bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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 made available under aCC-BY-NC-ND 4.0 International license.

1 Live-cell single particle tracking of PRC1 reveals a highly dynamic system with 2 low target site occupancy

3

4 Miles K. Huseyin and Robert J. Klose*

5 Department of Biochemistry, University of Oxford, Oxford, United Kingdom

6 *For correspondence: [email protected]

7

8 Abstract

9 Polycomb repressive complex 1 (PRC1) is an essential chromatin-based repressor of gene 10 transcription. However, how PRC1 engages with chromatin to identify its target genes and 11 achieve gene repression remains poorly defined, representing a major hurdle to our 12 understanding of Polycomb system function. Here we use genome engineering and single 13 particle tracking to dissect how PRC1 binds to chromatin in live mouse embryonic stem cells. 14 We reveal that PRC1 is highly dynamic, with only a small fraction stably interacting with 15 chromatin. By integrating subunit-specific dynamics, chromatin binding, and abundance 16 measurements, we discover that PRC1 exhibits surprisingly low occupancy at target sites. 17 Furthermore, we employ perturbation approaches to uncover how specific components of 18 PRC1 define its kinetics and chromatin binding. Together, these discoveries provide a 19 quantitative understanding of chromatin binding by PRC1 in live cells, and suggests that 20 chromatin modification, as opposed to PRC1 complex occupancy, is central to gene 21 repression.

22

23 Introduction

24 Eukaryotic DNA is wrapped around histones to form nucleosomes and chromatin that organise 25 and package the genome within the confines of the nucleus. In addition to this packaging role, 26 chromatin and its post-translational modification can also profoundly influence gene 27 transcription (Atlasi and Stunnenberg, 2017; Kouzarides, 2007; Musselman et al., 2012). 28 Therefore, significant effort has been placed on studying how chromatin-modifying enzymes 29 regulate gene expression. However, in many cases, the mechanisms that enable these 30 enzymes to bind chromatin and identify their appropriate target sites remains poorly 31 understood. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

32 Chromatin-based regulation of gene transcription is typified by the Polycomb repressive 33 system, which is essential for normal gene regulation during animal development (Blackledge 34 et al., 2015; Di Croce and Helin, 2013; Schuettengruber et al., 2017; Simon and Kingston, 35 2009; Vidal, 2019; Voigt et al., 2013). This system is comprised of two central histone- 36 modifying protein complexes, Polycomb repressive complex 1 (PRC1) and PRC2. PRC1 is an 37 E3 ubiquitin ligase that mono-ubiquitylates histone H2A on lysine 119 (H2AK119ub1) (de 38 Napoles et al., 2004; H. Wang et al., 2004). PRC2 is a methyltransferase that methylates 39 histone H3 at lysine 27 (H3K27me1/2/3) (Cao et al., 2002; Czermin et al., 2002; Kuzmichev 40 et al., 2002; Müller et al., 2002). Polycomb complexes bind to their target genes and create 41 Polycomb chromatin domains that are characterised by enrichment of H2AK119ub1, 42 H3K27me3, and the Polycomb repressive complexes themselves (Boyer et al., 2006; Bracken 43 et al., 2006; Lee et al., 2006; Mikkelsen et al., 2007). Once formed, Polycomb chromatin 44 domains are thought to create structural effects on chromatin that repress transcription by 45 limiting access of transcriptional regulators to gene promoters (Eskeland et al., 2010; Francis 46 et al., 2004, 2001; Grau et al., 2011; Isono et al., 2013; King et al., 2002; Lau et al., 2017; 47 Schoenfelder et al., 2015; Shao et al., 1999). However, our understanding of the mechanisms 48 that enable Polycomb complexes to recognise and bind to target sites in vivo remains 49 rudimentary and the extent to which structural effects underpin gene repression remains 50 unclear.

51 Dissecting Polycomb complex targeting mechanisms and their function on chromatin has been 52 challenging because PRC1 and 2 are composed of multiple compositionally diverse 53 complexes. This is exemplified by PRC1, which exists as at least six distinct complexes in 54 mammals. The composition of PRC1 complexes is defined by the PCGF protein (PCGF1-6) 55 that dimerises with either RING1A or RINGB to form the catalytic core. PCGF proteins then 56 interact with auxiliary proteins that regulate catalysis and have unique chromatin binding 57 activities (Endoh et al., 2017; Farcas et al., 2012; Gao et al., 2012; Hauri et al., 2016; He et 58 al., 2013; Junco et al., 2013; Kloet et al., 2016; Morey et al., 2013; Pintacuda et al., 2017; 59 Rose et al., 2016; Tavares et al., 2012; Wu et al., 2013). Based on subunit composition, PRC1 60 complexes are often further separated into canonical and variant forms. Canonical PRC1 61 complexes form around PCGF2 or 4, and interact with CBX and PHC proteins. The CBX 62 proteins have a chromobox domain that binds to H3K27me3 and therefore canonical PRC1 63 complexes are readers of PRC2 catalytic activity and occupy target sites enriched for 64 H3K27me3 (Bernstein et al., 2006; Fischle et al., 2003; Zhen et al., 2016). Canonical PRC1 65 complexes also enable the formation of long-range interactions between Polycomb chromatin 66 domains and create more localised chromatin structures that are proposed to inhibit gene 67 expression (Bonev et al., 2017; Gonzalez et al., 2014; Isono et al., 2013; Kim et al., 2002; King bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

68 et al., 2002; Kundu et al., 2017). In contrast, variant PRC1 complexes form primarily around 69 PCGF1/3/5/6 and interact with RYBP or YAF2 in place of CBX proteins (Gao et al., 2012; 70 Morey et al., 2013; Tavares et al., 2012). Therefore, variant PRC1 complexes do not recognise 71 H3K27me3 and instead have distinct chromatin binding activities. For example, PCGF1- and 72 PCGF6-PRC1 incorporate auxiliary proteins with DNA binding activity that target these 73 complexes to gene promoters (Endoh et al., 2017; Farcas et al., 2012; Gearhart et al., 2006; 74 He et al., 2013; Wu et al., 2013). Furthermore, variant PRC1 complexes are highly active E3 75 ubiquitin ligases and define most H2AK119ub1 in vivo (Fursova et al., 2019; Gao et al., 2012; 76 Rose et al., 2016; Taherbhoy et al., 2015).

77 A series of mechanisms, identified mostly from in vitro biochemical experiments, have been 78 proposed to explain how individual PRC1 complexes bind to chromatin (Blackledge et al., 79 2015; Schuettengruber et al., 2017; Simon and Kingston, 2013). The contributions of these 80 mechanisms to chromatin binding in vivo have primarily been examined by ensemble fixation- 81 based approaches like chromatin immunoprecipitation coupled with massively parallel 82 sequencing (ChIP-seq). This has provided a static snapshot of PRC1 complex distribution 83 throughout the genome and perturbation experiments have provided some information about 84 the mechanisms that enable specific PRC1 complexes to bind chromatin. However, ChIP-seq 85 is blind to kinetics and cannot directly compare and quantitate the chromatin binding activities 86 of individual PRC1 complexes. As such, we currently lack a quantitative model to describe 87 chromatin binding and the function of PRC1 in live cells. This represents a major conceptual 88 gap in our understanding of the Polycomb system and its role in gene regulation.

89 To begin addressing this, live cell fluorescence recovery after photo bleaching (FRAP) 90 approaches have been used to study chromatin binding for a number of Polycomb proteins in 91 Drosophila (Ficz et al., 2005; Fonseca et al., 2012; Steffen et al., 2013). These important 92 studies revealed that Polycomb proteins interact with chromatin more dynamically than was 93 expected from in vitro chromatin binding experiments, and similar conclusions have emerged 94 from FRAP experiments in mammalian cell culture systems (Almeida et al., 2017; Ren et al., 95 2008; Vandenbunder et al., 2014; Youmans et al., 2018). However, FRAP infers single 96 molecule kinetics from ensemble measurements and therefore can overlook essential 97 chromatin binding behaviours that are only evident when individual molecules are directly 98 observed (Mazza et al., 2012; McNally, 2008; Mueller et al., 2010). To overcome this limitation, 99 several recent studies have leveraged single-molecule live-cell imaging approaches to shed 100 new light on how previously proposed chromatin binding mechanisms shape PRC2 dynamics 101 (Tatavosian et al., 2018; Youmans et al., 2018). Similar approaches have also provided initial 102 descriptions of chromatin binding by CBX proteins, which are specific to canonical PRC1 103 complexes (Tatavosian et al., 2018; Zhen et al., 2016). However, because PRC1 is composed bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

104 of a diverse set of canonical and variant PRC1 complexes, an understanding of chromatin 105 binding by PRC1 in live cells remains absent, presenting a major barrier to understanding how 106 PRC1 regulates gene expression in vivo.

107 To overcome this, here we combine genome editing and single particle tracking (SPT) to 108 quantify and dissect chromatin binding of PRC1 in live mouse embryonic stem cells (ESCs). 109 This reveals that PRC1 is highly dynamic, with a small number of molecules displaying stable 110 binding to chromatin. By quantifying absolute PRC1 complex numbers and integrating 111 genomic information, we estimate maximum target site occupancy and discover that most 112 PRC1 target genes are sparsely bound by PRC1 complexes. In dissecting the mechanisms 113 that underpin chromatin binding by PRC1, we discover that interaction between its catalytic 114 core and the nucleosome contributes little to chromatin binding, indicating that the observed 115 binding behaviours of PRC1 must be defined by auxiliary subunits that are specific to individual 116 PRC1 complexes. By systematically characterising chromatin binding and occupancy by 117 individual PRC1 complexes, we reveal how distinct chromatin binding modalities are related 118 to the activity and function of PRC1. Furthermore, using genetic perturbation approaches, we 119 dissect the contribution of canonical and variant PRC1-specific targeting mechanisms to their 120 chromatin-binding activities. Together, these new discoveries provide a quantitative 121 understanding of chromatin binding by PRC1 complexes in vivo and have important 122 implications for our understanding of PRC1-mediated gene repression.

123

124 Results

125 Live-cell imaging reveals that a small fraction of PRC1 is bound to chromatin

126 To study PRC1 in live cells we used CRISPR/Cas9-mediated genome editing to engineer a 127 HaloTag (Los et al., 2008) into both alleles of the endogenous Ring1b gene in mouse 128 embryonic stem cells (ESCs). We chose RING1B as a proxy for PRC1 complex behaviour 129 because its paralogue RING1A is very lowly expressed in ESCs, RING1B forms the core 130 subunit of both canonical and variant PRC1 complexes, and its chromatin binding has been 131 studied extensively by fixation-based approaches (Farcas et al., 2012; Gao et al., 2012; Hauri 132 et al., 2016; Kloet et al., 2016; Ku et al., 2008; Tavares et al., 2012; Wang et al., 2010). 133 Importantly, addition of a HaloTag to RING1B (RING1B-HaloTag) did not alter RING1B 134 expression, H2AK119 ubiquitylation, or PRC1 complex formation (Figure 1A and B). When a

135 HaloTag compatible dye (JF549) (Grimm et al., 2017) was applied to RING1B-HaloTag cells 136 this allowed us to image RING1B in the nucleus of live cells (Figure 1C), and we observed a 137 distribution of RING1B signal that was similar to previous reports (Almeida et al., 2017; Isono 138 et al., 2013; Ren et al., 2008). bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

139 To measure the diffusion and chromatin binding of PRC1, we labelled RING1B-HaloTag with

140 a photoactivatable dye (PA-Halo-JF549) (Lavis et al., 2016) and performed high temporal 141 resolution (15 ms exposure) SPT with highly inclined and laminated optical sheet (HILO) 142 microscopy (Tokunaga et al., 2008) (Figure 2A and B). When the resulting tracks were 143 analysed with Spot-On (Hansen et al., 2018), a three-state kinetic model with immobile, slowly 144 diffusing, and fast diffusing molecules fit the data well. In contrast, a two-state model with only 145 immobile and diffusing molecules fit the data poorly (Figure S1A). These observations suggest 146 there is an immobile fraction of PRC1 which is bound to chromatin, a slowly diffusing fraction 147 that may correspond to transient interactions with chromatin or confinement within areas of 148 the nucleus, and a fraction that is freely diffusing (Izeddin et al., 2014). Fitting our SPT 149 measurements to this model revealed that 20% of RING1B was chromatin-bound, with the 150 remainder existing in a slowly or freely diffusing state (Figure 2C). To contextualise these 151 measurements, we generated cell lines stably expressing HaloTag-histone H2B, which is 152 stably incorporated into chromatin, or HaloTag-with a nuclear localisation signal, which does 153 not bind specifically to chromatin. SPT measurements using these cells yielded different 154 behaviours for each protein that were consistent with previous observations (Hansen et al., 155 2018) (Figure S1B). HaloTag-H2B had a bound fraction of 59% whereas HaloTag-NLS had a 156 bound fraction of 11%, revealing that the bound fraction of RING1B was towards the lower 157 end of these two extremes (Figure 2C). Therefore, our SPT measurements reveal that 158 RING1B, and therefore PRC1, exists predominantly in diffusing states and has only a small 159 chromatin-bound fraction.

160 Only a fraction of chromatin bound PRC1 displays stable binding

161 High temporal resolution SPT demonstrated that 20% of RING1B was bound to chromatin. 162 However, these imaging conditions did not allow us to measure the length of more stable 163 chromatin binding events because individual molecules photobleach within a few seconds. To 164 overcome this limitation, we carried out SPT using reduced laser power with longer exposure 165 times (0.5 s). Under these imaging conditions, diffusing molecules blur while stably bound 166 molecules are visible and photobleach less rapidly (Figure 2D). Using this approach, the 167 distribution of track lengths for chromatin-bound RING1B molecules fit well to a biexponential 168 decay modelling short and long-lived events (Figure 2E). The mean survival times obtained 169 from this model are influenced by photobleaching. Therefore we used HaloTag-H2B, which 170 binds stably to chromatin on the order of hours (Kimura and Cook, 2001), as a control to 171 correct for the influence of photobleaching (Hansen et al., 2017; Uphoff et al., 2013). However, 172 we were still unable to accurately estimate the duration of long binding events as the observed 173 dwell times for RING1B were similar to the photobleaching time (Figure 2E). Despite this 174 limitation, our measurements revealed that 38% of observed binding events were stable and bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

175 some molecules remained bound to chromatin for up to, and beyond, 100 s. To extend the 176 time that we could image long binding events, we reduced the acquisition frequency to 30 s 177 intervals. Under these conditions, we observed some RING1B molecules that survived for 178 hundreds of seconds, but photobleaching still limited our capacity to accurately estimate stable 179 binding times (Figure 2F and G). These observations demonstrate that while only 20% of 180 PRC1 is bound to chromatin, and less than half of these binding events are stable, those 181 molecules that bind stably do so for relatively long periods of time.

182 To investigate whether the stable binding events we observe in SPT are representative of 183 PRC1 behaviour at known sites of chromatin interaction, we focussed on Polycomb bodies. 184 Polycomb bodies are cytologically distinct foci that correspond to a subset of PRC1-bound 185 regions of the genome, including the Hox gene clusters, that have very large Polycomb 186 chromatin domains (Boyle et al., 2019; Isono et al., 2013; Ren et al., 2008; Saurin et al., 1998; 187 Wani et al., 2016). Live-cell imaging of RING1B revealed approximately one hundred 188 Polycomb bodies per nucleus (Figure 2H). Using FRAP measurements, we examined 189 RING1B dynamics and chromatin binding in Polycomb bodies or equally sized regions of the 190 nucleoplasm. This revealed slower and less complete recovery of fluorescence signal in 191 Polycomb bodies, demonstrating that known PRC1 interaction sites also display stable 192 RING1B binding (Figure 2I). However, our measurements revealed that Polycomb bodies 193 accounted for just 1.3% of the total nuclear volume and that RING1B fluorescence signal 194 inside Polycomb bodies was only 1.3-fold more intense than the surrounding nucleus (Figure 195 2J). This means that only 1.7% of total PRC1 signal originates from Polycomb bodies, 196 consistent with evidence from ChIP-seq experiments that PRC1 also binds more stably to 197 thousands of other smaller target sites in the genome (Blackledge et al., 2020; Endoh et al., 198 2012; Kloet et al., 2016; Morey et al., 2013). Therefore, PRC1 binds chromatin more stably at 199 Polycomb target sites and our SPT approach allows us to capture these events in live cells 200 throughout the genome.

201 Absolute PRC1 quantification estimates low occupancy at PRC1 target sites

202 There is currently very limited quantitative understanding of PRC1 abundance and occupancy 203 at target sites in the genome. However, the biochemical makeup of Polycomb chromatin 204 domains will have important implications for how the Polycomb system functions to repress 205 gene expression. Having quantified the fraction of RING1B that is bound to chromatin in live 206 cells, we wanted to use this new information to estimate the maximum number of molecules 207 of PRC1 that might bind to a typical target site. To achieve this, we first quantified the number 208 of RING1B molecules in ESCs using a biochemical approach (Cattoglio et al., 2019) that 209 compares in-gel fluorescence of RING1B-HaloTag from a defined number of cells to the bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

210 fluorescence of a recombinant HaloTag protein of known concentration (Figure 3A). From this 211 we calculated that there were ~63,000 molecules of RING1B per cell, which is similar to 212 estimates in other cell types (Beck et al., 2011; Schwanhäusser et al., 2011; Wiśniewski et al., 213 2014). SPT demonstrated that only 20% of RING1B was bound to chromatin (Figure 2C), 214 which corresponds to ~12,600 bound molecules. To estimate the maximum number of 215 molecules that could possibly bind to RING1B-occupied target sites identified in ChIP-seq 216 (18,643 sites) (Blackledge et al., 2020), we assumed that the bound fraction of RING1B was 217 concentrated exclusively in these sites. Based on this liberal assumption, we estimate that 218 there would be on average 0.1 RING1B molecules for every kilobase of RING1B-enriched 219 chromatin. However, the distribution of RING1B ChIP-seq signal varies significantly across its 220 binding sites in the genome (Figure 3B and C). Even if this distribution is taken into 221 consideration, sites in the top decile of RING1B density would still have fewer than 0.3 222 molecules per kb. Despite being very small, these occupancy values will be an overestimation 223 because we know that PRC1 binding events also occur away from sites identified in RING1B 224 ChIP-seq as evidenced by pervasive H2AK119ub1 throughout the genome (Fursova et al., 225 2019; Lee et al., 2015). Therefore, our calculations suggest that PRC1 occupancy at target 226 sites in the genome is on average very low.

227 Interaction between the PRC1 catalytic core and the nucleosome is not a central 228 determinant of chromatin binding

229 Having characterised chromatin binding by PRC1 in SPT, we then set out to dissect the 230 mechanisms that define this behaviour. RING1B binds to the nucleosome in a specific 231 orientation and through a conserved set of residues to ubiquitylate H2A. Mutating these 232 residues renders RING1B incapable of binding to nucleosomes and catalysing H2AK119ub1 233 in vitro (Bentley et al., 2011; McGinty et al., 2014). Therefore, we reasoned that nucleosome 234 interaction might contribute centrally to chromatin binding in vivo. To test this, we substituted 235 key residues in the nucleosome-binding domain (RING1BNBM) and stably expressed 236 RING1BNBM-HaloTag or wild type RING1B-HaloTag in a cell line that lacks RING1A and where 237 removal of endogenous RING1B can be induced by addition of tamoxifen (PRC1CKO) (Figure 238 4A and B) (Bentley et al., 2011; Blackledge et al., 2020; McGinty et al., 2014). Following 239 tamoxifen treatment, cells expressing wild type RING1B-HaloTag supported normal 240 H2AK119ub1, cell morphology, and pluripotency (Figures 4B and S2A). Importantly, 241 RING1BNBM failed to maintain these features, demonstrating it was non-functional in vivo. We 242 then carried out SPT in the presence of endogenous RING1B to maintain cell viability. This 243 revealed that the fraction of bound and stably bound molecules of RING1BNBM was almost 244 identical to that of the wild type protein (Figure 4C and D) and stable binding remained beyond 245 the photobleaching limit of the experiment, demonstrating that the interaction between bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

246 RING1B and the nucleosome does not contribute centrally to chromatin binding by PRC1. In 247 agreement with this, a minimal catalytic domain that does not interact with PRC1 auxiliary 248 factors, but is capable of depositing H2AK119ub1, had a chromatin bound fraction (8%) that 249 was indistinguishable from the HaloTag-NLS control (Figures S2B, 4E and F) (Bentley et al., 250 2011; Blackledge et al., 2014; Buchwald et al., 2006; Cooper et al., 2016). Therefore, we 251 conclude that interaction between the catalytic core of PRC1 and the nucleosome does not 252 contribute centrally to chromatin binding, and propose instead that this must rely on auxiliary 253 proteins in PRC1 complexes.

254 PCGF2-PRC1 is the most abundant PRC1 complex in ESC

255 PCGF proteins dimerise with RING1B and interact with auxiliary PRC1 subunits that are 256 thought to contribute to chromatin binding by PRC1 (Gao et al., 2012; Hauri et al., 2016; Junco 257 et al., 2013; Kloet et al., 2016). Therefore, we reasoned that understanding the composition 258 of PRC1 complexes and defining their individual chromatin binding behaviours in live cells 259 would be required to discover the mechanisms that underpin PRC1 targeting and function in 260 gene repression. To achieve this, we added a HaloTag to both alleles of the endogenous 261 Pcgf1, Pcgf2, Pcgf3, and Pcgf6 genes (Figures 5A and B, and S3A, B, C and E). PCGF4 is 262 not expressed in ESCs so we excluded it from our analysis. PCGF3 and 5 form nearly identical 263 PRC1 complexes and PCGF5 are expressed at similar or slightly lower levels, so PCGF3 was 264 used as a proxy for PCGF3/5-PRC1 complex behaviour (Gao et al., 2012; Hauri et al., 2016; 265 Kloet et al., 2016). In ESCs, canonical PRC1 complexes form around PCGF2 and 266 predominantly contain CBX7, whereas variant PRC1 complexes form mostly around PCGF1, 267 3, 5, and 6 and predominantly contain RYBP (Figure 5A) (Gao et al., 2012; Hauri et al., 2016; 268 Kloet et al., 2016; Morey et al., 2013; Tavares et al., 2012). Therefore, to capture chromatin 269 binding by canonical and variant PRC1 complexes, we added a HaloTag to both alleles of the 270 Cbx7 and Rybp genes (Figures 5B, S3D and F).

271 To determine the abundance of individual PRC1 complexes in ESCs we first quantified the 272 number of PCGF molecules and in parallel determined their relative levels by fluorescence 273 microscopy. This showed that individual PCGF molecules were expressed at lower levels than 274 RING1B, consistent with each PCGF complex constituting a fraction of total PRC1 (Figure 275 5C). PCGF2 was most abundant with ~26000 molecules per cell and corresponded to 40% of 276 total RING1B. The number of CBX7 molecules (~19,000) was similar to that of PCGF2, 277 suggesting that canonical PCGF2/CBX7-PRC1 complexes are the predominant form of PRC1 278 in ESCs. Differences in the absolute number of PCGF2 and CBX7 molecules likely results 279 from some PCGF2 forming variant complexes (Gao et al., 2012; Morey et al., 2013; Tavares 280 et al., 2012) and the incorporation of other more lowly expressed CBX proteins in PCGF2- bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

281 PRC1 complexes (Hauri et al., 2016; Kloet et al., 2016). In contrast, PCGF proteins that 282 exclusively form variant PRC1 complexes were less abundant. There were ~14,000 PCGF6, 283 ~5000 PCGF1, and ~3000 PCGF3 molecules per cell. Given that PCGF3 and PCGF5 are 284 expressed at similar levels (Figure S3C) (Fursova et al., 2019; Kloet et al., 2016; Scelfo et al., 285 2019), we estimate that there are ~6000 PCGF3/5 molecules per cell. At ~30,000 molecules 286 per cell, RYBP was slightly more abundant than PCGF1, 3, 5 and 6 together, in agreement 287 with a small amount of PCGF2 forming variant PRC1 complexes (Gao et al., 2012; Hauri et 288 al., 2016; Morey et al., 2013; Tavares et al., 2012). Importantly, the total sum of PCGF 289 molecules, or CBX7/RYBP molecules, was similar to the number of RING1B molecules, 290 consistent with RING1B forming a core scaffold for PRC1 complex assembly (Buchwald et al., 291 2006; Francis et al., 2001; Fursova et al., 2019; Li et al., 2006; Suzuki et al., 2002; H. Wang 292 et al., 2004; Wang et al., 2010). This detailed quantification of canonical and variant PRC1 293 subunit abundance suggests that canonical PRC1 is the most abundant form of PRC1 in 294 ESCs.

295 Canonical PRC1 is characterised by stable chromatin binding and is restricted to a 296 subset of PRC1 bound sites

297 Based on our quantification of PRC1 complex abundances, we then set out to discover how 298 these complexes enable the chromatin binding behaviour and function of PRC1. Initially we 299 focussed on the canonical PRC1 components PCGF2/CBX7 and carried out SPT. This 300 revealed that 18% of PCGF2 and CBX7 were chromatin bound, which was similar to RING1B 301 (20%) (Figure 6A). Survival time imaging revealed that PCGF2 and CBX7 had a slightly larger 302 fraction of stably bound molecules (44% and 50%) than RING1B (38%) with their stable 303 binding time estimations also being limited by photobleaching (Figure 6B). It has been shown 304 that canonical PRC1 is important for the formation of Polycomb bodies (Isono et al., 2013; 305 Plys et al., 2019; Tatavosian et al., 2019). Consistent with this, when we examined the 306 distribution of PCGF2 in the nucleus it was more enriched (2.3-fold) in Polycomb bodies than 307 RING1B (Figure S4A) and FRAP of PCGF2 in these regions displayed similar kinetics to 308 RING1B (Figure S4B). We estimate that on average each Polycomb body has 9 PCGF2 and 309 10 RING1B molecules, suggesting that these foci are mostly composed of canonical PRC1 310 complexes and is consistent with the more prominent enrichment of PCGF2/CBX7 in these 311 foci compared to variant PRC1 subunits (Figure 5B). However, despite the visible enrichment 312 of PCGF2 and RING1B in Polycomb bodies, we estimate that there are relatively few 313 molecules bound at these sites, as even the brightest foci correspond to not more than ~50 314 PRC1 molecules, and that these regions correspond only to a small fraction of total PRC1 315 (1.7% for RING1B and 3.6% for PCGF2). Together, these observations suggest that while bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

316 canonical PRC1 is more restricted to Polycomb bodies than PRC1 as a whole, Polycomb 317 bodies in ESCs do not correspond to large accumulations of Polycomb proteins.

318 Previous ChIP-seq analysis has shown that the distributions of PCGF2 and RING1B are not 319 identical (Blackledge et al., 2020; Fursova et al., 2019; Kloet et al., 2016; Morey et al., 2013; 320 Scelfo et al., 2019). By combining information detailing PCGF2 binding from ChIP-seq analysis 321 and our estimates of the number of bound molecules in live-cell imaging, we found that PCGF2 322 occupancy was most dense at Polycomb target sites in the top decile of RING1B density (0.09 323 molecules per kb). PCGF2 occupancy then decayed quickly (to between 0.01 and 0.03 324 molecules per kb) at sites with lower levels of RING1B (Figure 6C and D). PRC2 is also most 325 enriched at high density RING1B sites in ChIP-seq and present in Polycomb bodies 326 (Blackledge et al., 2020) which have elevated levels of H3K27me3. This is consistent with the 327 restricted nature of PCGF2 occupancy relying on CBX proteins in canonical PRC1 complexes 328 binding to H3K27me3 (Kloet et al., 2016; Morey et al., 2013; Tavares et al., 2012).

329 Variant PRC1 complexes are characterised by distinct dynamics and chromatin binding

330 Having demonstrated that canonical PRC1 displays stable chromatin binding and occupancy 331 at a restricted subset of PRC1 bound sites, we then set out to examine the behaviour of variant 332 PRC1 complexes and to determine their contribution to PRC1 complex dynamics and 333 chromatin binding. Interestingly, SPT revealed that variant PRC1 complexes had dynamics 334 that were distinct from canonical PRC1. PCGF3 and PCGF6 had smaller bound fractions than 335 RING1B (13% and 17% respectively) and were more similar to that of RYBP (13%) (Figure 336 7A). In contrast, PCGF1 had the highest bound fraction (26%) of any PCGF protein. Survival 337 time imaging revealed that, although all variant PRC1 subunits had similar stably bound 338 fractions to RING1B, the average length of stable binding events was between 30 and 50 339 seconds for RYBP, PCGF3, and PCGF6 (Figure 7B and C). This is much shorter than the 340 stable binding events observed for canonical PRC1, which extended beyond the 341 photobleaching limit of our experiments. In contrast, PCGF1 displayed long binding times that 342 were indistinguishable from canonical PRC1. This indicates that the PCGF1-PRC1 complex 343 binds to chromatin more frequently and stably than other variant PRC1 complexes.

344 The differences in dynamics between variant PRC1 complexes suggest they must have 345 distinct chromatin binding mechanisms. We reasoned that contextualising these behaviours 346 was important as variant PRC1 complexes are responsible for almost all H2AK119ub1 in the 347 genome and have recently been proposed to be central to PRC1-mediated gene repression 348 (Fursova et al., 2019; Scelfo et al., 2019). We first considered PCGF3 dynamics in the context 349 of its known role in depositing low levels of H2AK119ub1 indiscriminately across the genome 350 (Fursova et al., 2019). Our attempts to ChIP-seq PCGF3/5 have yielded no obvious bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

351 enrichment profile, with similar approaches yielding very few enriched sites (Scelfo et al., 352 2019). An inability to capture PCGF3/5-PRC1 by ChIP is consistent with our SPT 353 measurements showing that PCGF3 has a small chromatin-bound fraction and less stable 354 binding than other PCGF molecules. Furthermore, this is consistent with previous reports 355 suggesting that fixation based chromatin binding assays fail to effectively capture transient 356 chromatin binding events (Schmiedeberg et al., 2009; Teves et al., 2016). More importantly 357 however, we posit that the dynamic chromatin interactions that PCGF3-PRC1 engages in may 358 be ideally suited to the unique role PCGF3/5 play in depositing H2AK119ub1 pervasively 359 throughout the genome (Fursova et al., 2019) In line with this, we estimate that there are 5.9 360 x 106 H2AK119ub1 molecules in a diploid ESC genome and H2AK119ub1 has a half-life of 361 ~90 minutes (Seale, 1981), necessitating replacement of 750 molecules of H2AK119ub1 by 362 PRC1 each second. Approximately 50% of H2AK119ub1 is deposited by PCGF3/5-PRC1 363 complexes (Fursova et al., 2019), of which we estimate there are ~6000 molecules. Each 364 PCGF3/5-PRC1 complex would therefore need to carry out on average at least one 365 ubiquitylation event every 16 s, which is on a similar scale to the stable binding times we have 366 measured for PCGF3 in live cells. This approximation suggests that the highly dynamic nature 367 of PCGF3 relative to other PRC1 subunits may be necessary for the role it plays in depositing 368 H2AK119ub1 throughout the genome.

369 In contrast to PCGF3/5, ChIP-seq approaches have shown that PCGF1- and PCGF6-PRC1 370 complexes bind predominantly at PRC1 target genes. The capacity to capture these proteins 371 in ChIP-seq is consistent with their higher bound fraction and longer binding times in SPT 372 experiments. By integrating PCGF1 protein numbers, the bound fraction from SPT, and ChIP- 373 seq distributions, we estimate a narrow range of PCGF1 occupancy (between 0.01 to 0.02 374 molecules per kb) across PRC1 target sites (Figure 7D and E). Despite the distribution of 375 PCGF6 ChIP-seq signal correlating slightly better with RING1B at PRC1 target sites, similar 376 estimations for PCGF6 show that even at sites with the highest density of PCGF6, there would 377 on average only be 0.04 molecules of PCGF6 per kb (Figure 7D and E). Therefore, in contrast 378 to canonical PRC1, variant PCGF1 and PCGF6-PRC1 complexes have a more monotone 379 distribution across PRC1 target sites and even lower occupancy levels. This suggests that 380 low-level occupancy of variant PRC1 complexes at target sites is sufficient to maintain 381 enrichment of H2AK119ub1 and support gene repression.

382 PRC2 and H3K27me3 regulate the frequency of chromatin binding by canonical PRC1

383 Our SPT experiments demonstrated that PCGF2 and PCGF1-PRC1 complexes had similar 384 stable chromatin binding. However, these PRC1 complexes have different distributions at 385 target sites in the genome, are proposed to utilise distinct chromatin binding mechanisms, and bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

386 contribute uniquely to Polycomb system function. Therefore, we were keen to explore the 387 underlying mechanisms that enable these kinetically similar chromatin-binding behaviours but 388 lead to distinct targeting and function. Initially we focussed on the canonical PCGF2-PRC1 389 complex because its CBX proteins bind H3K27me3 at PRC1 target sites and this modification 390 is known to be important for occupancy as measured by ChIP (Boyer et al., 2006; Leeb et al., 391 2010; Schoeftner et al., 2006; Tavares et al., 2012). To examine how H3K27me3 affects 392 PCGF2 dynamics and chromatin binding in live cells we used the recently developed 393 degradation tag (dTAG) system (Nabet et al., 2018) to induce disruption of the enzyme 394 complex that places H3K27me3. To achieve this, we introduced a dTAG into both alleles of 395 the endogenous Suz12 gene. SUZ12 is a core structural subunit of PRC2 and is required for 396 its histone methyltransferase activity (Cao and Zhang, 2004; Jiao and Liu, 2015; Pasini et al., 397 2004). Treatment of dTAG-SUZ12 cell lines with the dTAG-13 compound caused rapid 398 degradation of SUZ12 and a complete loss of H3K27me3 by 96 hours of treatment (Figure 8A 399 and B). We used this approach to induce removal of SUZ12 in PCGF2-HaloTag and RING1B- 400 HaloTag cell lines while using a HaloTag-PCGF6 cell line as a control for a PRC1 complex 401 that lacks H3K27me3-binding CBX proteins.

402 We then carried out SPT prior to and following SUZ12/H3K27me3 depletion. After dTAG-13 403 treatment, the bound fraction of PCGF2 and RING1B was significantly reduced, in agreement 404 with observations from ChIP experiments that loss of H3K27me3 reduces canonical PRC1 405 occupancy (Figure 8C) (Boyer et al., 2006; Leeb et al., 2010; Schoeftner et al., 2006; Tavares 406 et al., 2012; L. Wang et al., 2004). Importantly, PCGF6-PRC1 chromatin binding was 407 unaffected. Interestingly, despite a reduction in the total bound fraction of RING1B and 408 PCGF2, there was no significant effect on the stability of the binding events that did occur 409 (Figure 8D and E). Furthermore, when we examined Polycomb bodies, there was only a 410 modest effect on the number of PCGF2 foci and a slight downwards shift in the distribution of 411 PCGF2 foci volumes (Figures 8F, G and H, and S5A and B). This indicates that H3K27me3 412 recognition is primarily required to increase the frequency with which canonical PRC1 binds 413 chromatin, but not the stability of these binding events when they occur. Furthermore, it 414 suggests that H3K27me3-independent mechanisms must predominate in supporting stable 415 chromatin binding by canonical PRC1.

416 The C-terminal RAWUL domain of PCGF1 is required for stable chromatin binding

417 Although PCGF1 exhibited similar binding stability to canonical PRC1 subunits, these PRC1 418 complexes lack shared auxiliary subunits. This suggested the mechanisms that support stable 419 chromatin binding by PCGF1-PRC1 must be distinct from those of PCGF2-PRC1. Therefore, 420 we were keen to define how PCGF1 interfaces with chromatin. PCGF1 can be separated into bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

421 two distinct regions based on domain conservation (Figure S6A) (Gearhart et al., 2006; Gong 422 et al., 2005; Wong et al., 2016). The N-terminus of PCGF1 contains a RING domain through 423 which it heterodimerises with RING1B, forming a RING-RING domain dimer that is responsible 424 for the ubiquitin ligase activity of the complex. In contrast, the C-terminus of PCGF1 contains 425 a RAWUL domain which interacts with the PCGF1-PRC1 complex specific proteins KDM2B 426 and BCOR/BCORL1 (Gearhart et al., 2006; Junco et al., 2013; Rose et al., 2016; Wong et al., 427 2016). In order to determine which of these domains and their interactions might explain the 428 chromatin binding behaviour of PCGF1 we generated cell lines stably expressing exogenous 429 HaloTag-PCGF1 fusion proteins corresponding to the full-length protein and its N- and C- 430 terminal domains (Figures 9A and S6A and B). Full-length PCGF1 exhibited stable binding 431 that was comparable to that of the endogenous protein, although it had a smaller bound 432 fraction (Figure 9B and C). When we examined the behaviour of the N- and C-terminal 433 fragments, the bound fraction for each was similar to the full-length protein, with a modest 434 reduction in the binding frequency of the C-terminal fragment (Figure 9B). Strikingly, however, 435 we could not detect stable binding of the N-terminal fragment, while the C-terminal fragment 436 almost completely recapitulated the behaviour of the full-length protein (Figure 9C). Together, 437 these observations indicate that stable chromatin binding by PCGF1 requires the C-terminal 438 RAWUL domain, possibly through the chromatin binding activities of KDM2B and 439 BCOR/BCORL, but not its RING1B dimerization domain. Therefore, in contrast to canonical 440 PRC1 that requires interactions with H3K27me3 to increase the frequency but not stability of 441 binding to chromatin, we discover that the RAWUL domain of PCGF1 enables stable binding, 442 but does not define the frequency of binding. Together these observations indicate that 443 auxiliary PRC1 complex components contribute in distinct ways to the kinetics of chromatin 444 binding and that these important behaviours can only be uncovered using single-molecule live 445 cell approaches.

446

447 Discussion

448 Here, using endogenous protein tagging and single-molecule imaging, we systematically 449 characterise the behaviour of PRC1 in live cells. We discover that PRC1 complexes are highly 450 dynamic and only a small fraction binds stably to chromatin (Figures 1 and 2). By quantifying 451 the number of RING1B molecules in single cells, the fraction that are chromatin bound, and 452 estimating their occupancy throughout the genome, we discover that a surprisingly small 453 number of PRC1 complexes are bound at target genes (Figure 3). We show that chromatin 454 binding by PRC1 does not rely on the intrinsic nucleosome binding activity of its catalytic core, 455 suggesting a central role for auxiliary proteins in chromatin binding by PRC1 (Figure 4). By bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

456 testing how different PRC1 complexes create the observed dynamics of PRC1, we discover 457 that canonical PCGF2-PRC1 is most abundant, constitutes the majority of chromatin-bound 458 PRC1, and is enriched in Polycomb bodies (Figures 5 and 6). Conversely, we discover that 459 variant PRC1 complexes, made up of PCGF1, 3, or 6, are less numerous and have distinct 460 chromatin binding properties that correspond to their target site recognition and roles in 461 depositing H2AK119ub1 (Figure 7). Building on these new findings, we perturb proposed 462 PRC1 targeting mechanisms, and uncover a role for H3K27me3 in increasing the frequency 463 with which canonical PRC1 binds chromatin, but not its capacity to bind stably to chromatin 464 (Figure 8). In contrast, we reveal that stable chromatin binding by the variant PCGF1-PRC1 465 complex relies on the RAWUL domain of PCGF1, which is known to interact with DNA binding 466 factors (Figure 9). Together, these new findings provide a detailed quantitative understanding 467 of the dynamics of PRC1 complexes and their interactions with chromatin in live cells, and 468 reveals that a surprisingly small number of PRC1 complexes and low target gene occupancy 469 is sufficient to support H2AK119ub1 and gene repression by the Polycomb system.

470 Although Polycomb repressive complexes have been extensively studied, the mechanisms by 471 which they affect transcriptional repression have remained poorly understood (Blackledge et 472 al., 2015; Vidal, 2019). PRC1 catalyses H2AK119ub1, and this has been implicated in 473 transcriptional repression (Blackledge et al., 2020, 2014; Endoh et al., 2012; Tamburri et al., 474 2020; Tsuboi et al., 2018). However, PRC1 protein complexes also occupy target genes and 475 have been proposed to repress gene expression through a number of more direct and 476 catalysis-independent mechanisms. These include limiting chromatin accessibility, supporting 477 chromatin compaction/interaction, or possibly through phase separation (Eskeland et al., 478 2010; Francis et al., 2004, 2001; Grau et al., 2011; Isono et al., 2013; King et al., 2002; Kundu 479 et al., 2017; Lau et al., 2017; Plys et al., 2019; Tatavosian et al., 2019). Based on these 480 observations, it has been debated whether PRC1 primarily uses H2AK119ub1 or catalysis- 481 independent functions for transcriptional repression. If PRC1 primarily represses transcription 482 via catalysis-independent mechanisms, one might expect a requirement for high-level physical 483 occupancy of PRC1 at target genes. However, from our quantitative measurements of PRC1 484 subunit abundance and chromatin binding in live cells, we estimate there are a surprisingly 485 small number of PRC1 complexes bound to target sites at any given time. This is also true if 486 one focuses on canonical PRC1 complexes that are responsible for effects on chromatin 487 compaction and architecture (Eskeland et al., 2010; Francis et al., 2004; Grau et al., 2011; 488 Kundu et al., 2017; Lau et al., 2017; Schoenfelder et al., 2015). Furthermore, when we 489 examine Polycomb bodies, which have been proposed to correspond to phase separated 490 entities (Banani et al., 2017), the average number and concentration (10 molecules, 130 nM) 491 of PRC1 complexes inside them is below that which is thought to be required to support phase bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

492 separation (Erdel et al., 2020; Plys et al., 2019; Tatavosian et al., 2019). Therefore, the 493 sparseness of PRC1 occupancy at Polycomb target sites and in Polycomb bodies, suggests 494 that structural effects and phase separation are likely not the primary determinants of PRC1- 495 mediated gene repression in ESCs. These inferences are consistent with our previous findings 496 that deletion of PCGF2/4, which disrupts canonical PRC1 complexes, has a limited effect on 497 gene repression by PRC1 in this cell type (Fursova et al., 2019) and that Polycomb complexes 498 do not appear to limit chromatin accessibility (King et al., 2018). Furthermore, it is consistent 499 with the fact that CBX7, which in ESCs is the predominant CBX protein in canonical PRC1 500 complexes, does not exhibit phase separating behaviour in vitro (Grau et al., 2011; Plys et al., 501 2019). Interestingly, it has also been reported in Drosophila that there are very low cellular 502 concentrations of PRC1 complex components in the developing embryo (Bonnet et al., 2019; 503 Fonseca et al., 2012; Steffen et al., 2013). Unless the stably bound fraction of molecules in 504 these systems is much higher, it would also be difficult to rationalise how physical or structural 505 effects via the occupancy of these complexes on chromatin could be the primary effector of 506 PRC1-mediated gene repression.

507 In contrast to canonical PRC1, removal of variant PRC1 complexes in ESCs causes 508 derepression of thousands of Polycomb target genes (Fursova et al., 2019; Scelfo et al., 509 2019). A central distinction between canonical and variant PRC1 complexes is their catalytic 510 activity. Canonical PRC1 complexes have limited capacity to catalyse H2AK119ub1 in vitro 511 and contribute virtually nothing to H2AK119ub1 in vivo (Fursova et al., 2019; Rose et al., 2016; 512 Taherbhoy et al., 2015). In contrast, variant PRC1 complexes are highly active in vitro and 513 account for almost all H2AK119ub1 in vivo. Catalytic mutation of PRC1 causes Polycomb 514 target gene derepression, despite variant PRC1 complex occupancy at target sites being 515 largely unaffected as measured by ChIP-seq (Blackledge et al., 2020; Tamburri et al., 2020). 516 This suggests that H2AK119ub1, as opposed to variant PRC1 complex occupancy, may play 517 a central and potentially direct role in gene repression. In agreement with this possibility, we 518 estimate at RING1B sites with the highest density of H2AK119ub1 that between 60-100% of 519 nucleosomes have at least one H2AK119ub1. We envisage that this level of histone 520 modification, even if dynamically turned over, would be compatible with persistent 521 transcriptional repression. Therefore, we propose that H2AK119ub1, as opposed to PRC1 522 occupancy, is primarily responsible for PRC1-dependent gene repression in ESCs. In future 523 work, it will be important to measure the number and kinetic parameters of canonical and 524 variant PRC1 complexes during differentiation and in more committed cell types. This will allow 525 us to understand whether the general principles we observe in ESCs are recapitulated, or if 526 canonical PRC1 complexes play a more direct and active role in gene repression as has been bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

527 reported previously (Cohen et al., 2018; Dietrich et al., 2007; Isono et al., 2013; Lau et al., 528 2017; Tsuboi et al., 2018).

529 Our single particle tracking experiments revealed that PRC1 complexes are for the most part 530 highly dynamic. In the case of variant PRC1 complexes, this behaviour may be important for 531 catalysis and the differences in the dynamics of individual variant PRC1 complexes may be 532 required to achieve the appropriate distribution of H2AK119ub1 in the genome. For example, 533 we previously demonstrated that H2AK119ub1 is present across the genome at low levels and 534 that this is primarily dependent on PCGF3/5-PRC1 complexes (Fursova et al., 2019; Lee et 535 al., 2015). PCGF3/5 proteins are not significantly enriched at Polycomb target genes (Scelfo 536 et al., 2019) and SPT revealed that the PCGF3-PRC1 complex had the smallest bound fraction 537 and shortest binding times of all PRC1 complexes. Therefore, we propose that PCGF3/5- 538 PRC1 complexes may utilise a highly dynamic hit-and-run mechanism to deposit low levels of 539 H2AK119ub1 indiscriminately throughout the genome. In contrast, PCGF1- and PCGF6- 540 PRC1 complexes show more specific enrichment at gene promoters as measured by ChIP- 541 seq (Fursova et al., 2019; Gao et al., 2012; Scelfo et al., 2019). This appears to be reflected 542 in our SPT measurements where they display a higher bound fraction or more stable binding 543 to chromatin than PCGF3-PRC1. This likely corresponds to the fact that PCGF1- and PCGF6- 544 PRC1 complexes incorporate auxiliary subunits that have DNA or chromatin binding activities, 545 which cause generic targeting to gene promoters. In agreement with this, removal of the C- 546 terminal RAWUL domain of PCGF1, which interacts with chromatin binding subunits BCOR 547 and KDM2B (Farcas et al., 2012; Gearhart et al., 2006; He et al., 2013; Wang et al., 2018; Wu 548 et al., 2013), disrupted stable chromatin binding by PCGF1. Therefore, in contrast to 549 PCGF3/5-PRC1 complexes, we propose that PCGF1 and 6-PRC1 complexes more frequently 550 occupy gene promoters via their site-specific chromatin binding activities, which provides an 551 opportunity to enrich H2AK119ub1 at these sites. Interestingly, however, most sites occupied 552 by PCGF1 and 6-PRC1 complexes do not display elevated H2AK119ub1 when analysed by 553 ChIP-seq (Fursova et al., 2019; Scelfo et al., 2019). We have previously proposed that 554 transcription may block deposition of H2AK119ub1 (Blackledge et al., 2015; Klose et al., 555 2013), meaning that despite PCGF1 and PCGF6-PRC1 complexes binding to and sampling 556 the majority of gene promoters, H2AK119ub1 would only accumulate at sites lacking 557 transcription in order to maintain, as opposed to initiate, a transcriptionally repressed state.

558 In summary, our quantitative single molecule measurements of PRC1 dynamics and 559 chromatin binding in live cells, coupled with estimation of genome-wide target site occupancy, 560 provide important new insight into how PRC1 interacts with chromatin. Furthermore, it 561 provides new evidence to suggest that H2AK119ub1, as opposed to PRC1 occupancy and 562 non-catalytic functions, is the primary determinant of PRC1-mediated repression in ESCs. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

563 Materials and Methods

Key Resources Table

Reagent type Source or (species) or Designation Identifiers Additional information reference resource

Rabbit monoclonal Cell Signaling Cat# Antibody (1:2000) anti-H2AK119ub1 Technology D27C4

Rabbit monoclonal Cell Signaling Cat# Antibody (1:2000) anti-H3K27me3 Technology C36B11

Rabbit monoclonal Cell Signaling Cat# Antibody (1:2000) anti-RING1B (WB) Technology D22F2

Mouse monoclonal (Atsuta et al., Antibody 3 µg anti-RING1B (IP) 2001)

Rabbit monoclonal Cell Signaling Cat# Antibody (1:1000) anti-SUZ12 Technology D39F6

Mouse monoclonal Cell Signaling Cat# Antibody (1:1000) anti-H3 Technology 96C10

Rabbit polyclonal LifeSpan Cat# LS- Antibody (1:1000) anti-PCGF6 BioSciences C482495

Rabbit polyclonal Cat# sc- Antibody anti-PCGF2 (Mel- Santa Cruz (1:1000) 10744 18)

In house Rabbit polyclonal Antibody (Blackledge et (1:1000) anti-PCGF1 al., 2014)

Rabbit monoclonal Cat# Antibody anti- Abcam (1:1000) ab201510 PCGF3+PCGF5

Rabbit polyclonal Cat# 07- Antibody Millipore (1:1000) anti-CBX7 981

Rabbit monoclonal Cat# Antibody Abcam (1:3000) anti-BRG1 ab110641

Mouse monoclonal Antibody Abcam Cat# ab818 (1:3000) anti-TBP

Mouse monoclonal Antibody Sigma Cat# F1804 (1:1000) anti-FLAG

Chemical compound, Proteinase K Sigma Cat# P4850 drug

Chemical (Z)-4- Cat# compound, Sigma Hydroxytamoxifen H7904 drug bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

Chemical (Nabet et al., compound, dTAG-13 2018) drug

Chemical Cat# compound, Halo-TMR Promega G8251 drug

Chemical (Grimm et al., compound, Halo-JF549 2017) drug

Chemical (Lavis et al., compound, PA-Halo-JF549 2016) drug

Chemical HaloTag Standard Cat# compound, Promega Protein G4491 drug

Chemical Lipofectamine Thermo Fisher Cat# compound, 3000 Scientific L3000015 drug

Cell line (Mus Mouse ESC: (Blackledge et

musculus) PRC1CKO al., 2020)

See Materials and Methods, Cell line (Mus Mouse ESC: Endogenous tagging of genes This paper musculus) RING1B-HaloTag via CRISPR/Cas9-mediated HDR

See Materials and Methods, Mouse ESC: Stable expression of HaloTag Cell line (Mus PRC1CKO This paper fusion proteins through musculus) TIGRERING1B-HaloTag CRISPR/Cas9-mediated editing of the TIGRE locus

Mouse ESC: See Materials and Methods, Stable expression of HaloTag Cell line (Mus PRC1CKO This paper fusion proteins through musculus) TIGRERING1B-NBM- HaloTag CRISPR/Cas9-mediated editing of the TIGRE locus

See Materials and Methods, Stable expression of HaloTag Cell line (Mus Mouse ESC: This paper fusion proteins through musculus) TIGRERPCD-HaloTag CRISPR/Cas9-mediated editing of the TIGRE locus

See Materials and Methods, Cell line (Mus Mouse ESC: Endogenous tagging of genes This paper musculus) HaloTag-PCGF1 via CRISPR/Cas9-mediated HDR

See Materials and Methods, Cell line (Mus Mouse ESC: Endogenous tagging of genes This paper musculus) PCGF2-HaloTag via CRISPR/Cas9-mediated HDR

See Materials and Methods, Cell line (Mus Mouse ESC: Endogenous tagging of genes This paper musculus) HaloTag-PCGF3 via CRISPR/Cas9-mediated HDR bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

See Materials and Methods, Cell line (Mus Mouse ESC: Endogenous tagging of genes This paper musculus) HaloTag-PCGF6 via CRISPR/Cas9-mediated HDR

See Materials and Methods, Cell line (Mus Mouse ESC: Endogenous tagging of genes This paper musculus) CBX7-HaloTag via CRISPR/Cas9-mediated HDR

See Materials and Methods, Cell line (Mus Mouse ESC: Endogenous tagging of genes This paper musculus) HaloTag-RYBP via CRISPR/Cas9-mediated HDR

Mouse ESC: See Materials and Methods, Cell line (Mus Endogenous tagging of genes RING1B-HaloTag This paper musculus) via CRISPR/Cas9-mediated dTAG-SUZ12 HDR

Mouse ESC: See Materials and Methods, Cell line (Mus Endogenous tagging of genes PCGF2-HaloTag This paper musculus) via CRISPR/Cas9-mediated dTAG-SUZ12 HDR

Mouse ESC: See Materials and Methods, Cell line (Mus Endogenous tagging of genes HaloTag-PCGF6 This paper musculus) via CRISPR/Cas9-mediated dTAG-SUZ12 HDR

See Materials and Methods, Stable expression of HaloTag Cell line (Mus Mouse ESC: This paper fusion proteins through musculus) TIGREHaloTag-PCGF1 CRISPR/Cas9-mediated editing of the TIGRE locus

See Materials and Methods, Mouse ESC: Stable expression of HaloTag Cell line (Mus TIGREHaloTag-PCGF1 This paper fusion proteins through musculus) 1-139 CRISPR/Cas9-mediated editing of the TIGRE locus

See Materials and Methods, Mouse ESC: Stable expression of HaloTag Cell line (Mus TIGREHaloTag-PCGF1 This paper fusion proteins through musculus) 140-259 CRISPR/Cas9-mediated editing of the TIGRE locus

Sequence- CRISPR/Cas9 This paper GCTCATTTGTGCTCCTTGGT based reagent sgRNA for Ring1b

Sequence- CRISPR/Cas9 This paper GCCTCATCGCGATCGCAATC based reagent sgRNA for Pcgf1

Sequence- CRISPR/Cas9 This paper CCCTTTCCTCAAGGGGGGCA based reagent sgRNA for Pcgf2

Sequence- CRISPR/Cas9 This paper ATCTTCCTGGTCAACATCTT based reagent sgRNA for Pcgf3

Sequence- CRISPR/Cas9 This paper TCTGATCCCCGCCATGGACG based reagent sgRNA for Pcgf6

Sequence- CRISPR/Cas9 This paper CTATTCACAGCTTCTCGTTG based reagent sgRNA for Cbx7 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

Sequence- CRISPR/Cas9 This paper GCCCATGGTCATGGACGGGC based reagent sgRNA for Rybp

Sequence- CRISPR/Cas9 This paper GTGCTTCTGAGGCGCCATCG based reagent sgRNA for Suz12

CRISPR/Cas9 Sequence- sgRNA for TIGRE This paper ACTGCCATAACACCTAACTT based reagent locus

Software, (Ramírez et al., deepTools (v3.0.1) algorithm 2014)

Software, (Zhang et al., MACS2 (v2.1.1) algorithm 2008)

Software, UCSC Genome (Kent et al.,

algorithm Browser 2002)

Software, (Ollion et al., TANGO algorithm 2013)

Software, cellSens Olympus algorithm

Software, (Uphoff et al., Stormtracker algorithm 2013)

Software, MATLAB MathWorks algorithm

Software, (Schindelin et Fiji algorithm al., 2012)

(Blackledge et Sample numbers: GSM3891386, Other RING1B ChIP-seq GSE132752 al., 2020) GSM3891387, GSM3891388

(Blackledge et Sample numbers: GSM3891362, Other PCGF1 ChIP-seq GSE132752 al., 2020) GSM3891363, GSM3891364

(Blackledge et Sample numbers: GSM3891368, Other PCGF2 ChIP-seq GSE132752 al., 2020) GSM3891369, GSM3891370

(Blackledge et Sample numbers: GSM3891374, Other PCGF6 ChIP-seq GSE132752 al., 2020) GSM3891375, GSM3891376

H2AK119ub1 ChIP- (Blackledge et Sample numbers: GSM3891325, Other GSE132752 seq al., 2020) GSM3891326, GSM3891327 564 565 Cell culture

566 All mouse ESC lines used in this study were grown at 37 °C and 5% CO2 on gelatinised plates. 567 Dulbecco’s Modified Eagle Medium (Gibco) was supplemented with 15% fetal bovine serum 568 (Labtech), 2 mM L-glutamine (Life Technologies), 1x penicillin-streptomycin solution (Life 569 Technologies), 1x non-essential amino acids (Life Technologies), 0.5 mM beta- 570 mercaptoethanol (Life Technologies), and 10 ng/mL leukemia inhibitory factor. Cells were bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

571 passaged using trypsin-EDTA (0.25%, Gibco) with 2% chicken serum. Cells were regularly 572 tested for the presence of mycoplasma.

573 To induce conditional knockout, PRC1CKO cells were treated with 800 nM 4-hydroxytamoxifen 574 (OHT) for 72 hours. To induce and maintain degradation and depletion of SUZ12, dTAG- 575 SUZ12 were treated with 100 nM dTAG-13 for 96 hours.

576 Alkaline phosphatase staining

577 The PRC1CKO line, and the same line expressing either RING1B-HaloTag or RING1BNBM- 578 HaloTag were plated and grown in the presence or absence of OHT for 72 hours. 10,000 cells 579 per condition were plated out on a 10 cm plate after counting using a Countess II cell counter 580 (Invitrogen). Cells were grown for 4 days and then fixed with 2% paraformaldehyde. Fixed 581 cells were stained using AP staining solution (100 mM Tris-HCl pH 9.0, 100 mM NaCl, 5 mM

582 MgCl2, 0.4 mg/mL napthol phosphate N-5000, 1 mg/mL Fast Violet B salt) for 10 min, rinsed 583 with PBS and distilled water, and then air-dried. Three biological replicates were carried out 584 and at least 100 colonies were counted for each condition in each replicate.

585 Stable cell line generation by random integration

586 To generate cell lines expressing H2B-HaloTag and HaloTag-3xNLS, the corresponding 587 cDNAs were cloned into an expression plasmid driven by the CAG promoter that also 588 contained an internal ribosome entry site that allowed co-expression of the G418 resistance 589 gene. In order to randomly integrate this expression cassette, ESCs in one well of a 6-well 590 plate were transfected with 10 μg of expression vector using Lipofectamine 3000 591 (ThermoFisher) according to the manufacturer’s instructions. The following morning, 592 transfected cells were passaged onto new plates at a range of cell densities. 5 hours later, 593 400 μg/ml of G418 was applied. Cells were grown in the presence of G418 to select for stable 594 integration events and approximately one week later resistant colonies were picked and 595 expanded on 96-well plates. 96 well plates of putative positive clones were labelled with 500 596 nM Halo-TMR dye (Promega) for 15 minutes at 37 °C, washed in label free medium three 597 times, and then incubated in medium without dye for 30 minutes at 37 °C. Individual clones 598 with stable HaloTag fusion protein expression were identified by fluorescence microscopy.

599 Endogenous tagging of genes via CRISPR/Cas9-mediated HDR

600 In order to introduce tags into the endogenous copies of genes using CRISPR-mediated 601 genome editing and HDR, sgRNAs were designed using the CRISPOR online tool (Haeussler 602 et al., 2016) and cloned into pSptCas9(BB)-2A-Puro(PX459)-V2.0 (Addgene #62988) as 603 previously described (Ran et al., 2013). Targeting constructs for homology directed repair bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

604 (HDR) were engineered to lack the guide sequence and were assembled from PCR products 605 by Gibson assembly (Gibson Assembly Master Mix kit, New England Biolabs).

606 To generate endogenous HaloTag or dTAG fusion cell lines, targeting constructs were 607 generated to modify the endogenous locus such that it encoded the HaloTag or dTAG with a 608 flexible polypeptide fused to either the N- or C-terminus of the gene of interest.

609 Targeting was carried out by transfecting 3 μg of each Cas9 guide and 10 g of targeting 610 construct into ESCs in one well of a 6-well plate using Lipofectamine 3000 (ThermoFisher) 611 according to the manufacturer’s instructions. The following morning, transfected cells were 612 passaged onto new plates at a range of cell densities. 5 hours later puromycin was applied to 613 cells at 1 μg/ml to select for transfected cells.

614 After 48 hours of selection puromycin was removed and approximately one week later 615 individual clones were picked and expanded onto 96 well plates. Alternatively, cells were 616 labelled with 500 nM Halo-TMR dye (Promega) as described above, and FACS sorted to 617 enrich for cells expressing a Halo-Tag. FACS sorted cells were then plated at low density and 618 approximately 1 week later individual clones were picked and expanded onto 96 well plates.

619 Appropriately edited clones were then identified by PCR screening from genomic DNA. In 620 most cases, a PCR screen with a primer inside the tag sequence and a primer outside of the 621 targeting construct was used to identify clones that had undergone HDR. Then a primer pair 622 flanking the targeting construct was used amplify genomic DNA in clones that showed 623 evidence of HDR to test whether they were heterozygote or homozygote. When compared to 624 parental lines, lines that where HDR had occurred in both alleles produced a product that was 625 longer by exactly the size of the inserted sequence and showed no evidence of a wild type 626 PCR product. The integrity of homozygote clones was then further validated by sequencing 627 the PCR products to ensure the expected HDR event had occurred, western blotting, and 628 fluorescence microscopy for HaloTag cell lines.

629 Stable expression of HaloTag fusion proteins through CRISPR/Cas9-mediated editing of the 630 TIGRE locus

631 To generate cell lines expressing RING1B, RING1B nucleosome binding mutant (NBM, R81A, 632 K93A, K97A, K98A), RPCD, PCGF1, PGF1 1-139, and PCGF1 140-259 HaloTag fusion 633 proteins, targeting constructs were engineered to insert an expression cassette into the TIGRE 634 locus (Zeng et al., 2008). An sgRNA for the TIGRE locus was cloned into pSptCas9(BB)-2A- 635 Puro(PX459)-V2.0 (Addgene #62988) as previously described (Ran et al., 2013).

636 Targeting constructs contained the cDNA of interest cloned next to a CAG promoter and fused 637 to the HaloTag sequence with a FLAG 2xStrep-tag II tag to enable immunoblotting. A triple bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

638 SV40 NLS sequence was also included in the RPCD and all PCGF expression constructs to 639 ensure they localised to the nucleus.

640 Targeting was carried out by transfecting 3 μg of each Cas9 guide and 10 g of targeting 641 construct into ESCs in one well of a 6-well plate using Lipofectamine 3000 (ThermoFisher) 642 according to the manufacturer’s instructions. The following morning, transfected cells were 643 passaged onto new plates at a range of cell densities. After 5 h, selection was applied using 644 1 μg/ml puromycin. Selection was maintained for 48 hours, and then surviving clones grown 645 out.

646 Halo-Tag expressing cells were enriched by FACS as described above and colonies were 647 transferred to 96-well plates. Clones that had acquired the expression cassette via HDR at the 648 TIRGRE locus were identified by a PCR screen with a primer inside the expression cassette 649 and outside of the targeting construct. Then a primer pair flanking the targeting construct was 650 used amplify genomic DNA in clones that showed evidence of HDR to test whether they were 651 heterozygotes or homozygotes as described above. To limit the level of exogenous expression 652 heterozygous clones were selected and further validated by sequencing the PCR products to 653 ensure the expect HDR event had occurred, western blotting, and fluorescence microscopy.

654 Nuclear and histone extracts

655 Cell pellets were resuspended in 10 volumes of buffer A (10 mM Hepes pH 7.9, 1.5 mM MgCl2, 656 10 mM KCl, 0.5 mM DTT, 0.5 mM PMSF, and cOmplete protease inhibitor (PIC, Roche)) and 657 incubated for 10 min on ice. After centrifugation at 1500 g for 5 min, the cell pellet was 658 resuspended in 3 volumes of buffer A supplemented with 0.1% NP-40. Following inversion, 659 nuclei were pelleted again at 1500 g for 5 min. Pelleted nuclei were resuspended in 1 volume 660 of buffer C (250 mM NaCl, 5 mM Hepes pH 7.9, 26% glycerol, 1.5 mM MgCl2, 0.2 mM EDTA, 661 0.5 mM DTT and 1x PIC). The volume of the nuclear suspension was measured and NaCl 662 concentration increased to 400 mM by dropwise addition. Nuclei were incubated at 4 °C for 1 663 h to extract nuclear proteins, which were recovered as the supernatant after centrifugation at 664 18,000 g for 20 min. Protein concentration was determined by Bradford assay (BioRad).

665 Histone extracts were prepared by washing pelleted cells in RSB (10 mM Tris HCl pH 7.4, 10 666 mM NaCl, 3 mM MgCl2 and 20 mM NEM), followed by centrifugation at 240 g for 5 min and 667 resuspension in RSB buffer supplemented with 0.5% NP-40. Following incubation on ice for 668 10 min, cells were centrifuged at 500 g for 5 minutes. The nuclear pellet was resuspended in 669 5 mM MgCl2, an equal volume of 0.8 M HCl was added, and incubated on ice for 20 min to 670 extract histones. The supernatant was taken after centrifugation for 20 min at 18,000 g, and 671 histones precipitated by addition of TCA up to 25% by volume and incubation on ice for 30 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

672 min. Histones were pelleted by centrifugation at 18,000 g for 15 min, and the pellet washed 673 with cold acetone twice. The histone pellet was resuspended by vortexing in 1x SDS loading 674 buffer (2% SDS, 100 mM Tris pH 6.8, 100 mM DTT, 10% glycerol and 0.1% bromophenol 675 blue) and boiling at 95 ◦C for 5 min. For histone extractions where H2AK119ub1 was to be 676 analysed, all buffers were supplemented with 20 mM NEM.

677 Immunoprecipitation

678 Immunoprecipitations were performed using 500 µg of nuclear extract. Extracts were diluted 679 to 550 µL with BC150 (150 mM KCl, 10% glycerol, 50 mM HEPES (pH 7.9), 0.5 mM EDTA, 680 0.5 mM DTT, 1x PIC) and incubated with antibody against the protein of interest overnight at 681 4 °C. Protein A beads (Repligen) were used to capture antibody-bound protein at 4 °C for 1 h. 682 Beads were pelleted at 1000xg, and washed 3 times with BC300 (300 mM KCl, 10% glycerol, 683 50 mM HEPES (pH 7.9), 0.5 mM EDTA, 0.5 mM DTT). For western blotting, beads were 684 resuspended in 2x SDS loading buffer and boiled at 95 ◦C for 5 min, and the supernatant taken 685 as the immunoprecipitate. 1x SDS loading buffer was added to input samples which were also 686 incubated at 95°C for 5 mins, prior to SDS-PAGE and western blot analysis.

687 Immunoblotting

688 Protein extracts were mixed with SDS loading buffer and boiled at 95 ◦C for 5 min. Proteins 689 were resolved by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). Gels (0.1% SDS, 690 0.1% ammonium persulphate (Sigma), 0.1% TEMED (Sigma), 400 mM Tris HCl pH 8.8) were 691 cast using the Mini-Protean Tetra Cell system (BioRad). Depending on the size of the protein 692 of interest, resolving gels between 6% and 15% acrylamide/bis-acrylamide (BioRad) were 693 used. The stacking gel contained 5% acrylamide/bis-acrylamide and 125 mM Tris HCl pH 6.8. 694 Gels were run at 200 V in 1x SDS-PAGE running buffer (25mM Tris, 192mM glycine and 0.1% 695 SDS). Proteins were transferred to nitrocellulose membrane by semi-dry transfer using the 696 Trans-Blot Turbo Transfer System (BioRad, 1.5 A for 7 min). Membranes were blocked in 5% 697 milk in 1x PBS with 0.1% Tween 20 (PBST, Fisher) for 30 min at room temperate. Primary 698 antibody incubations were carried out overnight at 4 ◦C in the same buffer. Membranes were 699 washed for 3x 5 min with PBST, and then incubated with secondary antibody for 1 h in 5% 700 milk in PBST. After 3x 5 min PBST washes and one PBS rinse, membranes were imaged 701 using an Odyssey Fc system (LI-COR).

702 Fluorescence-based protein quantification

703 We quantified the number of proteins in cells as described previously (Cattoglio et al., 2019). 704 Briefly, cells were trypsinised, pelleted, and counted with a Countess II cell counter. 5 x 106 705 cells were labelled with 500 nM HaloTMR for 15 min at 37 °C which causes near-quantitative bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

706 labelling of HaloTag proteins in live cells (Yoon et al., 2016). Cells were pelleted, counted 707 again, and 3 x 106 labelled cells were reserved as a pellet. HaloTag Standard Protein 708 (Promega) was labelled with 10 fold molar excess (5 µM) Halo-TMR dye at 4 °C for 15 min. 709 Pelleted cells were lysed by boiling in 1x SDS loading buffer at 95 °C for 10 min and volumes 710 equivalent to 1 x 105 and 2.5 x 105 cells were resolved by SDS-PAGE alongside known 711 quantities of HaloTag Standard Protein. Gels were imaged using a Typhoon FLA 7000 712 (Fujifilm) using the 532nm laser. Band intensities were quantified using Image Studio software 713 (LI-COR). Calculated numbers of molecules and estimated nuclear concentrations are given 714 in Table S1.

715 Single particle tracking

716 Cells for single particle tracking were plated onto gelatinised and glycine-coated 35 mm petri 717 dishes containing a 14 mm No. 1.5 coverglass (MatTek, #P35G-1.5-14-C) at least 5 hours 718 before imaging. Cells were labelled using 100 nM PA-Halo-JF549 (Grimm et al., 2017) for 15 719 min at 37 °C. They were then washed 3 times with Fluorobrite DMEM (Thermo Fisher 720 Scientific) supplemented as described for general ESC culture above. Cells were incubated 721 for a further 30 min in supplemented Fluorobrite DMEM at 37 °C and washed once more 722 before imaging.

723 Single molecule tracking experiments were carried out using a custom TIRF/HILO microscope 724 as described in (Wegel et al., 2016). The angle of the excitation beam is controlled by 725 translation of the position of the focus in the objective (Olympus 100x NA1.4), and was 726 positioned in each experiment to maximise the signal-to-noise ratio. Sample temperature was 727 maintained at 37 °C using an objective collar heater and heated stage, while the pH of sample 728 medium was maintained by addition of HEPES pH 7.4 to a concentration of 30 mM. For rapid 729 tracking of diffusing and bound molecules, movies were acquired with continuous 561 nm and 730 405 nm excitation at 22 mW and 5 mW intensity at the fibre output, respectively, with a 15 ms 731 exposure time. Movies were acquired for 4000 frames and contained multiple nuclei. 732 Experiments were performed in at least three biological replicates of at least 10 movies each, 733 with each movie containing multiple cells.

734 To measure binding times of single molecules, movies were acquired with a frame rate of 2 735 Hz and exposure time of 0.5 s for 600 frames or at 0.03 Hz with an exposure time of 1 s for 736 30 frames. Acquisition was started after activating sufficient numbers of fluorophores using 737 the 405nm laser and carried out with 561nm excitation at 0.1mW. Experiments were 738 performed in at least two biological replicates of eight movies each, with each movie containing 739 multiple cells.

740 Confocal imaging bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

741 Polycomb body imaging, FRAP experiments, and relative fluorescence intensity 742 measurements were carried out with a Spinning Disk Confocal microscope. Cells were plated 743 on gelatinised 35 mm petri dishes containing a 14 mm No. 1.5 coverglass (MatTek, #P35G- 744 1.5-14-C) at least 5 hours before imaging. Prior to imaging, cells were labelled with 500 nm 745 Halo-JF549 for 15 min at 37 °C, followed by 3 washes, changing medium to Fluorobrite DMEM 746 (Thermo Fisher Scientific) supplemented as described for general ESC culture above. Cells 747 were then incubated for a further 30 min at 37 °C in supplemented Fluorobrite DMEM to 748 remove excess dye. For Polycomb body imaging, 10 µg ml−1 Hoechst 33258 (Thermo Fisher 749 Scientific) was also added to the medium during this incubation. Cells were washed once more 750 with Fluorobrite DMEM before imaging. Confocal microscopy was performed on an IX81 751 Olympus microscope connected to a spinning disk confocal system (UltraView VoX 752 PerkinElmer) using an EMCCD camera (ImagEM, Hamamatsu Photonics) in a 37 °C heated, 753 humidified, CO2-controlled chamber.

754 Polycomb body imaging

755 Z-stacks for imaging Polycomb bodies were acquired on the above described spinning disk 756 confocal system using an Olympus PlanApo 100x/1.4 N.A. oil-immersion objective heated to 757 37 °C, and Volocity software (PerkinElmer). Z-stacks for Polycomb body quantification were 758 acquired by imaging Halo-JF549 with a 568 nm laser at 1.25 s exposure and 15% laser power, 759 while Hoechst was imaged with a 405 nm laser at 250 ms exposure and 20% laser power. Z- 760 stacks were acquired at 150 nm intervals. Experiments were performed in at least two 761 biological replicates of at least 15 cells each.

762 FRAP imaging

763 FRAP experiments were performed on the above described spinning disk confocal system 764 using an Olympus PlanApo 60x/1.4 N.A. oil-immersion objective heated to 37 °C, and Volocity 765 software (PerkinElmer). Spot FRAP was performed by acquiring 10 prebleach frames at 1 s 766 intervals, and then bleaching a circle of 1.4 µm in diameter in either the nucleoplasm or a 767 Polycomb body using the 568nm laser at 100% power. Post-bleaching images were acquired 768 at 0.5 s or 1 s intervals for 10 min. Images were acquired using the 568 nm laser at 400 ms 769 exposure and 15% laser power. Experiments were performed in two biological replicates of at 770 least 10 cells each for Polycomb body and non-body regions.

771 Relative fluorescence intensity measurements

772 Images to quantify relative fluorescence intensities within nuclei were acquired using the 773 above described spinning disk confocal system with an Olympus PlanApo 60x/1.4 N.A. oil- 774 immersion objective heated to 37 °C, and Volocity software (PerkinElmer). Single Z-slices of bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

775 nuclei were acquired using the 568 nm laser at 400 ms exposure and 15% laser power. Nuclei 776 were segmented manually, and mean fluorescence calculated and normalised to that of 777 RING1B. Calculated numbers of molecules are given in Table S1.

778 Tracking and localisation of SPT data

779 Analysis of SPT data was performed in MATLAB (MathWorks) using previously described 780 Stormtracker software (Uphoff et al., 2013). Fluorescent molecules were detected in each 781 frame based on an intensity threshold and localisations of molecules determined to 25 nm 782 precision by fitting an elliptical Gaussian point spread function. Localisations in consecutive 783 frames of movies within a radius of 7.7 µm were linked to form tracks. Localisations were 784 permitted to be lost for single frames within tracks as a result of fluorophore blinking or loss of 785 focus through a memory parameter.

786 Calculation of apparent diffusion coefficients

787 Tracks with at least 5 localisations were used to determine apparent diffusion coefficients (D*) 788 based on the mean-squared displacement (MSD) of the total track. D* was calculated by

MSD 휎2 789 D* = − (4푑푡) 푑푡

790 where dt is the time between frames, with correction for localisation uncertainty σ.

791 Spot-On analysis of tracks

792 Analysis of tracks to determine the fraction of molecules in bound and diffusing states was 793 performed using the online Spot-On tool (Hansen et al., 2018). Jump length distributions were 794 generated using a bin width of 0.01 µm with 8 time points and 4 jumps permitting a maximum 795 jump length of 5.05 µm. Jump length distributions were fitted with a three-state kinetic model 796 with a localisation error of 40 nm, using Z correction with dZ = 0.7 and cumulative density 797 function fitting with three iterations. Tracks from individual movies, each containing multiple 798 cells, were fitted separately. All statistical analyses were performed on individual movie fits. 799 Bound fractions and estimated numbers of bound proteins for PRC1 subunits are given in 800 Table S1.

801 Single molecule binding time analysis

802 Tracking movies for measuring single molecule binding times were localised and tracked as 803 described above, using a tracking radius of 1.92 µm for 2 Hz imaging or 6.72 µm for 0.03 Hz 804 imaging. These limits were determined based on the measured displacements of stably bound 805 H2B-HaloTag molecules due to chromatin diffusion. Track lengths of stationary molecules 806 were used to determine apparent dwell times of chromatin bound molecules. Mean survival bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

807 times were determined by fitting the observed distribution of apparent dwell times with a 808 double exponential function. In order to estimate binding times from these measurements, 809 mean survival times were corrected for events that result in loss of tracks from reasons other 810 than unbinding, such as photobleaching of dye molecules or diffusion of chromatin out of focus 811 (Hansen et al., 2017; Uphoff et al., 2013). This was done by measuring the survival time for 812 H2B-HaloTag, which is stably bound for hours and will therefore only rarely be observed to 813 unbind, with all other tracks being limited by photobleaching or chromatin diffusion. The 814 survival time of H2B-HaloTag, tBleach (typically approximately 50 s), was measured for each 815 experiment to account for changes in imaging conditions. Binding times were calculated from 816 fitted dwell time constants as follows:

tDwell × tBleach 817 tBound = tBleach − tDwell

818 When tDwell exceeded tBleach, or was close enough to tBleach such that tBound was more 819 than double tBleach, binding times were not estimated and are indicated as likely exceeding 820 100 s. Estimated stable binding times and fractions of observed binding events which were 821 stable for PRC1 subunits are given in Table S1.

822 FRAP analysis

823 To measure fluorescence recovery, intensity measurements were made with Fiji and ImageJ 824 (Schindelin et al., 2012). Fluorescence intensity was measured in the bleached region and a 825 corresponding unbleached region within the same cell. From each, the intensity of a 826 background region outside of cells was subtracted. The relative intensity of bleached and 827 unbleached regions was calculated by dividing background-corrected unbleached intensity by 828 background-corrected bleached intensity. Prebleach measurements were used to normalise 829 all values so that the mean of the relative prebleach intensity was 1. The mean, normalised 830 relative intensity of all repeats was plotted with error bars indicating standard error of the 831 mean.

832 Polycomb body analysis

833 To segment Polycomb bodies in individual live cells for analysis, nuclei were first manually 834 segmented in 3D based on Hoechst fluorescence in the 405 nm channel using TANGO in 835 ImageJ (Ollion et al., 2013). The 561 nm channels of the same Z-stacks were deconvolved 836 using Olympus cellSens software (constrained iterative deconvolution, 5 cycles). Deconvolved 837 561 nm z-stacks were masked using outputs from TANGO, and individual Polycomb bodies 838 identified using a custom script. Briefly, segmented nuclei were background subtracted using 839 a 4 px rolling ball and a mask of Polycomb bodies generated using Otsu thresholding. The 3D bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

840 Objects Counter plugin in ImageJ was used to quantify the properties of the masked Polycomb 841 bodies, and its outputs (object volume, number, mean and total intensity) were processed and 842 analysed using a custom R script to compile all data and calculate any values derived from 843 the 3D Objects Counter measurements (fluorescence enrichment in foci, fractions of total 844 nuclear volume and fluorescence in foci). The mean background intensity of each image stack, 845 measured from a region containing no cells, was subtracted from all measured fluorescence 846 intensities from the same stack. Apparent foci with volumes larger than 1.5 µm3 were excluded 847 from measurements, as these likely resulted from segmentation identifying regions of the 848 nucleus with fluctuations in fluorescence intensity outside of Polycomb bodies. Quartiles of 849 foci by volume or mean fluorescence intensity were assigned based on the distribution of foci 850 in untreated cells, and quartile boundaries were used to classify foci in treated cells. 851 Enrichment of signal within Polycomb bodies was calculated for each cell by comparison of 852 mean fluorescence densities in all foci to mean fluorescence density for the rest of the nucleus. 853 Estimates of numbers and concentrations of molecules in foci were made by determining the 854 fluorescence intensity equivalent to a single molecule based on total nuclear fluorescence 855 signal and measured numbers of proteins per cell (see above), and then taking into account 856 the total fluorescence signal and volume of each Polycomb body.

857 Estimating the number of H2AK119ub1 molecules and PRC1 catalytic activity

858 To estimate the number of H2AK119ub1 molecules and the rate of catalysis by PCGF3/5- 859 PRC1 we first estimated the number of histone H2A molecules in nucleosomes in a 2n mouse 860 genome. To achieve this we used previously published ((Teif et al., 2012) genome-wide 861 measurements of the mean linear distance between nucleosomes (186.1 bp) in mouse ESCs 862 and the size of the 2n mouse genome (5,461,710,950 base pairs) to estimate the total number 863 of nucleosomes as follows:

2n genome size 864 Total 2n nucleosomes = Nucleosome spacing

865 Based on this calculation we estimate that there are 2.9 x 107 nucleosomes. In each 866 nucleosome there are 2 histone H2A molecules, meaning there are 5.9 x 107 histone H2A 867 molecules. In ESCs ~10% of H2A is monoubiquitylated on K119 ( (Fursova et al., 2019) so 868 we estimate there are 5.9 x 106 molecules of H2AK119ub1. In mammalian cells with a similar 869 fraction of H2AK119ub1, the decay half time of this modification was previously determined to 870 be 90 minutes (Seale, 1981). On this basis, we estimated a decay constant of 1.3 x 10-4 s-1 871 which would correspond to loss of 750 H2AK119ub1 modifications per second, and the same 872 rate of replacement by PRC1, assuming the system exists in a steady state. Based on 873 previous findings that PCGF3/5-PRC1 contribute approximately half of global H2AK119ub1 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

874 (Fursova et al., 2019), we calculated how frequently each PCGF3/5-PRC1 complex should 875 deposit H2AK119ub1 to maintain this rate, assuming approximately equal levels of PCGF3 876 and PCGF5 (total 6000 molecules):

877 Frequency of deposition by PCGF3/5 Deubiquitylation rate 1 878 = 1 ÷ ( × ) 2 Total PCGF3/5 molecules

879 On this basis we estimate that each PCGF3/5 complex should deposit a H2AK119ub1 880 modification every 16 s to maintain H2AK119ub1 levels.

881 Analysis of ChIP-seq data

882 Aligned and processed ChIP-seq data for RING1B, PCGF1, PCGF2, PCGF6 and 883 H2AK119ub1 in ESCs was obtained from (Blackledge et al., 2020) (GEO: GSE132752; 884 samples used are listed in the Key Resources table). As described previously (Blackledge et 885 al., 2020), RING1B peaks (18,643) correspond to regions of the genome that have a peak of 886 RING1B that is lost following conditional removal of RING1B as quantified using calibrated 887 ChIP-seq analysis. Raw ChIP-seq data were processed as described in Blackledge et al., 888 2020. To visualise ChIP-seq distributions, genome coverage tracks were generated using the 889 pileup function from MACS2 (Zhang et al., 2008) and viewed using the UCSC genome browser 890 (Kent et al., 2002). Read counts at RING1B peaks were extracted using multiBamSummary 891 from deeptools (“–outRawCounts”) (Ramírez et al., 2014).

892 Estimation of maximal protein occupancy

893 To estimate the maximal number of molecules that could occupy RING1B peaks, we used 894 total molecule number measurements from biochemical experiments, the fraction of bound 895 molecules from SPT experiments, and read counts in RING1B peaks from ChIP-seq 896 experiments. To estimate the number of chromatin bound molecules, the bound faction from 897 SPT experiments was multiplied by the total number of molecules. The number of bound 898 molecules at each peak was estimated as follows:

Total bound molecules 899 Bound molecules in peak = Reads in peak × Reads in all peaks

900 The density of molecules per kilobase was estimated by dividing the number of molecules in 901 a given peak by the length of the peak. Read counts and calculated densities for all peaks are 902 available in Table S2. We would also like to emphasise that our calculations will inevitably 903 result in an overestimation of the number molecules bound at RING1B peaks due to the two 904 central assumptions that were made in order to capture what we would consider to be the 905 maximal possible occupancy. The first assumption was that binding events would be restricted bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

906 to RING1B peaks. The limitation of this assumption is that PRC1 must also bind chromatin 907 outside of ChIP-seq peaks as H2AK119ub1 occurs, albeit at reduced levels, throughout the 908 genome (Fursova et al., 2019; Lee et al., 2015). The second assumption we made was that 909 each RING1B peak is present as two copies because ESCs are diploid (2n). However, our 910 ESC cultures are not synchronous, so some cells will have replicated their DNA but not yet 911 divided, meaning that a subset of cells will have DNA content greater than 2n. Despite these 912 assumptions, and the overestimations that will inevitably be inherent to them, we reasoned 913 that our maximal estimates would be useful in contextualising possible PRC1 functions at 914 target sites, particularly given the structural roles that PRC1 is proposed to have at these loci.

915 Estimation of H2AK119ub1 levels

916 In order to estimate the number of H2AK119ub1-modified nucleosomes at RING1B peaks we 917 multiplied the estimated total number of H2AK119ub1 molecules (see above) by the fraction 918 of the genome that can be uniquely mapped using our sequencing strategy (0.88, 2.4 x 109 919 bp) (Crusoe et al., 2015) and then multiplied this by the fraction of H2AK119ub1 ChIP-seq 920 reads that fall in RING1B peaks as follows:

Reads in peaks 921 H2AK119ub1 in peaks = Total H2AK119ub1 × Mappable genome fraction × Total reads

922 Based on this calculation we estimate that there are 3.3 x 105 H2AK119ub1 molecules in 923 RING1B peaks. We then estimated the number of H2AK119ub1 modifications at each peak 924 as follows:

Total H2AK119ub1 × Mappable genome fraction 925 H2AK119ub1 at peak = Reads at peak × Total reads

926 The density of H2AK119ub1 modifications per kilobase at peaks was calculated by dividing 927 the estimated number of H2AK119ub1 modifications by the peak length.

928

929 Acknowledgements

930 Work in the Klose lab is supported by the Wellcome Trust (109102/Z/15/Z to M.K.H, 931 209400/Z/17/Z to R.J.K.), the European Research Council (681440), and the Lister Institute 932 of Preventive Medicine. We thank Aleksander Szczurek, Mathew Stracy and Stephan Uphoff 933 for their contributions to the analysis of microscopy data. We gratefully acknowledge the 934 Micron Advanced Bioimaging Unit (supported by Wellcome Strategic Awards 091911/B/10/Z 935 and 107457/Z/15/Z) for their support and assistance with all practical microscopy in this bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

936 work. We thank Anders Hansen, Nadezda Fursova, Neil Blackledge and Paula Dobrinić for 937 critical reading of the manuscript.

938

939 Competing interests

940 The authors declare no competing interests.

941

942 References

943 Almeida M, Pintacuda G, Masui O, Koseki Y, Gdula M, Cerase A, Brown D, Mould A, Innocent C, 944 Nakayama M, Schermelleh L, Nesterova TB, Koseki H, Brockdorff N. 2017. PCGF3/5-PRC1 945 initiates Polycomb recruitment in X chromosome inactivation. Science 356:1081–1084. 946 doi:10.1126/science.aal2512 947 Atlasi Y, Stunnenberg HG. 2017. The interplay of epigenetic marks during stem cell differentiation 948 and development. Nat Rev Genet 18:643–658. doi:10.1038/nrg.2017.57 949 Atsuta T, Fujimura S, Moriya H, Vidal M, Akasaka T, Koseki H. 2001. Production of Monoclonal 950 Antibodies Against Mammalian Ring1B Proteins. Hybridoma 20:43–46. 951 doi:10.1089/027245701300060427 952 Banani SF, Lee HO, Hyman AA, Rosen MK. 2017. Biomolecular condensates: organizers of cellular 953 biochemistry. Nat Rev Mol Cell Biol 18:285–298. doi:10.1038/nrm.2017.7 954 Beck M, Schmidt A, Malmstroem J, Claassen M, Ori A, Szymborska A, Herzog F, Rinner O, Ellenberg J, 955 Aebersold R. 2011. The quantitative proteome of a human cell line. Mol Syst Biol 7:549. 956 doi:10.1038/msb.2011.82 957 Bentley ML, Corn JE, Dong KC, Phung Q, Cheung TK, Cochran AG. 2011. Recognition of UbcH5c and 958 the nucleosome by the Bmi1/Ring1b ubiquitin ligase complex. EMBO J 30:3285–3297. 959 doi:10.1038/emboj.2011.243 960 Bernstein E, Duncan EM, Masui O, Gil J, Heard E, Allis CD. 2006. Mouse polycomb proteins bind 961 differentially to methylated histone H3 and RNA and are enriched in facultative 962 heterochromatin. Mol Cell Biol 26:2560–9. doi:10.1128/MCB.26.7.2560-2569.2006 963 Blackledge NP, Farcas AM, Kondo T, King HW, McGouran JF, Hanssen LLP, Ito S, Cooper S, Kondo K, 964 Koseki Y, Ishikura T, Long HK, Sheahan TW, Brockdorff N, Kessler BM, Koseki H, Klose RJ. 965 2014. Variant PRC1 complex-dependent H2A ubiquitylation drives PRC2 recruitment and 966 polycomb domain formation. Cell 157:1445–1459. doi:10.1016/j.cell.2014.05.004 967 Blackledge NP, Fursova NA, Kelley JR, Huseyin MK, Feldmann A, Klose RJ. 2020. PRC1 Catalytic 968 Activity Is Central to Polycomb System Function. Mol Cell 77:857-874.e9. 969 doi:10.1016/j.molcel.2019.12.001 970 Blackledge NP, Rose NR, Klose RJ. 2015. Targeting Polycomb systems to regulate gene expression: 971 modifications to a complex story. Nat Rev Mol Cell Biol 16:643–649. doi:10.1038/nrm4067 972 Bonev B, Cohen NM, Szabo Q, Fritsch L, Papadopoulos GL, Lubling Y, Xu X, Lv X, Hugnot J-P, Tanay A, 973 Cavalli G. 2017. Multiscale 3D Genome Rewiring during Mouse Neural Development. Cell 974 171:557-572.e24. doi:10.1016/j.cell.2017.09.043 975 Bonnet J, Lindeboom RGH, Pokrovsky D, Stricker G, Çelik MH, Rupp RAW, Gagneur J, Vermeulen M, 976 Imhof A, Müller J. 2019. Quantification of Proteins and Histone Marks in Drosophila Embryos 977 Reveals Stoichiometric Relationships Impacting Chromatin Regulation. Dev Cell 51:632- 978 644.e6. doi:10.1016/j.devcel.2019.09.011 979 Boyer LA, Plath K, Zeitlinger J, Brambrink T, Medeiros LA, Lee TI, Levine SS, Wernig M, Tajonar A, Ray 980 MK, Bell GW, Otte AP, Vidal M, Gifford DK, Young RA, Jaenisch R. 2006. Polycomb complexes bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

981 repress developmental regulators in murine embryonic stem cells. Nature 441:349–353. 982 doi:10.1038/nature04733 983 Boyle S, Flyamer IM, Williamson I, Sengupta D, Bickmore WA, Illingworth RS. 2019. A Central Role for 984 Canonical PRC1 in Shaping the 3D Nuclear Landscape. doi:10.1101/2019.12.15.876771 985 Bracken AP, Dietrich N, Pasini D, Hansen KH, Helin K. 2006. Genome-wide mapping of Polycomb 986 target genes unravels their roles in cell fate transitions. Genes Dev 20:1123–1136. 987 doi:10.1101/gad.381706 988 Buchwald G, van der Stoop P, Weichenrieder O, Perrakis A, van Lohuizen M, Sixma TK. 2006. 989 Structure and E3-ligase activity of the Ring-Ring complex of Polycomb proteins Bmi1 and 990 Ring1b. EMBO J 25:2465–2474. doi:10.1038/sj.emboj.7601144 991 Cao R, Wang L, Wang H, Xia L, Erdjument-Bromage H, Tempst P, Jones RS, Zhang Y. 2002. Role of 992 Histone H3 Lysine 27 Methylation in Polycomb-Group Silencing. Science 298:1039–1043. 993 doi:10.1126/science.1084274 994 Cao R, Zhang Y. 2004. SUZ12 Is Required for Both the Histone Methyltransferase Activity and the 995 Silencing Function of the EED-EZH2 Complex. Mol Cell 15:57–67. 996 doi:10.1016/j.molcel.2004.06.020 997 Cattoglio C, Pustova I, Walther N, Ho JJ, Hantsche-Grininger M, Inouye CJ, Hossain MJ, Dailey GM, 998 Ellenberg J, Darzacq X, Tjian R, Hansen AS. 2019. Determining cellular CTCF and cohesin 999 abundances to constrain 3D genome models. eLife 8. doi:10.7554/eLife.40164 1000 Cohen I, Zhao D, Bar C, Valdes VJ, Dauber-Decker KL, Nguyen MB, Nakayama M, Rendl M, Bickmore 1001 WA, Koseki H, Zheng D, Ezhkova E. 2018. PRC1 Fine-tunes Gene Repression and Activation to 1002 Safeguard Skin Development and Stem Cell Specification. Cell Stem Cell 22:726-739.e7. 1003 doi:10.1016/j.stem.2018.04.005 1004 Cooper S, Grijzenhout A, Underwood E, Ancelin K, Zhang T, Nesterova TB, Anil-Kirmizitas B, Bassett 1005 A, Kooistra SM, Agger K, Helin K, Heard E, Brockdorff N. 2016. Jarid2 binds mono- 1006 ubiquitylated H2A lysine 119 to mediate crosstalk between Polycomb complexes PRC1 and 1007 PRC2. Nat Commun 7:13661–13661. doi:10.1038/ncomms13661 1008 Crusoe MR, Alameldin HF, Awad S, Boucher E, Caldwell A, Cartwright R, Charbonneau A, 1009 Constantinides B, Edvenson G, Fay S, Fenton J, Fenzl T, Fish J, Garcia-Gutierrez L, Garland P, 1010 Gluck J, González I, Guermond S, Guo J, Gupta A, Herr JR, Howe A, Hyer A, Härpfer A, Irber L, 1011 Kidd R, Lin D, Lippi J, Mansour T, McA’Nulty P, McDonald E, Mizzi J, Murray KD, Nahum JR, 1012 Nanlohy K, Nederbragt AJ, Ortiz-Zuazaga H, Ory J, Pell J, Pepe-Ranney C, Russ ZN, Schwarz E, 1013 Scott C, Seaman J, Sievert S, Simpson J, Skennerton CT, Spencer J, Srinivasan R, Standage D, 1014 Stapleton JA, Steinman SR, Stein J, Taylor B, Trimble W, Wiencko HL, Wright M, Wyss B, 1015 Zhang Q, zyme en, Brown CT. 2015. The khmer software package: enabling efficient 1016 nucleotide sequence analysis. F1000Research 4:900. doi:10.12688/f1000research.6924.1 1017 Czermin B, Melfi R, McCabe D, Seitz V, Imhof A, Pirrotta V. 2002. Drosophila enhancer of Zeste/ESC 1018 complexes have a histone H3 methyltransferase activity that marks chromosomal Polycomb 1019 sites. Cell 111:185–196. doi:10.1016/S0092-8674(02)00975-3 1020 de Napoles M, Mermoud JE, Wakao R, Tang YA, Endoh M, Appanah R, Nesterova TB, Silva J, Otte AP, 1021 Vidal M, Koseki H, Brockdorff N. 2004. Polycomb Group Proteins Ring1A/B Link 1022 Ubiquitylation of Histone H2A to Heritable Gene Silencing and X Inactivation. Dev Cell 7:663– 1023 676. doi:10.1016/J.DEVCEL.2004.10.005 1024 Di Croce L, Helin K. 2013. Transcriptional regulation by Polycomb group proteins. Nat Struct Mol Biol 1025 20:1147–1155. doi:10.1038/nsmb.2669 1026 Dietrich N, Bracken AP, Trinh E, Schjerling CK, Koseki H, Rappsilber J, Helin K, Hansen KH. 2007. 1027 Bypass of senescence by the polycomb group protein CBX8 through direct binding to the 1028 INK4A-ARF locus. EMBO J 26:1637–1648. doi:10.1038/sj.emboj.7601632 1029 Endoh M, Endo TA, Endoh T, Isono Kyoichi, Sharif J, Ohara O, Toyoda T, Ito T, Eskeland R, Bickmore 1030 WA, Vidal M, Bernstein BE, Koseki H. 2012. Histone H2A mono-ubiquitination is a crucial step bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1031 to mediate PRC1-dependent repression of developmental genes to maintain ES cell identity. 1032 PLoS Genet 8. doi:10.1371/journal.pgen.1002774 1033 Endoh M, Endo TA, Shinga J, Hayashi K, Farcas A, Ma K-WW, Ito S, Sharif J, Endoh T, Onaga N, 1034 Nakayama M, Ishikura T, Masui O, Kessler BM, Suda T, Ohara O, Okuda A, Klose R, Koseki H. 1035 2017. PCGF6-PRC1 suppresses premature differentiation of mouse embryonic stem cells by 1036 regulating germ cell-related genes. eLife 6:1–26. doi:10.7554/eLife.21064 1037 Erdel F, Rademacher A, Vlijm R, Tünnermann J, Frank L, Weinmann R, Schweigert E, Yserentant K, 1038 Hummert J, Bauer C, Schumacher S, Al Alwash A, Normand C, Herten D-P, Engelhardt J, 1039 Rippe K. 2020. Mouse Heterochromatin Adopts Digital Compaction States without Showing 1040 Hallmarks of HP1-Driven Liquid-Liquid Phase Separation. Mol Cell S1097276520300757. 1041 doi:10.1016/j.molcel.2020.02.005 1042 Eskeland R, Leeb M, Grimes GR, Kress C, Boyle S, Sproul D, Gilbert N, Fan Y, Skoultchi AI, Wutz A, 1043 Bickmore WA. 2010. Ring1B Compacts Chromatin Structure and Represses Gene Expression 1044 Independent of Histone Ubiquitination. Mol Cell 38:452–464. 1045 doi:10.1016/J.MOLCEL.2010.02.032 1046 Farcas AM, Blackledge NP, Sudbery I, Long HK, McGouran JF, Rose NR, Lee S, Sims D, Cerase A, 1047 Sheahan TW, Koseki H, Brockdorff N, Ponting CP, Kessler BM, Klose RJ. 2012. KDM2B links 1048 the polycomb repressive complex 1 (PRC1) to recognition of CpG islands. eLife 2012:1–26. 1049 doi:10.7554/eLife.00205 1050 Ficz G, Heintzmann R, Arndt-Jovin DJ. 2005. Polycomb group protein complexes exchange rapidly in 1051 living Drosophila. Dev Camb Engl 132:3963–76. doi:10.1242/dev.01950 1052 Fischle W, Wang Y, Jacobs SA, Kim Y, Allis CD, Khorasanizadeh S. 2003. Molecular basis for the 1053 discrimination of repressive methyl-lysine marks in histone H3 by polycomb and HP1 1054 chromodomains. Genes Dev 17:1870–1881. doi:10.1101/gad.1110503 1055 Fonseca JP, Steffen PA, Müller S, Lu J, Sawicka A, Seiser C, Ringrose L. 2012. In vivo Polycomb kinetics 1056 and mitotic chromatin binding distinguish stem cells from differentiated cells. Genes Dev 1057 26:857–871. 1058 Francis NJ, Kingston RE, Woodcock CL. 2004. Chromatin compaction by a polycomb group protein 1059 complex. Science 306:1574–7. doi:10.1126/science.1100576 1060 Francis NJ, Saurin AJ, Shao Z, Kingston RE. 2001. Reconstitution of a Functional Core Polycomb 1061 Repressive Complex. Mol Cell 8:545–556. doi:10.1016/S1097-2765(01)00316-1 1062 Fursova NA, Blackledge NP, Nakayama M, Ito S, Koseki Y, Farcas AM, King HW, Koseki H, Klose RJ. 1063 2019. Synergy between Variant PRC1 Complexes Defines Polycomb-Mediated Gene 1064 Repression. Mol Cell 74:1020-1036.e8. doi:10.1016/J.MOLCEL.2019.03.024 1065 Gao Z, Zhang J, Bonasio R, Strino F, Sawai A, Parisi F, Kluger Y, Reinberg D. 2012. PCGF Homologs, 1066 CBX Proteins, and RYBP Define Functionally Distinct PRC1 Family Complexes. Mol Cell 1067 45:344–356. doi:10.1016/j.molcel.2012.01.002 1068 Gearhart MD, Corcoran CM, Wamstad JA, Bardwell VJ. 2006. Polycomb group and SCF ubiquitin 1069 ligases are found in a novel BCOR complex that is recruited to BCL6 targets. Mol Cell Biol 1070 26:6880–9. doi:10.1128/MCB.00630-06 1071 Gong Y, Wang X, Liu J, Shi L, Yin B, Peng X, Qiang B, Yuan J. 2005. NSPc1, a mainly nuclear localized 1072 protein of novel PcG family members, has a transcription repression activity related to its 1073 PKC phosphorylation site at S183. FEBS Lett 579:115–121. doi:10.1016/j.febslet.2004.11.056 1074 Gonzalez I, Mateos-Langerak J, Thomas A, Cheutin T, Cavalli G. 2014. Identification of Regulators of 1075 the Three-Dimensional Polycomb Organization by a Microscopy-Based Genome-wide RNAi 1076 Screen. Mol Cell 54:485–499. doi:10.1016/j.molcel.2014.03.004 1077 Grau DJ, Chapman BA, Garlick JD, Borowsky M, Francis NJ, Kingston RE. 2011. Compaction of 1078 chromatin by diverse polycomb group proteins requires localized regions of high charge. 1079 Genes Dev 25:2210–2221. doi:10.1101/gad.17288211 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1080 Grimm JB, Muthusamy AK, Liang Y, Brown TA, Lemon WC, Patel R, Lu R, Macklin JJ, Keller PJ, Ji N, 1081 Lavis LD. 2017. A general method to fine-tune fluorophores for live-cell and in vivo imaging. 1082 Nat Methods 14:987–994. doi:10.1038/nmeth.4403 1083 Haeussler M, Schönig K, Eckert H, Eschstruth A, Mianné J, Renaud J-B, Schneider-Maunoury S, 1084 Shkumatava A, Teboul L, Kent J, Joly J-S, Concordet J-P. 2016. Evaluation of off-target and 1085 on-target scoring algorithms and integration into the guide RNA selection tool CRISPOR. 1086 Genome Biol 17:148. doi:10.1186/s13059-016-1012-2 1087 Hansen AS, Pustova I, Cattoglio C, Tjian R, Darzacq X. 2017. CTCF and cohesin regulate chromatin 1088 loop stability with distinct dynamics. eLife 6:1–33. doi:10.7554/eLife.25776 1089 Hansen AS, Woringer M, Grimm JB, Lavis LD, Tjian R, Darzacq X. 2018. Robust model-based analysis 1090 of single-particle tracking experiments with Spot-On. eLife 7. doi:10.7554/eLife.33125 1091 Hauri S, Comoglio F, Seimiya M, Paro R, Gstaiger M, Beisel C. 2016. A High-Density Map for 1092 Navigating the Human Polycomb Complexome. Cell Rep 17:583–595. 1093 doi:10.1016/j.celrep.2016.08.096 1094 He J, Shen L, Wan M, Taranova O, Wu H, Zhang Y. 2013. Kdm2b maintains murine embryonic stem 1095 cell status by recruiting PRC1 complex to CpG islands of developmental genes. Nat Cell Biol 1096 15:373–384. doi:10.1038/ncb2702 1097 Isono K, Endo TA, Ku M, Yamada D, Suzuki R, Sharif J, Ishikura T, Toyoda T, Bernstein B, Koseki H. 1098 2013. SAM domain polymerization links subnuclear clustering of PRC1 to gene silencing. Dev 1099 Cell 26:565–577. doi:10.1016/j.devcel.2013.08.016 1100 Izeddin I, Récamier V, Bosanac L, Cissé II, Boudarene L, Dugast-Darzacq C, Proux F, Bénichou O, 1101 Voituriez R, Bensaude O, Dahan M, Darzacq X. 2014. Single-molecule tracking in live cells 1102 reveals distinct target-search strategies of transcription factors in the nucleus. eLife 1–27. 1103 doi:10.7554/eLife.02230 1104 Jiao L, Liu X. 2015. Structural basis of histone H3K27 trimethylation by an active polycomb repressive 1105 complex 2. Science 350:aac4383–aac4383. doi:10.1126/science.aac4383 1106 Junco SE, Wang R, Gaipa JC, Taylor AB, Schirf V, Gearhart MD, Bardwell VJ, Demeler B, Hart PJ, Kim 1107 CA. 2013. Structure of the polycomb group protein PCGF1 in complex with BCOR reveals 1108 basis for binding selectivity of PCGF homologs. Structure 21:665–671. 1109 doi:10.1016/j.str.2013.02.013 1110 Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, Haussler D. 2002. The Human 1111 Genome Browser at UCSC. Genome Res 12:996–1006. doi:10.1101/gr.229102 1112 Kim CA, Gingery M, Pilpa RM, Bowie JU. 2002. The SAM domain of polyhomeotic forms a helical 1113 polymer. Nat Struct Biol 9:453–457. doi:10.1038/nsb802 1114 Kimura H, Cook PR. 2001. Kinetics of core histones in living human cells: little exchange of H3 and H4 1115 and some rapid exchange of H2B. J Cell Biol 153:1341–53. doi:10.1083/jcb.153.7.1341 1116 King HW, Fursova NA, Blackledge NP, Klose RJ. 2018. Polycomb repressive complex 1 shapes the 1117 nucleosome landscape but not accessibility at target genes. Genome Res 28:1494–1507. 1118 doi:10.1101/gr.237180.118 1119 King IFG, Francis NJ, Kingston RE. 2002. Native and Recombinant Polycomb Group Complexes 1120 Establish a Selective Block to Template Accessibility To Repress Transcription In Vitro. Mol 1121 Cell Biol 22:7919–7928. doi:10.1128/MCB.22.22.7919-7928.2002 1122 Kloet SL, Makowski MM, Baymaz HI, van Voorthuijsen L, Karemaker ID, Santanach A, Jansen PWTC, 1123 Di Croce L, Vermeulen M. 2016. The dynamic interactome and genomic targets of Polycomb 1124 complexes during stem-cell differentiation. Nat Struct Mol Biol 23:682–90. 1125 doi:10.1038/nsmb.3248 1126 Klose RJ, Cooper S, Farcas AM, Blackledge NP, Brockdorff N. 2013. Chromatin Sampling-An Emerging 1127 Perspective on Targeting Polycomb Repressor Proteins. PLoS Genet 9:e1003717–e1003717. 1128 doi:10.1371/journal.pgen.1003717 1129 Kouzarides T. 2007. Chromatin Modifications and Their Function. Cell 128:693–705. 1130 doi:10.1016/j.cell.2007.02.005 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1131 Ku M, Koche RP, Rheinbay E, Mendenhall EM, Endoh M, Mikkelsen TS, Presser A, Nusbaum C, Xie X, 1132 Chi AS, Adli M, Kasif S, Ptaszek LM, Cowan CA, Lander ES, Koseki H, Bernstein BE. 2008. 1133 Genomewide analysis of PRC1 and PRC2 occupancy identifies two classes of bivalent 1134 domains. PLoS Genet 4:e1000242–e1000242. doi:10.1371/journal.pgen.1000242 1135 Kundu S, Ji F, Sunwoo H, Jain G, Lee JT, Sadreyev RI, Dekker J, Kingston RE. 2017. Polycomb 1136 Repressive Complex 1 Generates Discrete Compacted Domains that Change during 1137 Differentiation. Mol Cell 65:432-446.e5. doi:10.1016/j.molcel.2017.01.009 1138 Kuzmichev A, Nishioka K, Edrjument-Bromage H, Tempst P, Reinberg D. 2002. Histone 1139 methyltransferase activity associated with a human multiprotein complex containing the 1140 Enhancer of Zeste protein. Genes Dev 16:2893–2905. doi:10.1101/gad.1035902.repression 1141 Lau MS, Schwartz MG, Kundu S, Savol AJ, Wang PI, Marr SK, Grau DJ, Schorderet P, Sadreyev RI, 1142 Tabin CJ, Kingston RE. 2017. Mutation of a nucleosome compaction region disrupts 1143 Polycomb-mediated axial patterning. Science 355:1081–1084. doi:10.1126/science.aah5403 1144 Lavis LD, Grimm JB, English BP, Choi H, Muthusamy AK, Mehl BP, Dong P, Brown TA, Lippincott- 1145 Schwartz J, Liu Z, Lionnet T. 2016. Bright photoactivatable fluorophores for single-molecule 1146 imaging. Nat Methods 13:985–985. doi:10.1101/066779 1147 Lee HG, Kahn TG, Simcox A, Schwartz YB, Pirrotta V. 2015. Genome-wide activities of Polycomb 1148 complexes control pervasive transcription. Genome Res 25:1170–1181. 1149 doi:10.1101/gr.188920.114 1150 Lee TI, Jenner RG, Boyer LA, Guenther MG, Levine SS, Kumar RM, Chevalier B, Johnstone SE, Cole 1151 MF, Isono Kyoichi, Koseki H, Fuchikami T, Abe K, Murray HL, Zucker JP, Yuan B, Bell GW, 1152 Herbolsheimer E, Hannett NM, Sun K, Odom DT, Otte AP, Volkert TL, Bartel DP, Melton DA, 1153 Gifford DK, Jaenisch R, Young RA. 2006. Control of Developmental Regulators by Polycomb in 1154 Human Embryonic Stem Cells. Cell 125:301–313. doi:10.1016/j.cell.2006.02.043 1155 Leeb M, Pasini D, Novatchkova M, Jaritz M, Helin K, Wutz A. 2010. Polycomb complexes act 1156 redundantly to repress genomic repeats and genes. Genes Dev 24:265–276. 1157 doi:10.1101/gad.544410 1158 Li Z, Cao R, Wang M, Myers MP, Zhang Y, Xu R-M. 2006. Structure of a Bmi-1-Ring1B Polycomb 1159 Group Ubiquitin Ligase Complex. J Biol Chem 281:20643–20649. 1160 doi:10.1074/jbc.M602461200 1161 Los GV, Encell LP, McDougall MG, Hartzell DD, Karassina N, Zimprich C, Wood MG, Learish R, Ohana 1162 RF, Urh M, Simpson D, Mendez J, Zimmerman K, Otto P, Vidugiris G, Zhu J, Darzins A, 1163 Klaubert DH, Bulleit RF, Wood KV. 2008. HaloTag: A novel protein labeling technology for cell 1164 imaging and protein analysis. ACS Chem Biol 3:373–382. doi:10.1021/cb800025k 1165 Madeira F, Park Y mi, Lee J, Buso N, Gur T, Madhusoodanan N, Basutkar P, Tivey ARN, Potter SC, Finn 1166 RD, Lopez R. 2019. The EMBL-EBI search and sequence analysis tools APIs in 2019. Nucleic 1167 Acids Res 47:W636–W641. doi:10.1093/nar/gkz268 1168 Mazza D, Abernathy A, Golob N, Morisaki T, McNally JG. 2012. A benchmark for chromatin binding 1169 measurements in live cells. Nucleic Acids Res 40:1–13. doi:10.1093/nar/gks701 1170 McGinty RK, Henrici RC, Tan S. 2014. Crystal structure of the PRC1 ubiquitylation module bound to 1171 the nucleosome. Nature 514:591–6. doi:10.1038/nature13890 1172 McNally JG. 2008. Quantitative FRAP in Analysis of Molecular Binding Dynamics In Vivo. Methods Cell 1173 Biol, Fluorescent Proteins 85:329–351. doi:10.1016/S0091-679X(08)85014-5 1174 Mikkelsen TS, Ku M, Jaffe DB, Issac B, Lieberman E, Giannoukos G, Alvarez P, Brockman W, Kim T-K, 1175 Koche RP, Lee W, Mendenhall E, O’Donovan A, Presser A, Russ C, Xie X, Meissner A, Wernig 1176 M, Jaenisch R, Nusbaum C, Lander ES, Bernstein BE. 2007. Genome-wide maps of chromatin 1177 state in pluripotent and lineage-committed cells. Nature 448:553–560. 1178 doi:10.1038/nature06008 1179 Morey L, Aloia L, Cozzuto L, Benitah SA, Di Croce L. 2013. RYBP and Cbx7 Define Specific Biological 1180 Functions of Polycomb Complexes in Mouse Embryonic Stem Cells. Cell Rep 3:60–69. 1181 doi:10.1016/j.celrep.2012.11.026 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1182 Mueller F, Mazza D, Stasevich TJ, McNally JG. 2010. FRAP and kinetic modeling in the analysis of 1183 nuclear protein dynamics: what do we really know? Curr Opin Cell Biol 22:403–411. 1184 doi:10.1016/j.ceb.2010.03.002 1185 Müller J, Hart CM, Francis NJ, Vargas ML, Sengupta A, Wild B, Miller EL, O’Connor MB, Kingston RE, 1186 Simon JA. 2002. Histone methyltransferase activity of a Drosophila Polycomb group 1187 repressor complex. Cell 111:197–208. doi:10.1016/S0092-8674(02)00976-5 1188 Musselman CA, Lalonde M-E, Côté J, Kutateladze TG. 2012. Perceiving the epigenetic landscape 1189 through histone readers. Nat Struct Mol Biol 19:1218–1227. doi:10.1038/nsmb.2436 1190 Nabet B, Roberts JM, Buckley DL, Paulk J, Dastjerdi S, Yang A, Leggett AL, Erb MA, Lawlor MA, Souza 1191 A, Scott TG, Vittori S, Perry JA, Qi J, Winter GE, Wong K-K, Gray NS, Bradner JE. 2018. The 1192 dTAG system for immediate and target-specific protein degradation. Nat Chem Biol 14:431– 1193 441. doi:10.1038/s41589-018-0021-8 1194 Ollion J, Cochennec J, Loll F, Escudé C, Boudier T. 2013. TANGO: a generic tool for high-throughput 1195 3D image analysis for studying nuclear organization. Bioinformatics 29:1840–1841. 1196 doi:10.1093/bioinformatics/btt276 1197 Pasini D, Bracken AP, Jensen MR, Denchi EL, Helin K. 2004. Suz12 is essential for mouse development 1198 and for EZH2 histone methyltransferase activity. EMBO J 23:4061–4071. 1199 doi:10.1038/sj.emboj.7600402 1200 Pintacuda G, Wei G, Roustan C, Kirmizitas BA, Solcan N, Cerase A, Castello A, Mohammed S, 1201 Moindrot B, Nesterova TB, Brockdorff N. 2017. hnRNPK Recruits PCGF3/5-PRC1 to the Xist 1202 RNA B-Repeat to Establish Polycomb-Mediated Chromosomal Silencing. Mol Cell 68:955- 1203 969.e10. doi:10.1016/J.MOLCEL.2017.11.013 1204 Plys AJ, Davis CP, Kim J, Rizki G, Keenen MM, Marr SK, Kingston RE. 2019. Phase separation of 1205 Polycomb-repressive complex 1 is governed by a charged disordered region of CBX2. Genes 1206 Dev. doi:10.1101/gad.326488.119 1207 Ramírez F, Dündar F, Diehl S, Grüning BA, Manke T. 2014. deepTools: a flexible platform for 1208 exploring deep-sequencing data. Nucleic Acids Res 42:W187–W191. doi:10.1093/nar/gku365 1209 Ran FA, Hsu PPD, Wright J, Agarwala V, Scott D a, Zhang F. 2013. Genome engineering using the 1210 CRISPR-Cas9 system. Nat Protoc 8:2281–308. doi:10.1038/nprot.2013.143 1211 Ren X, Vincenz C, Kerppola TK. 2008. Changes in the distributions and dynamics of polycomb 1212 repressive complexes during embryonic stem cell differentiation. Mol Cell Biol 28:2884–95. 1213 doi:10.1128/MCB.00949-07 1214 Rose NR, King HW, Blackledge NP, Fursova NA, Ember KJI, Fischer R, Kessler BM, Klose RJ. 2016. RYBP 1215 stimulates PRC1 to shape chromatin-based communication between Polycomb repressive 1216 complexes. eLife 5:1–29. doi:10.7554/eLife.18591 1217 Saurin AJ, Shiels C, Williamson J, Satijn DP, Otte AP, Sheer D, Freemont PS. 1998. The human 1218 polycomb group complex associates with pericentromeric heterochromatin to form a novel 1219 nuclear domain. J Cell Biol 142:887–98. doi:10.1083/jcb.142.4.887 1220 Scelfo A, Fernández-Pérez D, Tamburri S, Zanotti M, Lavarone E, Soldi M, Bonaldi T, Ferrari KJ, Pasini 1221 D. 2019. Functional Landscape of PCGF Proteins Reveals Both RING1A/B-Dependent-and 1222 RING1A/B-Independent-Specific Activities. Mol Cell 74:1037-1052.e7. 1223 doi:10.1016/J.MOLCEL.2019.04.002 1224 Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, Preibisch S, Rueden C, 1225 Saalfeld S, Schmid B, Tinevez J-Y, White DJ, Hartenstein V, Eliceiri K, Tomancak P, Cardona A. 1226 2012. Fiji: an open-source platform for biological-image analysis. Nat Methods 9:676–682. 1227 doi:10.1038/nmeth.2019 1228 Schmiedeberg L, Skene P, Deaton A, Bird A. 2009. A Temporal Threshold for Formaldehyde 1229 Crosslinking and Fixation. PLoS ONE 4:e4636–e4636. doi:10.1371/journal.pone.0004636 1230 Schoeftner S, Sengupta AK, Kubicek S, Mechtler K, Spahn L, Koseki H, Jenuwein T, Wutz A. 2006. 1231 Recruitment of PRC1 function at the initiation of X inactivation independent of PRC2 and 1232 silencing. EMBO J 25:3110–3122. doi:10.1038/sj.emboj.7601187 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1233 Schoenfelder S, Sugar R, Dimond A, Javierre B, Armstrong H, Mifsud B, Dimitrova E, Matheson L, 1234 Tavares-Cadete F, Furlan-Magaril M, Segonds-Pichon A, Jurkowski W, Wingett SW, Tabbada 1235 K, Andrews S, Herman B, LeProust E, Osborne CS, Koseki H, Fraser P, Luscombe NM, Elderkin 1236 S. 2015. Polycomb repressive complex PRC1 spatially constrains the mouse embryonic stem 1237 cell genome. Nat Genet 47:1179–86. doi:10.1038/ng.3393 1238 Schuettengruber B, Bourbon H-M, Croce LD, Cavalli G. 2017. Genome Regulation by Polycomb and 1239 Trithorax: 70 Years and Counting. Cell 171:34–57. doi:10.1016/J.CELL.2017.08.002 1240 Schwanhäusser B, Busse D, Li N, Dittmar G, Schuchhardt J, Wolf J, Chen W, Selbach M. 2011. Global 1241 quantification of mammalian gene expression control. Nature 473:337–342. 1242 doi:10.1038/nature10098 1243 Seale RL. 1981. Rapid turnover of the histone-ubiquitin conjugate, protein A24. Nucleic Acids Res 1244 9:3151–3158. doi:10.1093/nar/9.13.3151 1245 Shao Z, Raible F, Mollaaghababa R, Guyon JR, Wu C, Bender W, Kingston RE. 1999. Stabilization of 1246 Chromatin Structure by PRC1, a Polycomb Complex. Cell 98:37–46. doi:10.1016/s0092- 1247 8674(00)80604-2 1248 Simon JA, Kingston RE. 2013. Occupying Chromatin: Polycomb Mechanisms for Getting to Genomic 1249 Targets, Stopping Transcriptional Traffic, and Staying Put. Mol Cell 49:808–824. 1250 doi:10.1016/j.molcel.2013.02.013 1251 Simon JA, Kingston RE. 2009. Mechanisms of polycomb gene silencing: knowns and unknowns. Nat 1252 Rev Mol Cell Biol 10:697–708. doi:10.1038/nrm2763 1253 Steffen PA, Fonseca JP, Gänger C, Dworschak E, Kockmann T, Beisel C, Ringrose L. 2013. Quantitative 1254 in vivo analysis of chromatin binding of Polycomb and Trithorax group proteins reveals 1255 retention of ASH1 on mitotic chromatin. Nucleic Acids Res 41:5235–5250. 1256 doi:10.1093/nar/gkt217 1257 Suzuki M, Mizutani-Koseki Y, Fujimura Y-I, Miyagishima H, Kaneko T, Takada Y, Akasaka T, Tanzawa 1258 H, Takihara Y, Nakano M, Masumoto H, Vidal M, Isono K, Koseki H. 2002. Involvement of the 1259 Polycomb-group gene Ring1B in the specification of the anterior-posterior axis in mice. Dev 1260 Camb Engl 129:4171–4183. 1261 Taherbhoy AM, Huang OW, Cochran AG. 2015. BMI1–RING1B is an autoinhibited RING E3 ubiquitin 1262 ligase. Nat Commun 6. doi:10.1038/ncomms8621 1263 Tamburri S, Lavarone E, Fernández-Pérez D, Conway E, Zanotti M, Manganaro D, Pasini D. 2020. 1264 Histone H2AK119 Mono-Ubiquitination Is Essential for Polycomb-Mediated Transcriptional 1265 Repression. Mol Cell 77:840-856.e5. doi:10.1016/j.molcel.2019.11.021 1266 Tatavosian R, Duc HN, Huynh TN, Fang D, Schmitt B, Shi X, Deng Y, Phiel C, Yao T, Zhang Z, Wang H, 1267 Ren X. 2018. Live-cell single-molecule dynamics of PcG proteins imposed by the DIPG 1268 H3.3K27M mutation. Nat Commun 9:2080–2080. doi:10.1038/s41467-018-04455-7 1269 Tatavosian R, Kent S, Brown K, Yao T, Duc HN, Huynh TN, Zhen CY, Ma B, Wang H, Ren X. 2019. 1270 Nuclear condensates of the Polycomb protein chromobox 2 (CBX2) assemble through phase 1271 separation. J Biol Chem 294:1451–1463. doi:10.1074/jbc.RA118.006620 1272 Tavares L, Dimitrova E, Oxley D, Webster J, Poot R, Demmers J, Bezstarosti K, Taylor S, Ura H, Koide 1273 H, Wutz A, Vidal M, Elderkin S, Brockdorff N. 2012. RYBP-PRC1 complexes mediate H2A 1274 ubiquitylation at polycomb target sites independently of PRC2 and H3K27me3. Cell 148:664– 1275 678. doi:10.1016/j.cell.2011.12.029 1276 Teif VB, Vainshtein Y, Caudron-Herger M, Mallm J-P, Marth C, Höfer T, Rippe K. 2012. Genome-wide 1277 nucleosome positioning during embryonic stem cell development. Nat Struct Mol Biol 1278 19:1185–1192. doi:10.1038/nsmb.2419 1279 Teves SS, An L, Hansen AS, Xie L, Darzacq X, Tjian R. 2016. A dynamic mode of mitotic bookmarking 1280 by transcription factors. eLife 5. doi:10.7554/eLife.22280 1281 Tokunaga M, Imamoto N, Sakata-Sogawa K. 2008. Highly inclined thin illumination enables clear 1282 single-molecule imaging in cells. Nat Methods 5:159–161. doi:10.1038/NMETH.1171 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1283 Tsuboi M, Kishi Y, Yokozeki W, Koseki H, Hirabayashi Y, Gotoh Y. 2018. Ubiquitination-Independent 1284 Repression of PRC1 Targets during Neuronal Fate Restriction in the Developing Mouse 1285 Neocortex. Dev Cell 47:758-772.e5. doi:10.1016/j.devcel.2018.11.018 1286 Uphoff S, Reyes-Lamothe R, Garza de Leon F, Sherratt DJ, Kapanidis AN. 2013. Single-molecule DNA 1287 repair in live bacteria. Proc Natl Acad Sci 110:8063–8068. doi:10.1073/pnas.1301804110 1288 Vandenbunder B, Fourre N, Leray A, Mueller F, Volkel P, Angrand PO, Heliot L. 2014. PRC1 1289 components exhibit different binding kinetics in Polycomb bodies. Biol Cell 106:111–125. 1290 doi:10.1111/boc.201300077 1291 Vidal M. 2019. Polycomb Assemblies Multitask to Regulate Transcription. Epigenomes 3:12–12. 1292 doi:10.3390/epigenomes3020012 1293 Voigt P, Tee W-W, Reinberg D. 2013. A double take on bivalent promoters. Genes Dev 27:1318– 1294 1338. doi:10.1101/gad.219626.113 1295 Wang H, Wang L, Erdjument-Bromage H, Vidal M, Tempst P, Jones RS, Zhang Y. 2004. Role of histone 1296 H2A ubiquitination in Polycomb silencing. Nature 431:873–878. doi:10.1038/nature02926 1297 Wang L, Brown JL, Cao R, Zhang Y, Kassis JA, Jones RS. 2004. Hierarchical Recruitment of Polycomb 1298 Group Silencing Complexes. Mol Cell 14:637–646. doi:10.1016/j.molcel.2004.05.009 1299 Wang R, Taylor AB, Leal BZ, Chadwell LV, Ilangovan U, Robinson AK, Schirf V, Hart PJ, Lafer EM, 1300 Demeler B, Hinck AP, McEwen DG, Kim CA. 2010. Polycomb Group Targeting through 1301 Different Binding Partners of RING1B C-Terminal Domain. Structure 18:966–975. 1302 doi:10.1016/j.str.2010.04.013 1303 Wang Z, Gearhart MD, Lee Y-W, Kumar I, Ramazanov B, Zhang Y, Hernandez C, Lu AY, Neuenkirchen 1304 N, Deng J, Jin J, Kluger Y, Neubert TA, Bardwell VJ, Ivanova NB. 2018. A Non-canonical BCOR- 1305 PRC1.1 Complex Represses Differentiation Programs in Human ESCs. Cell Stem Cell 22:235- 1306 251.e9. doi:10.1016/j.stem.2017.12.002 1307 Wani AH, Boettiger AN, Schorderet P, Ergun A, Münger C, Sadreyev RI, Zhuang X, Kingston RE, 1308 Francis NJ. 2016. Chromatin topology is coupled to Polycomb group protein subnuclear 1309 organization. Nat Commun 7:10291–10291. doi:10.1038/ncomms10291 1310 Wegel E, Göhler A, Lagerholm BC, Wainman A, Uphoff S, Kaufmann R, Dobbie IM. 2016. Imaging 1311 cellular structures in super-resolution with SIM, STED and Localisation Microscopy: A 1312 practical comparison. Sci Rep 6:27290–27290. doi:10.1038/srep27290 1313 Wiśniewski JR, Hein MY, Cox J, Mann M. 2014. A “Proteomic Ruler” for Protein Copy Number and 1314 Concentration Estimation without Spike-in Standards. Mol Cell Proteomics 13:3497–3506. 1315 doi:10.1074/mcp.M113.037309 1316 Wong SJ, Gearhart MD, Taylor AB, Nanyes DR, Ha DJ, Robinson AK, Artigas JA, Lee OJ, Demeler B, 1317 Hart PJ, Bardwell VJ, Kim CA. 2016. KDM2B Recruitment of the Polycomb Group Complex, 1318 PRC1.1, Requires Cooperation between PCGF1 and BCORL1. Structure 24:1795–1801. 1319 doi:10.1016/j.str.2016.07.011 1320 Wu X, Johansen JV, Helin K. 2013. Fbxl10/Kdm2b Recruits Polycomb Repressive Complex 1 to CpG 1321 Islands and Regulates H2A Ubiquitylation. Mol Cell 49:1134–1146. 1322 doi:10.1016/J.MOLCEL.2013.01.016 1323 Yoon YJ, Wu B, Buxbaum AR, Das S, Tsai A, English BP, Grimm JB, Lavis LD, Singer RH. 2016. 1324 Glutamate-induced RNA localization and translation in neurons. Proc Natl Acad Sci 1325 113:E6877–E6886. doi:10.1073/pnas.1614267113 1326 Youmans DT, Schmidt JC, Cech TR. 2018. Live-cell imaging reveals the dynamics of PRC2 and 1327 recruitment to chromatin by SUZ12-associated subunits. Genes Dev. 1328 doi:10.1101/GAD.311936.118 1329 Zeng H, Horie K, Madisen L, Pavlova MN, Gragerova G, Rohde AD, Schimpf BA, Liang Y, Ojala E, 1330 Kramer F, Roth P, Slobodskaya O, Dolka I, Southon EA, Tessarollo L, Bornfeldt KE, Gragerov 1331 A, Pavlakis GN, Gaitanaris GA. 2008. An Inducible and Reversible Mouse Genetic Rescue 1332 System. PLoS Genet 4. doi:10.1371/journal.pgen.1000069 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1333 Zhang Y, Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, Nussbaum C, Myers RM, Brown M, 1334 Li W, Liu XS. 2008. Model-based Analysis of ChIP-Seq (MACS). Genome Biol 9:R137–R137. 1335 doi:10.1186/gb-2008-9-9-r137 1336 Zhen CY, Tatavosian R, Huynh TN, Duc HN, Das R, Kokotovic M, Grimm JB, Lavis LD, Lee J, Mejia FJ, Li 1337 Y, Yao T, Ren X. 2016. Live-cell single-molecule tracking reveals co-recognition of H3K27me3 1338 and DNA targets polycomb Cbx7-PRC1 to chromatin. eLife 5:1–36. doi:10.7554/eLife.17667 1339 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) (C) IP Antibody Input RING1B FLAG

RING1B HT

WT RING1B-HT WT RING1B-HT WT kDa WT RING1B-HT 70 RING1B-HT RING1B-HT-JF549 RING1B RING1B 55

RING1B PCGF1

TBP PCGF2 25 H2AK119ub1 PCGF3 PCGF5 H3 15 PCGF6 RYBP CBX7

Figure 1 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1340 Figure 1 –Endogenous Halo-tagging of RING1B enables live-cell imaging of PRC1.

1341 (A) Western blots for RING1B (upper panels) and H2AK119ub1 (lower panels) comparing 1342 wild type and homozygous RING1B-HaloTag cell lines. TBP and H3 were used as 1343 loading controls. 1344 (B) Immunoprecipitation of RING1B from WT or RING1B-HT ESCs followed by western 1345 blotting for PRC1 subunits to confirm normal complex formation. For WT ESCs, a 1346 control IP was performed using a FLAG antibody.

1347 (C) Single Z-slice of a RING1B-HT ESC labelled with JF549. Scale bar = 5 μm. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) (C) 67 Hz 80

60

0 ms 15 ms 40

20 Percentage bound Percentage

0 30 ms 45 ms

HT-H2B RING1B-HT HT-3xNLS (D) (E) 2 Hz 1.0 2 Hz 0.8

H2B 0.6 0 s 0.5 s H2B fit RING1B

1 - CDF 0.4 RING1B fit

0.2

0 1 s 100 s 0 20 40 60 80 100 Time (s) (F) (G) 0.033 Hz 1.0 0.033 Hz 0.8

H2B 0.6 0 s 30 s H2B fit RING1B

1 - CDF 0.4 RING1B fit

0.2

0 60 s 300 s 0 200 400 600 800 Time (s) (H) (I) (J) 1.2 RING1B-HT 1.8 Nucleus 1 1.5 Polycomb body 0.8

0.6 1.2

0.4

Mean enrichment 0.9 Recovered fraction 0.2 Polycomb body Maximum intensity Non-Polycomb body 0.6 0 projection -100 50 100 150 200 Body Time (s) Non-Body Figure 2 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1348 Figure 2 – PRC1 is highly dynamic with a small stably chromatin-bound fraction.

1349 (A) Example frames from a 67 Hz exposure SPT experiment showing a single RING1B 1350 molecule. Tracked steps of the molecule between frames are superimposed in red. 1351 Scale bar = 1 μm. 1352 (B) Example tracks of individual molecules from a single 15 ms exposure SPT movie 1353 showing a range of diffusive behaviours. Scale bar = 3 μm. 1354 (C) A bar chart indicating the percentage of RING1B, H2B and HaloTag-NLS molecules in 1355 the bound state determined from SPT using Spot-On. Error bars represent SEM for n 1356 = 3 biological replicates of at least 10 movies each. Indicated p-values were calculated 1357 using Student’s t test. 1358 (D) Example frames from a 2 Hz exposure SPT experiment showing a single molecule 1359 visible for 100 s indicated by white arrowheads. Scale bar = 1 μm. 1360 (E) Dwell time distributions (1 – cumulative distribution function, CDF) and fitted 1361 biexponential decay curves for immobile RING1B molecules for a representative 1362 biological replicate of movies acquired as in (D). H2B is included as a photobleaching 1363 control for highly stable binding. n = 3 biological replicates of 8 videos each. 1364 (F) Example frames from a 0.033 Hz exposure SPT experiment showing a single molecule 1365 visible for 300 s indicated by white arrowheads. Scale bar = 1 μm. 1366 (G) Dwell time distributions (1 – cumulative distribution function, CDF) and fitted 1367 biexponential decay curves for immobile RING1B molecules for a representative 1368 biological replicate of movies acquired as in (F). H2B is included as a photobleaching 1369 control for highly stable binding. n = 2 biological replicates of 5 movies each. 1370 (H) Schematic of Polycomb bodies in the nucleus (left panel). Maximum intensity

1371 projection of nuclear RING1B-HT-JF549 signal acquired by spinning disk microscopy. 1372 The yellow arrowhead indicates a single Polycomb body. Scale bar = 5 μm (right 1373 panel). 1374 (I) FRAP recovery curves for RING1B-HaloTag in regions containing a Polycomb body or 1375 elsewhere in the nucleus (Non-Polycomb body). The recovered fraction was measured 1376 relative to initial fluorescence intensity and corrected using an unbleached region. The 1377 error bars denote SEM for 20 cells each for Polycomb body and Non-Polycomb body 1378 regions across 2 biological replicates. 1379 (J) A box plot illustrating the relative mean fluorescence signal per unit volume of all 1380 RING1B Polycomb bodies in each cell (Body) compared to the remaining nuclear 1381 volume in the same cell (Non-Body), normalised to the median non-Polycomb body 1382 fluorescence. n = 63 cells. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B)

0.4

0.3 HaloTag Standard Protein RING1B-HT 0.2 kDa 0.05 0.01 µg 2.5 1 x 105 cells 70

0.1

55 Bound molecules per kb

0.0 1 2 3 4 5 6 7 8 9 10 RING1B density (C) 20 kb

Arl2bp Ccl22 Pllp Cx3cl1

RING1B ChIP-seq

RING1B Peaks

Length (bp) 905 3871 1939 Molecules/kb 0.10 0.15 0.08 Total molecules 0.09 0.58 0.16

Figure 3 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1383 Figure 3 – RING1B has low target site occupancy.

1384 (A) An example of an in-gel fluorescence image of TMR-labelled HaloTag Standard 1385 Protein loaded in known quantities (left panel) and RING1B-HT-TMR from whole-cell 1386 lysates of a known number of cells (right panel). Fluorescence intensity was used to 1387 quantify the number of molecules present (Cattoglio et al., 2019). 1388 (B) A box plot illustrating the number of RING1B molecules that are estimated to be bound 1389 per kilobase of DNA at RING1B peaks from ChIP-seq in ESCs (Blackledge et al., 1390 2020). Read distributions from ChIP-seq were used to segregate bound RING1B 1391 molecules into density deciles over RING1B peaks (18,643). 1392 (C) A genomic snapshot of RING1B ChIP-seq illustrating the varying distribution of 1393 RING1B binding between individual peaks. The values shown below indicate the 1394 length of the RING1B peak, the estimated density of RING1B molecules bound at each 1395 peak, and the estimated number of molecules bound per allele to each peak assuming 1396 a 2n genome. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) PRC1CKO +OHT - + - + - + OHT RING1B RING1B -- + + -- RING1B-HaloTag ---- + + RING1BNBM-HaloTag + RING1B-HaloTag

RING1B HT WT RING1B

RING1B or TBP Nucleosome RING1B HT Binding H2AK119ub1 Mutant H3 (C) (D) 20 0.6 0.72 0.51 18 0.5 16 14 0.4 12 10 0.3 8 6 0.2 Percentage bound Percentage

4 bound stably Fraction 0.1 2 0 0 RING1B RING1BNBM RING1B RING1BNBM (E) (F) 20 < 0.0001 18 16 Full-length RING1B HT 14 RING1B 12 10 8 Minimal catalytic RPCD HT 6 Percentage bound Percentage fusion 4 2 0 RING1B RPCD

Figure 4 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1397 Figure 4 – Interaction between the PRC1 catalytic core and the nucleosome is not a 1398 central determinant of chromatin binding.

1399 (A) A schematic illustrating the cell line used for exogenous expression of RING1B- 1400 HaloTag fusions. Either WT RING1B or a nucleosome binding mutant (NBM) fused to 1401 the HaloTag was stably expressed in Ring1a-/-; Ring1bfl/fl (PRC1CKO) cells. Addition of 1402 OHT causes removal of endogenous RING1B. 1403 (B) Western blots showing loss of endogenous RING1B and H2AK119ub1 following 1404 addition of OHT in PRC1CKO cells and rescue of H2AK119ub1 by RING1B-HaloTag but 1405 not RING1BNBM-HaloTag. TBP and H3 are included as loading controls. 1406 (C) A bar chart indicating the percentage of RING1B and RING1BNBM molecules in the 1407 bound state as determined by SPT. Error bars represent SEM for n = 3 biological 1408 replicates of at least 30 movies each. Indicated p-values were calculated using 1409 Student’s t test. 1410 (D) A bar chart showing the fraction of RING1B and RING1BNBM molecules that exhibit 1411 stable binding. Error bars represent SEM for n = 2 biological replicates of 8 movies 1412 each. Indicated p-values were calculated using Student’s t test. 1413 (E) Schematic showing the minimal RING1B PCGF4 catalytic domain (RPCD) formed by 1414 fusion of the RING domains of RING1B and PCGF4. 1415 (F) A bar chart of percentages of RING1B and RPCD molecules in the bound state. Error 1416 bars represent SEM for n = 3 biological replicates of at least 30 movies each. Indicated 1417 p-values were calculated using Student’s t test. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) Canonical Variant Canonical Variant PCGF2-HT HT-PCGF1 HT-PCGF3 PCGF1 RYBP

PCGF2/4 PCGF3/5 RING1B CBX7 RYBP CBX7-HT HT-PCGF6 HT-RYBP

PCGF6 RYBP

(C) 1.4 80000 Relative fluorescence 70000 1.2 Molecules per cell 60000 1.0 50000 0.8 40000 0.6 30000 0.4 20000 cellMoleculesper Relativefluorescence 0.2 10000 0 0

CBX7 RYBP RING1B PCGF1 PCGF2 PCGF3 PCGF6

Figure 5 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1418 Figure 5 – PCGF2-PRC1 is the most abundant PRC1 complex in ESC.

1419 (A) A schematic illustrating the central canonical and variant PRC1 complexes. 1420 (B) Representative images showing a single Z-slice from ESCs expressing HaloTag

1421 fusions of PRC1 subunits labelled with JF549. Scale bar = 5 μm. 1422 (C) Bar chart showing relative nuclear fluorescence, measured by spinning disk 1423 microscopy, and molecules per cell, measured by in-gel fluorescence, for PRC1 1424 subunits. Error bars indicate standard deviation for n > 35 cells (relative fluorescence) 1425 or n = 3 independent biological replicates of in-gel quantifications (molecules per cell). bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) (C)

PCGF2 25 0.10 0.6 0.0005 0.20 0.25 0.02 20 0.5 0.15 15 0.4 0.3 0.10 10 0.2 0.05

Percentage boundPercentage 5 Fraction stably bound stably Fraction 0.1 Bound molecules per kb 0 0 0 1 2 3 4 5 6 7 8 9 10 CBX7 CBX7 PCGF2 RING1B PCGF2 RING1B RING1B density

(D) 20 kb Barhl1 Ttf1 Setx Ntng2

RING1B

Molecules 1.50 0.14 0.04 0.16 0.20 3.21

PCGF2

Molecules 0.86 0.04 0.01 0.02 0.04 1.68

Figure 6 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1426 Figure 6 – Canonical PRC1 exhibits stable chromatin binding and is restricted to a 1427 subset of Polycomb sites.

1428 (A) A bar chart of the percentage of RING1B, CBX7 and PCGF2 molecules in the bound 1429 state as determined by SPT. Error bars represent SEM for n = 3 biological replicates 1430 of at least 10 movies each. Indicated p-values were calculated using Student’s t test. 1431 (B) A bar chart indicating the fraction of RING1B, CBX7 and PCGF2 molecules exhibiting 1432 stable binding. Error bars represent SEM for at least 2 biological replicates of 8 movies 1433 each. Indicated p-values were calculated using Student’s t test. 1434 (C) A box plot illustrating the number of PCGF2 molecules that are estimated to be bound 1435 per kilobase of DNA at RING1B peaks from ChIP-seq in ESCs (Blackledge et al., 1436 2020). Read distributions from ChIP-seq were used to segregate bound PCGF2 1437 molecules into density deciles over RING1B peaks. 1438 (D) A genomic snapshot illustrating RING1B and PCGF2 ChIP-seq signal at two strong 1439 PRC1 peaks sites and four weaker peaks illustrating the more restricted nature of 1440 PCGF2 binding. Values shown below the ChIP-seq signal indicate the estimated 1441 number RING1B and PCGF2 molecules bound at a single allele for the indicated peak. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) < 0.0001 0.04 35 0.0007 0.6 0.09 < 0.0001 0.07 30 0.02 0.5 0.58 25 0.4 20 0.3 15 10 0.2

Percentage boundPercentage 0.1

5 bound stably Fraction 0 0

RYBP RYBP RING1B PCGF1PCGF3PCGF6 RING1B PCGF1PCGF3PCGF6 (C) (D)

100 0.20 PCGF1 PCGF2 75 0.15 PCGF6

50 * * 0.10

25 0.05 Stable binding time (s) time bindingStable Bound molecules per kb 0 0 1 2 3 4 5 6 7 8 9 10 RYBP PCGF1PCGF3PCGF6 RING1B RING1B density (E) 10 kb Ss18l1 Hrh3

Psma7 Mtg2 Gm17180

RING1B

Molecules 0.22 0.10 0.80

PCGF1

Molecules 0.07 0.02 0.14

PCGF2

Molecules 0.05 0.02 0.29

PCGF6

Molecules 0.05 0.04 0.14

Figure 7 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1442 Figure 7 – Variant PRC1 subunits exhibit distinct dynamics and broad binding to PRC1 1443 targets.

1444 (A) A bar chart indicating the percentage of RING1B, RYBP, PCGF1, PCGF3 and PCGF6 1445 molecules in the bound state determined by SPT. Error bars represent SEM for n = 3 1446 biological replicates of at least 10 movies each. Indicated p-values were calculated 1447 using Student’s t test. 1448 (B) A bar chart indicating the fraction of RING1B, RYBP, PCGF1, PCGF3 and PCGF6 1449 molecules that exhibit stable binding. Error bars represent SEM for at least 2 biological 1450 replicates of 8 movies each. Indicated p-values were calculated using Student’s t test. 1451 (C) A bar chart indicating the stable binding times for RING1B, RYBP, PCGF1, PCGF3 1452 and PCGF6 after photobleaching correction in 2 Hz tracking experiments. Asterisks 1453 indicate proteins for which stable binding exceeded the photobleaching time. Error 1454 bars represent SEM for at least 2 biological replicates of 8 movies each. 1455 (D) A box plot illustrating the number of PCGF1, PCGF2, PCGF6 molecules that are 1456 estimated to be bound per kilobase of DNA at RING1B peaks from ChIP-seq in ESCs 1457 (Blackledge et al., 2020). Read distributions from ChIP-seq were used to segregate 1458 bound molecules into density deciles over RING1B peaks. 1459 (E) A genomic snapshot of RING1B, PCGF1, PCGF2 and PCGF6 ChIP-seq illustrating 1460 the differing distributions of PRC1 complexes across RING1B peaks. Values shown 1461 below the ChIP-seq signal indicate the estimated number of RING1B, PCGF1, PCGF2 1462 and PCGF6 molecules bound at a single allele for the indicated peak. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) dTAG-SUZ12 dTAG-SUZ12 WT RING1B-HT PCGF2-HT HT-PCGF6 kDa 0 1 h 24 h 72 h 0 1 h 24 h 72 h 0 1 h 24 h 72 h dTAG-13 130 dTAG-SUZ12 dTAG SUZ12 SUZ12 100* SUZ12

TBP + dTAG-13 dTAG-SUZ12 WT RING1B-HT PCGF2-HT HT-PCGF6 dTAG SUZ12 0 48 h 72 h 96 h 0 48 h 72 h 96 h 0 48 h 72 h 96 h dTAG-13 H3K27me3 H3

(C) (D) (E) 0.6 100 25 <0.0001 < 0.0001 0.82 0.13 0.16 0.29 0.5 20 75 0.4 15 0.3 50 **** 10 0.2 25 5

Percentage boundPercentage 0.1 Fraction stably bound stably Fraction Stable binding time (s) time bindingStable 0 0 0 dTAG-13 +- +- +- dTAG-13 +- +- +- dTAG-13 +- +- +- RING1B PCGF2 PCGF6 RING1B PCGF2 PCGF6 RING1B PCGF2 PCGF6 (F) (G) RING1B PCGF2 PCGF6 UNT dTAG-13 250 0.11 300 0.002 0.87

200 200 200 150 RING1B 100 100 100

Foci per nucelus 50

0 0 0

UNT UNT UNT PCGF2 dTAG-13 dTAG-13 dTAG-13 (H) PCGF2

100 0.003 0.01 0.008 0.003

PCGF6 50 Foci per nucleus

0 Q1 Q2 Q3 Q4 Focus volume

Figure 8 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1463 Figure 8 – PRC2 and H3K27me3 contribute to the binding frequency of canonical PRC1 1464 but not its stability.

1465 (A) A schematic illustrating the dTAG-SUZ12 system. SUZ12 is tagged endogenously and 1466 is depleted by addition of dTAG-13. 1467 (B) Western blots for RING1B-HT, PCGF2-HT and HT-PCGF6 cell lines with the dTAG- 1468 SUZ12 system showing rapid (< 1 hr) depletion of SUZ12 and loss of H3K27me3 after 1469 96 hrs of dTAG-13 treatment. TBP and H3 are included as loading controls. * indicates 1470 a nonspecific band present in the parental cell line which remains after SUZ12 1471 depletion. 1472 (C) A bar chart indicating the percentage of RING1B, PCGF2, and PCGF6 molecules in 1473 the bound state with and without 96 hr dTAG-13 treatment, as determined from SPT. 1474 Error bars represent SEM for n = 3 biological replicates of 30 movies each. Indicated 1475 p-values were calculated using Student’s t test. 1476 (D) A bar chart indicating the fraction of RING1B, PCGF2, and PCGF6 molecules 1477 exhibiting stable binding with and without 96 hr dTAG-13 treatment. Error bars 1478 represent SEM for n = 2 biological replicates of 8 movies each. Indicated p-values were 1479 calculated using Student’s t test. 1480 (E) A bar chart indicating the stable binding times for RING1B, PCGF2, and PCGF6 with 1481 and without 96 hr dTAG-13 treatment. Asterisks indicate proteins for which stable 1482 binding exceeded the photobleaching time. Error bars represent SEM for n = 2 1483 biological replicates of 8 movies each. 1484 (F) Maximum intensity projections of nuclear RING1B, PCGF2 and PCGF6 signal with and 1485 without 96 hr dTAG-13 treatment. Yellow arrowheads indicate single Polycomb bodies. 1486 Scale bar = 5 μm. 1487 (G) Box plots showing the number of Polycomb bodies detectable for RING1B, PCGF2 1488 and PCGF6 with and without 96 hr dTAG-13 treatment. n > 50 cells per condition from 1489 2 biological replicates. Indicated p-values were calculated using Student’s t test. 1490 (H) Box plots comparing the number of nuclear foci counted for PCGF2 with and without 1491 96 hr dTAG-13 treatment. Foci were divided into quartiles based on focus volumes in 1492 untreated cells. n > 50 cells per condition from 2 biological replicates. Indicated p- 1493 values were calculated using Student’s t test. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) (C)

0.96 0.6 20 0.03 PCGF1 0.03 RING RAWUL 0.5 1 259 15 0.4

1 139 10 0.3 0.2 140 259 NM 5 Percentage bound Percentage 0.1 Fraction stably bound stably Fraction 0 0 FL 1-139 140-259 FL 1-139 140-259 PCGF1 PCGF1

Figure 9 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1494 Figure 9 – The C-terminal RAWUL domain of PCGF1 is required for stable chromatin 1495 binding

1496 (A) A schematic illustrating the PCGF1 constructs used for exogenous expression. The 1497 RING (aa 47 – 86) and RAWUL domains (aa 190 – 253) are indicated. 1498 (B) A bar chart indicating the percentage of each PCGF1 fusion molecule (FL = full length) 1499 in the bound state. Error bars represent SEM for n = 3 biological replicates of 30 movies 1500 each. Indicated p-values were calculated using Student’s t test. 1501 (C) A bar chart indicating the fraction each PCGF1 fusion molecule exhibiting stable 1502 binding. Stable binding of PCGF1 1 – 129 was not measurable (NM). Error bars 1503 represent SEM for n = 2 biological replicates of 8 movies each. Indicated p-values were 1504 calculated using Student’s t test. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) RING1B-HT

P(r) Two state model Three state model

Δτ = 1 dt

Δτ = 2 dt

Δτ = 3 dt

Δτ = 4 dt

Δτ = 5 dt

Δτ = 6 dt

Δτ = 7 dt

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 Jump distance (µm) (B) RING1B-HT H2B-HT HT-3xNLS 300 350 450 400 250 300 350 250 200 300 200 250 150 150 200 100 100 150 100

Probability density 50 Probability density 50 Probability density 50 0 0 0 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 2 −1 2 −1 2 −1 log2(D*) (µm s ) log2(D*) (µm s ) log2(D*) (µm s )

Figure S1 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1505 Figure S1. Related to Figure 2.

1506 (A) Example histograms of apparent diffusion coefficients (D*, μm3 s-1) calculated from 1507 tracking experiments for individual RING1B, H2B and HaloTag-3xNLS molecules. 1508 Tracks analysed are from a single representative replicate of at least 10 movies per 1509 protein. 1510 (B) Jump length histograms for a single replicate of RING1B SPT overlaid with two-state 1511 and three-state models (dashed lines) produced by Spot-On. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) Round, AP+ Flat, AP- Dispersed, AP-

RPCD-HT - + 100 100 100 FLAG PRC1CKO PRC1CKO RING1B-HT PRC1CKO RING1BNBM-HT 80 80 80 HDAC1 60 60 60 40 40 40 20 20 20 % total % total colonies 0 0 0

Flat, AP- Flat, AP- Flat, AP- Round, AP+ Round, AP+ Round, AP+ Dispersed, AP- Dispersed, AP- Dispersed, AP-

Figure S2 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1512 Figure S2. Related to Figure 4.

1513 (A) Alkaline phosphatase (AP) staining analysis of PRC1CKO ESC colonies with and 1514 without 72 hr OHT treatment. Top panel: Example images of three classified colony 1515 types: round, AP staining positive; flat, AP staining negative; dispersed, AP staining 1516 negative. Bottom panel: bar charts of the numbers of ESC colonies for PRC1CKO cells, 1517 and the same cells exogenously expressing RING1B-HT or RING1BNBM-HT. Error bars 1518 indicate SEM from n = 3 biological replicates. 1519 (B) Western blot showing exogenous expression of RPCD-HaloTag using a FLAG epitope 1520 included in the expressed fusion protein. HDAC1 is included as a loading control. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) (C)

kDa WT HT- PCGF1 kDa WT HT-PCGF3 HT-PCGF3 kDa WT PCGF2-HT HT-PCGF1 55 55* 70 PCGF2-HT * PCGF1 PCGF3/5 PCGF2 55 PCGF2 35 35 PCGF3 25 PCGF1 PCGF5 25 TBP TBP TBP (D) (E) (F)

kDa WT CBX7-HT 55 CBX7-HT WT HT-PCGF6 kDa kDa WT HT-RYBP HT-PCGF6 70 CBX7 70 HT-RYBP 35 55 PCGF6 55 RYBP 25 CBX7 PCGF6 35 RYBP TBP TBP TBP

Figure S3 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1521 Figure S3. Related to Figure 5.

1522 (A – F) Western blots showing homozygous tagging of PRC1 subunits. TBP is included as 1523 a loading control. Nonspecific bands in (A) and (C) are indicated by *. Figure S4 (A) bioRxiv preprint (which wasnotcertifiedbypeerreview)istheauthor/funder,whohasgrantedbioRxivalicensetodisplaypreprintinperpetuity.It

Mean enrichment 1.0 1.5 2.0 2.5 3.0 doi: Body PCGF2 https://doi.org/10.1101/2020.04.25.061358 Non-Body made availableundera (B)

Recovered fraction 0.2 0.4 0.6 0.8 1.2 1 0 -10 0 1 CC-BY-NC-ND 4.0Internationallicense ; 01010200 150 100 50 this versionpostedApril25,2020. PCGF2 Time (s) Time Non-Polycomb body Polycomb body . The copyrightholderforthispreprint bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1524 Figure S4. Related to Figure 6.

1525 (A) A box plot illustrating the relative mean fluorescence signal per unit volume of all 1526 PCGF2 Polycomb bodies in each cell (Body) compared to the remaining nuclear 1527 volume in the same cell (Non-Body), normalised to the median non-Polycomb body 1528 fluorescence. n = 36 cells. 1529 (B) FRAP recovery curves for PCGF2-HT in regions containing a Polycomb body or 1530 elsewhere in the nucleus (Non-Polycomb body). The recovered fraction was measured 1531 relative to initial fluorescence intensity, corrected using an unbleached region. Error 1532 bars denote SEM for n > 30 cells per condition. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) (B) RING1B PCGF6

0.005 0.04 0.68 0.29 80 0.87 0.74 0.90 0.29 75 60 50 UNT 40 dTAG-13

25 20 Foci per nucleus

0 0 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Focus volume Focus volume

Figure S5 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1533 Figure S5. Related to Figure 8.

1534 (A) A box plot comparing numbers of nuclear foci counted for RING1B with and without 96 1535 hr dTAG-13 treatment, with foci divided into quartiles based on focus volumes in 1536 untreated cells. n > 50 cells per condition from 2 biological replicates. Indicated p- 1537 values were calculated using Student’s t test. 1538 (B) As in (A) for PCGF6. bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

(A) PCGF1 RING domain

RAWUL domain

(B) HaloTag-PCGF1

kDa - FL 1 - 139140 - 259 70 FLAG 55 HDAC1

Figure S6 bioRxiv preprint doi: https://doi.org/10.1101/2020.04.25.061358; this version posted April 25, 2020. The copyright holder for this preprint (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 made available under aCC-BY-NC-ND 4.0 International license.

1539 Figure S6. Related to Figure 9.

1540 (A) Alignments of PCGF1 amino acid sequences across vertebrate organisms produced 1541 by Clustal Omega (Madeira et al., 2019). Residues shaded in black are identical across 1542 50% of sequences. Residues which are similar are shaded in grey. Positions of RING 1543 (blue) and RAWUL (red) domains in the Mus musculus sequence are annotated. Red 1544 line indicates the site at which the amino acid sequence was split to generate N- and 1545 C-terminal truncations. 1546 (B) Western blot showing exogenous expression of HaloTag fusions of full length PCGF1 1547 (FL) and N- (1 – 139) and C-terminal (140 – 259) truncations, using a FLAG epitope 1548 included in the expressed fusion protein. HDAC1 is included as a loading control.