Direct measurement of sub-kilobase chromatin structure reveals that linker histone H1 broadly compacts chromatin, with differential impact amongst epigenetic states

19 20 Chromatin compaction by linker histone H1 family proteins is a long-standing model for 21 transcriptional repression. However, the biophysical and conformational details of such 22 compaction in situ, at the kilobase- and sub-kilobase length scale relevant to the activity of 23 transcriptional regulatory elements, remain under debate. Rather than inferring such compaction 24 from indirect measurements of features like DNA accessibility, we sought to directly probe sub-25 kilobase contacts between nearby nucleosomes. We developed an improved version of 26 radiation-induced correlated cleavage with sequencing (RICC-seq), which we term RICC-seq 27 2.0, and used it in parallel with Micro-C to cross-validate our measurements of chromatin 28 structure in both diverse cell types with different levels of linker histone and different levels of 29 chromatin compaction, as well as a CRISPRi system for pan-H1 depletion. Using this system, 30 we find that chromatin fiber de-compaction upon H1 depletion is global across the genome, 31 reducing the contrast in inter-nucleosome contacts between acetylated chromatin and the rest 32 of the genome. Surprisingly, this does not dramatically change higher-order chromatin 33 organization such as nuclear compartments. Nevertheless, we observe a broad increase in 34 accessibility at tens of thousands of sites and an increase in expression of over a thousand 35 genes, which are enriched in polycomb repressive complex targets. Investigating the local 36 chromatin compaction at upregulated genes as opposed to genes that do not change 37 transcription, we observe that upregulated genes are not specifically de-compacted. Rather, our 38 data support a model in which linker histone globally induces local compaction of nucleosome 39 contacts and an increase in linker lengths, and repression by PRC1/2 is particularly dependent 40 on these local features of chromatin architecture. 41 42 43 44 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 3 Main Text 45 46 Compaction of the chromatin fiber has been invoked as a mechanism for transcriptional 47 repression since the earliest investigations into the organization of chromatin fibers in vivo and 48 in vitro1–4. In situ, microscopy shows that the density of silent heterochromatin is significantly 49 higher than the density of transcriptionally active euchromatin5,6. In vitro, chromatin fibers with 50 regularly spaced nucleosomes collapse into compact 30 nm-diameter structures, supporting the 51 “30-nm fiber” model of transcriptional repression7. H1 linker histones are a family of proteins 52 essential for development in metazoans, which bind the dyad positions of core nucleosomes via 53 their globular domain and interact with the linker DNA entering and exiting the nucleosome via 54 their unstructured, positively charged N-terminal and C-terminal domains 8. In a variety of 55 chromatin reconstitution experiments, linker histones have been shown to promote the 56 compaction of chromatin fibers or chromatin domains into higher density structures 7,9–12. 57 58 Although the simple 30-nm fiber model dominated the field for decades, efforts to assess in vivo 59 or in situ chromatin fiber structure enabled by advances in electron microscopy and X-ray 60 scattering did not identify the expected long-range regular higher-order helices posed by the 30-61 nm fiber model 13–16. Chromatin was therefore proposed to be an unstructured, liquid-like 62 “polymer melt” of nucleosomes3. This updated model of chromatin as a liquid is consistent with 63 the recently observed propensity of unmodified chromatin to form phase-separated liquid 64 condensates12,17. However, even in condensates, local structural motifs of chromatin fibers 65 dictated by nucleosome modifications and the geometry of inter-nucleosome stacking 66 interactions dictated by linker DNA lengths and architectural proteins such as linker histones can 67 modulate phase separation behavior17. For example, linker histones were shown to increase the 68 density of chromatin condensates12, and sequencing-based methods for mapping local 69 chromatin interactions, such as Micro-C and RICC-seq, found short-range zig-zag 70 tetranucleosome folding signatures18–20. Super-resolution imaging found that chromatin fibers 71 consist of small clusters of nucleosomes, termed “clutches”, the size of which is modulated by 72 factors including linker histones21. In vitro FRET measurements and simulations both point to 73 such clusters or tetranucleosome motifs being highly dynamic22,23. 74 75 A full understanding of chromatin fiber structure and behavior, and the regulation of its 76 interactions with the proteins that carry out DNA-based processes, including DNA replication, 77 transcription, and DNA repair, therefore requires us to reconcile the long-range disorder and the 78 potential local order of chromatin. This is particularly important as the local interactions of 79 nucleosomes determine the accessibility and binding affinity of individual loci to these proteins. 80 For example, the spacing of nucleosomes, which, as a result of DNA’s helical nature and its 81 relative stiffness on the length scale of typical inter-nucleosome linker lengths (~30-70 bp), 82 strongly determines the geometry of nearby nucleosome stacking, is tightly controlled by 83 nucleosome remodelers, forming regular arrays in some areas of the genome17,24–28. Linker 84 histones are one of the strongest determinants of nucleosome spacing, with high linker histone 85 expression correlating with longer average linker lengths and a large nucleosome repeat length 86 (NRL)29. Long linker segments can create boundaries between nucleosome interaction 87 domains30. Structural studies show that chromatin modifiers (“writers”) can bridge nucleosomes 88 as they propagate the histone modification from one nucleosome to another, either alone, like 89 the polycomb repressive complex 2 (PRC2) or through dimerization31,32, as is the case for the 90 H3K9me3-binding protein HP1. The efficiency of deposition of H3K27me3 by PRC2 in vitro is 91 enhanced by compacted chromatin fibers6,33. On the other hand, the efficiency of the 92 modification H3K36me2 by the methyltransferase NSD2 is inhibited by chromatin fiber 93 compaction6. 94 95 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 4 We therefore sought to understand how local chromatin fiber compaction on the scale of a few 96 nucleosomes—the length scale relevant to protein binding at regulatory regions on DNA—is 97 regulated by linker histone in different chromatin contexts and what its consequences are for 98 transcriptional repression. Locus-specific chromatin compaction in situ is very difficult to 99 measure, and proxies for compaction such as DNA accessibility measured by ATAC-seq have 100 historically been used instead. To address this challenge, we sought to use two complementary 101 approaches that rely on different fundamental operating principles: Micro-C18 and radiation-102 induced correlated cleavage with sequencing (RICC-seq)20. 103 104 Micro-C, which relies on cell crosslinking, micrococcal nuclease (MNase) digestion of chromatin, 105 and proximity ligation of DNA ends, can probe nucleosome-nucleosome contacts and chromatin 106 organization on the kilobase scale, and differences in local contacts, such as a zig-zag-like 107 signature in nucleosome contact probabilities, between yeast and mouse embryonic stem cells 108 (mESCs) 18. This is a priori attractive as a potential measure of compaction, but uncertainties 109 about artifacts caused by sequence and particularly by the accessibility bias of MNase cleavage 110 make it difficult to determine whether the differential signal observed is due to compaction or 111 accessibility. 112 113 This uncertainty motivated our use of an orthogonal method to Micro-C in order to validate 114

and gain more sensitivity to local changes in nucleosome contacts. RICC-seq20 relies on 115 spatial clusters of DNA damage events, within a few nanometers of each other, that produce 116 characteristic single-stranded DNA fragment lengths in irradiated cells. The peaks in the 117 fragment length distribution (FLD) reflect the lengths of frequently occurring DNA loops 118 spanning self-contact points that are simultaneously cleaved within a diameter of ~8 nm. The 119 primary peaks observed in RICC-seq FLDs from human fibroblasts correspond to a single DNA 120 wrap around a nucleosome (~78 nt), a full nucleosome unit (~180 nt), and contacts between the 121 DNA gyres of stacked alternating nucleosomes (~270 nt and ~360 nt). Using chromatin fiber 122 simulations, we explored how the locations and strengths of these peaks vary with chromatin 123 fiber geometry, indicating that RICC-seq FLDs have the potential to be sensitive to nucleosome 124 spacing, nucleosomal DNA wrapping (which alters DNA entry/exit angles), and the strength of 125 attractive interactions between nucleosomes. This indicated to us that RICC-seq should be able 126 to detect the effects of linker histone H1 on local chromatin compaction, beyond what is already 127 known about its effects on nucleosome spacing and linker lengths. 128 129 Before applying RICC-seq to this problem, we had to overcome its limitations: the protocol was 130 long, requiring more than a week to complete, did not compensate for sample-to-sample 131 variations in DNA fragment length capture bias to enable quantitative comparisons between 132 different samples, and exhibited significant sequence bias in the final libraries. Here, we develop 133 an optimized RICC-seq 2.0 protocol that solves these challenges, and apply it, together with 134 Micro-C, to measure DNA-DNA contacts on the sub-kilobase length scale in cells with varying 135 levels of H1 linker histones. We find that linker histone has a dramatic genome-wide effect on 136 kilobase-scale chromatin compaction, and that changes in accessibility and transcription are 137 concentrated in regions silenced by the polycomb repressive complexes (PRC1/2). 138 139 140 141 142 143 144 145 146 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 5

147 148 RICC-seq 2.0 improves robustness, reduces sequence bias, and allows cross-sample 149 comparison 150 151 In order to use RICC-seq to assess the chromatin fiber compaction effects of linker histone 152 across cell types and perturbation conditions, we improved on the original RICC-seq method in 153 several ways. 154 155 First, we addressed a challenge we encountered when scaling up the RICC-seq protocol to 156 larger numbers of samples and conditions: library preparations would sometimes fail, yielding 157 no peaks in the fragment length distribution. The original protocol used a high heat denaturation 158 step to dissociate radiation-cleaved single-stranded DNA (ssDNA) fragments from the higher 159 molecular weight genomic DNA prior to elution. We found that this heat-elution method caused 160 the appearance of a large number of additional ssDNA breaks at heat-labile sites throughout the 161 genome, overwhelming the DNA cleavage signal from the original radiation-induced breaks. 162 These heat-labile sites have been previously documented as a product of DNA irradiation34. 163 Small variations in the precise timing of heat denaturation would cause more or fewer of these 164 breaks, leading to a lack of robustness in the RICC-seq protocol. Replacement of heat 165 denaturation with high-pH (NaOH incubation) denaturation and avoidance of high heat (above 166 65˚C) in subsequent library processing was sufficient to generate a more robust ssDNA elution 167 and library preparation (Figure 1a-b). 168 169 Second, we addressed another challenge of using RICC-seq—the length and complexity of the 170 protocol—which was partly due to the necessity for end-repair in agarose plugs and multiple 171 gel-based size selections and amplifications to strike a balance between maintaining as much of 172 the insert size distribution as possible while removing dimers of ligated sequencing adapters. To 173 streamline sequencing adapter ligation to the eluted ssDNA, we used the Single Reaction 174 Single-stranded LibrarY (SRSLY) protocol developed for ancient DNA and cell-free DNA 175 sequencing, which uses single-strand binding protein (SSB) to stabilize and blocked adapters 176 with random-heptamer splint overhangs to capture the end-repaired ssDNA fragments35. This 177 allowed us to proceed from ligation to PCR without the need for size selection (Figure 1a). 178 179 Together, these changes produced a more robust RICC-seq 2.0 protocol that can capture 180 ssDNA fragments from irradiated cells across a broader range of GC contents and fragment 181 lengths (Figure 1c). In particular, RICC-seq 2.0 demonstrates an improved efficiency of capture 182 for fragments that are both long and GC-rich (Figure 1d). 183 184 Third, because RICC-seq can be sensitive to the length bias introduced by sample handling and 185 PCR, we developed a spike-in and normalization strategy to account for such sample-specific 186 biases and allow us to quantitatively compare samples across experiments, cell types and 187 perturbations. To create a “standard candle” library, we digested Schizosaccharomyces pombe 188 chromatin with MNase into a nucleosome ladder, quantified its FLD using capillary 189 electrophoresis prior to library preparation. Known quantities of this S. pombe spike-in were 190 then added to RICC-seq libraries prior to SRSLY adapter ligation and library preparation (Figure 191 1a). Spike-in reads were computationally isolated to calculate their own FLD and a length bias 192 correction factor was calculated by fitting an exponential to the ratio of the post-sequencing FLD 193 and pre-sequencing FLD (Figure 1e). The absolute amount of spike-in was used to scale FLDs 194 for comparisons between no-irradiation controls, irradiated genomic DNA, and irradiated cell 195 samples (Figure 1f). The length-dependent correction factor (Figure 1e) was then used to 196 correct length bias in RICC-seq FLDs (Figure 1f). Lastly, to enhance the contrast of peaks in the 197 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 6 FLD over the background of DNA fragments caused by random, uncorrelated breaks, we 198 calculated the ratio of the scaled and corrected FLD from irradiated cells to the scaled and 199 corrected FLD from irradiated genomic DNA from the same cell sample (maintained in 0.5M Tris 200 pH 8.0 as a quencher for radiation-induced radicals, approximating intracellular quenching) 201 (Figure 1f). These procedures allowed us to directly compare replicates and different 202 experimental samples that may have been subject to different length biases (Figure 1g). 203 204 RICC-seq 2.0 is sensitive to chromatin compaction differences across species 205 206 The combination of a more robust protocol and spike-in normalization allowed us to apply RICC-207 seq 2.0 (Figure 1) to a broad range of cell samples, including budding yeast (Figure 2a). 208 Budding yeast does not express a canonical member of the H1 linker histone family, but 209 expresses Hho1p, which is homologous to the H5 linker histone found in chicken erythrocytes 210 and binds nucleosomes 36. However, its expression level is much lower than mammalian linker 211 histones: a ratio of 0.3 molecules per nucleosome 29,36. Budding yeast therefore has short inter-212 nucleosome linkers and a short NRL, and a largely open chromatin conformation with little clear 213 distinction between euchromatin and heterochromatin as is found in metazoans. At the other 214 extreme, the linker histone to nucleosome ratio rises even higher to ~1.3 in the transcriptionally 215 inactive nuclei of chicken erythrocytes, in which the primary histone variant is H529. Chicken 216 erythrocytes have been used as a model system for highly compacted chromatin fibers, as they 217 represent one of the few cell types in which electron microscopy reveals structures resembling 218 30-nm diameter fibers, albeit more disordered ones than reconstituted in vitro 5,9,37. 219 220 Seeking to understand the dynamic range of RICC-seq 2.0 FLDs as a function of varying 221 compaction and linker histone levels, we applied it to four sample types: S. cerevisiae, human 222 BJ-5ta fibroblasts, naïve mouse B cells, and chicken erythrocytes (Figure 2a-h). After correcting 223 for length bias and calculating the ratio of the irradiated cell sample FLDs to their corresponding 224 irradiated genomic DNA FLDs, we compared them directly (Figure 2i). We found two main 225 effects. First, moving from budding yeast, with an average NRL of 163 bp, to BJ-5ta human 226 fibroblasts, with an average NRL of 186 38, mouse B cells with a NRL estimated to be ~192 227 (based on human lymphoblastoid cells28), to chicken erythrocytes, with an average NRL of 212 228 bp 29, we observed a shift in the location of the higher-order contact (third and fourth) peaks of 229 the RICC-seq FLD toward longer fragment lengths, consistent with the increase in FLD. 230 Importantly, we also observed that the inter-nucleosome contact peaks were more prominent 231 compared to the sub-nucleosomal (first) and mono-nucleosome (second) peak in cell types with 232 more compact chromatin, such as human fibroblasts and to a greater extent, mouse B-cells. 233 Chicken erythrocytes had the most extreme example, with a high fourth peak at ~400 nt. 234 235 The correlation between the linker histone level and the inter-nucleosome stacking signal, which 236 we interpret as a measure of local chromatin fiber compaction differences between cell types, 237 motivated us to perturb the linker histone level in a well-characterized system to more precisely 238 analyze its effects, context dependence and functional consequences. 239 240 Linker histone depletion by CRISPRi leads to genome-wide reduction of nucleosome 241 repeat length and loss of zig-zag alternating nucleosome contacts 242 243 Using a doxycycline-inducible dCas9 K562 cell line, we designed CRISPRi guides against the 244 four H1 subtypes that are the most abundantly expressed in K562 cells: H1.2, H1.3, H1.4 an 245 H1.5, as well as scrambled controls (Figure 3a). The guide RNAs were stably transfected. 246 dCas9 induction for five days led to a reduction in the H1:nucleosome ratio from ~0.75 to ~0.2, 247 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 7 as quantified by HPLC (Figure 3b-c). Over the five-day timeline of the experiment, cell doubling 248 time was not qualitatively different, indicating maintenance of viability (Figure 3d). 249 250 We applied Micro-C (Figure 3e-f) to cells with CRISPRi-depleted H1 (H1-low) and control cells 251 expressing scrambled CRISPR guides (scr-CTRL) at the 5-day time point. Due to potential 252 differences in global accessibility upon reduction of linker histone and the sensitivity of Micro-C, 253 as with other MNase-based assays, to the precise MNase concentration, we titrated and 254 optimized the MNase concentration for each condition independently until similar chromatin 255 digestion profiles were obtained, as assayed by capillary electrophoresis. 256 257 Analysis of the short-range (< 1.5 kb) Micro-C contact probability curves (Figure 3f) revealed 258 that both scr-CTRL cells and H1-low cells exhibit a series of peaks corresponding to contacts 259 between integer nucleosome steps proceeding down the fiber (N+1, N+2, …), with two main 260 differences between the curves (Figure 3g-h). Most prominent is a shift in peak location 261 corresponding to a drop in the NRL upon H1 depletion (Figure 3g,i). However, a second, more 262 subtle but significant effect is a difference in the relative heights of the contact frequency peaks. 263 While scr-CTRL cells exhibit a staircase-like pattern in which pairs of contact peaks are of 264 similar height (N+2 and N+3, N+4 and N+5, …), this pattern was subdued and the peak heights 265 approached a smoothly decreasing function in the H1-low cells (Figure 3h). We quantified this 266 through the ratio of odd and even nucleosome contact probability peaks and found the effect to 267 be significant across our biological replicates (Figure 3j). 268 269 Although matching the global MNase digestion profiles between samples should mitigate some 270 of the accessibility biases of Micro-C on a genome-wide scale, concerns that the Micro-C results 271 may not fully reflect local folding of the chromatin fiber nevertheless remain. To validate that our 272 observed change in not only NRL but also nucleosome contact (and hence chromatin fiber 273 folding) patterns are not an artifact of differential MNase digestion, we applied the RICC-seq 2.0 274 protocol to the same cells, using a S. pombe spike-in to normalize fragment length histograms 275 between samples. RICC-seq does not rely on enzymatic digestion and should therefore not be 276 influenced in the same way by changes in the accessibility to proteins. Its cleavage events are 277 mediated by ionizing radiation that penetrates the whole nucleus and by highly diffusible 278 species—primarily, hydroxyl radicals20. In genome-wide analysis, we found that the RICC-seq 279

corroborated our findings from Micro-C (Figure 3k-p). The inter-nucleosome peaks shift 280 to lower fragment lengths in the H1-low RICC-seq FLD, indicating a lower NRL (Figure 3k), and 281 the strength of the fourth peak, which was most strongly correlated with chromatin compaction 282 and linker histone levels in our cross-species comparison (Figure 2), dropped significantly 283 (Figure 3k-l). Smaller significant changes were also present in the second (mono-nucleosome) 284 RICC-seq FLD peak (Figure 3k-l), but we do not draw a strong conclusion from this segment of 285 the FLD because it exhibited more variability between biological replicates. 286 287 We then sought to determine to what extent these effects on the short-range chromatin fiber 288 compaction evident in the RICC-seq data were a direct result of linker histone depletion, as 289 opposed to indirect effects, such as from cell stress responses. We performed a washout 290 experiment in which the H1-low and scr-CTRL cells were depleted of H1 for five days, as before, 291 and then cultured in doxycycline-free media for five more days to allow for H1 levels to return. 292 We found that the strength and location of the fourth peak returned upon dCas9 washout 293 (Figure 3m-p). Overall, this led us to conclude that linker histone H1 has a direct effect on short-294 range stacking between alternating nucleosomes—nucleosome N to N+2 zig-zag contacts—in 295 the context of intact chromatin. 296 297 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 8 Short-range zig-zag stacking contrast between euchromatin and heterochromatin 298 depends on linker histone levels 299 300 Next, we asked how the dependence of nucleosomal zig-zag contacts depend on the local 301 epigenetic context. Segmenting the Micro-C contacts by overlap with histone mark ChIP-seq 302 peaks—H3K27 acetylation to mark active promoters and enhancers, and H3K27 trimethylation 303 and H3K9 trimethylation to mark the two primary types of heterochromatin—we found that there 304 was a subtle change in the zig-zag signal of the first eight Micro-C contact peaks (Figure 4a-b). 305 Quantitating the zig-zag signature using the odd-even peak height ratio, we found that there 306 were differences in compaction between the three chromatin states, with H3K9me3 chromatin 307 having the strongest zig-zag and H3K27me3 the weakest (Figure 4c). H1 depletion reduced the 308 zig-zag signature such that the H3K27me3 heterochromatin in H1-low cells had a similar level to 309 H3K27 acetylated chromatin in scr-CTRL cells (Figure 4c). 310 311 We cautiously interpret the zig-zag signature in short-range Micro-C contact data on a global 312 level as a measure of short-range chromatin compaction because differences in the propensity 313 for cleavage by MNase can be controlled at a global level by tuning the MNase concentration. 314 However, artifacts caused by differential digestion by MNase cannot be mitigated if they occur 315 between different sets of genomic loci within the same sample, as would be expected for 316 heterochromatic versus euchromatic loci. Indeed, we observed that in libraries with different 317 extents of digestion, as measured by the effective fragment size of the mononucleosome peak 318 (Figure 4d), the zig-zag signature depended on the amount of digestion, with an inverse 319 correlation between the strength of the zig-zag signature and the size of the mononucleosome 320 fragment (Figure 4d-e). We therefore concluded that Micro-C is not a reliable measure of 321 differences true chromatin compaction within the same sample, and validation of results by an 322 orthogonal method is needed. 323 324 We analyzed our RICC-seq 2.0 data segmented by epigenetic state in the same mode, in order 325 to determine whether the patterns of zig-zag contacts suggested by the Micro-C data could be 326 orthogonally validated by a non-enzymatic method (Figure 4f-k). We monitored the irradiated 327 genomic DNA (gDNA) control from both scr-CTRL and H1-low cells to ensure that the peak 328 changes we observed were not driven by pre-existing DNA damage that could be differential 329 between genomic loci (Figure 4h,i). Although we observed some weak peaks in the H1-low 330 gDNA control consistent with small amounts of contaminating DNA fragments with damage 331 between nucleosomes, they were not correlated with the irradiated cell peaks in a way that 332 would explain the observed differences. To normalize against differences in the gDNA, we 333 calculated the ratio between the length bias-corrected, epigenetic state-specific RICC-seq cell 334 FLDs to the similarly corrected gDNA FLDs (Figure 4j-k). The results we observed partially 335 agree with Micro-C data. Interpreting the strength of the fourth peak (~330-420 nt) as a measure 336 of the population-averaged stacking of alternating nucleosomes by zig-zag chromatin fiber 337 compaction, we found that acetylated chromatin indeed does have very low compaction. 338 However, the level of compaction between H3K27me3 heterochromatin and H3K9me3 339 heterochromatin appears similar by RICC-seq, as opposed to the higher H3K9me3 compaction 340 suggested by Micro-C. The location of the fourth peak is shifted (Figure 4j, arrow) between 341 H3K9me3 and H3K27me3 chromatin, consistent with the difference in NRL observed by Micro-342 C. 343 344 The difference in local chromatin fiber compaction between heterochromatic regions and 345 acetylated euchromatic regions is consistent with the removal of linker histone H1 from 346 acetylated chromatin causing unfolding of the fiber 10,39. We therefore compared the epigenetic 347 state compaction landscape between scr-CTRL cells and the H1-low cells. We found that most 348 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 9 of the difference in chromatin fiber compaction between epigenetic states in the scr-CTRL 349 landscape is gone in H1-low cells (Figure 4g,k). 350 351 Changes to long-range chromatin compartments, domains, and loops are minimal after 352 five-day H1 depletion 353 354 The dramatic loss of local chromatin fiber compaction upon H1 depletion observed with short-355 range Micro-C curves and RICC-seq 2.0 motivated us to ask how this short-range 356 decompaction relates to long-range chromosome folding features. We performed Hi-C to obtain 357 a sensitive measure of long-range compartment changes. The large-scale (1 Mb resolution) 358 balanced contact matrices appeared similar between scr-CTRL and H1-low cells (Figure 5a) 359 and indeed, HiC-Rep analysis at 500 kb resolution showed that the difference between 360 conditions was comparable to the difference between replicates (Figure 5b). 361 362 We then analyzed compartment changes between scr-CTRL and H1-low using the compartment 363 score (c-score). We found that, in contrast to what was observed using the same analysis for H1 364 depletion by conditional triple knockout in mouse T-cells6, the changes in c-score in K562 cells 365 depleted of H1 by five days of CRISPRi were subtle, and only weakly weighted toward B-to-A 366 transitions (Figure 5c). 367 368 Visual analysis of chromatin domains shows little change with H1 depletion (Figure 5d), and 369 calls of domain boundaries location (Figure 5e) and strength (Figure 5f) showed that there are 370 no substantial domain changes on the global scale. Calling chromatin loops showed a general 371 loss of loops with H1 depletion (Figure 5i), though the low specificity of loop calling suggests this 372 may in fact reflect an overall weaking of loop strength (Figure 5h). This small loss of loop 373 strength affects both CTCF and non-CTCF loops(Figure 5j). 374 375 Transcriptional de-repression upon H1 depletion preferentially occurs in polycomb 376 repressive complex target genes 377 378 Considering the relatively subtle changes in long-range genome organization, we next 379 wondered about the effects of global de-compaction of chromatin and the loss of compaction 380 contrast between epigenetic states on functional outcomes like transcriptional regulation, and its 381 associated features such as DNA accessibility and histone modifications. 382 383 We performed poly(A)-capture RNA-seq on scr-CTRL and H1-low cells at five days of H1 384 depletion to compare against our chromatin compaction results. We found that that the vast 385 majority of changing genes were up-regulated in their transcription (1525 significantly 386 upregulated and 32 downregulated with p 1) (Figure 6a). To 387 determine which regulators may be responsible for the changes in gene expression, we 388 performed ChEP-MS to identify changes in protein abundance on chromatin (Figure 6b). We 389 found that the transcription factor GATA1, which is highly expressed in K562 cells, dramatically 390 increased its association with chromatin in H1-low cells, while the BAF complex component 391 SMARCC2, the chromatin-binding nucleoporin NUP153, the H3K4-targeting histone 392 demethylase KDM1B and its methyltransferase KMT2A, the neuron-specific transcription factor 393 TBR1, the polycomb repressive complex2 (PRC2) member SUZ12, and the repressive CBX1 394 (HP1-beta) protein were decreased in their association. We next used ENRICHR to determine 395 the upstream regulators most likely to explain the change in gene expression (Figure 6c) and 396 followed up with top hits GSEA analyses (Figure 6d). We found that the most significantly 397 upregulated gene sets upon H1 depletion are those regulated by the PRC2 complex member 398 SUZ12 and PRC1 complex members CBX8 and CBX2, which are respectively involved in 399 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 10 depositing and sensing the H3K27me3 histone mark. Surprisingly, although the association of 400 GATA1 with chromatin was highly significant, GATA1 targets were not enriched in the 401 upregulated gene set (Figure 6c,e). 402 403 Next, we investigated how chromatin accessibility responds to loss of H1 using ATAC-seq at 5 404 days of CRISPRi. We found that as with transcription, differential accessibility is biased toward 405 gains across thousands of ATAC-seq peaks, which are more enriched in distal intergenic 406 regions relative to peaks with stable accessibility (Figure 6f,g). 407 408 The increase in accessibility and the de-repression of PRC1/2 target genes led us to 409 hypothesize that loss of H3K27me3 upon H1 depletion may explain the observed increase in 410 transcription. CUT&Tag for H3K27me3 showed widespread changes, with regions changing in 411 both directions but dominated by a loss of H3K27me3 (Figure 6h,i). To tie accessibility changes 412 to epigenetic state, we then asked where the newly accessible sites fell, relative to the existing 413 epigenetic context. We then quantified changes in accessibility in regions that lost H3K27me3 414 compared to those where the signal remained unchanged, as well as in other regions marked 415 by several additional epigenetic marks (Roadmap Epigenomics 40) (Figure 6k). Consistent with 416 the gene regulation results, we saw that the regions with net increases in accessibility are 417 heterochromatic—those marked by H3K27me3 in scr-CTRL or parental K562 cells and those 418 marked by H3K9me3 in parental K562 cells 40 (Figure 6k). However, the fold-change of 419 accessibility in regions losing H3K27me3 was not higher when compared to all H3K27me3 420 regions. Similarly, when we investigated the change in H2K27me3 between control and H1-low 421 cells specifically focusing on genes that increased in transcription, we found that the local 422 H3K27me3 landscape stayed at a similar level (Figure 6l). What was notable, was that the 423 genes that were upregulated upon H1 loss had a much higher level of H3K27me3 signal near 424 their promoters than genes that were not de-repressed, regardless of H1 depletion (Figure 6l). 425 426 Together, these results suggest that gene de-repression and the gain of accessibility does not 427 require complete local loss of promoter proximal H3K27me3 and that the mechanism of de-428 repression is not simply a direct consequence of local H3K27me3 loss. 429 430 431 Chromatin de-compaction by H1 depletion is genome-wide, except for regions that were 432 already de-compacted and accessible 433 434 We next looked at the regions that gain accessibility in H1-low versus scr-CTRL–spanning 435 promoter proximal and distal sites. We found that H3K27me3 signal flanking these peaks of 436 accessibility is largely maintained in H1-low (Figure 7a), indicating that accessibility gains do not 437 generally require local depletion of H3K27me3 at these regulatory elements. The local 438 difference in the CUT&Tag signal observed is likely to be driven by the change in accessibility, 439 as a sharp CUT&Tag peak at the center of the newly opened ATAC-seq peaks (Figure 7a). 440 441 We then turned to RICC-seq as a measure of chromatin compaction to determine whether there 442 are focal changes in chromatin compaction at regions where DNA accessibility or H3K27 443 trimethylation are changing. We did not observe any change in chromatin compaction by RICC-444 seq between genomic regions that lose K27me3 and those that do not (Figure 7b). The primary 445 difference remains between the scr-CTRL and the H1-low sample at all H3K27me3-marked 446 sites, regardless of their change in the histone mark between the two conditions. 447 448 For genomic regions that change accessibility, on the other hand, we did observe changes in 449 compaction (Figure 7c-e). We found a distinction in the average RICC-seq FLD between 450 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 11 unchanging ATAC-seq peaks (Figure 7c) and those that become accessible upon H1 depletion 451 (Figure 7d). Unchanging ATAC-seq peaks are already quite decompacted, with FLDs similar to 452 genome-wide acetylated chromatin (Figure 7e) and resembling the decompacted chromatin of 453 budding yeast (Figure 2). The regions that gain accessibility, however, began with a FLD very 454 similar to the genome-wide average and decompacted to a FLD comparable to the genome-455 wide H1-low FLD upon H1 depletion (Figure 7e). 456 457

458 459 We set out to understand the relationship between chromatin structure, transcriptional 460 regulation, DNA accessibility and histone marks. By improving the RICC-seq protocol to RICC-461 seq 2.0, we obtained a protocol that could be reliably applied to a variety of sample types with 462 varying levels of chromatin compaction. We verified that cell types with very different NRLs and 463 global levels of chromatin compaction, spanning de-compacted budding yeast cells through 464 mammalian cell types and hyper-compacted, transcriptionally inactive chicken erythrocytes, 465 produced different RICC-seq FLDs, demonstrating that the method is sensitive to changes in 466 chromatin compaction. 467 468 We were particularly motivated to make direct in situ measurements of chromatin compaction in 469 a model of linker histone depletion because linker histone H1 has so often been invoked as an 470 architectural protein that uses chromatin compaction as its mechanism for broad-based 471 transcriptional repression. Our results show that a dramatic reduction in total H1 levels leads to 472 not only an increase in accessibility at thousands of sites and the upregulation of thousands of 473 genes, but it also causes chromatin decompaction at the tri-nucleosome length scale, which we 474 could measure using two orthogonal methods—Micro-C and RICC-seq 2.0. We did not see 475 strong changes in long-range chromatin organization over the same time scale, suggesting that 476 that chromatin structure at this scale is not directly coupled to H1 density and transcriptional 477 regulation. This underscores the importance of maintenance of chromatin compaction by linker 478 histone in regulating both the accessibility of many sites across the genome and the 479 transcriptional repression of a large set of genes. We also observed a modest but broad-based 480 loss in H3K27me3 signal as measured by CUT&Tag, indicating that in this system, linker histone 481 plays a role in the maintenance of the H3K27me3 mark, as has been observed in other 482 systems, including T-cells6 and B-cells41, as well as in K562 cells in which H1 is depleted via 483 CRAMP1 knockout42. 484 485 Surprisingly, the changes in chromatin compaction upon linker histone depletion are remarkably 486 uniform across the genome. Although DNA accessibility is particularly enriched in H3K27me3-487 decorated regions of the genome upon H1 depletion, this is not accompanied by specific 488 decompaction at H3K27me3 regions, any more than is happening in the rest of the genome. 489 RICC-seq is, however, sensitive to changes in compaction elsewhere. We observed a difference 490 in the FLD shift between regions that maintained accessibility and those that gained it—those 491 with pre-existing accessibility had a more de-compacted FLD in control cells, and experienced 492 only a modest change in the FLD and in the compaction contact peak upon H1 depletion. 493 494 Our results support a model in which linker histone H1 is not locally inducing compaction at 495 particular loci, but is rather working genome-wide to compact most chromatin, with the 496 exception of acetylated regions where it is removed and chromatin can de-compact10,39. This is 497 consistent with FRAP data showing that in vivo, linker histones are remarkably dynamic43, which 498 would permit their broad distribution across the genome, and with electron microscopy and 499 super-resolution microscopy, which show a broad change in both chromatin density and 500 nucleosome clutch size 6,21. There may be variability in H1 density between chromatin types, 501 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 12 and chromatin compaction at the tri-nucleosome scale may have different sensitivity to the linker 502 histone level as compared to other architectural or functional features. Indeed, some threshold 503 effects in chromatin fiber architecture have been observed with linker histone density changes in 504 silico44. 505 506 Overall, we find that H1 acts to modulate the global nucleosome repeat length and local 507 compaction of chromatin, but that some chromatin states may be more dependent on this 508 compaction than others. PRC1/2 repression of both accessibility and gene expression is 509 particularly sensitive. The exact nature of this sensitivity may be a combination of H1’s effects 510 on both chromatin compaction and NRL. In vitro experiments show that PRC2 deposition of 511 H3K27me3 preferentially occurs on long-NRL chromatin45, but compaction also promotes 512 deposition of H3K27me3 and prevents deposition of its antagonistic mark H3K36me2 6. 513 514 Our results highlight that there is a close regulatory relationship between H1-dependent sub-515 kilobase chromatin compaction, DNA accessibility, histone marks and transcriptional 516 regulation—and that it is more immediate than the long-range compartmentalization of the 517 nucleus. In a system with as much complexity and redundancy as chromatin, directly measuring 518 chromatin compaction as a distinct physical variable, rather than inferring it from other methods, 519 can help define more precise mechanistic models of transcriptional repression. 520 521 Disclosures 522 523 V.I.R. is a co-inventor on a patent application covering a chromatin conformation capture 524 method. 525 526 Acknowledgments 527 528 We would like to thank the Risca Lab, and Skoultchi Lab, and the members of the Rockefeller 529 Chromatin Supergroup, as well as Ari Melnick, Ethel Cesarman, and Yael David for helpful 530 discussions. We also thank the support of the Rockefeller University Genomics Resource 531 Center and High Performance Computing Resource Center. This work was supported by a NIH 532 New Innovator Award to V.I.R. (DP2GM150021), a Rita Allen Foundation Scholar Award to 533 V.I.R., a Hirschl/Weill-Caulier Career Scientist Award to V.I.R., a NSERC post graduate 534 scholarship award to H.C., and a HFSP Postdoctoral Fellowship to A.O. H.D. P. and A.I.S. were 535 supported by NIH grant R01GM147165 and D.V.F. by NIH grant R01HD114814. 536 537 538 539 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 13 Figure Legends 540 541 Figure 1. Improved RICC-seq 2.0 protocol reduces sequence bias, improves recovery of 542 long high-GC fragments and allows quantitative comparison between samples. 543 544 a) Schematic of the new RICC-seq protocol, incorporating the SRSLY ssDNA library 545 preparation, including spike-in. 546 b) Schematic depicting fragments produced by RICC-seq and the nucleosome-nucleosome 547 contacts that generate the corresponding peak distributions. 548 c) Fragment-length distribution plot over increasing genome %GC with fragments mapping to 549 Roadmap Epigenomics 40 H3K27me3, H3K9me3, and H3K27ac peaks shown for RICC-seq 1.0 550 and 2.0 methods. 551 d) Contour plot of fragment lengths captured and %GC of representative BJ sample from RICC-552 seq 1.0 and RICC-seq 2.0. 553 e) MNase-digested fission yeast chromatin ladder is shown before sequencing as a Capillary 554 electrophoresis (TapeStation) trace and after sequencing as a Fragment Length Distribution 555 (FLD), both smoothed (5 nt rolling average) and aligned by the falloff of the mononucleosome 556 peak. Ratio of post- to pre-sequencing distributions shown with exponential curve fit after 300bp 557 which is extrapolated back and used to correct samples within an experiment. 558 f) Representative BJ-5ta fibroblast sample with the respective corrections applied 559 g) Multiple BJ fibroblast samples shown with the respective corrections applied n=2 biological 560 replicates shown each with n=2 technical replicates. 561 562 563 Figure 2. RICC-seq 2.0 applied across organisms with increasing H1 levels and 564 chromatin compaction reveals an increase in nucleosome stacking contacts. 565 566 a,c,g,e) Representative spike-in–scaled FLDs for each species before length-bias correction 567 b,d,h,f) Replicate FLDs after length-bias correction with 95% CIs, max-normalized to the 568 mononucleosome signal; condition means shown with 95% CI 569 i) Per-organism corrected replicate FLDs averages shown. Chicken n=2 technical replicates, 570 Mouse B cell n=2 technical replicates, BJ n=3 technical replicates of 2 biological replicates, 571 yeast n=3 technical replicates of two biological replicates. 572 573 Figure 3. H1 depletion shortens nucleosome repeat length and reduces contacts 574 associated with nucleosome-nucleosome stacking interactions, which reemerge upon 575 wash-out. 576 577 a) Experimental design for generating H1-low and scrambled control (scr-CTRL) cells, with five 578 days doxycycline (dox) induction and matched 5 days dox-washout (rescue) conditions for 579 induction of dCas9 in cells with constitutively expressed, stably transfected CRISPRi guides 580 RNAs targeting H1.2, H1.3, H1.4, and H1.5 (H1-low) or scrambled guides (scr-CTRL). 581 b) HPLC of H1 subtypes in H1-low cells relative to scr-CTRL. 582 c) Quantification of HPLC of H1:H2B ratio depletion on day 5 of dox induction. 583 d) Doubling time shown for samples before and after induction on dox for scr-CTRL (blue) and 584 H1-low (pink). n=1 biological replicate. 585 e-f) Schematic of Micro-C workflow and resulting short-range contact probability histogram. 586 g) Micro-C contact frequency curve and h) maximum contacts for scr-CTRL and H1-low 587 conditions i) nucleosome repeat length (NRL) quantification n=2, j) odd/even contact frequency 588 peak maximum quantification of n=2 biological replicates shown. 589 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 14 k) RICC-seq fragment length distribution in scr-CTRL and H1-low cells, spike-in–scaled, depth-590 matched within biological replicate, normalized to the 180 nt peak maximum, interpolated to a 591 common 55–450 bp grid, and smoothed (10 nt rolling average). Shaded bands denote 95% CI 592 (t-interval) across the relevant replicates n= 3 biological replicates shown. 593 l) Ratio taken over scr-CTRL and H1-low; pink boxes mark contiguous Welch-significant runs 594 (p<0.05; calculated within 5 nt windows) for the indicated pairwise comparison. 595 m) Condition means across dox washout rescue experiment (scr-CTRL vs H1-low). n=2 H1 low 596 with dox, n=2 H1low dox washout, n= 2 Scr with dox n=1 scr-CTRL dox washout. n: technical 597 replicates. 598 n) Ratio of technical replicate means from (m) ±95% CI for scr-CTRL, scr-CTRL dox-off. 599 o) Ratio of technical replicate means from (m) ±95% CI for H1-low and H1-low dox-off. 600 p) Ratio of technical replicate means from (m) ±95% CI for H1-low dox-off/scr-CTRL dox-off; 601 shaded segments denote Welch test-significant runs (p<0.05; 5 nt windows). 602 603 604 Figure 4: H1 depletion phenocopies compaction structure of active chromatin. 605 606 Nucleosome-nucleosome contacts as measured by Micro-C in epigenetic regions defined by 607 published WT K562 ChIP datasets in a) scr-CTRL b) and H1-low conditions 40. Two biological 608 replicates shown each. 609 c) Ratio of N/N+odd contacts and N/N+even nucleosome contacts in epigenetic state regions. 610 Error bar: standard deviation between biological replicates, n=2. 611 d) Capillary electrophoresis (TapeStation) traces of the fragment size distribution produced by 612 MNase titration with different amounts of enzyme. The estimated fragment length of the 613 mononucleosome peak is indicated. 614 e) Micro-C contact probability curves for the libraries obtained from the MNase titration in (d). 615 f-g) RICC-seq FLDs from irradiated cells and matched gDNA controls in h-i) 616 j-k) scr-CTRL cells and H1-low cells, subset by histone mark 40 gapped peaks. 1 biological 617 replicate shown. 618 619 Figure 5: Long-range chromatin structure is only weakly affected by 5-day H1 depletion 620 621 a) HiC contact maps of scr-CTRL and H1-low cells, ICE corrected by cooltools. Chr 4, 1 Mb 622 resolution, 80-100 M paired end reads per HiC replicate. 623 b) Genome-wide reproducibility analysis within (teal and pink) and between (gray) conditions, 624 using HiC-Rep at 500 kb resolution. 625 c) Compartment scores (c-scores) in matched genomic bins between scr-CTRL and H1-low 626 HiC contact data at 100 kb resolution. C-score is calculated with cscoretool and compartment 627 shifts are defined as |∆c-score| > 0.2, compartment changes where a bin’s c-score changes sign 628 and A-shift or B-shift where bins shift within the same compartment. 629 d) Example contact maps showing domains in scr-CTRL and H1-low Micro-C. Chr 4 zoom, 100 630 kb resolution, 80-100 M paired end reads per HiC replicate 631 e) Example domain boundaries called by cooltools in scr-CTRL and H1-low Micro-C. Chr 1: 32 632 Mb – 34.7 Mb. 633 f) Genome-wide boundary strength distributions for different window sizes for scr-CTRL and 634 H1-low Micro-C. 635 g) Example contact maps showing loops in scr-CTRL and H1-low Micro-C. Chr 4 zoom, 10 kb 636 resolution, 80-100 M paired end reads per HiC replicate. 637 h) Aggregate peak analysis of loops called by HICCUPS centered around loops called only in 638 scr-CTRL, only in H1-low, or in both datasets. 639 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 15 i) Overlap of loops called by juicertools HICCUPS in scramble control and H1-low Micro-C. 640 Overlap is defined as both anchors having 90% overlap. 641 j) Aggregate peak analysis of loops called by HICCUPS centered around CTCF loops, defined 642 as those overlapping RAD21 and SMC ChIP peaks (called by juicertools motif), and non-CTCF 643 loops. APA is site +/- 10 kb with 1 kb bins. 644 645 Figure 6: Transcriptional upregulation and increase in accessible chromatin noted upon 646 H1 depletion while opportunistic TF binding does not drive expression changes. 647 a) ERCC spike-in normalized RNA-seq. 1525 genes are significantly upregulated upon H1 648 depletion with |log2(fold-change)| >1 and padj < 0.05 (Benjamini-Hochberg). H1-5, H1-2, H1-3 649 among the significantly downregulated genes at –1.64, -1.69 and –1.39 log2 fold change 650 respectively. 651 b) CHEP-seq volcano plot of H1-low vs. scr-CTRL proteins associated with chromatin, n=3 652 biological replicates for each condition, significance threshold defined as |log2(fold-change)| >1. 653 c) ENRICHR analysis of transcription factors (TFs) with significantly enriched targets in 654 upregulated genes in H1-low vs. scr-CTRL. TF target gene sets based on ENCODE TF 2015 655 ChIP-seq peak overlap with target TSS in K562 cells. 656 d) Gene set Enrichment analysis (GSEA) of top hits identified by ENRICHR (c) based on 657 Harmonizome ENCODE Transcription Factor Targets database. Normalized enrichment score 658 (NES): CBX8 1.615, CBX2 1.56 and SUZ12 1.17, respectively. 659 e) GSEA of GATA1 (identified by CHEP-seq, (b)) target genes based on H1-low vs. scr-CTRL 660 fold-change based on Harmonizome ENCODE Transcription Factor Targets database. 661 NES:0.859 p=1.00. 662 f) ATAC-seq peaks base mean over fold change with 2678 peaks significantly upregulated in 663 H1-low condition. |log2(fold-change)| >1 and padj. <0.05 (Benjamini-Hochberg). 664 g) ATAC-seq peaks annotated by genomic features. 665 h) M-A plot of H3K27me3 CUT&Tag fold-change at H3K27me3 CUT&Tag peaks. 2453 peaks 666 significantly downregulated in H1 low condition (|log2(fold-change)|>1 and padj. <0.05, Benjamini-667 Hochberg criterion). 668 i) H3K27me3 H1-low upregulated, H1-low downregulated and unchanging peaks annotated by 669 genomic feature. 670 j) H3K27me3 signal over H1-low downregulated and all H3K27me3 peaks. Peak centers 671 shown +/- 8kb with a bin size 100 bp. 672 k) Enrichment of H1-low/scr-CTRL Log2FC ATAC-seq signal over H3K27me3 CUT&Tag peaks 673 compared to published Roadmap Epigenomics ChIP-seq peaks for K562 40 674 l) H3K27me3 signal over upregulated vs unchanging genes measured by RNA-seq. Signal 675 plot over +/-5 kb around TSS of genes with a bin size of 100 bp. 676 677 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 16 Figure 7: Chromatin fiber compaction is unaffected by H3K27me3 depletion but varies 678 with changes in accessibility 679 a) ATAC-seq accessibility signal over Cut&Tag H3K27me3 peaks +/- 5 kb around peak centers 680 shown with a bin size of 100 bp. 681 b) RICC-seq FLD plot over downregulated vs unchanging H3K27me3 peaks in H1-low vs scr-682 CTRL. n= 1 biological replicate shown as ratio over subset genomic DNA control. Signal 683 normalized to mononucleosome peak. 684 c) RICC-seq FLD plot over unchanging ATAC-seq peaks and genome-wide smoothed (30 nt 685 rolling average). n=1 biological replicate shown corrected by biological replicate-specific 686 correction factor and ratio over similarly subset and corrected sample-matched genomic DNA 687 control. Signal normalized to mononucleosome peak. 688 d) RICC-seq FLD plot over upregulated ATAC-seq peaks and genome-wide, as in (c). 689 e) RICC-seq FLD plot over upregulated ATAC-seq peaks, H3K27me3, and H3K27ac, as in (c). 690 691

692 693 Cell culture and preparation 694 695 BJ-5ta Fibroblasts 696 Cells were grown in DMEM supplemented with 10% FBS and 1% Penicillin-Streptomycin. Cells 697 passaged every three days at 1:3 splits when cells are about 75% confluent. To harvest, cells 698 were contact inhibited and trypsinized with 4 ml Trypsin to lift then quench with 8 mL media. 699 Cells were washed and spun down with PBS and 2 million cells per technical replicate 700 harvested per plug. 701 702 Budding yeast 703 For RICC-seq experiments on budding yeast, we used Saccharomyces cerevisiae W303 704 RAD5+ wild-type strains (Gift from Xiaolan Zhao). Overnight starter cultures were diluted into 705 fresh YPD to an initial OD600 of ~0.1 and grown at 30°C with shaking for at least 4 hours to allow 706 cells to enter mid-log growth phase with an OD600 of 0.5–1.0. Cells were harvested in 50 mL 707 conical tubes by centrifugation at ~900 × g for 3 min, washed once in 25 mL PBS, transferred to 708 1.5 mL microcentrifuge tubes, and washed again in 1 mL PBS (30 s at ~900 × g between 709 washes). Pellets were resuspended in PBS and mixed 1:1 with molten low-melting-point (LMP) 710 agarose. Yeast cells were embedded directly in agarose plugs without prior zymolyase 711 treatment, as pilot experiments indicated it was not required for efficient lysis and downstream 712 processing. The cell–agarose suspensions were immediately cast into plug molds and solidified 713 on ice for 10 min. Plugs were released into 2 mL tubes (3 plugs per tube) and irradiated on ice 714 in a 50 mL conical tube with 1000 Gy X-ray ionizing radiation at ~120 Gy/min over the course of 715 8 minutes and 20 seconds, while non-irradiated tubes were kept on ice for the same duration. 716 Following irradiation, plugs were immediately incubated in 950 µL RICC lysis buffer 717 supplemented with 50 µL Proteinase K at 25°C with gentle shaking for 48 h. After cell lysis, the 718 rest of the RICC-seq protocol proceeded as described further below. 719 720 721 Chicken Red blood cell Whole chicken blood was ordered from Pel-Freez Biologicals, Whole 722 Chicken Blood, Non-Sterile with Alsever’s Media, Cat. No. 33133-1. Cell concentration was 723 determined using a hemocytometer corresponding to ~9.8 × 10^8 cells/mL. A 5 mL aliquot was 724 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 17 transferred to a 15 mL conical tube and pelleted at 200 × g for 5 min, the supernatant was 725 removed, and the cells were washed twice in PBS at 200 × g, 5 min for each wash. The pellet 726 was spun again at 200 × g for 5 min and kept on ice, then resuspended in 2.5 mL PBS to 727 generate a suspension at 1.5 × 10^2 cells/mL. Cells were aliquoted at 750 µL into 2 mL tubes, 728 equilibrated at 37°C for 1 min, and mixed 1:1 (v/v) with pre-warmed 2% low–melting point 729 agarose. The solution was pipetted carefully into plug molds, avoiding bubbles and allowed to 730 solidify on ice. Plugs were then transferred into 400 µL cold PBS and either irradiated with 300 731 Gy ionizing radiation on ice or kept on ice as non-irradiated 0 Gy and PLC controls. Following 732 irradiation, plugs were incubated in 1,170 µL lysis buffer supplemented with 30 µL Proteinase K; 733 samples were kept on ice for 2–3 h post-lysis and then transferred to room temperature for 734 overnight incubation. 735 736 K562 H1 depletion and scrambled control cell culture and induction K562 cells expressing a 737 dox-inducible dCas9-KRAB-P2A-mCherry were generated by lentiviral transduction with the 738 TET-ON vector pAAVS1-NDi-CRISPRi (addgene #73497). Transduced cells were selected with 739 200ug/ml G418 and inducible dCas9-KRAB-P2A-mCherry cells were further selected with 740 fluorescence-activated cell sorting (FACS) after 3 days of Doxycycline treatment (1ug/ml). 741 Selected cells were allowed to grow in the absence of doxycycline and then transduced with 742 pU6-sgRNA EF1Alpha-puro-T2A-BFP (addgene #60955) which was engineered with 4 in 743 tandem U6-sgRNA expression cassetes, each expressing a subtype specific H1 sgRNA namely 744 H1.5 (GGCAGGAGCGGTTTCCGACA), H1.2 (GGCTGCCGCCGGCTATGATG), H1.3 745 (GGCTGCCGCCGGCTATGATG) and H1.4 (GGCCAAGCCTAAGGCTAAAA). As control, cells 746 were also transduced with a non targeting pU6-sgRNA EF1Alpha-puro-T2A-BFP 747 (GCACTACCAGAGCTAACTCA). Cells expressing constitutive high levels of sgRNAs were 748 selected by combining puromycin selection (10ug/ml) and FACS to collect the top BFP positive 749 cells. H1 depletion was induced by supplementing culture media with doxycycline 1ug/ml for 5 750 days and H1 content assayed by RP-HPLC of acid extracted histones. 751 752 Cells were grown in IMDM media (Thermo Scientific #12440046) supplemented with 10% heat-753 inactivated FBS (Sigma-Aldrich F4135) and 1% Penicillin-Streptomycin. Cells were passaged 754 every two days, seeding 2.4 M cells in a T150 flask. Starting five days before experimental 755 harvest, doxycycline was added to the growth media of both H1-low and scr-CTRL cells to a 756 concentration of 3 mg/mL. Cells were grown in the doxycycline media for five days, following the 757 regular splitting schedule, then harvested for experiments on day five. 2 million cells per 758 technical replicate harvested. 759 760 Naïve B cell isolation 761 Spleens from wild-type C57BL/6J mice (Jackson Laboratories, strain 000664) were 762 mechanically dissociated and passed through a 40-µm strainer. Red blood cells were lysed 763 using ACK buffer (Lonza). Resting B cells were then enriched by negative selection using anti-764 CD43 (Ly48) magnetic microbeads (MACS, Miltenyi Biotech), according to the manufacturer’s 765 instructions. Briefly, the cell suspension was incubated with 30 µl of CD43 magnetic beads 766 diluted in 270 µl of PBS per spleen for 20 mins at 4 °C. The mixture was then applied to an LS 767 MACS Column on MACS Separator. The flow-through containing naïve B cells was collected 768 and resuspended in PBS supplemented with 0.5% bovine serum albumin (BSA) and 2 mM 769 EDTA. 770 771 772 773 774 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 18 RICC-seq 2.0 library preparation 775 776 RICC-seq 2.0 was performed on multiple different cell types each with their own growing 777 techniques and harvest conditions as outlined above. Adherent cells were grown to ~100% 778 confluence and rested ~1 day to enrich G0. Cells were were washed in warm PBS (Gibco 779 14190-144), detached with 0.05% Trypsin-EDTA or Accutase at 37 °C (Invitrogen 25300-054), 780 counted with trypan blue (Gibco 15-250-061), pooled, and pelleted (200–300 RCF, 4 min). 781 Pellets were resuspended in PBS to 50 M cells/mL and embedded to final concentration 1% 782 low-melt agarose (Sigma Type VII-A, A0701-25G) kept at 37 °C, using Bio-Rad plug molds 783 (1703713). 2 million cells per plug were used for the K562 cells harvested 5 days post 784 doxycycline induction. 785 786 Plugs were irradiated to a total dose of 300 Gy for most conditions except 1000 Gy for yeast 787 and lysed overnight (24-48 h, RT/20 °C with shaking) in RICC lysis buffer containing 20% N-788 lauroylsarcosine (Sigma L7414-50ML), Proteinase K (NEB P8107), and EDTA (Thermo 789 15575020), then washed for ~5 h as follows: TE + 1 mM PMSF, 30 min at 4 °C; TE + 1 mM 790 PMSF, 45 min at 4 °C; TE, 60 min RT; TE, 60 min RT; TE + RNase A 0.1 mg/mL (Sigma R4642-791 250MG), 45 min at 37 °C; TE + RNase A 0.1 mg/mL, 45 min at 37 °C; TE, 60 min RT. 792 793 Genomic DNA control plugs were equilibrated in 0.5 M Tris-HCl pH 8 on ice (three 15-min 794 exchanges, then +400 µL Tris) and irradiated identically, followed by the same washes. ssDNA 795 was eluted by incubating plugs in 0.1 N NaOH (200 µL, 15 min), neutralized with 1 M Tris-HCl 796 pH 7.5 (100 µL) + 2 mM EDTA (RT, ~4 h), and purified with the Zymo RNA Clean & 797 Concentrator-5 kit (R1016) using modified speeds (binds at 3,800 RCF; washes at 10,000 798 RCF); columns were loaded with 600 µL RNA Binding Buffer then 900 µL absolute ethanol and 799 eluted twice in 10 µL 10 mM Tris pH 8 (total ~18 µL). K562 eluate samples were purified with 800 custom-made carboxyl-coated beads to facilitate batch processing46. 801 802 Spike-ins were included post elution into each sample before library preparation. A genomic 803 yeast DNA ladder digested with MNase was included for cross-experiment comparability. 804 805 Ends were dephosphorylated with rSAP (NEB M0371) in CutSmart buffer (21 µL total; 37 °C, 1 806 h), then 5′-phosphorylated with T4 PNK (Enzymatics Y9040L) in the presence of DTT (125 mM 807 stock to 5 mM final; Sigma 43815-1G) and ATP (500 µM stock to 5 µM final) in a 25 µL reaction 808 (37 °C, 1 h); MgCl₂ (Invitrogen AM9530G) and spermidine (Sigma 85558-1G). 809 810 SRSLY ligation was done with SRSLY P5/P7 adapters, T4 DNA ligase (Enzymatics L6030-LC-L) 811 with 1× ligase buffer and 18.5% PEG-8000 (50% stock) at 37 °C for 1 h; adapters were added at 812 ~250 nM each. Post-ligation clean-up used Zymo R1016 as above, eluting in 12 µL 10 mM Tris 813 pH 8. K562 were cleaned up using custom-made size selection beads46. 814 815 Libraries were barcoded by PCR in 50 µL with KAPA HiFi PCR Mix (2×; KAPA KK2602), 2 µM 816 each i5/i7 barcode primer, and 21 µL template; cycling was 98 °C 3 min; 5 initial cycles of 98 °C 817 30 s, 65 °C 30 s, 72 °C 1 min; 72 °C 1 min. A side-reaction qPCR on a 5 µL aliquot determined 818 additional cycles to retain exponential amplification (BJ, K562 typically 12 total cycles). gDNA 819 control, 0 Gy, and irradiated conditions were processed in parallel through all steps. Rad source 820 1800 Q4 X-ray Irradiator used delivering a dose rate of 123.1 Gy/min on the top shelf. BJ, K562, 821 Chicken red blood cells were irradiated with a total dose of 300 Gy and Budding Yeast at 1000 822 Gy. 823 824 825 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 19 826 827 Hi-C library preparation 828 829 HiC was performed using the HiC 3.0 protocol followed by NEBNext Ultra II library preparation. 830 Briefly, pellets of 5 M cells were crosslinked first with 1% formaldehyde in HBSS for 10 minutes, 831 then with 3 mM DSG for 40 minutes, and subsequently snap frozen in liquid nitrogen. Cells were 832 lysed with a Dounce homogenizer in a buffer containing 0.2% NP-40, followed by chromatin 833 solubilization by SDS, quenched with Triton X-100. Chromatin was digested with a cocktail of 834 DdeI (400 U) and DpnII (400 U), incubated overnight at 37°C. DNA ends were then biotinylated 835 using Klenow polymerase and a dNTP mix containing biotin-14-dATP, then proximity ligation 836 was performed using T4 DNA ligase. After biotin fill-in and proximity ligation, crosslinks were 837 reversed and proteins digested with an overnight 65°C incubation with Proteinase K. DNA was 838 then purified and concentrated via phenol-chloroform extraction followed by ethanol precipitation 839 and passage through Amicon filter tubes. End repair was performed with T4 polymerase and 840 dATP/dGTP, then DNA was sonicated to obtain a distribution of sequenceable DNA lengths, 841 which was further size selected using AMPure XP beads. At this point, we switched to NEBNext 842 Ultra II library preparation, following the manufacturer’s protocol, then sequenced samples on 843 an Illumina NovaSeq X Plus platform to a depth of 50 - 90 million read pairs per replicate for two 844 technical replicates. 845 846 In-situ Micro-C library preparation 847 848 Two biological replicates of Micro-C were performed using an in-situ proximity ligation-based 849 protocol, adapted for our purposes from previously published Micro-C protocols, as described 850 below. 851 852 Crosslinking 853 Pellets of 10 million cells were crosslinked with 1% formaldehyde for 10 minutes, quenched with 854 1 M Tris-HCl pH 7.5, washed with DPBS, then crosslinked with 3 mM DSG (synthesized in 855 house) for 45 minutes. DSG crosslinking was quenched with 1 M Tris-HCl pH 7.5, and cells 856 were washed with DPBS before snap freezing in liquid nitrogen. 857 858 MNase titration 859 Cell pellets were thawed on ice and resuspended in buffer MB128. The cell suspension was split 860 into five aliquots of 2 million nominal cells, 2-3 of which aliquots were used for MNase titration, 861 and 2-3 of which were used as experimental samples. Experimental samples were kept on ice 862 while the MNase titration was took place (typically about three hours). Titration was carried out 863 by extracting nuclei by incubating cells in MB1 for 20 minutes, spinning down and resuspending 864 nuclei in MB1, then adding varying volumes of MNase (NEB M0247S) to aliquots. Typically, 865 volumes between 2 and 6 uL of 1:10 diluted MNase produced the desired level of digestion. 866 MNase digestion was performed for 10 minutes at 37°C on a thermomixer, then stopped with 867 EGTA and incubation at 65°C. Digested chromatin was treated with Proteinase K (NEB P8107) 868 and RNase A (Sigma-Aldrich R4642) and incubated for 2 hours at 65°C to reverse crosslinks 869 and digest proteins. DNA was then purified with phenol-chloroform extraction followed by 870 passage through a Zymo DNA Clean & Concentrator kit. The digestion profiles of the titration 871 samples were assessed by quantifying the DNA by Qubit and running 1 ng on a TapeStation. 872 MNase concentrations leading to a major mononucleosome peak and small dinucleosome peak 873 were identified and used for the full protocol with the remaining aliquots left on ice. 874 875 Micro-C 876 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 20 Nuclei were spun down, resuspended in MB1 with the predetermined concentration of MNase, 877 and digested as in the titration. The digested sample was washed with MB228 then end repaired 878 and end labeled with PNK (NEB M0201L) and Klenow polymerase (NEB M0210M) with 879 biotinylated dATP and dCTP. After enzyme inactivation with EDTA and incubation at 65°C, the 880 sample was washed with MB328. Proximity ligation was carried out using T4 ligase (NEB 881 M0202L) and unligated fragments were digested with Exonuclease III (NEB M0206L). The 882 sample was then incubated overnight with Proteinase K and RNase A at 65°C to reverse 883 crosslinks and digest proteins. DNA was purified with phenol-chloroform extraction followed by 884 passage through a Zymo DNA Clean & Concentrator kit, then ligated mononucleosome pairs 885 were selected for by running the sample on a 2% agarose gel, excising and extracting the 886 dinucleosome-sized band, and pulling down with streptavidin beads. Library preparation was 887 then carried out using the NEBNext Ultra II kit, following manufacturer’s instructions. Libraries 888 were sequenced twice on an Illumina Novaseq X Plus to approximately 100-400 million 2 x 150 889 bp paired end reads per replicate per run. Reads from the two runs were combined for a final 890 depth of 500 - 600 million reads per technical replicate, or 1.6 - 2 billion reads per biological 891 replicate. 892 893 ATAC-seq library preparation 894 895 ATAC-seq was performed following the Omni-ATAC-seq protocol, using Tn5 made in-house with 896 the Open-Tn5 method. Tn5 was loaded with Illumina adaptors by incubating equal volumes of 1 897 mg/mL Tn5 and 1 𝜇M adaptor together for 10 minutes at room temperature immediately before 898 being used for tagmentation. Briefly, the Omni-ATAC-seq protocol involved harvesting cells in 899 aliquots of 50,000 cells – in this case, we did two separate harvests (biological replicates) with 900 3-5 aliquots (technical replicates) per condition – followed by light permeabilization in a buffer 901 containing 0.1% NP-40, 0.1% Tween, and 0.01% digitonin. The samples were then tagmented 902 with Tn5 at 37°C for 30 minutes, cleaned up with a Zymo DNA Clean & Concentrate kit, and 903 PCR amplified with barcoded primers for an optimized number of cycles determined with a side 904 qPCR reaction. Following amplification, libraries were cleaned up, quantified, and sequenced on 905 an Illumina NextSeq 2000 platform to a depth of 30 - 50 million read pairs per technical 906 replicate. 907 908 CUT&Tag library preparation 909 910 Cut&Tag libraries were prepared using the Epicypher protocol. K562 cells were lightly fixed 911 using 0.1% formaldehyde for 1 minute, spun down resuspended in cell freezing media and 912 frozen down post day 5 doxycycline induction of H1 depletion. Pellets were then hawed, spun 913 down, washed in PBS and 100,000 cells per technical replicate counted. The cut and tag 914 protocol proceeded as outlined in the Epicypher v1.7 protocol. TN5 for these CUT&Tag libraries 915 was made in house using our previously published protocol47. The antibody used for mapping 916 H3K27me3 was CST 9733 at a 1:100 final concentration. 917 918 RNA extraction and RNA-seq library preparation 919 920 Triplicate cultures of cells expressing H1 sgRNAs and Scramble sgRNA were induced with 921 Doxycycline 1ug/ml for 5 days. Cells were backdiluted during treatment in order to maintain cell 922 cultures in exponential growth phase. Five mmillion cells were collected on day 5. One million 923 cells were ressuspended in Trizol for RNA extraction and four million cells used for acid extraction 924 and HPLC for validation of H1 depletion. Total RNA was extracted using Direct-zol RNA Miniprep 925 Kit (Zymo Research). Standard mRNA -Seq (poly(A) selection) was performed at Azenta Life 926 Sciences. Libraries were performed incorporating unique molecular identifers during adapter 927 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 21 ligation and External RNA Controls Consortium (ERCC) spike -ins were added to ech sample 928 before reverse transcription. 929 930 Chromatin Enrichment for Proteomics (ChEP) 931 932 Chromatin-bound proteins were isolated using the ChEP protocol as described by Kustatscher et 933 al.48 with minor modifications. Briefly, cells were formaldehyde -crosslinked (1%, 10 min), nuclei 934 were isolated under hypotonic conditions, and chromatin was enriched by sequential high -935 stringency washes under denaturing conditions (2% SDS, 8 M urea). Crosslinked chromatin was 936 sonicated and used for quantitative Mass Spectrometry analysis. 937 938 Mass-spectrometry 939 940 Samples were alkylated with 30mM IAA for 45min at RT in the dark. Reactions were then desalted 941 into 50mM NH4HCO3 using ZebaSpin 7k columns (ThermoFisher) and eluates were 942 supplemented with trypsin (0.1mg/ml) and digested for 2h at 37C. At the end of the 2h, samples 943 were supplemented with additional trypsin and digestions allowed to proceed overnight. 944 Digestions were quenched with 1% formic acid, dried in SpeedVac and then resuspended in 130 945 µl MS Sample Buffer (0.1% formic acid, 1% acetonitrile in water). 946 947 LCMS analyses were performed on a TripleTOF 5600+ mass spectrometer (AB SCIEX) coupled 948 with M5 MicroLC system (AB SCIEX/Eksigent) and PAL3 autosampler. LC separation was 949 performed in a trap-elute configuration, which consists of a trap column (LUNA C18(2), 100 Å, 5 950 μm, 20 X 0.3 mm cartridge, Phenomenex) and an analytical column (Kinetex 2.6 μm XB-C18, 100 951 Å, 50 X 0.3 mm microflow coumn, Phenomenex). The mobile phase consisted of water with 0.1% 952 FA (phase A) and 100% ACN containing 0.1% FA (phase B). 953 954 Peptides in MS Sample Buffer were injected into a 50-μl sample loop, trapped and cleaned on 955 the trap column with 3% mobile phase B at a flow rate of 25 μl/min for 4 min before being 956 separated on the analytical column with a gradient elution at a flow rate of 5 μl/min. The gradient 957 was set as follows: 0–24 min: 3% to 35% phase B, 24–27 min: 35% to 80% phase B, 27–32 958 min: 80% phase B, 32–33 min: 80% to 3% phase B, and 33–38 min at 3% phase B. An equal 959 volume of each sample (30 μl) was injected four times, once for information-dependent 960 acquisition (IDA), immediately followed by DIA/SWATH in triplicate. Acquisitions of distinct 961 samples were separated by a blank injection (80 µl MS Sample Buffer) to prevent sample 962 carryover. The mass spectrometer was operated in positive ion mode with EIS voltage at 5200 963 V, Source Gas 1 at 30 psi, Source Gas 2 at 20 psi, Curtain Gas at 25 psi, and source 964 temperature at 200°C. 965 966 RICC-seq data processing 967 968 RICC-seq library alignment and fragment length distribution (FLD) generation 969 Alignment. Illumina FASTQ paired end reads were aligned with Bowtie2 to prebuilt 970 Bowtie2Index references (e.g., hg38/hg19/mm10), applying a mapping-quality cutoff (MAPQ ≥ 971 30) and optional removal of reference blacklist regions (defaults provided per genome when 972 available). Per-sample reads to the yeast genome E2F were also aligned and used for 973 downstream spike in correction and length bias normalization. 974 Subsetting. For each input BAM, we performed a name sort (samtools sort -n) to ensure proper 975 pairing semantics for downstream intersection. Paired-end alignments were then intersected 976 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 22 with each provided BED/peak set using bedtools pairtobed (criterion: overlap of the paired 977 fragment span with the feature set; -type ospan). The script iterates over all BAM×BED 978 combinations. The resulting per-combination subset BAMs (reads whose paired fragment spans 979 overlapped the specified regions) were carried forward to downstream analyses. 980 981 Fragment length histogram generation. For each input BAM, we computed paired-end insert-982 size distributions with Picard Toolkit (Broad Institute) CollectInsertSizeMetrics. The task 983 excludes PCR/optical duplicates. Picard was invoked with DEVIATIONS=10.0 to capture long-984 tail fragments up to mean ± 10 SD, MINIMUM_PCT=0.05 to require ≥5% of pairs for a stable 985 estimate, and HISTOGRAM_WIDTH=700. For each BAM, the script writes (i) a PDF histogram 986 _hist.pdf and (ii) a tabulated metrics log 987 _full_hist_graphwithoutdups.log (median/mean/stdev, read counts, and percentile 988 cutoffs) for downstream QC. The log files for both spike in controls and samples were then 989 compiled to use for plotting Fragment Length Distributions (FLD) and correcting. 990 FLD normalization and correction 991 Fragment length distributions were corrected in two stages. First, per-sample spike-in 992 normalization was applied using a biological replicate-specific scaling factor defined as the ratio 993 of the mean spike-in read depth of all technical replicates within a biological replicate (REF𝑏) to 994 the individual sample’s spike-in depth (𝐷𝑖). Each replicate distribution was multiplied by 995 REF𝑏/𝐷𝑖to equalize spike-in coverage across replicates. 996 Second, to correct for fragment-length bias, we used the spike-in–scaled curve and its length-997 bias–corrected counterpart calculated by comparing the spike-in ladder FLD fragment loss after 998 sequencing to the ladder input on TapeStation. For each technical replicate, a length-bias 999 correction factor was computed per base pair. Within each biological replicate, these per-1000 technical replicate correction curves were averaged to obtain a mean biological replicate-1001 specific falloff profile which was used to correct each averaged biological replicate curve. 1002 Each curve was then normalized to the signal of the mononucleosome at 180 bp and lightly 1003 smoothed (10 bp rolling average). These biological replicate-level normalized profiles were used 1004 to compute condition means ± 95 % confidence intervals and to perform per-base Welch’s t-1005 tests, with significant contiguous regions (≥ 5 bp) highlighted in downstream figures. 1006 To summarize a condition (e.g., SCRM or dH1), we stacked all available replicates from that 1007 condition and compute the condition mean curve as the pointwise average across biological 1008 replicates. 95% confidence intervals were obtained using a Student’s t interval across biological 1009 replicates, i.e., mean ± 𝑡0.975, 𝑛−1 ⋅ SD/√𝑛, where 𝑛 is the number of biological replicates at that 1010 bp (using exact 𝑡for small 𝑛, ~1.96 as 𝑛grows). For between-condition comparisons we plotted 1011 the pointwise dH1/SCRM mean ratio and assessed significance at each bp with a two-sided 1012 Welch’s t-test computed over the underlying biological replicate curves, shading only significant 1013 segments of length ≥ 5 bp as significant differences. 1014 All shown subset RICC-seq curves were length bias corrected by the previously calculated 1015 respective biological replicate correction factor and divided by the corrected genomic DNA 1016 control (PLC) curve to produce the signal over the genomic background per condition. 1017 HiC and Micro-C analysis 1018 1019 Alignment and QC filtering 1020 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 23 Basic alignment and quality control (QC) filtering for both HiC and Micro-C was done based on 1021 the Dovetail Genomics analysis pipeline (micro-c.readthedocs.io), in which ligation pairs were 1022 aligned with bwa mem (v 0.7.17) using two-sided alignment, then valid ligation events were 1023 identified with pairtools parse (v 0.3.0). PCR duplicates were removed and final bam and pairs 1024 files were created with pairtools split. From pairs files, ICE balanced mcool files with a base 1025 resolution of 500 bp were made using cooler cload and zoomify (v 0.8.6) with default 1026 parameters, and hic files with a base resolution of 500 bp were made using juicer_tools pre (v 1027 1.22.01) with default parameters. Pairs, mcool, and hic files were then used for subsequent 1028 analyses, described below. 1029 1030 HiC analysis 1031 P(s) curves were made from balanced mcool files at 10 kb resolution using cooltools (v 0.5.4) 1032 expected_cis with smoothing. For correlation analysis with HiCRep49 , we converted our HiC 1033 pairs files to full contact matrices at 500 kb resolution using juicer and straw (v 1.6)50, then 1034 computed pairwise SCC scores between all replicates, within and between conditions. 1035 Chromosome compartment analysis was carried out using the cscoretool package (v1.1)51 . 1036 First, cscoretool was run on each chromosome for each condition, using pairs files at 100 kb 1037 resolution. We next calculated Spearman’s correlation coefficient for each chromosome’s c-1038 scores with H3K36me3 ChIP signal (from ENCODE52 ENCSR000AKR) and flipped the sign of 1039 the c-score for negatively correlated chromosomes so that positive c-scores consistently 1040 correspond to the gene-dense A compartment (supp fig ref). Chromosomes for which correlation 1041 did not pass a significance threshold of p < 0.01 were excluded from the analysis. Differential 1042 compartment scores were found by subtracting scrambled control c-scores from H1-low c-1043 scores in matched genomic bins, and shifts were defined as |∆ c-score | > 0.25, and negative to 1044 positive = B to A, positive to negative = A to B, increasing within positive or negative = A-shifted, 1045 decreasing within positive or negative = B-shifted. 1046 1047 Micro-C domain and loop calling 1048 Domain analysis was performed using the cooler/cooltools suites. From balanced mcool files, 1049 the cooltools insulation tool was used to find insulation strength and call domain boundaries at a 1050 base resolution of 10 kb. Loop calling was performed on hic files filtered to remove inward 1051 ligations (see below). Loops were called at 5 kb resolution using the juicer hiccups tool with KR 1052 normalization. CTCF loops were identified with juicer motifs using publicly available SMC3 1053 (ENCSR000EGW), RAD1 (ENCSR000FAD), and CTCF (ENCSR000EGM) ChIP-seq in K56252. 1054 Differential loops were found by calling loops on biological replicates, finding reproducible loops 1055 within conditions, and then comparing the consensus loop lists between conditions. Aggregate 1056 peak analysis was performed using the juicer apa tool. 1057 1058 Short-range contact probability 1059 Contact probability is calculated from pairs files by first separating reads based on ligation 1060 orientation, subtracting positions of cis pairs to find contact distance, then plotting a histogram of 1061 those contact distances. It is necessary to separate pairs by ligation orientation because contact 1062 probability curves for each orientation are shifted relative to each other – this is due to the fact 1063 that read positions will correspond to the entry or exit of each nucleosome in a pair depending 1064 on their ligation orientation, and accordingly, the contact distance will include or not the lengths 1065 of the nucleosomes. Pairs are separated by ligation orientation based on the strand information 1066 as follows: +/- pairs were designated “inward” ligations, -/+ pairs were designated “outward” 1067 ligations, and +/+ or -/- pairs were designated “tandem” ligations. To look at short-range contact 1068 probability in specific genomic regions, bam files were first intersected with bed regions using 1069 bedtools intersect (v 2.30.0) (cite), then pairs with read IDs matching reads in the intersected 1070 bam were used for contact distance calculation. 1071 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 24 1072 Nucleosome contact peaks are called from the contact probability curves by finding where the 1073 second derivative of the curve is negative. NRL is calculated from these peaks by finding the 1074 average basepair distance between maxima and N/N+odd:N/N+even ratios of nucleosome 1075 contacts are calculated by summing N/N+3 and N/N+5 contact probabilities, summing N/N+2 1076 and N/N+4 contact probabilities, and diving the two values. A higher ratio indicates an 1077 enrichment of N/N+odd contacts. 1078 1079 RNA-seq analysis 1080 1081 RNA-seq data was analyzed by first extracting UMIs and filtering for unique reads using fastp (v 1082 0.24.0) (cite) with parameters --umi_loc per_read --umi_skip 2 --umi_len 5 to match the UMI 1083 scheme used by Genewiz. The UMI-filtered data was then aligned to an index composed of 1084 combined GRCh38 and ERCC transcriptomes using STAR alignment (v 2.7.11b) (cite) 1085 (parameters: --outFilterType BySJout --outFilterMultimapNmax 15 --alignSJoverhangMin 8 --1086 alignSJDBoverhangMin 1 --outFilterMismatchNmax 500 --outFilterMismatchNoverReadLmax 1087 0.05 --alignIntronMin 20 --alignIntronMax 1000000). A counts matrix of paired-end fragments 1088 over genes was made from the aligned reads using featureCounts (subread v 2.0.6) (Liao Y, 1089 Smyth GK and Shi W (2014)) and the combined gencode.v38-ERCC transcriptome, then 1090 DESeq2 (v 1.42.0) was run on genes with at least 10 total counts, using ERCC genes as the 1091 control gene set with estimateSizeFactors. 1092 1093 1094 ATAC-seq data processing 1095 1096 ATAC-seq data was aligned using Bowtie253 filtered, and shift-corrected using deeptools 1097 alignmentSieve parameter, which shifts the plus strand by +4 and the minus strand by –5 to 1098 account for the Tn5 homodimer which leaves 9bp of DNA between the two Tn5 molecules. 1099 Peaks were called using macs254,55 reproducible peaks were found between replicates using 1100 Irreproducible discover rate (IDR)56 , and a master peak set was made by listing reproducible 1101 peaks from both conditions and concatenating adjacent peaks. GenomicRanges,, an R 1102 package, is used to load the master peak for further downstream analysis in R. Differential peak 1103 analysis was carried out with DESeq2 separately on peaks within 5 kb of a TSS and those 1104 further than 5 kb from any TSS. Fragment length distributions were found from filtered bam files 1105 using samtools57. 1106 1107 Fragment length distributions (FLD) were plotted over ATAC-seq FLD bed files that intersected 1108 K562 ChromHmm regions. To get bed files that consisted of accurate fragment lengths, we 1109 made ATAC-seq bams that had replicates merged by condition using samtools, then bedpe files 1110 were created using bedtools bamtobed with the parameter bedpe. Finally, using awk to get the 1111 chromosome name, forward read start coordinates and reverse read end coordinates, which 1112 represents the true length of a fragment read and saved that to a bed file. Next, to get the 1113 fragments that intersected chromHmm regions, we used bedtools intersect to find and record 1114 only the ATAC-seq true fragments that overlapped any of the chromHmm regions. This 1115 represents our ATAC-seq fragments found in chromHmm regions bed file. 1116 1117 1118 CUT&Tag data processing 1119 1120 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 25 CUT&Tag data was aligned with BWA mem58, filtered, with duplicates removed. Peaks were 1121 called using Sicer259, then concatenated the peak set from each replicate from both conditions 1122 to create a master peak set. 1123 1124 Bigwig files were created using deeptools60 bamCoverage with reads extended and CPM 1125 normalized. 1126 1127 Differential peak analysis was done with DESeq261 using the master peak set as regions for 1128 aligned reads to be tallied over. Up and down peaks are those that pass two thresholds; 1) Padj 1129 value |2|. Principle component analysis (PCA) plots were 1130 generated using DESeq2 results to show the variance between conditions and replicates to 1131 ensure unwanted batch effects were not playing a pivotal role underlying the data. 1132 1133 ChIPseeker62,63 is used to annotate the differential peaks. This gives insights into the distribution 1134 of peaks in various regions of the genome, as seen in the legend of the plot labeled features. 1135 1136 Heat maps and profile plots were generated using deeptools. To show the reliability of peaks 1137 called, we took the differential peaks and plotted the bigwig signal over the center of said 1138 differential peaks with 5kb up and down stream. This region illustrates the difference in signal 1139 between the two conditions H1low and Scrambled. 1140 We also plotted the Cut&Tag H3K27me3 signal over Pro-seq nascent transcription regions to 1141 discover if H1 linker histone affecting compaction plays a pivotal role in nascent transcription. 1142 Another profile/heatmap plot generated by our pipeline shows our Cut&Tag H3K27me3 signal 1143 over differential ATAC-seq peaks. The interaction between chromatin compaction state and loss 1144 of H1 illustrates that as we lose linker histone, there is more accessibility. 1145 1146 Nextflow pipeline data processing 1147 1148 Nextflow64 is used to create a reproducible and scalable pipeline that incorporates many of the 1149 tools mentioned in the methods section. We have two pipelines engineered to handle 1150 epigenomic sequencing techniques. The first Risca Lab pipeline (NEXDEP) can process fastq 1151 reads from ATAC-seq, Cut&Tag, Cut&Run, ChIP-seq assays to align, filter, give quality control 1152 metrics and preprocess data to produce bam files with sequence alignment information. The 1153 second Risca Lab pipeline was engineered specifically to call peaks from previously mentioned 1154 assays and provide downstream analysis and plots such as heatmaps, MA-plots, PCA plots, 1155 and peak annotation information, along with many other custom analytical Nextflow workflow 1156 techniques that partially aided in completion of this study. 1157 1158 IDA and data analyses 1159 1160 IDA was performed to generate reference spectral libraries for SWATH data quantification. The 1161 IDA method was set up with a 200 ms TOF -MS scan from 300 to 1,250 Da, followed by MS/MS 1162 scans in a high-sensitivity mode from 100 to 1,500 Da of the top 25 precursor ions above 100 cps 1163 threshold (80 ms accumulation time, 100 ppm mass tolerance, rolling collision energy, and 1164 dynamic accumula -tion) for charge states (z) from +2 to +5. IDA files were searched using 1165 ProteinPilot (version 5.0.2, ABSciex) with a default setting for tryptic digest and IAA alkylation 1166 against a protein sequence data-base. 1167 1168 The Homo sapiens proteome FASTA file (82,493 protein entries, UniProt UP000005640) 1169 augmented with sequences for common contaminants was used as a reference for the search. 1170 Up to two missed cleavage sites were allowed. Mass tolerance for precursor and fragment ions 1171 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 26 was set to 100 ppm. A false discovery rate (FDR) of 5% was used as the cutoff for peptide 1172 identification. 1173 1174 1175 1176 1177 SWATH acquisitions and data analyses 1178 1179 For SWATH (SWATH-MS, Sequential Window Acquisition of All Theoretical Mass Spectra) acqui-1180 sitions (Zhu et al., 2014), one 50-ms TOF-MS scan from 300 to 1,250 Da was performed, followed 1181 by MS/MS scans in a high-sensitivity mode from 100 to 1,500 Da (15 ms accumulation time, 100 1182 ppm mass tolerance, +2 to +5 z, rolling collision energy) with a variable -width SWATH window 1183 (Zhang et al., 2015). DIA data were quantified using PeakView (version 2.2.0.11391, ABSciex) 1184 with SWATH Ac -quisition MicroApp (version 2.0.1.2133, ABSciex) against selected spectral 1185 libraries generated in Pro -tein-Pilot. Retention times for individual SWATH acquisitions were 1186 calibrated using 25 or more pep-tides for plectin (PLEC, UniProt Q15149) and myosin-9 (MYH9, 1187 UniProt P35579), two abundant pro-teins that were highly representative in the IDA ion library and 1188 all SWATH acquisitions. The following software settings were utilized: up to 25 peptides per 1189 protein, 6 transitions per peptide, 95% peptide confidence threshold, 5% FDR for peptides, XIC 1190 extraction window 10 minutes, and XIC width 100 ppm. 1191 1192 1193 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 27

1194 1195 1. Felsenfeld, G. & McGhee, J. D. Structure of the 30 nm chromatin fiber. Cell 44, 375–377 1196 (1986). 1197 2. Li, G. & Reinberg, D. Chromatin higher-order structures and gene regulation. Curr. Opin. 1198 Genet. Dev. 21, 175–186 (2011). 1199 3. Maeshima, K., Iida, S. & Tamura, S. Physical Nature of Chromatin in the Nucleus. Cold 1200 Spring Harb. Perspect. Biol. 13, (2021). 1201 4. Mansisidor, A. R. & Risca, V. I. Chromatin accessibility: methods, mechanisms, and 1202 biological insights. Nucleus 13, 236–276 (2022). 1203 5. Ou, H. D. et al. ChromEMT: Visualizing 3D chromatin structure and compaction in 1204 interphase and mitotic cells. Science 357, (2017). 1205 6. Willcockson, M. A. et al. H1 histones control the epigenetic landscape by local chromatin 1206 compaction. Nature 589, 293–298 (2021). 1207 7. Routh, A., Sandin, S. & Rhodes, D. Nucleosome repeat length and linker histone 1208 stoichiometry determine chromatin fiber structure. Proc. Natl. Acad. Sci. U. S. A. 105, 1209 8872–8877 (2008). 1210 8. Fyodorov, D. V., Zhou, B.-R., Skoultchi, A. I. & Bai, Y. Emerging roles of linker histones in 1211 regulating chromatin structure and function. Nat. Rev. Mol. Cell Biol. 19, 192–206 (2018). 1212 9. Robinson, P. J. J. & Rhodes, D. Structure of the “30 nm” chromatin fibre: a key role for the 1213 linker histone. Curr. Opin. Struct. Biol. 16, 336–343 (2006). 1214 10. Robinson, P. J. J. et al. 30 nm chromatin fibre decompaction requires both H4-K16 1215 acetylation and linker histone eviction. J. Mol. Biol. 381, 816–825 (2008). 1216 11. Song, F. et al. Cryo-EM study of the chromatin fiber reveals a double helix twisted by 1217 tetranucleosomal units. Science 344, 376–380 (2014). 1218 12. Gibson, B. A. et al. Organization of Chromatin by Intrinsic and Regulated Phase 1219 Separation. Cell 179, 470-484.e21 (2019). 1220 13. Nishino, Y. et al. Human mitotic chromosomes consist predominantly of irregularly folded 1221 nucleosome fibres without a 30-nm chromatin structure. EMBO J. 31, 1644–1653 (2012). 1222 14. Eltsov, M., MacLellan, K. M., Maeshima, K., Frangakis, A. S. & Dubochet, J. Analysis of 1223 cryo-electron microscopy images does not support the existence of 30-nm chromatin fibers 1224 in mitotic chromosomes in situ. Proceedings of the National Academy of Sciences 105, 1225 19732–19737 (2008). 1226 15. Maeshima, K., Hihara, S. & Eltsov, M. Chromatin structure: does the 30-nm fibre exist in 1227 vivo? Curr. Opin. Cell Biol. 22, 291–297 (2010). 1228 16. Luger, K., Dechassa, M. L. & Tremethick, D. J. New insights into nucleosome and 1229 chromatin structure: an ordered state or a disordered affair? Nat. Rev. Mol. Cell Biol. 13, 1230 436–447 (2012). 1231 17. Chen, L. et al. Nucleosome spacing can fine-tune higher-order chromatin assembly. Nat. 1232 Commun. 16, 6315 (2025). 1233 18. Hsieh, T.-H. S. et al. Resolving the 3D Landscape of Transcription-Linked Mammalian 1234 Chromatin Folding. Mol. Cell 78, 539-553.e8 (2020). 1235 19. Hsieh, T.-H. S. et al. Mapping Nucleosome Resolution Chromosome Folding in Yeast by 1236 Micro-C. Cell 162, 108–119 (2015). 1237 20. Risca, V. I., Denny, S. K., Straight, A. F. & Greenleaf, W. J. Variable chromatin structure 1238 revealed by in situ spatially correlated DNA cleavage mapping. Nature 541, 237–241 1239 (2017). 1240 21. Ricci, M. A., Manzo, C., García-Parajo, M. F., Lakadamyali, M. & Cosma, M. P. Chromatin 1241 fibers are formed by heterogeneous groups of nucleosomes in vivo. Cell 160, 1145–1158 1242 (2015). 1243 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 28 22. Ding, X., Lin, X. & Zhang, B. Stability and folding pathways of tetra-nucleosome from six-1244 dimensional free energy surface. Nat. Commun. 12, 1091 (2021). 1245 23. Kilic, S. et al. Single-molecule FRET reveals multiscale chromatin dynamics modulated by 1246 HP1α. Nat. Commun. 9, 235 (2018). 1247 24. Portillo-Ledesma, S. et al. Nucleosome Clutches are Regulated by Chromatin Internal 1248 Parameters. J. Mol. Biol. 433, 166701 (2021). 1249 25. Portillo, S., Tsao, L. H. & Schlick, T. Nucleosome Clutches in Chromatin are Tightly 1250 Regulated by Nucleosome Positions and Linker Histone Density. Biophysical Journal vol. 1251 118 8a–9a Preprint at https://doi.org/10.1016/j.bpj.2019.11.3370 (2020). 1252 26. Collepardo-Guevara, R. & Schlick, T. Chromatin fiber polymorphism triggered by variations 1253 of DNA linker lengths. Proc. Natl. Acad. Sci. U. S. A. 111, 8061–8066 (2014). 1254 27. Abdulhay, N. J. et al. Massively multiplex single-molecule oligonucleosome footprinting. 1255 Elife 9, (2020). 1256 28. Gaffney, D. J. et al. Controls of nucleosome positioning in the human genome. PLoS Genet. 1257 8, e1003036 (2012). 1258 29. Woodcock, C. L., Skoultchi, A. I. & Fan, Y. Role of linker histone in chromatin structure and 1259 function: H1 stoichiometry and nucleosome repeat length. Chromosome Research vol. 14 1260 17–25 Preprint at https://doi.org/10.1007/s10577-005-1024-3 (2006). 1261 30. Oberbeckmann, E., Quililan, K., Cramer, P. & Oudelaar, A. M. In vitro reconstitution of 1262 chromatin domains shows a role for nucleosome positioning in 3D genome organization. 1263 Nat. Genet. 56, 483–492 (2024). 1264 31. Grau, D. et al. Structures of monomeric and dimeric PRC2:EZH1 reveal flexible modules 1265 involved in chromatin compaction. Nat. Commun. 12, 714 (2021). 1266 32. Leicher, R. et al. Single-molecule and in silico dissection of the interaction between 1267 Polycomb repressive complex 2 and chromatin. Proc. Natl. Acad. Sci. U. S. A. 117, 30465–1268 30475 (2020). 1269 33. Yuan, W. et al. Dense chromatin activates Polycomb repressive complex 2 to regulate H3 1270 lysine 27 methylation. Science 337, 971–975 (2012). 1271 34. Stenerlöw, B., Karlsson, K. H., Cooper, B. & Rydberg, B. Measurement of prompt DNA 1272 double-strand breaks in mammalian cells without including heat-labile sites: results for cells 1273 deficient in nonhomologous end joining. Radiat. Res. 159, 502–510 (2003). 1274 35. Troll, C. J. et al. A ligation-based single-stranded library preparation method to analyze cell-1275 free DNA and synthetic oligos. BMC Genomics 20, 1023 (2019). 1276 36. Freidkin, I. & Katcoff, D. J. Specific distribution of the Saccharomyces cerevisiae linker 1277 histone homolog HHO1p in the chromatin. Nucleic Acids Res. 29, 4043–4051 (2001). 1278 37. Scheffer, M. P., Eltsov, M. & Frangakis, A. S. Evidence for short-range helical order in the 1279 30-nm chromatin fibers of erythrocyte nuclei. Proc. Natl. Acad. Sci. U. S. A. 108, 16992–1280 16997 (2011). 1281 38. Clerkin, A. B., Pagane, N., West, D. W., Spakowitz, A. J. & Risca, V. I. Determining 1282 mesoscale chromatin structure parameters from spatially correlated cleavage data using a 1283 coarse-grained oligonucleosome model. bioRxiv (2024) doi:10.1101/2024.07.28.605011. 1284 39. Shimada, M. et al. Gene-Specific H1 Eviction through a Transcriptional 1285 Activator→p300→NAP1→H1 Pathway. Mol. Cell 74, 268-283.e5 (2019). 1286 40. Roadmap Epigenomics Consortium et al. Integrative analysis of 111 reference human 1287 epigenomes. Nature 518, 317–330 (2015). 1288 41. Yusufova, N. et al. Histone H1 loss drives lymphoma by disrupting 3D chromatin 1289 architecture. Nature 589, 299–305 (2021). 1290 42. Matthews, R. E. et al. CRAMP1 drives linker histone expression to enable Polycomb 1291 repression. Mol. Cell 85, 2503-2516.e8 (2025). 1292 43. Misteli, T., Gunjan, A., Hock, R., Bustin, M. & Brown, D. T. Dynamic binding of histone H1 to 1293 chromatin in living cells. Nature 408, 877–881 (2000). 1294 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 29 44. Portillo-Ledesma, S., Wagley, M. & Schlick, T. Chromatin transitions triggered by LH density 1295 as epigenetic regulators of the genome. Nucleic Acids Res. 50, 10328–10342 (2022). 1296 45. Lee, C.-H. et al. Distinct stimulatory mechanisms regulate the catalytic activity of polycomb 1297 repressive complex 2. Mol. Cell 70, 435-448.e5 (2018). 1298 46. Canaj, H. et al. Deep profiling reveals substantial heterogeneity of integration outcomes in 1299 CRISPR knock-in experiments. bioRxiv 841098 (2019) doi:10.1101/841098. 1300 47. Soroczynski, J. et al. OpenTn5: Open-Source Resource for Robust and Scalable Tn5 1301 Transposase Purification and Characterization. bioRxivorg (2024) 1302 doi:10.1101/2024.07.11.602973. 1303 48. Kustatscher, G., Wills, K. L. H., Furlan, C. & Rappsilber, J. Chromatin enrichment for 1304 proteomics. Nat. Protoc. 9, 2090–2099 (2014). 1305 49. Yang, T. et al. HiCRep: assessing the reproducibility of Hi-C data using a stratum-adjusted 1306 correlation coefficient. Genome Res. 27, 1939–1949 (2017). 1307 50. Durand, N. C. et al. Juicer provides a one-click system for analyzing loop-resolution Hi-C 1308 experiments. Cell Syst. 3, 95–98 (2016). 1309 51. Zheng, X. & Zheng, Y. CscoreTool: fast Hi-C compartment analysis at high resolution. 1310 Bioinformatics vol. 34 1568–1570 Preprint at https://doi.org/10.1093/bioinformatics/btx802 1311 (2018). 1312 52. ENCODE Project Consortium. An integrated encyclopedia of DNA elements in the human 1313 genome. Nature 489, 57–74 (2012). 1314 53. Langmead, B., Wilks, C., Antonescu, V. & Charles, R. Scaling read aligners to hundreds of 1315 threads on general-purpose processors. Bioinformatics 35, 421–432 (2019). 1316 54. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008). 1317 55. Gaspar, J. M. Improved peak-calling with MACS2. bioRxiv (2018) doi:10.1101/496521. 1318 56. Li, Q., Brown, J. B., Huang, H. & Bickel, P. J. Measuring reproducibility of high-throughput 1319 experiments. aoas 5, 1752–1779 (2011). 1320 57. Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–1321 2079 (2009). 1322 58. Li, H. & Durbin, R. Fast and accurate long-read alignment with Burrows-Wheeler transform. 1323 Bioinformatics 26, 589–595 (2010). 1324 59. Xu, S., Grullon, S., Ge, K. & Peng, W. Spatial clustering for identification of ChIP-enriched 1325 regions (SICER) to map regions of histone methylation patterns in embryonic stem cells. 1326

Mol. Biol. 1150, 97–111 (2014). 1327 60. Ramírez, F. et al. deepTools2: a next generation web server for deep-sequencing data 1328 analysis. Nucleic Acids Res. 44, W160-5 (2016). 1329 61. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for 1330 RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014). 1331 62. Wang, Q. et al. Exploring epigenomic datasets by ChIPseeker. Curr. Protoc. 2, e585 1332 (2022). 1333 63. Yu, G., Wang, L.-G. & He, Q.-Y. ChIPseeker: an R/Bioconductor package for ChIP peak 1334 annotation, comparison and visualization. Bioinformatics 31, 2382–2383 (2015). 1335 64. Di Tommaso, P. et al. Nextflow enables reproducible computational workflows. Nat. 1336 Biotechnol. 35, 316–319 (2017). 1337 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint a d c e f g RICC-seq 1.0 RICC-seq 2.0 Paired-End Sequencing Remove 3’ and 5’ phosphates with rSAP P OH Denature with 0.1M NaOH and elute ssDNA SRSLY adapter ligation & column clean-up -OH P-P--OH NNNNNNN NNNNNNNOH- NNNNNN NNNNNNOH- (Predominant unwanted product: 3’ adapter dimer) (Illumina Read 1) (Illumina Read 2’) NNNNNN NNNNNN Barcoding PCR Barcode Wash Add back 5’ phosphates with PNK P Gamma- irradiate at 0 °C Agarose- embedded cells L yse, wash Soak in 5M Tris pH 8.0 Add spike-in DNA and SSB to stabilize Incubate 65˚C to inactivate rSAP and re-denature P P OH OHOH P OH OHP P OH POH Add SRSLY sequencing adapters (blocked) Bead size selection 100 200 300 400 500 600 700 F ragment Length (nt) 0 200 400 600 800 1000Spike-in molarity 0 100 200 300 400 500 600 700 F ragment Length (nt) 0.0 0.2 0.4 0.6 0.8 1.0 100 200 300 400 500 600 700 F ragment Length (nt) 0.0 0.2 0.4 0.6 0.8 1.0 Pre-seq Post-seq 100 200 300 400 500 600 700 F ragment Length (nt) 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Fit ≥300 bp Spike-in Pre-sequencing Spike-in Post-sequencing Spike-in sequencing counts Aligned spike-in fragment densities Ratio of post- to pre-sequencing Counts uniformly scaled by spike-in Coutns corrected for length bias by spike-in Ratio of 300 Gy to genomic DNA (PLC) Counts uniformly scaled by spike-in Coutns corrected for length bias by spike-in Ratio of 300 Gy to genomic DNA Fr agment Length (bp) 0 250 500 750 1000 20 40 60 80 20 40 60 80 RICC 1.0 RICC 2.0 % GC % GC RICC 2.0 Workflow Sequencing countsFragment length (nt) b e- electrons OH radicals cause base damage and strand breaks within a ~3.5 nm radius X-ray Ionization event e- 100 200 300 400 500 F ragment Length (nt) 100 200 300 400 500 Fragment Length (nt) 100 200 300 400 500 Fragment Length (nt) 2000 4000 6000 8000 10000 12000 0 2000 4000 6000 8000 10000 0 100 200 300 400 500 Fragment Length (nt) BJ fibroblast exp.1 BJ fibroblast exp.2 PLC 100 200 300 400 500 Fragment Length (nt) 100 200 300 400 500 Fragment Length (nt) 2000 4000 6000 8000 10000 0 0.5 0.75 1.00 1.25 1.75 0.25 0 1.5 2.0 2.5 3.0 3.5 1.0 0.5 1000 2000 3000 4000 5000 0 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 100 150 200 250 300 350 400 450 0 2000 4000 6000 Irr (raw) PLC (raw) 100 150 200 250 300 350 400 450 0.0 0.5 1.0 1.5 2.0 2.5 Yeast ±95% Yeast mean 100 150 200 250 300 350 400 450 0 1000 2000 3000 4000 5000 100 150 200 250 300 350 400 450 0.0 0.5 1.0 1.5 2.0 100 150 200 250 300 350 400 450 0 1000 2000 3000 4000 5000 Chicken erythrocyte 100 150 200 250 300 350 400 450 0.0 0.5 1.0 1.5 2.0 2.5 100 150 200 250 300 350 400 450 F ragment Length (nt) 0 2000 4000 6000 100 150 200 250 300 350 400 450 F ragment Length (nt) 0 1 2 3 Fragment Length (nt) F ragment Length (nt) Fragment Length (nt) 100 150 200 250 300 350 400 450 F ragment Length (nt) 2 Ratio of length-corrected counts: Cells / genomic DNA Budding Yeast + 95% CI BJ Fibroblast + 95% CI Mouse B-Cell + 95% CI Chicken Erythrocyte + 95% CI 3 1 0.6 a c b d e f g h i Fragment countsFragment countsFragment counts Fragment counts Counts ratio (cells / gDNA) (length bias-corrected) Counts ratio (cells / gDNA) (length bias-corrected) Counts ratio (cells / gDNA) (length bias-corrected) Counts ratio (cells / gDNA) (length bias-corrected) Mouse naive B-cell Human skin fibroblast (BJ) Budding yeast BJ fibroblast ±95% BJ fibroblast mean Chicken erythrocyte ±95% Chicken erthrocyte mean Mouse B-cell ±95% Mouse B-cell mean F ragment Length (nt) Fragment Length (nt) Fragment Length (nt) .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint a f k WT K562 rtTA G418 + G418 + dox sort for GFP BFP puro scrambled sgRNA BFP puro H1.2 - H1.5 sgRNA or + puro sort for BFP scr CTRL K562 WT levels of H1 with or without doxycycline H1-low K562 depleted of H1 upon doxycycline induction dCas9-KRAB-GFP TRE 0 0 .2 0.4 0.6 0.8H1:H2B ratio scr +dox H1-low +dox 0 5 1 0 15 20 25 30 scr -dox H1-low -dox scr +dox H1-low +dox Doubling time (hours) H2B H1cde H1b scr CTRL H1-low b c d 0.001 0.002 0 500 1000 1500 Micro-C contact frequency H1-low scr CTRL 0.001 0.002 2 4 6 Micro-C peak frequency H1-low scr CTRL 160 170 180 190 200NRL (bp) H1-low scr-CTRL 0.9 1.0Odd/Even contactsH1-low scr-CTRL 0.001 0.002 0 500 1000 1500 Genomic distance (bp) Micro-C contact frequency scr CTRL H1-low g h i j Genomic distance (bp) Nucleosome peak number 8 1 2 3 4 Ratio of corrected RICC-seq FLDs (Cells/gDNA) scr-CTRL mean (n=3) scr-CTRL 95% CI H1-low mean (n=3) H1-low 95% CI 0.7 0.8 0.9 1.0 1.1 H1-low / scr-CTRL H1-low / scr-CTRL (means) Welch p<0.05 H1-low / scr-CTRL l 100 150 200 250 300 350 400 450 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 Ratio of corrected RICC-seq FLDs (Cells / gDNA) scr-CTRL (n=2; with 95% CI) scr-CTRL dox-off (n=1; with 95% CI) H1-low (n=2; with 95% CI) H1-low dox-off (n=2; with 95% CI) 320 340 360 380 400 420 F ragment length (nt) 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25 100 150 200 250 300 350 400 450 1.00 1.25 1.50 1.75 scr-CTRL dox-off / scr-CTRL 100 150 200 250 300 350 400 450 1.0 1.2 H1-low dox-off / H1-low Welch p<0.05 100 150 200 250 300 350 400 450 0.6 0.8 1.0 1.2 H1-low dox-off / scr-CTRL dox-off Welch p<0.05 F ragment length (nt) 100 150 200 250 300 350 400 450 Fragment length (nt) 100 150 200 250 300 350 400 450 Fragment length (nt) m n o p FA and DSG crosslink MNase-digest DNA biotinylate, ligate DNA e Micro-C workflow Fragment length (nt) .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint Fragment length (Capillary electrophoresis A.U.) 5e−04 1e−03 0 500 1000 1500 Contact distance (bp) Micro-C contact probability DNA stain intensity (A.U.) 131 bp (1.0 uL MNase) 104 bp (1.2 uL MNase) 124 bp (0.8 uL MNase) k a i jg e h f d c 0.001 0.002 2 4 6 8 H3K27ac H3K27me3 H3K9me3 0.001 0.002 2 4 6 8 Nucleosome contact number Odd/Even contacts H1-low scr-CTRL 0.9 1.0 1.1 H3K27acH3K27me3H3K9me3H3K27acH3K27me3H3K9me3 scr-CTRL H1-low 4.2 5.8 7.4 9.0 0 100 200 300 400 500 5 6 7 8 0 100 200 300 400 500 Fragment Size (nt) Genome H3K9me3 H3K27me3 H3K27ac 5 6 7 0 100 200 300 400 500 5.8 7.2 8.6 10.0 0 100 200 300 400 500 Matched gDNA Control RICC-seq log2(Fragments/Mb) Matched gDNA Control scr-CTRL H1-low scr-CTRL H1-low Micro-C peak probability 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 full genome H3k9me3 H3k27me3 H3k27ac 100 150 200 250 300 350 400 450 Fragment Size (nt) 0.5 1.0 1.5 2.0 2.5 3.0 Ratio of corrected RICC-seq FLDs (Cells/gDNA) Ratio of corrected RICC-seq FLDs (Cells/gDNA) 100 150 200 250 300 350 400 450 Fragment Size (nt) b .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint a 0.965 0.970 0.975 0.980 HiC-Rep Spearman Correlation H1-low scr within condition between conditions c d b j scr H1-low 1 Mb 1 Mb −1.0 −0.5 0.0 0.5 1.0 −1.0 −0.5 0.0 0.5 1.0 scr-CTRL c score H1−low c score A to B B to A B-shift A-shift None scr H1-low # boundaries boundary strength H1-low 50 kb 100 kb 250 kb # boundaries 50 kb 100 kb 250 kb scr scr H1-low 10 kb 10 kb e f Scr-only loops H1-low-only loops Scr Scr H1-lowH1-low scr H1-low 5 kb 5 kb Common loops Scr H1-low 2520 loops 2520 loops 836 loops 836 loops 1872 loops 1872 loops 2998 loops 2998 loops 11960 loops 11960 loops CTCF: 23.8% decrease in loop number non-CTCF: 28.0% decrease in loop number Scr CTCF loops Scr non-CTCF loops Scr Scr H1-lowH1-low Scr 2520 H1- lo w 8361872 g h i .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint 0.0 0.1 0.2 0.3Enrichment ScoreFDR: 1.000 Pval: 1.000 NES: 0.859 0 2500 5000 7500 10000125001500017500 R ank in Ordered Dataset -5.0 -2.5 0.0 2.5Ranked list metric Pos Neg Zero score at 13365 GATA1 Targets un- changing down in H1-low up in H1-low 0 25 50 75 100 K27me3 CUT&Tag peaks (%) Promoter (<=1kb) Promoter (1−2kb) Promoter (2−3kb) Promoter (3−4kb) Promoter (4−5kb) 5' UTR 3' UTR 1st Exon Other Exon 1st Intron Other Intron Do wnstream (<=300) Distal Intergenic a c g j f h e 2.5 0.0 2.5 5.0 1e+01 1e+02 1e+03 1e+04 1e+05 log10 Base Mean log2 Fold-change H1-low/scr-CTRL not significant padj < 0.05 padj 1 1525 32 CHEP -Log10 P-value (H1-low vs. scr-CTRL) 3 3 2 1 0 0-6 -3 CHEP log2 Fold-change (H1-low vs. scr-CTRL) p-value log2fc & p-value b 0.0 0.2 0.4 0.6Enrichment ScoreFDR: 0.000 Pval: 0.000 NES: 1.568 0 2500 5000 7500 10000 125001500017500 R ank in Ordered Dataset -5.0 -2.5 0.0 2.5Ranked list metric Pos Neg Zero score at 13365 CBX2 Targets 0.0 0.2 0.4Enrichment ScoreFDR: 0.000 Pval: 0.000 NES: 1.178 0 2500 5000 7500 1000012500 15000 17500 R ank in Ordered Dataset -5.0 -2.5 0.0 2.5Ranked list metric Pos Neg Zero score at 13365 SUZ12 Targets un- changing up in H1-low 0 25 50 75 100 ATAC-seq peaks (%) Promoter (<=1kb) Promoter (1−2kb) Promoter (2−3kb) Promoter (3−4kb) Promoter (4−5kb) 5' UTR 3' UTR 1st Exon Other Exon 1st Intron Other Intron Do wnstream (<=300) Distal Intergenic0.0 2.5 5.0 10 0.5 10 1 10 1.5 10 2 10 2.5 10 3 10 3.5 ATAC-seq peak Base Mean Upregulated peaks: 2678 (5.4%) Do wnregulated peaks: 11 (0%) padj < 0.05 only: 5142 (10.4%) Not significant: 41419 (84.1%) 10 1 10 2 10 3 10 4 H3K27me3 CUT&Tag Broad Peak Base Mean H3K27me3 Log2 Fold-change H1-low vs scr-CTRL Upregulated peaks: 482 (3%) Downregulated peaks: 2453 (15.1%) pad j < 0.05 only: 4060 (25%) Not significant: 9222 (56.9%) −5.0 −2.5 0.0 2.5 d i -8.0 center 8.0Kb 1.0 1.5 2.0 2.5 3.0 3.5 4.0 -8.0 center 8.0Kb scr-CTRL H3K27me3 H1-low H3K27me3 -8.0 center 8.0Kb -8.0 center 8.0Kb 0 2 4 6 8 10 12 H1 low H3K27me3 Scrambled H3K27me3 Down H3K27me3 peaks All H3K27me3 peaks RPGC-normalized H3K27me3 CUT&Tag coverage -5.0 TSS 5.0Kb 1.0 1.5 2.0 2.5 3.0 -5.0 TSS 5.0Kb scr-CTRL H3K27me3 H1-low H3K27me3 scr-CTRL H3K27me3 -5.0 TSS 5.0Kb H1-low-upregulated genes H1-low H3K27me3 -5.0 TSS 5.0Kb Unchanging genes 0 2 4 6 8 10 k −0.2 −0.1 0.0 0.1 0.2 H3K27me3 all H3K27me3 down DNAse narrow peak H2A.Z narrow peak H3K27ac narrow peak H3K27me3 broad peak H3K36me3 broad peak H3K4me1 narrow peak H3K4me2 narrow peak H3K4me3 narrow peak H3K9me3 broad peak ATAC-seq H1-low vs. scr-CTRL log2 Fold-change Accessibility enriched Accessibility depleted Roadmap K562 ChIP-seq dataCut and Tag l TF Adjusted P-value Odds Ratio CBX8 2.99e-35 2.89 CBX2 2.25e-30 2.73 SUZ12 1.29e-24 2.42 GATA1 1.00e+00 0.38 ... ENRICHR ANAL YSIS OF UPSTREAM REGULATORS FOR GENES UP IN H1-low (K562 ENCODE TF 2015) 0.0 0.2 0.4 0.6Enrichment ScoreFDR: 0.000 Pval: 0.000 NES: 1.615 0 2500 5000 7500 10000125001500017500 R ank in Ordered Dataset -5.0 -2.5 0.0 2.5Ranked list metric Pos Neg Zero score at 13365 CBX8 Targets ATAC-seq Log2 Fold-change H1-low vs scr-CTRL RPGC-normalized H3K27me3 CUT&Tag coverage Upregulated genes: 1525/22978 = 6.6% Downregulated genes: 32/22978 = 0.139% .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint a b 100 150 200 250 300 350 400 450 Fragment Size (nt) 0.5 1.0 1.5 2.0 2.5 Corrected RICC-seq 2.0 FLD ratio signal norm. to mono. nuc. (cells / gDNA) 100 150 200 250 300 350 400 450 Fragment Size (nt) 0.5 1.0 1.5 2.0 2.5 100 150 200 250 300 350 400 Fragment Size (nt) 0.5 1.0 1.5 2.0 2.5 450 Corrected RICC-seq 2.0 FLD ratio signal norm. to mono. nuc. (cells / gDNA) 100 150 200 250 300 350 400 450 Fragment Size (nt) scr-CTRL- down peaks scr-CTRL- unchan ging H1-low- unchanging H1-low- down peaks K27me3 Differential Peaks 0.5 1.0 1.5 2.0 2.5 3.0 3.5 c scr-CTRL- full gen ome H1-low- full genome H1-low-ATAC unch. scr-CTRL- ATAC unch. scr-CTRL- full genome H1-low- full genome H1-low-ATAC up scr-CTRL- ATAC up H1-low-H3K27me3 scr-CTRL- H3K27ac scr-CTRL- H3K27me3 H1-low- H3K27ac H1-low- ATAC up scr-CTRL- ATAC up d e -10.0 center 10.0Kb 1.5 2.0 2.5 3.0 3.5 4.0 -10.0 center 10.0Kb scr-CTRL H3K27me3 up A TAC peaks H1low H3K27me3 unchanging A TAC peaks 0 2 4 6 8 10 12 14 -10.0 center 10.0Kb-10.0 center 10.0Kb scr-CTRL H3K27me3 H1low H3K27me3 Cut&Tag insertions Cut&Tag normalized insertion density .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted December 23, 2025. ; https://doi.org/10.64898/2025.12.19.695525doi: bioRxiv preprint