their specific perinuclear location during heat stress; whereas SUN1 and NUP1 did not (Fig. 2a-c and Extended Data Fig. 1). The detachment of these proteins from the nuclear periphery was not due to degradation, since protein degradation or cleavage products were undetectable (Fig. 2d). In fact, CRWN4 appeared with higher abundance after heat treatment (Fig. 2d). These findings indicate that the plant nuclear lamina disassembles during the heat-stress response, along with relocalization of its multiple components into the nuclear interior.
Previous functional studies on Arabidopsis nuclear lamina component genes indicated that they participated in regulating nuclear size and shape 7,8,14 . Therefore, we asked whether the disassembly of nuclear lamina during heat shock results in concomitant changes in nuclear morphology, resembling those observed in lamin mutant plants. Compared to control plants, heat-stressed plants surprisingly maintained nuclear shape and size ( Supplementary Fig. 2). Statistical analyses on are highly conserved in the plant kingdom, but they do not share any sequence homology with animal nuclear lamins. Amongst the NMCP(s), in each plant species studied so far, there is at least one showing preferential localization at the nuclear periphery 2,4-12 . While the sole NMCP homologue in the basal land plant Marchantia appears to be dispensable for vegetative growth 10 , NMCP genes in Arabidopsis are essential for plant viability 13 . Genetic crosses revealed functional redundancy of Arabidopsis NMCP genes, which are also named CRWNs (CROWDED NUCLEI). In an earlier study by Wang and colleagues, it was found that quadruple crwn mutants, as well as some triple crwn mutants, could not be recovered from a segregating population, indicating the requirement of a minimum level of CRWN activities to complete the plant life cycle 13 .
Arabidopsis CRWN1 and CRWN4 proteins are located at the nuclear periphery. Mutant plants lacking these CRWN proteins develop spherical nuclei that are noticeably smaller than those in wild-type plants 7,8 . On the other hand, overexpression of CRWN1 promotes nuclear deformation, leading to the formation of ring-like and bleb-like structures on the nuclear envelope 14 . Transcriptomic analyses of crwn single and double mutants revealed that the gene expression profile of these mutants was featured with ectopic activation of pathways related to stress responses 15 . Several transcription factors and coregulators were found to interact with CRWN1, suggesting the involvement of the plant nuclear lamina in regulating gene expression 16,17 . Reminiscent of the crwn mutant transcriptome, mutants of lamin-interacting membrane proteins in the PNET2 family also show activated stress pathways 18 .
The Arabidopsis nuclear lamina selectively interacts with chromatin regions, creating a non-random chromatin distribution pattern at the nuclear periphery where repressed chromatin is enriched 19,20 . Recent work showed that CRWN1 and CRWN4 help in establishing a scattered distribution of centromeres in the nucleus 21 . At a chromosomal level, plant chromatin organization in interphase nuclei displays flexibilities. Many developmental cues and environmental factors, such as dedifferentiation 22 , leaf development 23 , seedling growth 24 , floral transition 25 , seed development 26 , light intensity 23,27 , microbial infection 28 and temperature stress 29 , can trigger global rearrangement of chromatin, demonstrating a tight connection between the structural arrangement of chromatin and its activities. In addition, scattered evidence implies that the plant nuclear lamina might undergo active turnover 16,30 . However, given the role of NMCP genes (that is, CRWN1 and CRWN4) in organizing chromatin at the nuclear periphery 20 , it remains unclear whether the chromatin-nuclear periphery interactions are dynamic when the global chromatin organization patterns are being altered in plants.
In this study, we show that plant nuclear lamina disassembles substantially following several types of abiotic stress treatment, including heat stress. During the heat-stress response, chromatin domains, initially tethered to the nuclear envelope, remained largely occupied by CRWN1 and become scattered in the inner nuclear space. We further show that during heat stress, CRWN1 proteins function as structural factors to guide the genome to adopt an alternative folding conformation. Also, CRWN1 acts as a negative transcriptional coregulator, modulating transcriptional reprogramming in response to heat stress.

Changes in chromatin positioning under heat stress
Our recent work on CRWN1 and CRWN4 indicated that they were required for establishing specific perinuclear chromatin localization patterns 20 . At the nuclear periphery, CRWN1 proteins preferentially interact with chromatin regions having low accessibility, which overlap with inactive protein-coding genes and transposons 20 . Given the fact that plants show extensive chromatin-based transcriptional reprogramming during stress responses 31 , we asked whether stress could trigger changes in the chromatin-nuclear lamina interaction patterns. Of many widely studied abiotic stress conditions, we chose heat stress in our Article https://doi.org/10.1038/s41477-023-01457-2 nuclear sizes revealed a clear difference in nuclear morphology between heat-stressed wild-type and crwn1 plants, indicating that nuclear lamin proteins, albeit having distinct nuclear localization under heat stress, are able to maintain nuclear structure ( Supplementary Fig. 2).
We further asked if other stress conditions were able to trigger similar changes to the nuclear lamina. By treating plants with salt or osmotic stresses, we also observed remarkable detachment of CRWN1, CRWN4 and KAKU4 from the nuclear periphery (Fig. 2a-c). On the    contrary, SUN1 and NUP1 retained their perinuclear localization under these stress conditions (Extended Data Fig. 1). Interestingly, after prolonged stress exposure, for example, in seedlings continuously facing 50 mM NaCl stress from onset of germination, the abundance of CRWN1 dropped and the remaining protein was distributed throughout the nucleoplasm (Extended Data Fig. 2). Taken together, these results reveal the dynamic nature of the Arabidopsis nuclear lamina; upon perceiving different stress conditions, it disassembles and releases its component proteins into the nuclear interior.

Dynamic lamin-chromatin interactions under heat stress
Following up on our findings that PLAD chromatin remained in direct contact with CRWN1 proteins after the disassembly of the nuclear lamina in heat response (Figs. 1 and 2), we sought to gain a better view on the global interactions between chromatin and CRWNs in this process. To this end, we performed ChIP-seq experiments by using the CRWN1:2HA and CRWN4:2HA tagging lines in their respective mutant genetic backgrounds (as mentioned in the previous section). The CRWN4:2HA tagging line carried a genomic CRWN4 sequence fused with tandem HA repeats, which could fully rescue the nuclear morphology phenotypes (smaller and round nuclei) in crwn4 ( Supplementary Fig. 3). Principal component analysis of ChIP-seq sample reads distribution patterns revealed noticeable changes in CRWN1-chromatin interactions before and after heat stress, but not in CRWN4:2HA-chromatin interactions ( Supplementary Fig. 4). In addition, close examination of ChIP-seq signals across individual chromosomes indicated that Article https://doi.org/10.1038/s41477-023-01457-2 CRWN1-chromatin interactions were much more prominent compared to CRWN4 (Fig. 3a, Extended Data Fig. 3 and Supplementary Data 1). The majority of CRWN4:2HA-bound chromatin, which was also enriched with CRWN1:2HA, was located in centromeric and pericentromeric regions (Fig. 3b). We also observed that heat stress influenced CRWN1-and CRWN4-chromatin interactions differentially. Compared to control plants, heat-stressed plants showed stronger CRWN1:2HA-chromatin interactions in pericentromeric regions, but the interactions in chromosome arms became weaker ( Fig. 3a and Extended Data Fig. 3c). Such a shift of CRWN1:2HA-chromatin interaction patterns suggests that the decondensation of pericentromeric regions, triggered by heat stress, promotes the formation of new contacts with CRWN1:2HA. A similar amount of genomic regions were enriched with CRWN1 before and after heat stress (

CRWN1-chromatin interactions mediate gene expression control
Next, we sought to explore whether dynamic CRWN1:2HA-chromatin interactions are associated with specific features of gene expression. Under both control and heat-stress conditions, CRWN1:2HA showed a negative association with gene expression (Fig. 4a). Among all the genes bound by CRWN1:2HA, silenced or weakly expressed genes exhibited stronger ChIP-seq signals than highly expressed ones, and the differences in ChIP-seq signal strength were located around the 5′ end of genes (Fig. 4a). This observation is in line with our previous report showing that CRWN1-chromatin interactions are negatively correlated to chromatin accessibility, which was a universal feature of the 5′ end of actively expressed genes 20 . We also analysed expression changes of genes, which showed dynamic changes in CRWN-chromatin interactions under heat stress. According to the annotated chromatin domains enriched by CRWN1 under mock (M) and heat (H) conditions, we refer to genes overlapping with them as 'M-' or 'H-specific' or 'MH' (Supplementary Data 1). Compared to genes bound by CRWN1:2HA in both conditions (that is, MH), those losing their interactions during heat stress (that is, M-specific) tended to show elevated expression (Fig. 4b).
Among those M-specific genes, 57 and 34 showed significant up-and downregulation under heat stress, respectively (Supplementary Data 2). Of the 57 upregulated genes, three (AT3G46230, AT4G10250 and AT4G21320) were annotated as closely linked to heat response. The upregulation of these three genes under heat stress was confirmed by reverse transcription quantitative real-time PCR (RT-qPCR), and ChIP-qPCR experiments confirmed substantial decrease in CRWN1chromatin interactions at two of the three loci (Extended Data Fig. 4a-c). Similarly, of the genes in the MH group, those annotated as upregulated differentially expressed genes (DEGs) showed attenuated interactions with CRWN1:2HA under heat stress (Fig. 4c). These results reveal a correlation between the dissociation of CRWN1 from chromatin and gene upregulation during the heat-stress response. We next asked whether CRWN1/4 target genes could be differentiated from non-target genes according to transcriptional activities. However, by comparing changes in gene expression, we did not observe any notable difference between these two groups of genes (Extended Data Fig. 4d). Furthermore, CRWN1/4 target genes did not show a systematic shift towards up-or downregulation in crwn1 crwn4 mutants, implying that CRWN1/4-chromatin interactions alone are not sufficient to drive gene expression change (Extended Data Fig. 4d). It is also possible that the complete loss of CRWN1 and CRWN4 proteins in the double mutant plants led to the establishment of new networks among chromatin and other nucleoskeleton proteins (for example, CRWN2 and CRWN3), masking gene expression changes associated with dynamic CRWN1/4chromatin interactions in wild-type plants.

CRWNs mediate chromatin organization under heat stress
The Arabidopsis CRWN1 and CRWN4 have recently been shown to modulate perinuclear chromatin positioning and pericentromere structure 20,21 . We applied the Hi-C method for genome-wide detection of chromatin interactions to examine whether CRWN1 and CRWN4 play a structural role in mediating the changes of chromatin organization under heat stress 32 . In total, Hi-C maps of wild type and crwn1 crwn4 under control and heat-stress conditions were generated ( Fig. 5 and Extended Data Fig. 5). In line with recent studies, we observed a reduction of chromatin contacts within pericentromeres as well as reduction of many distal intrachromosomal contacts in heat-stressed wild-type plants (Extended Data Fig. 6a). Under heat stress, PLADs exhibited reduced intrachromosomal contacts (Extended Data Fig. 6b), which agrees with our observation from FISH that such a stress condition resulted in the dispersal of PLADs in nucleoplasm and reduction of chromatin contact strength among them (Fig. 1d). Consistent with a recent report by Sun and colleagues, as shown in Extended Data Fig. 6a, we also observed that heat stress caused a remarkable reduction of contact amongst KNOT ENGAGED ELEMENTS (KEEs), which had strong intra-and interchromosomal interactions 32,35 . A closer examination of KEEs in all Hi-C maps revealed that KEE patterns were highly similar in wild type and crwn1 crwn4 under the same growth condition, suggesting that CRWNs do not play a structural role in these genomic regions (Extended Data Fig. 6c).
The crwn1 crwn4 mutant Hi-C map revealed changes in pericentromeric regions (Extended Data Fig. 7), which is consistent with a recent report showing stronger chromatin contacts within and among individual pericentromeres 21 . Analysis of Hi-C data allows one to recognize chromatin regions that are spatially linked to or separated from each other. In most cases, a binary annotation strategy is applied to group chromatin regions into two spatial compartments 36 . In the nucleus including Arabidopsis, these two spatial chromatin compartments (named 'A/B compartment' when it was first applied to human Hi-C data) differ in gene expression and epigenetic landscape [35][36][37] . For chromosome arm regions, we found that neither heat stress nor the loss of CRWNs resulted in drastic changes in the chromatin compartmentalization identity, but they both enhanced chromatin contacts between different spatial compartments ( Fig. 5b-d). These changes appeared strongest in heat-stressed crwn1 crwn4. Taking chromosome 1 right arm as an example, among all the plant samples, the Hi-C map of heat-stressed crwn1 crwn4 showed the least correlation between spatial compartments (Fig. 5a). Furthermore, CRWN1 displayed preferential enrichment of chromatin binding with the B compartment, and chromatin in heat-stressed crwn1 crwn4 showed noticeably the weakest compartmentalization, indicating that CRWNs play a structural role in regulating genome organization under heat stress (Figs. 5d and 6a).
Further support for a structural role of CRWNs in modulating chromatin organization under heat stress was provided by the observation that the heterochromatic pericentromeres acquired more chromatin contacts with chromosome arms under heat stress. Concomitant changes in folding conformation could be reflected by the interaction decay exponents (IDEs) describing chromatin interaction as a function of loci distance (Fig. 6b and Extended Data Fig. 6a) 32 . For each chromosome, heat stress resulted in steeper decrease in chromatin contacts Article https://doi.org/10.1038/s41477-023-01457-2 within the pericentromere at long distances, indicating that this stress condition led to heterochromatin decondensation (Fig. 6b). Accordingly, further inspection of Hi-C maps suggested that the decondensed pericentromeres acquired more chromatin contacts with chromosome arms (Fig. 6c, highlighted with dotted-line boxes). Interestingly, the genomic regions that located in chromosome arms and showing stronger chromatin contacts with pericentromeres under heat stress tended to be associated with CRWN1 binding (Fig. 6d, black curves). For crwn1 crwn4 under heat stress, compared to wild-type plants, pericentromeres showed decreased chromatin contacts with chromosome arms, which was strongly correlated with CRWN1 ChIP-seq signals (Fig. 6d, red curves, and Extended Data Fig. 8). Taken together, these results indicate that the dynamic CRWN1-chromatin interactions participate in shaping genome organization under heat stress.

Discussion
In this work, we present data highlighting the dynamic nature of the plant nuclear lamina in response to various stress conditions, such as heat, salt and osmotic stresses (Fig. 2). Since multiple nuclear lamina components interact with each other 14,38 , one would speculate that the assembly of the nuclear lamina relies on the presence of all these proteins. However, the localization of single lamin protein species at the nuclear periphery turned out to be independent of other nuclear lamina components. For example, KAKU4 protein localization remains at the nuclear periphery in crwn1 crwn4 double mutant 14 . These results prompted us to speculate that the nuclear localization patterns of lamin proteins are largely independent of each other, and during stress adaptation their disassembly was triggered in parallel. To gain a better understanding on how the plant nuclear lamina responds to stresses, it will be intriguing to further identify which parts or which features of individual lamin proteins enable their dynamic behaviour.
To our surprise, the disassembly of the Arabidopsis nuclear lamina under heat stress does not lead to apparent changes in nuclear size and shape, albeit multiple studies have clearly demonstrated that nuclear lamin components, for example, CRWN1, CRWN4 and KAKU4, are indispensable for proper nuclear morphology 7,8,14 . Although we could not verify in our experiments whether the nuclear periphery was free of nuclear lamina under stress, our observation strongly suggests that the localization of tested nuclear lamin components at the nuclear periphery is not necessary to maintain nuclear morphology. Nuclear size in animals is known to be controlled at multiple phases of the cell cycle, including the stage of building the nuclear envelope   in newly formed daughter cells (recently reviewed by Cantwell and Dey 39 ). Our results imply that CRWNs and KAKU4 proteins play their structural roles concerning nuclear morphology before interphase. During interphase they actively participate in mediating stress responses. In line with this notion, both CRWN1 and CRWN4 assemblies were observed at telophase, when the nuclear envelope reconstruction starts 8 .
The dynamics of the Arabidopsis nuclear lamina suggests several future exciting avenues for further research, concerning the underlying regulatory mechanisms, spatial chromatin organization and transcriptional regulation. First, multiple studies on CRWN1 indicate that the plant nuclear lamin component responds to stresses in a highly variable manner. CRWN1 is degraded after infection with virulent bacteria, which is probably mediated by the salicylic acid signalling pathway 16 . Article https://doi.org/10.1038/s41477-023-01457-2 On the other hand, it was shown by Sakamoto and colleagues that CRWN1 proteins remained stably localized at the nuclear periphery in roots when plants were stressed with excess copper ions 38 . Here, we reveal that heat-stress treatment causes CRWN1 (as well as other nuclear lamina components CRWN4 and KAKU4) to be redistributed to the nucleoplasm in the absence of protein degradation (Fig. 2). It is unknown how CRWN1 proteins respond differently to these stress stimuli. However, on the basis of the available information on metazoan nuclear lamins 40  which could be useful in elucidating how the nuclear lamina reacts differently to diverse stimuli. Second, plant nuclear lamina components under various stress conditions may interact with diverse genomic regions. In this study, we focused on CRWN1 and found noticeable gain and loss of CRWN1-chromatin interactions during the nuclear lamina disassembly (Fig. 3c), which correlated to changes in gene expression (Fig. 4). Accompanied by changes in spatial distribution, the nuclear lamin proteins may switch to interact with new protein partners. A recent proteomic approach using biotin proximity labelling indicated that KAKU4 interacts with nucleosomes (presumably from chromatin) at the nuclear periphery 18 . Since KAKU4 proteins detached from the nuclear envelope after perceiving stresses (Fig. 2), we speculate that KAKU4 might bind to different genomic loci before and after the nuclear lamina disassembly. In addition, whether CRWN2 and CRWN3, which are homologues of CRWN1 but located throughout the nucleoplasm 8,13 , interact directly with chromatin remains unknown. Nevertheless, we propose that this would be the case since CRWN1 was reported to interact with all its homologues (that is, other CRWNs) in vivo 38 .
Third, given a strong correlation between the localization of chromatin at the plant nuclear periphery and suppressed gene expression 19,20 , it is of great interest to explore to what extent changes in the nuclear lamina at the nuclear periphery are connected to transcriptional regulation in this nuclear compartment. It has been noted that a few genes showed propensity to be repositioned to the nuclear periphery along with transcriptional activation/upregulation, examples of which include copper-associated gene in response to copper toxicity and the CAB gene cluster (encoding chlorophyll a/b-binding proteins) in response to red/far-red light 38,43 . Is such transcriptional regulation associated with changes at the nuclear periphery so that it favours transcription of these genes? Will the disassembly of the plant nuclear lamina convert the nuclear periphery to a place harbouring more active euchromatin? Although such a scenario has not been described in plants, case studies focusing on rod photoreceptor nuclei from nocturnal animals show that withdrawal of lamin-chromatin

Plant materials and growth conditions
Surface-sterilized Arabidopsis seeds were sowed on vertical half-strength Murashige and Skoog media plates containing 1% sucrose and 0.3% phytagel. After being stratified at 4 °C for 3 d, the plant materials were grown in a growth chamber (MLR-352-PE from PHCbi) set at 21 °C and long day (16 h light/8 h dark) conditions. Mutants used in this study were crwn1-1 (SALK_025347), crwn4-1 (SALK_079296) and kaku4-2 (SALK_076754), which were ordered from the Nottingham Arabidopsis Stock Centre. For heat-stress treatment, 11-day-old plants were transferred to another growth chamber of the same model and with identical settings, except that the temperature was set at 37 °C when lights were switched on. Plants were collected three days later for analyses. For salt and osmotic stress treatment, 12-day-old plants were transferred to half-strength Murashige and Skoog plates supplemented with 150 mM sodium chloride or 5% PEG 8000, respectively. Plants were collected two days later for analyses.
Plant nuclei were extracted essentially as described in our earlier work 46 . The subsequent nuclei sorting was performed with a S3e Cell Sorter (Bio-Rad) according to our established protocol 10 . The extracted nuclei were stained with 0.5 µM DAPI to reveal their ploidy levels; if not otherwise specified, only 2C nuclei were collected for downstream experiments.
For the KAKU4:GFP construct, the genomic region containing 2 kb upstream of KAKU4 and its coding sequence was amplified with oligonucleotides 5′-GCATAGAACG AGGAATACAGG-3′ and 5′-CTGCCTCCTGCAGCTCCGGATT TGGCCCGTCCTTTGCCTC-3′, and the GFP complementary DNA was amplified with oligonucleotides 5′-TCCGGAGCT GCAGGAGGCA GCGCGGCCGCT GTGAGCAAGGG-3′ and 5′-TTATCCGGACT TGTACAGCTCG-3′. These two PCR fragments were purified and cloned into the pFK206 vector with a Gibson assembly reaction, by which the GFP sequence was appended to the C terminus of KAKU4.

FISH and immunohistostaining
Bacterial artificial chromosome (BAC) probes were generated by the nick translation DNA labelling system (Roche, catalogue no. 11745808910). The green and red probes described in this study were labelled by digoxigenin (DIG) and 2,4-dinitrophenyl (DNP), respectively. The labelled BACs were pooled in a hybridization mix (50% formamide, 10% dextran sulfate, 2× SSC, 50 mM sodium phosphate (pH 7.0)). The working concentration of each labelled BAC was 1 ng µl −1 . Individual BACs are listed in Supplementary  Table 1. FISH experiments were performed according to ref. 48. Briefly, around 5,000 nuclei in 20 µl PBS buffer were incubated at 65 °C for 30 min. The nuclei were subsequently mixed with 10 µl of 0.1 mg ml −1 RNase A and spread into a circle drawn on a glass slide using an ImmEdge pen. After incubating the slides at 37 °C in a hybridizer (ThermoBrite, model 07J91-020) for 1 h, the slides were dipped up and down for 1 min each in a graded ethanol series (30%, 60%, 80%, 90%, 95%, 100% EtOH) for dehydration. For immunohistostaining, an antigen retrieval step was performed by incubating the slides in boiling solution (10 mM sodium citrate at pH 6.0) for 12 min in a microwave oven at 700 W. After antigen retrieval, slides were postfixed in 4% formaldehyde solution for 10 min, followed by dehydration in a graded ethanol series and then air-dried. The treated slides were used for FISH or immunohistostaining experiments. For FISH experiments, the probe hybridization, slide washing and signal detection steps were performed according to our published protocol with minor changes 19 . DIG-labelled probes were detected with 1:10 diluted DIG Alexa Fluor 488-conjugated mouse antibody (Biotechne, catalogue no. IC7520G), and DNP-labelled probes were detected with 1:500 diluted DNP rabbit antibody (Ther-moFisher, catalogue no. 04-8300) and 1:150 diluted anti-rabbit Alexa Fluor 546-conjugated goat antibody (ThermoFisher, catalogue no. A-11035). For immunohistostaining, the HA-tagged protein of interest was detected with 1:500 diluted HA tag Alexa Fluor 647 conjugated mouse antibody (ThermoFisher, catalogue no. 26183-A647). After antibody incubation, the slides were washed with 4× SSC 0.2% Tween 20 in a foil-wrapped jar at room temperature, three times for 5 min each. Finally, slides were mounted with 5 µl SlowFade Diamond Antifade Mountant (Invitrogen, catalogue no. S36964).

Microscopy and image processing
Confocal images were captured with a Zeiss LSM 700 system. For FISH experiments, a single image was taken from the central focal plane of individual nuclei. Image analyses were performed with ImageJ 49 . The distance between a FISH signal spot and the nuclear periphery was approximated as the distance between its estimated barycentre and the edge of the DAPI staining. We noticed that a fraction of the observed nuclei showed split or scattered FISH signal patterns, making it difficult to estimate the distance between the probed genomic region to the nuclear envelope ( Supplementary Fig. 1). These nuclei were excluded from the analysis. During FISH image acquisition, we also excluded those nuclei in which FISH signals were apparently located close to the top or the bottom of nuclei, because they would largely mislead distance calculations. These nuclei could be easily recognized by moving the sample stage along the z axis to reveal that the focal plane containing FISH signals could only capture DAPI signals with a smaller area (Supplementary Fig. 5a). For immunostaining, an image of the central focal plane of a nucleus was taken to analyse the distribution of the protein of interest. Images of chromosome painting were acquired with a Zeiss LSM 880 system. Quantification of nuclear size was performed with images of DAPI-stained nuclei taken with an Olympus IX83 fluorescence microscope. The Olympus cellSens software (v.3.10.12201.0) was used to measure the area of nuclei in the images, which was used as an approximation to nuclear size.

Chromatin immunoprecipitation and library sequencing
Arabidopsis shoots were fixed under vacuum for 30 min with 1% formaldehyde in MC buffer (10 mM potassium phosphate, pH 7.0, 50 mM NaCl, 0.1 M sucrose) at room temperature. Fixation was terminated by replacing the solution with 0.15 M glycine in MC buffer under vacuum for 10 min at room temperature. Approximately 1 g of fixed tissue was homogenized and resuspended in nuclei isolation buffer (20 mM HEPES, pH 8.0, 250 mM sucrose, 1 mM MgCl 2 , 5 mM KCl, 40% glycerol, 0.25% Triton X-100, 0.1 mM PMSF, 0.1% 2-mercaptoethanol) and filtered with double-layered miracloth (Millipore). Isolated nuclei were resuspended in 0.5 ml sonication buffer (10 mM potassium phosphate, pH 7.0, 0.1 mM NaCl, 0.5% sarkosyl, 10 mM EDTA), and chromatin was sheared by sonication with a QSONICA sonicator Q800R3 to achieve average fragment size around 400 base pairs (bp). Next, 50 µl 10% Triton X-100 was mixed with the sonicated sample, and 25 µl of the mixture was saved as input sample. The rest of the sheared chromatin was mixed with an equal volume of immunoprecipitation (IP) buffer (

Gene expression analyses
RNA sequencing (RNA-seq) was performed with two biological replicates per sample. Total RNA was extracted from aerial parts of seedlings using a RNeasy Plant Mini Kit (Qiagen). RNA-seq library preparation was performed as previously described 10 . RNA-seq sequencing reads were aligned against the Araport11 annotation using TopHat2 (v.2.1.1) with default parameters, and were further assigned to genes using the R package GenomicAlignments 51-53 . Differentially expressed genes were identified with the R package DESeq2 (ref. 54). We used criteria of false discovery rate smaller than 0.01 and log 2 (expression fold change) more than 1.6 to call upregulated and downregulated genes. Details of the reads count table, gene expression measurement (in reads per kilobase per million mapped reads) and differentially expressed genes can be found in Supplementary Data 2.
For individual gene expression analysis, total RNA was isolated from samples using the RNeasy Plant Mini Kit (Qiagen) following the manufacturer's instructions. The RNA was then treated with DNase I (Thermo Scientific) to remove any contaminating DNA. Subsequently, the RNA was reverse transcribed to cDNA using SuperScript II Reverse Transcriptase (Invitrogen). Quantitative RT-PCR was conducted on a CFX96 real-time system (Bio-Rad) using qPCRBIO SyGreen Mix ROX (Lo-ROX) (PCR Biosystems) with gene-specific primers listed in Supplementary Table 2.

In situ Hi-C
In situ Hi-C libraries were prepared essentially as previously described 55 . In total, two Hi-C library replicates for each sample were made, and for each replicate around 0.5 g of fixed sample was homogenized for nuclei isolation. Nuclei were resuspended with 150 µl 0.5% SDS and split into three tubes. After penetration at 62 °C for 5 min, SDS was quenched by adding 145 µl water and 25 µl 10% Triton X-100, and incubated at 37 °C for 15 min. Subsequently, chromatin was digested overnight at 37 °C with 50 U DpnII (NEB) in each tube. The next day, DpnII was inactivated by incubating at 62 °C for 20 min. Then, sticky ends were filled in by adding 1 µl of 10 mM dTTP, 1 µl of 10 mM dATP, 1 µl of 10 mM dGTP, 10 µl of 1 mM biotin-14-dCTP, 29 µl water and 40 U Klenow fragment (Thermo Fisher), and incubated at 37 °C for 2 h. After adding 663 µl water, 120 µl 10× blunt-end ligation buffer (300 mM Tris-HCl, 100 mM MgCl 2 , 100 mM DTT, 1 mM ATP, pH 7.8) and 40 U T4 DNA ligase (Thermo Fisher), proximity ligation was carried out at room temperature for 4 h. Then, three tubes of ligation products were centrifuged, and nuclei pellets were resuspended and combined with 650 µl SDS buffer (50 mM Tris-HCl, 1% SDS, 10 mM EDTA, pH 8.0). After treatment with 10 µl proteinase K (Thermo Fisher) at 55 °C for 30 min, de-cross-linking was performed by adding 30 µl 5 M NaCl and incubating at 65 °C overnight. DNA was recovered and subsequently treated with RNase A at 37 °C for 30 min. After purification, 3-5 µg Hi-C DNA was topped to 130 µl with TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and sheared with a Q800R3 sonicator (QSONICA) by using the following setting: 25% amplitude, 15 s ON, 15 s OFF, pulse-on time for 4.5 min, to achieve fragment size shorter than 500 bp. Sonicated DNA was purified with Ampure beads to recover fragments longer than 300 bp. Then, with a 50 µl reaction volume, the DNA was mixed with 0.5 µl 10 mM dTTP, 0.5 µl 10 mM dATP and 5 U T4 DNA polymerase and incubated at 20 °C for 30 min to remove biotin from unligated DNA ends. After that, the DNA was purified with Ampure beads, and continued with end repair and adaptor ligation using the NEBNext Ultra II DNA Library Prep Kit (NEB). Ligated DNA was affinity purified with Dynabeads MyOne Streptavidin C1 beads (Invitrogen) as described 56 , and further amplified with 12 PCR cycles. The libraries were sequenced on an Illumina Novaseq instrument with 2 × 150 bp reads.
Reads mapping to the TAIR10 genome with Bowtie 2 (v.2.2.4), removal of PCR duplicates and reads filtering were performed as previously described 56 . Hi-C reads of each sample are summarized in Supplementary Data 3. Hi-C map normalization was performed by using an iterative matrix correction function in the 'HiTC' package in the R program 57 . For all Hi-C maps, the iterative normalization process was stopped when the eps value, which reflected how similar the matrices in two consecutive correction steps were, dropped below 1 × 10 −4 . The bin size setting for genome-wide Hi-C maps was 20 kb. In addition, the filtered Hi-C reads were used to create hic files with the juicer tool for interactive Hi-C map inspection 58

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
Short read data of in situ Hi-C, ChIP-seq and RNA-seq are publicly available at NCBI Sequence Read Archive under accession number PRJNA870030. Large datasets, such as normalized Hi-C matrices and BigWig ChIP-seq track files are available in the figshare repository, which are accessible at https://doi.org/10.6084/m9.figshare.21370560. v1. Source data are provided with this paper.