In vivo properties of Arabidopsis FCA condensates involved in RNA 3’ processing

biomolecular condensate, RNA processing, FLC, COOLAIR, noncoding RNA 15

16 The Arabidopsis RNA binding protein FCA is found in liquid-like nuclear condensates and promotes 17 proximal polyadenylation of specific nascent transcripts. To understand the in vivo interactions required 18 for these condensates we used single-particle tracking experiments on FCA expressed at endogenous 19 levels in live cells. These revealed FCA forms a core tetramer that multimerizes into higher-order 20 particles (~16 to 5 6 FCA molecules ), corresponding to condensates observed using confocal 21 microscopy. The coiled -coil protein FLL2, genetically required for FCA function and condensate 22 formation, showed co-localization primarily to the larger condensates. FCA regulates FLC, however, 23 its in vivo co-localization with an FLC lacO/LacI transgene was found to be infrequent. A missense 24 mutation in the FCA RRM domain showed RNA binding to FCA is necessary for function. This reduced 25 RNA-binding activity, attenuated FCA condensate formation, 3’ RNA processing and FLC repression, 26 however it did not influence the core tetramer. Our work points to a modular structure for FCA 27 condensates centred around a core of four FCA molecules , which multimerize to larger functionally 28 important condensates via associated RNA and FLL2 interaction, with only transient residency at their 29 site of action. 30 31

32 Cellular compartmentalization by biomolecular condensates is associated with diverse processes in 33 eukaryotic cells (Banani et al, 2017). While in vitro studies have highlighted their roles in increasing 34 local concentration and residence time of interacting components, the functional characteristics of 35 heterogeneous condensates in vivo remain unclear. Plants are an excellent system in which to study 36 biomolecular condensates (Field et al, 2023). For example, Arabidopsis FLOWERING LOCUS C (FLC), 37 a gene encoding a key flowering repressor, is regulated by co-transcriptional and antisense-mediated 38 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint chromatin silencing mechanism s (Fang et al , 2020 ; Xu et al , 2021 ) that involves multiple nuclear 39 biomolecular condensates with very different properties (Zhu et al, 2021; Fiedler et al, 2022; Berry et 40 al, 2017). One of those is a liquid-like condensate that involves an RNA-binding protein called FCA, 41 which directly interacts with FY , a Pfs2p/WDR33 homologue and a central component of the 42 polyadenylation machinery in eukaryotes (Simpson et al , 2003) . FCA contains two RRM (RNA 43 Recognition Motif) domains and a WW domain located between two predicted prion -like domains 44 (Macknight et al , 1997) . It was shown to form nuclear condensates through liquid-liquid phase 45 separation and to recruit RNA 3’ processing factors (Fang et al, 2019). Mutant screening identified a 46 coiled-coil protein, FLL2, required for the liquid-liquid phase separation of FCA nuclear condensates 47 and participation in FLC regulation (Fang et al, 2019). In vitro assays revealed that FCA condensates 48 were enhanced by addition of low amounts of Arabidopsis RNA. However, the immunoprecipitation of 49 COOLAIR nascent transcripts by FCA was not dependent on FLL2 (Fang et al, 2019). 50 51 We have continued our investigations of the properties of FCA liquid-like condensates and exploited a 52 single-molecule imaging technique, using Slimfield microscopy with an oblique illumination angle of 53 60 ± 5°degrees to enhance imaging contrast (Plank et al., 2009). This enables simultaneous tracking of 54 individual molecules within a wide range of biological condensates providing quantitative 55 measurements of molecule number and molecular motion behavior in plant nuclei (Shen et al, 2023). It 56 has been used to study biomolecular condensates in live cells including bacteria, yeast or mammalian 57 cells (Kent et al, 2020; Izeddin et al, 2014; Ladouceur et al, 2020; Biswas et al, 2022) and we adapted 58 it for living plant tissues to specifically study nuclear proteins (Bayle et al, 2021). 59 60 Our findings indicate that FCA condensates consist of multimers of a tetrameric FCA core, with sizes 61 ranging from approximately 16 to 56 molecules. The lower diffusion coefficient of the larger particles, 62 compared to core particles, suggests they correspond to the liquid-like condensates observed using 63 confocal microscopy. Notably, FLL2, essential for FCA function, predominantly co -localizes within 64 the larger condensates. However, sustained co -localization of large FCA condensates with the FLC 65 locus is observed only infrequently. The involvement of RNA in promoting FCA condensation was 66 shown through the identification of a missense mutation in the FCA second RRM domain . This 67 mutation reduced RNA-binding activity in vitro, significantly attenuated FCA condensate formation , 68 lowered RNA 3’ processing efficiency and prevented FLC repression but did not influence the core 69 tetramer. Our work points to a modular structure for FCA and RNA 3’ processing condensates, with a 70 core of four FCA molecules that multimerize via associated RNA and FLL2 interaction to larger 71 functionally important condensates. 72 73

74 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Single-particle tracking reveals physical properties of FCA condensates. 75 Biomolecular condensates are considered to increase local concentration of its components (Banani et 76 al, 2017; Shin & Brangwynne, 2017). To assess the consequence of condensation on FCA function, we 77 generated plants carrying an FCA -mScarlet-I translational fusion. A mScarlet-I coding sequence was 78 inserted into an FCA genomic clone and transgenic plant lines carrying single T-DNA insertion were 79 generated, where the fusion was expressed at similar level as the endogenous FCA gene (Fig EV1A and 80 B). The FCA mScarlet-I fusion formed liquid-like condensates (Fig 1A) and rescued the fca-9 mutant 81 phenotype (Fig EV1B). mScarlet-I is a bright red fluorescent protein known for its characteristics as a 82 rapidly maturing monomer (Bindels et al , 2016) and using this in conjunction with the Slimfield 83 microscope provided the capability to track very small FCA particles dynamically in vivo (Fig 1A). The 84 single-molecule sensitivity of Slim field in vivo, coupled with stepwise photobleaching of fluorescent 85 protein tags (Leake et al, 2006; Reyes-Lamothe et al, 2010), also enabled accurate quantification of the 86 number of molecules (stoichiometry) within the condensate (EV2; Jin et al, 2021). 87 88 Analysis of the single-particle tracking dataset identified a distinct peak representing approximately 4 89 molecules of FCA (Fig 1B) with other peaks in multiples of 4 from approximately 16 to 56 molecules. 90 To investigate whether these regular peaks follow specific patterns, we conducted simulations through 91 kernel density estimation to assess their periodicity (Leake et al, 2008). The estimate of the number of 92 molecules in a repeat unit depends on consistent stoichiometry intervals observed between nearest -93 neighbor peaks in the stoichiometry distribution. The periodicity peaks for FCA particles were found to 94 be approximately 4.2 and 8.7 (Fig 1C). 95 96 Particles can also exhibit different motion behaviour (Shen et al, 2023). We analysed the diffusion 97 coefficient of individual particles and plotted single particles with stoichiometry. Generally, particles 98 with low stoichiometry display variable mobility, while those with high stoichiometry show low 99 mobility (Fig 1D). These results suggest that the core module of an FCA particle is a tetramer with 100 larger condensates forming through the higher-order assembly of these tetrameric cores. 101 102 The co-localization of FCA and FLL2 is primarily observed in the prominent foci in 103 Arabidopsis. 104 105 Our previous study used transient assays suggested that FCA condensates serve as the sites for the 106 interaction between FCA, FLL2 and 3’ -end processing factors (Fang et al , 2019) . However, the 107 condensates formed by FCA and FLL2 exhibited different patterns in low resolution images of 108 Arabidopsis roots (F ang et al, 2019). To investigate the potential co -localization of FCA and FLL2 109 within the same nuclear condensates in vivo, we crossed plants carrying an FCA-mTurquoise2 transgene 110 (Xu et al, 2021) with plants carrying FLL2-eYFP transgene (Fang et al, 2019) and generated double 111 homozygotes. Utilizing 3D imaging and Airyscan confocal microscopy, we found that FCA and FLL2 112 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint generally reside in distinct fluorescent foci within the cell nucleus (Fig 2A) but co-localize within a few 113 larger foci (Fig 2B and D). We conducted a comprehensive analysis to determine the size distributions 114 for FCA, FLL2 and FCA/FLL2 -co-localized condensates (Fig 2 C). The data suggest a specific 115 association between FCA and FLL2 within larger condensates. 116 117 To determine if the se prominent FCA/FLL2 condensates directly associate with the FLC locus, we 118 crossed the FCA-mScarlet-I transgene into a line carrying a previously established FLC-lacO/LacI-119 eYFP system that enables spatial localization of the FLC locus in the nucleus (Rosa et al, 2013). FCA 120 condensates did not frequently co-localize with the FLC-lacO/LacI-eYFP marked locus (Fig EV3). 121 Only 2.7% of identified FCA condensates partially co-localized with the FLC-lacO/LacI-eYFP locus 122 suggesting that FCA condensates only transiently associate with the FLC locus. 123 124 FCA condensation is dependent on a functional RRM domain. 125 To further investigate the functional significance of the heterogeneity in FCA condensates we continued 126 a genetic screen for factors essential for FCA function (Liu et al, 2007). This identified a missense 127 mutation on the FCA overexpression transgene at amino acid 304 of the FCA protein , resulting in a 128 leucine to phenylalanine change (Fig EV1C). This alteration is located within the second RRM domain 129 of FCA, a domain conserved across various species (Macknight et al , 1997) . The impact of this 130 missense mutation on the RNA-binding activity was established through in vitro RNA-binding assays 131 (Fig EV1D). That the mutant came out of an FLC mis-expression screen and exhibited late flowering 132 shows the RNA-binding activity is important for FCA function (Fig EV1E). 133 134 To examine whether this RRM mutation affects FCA condensation, we generated transgene constructs 135 expressing an FCA wild-type genomic DNA fragment with (FCArrm) and without (FCAwt) the RRM 136 mutation fused with mScarlet-I and introduce them into fca-9 mutant (Fig EV1A). After selecting for 137 single insertion transgenic plants and homozygotes, we analysed the FCAwt condensates within the 138 nuclei of young roots using Airyscan confocal microscope. Plants containing t he FCArrm transgene 139 limited the size and number of condensates, displaying a more diffuse signal compared to plants 140 carrying FCAwt (Fig 3A and B). 141 142 In order to determine whether the RRM mutant alters the structure of FCA condensates, we also 143 conducted single-particle tracking with the FCArrm lines. We observed an obvious stoichiometry peak 144 of approximately 4 molecules for FCArrm, slightly higher than that for FCAwt (Fig 3C; Appendix Fig 145 S1). However, when focusing on particles with stoichiometries exceeding 8 molecules, the probability 146 density of FCArrm was found to be lower than FCAwt. FCArrm particles with more than 16 molecules 147 were scarce, however, the integrated nuclear intensity of fluorescent signal in FCArrm and FCAwt 148 showed no obvious change ( Appendix Fig S 2). This result suggests that the reduced RNA -binding 149 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint activity of FCA proteins decreases their likelihood of forming larger particles with a range of 16 to 56 150 molecules, consistent with the confocal imaging analysis. 151 152 To determine if the RRM mutant changed the physical properties of FCA condensates, w e compared 153 the diffusion coefficient s of individual particles with their stoichiometry (Fig 3 D). The mean 154 stoichiometry of FCArrm is lower than that of FCAwt, while the overall pattern of diffusion coefficients 155 is distributed towards slightly higher values (Fig 3E; Appendix Fig S3). Surprisingly, the periodic unit 156 of FCArrm only experiences a minor reduction compared to FCAwt, which implies that FCA proteins 157 with reduced RNA-binding activity retain the capacity to form the tetrameric core (Fig 3F). Thus, the 158 RNA-binding activity of FCA proteins affects the formation of larger condensates but does not affect 159 core protein assembly. 160 161 The functional RNA-binding domain is required for COOLAIR R-loop resolution and FCA auto-162 regulation. 163 We then used the FCAwt and FCArrm lines to further investigate the functional importance of RNA 164 binding. Compared to the FCAwt transgenic plants, FCArrm plants displayed higher levels of FLC 165 expression and therefore flower ed late (Fig 4 A and B , EV1B ). FCA associate s with COOLAIR, 166 recruiting 3’ processing factors to promote proximal COOLAIR polyadenylation, thus resolving a 167 COOLAIR-formed R-loop (Liu et al, 2010; Xu et al, 2021). The ratio of proximal-to-distal isoforms of 168 COOLAIR transcripts was reduced in the FCArrm lines compared to FCAwt consistent with reduced 169 R-loop resolution when RNA binding is defective (Fig 4 C). S9.6-DNA/RNA immunoprecipitation 170 followed by cDNA conversion (DRIPc)-qPCR analysis also revealed FCArrm has reduced ability to 171 resolve the COOLAIR-induced R-loop (Fig 4D). 172 173 FCA autoregulates itself promoting polyadenylation at the proximal sites in intron3 (Quesada et al, 174 2003), so we conducted Quant-seq analysis to capture polyadenylated FCA transcripts in Col-0, FCAwt, 175 FCArrm and fca-9. In FCArrm and fca-9, the re were significantly fewer reads of transcripts 176 corresponding to proximally polyadenylated FCA transcripts compared to Col-0. Conversely, reads at 177 the distally polyadenylated site of FCA increased (Fig 4E-F), all consistent with reduced FCA function. 178 RNA-binding activity is thus important for FCA condensation and for well-defined FCA functional 179 roles, promotion of proximal COOLAIR termination and R-loop resolution. 180 181

182 Our study provides insight on the functional properties of the dynamic liquid-like RNA 3’ processing 183 condensates containing the Arabidopsis nuclear RNA -binding protein FCA. We conducted single -184 particle tracking to determine the molecular stoichiometry and diffusivity of FCA condensates in plant 185 nuclei. This revealed insights into the particle assembly size and association of FCA with its protein 186 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint and RNA partners functioning in FLC repression. Our work points to a modular structure for FCA 187 condensates centred around a core of four FCA molecules, which multimerize to larger functionally 188 important condensates via associated RNA and FLL2 interaction, with only transient residency at their 189 site of action. 190 191 The missense mutation in the FCA RRM domain reduced RNA-binding activity, attenuated condensate 192 formation and prevented FLC repression but the stoichiometry of the smaller FCA particles was 193 unaffected. The lack of effect of RNA binding on the core modular structure is similar to previous 194 observations with ELA V , a well-characterized RNA-binding protein with three RRM domains. ELA V 195 multimerizes on its target RNAs and forms a defined complex (Soller & White, 2005). Future analysis 196 should examine how FCA core tetramers multimerize. The FCA prion-like domain (Chakrabortee et al, 197 2016), RRM domain (Macknight et al, 1997), WW domain (Simpson et al, 2003; Henderson et al, 2005) 198 and extended NH2 terminus initiating in a non -methionine start codon (Simpson et al, 2010) may all 199 play a role. In addition, m6A exerts function in the regulatory mechanisms of FLC and COOLAIR (Xu 200 et al, 2021; Wang et al, 2022; Sun et al, 2022; Amara et al, 2023); we need to understand if the dynamic 201 structures are affected by RNA modification and how this tetramer interfaces with its target RNAs . 202 Continued analysis of the role of the first RRM may also be productive as the protein phosphatase 203 SSU72 was found to antagonize FCA’s binding to COOLAIR in a manner dependent on the first RRM 204 motif (Yongke Tlan et al, 2019). 205 206 COOLAIR transcripts form dense ‘clouds’ at each FLC locus as judged by single-molecule fluorescence 207 in situ hybridization, which are enhanced upon cold exposure (Rosa et al., 2016). Whether the FCA 208 condensates associate with these RNA condensates will require RNA live imaging analysis. It is 209 interesting to speculate that COOLAIR molecules might undergo thermo-responsive phase transitions 210 as has been found recently for other RNAs (Wadsworth et al., 2023). Protein partners can tune this 211 phase behavio ur (Ruff et al ., 2020) , so this system has a lot of potential to uncover fundamental 212 principles of condensate functionality in vivo , and if transient association with the FLC locus is 213 sufficient for function. In mammalian cells, Mediator -bound enhancers only sporadically co -localize 214 with their target loci. This variability in co-localization can be attributed to a dynamic kissing model, 215 where a distal Mediator cluster interacts with the gene only at specific timepoints (Cho et al, 2018). 216 217 Our work demonstrates the power of studying biomolecular condensate functionality in genetically 218 tractable systems. Arabidopsis with its extensive genetic resources, ability to make transgenic lines and 219 small, transparent roots is particularly suitable for these studies, as illustrated by our ability to image 220 condensates in living roots with the single-molecule sensitivity of Slim field. Coupled with stepwise 221 photobleaching we could accurately quantify the number of molecules within condensates providing 222 mechanistic insights into the modular structure of FCA condensates , with and without various 223 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint functionally important partners. These tools now allow us to explore the role of FCA condensate 224 dynamics in response to various environmental exposures. 225 226

227 228 Plant materials and growth conditions 229 The wild type Col-0, fca-9 mutant allele, pFCA::FCA-mTurquoise and pFLL2::FLL2-eYFP transgenic 230 plants were described previously (Fang et al , 2020 ; Xu et al , 2021 ). To generate the pFCA::FCA -231 mScarlet-I transgenic plants, FCA genomic DNA fragments was amplified and insert into the pENTRY 232 vector. The mScarlet -I coding sequence was inserted before the stop codon via In -Fusion seamless 233 cloning. The FCA genomic DNA fragment fused mScarlet -I was transferred to pSLJ699I5 vector via 234 gateway cloning. The RRM mutation construct is modified from wild -type FCA genomic DNA 235 constructs via In-Fusion seamless cloning . These constructs were transformed into the fca-9 mutant. 236 Arabidopsis seeds were surface sterilized and sown on standard half-strength Murashige and Skoog (½ 237 MS) medium plate without glucose and stratified at 4 °C for 3 days before transferred to long -day 238 conditions (16-h light at 20 °C, 8-h darkness at 16 °C). 239 240 Confocal microscopy of FCA-mScarlet-I 241 Plants were grown vertically on half strength MS plates for 4-7 days. Confocal images were acquired 242 on a Zeiss LSM 880 upright microscope equipped with an Airyscan detector. Samples were imaged 243 through a C-Apochromat 63x/NA 1.2 water-immersion objective by exciting them at 561 nm and using 244 the Airyscan detector in super resolution mode. The fluorescence emission arrived at the detector after 245 passing through a filter that allowed wavelengths between 495 -550 nm and above 570 nm to pass 246 through. The voxel size was 0.06 × 0.06 × 0.26 µm (xyz) and the pixel dwell time was 1.78 µs. Z-stacks 247 comprised 28 slices and spanned a 7.13 µm range. After Airyscan processing, the image analysis was 248 performed in Arivis Vision4D ver. 4.1.0. (Zeiss) as described below. 249 250 Co-localization analysis of FCA and FLL2 251 Confocal images were acquired either on the Zeiss LSM 880 microscope mentioned above or a Zeiss 252 LSM 980 confocal microscope. For using LSM880, line by line sequential acquisition was employed. 253 Samples were imaged through a C -Apochromat 63x/NA 1.2 water-immersion objective and were 254 excited at 458 (mTurquoise) and 514 nm (eYFP). The fluorescence emission of the samples passed 255 through a filter that only allowed wavelengths in the 465-505 nm range and above 525 nm to reach the 256 Airyscan detector, which was operating in super resolution mode. The voxel size was 0.05 × 0.05 × 257 0.21 µm (xyz) and the pixel dwell time was 0.72 µs. Z-stacks comprised 25 slices spanning a 4.96 µm 258 range. 0.2 mm TetraSpeck microspheres (Thermo Fisher Scientific, UK) were imaged using these same 259 settings and the resulting images were used to correct the sample images for chromatic aberration. The 260 correction was performed in Zen (Zeiss, Germany) using the channel alignment method. 261 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint For using Zeiss LSM 980, frame by frame acquisition was employed. Samples were imaged through a 262 C-Apochromat 63x/NA 1.2 water-immersion objective and were excited at 445 (mTurquoise) and 514263 nm (eYFP). The fluorescence emission of eYFP was collected through a filter that only allowed 264 wavelengths in the 530-582 range to reach the Airyscan detector, whereas the fluorescence emission of 265 mTurquoise was collected through a filter that allowed wavelengths in the 422 -497 and 607-735 nm 266 ranges to reach the Airyscan detector. The detector was operating in super resolution mode. The voxel 267 size was 0.05 × 0.05 × 0.20 µm (xyz) and the pixel dwell time was 1.06 µs. Z-stacks comprised 30 slices 268 spanning a 5.8 µm range. 269 270 After Airyscan processing, t he image analysis was performed in Arivis Vision4D ver. 4.1.0. (Zeiss). 271 Firstly, the blob finder algorithm was applied to the mTurquoise2 channel using a diameter value of 272 0.35 mm, a probability threshold of 60%, and a split sensitivity of 50%. Then, the blob finder algorithm 273 was applied to the eYFP channel using a diameter value of 0.30 mm, a probability threshold of 50%, 274 and a split sensitivity of 50%. Blob finder is a segmentation algorithm that excels at finding round or 275 spherical objects in 2D or 3D, respectively. It uses a Gaussian scale to identify seeds for the objects and 276 subsequently applies a watershed algorithm to determine the boundaries of the objects. Afterwards, the 277 intersection between the output of the two blob finder operations was calculated. Finally, metrics such 278 as volume, mean intensity, and total intensity were computed for the objects generated by each of the 279 blob finder operations, as well as for their intersection. 280 281 Slimfield microscopy 282 The customized Slim field microscope was previously described ( Payne-Dwyer and Leake, 2022). 283 Continuous wave lasers (Coherent OBIS) delivered a de -expanded Gaussian beam mode (TEM00) to 284 the back aperture of an NA 1.49 Apo TIRF 100× oil immersion objective lens (Nikon). Single molecule 285 sensitivity was achieved using a fast sCMOS camera (Photometrics Prime95B) which triggered a 286 continuous wave laser for 10 ms exposure per frame at 80 Hz. Excitation and emission filters were 287 tailored to mScarlet-I (594/25 nm) for their respective red channels. The total magnification was 288 approximately 200×, resulting in an oversampled image pixel edge size of 53 nm. Nuclei were identified 289 using brightfield mode to determine the best focus and were subsequently captured for manual 290 segmentation later. Fluorescence acquisition settings in the green and red channels were pre-optimized 291 to prevent initial saturation and to ensure the detection of individual molecular brightness tracks for 292 mScarlet-I. These settings were then locked to prevent systematic variation in the characteristic single 293 molecule brightness. 294 295 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Single-particle tracking and intensity analysis 296 Fluorescent foci were identified from Slimfield image sequences and linked into tracks using 297 ADEMScode software ( Plank et al , 2009 ). Each of these foci were localised to a super -resolved, 298 subpixel precision, typically 40 nm. The random overlap probability was estimated using a nearest 299 neighbour distance approach (Payne-Dwyer et al, 2023). Analysis was spatially restricted to data lying 300 within a region of effectively uniform laser illumination (80% ± 9% s.d. peak intensity ) by cropping 301 images to an area of 240 × 184 pixels (12.7 × 9.8 μm). Fluorescence image sequences were further 302 cropped to specify individual nuclei, using manually segmented masks from brightfield images. 303 The 2D diffusion coefficient of each track ( 𝐷) was computed using the increase of mean-square 304 displacement across time interval according to a random walk model (Leake et al , 2008) . Focus 305 intensity was calculated by summing the pixel value s within a 5-pixel distance and subtracting an 306 averaged background level from the rest of a 17-pixel square (Leake et al , 2008) . The track 307 stoichiometry was found by extrapolating the first 5 foci intensities to account for photobleaching 308 (Syeda et al., 2019), then dividing this initial intensity by the characteristic single molecule brightness 309 (Wollman et al ., 2017) . The characteristic single molecule brightness linked to each reporter was 310 determined as 110 ± 15 (mean ± sem) for mScarlet-I based on the Chung-Kennedy-filtered terminal 311 intensity of tracks in each acquisition (Leake et al, 2003), thus providing an internal calibration for track 312 stoichiometry, periodicity, and nuclear protein copy number. 313 314 To calculate periodicity, the stoichiometries of all tracks within each nucleus were represented as a 315 kernel density distribution (Leake et al , 2014) , employing an empirical standard deviation of 0.6 316 molecules on the characteristic single molecule brightness. Peaks within this distribution were identified 317 using the MATLAB findpeaks function, and the intervals between nearest neighbour peaks were 318 computed (Payne-Dwyer and Leake, 2022). Raw estimations of the total number of molecules in each 319 nucleus were determined using ImageJ macros to integrate the pixel values inside segmented nuclei as 320 in previous work (Wollman and Leake, 2015; Payne-Dwyer and Leake, 2022). The nuclear protein copy 321 number of each reporter dataset was refined to exclude autofluorescence by calculating the difference 322 between mean integrated nuclear intensities of the labelled dataset and an unlabeled control, adjusted 323 proportionally to the ratio of mean nuclear segment areas. 324 325 Expression analysis 326 Total RNA was extracted from plant materials as described previously (Qüesta et al, 2016), treated with 327 TURBO DNase (Ambion) to eliminate DNA contamination, and reverse-transcribed using SuperScript 328 IV Reverse Transcriptase (Invitrogen) with gene -specific reverse primers. Quantitative PCR (qPCR) 329 analysis was conducted on a LightCycler480 II (ROCHE), and the qPCR data were normalized to the 330

genes UBC and PP2A. To measure the proximal ratio of COOLAIR, the level of the proximal 331 component was normalized to the distal COOLAIR. Primers were listed in Appendix Table S1. 332 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint 333 DNA−RNA hybrid immunoprecipitation followed by cDNA conversion (DRIPc) 334 The DRIP protocol is as described previously ( Xu et al, 2021) with modifications. In all, 3 g 10-day-335 old seedlings were harvested and grounded into a fine powder. The powder was suspended in 35 mL of 336 Honda buffer (20mM HEPES, 0.44M sucrose, 1.25% Ficoll, 2.5% dextran T40, 10mM MgCl 2, 0.5% 337 Triton x-100, 5 mM DTT, 1x protease inhibitor cocktail (Roche), filtered through one layer of Miracloth, 338 and centrifuged at 3500×g for 15 min. Nuclear pellets were resuspended in 1mL Honda buffer and 339 centrifuged at 8000×g for 1 min. Pellets were then resuspended in the lysis buffer (50mM Tris-HCl pH 340 8.0, 10mM EDTA, 1% SDS) supplied with 0.1 mg/mL proteinase K (AM2546, Invitrogen) and digested 341 at 55 °C overnight with gentle rotation. The mixture was phenol/chloroform extracted, followed by 342 nucleic acid precipitation with NaOAc and ethanol. The nucleic acid pellet was dissolved in water and 343 quantified with Qubit DNA quantification kit (Invitrogen). In general, 1 μg of nucleic acid was digested 344 by restriction enzymes including DdeI, NdeI and XbaI (NEB) overnight. The digested nucleic acid was 345 phenol/chloroform extracted, followed by nucleic acid precipitation with NaOAc and ethanol. Then the 346 nucleic acid was treated with RNAse III (Ambion) and then extracted by phenol/chloroform and ethanol 347 precipitation. The treated nucleic acid was then diluted ten times with dilution buffer (16.7mM Tris pH 348 7.5, 167mM NaCl, 2.2mM EDTA, 0.1% Triton X-100) and 10% was stored at −20 °C as input. In all, 349 5 μg of S9.6 antibody (1:100 dilution, ENH001, Kerafast) was added, then incubated overnight at 4 °C. 350 The next day, 50 μl Protein G Agarose (Invitrogen) was added and incubated for another 2 h. The 351 immunoprecipitants were washed five times with dilution buffer and twice with TE buffer, then were 352 eluted in 100 μl elution buffer (10mM Tris pH 7.5, 2 mM EDTA, 0.2% SDS, 100 ng/μl tRNA) at 55 °C 353 for 2 hours, together with input samples. The nucleic acids were precipitated with NaOAc, isopropanol, 354 and glycogen, dissolved in water. The nucleic acid samples were then treated with treated with TURBO 355 DNase (Ambion) to eliminate DNA contamination, and reverse -transcribed using SuperScript IV 356 Reverse Transcriptase (Invitrogen) with gene-specific reverse primers. All the samples including inputs 357 were subjected to qPCR analysis via LightCycler480 II (Roche). The data were normalized to 1% of 358 input. Primers were listed in Appendix Table S1. 359 360 Quant-seq 361 The Quant -seq protocol is as described previously ( Mateo-Bonmatí et al , 2024 ). The total RNAs 362 prepared for Quant-seq underwent purification using the Qiagen RNeasy miniprep kit (74106). Lexogen 363 GmbH (Austria) conducted the library preparation, sequencing, and data analysis. Sequencing-ready 364 libraries were created from 100 ng of input RNA using a QuantSeq 3 ’ mRNA-Seq Library Prep Kit 365 REV for Illumina (015UG009V0271), following standard procedures. The RNA integrity and quality 366 of indexed libraries were assessed on a Fragment Analyzer device (Agilent Technologies) using a DNF-367 471 RNA Kit and HS -DNA assay, respectively. Library quantification was carried out using a Qubit 368 dsDNA HS assay (Thermo Fisher). 369 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint A sequencing-ready pool of indexed libraries was then sequenced on an Illumina NextSeq 2000 with a 370 100-cycle cartridge, using the Custom Sequencing Primer (CSP). Verification of read quality utilized 371 FastQC version v0.11.7, and read adapter trimming employed cutadapt version 1.18 (Martin 2011). 372 Clean reads were mapped to the latest version of the Arabidopsis genome (TAIR10) using the splice -373 aware aligner STAR version 2.6.1a (Dobin et al, 2013). The raw reads have been deposited in the Short 374 Read Archive (SRA) under the reference (PRJNA1087342 and PRJNA1076161). 375 376 Electrophoretic Mobility Shift Assay 377 FCA cDNA with L304F mutation was amplified from the mutant as described in Figure EV1. 378 To generate RNA probes, dsDNA of the corresponding region was amplified through PCR (primers see 379 Appendix Table S1) with the T7/T3 phage polymerase promoter added at the 5’ end of the primer. 380 COOLAIR RNA was then generated by following a protocol provided by the MAXIscript® T7 381 Transcription Kit (Invitrogen). The resulting product was DNA digested, denatured, and separated on 382 5% TBE gel containing 8 M Urea. The band at the right size was cut out, sliced, and eluted in gel 383 extraction buffer (0.5 M NH4Ac, 1 mM EDTA, 0.2% SDS, 1xRNAsecure ™ (Invitrogen)) at 37ºC 384 overnight, followed by precipitation with isopropanol. 10 pm ol purified RNA was 3’biotinylated by 385 following the instructions provided by the RNA 3' End Biotinylation Kit (Pierce). The labelling 386 efficiency was determined to be at least 90%. For the gel shift assay, a 20 μl reaction (25 ng/μl protein, 387 2 nM probe, 250 ng/μl tRNA, 50 ng/μl heparin, 10 mM HEPES (pH 7.3), 120 mM KCL, 1 mM MgCl2, 388 1 mM DTT) was set up on ice. For the competition assay, 2x to 100x of unlabelled probe was included 389 in the reaction. The reaction with only GST was used as a negative control. The mixture was incubated 390 at room temperature for 5 min and resolved on 3.5% native TBE-acrylamide gel and transferred onto a 391 positively charged nylon membrane. The biotinylated RNA on the membrane was detected by 392 chemiluminescence according to a protocol provided with the Chemiluminescent Nucleic Acid 393 Detection Module (Pierce, 89880). Primers to generate probes were listed below. For the binding 394 reaction, 1 μg of different recombinant proteins was incubated with 2 pmol of biotinylated probe in 20 395 μL of 1× buffer containing 20 mM Tris (pH 7. 5), 100 mM NaCl, 1 0 mM MgCl 2, non-specific 396 competitors (1µg/µl heparin and 5µg tRNA) and 0.05% Triton X-100 at room temperature for 10 min. 397 The RNA -protein mixture was resolved on 6% 1× T BE acrylamide gel under 100 V for 50 min, 398 followed by electrophoretic transfer to positively charged nylon membranes (GE Healthcare). The 399 biotinylated RNA on the membrane was UV cross-linked and detected using chemiluminescent nucleic 400 acid detection following the manufacturer’s instructions (Thermo). 401 402

403 We thank Shuqin Chen and Tina Zhang for their excellent technical assistance and Hsuan Pai for her 404 artwork assistance. We also thank all members of the Dean and Howard groups. This work was 405 funded by the European Research Council Advanced Grant (EPISWITCH, 833254), Wellcome Trust 406 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint (210654/Z/18/Z), and the Royal Society Professorship (RP\R1\180002) to Caroline Dean, and EPSRC 407 grants (EP/T00214X/1 to C.D. and EP/T002166/1 and EP/W024063/1 to M.L). 408 409 Author contributions 410 G-J.J. and C.D. conceived the study. G -J.J., A.P.-D., R.M., Z. W., F.L. and S.G.L. performed the 411 investigation and analysed the data. G-J.J. and C.D. wrote the manuscript; all authors reviewed and 412 edited the manuscript. M.C.L. and C.D. supervised the study, acquired the funding, and were the project 413 administrators. 414 415

416 Amara U, Hu J, Cai J & Kang H (2023) FLK is an mRNA m6A reader that regulates floral transition 417 by modulating the stability and splicing of FLC in Arabidopsis. Mol Plant 16: 919–929 418 Banani SF, Lee HO, Hyman AA & Rosen MK (2017) Biomolecular condensates: Organizers of 419 cellular biochemistry. Nat. Rev. Mol. Cell Biol. 18: 285-298 420 Bayle V, Fiche JB, Burny C, Platre MP, Nollmann M, Martinière A & Jaillais Y (2021) Single-421 particle tracking photoactivated localization microscopy of membrane proteins in living plant 422 tissues. Nat Protoc 16: 1600–1628 423 Berry S, Rosa S, Howard M, Bühler M & Dean C (2017) Disruption of an RNA-binding hinge region 424 abolishes LHP1-mediated epigenetic repression. Genes Dev 31: 2115–2120 425 Bindels DS, Haarbosch L, Van Weeren L, Postma M, Wiese KE, Mastop M, Aumonier S, Gotthard 426 G, Royant A, Hink MA, et al (2016) mScarlet: A bright monomeric red fluorescent protein for 427 cellular imaging. Nat Methods 14: 53–56 428 Biswas S, Chen Z, Karslake JD, Farhat A, Ames A, Raiymbek G, Freddolino PL, Biteen JS & 429 Ragunathan K (2022) HP1 oligomerization compensates for low-affinity H3K9me recognition and 430 provides a tunable mechanism for heterochromatin-specific localization. Sci Adv 8 431 Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M & Gingeras TR 432 (2013) STAR: Ultrafast universal RNA-seq aligner. Bioinformatics 29: 15–21 433 Chakrabortee S, Kayatekin C, Newby GA, Mendillo ML, Lancaster A & Lindquist S (2016) 434 Luminidependens (LD) is an Arabidopsis protein with prion behavior. Proc Natl Acad Sci U S A 435 113: 6065–6070 436 Cho WK, Spille JH, Hecht M, Lee C, Li C, Grube V & Cisse II (2018) Mediator and RNA 437 polymerase II clusters associate in transcription-dependent condensates. Science 361: 412–415 438 Fang X, Wang L, Ishikawa R, Li Y, Fiedler M, Liu F, Calder G, Rowan B, Weigel D, Li P, et al 439 (2019) Arabidopsis FLL2 promotes liquid–liquid phase separation of polyadenylation complexes. 440 Nature 569: 265–269 441 Fang X, Wu Z, Raitskin O, Webb K, Voigt P, Lu T, Howard M & Dean C (2020) The 3 processing of 442 antisense RNAs physically links to chromatin-based transcriptional control. Proc Natl Acad Sci U 443 S A 117: 15316–15321 444 Fiedler M, Franco-Echevarría E, Schulten A, Nielsen M, Rutherford TJ, Yeates A, Ahsan B, Dean C 445 & Bienz M (2022) Head-to-tail polymerization by VEL proteins underpins cold-induced Polycomb 446 silencing in flowering control. Cell Rep 41 447 Field S, Jang GJ, Dean C, Strader LC & Rhee SY (2023) Plants use molecular mechanisms mediated 448 by biomolecular condensates to integrate environmental cues with development. Plant Cell 35: 449 3173–3186 450 Henderson IR, Liu F, Drea S, Simpson GG & Dean C (2005) An allelic series reveals essential roles 451 for FY in plant development in addition to flowering-time control. Development 132: 3597–3607 452 Izeddin I, Récamier V, Bosanac L, Cissé II, Boudarene L, Dugast-Darzacq C, Proux F, Bénichou O, 453 Voituriez R, Bensaude O, et al (2014) Single-molecule tracking in live cells reveals distinct target-454 search strategies of transcription factors in the nucleus. Elife 2014: 1–27 455 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Jin X, Lee JE, Schaefer C, Luo X, Wollman AJM, Payne-Dwyer AL, Tian T, Zhang X, Chen X, Li Y, 456 et al (2021) Membraneless organelles formed by liquid-liquid phase separation increase bacterial 457 fitness. Sci Adv 7 458 Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, 459 Žídek A, Potapenko A, et al (2021) Highly accurate protein structure prediction with AlphaFold. 460 Nature 596: 583–589 461 Kent S, Brown K, Yang C hsun, Alsaihati N, Tian C, Wang H & Ren X (2020) Phase-Separated 462 Transcriptional Condensates Accelerate Target-Search Process Revealed by Live-Cell Single-463 Molecule Imaging. Cell Rep 33: 108248 464 Ladouceur AM, Parmar BS, Biedzinski S, Wall J, Tope SG, Cohn D, Kim A, Soubry N, Reyes-465 Lamothe R & Weber SC (2020) Clusters of bacterial RNA polymerase are biomolecular 466 condensates that assemble through liquid-liquid phase separation. Proc Natl Acad Sci U S A 117: 467 18540–18549 468 Leake MC, Wilson D, Bullard B, Simmons RM & Bubb MR (2003) The elasticity of single kettin 469 molecules using a two-bead laser-tweezers assay. FEBS Lett 535: 55–60 470 Leake MC, Chandler JH, Wadhams GH, Bai F, Berry RM & Armitage JP (2006) Stoichiometry and 471 turnover in single, functioning membrane protein complexes. Nature 443: 355–358 472 Leake MC, Greene NP, Godun RM, Granjon T, Buchanan G, Chen S, Berry RM, Palmer T & Berks 473 BC (2008) Variable stoichiometry of the TatA component of the twin-arginine protein transport 474 system observed by in vivo single-molecule imaging. Proc Natl Acad Sci U S A 105: 15376–15381 475 Leake MC (2014) Analytical tools for single-molecule fluorescence imaging in cellulo. Phys Chem 476 Chem Phys 16: 12635–12647 477 Liu F, Quesada V, Crevillén P, Bäurle I, Swiezewski S & Dean C (2007) The Arabidopsis RNA-478 Binding Protein FCA Requires a Lysine-Specific Demethylase 1 Homolog to Downregulate FLC. 479 Mol Cell 28: 398–407 480 Liu F, Marquardt S, Lister C, Swiezewski S & Dean C (2010) Targeted 3′ processing of antisense 481 transcripts triggers arabidopsis FLC chromatin silencing. Macknight R, Bancroft I, Page T, Lister 482 C, Schmidt R, Love K, Westphal L, Murphy G, Sherson S, Cobbett C, et al (1997) FCA, a gene 483 controlling flowering time in arabidopsis, encodes a protein containing RNA-binding domains. 484 Cell 89: 737–7 485 Mateo-Bonmatí E, Montez M, Maple R, Fiedler M, Fang X, Saalbach G, Passmore LA & Dean CA 486 (2024) A CPF-like phosphatase module links transcription termination to chromatin silencing. 487 bioRxiv: 2023.07.06.547976 488 Payne-Dwyer AL & Leake MC (2022) Single-molecular quantification of flowering control proteins 489 within nuclear condensates in live whole Arabidopsis root. Methods in Molecular Biology 2476: 490 311-328. 491 Payne-Dwyer AL, Syeda AH, Shepherd JW, Frame L & Leake MC (2022) RecA and RecB: Probing 492 complexes of DNA repair proteins with mitomycin C in live Escherichia coli with single-molecule 493 sensitivity. J R Soc Interface 19 494 Payne-Dwyer A, Kumar G, Barrett J, Gherman LK, Hodgkinson M, Plevin M, Mackinder L, Leake 495 MC & Schaefer C (2023) Predicting Rubisco:Linker Condensation from Titration in the Dilute 496 Phase. arXiv:2301.05681 497 Plank M, Wadhams GH & Leake MC (2009) Millisecond timescale slimfield imaging and automated 498 quantification of single fluorescent protein molecules for use in probing complex biological 499 processes. Integr Biol 1: 602–612 500 Quesada V, Macknight R, Dean C & Simpson GG (2003) Autoregulation of FCA pre-mRNA 501 processing controls Arabidopsis flowering time. EMBO J 22: 3142–3152 502 Qüesta JI, Song J, Geraldo N, An H & Dean C (2016) Arabidopsis transcriptional repressor VAL1 503 triggers Polycomb silencing at FLC during vernalization. Science 353: 485–488 504 Reyes-Lamothe R, Sherratt DJ & Leake MC (2010) Stoichiometry and architecture of active DNA 505 replication machinery in escherichia coli. Science 328: 498–501 506 Rosa S, De Lucia F, Mylne JS, Zhu D, Ohmido N, Pendle A, Kato N, Shaw P & Dean C (2013) 507 Physical clustering of FLC alleles during polycomb-mediated epigenetic silencing in vernalization. 508 Genes Dev 27: 1845–1850 509 Rosa S, Duncan S & Dean C (2016) Mutually exclusive sense-antisense transcription at FLC 510 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint facilitates environmentally induced gene repression. Nat Commun 7: 1–7 511 Ruff KM, Dar F & Pappu R V . (2021) Ligand effects on phase separation of multivalent 512 macromolecules. Proc Natl Acad Sci U S A 118 513 Shen Z, Jia B, Xu Y, Wessén J, Pal T, Chan HS, Du S & Zhang M (2023) Biological condensates 514 form percolated networks with molecular motion properties distinctly different from dilute 515 solutions. Elife 12: 1–32 516 Shin Y & Brangwynne CP (2017) Liquid phase condensation in cell physiology and disease. Science 517 357 518 Simpson GG, Dijkwel PP, Quesada V, Henderson I & Dean C (2003) FY is an RNA 3′ end-519 processing factor that interacts with FCA to control the Arabidopsis floral transition. Cell 113: 520 777–787 521 Simpson GG, Laurie RE, Dijkwel PP, Quesada V, Stockwell PA, Dean C & Macknight RC (2010) 522 Noncanonical translation initiation of the Arabidopsis flowering time and alternative 523 polyadenylation regulator FCA. Plant Cell 22: 3764–3777 524 Soller M & White K (2005) ELAV Multimerizes on Conserved AU 4–6 Motifs Important for. Society 525 25: 7580–7591 526 Sonmez C, Bäurle I, Magusin A, Dreos R, Laubinger S, Weigel D & Dean C (2011) RNA 3′ 527 processing functions of Arabidopsis FCA and FPA limit intergenic transcription. Proc Natl Acad 528 Sci U S A 108: 8508–8513 529 Syeda AH, Wollman AJM, Hargreaves AL, Howard JAL, Bruning JG, McGlynn P & Leake MC 530 (2019) Single-molecule live cell imaging of Rep reveals the dynamic interplay between an 531 accessory replicative helicase and the replisome. Nucleic Acids Res 47: 6287–6298 532 Sun B, Bhati KK, Song P, Edwards A, Petri L, Kruusvee V, Blaakmeer A, Dolde U, Rodrigues V, 533 Straub D, et al (2022) FIONA1-mediated methylation of the 3’UTR of FLC affects FLC transcript 534 levels and flowering in Arabidopsis. PLoS Genet 18: 1–22 535 Varadi M, Anyango S, Deshpande M, Nair S, Natassia C, Yordanova G, Yuan D, Stroe O, Wood G, 536 Laydon A, et al (2022) AlphaFold Protein Structure Database: Massively expanding the structural 537 coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res 50: D439–538 D444 539 Wang C, Yang J, Song P, Zhang W, Lu Q, Yu Q & Jia G (2022) FIONA1 is an RNA N 6-540 methyladenosine methyltransferase affecting Arabidopsis photomorphogenesis and flowering. 541 Genome Biol 23: 1–29 542 Wadsworth GM, Zahurancik WJ, Zeng X, Pullara P, Lai LB, Sidharthan V, Pappu R V., Gopalan V & 543 Banerjee PR (2023) RNAs undergo phase transitions with lower critical solution temperatures. Nat 544 Chem 15: 1693–1704 545 Wollman AJM & Leake MC (2015) Millisecond single-molecule localization microscopy combined 546 with convolution analysis and automated image segmentation to determine protein concentrations 547 in complexly structured , functional cells , one cell at a time. Faraday Discuss 00: 1–24 548 Wollman AJM, Shashkova S, Hedlund EG, Friemann R, Hohmann S & Leake MC (2017) 549 Transcription factor clusters regulate genes in eukaryotic cells. Elife 6: 1–36 550 Xu C, Wu Z, Duan HC, Fang X, Jia G & Dean C (2021) R-loop resolution promotes co-551 transcriptional chromatin silencing. Nat Commun 12 552 Yongke Tlan, Zheng Han, Zhang Fel, Wang Shillang, Ji Xiaoru, Xu Chao, He Yuehul & Ding Yong 553 (2019) PRC2 recruitment and H3K27me3 deposition at FLC require FCA binding of COOLAIR. 554 Sci Adv 5: 1–12 555 Zhu P, Lister C & Dean C (2021) Cold-induced Arabidopsis FRIGIDA nuclear condensates for FLC 556 repression. Nature 599: 657–661 557 558 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Figure 1. Super-resolution and single-particle tracking characterize physical properties of FCA condensates A SlimVar microscopy of FCA-mScarlet-I (in magenta) in Arabidopsis root cells. Scale bar, 2 µm. B Distributions of stoichiometry (molecule number) of individual FCA particles. C Periodicity of FCA particle stoichiometry estimated from the most common spacing between neighboring peaks in each stoichiometry distribution. D The diffusion coefficient and stoichiometry of each FCA particles. The single-particle tracking analysis is from 3 biologically independent experiments. Results from individual replicate are provided in Appendix Figure S1. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Figure 2. The co-localization of FCA and FLL2 is primarily observed in a limited number of prominent foci in Arabidopsis A Representative 3D images of FCA and FLL2 colocalization. The FCA were identified and labelled in green and FLL2 labelled in magenta using Vision4D software. White labelling indicates the co-localization of FCA with FLL2. The enlarged images of co-localized foci were shown in insets at the bottom right corner. Scale bars, 1 μm. B Quantification of size of identified FCA, FLL2 and co-localized foci. The lines in the violin plots indicate the median and quartiles. ns, no significance. C Size and frequency distribution of FCA, FLL2 and co-localized foci. D Number of FCA, FLL2 or co-localized foci in each nuclei. The lines in the violin plots indicate the median and quartiles. Non-parametric Brunner-Munzel test. The quantitative analysis is based on 18 nuclei from 5 independent samples in 1 representative replicate. The labeled P-value was analyzed using the non-parametric Brunner-Munzel test. FCA FLL2 FCA ∩ FLL2 0 10 20 30foci number/nucleus FCA FLL2 FCA ∩ FLL2 0.00 0.05 0.10 0.15Volume (µm3) A FLL2-eYFP FCA-mTurquoise Merge Foci identification C B D 0.0001 0.0009 ns ns 0.00 0.05 0.10 0.15 0 10 20 30 40 Vollume (µm3) % Frequency FCA FLL2 FCA ∩ FLL2 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Figure 3. The RNA-binding activity of FCA modulates FCA condensates A Representative Airyscan confocal images of wild type FCA fused mScarlet-I (FCAwt) and FCA with L304F mutation (FCArrm). Scale bars, 5 μm. B Quantification of size and number of FCAwt and FCArrm condensates using Vision4D software. The lines in the violin plots indicate the median and quartiles. The quantitative analysis is based on 43 FCAwt nuclei and 65 FCArrm nuclei from 3 independent experiments. C Distributions of stoichiometry of individual FCAwt (in black) and FCArrm (in red) particles. D The diffusion coefficient and stoichiometry of each FCAwt (in black) and FCArrm (in red) particles. E Statistic analysis of stoichiometry and diffusion coefficient of FCAwt and FCArrm particles. F Periodicity of FCAwt (in black) and FCArrm (in red) particle stoichiometry. The single-particle tracking analysis is from 3 biologically independent experiments. The FCAwt data corresponds to that shown in Figure 1. The labeled P-value was analyzed using the non-parametric Brunner-Munzel test. Results from individual replicate are provided in Appendix Figure S1 and S3. FCAwt FCArrm 0 2 4 6 8 10 foci number/nucleus 0.0342 FCAwt FCArrm 0.0 0.1 0.2 0.3 0.0135 Volume (µm3) 0.0066 0.0026 0.0093 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Leaf number Col-0FCA WT FCA rrmfca-9 0 10 20 30 40 <0.0001 Col-0FCA rrm fca-9 0.0 0.5 1.0 1.5 Proximal: distal ratio (/Col-0) <0.0001 Col-0FCA WT FCA rrmfca-9 0 10 20 30 Relative expression level (/UBC/Col-0) 0.0006 4000 4500 5000 5500 6000 0 1 2 3 4 5 Distance from FLC TSS (bp) DRIPc-qPCR (/Input) Col-0 fca-9 FCAwt FCArrm FloweringA B D Spliced FLC FCA FCAwt FCArrm Col-0 fca-9 proximal E F distal A B C D E C Quant-seq read number 100 10 1 100 10 1 100 10 1 100 10 1 COOLAIR proximal: distal ratio 0.0 0.2 0.4 0.6 0.8 1.0Porximal/ total 0.0002 0.0 0.2 0.4 0.6 0.8 1.0Distal/ total 0.0002 Col-0FCAwtFCArrm fca-9 0 5 10 15Porximal/Distal 0.0057 .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint Figure 4. The missense mutation in the RRM of FCA only partially complements the fca-9 mutant phenotype A Flowering time of indicated plants grown in long-day photoperiod. Data are presented as mean ± s.d. (n = 10). B Expression of spliced FLC relative to UBC in the indicated plants. Data are presented as mean ± s.d. (n = 5). C The level of proximal isoforms of COOLAIR transcripts in the indicated plants relative to control (FCAwt). Data are presented as mean ± s.d. (n = 4). D DRIPc–qPCR analyzing COOLAIR R-loop in indicated plants. Data are mean ± s.d. from one representative replicate of three biological replicates. E Quant-seq experiment results from the FCA locus carried out on indicated seedlings. The vertical pale gray bars highlight proximal and distal polyA sites. F Quantification of reads corresponding to proximal and distal polyadenylated FCA transcripts. Data are presented as mean ± s.d. (n = 3). The labeled P-value was analyzed using the two-tailed t-test. .CC-BY-NC 4.0 International licensemade available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is The copyright holder for this preprintthis version posted April 7, 2024. ; https://doi.org/10.1101/2024.04.06.588283doi: bioRxiv preprint