Genome reorganization and its functional impact during breast cancer progression

Background Cancer cells undergo widespread changes in epigenetic patterns that mediate cancer compromised gene expression programs during cancer progression. However, the alterations in higher -order genome organization in which these changes occur and their functional implications are less well understood. To explore how chromatin structure and epigenetic parameters of genome architecture changes during cancer progression at a fine scale and genome -wide, we generated high -resolution Micro -C contact maps in non - malignant, pre-cancerous, and metastatic MCF10 breast cancer epithelial cells.

We profiled progression -associated reorganization of chromatin compartments, topologically associated domains (TADs), and chromatin loops, and also identified invariable chromatin features. We find large -scale compartmental shifts occur predominantly in early stages of cancer development, with more fine -scale structural changes in TADs and looping accumulating during the later transition to metastasis. We related these structural features to changes in gene expression, histone marks, and potential enhancers and found a large portion of diYerentially expressed genes physically connected to distal regulatory elements. While changes in chromatin loops were relatively rare during progression, diYerential loops were enriched for progression -associated genes, including those involved in proliferation, angiogenesis, and diYerentiation. Changes in either enhancer-promoter contacts or distal enhancer activity were accompanied by diYerential gene regulation, suggesting that changes in chromatin contacts are not n ecessary but can be suYicient for gene regulation.

Together, our results demonstrate a functionally relevant connection between gene regulation and genome remodeling at many key genes during cancer progression. KEY FINDINGS - The cancer genome is reorganized throughout cancer progression at the level of compartments, chromatin domains, and loops - Compartmental shifts occur in early stages of cancer development, with more fine - scale structural changes accumulating during metastasis - Chromatin domain boundaries are weakened during cancer progression - Many progression -regulated genes exhibit changes in distal enhancer histone modifications that are bridged by stable chromatin loops - Changes in enhancer activity or subtle changes in chromatin contacts can rewire enhancer-promoter connections to facilitate changes in gene expression - Prominent changes in chromatin loops occur at a small subset of diYerentially regulated genes during progression .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 3

The eukaryotic genome is highly organized in the cell nucleus. Amongst the most prominent structural features are kilobase-sized chromatin loops, medium-scale topologically associating domains (TADs) and higher -order compartments (1–5). How chromatin organization contributes to epigenetic control of gene regulation, including in physiological and pathological settings, such as cancer, remains only partially understood (6–9). A prominent mechanism to generate higher order chromatin structures is loop extrusion (10– 12). During this process, the multi-component cohesin complex is loaded onto chromatin and, using its intrinsic molecular motor activity, extrudes chromatin bidirectionally along the genome to form a loop or a domain until it encounters the major chromatin architectural protein CTCF bound to convergently oriented binding sites. The encounter of cohesin with bound CTCF stalls the extrusion process and generates a chromatin loop or TAD. While loop extrusion has universally been implicated in formation of chromatin loops and domains, formation of chromatin features by other mechanisms have also been observed, especially at a smaller scale (13–16). A common property of higher order chromatin folding is that the resulting loops, domains and compartments, bring distal genome elements into spatial proximity and into proximity of regulatory elements to their target genes has been implicated in gene regulation (2). For example, it has been suggested that one function of TADs is to facilitate the interaction of regulatory elements, particularly gene enhancers, with their target genes located within the same TAD (17–21). Similarly, long-range interactions via chromatin loops are thought to be essential at some loci to bring gene enhancers into proximity to their target promoters (22– 26). However, regulation of many gene loci also appears independent of chromatin organization, and enhancer-promoter proximity often does not correlate with gene activity (27–30). In fact, acute depletion of cohesin revealed genome-wide disruption of chromatin organization but surprisingly limited impact on gene expression (31). A possible explanation for these divergent findings is that chromatin organization may be functionally more relevant to bring about changes in gene activity rather than maintenance of gene expression as suggested by several cohesin depletion studies (30,32,33). Genome organization is likely relevant for cancer and its progression. Mutations in loop extrusion machinery, such as cohesin and the cohesin processivity factor NIPBL or at CTCF binding sites have been reported in many cancers (6,7,34–36). In addition, many structural variants (SVs) such as deletions , duplications, and translocations, have been documented in various cancer subtypes where SVs and the ensuing reorganization around them can lead to aberrant gene regulation (37–39). Despite these observations, the full extent of genome reorganization during cancer progression, and its functional consequences, remains largely unknown. Several studies primarily focused on large scale reorganizations have found changes in higher -order chromatin organization such as chromosome clustering and dynamic compartments, some .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 4 of which correlated with changes in diYerentially expressed oncogenes and enhancers (40– 48). Analysis of TADs in various cancers have found mixed results, with some studies pointing to increased TADs and gained boundaries and others observing more stable TAD organization or weakened boundaries (42,46,47,49,50). Furthermore, cancer-associated structural variants, such as chromosomal translocations, have been related to altered gene expression, for example via enhancer-hijacking (51,52). However, how local chromatin loops and TADs are restructured during oncogenic reprogramming and how these changes relate to cancer-associated gene expression has not been well documented. We address the question of how local and global chromatin organization changes during cancer progression, and how they related to cancer gene expression programs, by generating high-resolution Micro -C maps in the well-established MCF10 breast cancer progression model (53). This cancer model consists of three epithelial cell lines that were all originally derived from the same non -cancerous patient (54). MCF10A is an adherent epithelial pre- malignant cell line that spontaneously immortalized from the initial patient sample. Pre- malignant MCF10AT1 cells were derived from MCF10A cells by overexpressing a mutant Ha- Ras oncogene (55). When xenografted into immunocompromised mice , these cells form precancerous lesions . Metastatic MCF10CA1a cells are derived from metastatic tumors generated from xenografted MCF10AT1 cells (56). The MCF10 progression series represents a spectrum of cells that share a similar genetic background but are increasingly more cancerous and they have been widely used to study the genetic and epigenetic changes that occur during cancer progression and epithelial-mesenchymal transitions (EMT) (41,57–65). In this study, comparing fine-scale chromatin organization and other epigenetic features in the MCF10 cancer progression model has allowed us to identify stage-specific diYerences in genome reorganization and relate them to changes in gene expression, including of cancer progression-associated genes . Our results provide novel insights into the principles of chromatin-mediated gene regulation and into the dynamic structure-function relationship potentially contributing to genome regulation in cancer. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 5

Mapping global and local genome organization across breast cancer progression To understand how genome organization changes during cancer progression, we generated high-resolution (5 kb ) genome-wide maps of chromatin contact s using Micro -C in the MCF10A, MCF10AT1, and MCF10CA1a cancer progression series (Fig. 1A, Supp. Fig. 1 A). We obtained high-quality data with at least 1 billion Micro-C contacts per cell line , spread across 2 biological replicates with 4 technical replicates each (Supp Table 1). We identified features of chromatin organization at several levels, including large -scale reorganizations of compartments, medium -scale changes in topologically associating domains (TADs), and fine-scale changes in chromatin loops (Fig. 1B). We assigned A and B compartments using CALDER at a resolution of 10kb , TADs using SpectralTAD combined with FAN-C boundary insulation score calculations at 10kb, and loops using SIP at 5 - and 10kb resolution (see Methods for details) . Each cell type had similar percenta ges of the genome assigned to the active A (47.1-50.2%) or inactive B (49.7-52.9%) compartment, with the two cancer cell types MCF10AT1 and MCF10CA1a having a slightly higher proportion of A compartment designations (Fig. 1C). Similarly, we detected a similar number of TADs (between 7,459-7,825) and chromatin loops (between 15,713-17,332) in each cell line (Fig. 1D-E). Although the three cell lines are karyotypically similar due to their shared genetic background, they contain large-scale chromosomal duplications and translocations which were identified by SKY karyotyping analysis ( Supp. Fig . 1B), and numerical chromosome aberrations based on SKY and Micro -C sequencing depth analysis were included in the analysis of all chromatin features (see Methods; Supp. Fig. 1C ). After this correction, chromatin loop counts showed a high degree of reproducibility between technical and biological replicates (Supp. Fig. 1D ). We used this deep dataset to characterize structural features that are reshaped during cancer progression. Cancer progression reorganizes compartments, TADs, and chromatin loops Comparative analysis across the three cell types identified significant changes in all major chromatin features during cancer progression (Fig. 1, Supp. Fig. 2A-B, Supp. Table 2-4). At a large-scale, we detected changes in compartmentalization. We observe a general shift towards the more active A compartment in early cancer progression, with a larger portion of genomic regions becoming more A -like (31. 0%) compared to more B -like (26.0%) in the transition from MCF10A to MCF10AT1, while these changes are more balanced in the later transition from MCF10AT1 to MCF10CA1a (30.0% and 30.6% , respectively ). Interactions within the most A-like and B -like compartments were predominant in the pre -cancerous MCF10A cells (Fig 1F). However, in MCF10AT1 and MCF10CA1a, stronger interactions appear between more intermediate regions, suggesting a greater degree of intermixing that is consistent with increased compartmental heterogeneity which appears to occ ur early during cancer progression (Fig. 1F) (42,66). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 6 TAD boundaries represent genomic regions where upstream and downstream sequences are partially insulated from one another, with fewer contacts between them than within (1,5,19). We detected a total of 13,231 TADs across all three cell types, with 17,097 unique boundaries. TADs detected range in size from 190kb to as large as 3 .8 Mb, with a mean of 663kb and a median of 460kb (Supp. Fig. 2C). Assessing changes in insulation score (IS) at TAD boundaries revealed 3,392 (19.8%) boundaries where the degree of insulation changed significantly over the course of cancer progression . Because individual boundaries may be simultaneously used by multiple TADs, the total number of TADs which changed during progression is 5,084 (38.5%) (Fig. 1G). There are nearly three times as many boundary changes between later stages in cancer progression (1,693 diYerential boundaries between MCF10AT1 and MCF10CA1a) than early stages (567 between MCF10A and MCF10AT1 ). Interestingly, TAD boundaries that gained or lost insulation during progression showed a significant enrichment for weakened boundaries (7 1.2%) with far fewer boundaries exhibiting increased insulation strength as cancer progressed (28.8%; permutation test, Supp. Fig. 2D). This late-stage weakening of boundaries may reflect a more heterogeneous cell population as cancer progresses (6,67,68). Chromatin loops are formed by two distal genomic regions that are in more frequent contact than their surrounding or intervening sequences, indicated by higher contact frequency (1,3,10,11). We found 29,205 chromatin loops across all three cell lines, ranging in size from 50 Kb to 2 Mb, with a mean of 402kb and median of 270 Kb in length (Supp. Fig. 2C). 77.6% of loop anchors coincided with CTCF peaks, representing 95.0% of loops with at least one anchor bound by CTCF , and CTCF -bound loops were stronger and longer than non -CTCF loops (Supp. Fig. 2E-F). Loop boundaries often overlapped with TAD boundaries with 52.4% of TADs consisting of loop domains across all cell lines. However, a majority (73. 0%) of chromatin loops did not include TADs (Supp. Fig. 2G). TADs without loop interactions at their boundaries tended to be larger, while loops without TADs can be found at all sizes but are enriched for shorter loops (Supp Fig. 2H). To identify loops that changed significantly during cancer progression, we assessed changes in contact frequency among every loop in each cell type , correcting for karyotypic diYerences that result in diYerences in coverage between cell lines (see Methods) . We identified 1, 469 chromatin loops that change significantly over the course of cancer progression, including both weakened and strengthened contacts ( Fig. 1H), representing 5.0% of all identified loops. Unlike TADs there was a more balanced number of changes between early ( 1,004 diYerential loops between MCF10A and MCF10AT1) and late ( 1,204 between MCF10AT1 and MCF10CA1a) progression stages, as well as between strengthened (679 loops, 4 6.2% of all diYerential loops ) and weakened loops (790 loops, 53.8% of all diYerential loops ). Interestingly, only a small portion (1 9.0%) of diYerential loops were accompanied by changes in CTCF binding ( Supp Fig. 2I). Motif analysis of diYerential loop anchors revealed only weak motifs of various transcription factors enriched at the boundaries of gained and lost loops, although occupancy did not appear high enough to explain most of the changes we observe ( Supp Fig . 2 J). Weakened loops were often associated with a decrease in H3K27ac, a mark of active enhancers, consistent with the .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 7 notion that active enhancers can help recruit loop extrusion machinery ( Supp Fig. 2K; see below) (69,70). Taken together, these results demonstrate significant global changes in genome organization during cancer progression across multiple scales from chromatin compartments to loops. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 8 .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 9 Figure 1. Reorganization of compartments, TADs, and loops during breast cancer progression (A) A diagram of the experimental design. Three epithelial cell lines represent various stages of breast cancer progression; MCF10A are non -cancerous, MCF10AT1 are pre -malignant, and MCF10CA1a are metastatic. In each cell line we generated 5kb resolution Micro -C to identify features such as compartments, topologically associating domains (TADs), and chromatin loops. We overlapped these features with functional changes in gene expression from RNA -Seq, histone modifications and CTCF binding from ChIP -Seq, and chr omatin accessibility from ATAC-Seq. (B) Micro-C maps of a 2 Mb region of chromosome 1 in MCF10A (non-cancerous), MCF10AT1 (pre -cancerous), and MCF10CA1a (metastatic) cells at 5 kb resolution. Each map has annotations for loop calls, both static (black boxes) and diYerential (red boxes). Below e ach map is a track indicating compartment calls from CALDER (dark red is most A-like, dark blue is most B-like) as well as insulation scores tracks with static (grey) and diYerential (red) boundaries marked. Ribbons indicate TAD calls for each cell type. (C) Lengths of the genome assigned to each compartment in each cell type. (D) TAD and (E) loop calls from each cell type, colored by the number of maps they were initially detected in. (F) Saddle plots of interactions between regions of diYerent compartments in MCF10A, MCF10AT1, and MCF10CA1a. Bottom plots indicate the average eigenvector value for each compartment ventile. Plots shown are for chromos omes 2, 12, and 17 (see Methods). (G) Left; DiYerential TAD boundaries clustered by their timing of change, depicted in line plots and heatmap. Right; aggregate plots of weakened and strengthened TAD boundaries (n=100). (H) Left; DiYerential chromatin loops clustered by their timing of change, depicted in line plots and heatmap. Right; aggregate plots of weakened and strengthened loops (n=100). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 10 Persistent chromatin loops connect c ancer progression-regulated genes with distal regulatory features To explore the functional role of chromatin loops, we related them to gene expression and potential regulatory regions. We identified 17,185 expressed genes across all three cell types using RNA -Seq (see Methods) and 52,953 potential enhancer s as defined by overlapping histone H3K27ac ChIP-Seq and ATAC-Seq accessibility peaks, commonly used to identify enhancer regions (see Methods) (71–73). Approximately half of chromatin loops featured some combination of active gene promoters and enhancers within 10kb of loop anchors (Fig. 2A). We found that chromatin loops that connect two features (either enhancers or promoters; mean length 251-280 kb) are typically shorter than those that contain only one feature (mean length 338-372 kb) or none (Fig. 2B; mean length 495 kb; T-test p-value for all comparisons < 2.2e-16), with promoter-promoter loops being the smallest on average (mean length 251 kb). Interestingly, enhancer-promoter and enhancer -enhancer loops (mean counts 8.2, 8.2 , respectively ) were stronger than promoter-promoter loops (mean counts 7. 3) despite being longer on average, suggesting that epigenetic signatures associated with active enhancers may support stronger contacts (Fig. 2C; T-test p-value for both comparisons < 2.6e-7). We then explore d how diYerentially expressed genes relate to long-range chromatin interactions. We identified 8, 840 diYerentially expressed genes across all pairwise comparisons in the MCF10 cancer progression (Supp Fig. 3A-B). A similar number of genes changed in later stages of cancer development (4,968 between MCF10AT1 and MCF10CA1a) compared to early progression (4,773 between MCF10A and MCF10AT1) . Reassuringly, as expected from previous studies (41,64), genes associated with epithelial morphogenesis and cell adhesion were upregulated early during progression , whereas regulation of diYerentiation, tissue development, metabolism, and signal transduction genes was observed during later stages of progression (Supp Fig. 4B). These changes are consistent with the development of an intermediate and diverse pre -cancerous state early on during progression, while late changes are known to facilitate metastasis and support the epithelial-to-mesenchymal-like transition observed phenotypically among the progression series (Supp Fig. 3C-D) (74–77). To understand the regulatory modes of action of genes which were diYerentially expressed during cancer progression, we determined if they had gained or lost the activity-associated H3K27ac mark at their promoters or at distally looped enhancers. We found that while many genes only feature d a corresponding change in H3K27ac at their promoter (53.9% of upregulated and 28.6% of downregulated genes), a large percentage also showed changes in distal enhancer activity (28.8% of upregulated and 17.4% of downregu lated genes) , suggesting that enhancer loops may be playing an important functional role in control of these genes (Fig. 2D, F ). Comparing the direction of fold -change for genes and promoter H3K27ac, distal H3K27ac, or contact frequency with distal enhancers using Fisher’s Exact test revealed odds ratios significantly higher than 1 for all comparisons (9.4, 2.2, and 1.2, respectively for changes between MCF10A and MCF10CA1a), but that there was a stronger .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 11 association with promoters than enhancers or loop strength (Supp. Fig. 3E-F). This trend was similar for genes that were diYerentially regulated both early and late, suggesting that the role of chromatin loops is consistent across all stages of cancer progression. Comparing the changes in acetylation at all gene promoters and distal regulatory regions revealed that upregulated genes exhibit a significant increase in H3K27ac at distally looped enhancers, as well as a significant loss of repressive H3K27me3 marks ( Fig. 2E ; T-test p- value for both < 2.2e-16). This trend is less clear for downregulated genes, which feature both gained and lost H3K27ac at both enhancers and promoters as well as both gained and lost H3K27me3 at distal regions (Fig. 2G ). These results suggest that the static chromatin structures observed during the cancer progression process contribute to the control of diYerentially regulated genes, particularly among upregulated genes. Given that only 5% of loops changed significantly during progression (see Fig. 1) , it is not surprising that only a small percentage of diYerentially expressed genes exhibited significant changes in chromatin contact s with distal enhancers (2. 1% of upregulated and 2. 2% of downregulated genes; Fig. 2D, F). However, on average t here is no significant change in contact frequency between gene promoters and distal enhance rs. This trend was similar between both early- and late-regulated genes (Supp. Fig. 3G). Most diYerentially expressed genes are in regions where chromatin structure is essentially stable, reinforcing that persistent structural changes are not universally required for gene regulation. For example, the SPRY1 gene, which regulates cell growth and diYerentiation and has been shown to be upregulated in triple-negative breast cancer tumors (78), is upregulated between MCF10AT1 and MCF10CA1a, and is statically looped to distal enhancers that gain H3K27ac (Fig. 2H). Similarly, the SCNN1G gene, which encodes for a subunit of a sodium channel and is suppressed in head and neck squamous cell cancer (79), is downregulated between MCF10AT1 and MCF10CA1a, and is statically looped to distal enhancers that lose H3K27ac (Fig. 2I). In both cases, the contact frequency remains relatively constant despite changes in distal enhancer acetylation. Taken together, our results show that changes in gene expression are not necessarily accompanied by structural changes, and they suggest that stable chromatin loops may facilitate cancer progression by providing a pre -existing structure through which diYerentially regulated distal enhancers can act on target genes. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 12 Figure 2. Persistent chromatin loops connect diVerentially expressed genes to distal enhancers. (A) Percentages of loops designated as either promoter -promoter, enhancer -promoter, enhancer-enhancer, or single-sided promoter or enhancer loops. (B) Distributions of loop sizes by enhancer/promoter designations. P-values represent T-tests comparing the means of diYerent loop classes. Boxplots show the median (middle line), 25th and 75th quartiles (box perimeters), and range excluding outliers (dashed line whiskers). Outliers are defined as values that are over 1.5 times the interqua rtile range beyond the box bounds and are excluded from these plots . (C) Distributions of loop strength by enhancer/promoter designations. (D) Percentages of upregulated genes that have gained H3K27ac at promoters, .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 13 distal enhancers, both, or gained loops. P-values represent T-tests comparing the means of various loop sets . Non -significant (n.s.) represents p -values above 0.0 5. (E) Log2 fold - change of distal H3K27me3 (grey), distal H3K27ac (red), promoter H3K27ac (orange), gene expression (yellow), and loop strength (blue), when overlapped. Grey dots indicate features that do not change significantly, while colored points are significantly diYerential features. Boxplots are defined as in (B). P-values represent T-tests comparing the means of each class to 0. Non-significant (n.s.) represents p -values above 0.01. (F) Percentages of downregulated genes that have gained H3K27ac at promoters, distal enhancers, both, or gained loops. (G) Log2 fold -change of distal H3K27me3 (grey), distal H3K27ac (red), promoter H3K27ac (orange), gene expression (yellow), and loop strength (blue), when overlapped. Boxplot details as defined in (E). (H) An example of an upregulated gene (SPRY1) connected to gained enhancers by static loops. Black boxes show loop annotations. Red compartment tracks indicate A compartments, while blue indicate s B compartments. In CTCF signal tracks, red highlights indicate diYerential CTCF peaks. In H3K27ac and ATAC - Seq signal tracks, red highlights indicate diYerential enhancers as determined by changes in H3K27ac. Genes highlighted in black are diYerentially expressed. (I) An example of downregulated genes (SCNN1G, SCNN1B) connected to lost enhancers by static loops. Plot annotations are as described in (H). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 14 Changes in enhancer acetylation and enhancer-promoter contact are associated with changes in gene expression To begin to distinguish the eYects of enhancer-promoter contact and chromatin looping from enhancer activity eYects, we compared gene expression changes at looped and non- looped enhancer-promoter pairs. To do so, we used the activity-by-contact (ABC) model to predict function al enhancer-promoter pairs. ABC combines estimates of enhancer accessibility and activity from ATAC -Seq and H3K27ac ChIP -Seq with enhancer -promoter contact frequency from Micro-C data to generate an estimate of the likelihood of functional enhancer-gene interactions (see Methods) (80). This method allowed us to identify distal regulatory regions that are functionally linked to gene promoters without specifically requiring overlap with chromatin loops. For example, an enhancer and promoter may be in high contact as measured by Micro-C because they overlap with loop anchors, or because they are at close genomic distance. Applying the ABC model to our data identified 150,056 potential enhancer-promoter pairs across all three cell types. Of these, 53.4% are also promoters of other genes, 23.7% are within the bodies of other genes, and 22.9% are intergenic, and range in distance from 750 to 5M base pairs away from target gene promoters (Supp Fig. 4A). To better understand the relationship between contact frequency, enhancer activity, and gene expression, we asked how changes in enhancer activity or contact relate to gene expression at target promoters. Pairwise comparison of the MCF10 progression lines indicated that changes in both contact frequency and enhancer activity appear to drive changes in enhancer -promoter networks predicted by ABC (Fig. 3). For example, observing potential enhancers with changes in H3K27ac between MCF10CA1a and MCF10A reveals that these enhancers also exhibit a change in contact frequency and are associated with upregulation of target genes (Fig. 3A). We also find that changes in contact frequency are associated with increases in H3K27ac and correlate with higher gene expression (Fig. 3B). These results show that not only changes in either contact frequency and enhancer activity correlate with increased gene expression, but they also correlate with each other, suggesting a potentially linked functional role during enhancer-promoter communication. To then relate enhancer -promoter pairs to chromatin loops and to orthogonally assess whether chromatin loops are acting as a functional bridge for active enhancers, w e compared looped and non-looped enhancer-promoter pairs. Enhancer-promoter pairs that have changes in distal H3K27ac and are supported by chromatin loops correlate d with changes in gene regulation ( Fig. 3C). This eYect was stronger than distance-matched non- looped enhancer-promoter pairs , but similar to contact -matched non -looped pairs , suggesting that increased contact frequency caused by loop extrusion may contribute to the stronger correlation with gene expression ( Fig. 3D-E). Contact-matched non-looped pairs were closer on average to the looped pairs of similar contact frequency, while distance - matched non-looped pairs were in less -frequent contact than looped pairs of similar genomic distance (Fig. 3F-G). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 15 Comparing the distributions of target gene fold -change for these various sets of enhancer - promoter pairs reveals several trends (Fig. 3H ). First, pairs with significant changes in contact have a larger mean gene fold-change than pairs with significant changes in activity, suggesting that either can contribute to changes in enhancer -promoter functional pairing but that contact may have a particularly strong impact. Second, looped enhancer-promoter pairs have a comparable or larger mean gene fold-change to pairs with changes in activity or contact, suggesting again that chromatin loops may support functional enhancer-promoter pairs. Lastly, looped pairs have a similar mean gene fold-change as contact-matched pairs, which in turn have a higher mean gene fold-change than distance-matched pairs, suggesting that the increased contact frequency that chromatin loops provide to enhancer -promoter pairs may be a driving force for the functional connection. These tr ends hold true for all (tested) pairwise comparisons between cell types (Suppl Fig. 4B-C). Taken together, these findings demonstrate that not all gene regulatory changes are accompanied by chromatin reorganization, but when it occurs reorganization appears to facilitate functional changes. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 16 Figure 3. Changes in enhancer acetylation or enhancer -promoter contact are associated with changes in gene expression. Boxplots of distal enhancer H3K27ac (pink), enhancer -promoter contact (blue), and ABC score (purple), as well as gene log2 fold -change (yellow) for enhancer promoter pairs that feature (A) diYerential H3K27ac among enhancers, (B) diYerential enhancer -promoter contact frequency, and (C) diYerential H3K27ac for enhancer-promoter pairs supported by a chromatin loop. Boxplots in (D) and (E) represent sets of non-looped enhancer-promoter .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 17 pairs with diYerential H3K27ac that are matched to the looped set in (C) by contact and distance, respectively. Boxplots show the median (middle line), 25 th and 75th quartiles (box perimeters), and range excluding outliers (dashed line whiskers). Outliers are defined as values that are over 1.5 times the interquartile range beyond the box bounds and are excluded from these plots. P -values represent T -tests comparing the means of values in MCF10A and MCF10CA1a for enhancer activity, enhancer -promoter contact, and ABC Score, and T -tests comparing the mean of the value to 0 for gene log2 fold -change. (F) Contact distribution of all enhancer -promoter pairs (dashed line), compared to the looped enhancer-promoter pairs in (C, solid line), the contact-matched pairs in (D, blue shade), and the distance -matched pairs in (E, grey shade). (F) Distance distribution of all enhancer - promoter pairs (dashed line), compared to the looped enhancer -promoter pairs in (C, solid line), the contact -matched pairs in (D, blue shade), and the distance -matched pairs in (E, grey shade). (G) Summary boxplot of the gene log2 fold -change for the enhancer-promoter pairs previously shown in fig ures (A-E). P-values represent T-tests comparing the means of average gene log2 fold-changes values for diYerent sets of enhancer-promoter pairs. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 18 Changes in chromatin looping are enriched for progression-associated diVerentially expressed genes We next explored how changes in chromatin looping may functionally contribute to gene regulation during cancer progression. To do so, we compared changes in chromatin loop contacts to alterations in expression of progression-associated genes. Overall, while only a small subset of gene promoters overlaps with the anchors of diYerential chromatin loops (507 genes, 3.0% of expressed genes), those that do are significantly enriched for genes that are diYerentially expressed during cancer progression based on permutation analysis (331 diYerentially expressed genes, 65.3% of all diYerentially looped genes; Supp. Fig. 5A). We asked whether there was a relationship between the formation or loss of loops and diYerential expression of these loop-associated genes. We indeed find that diYerential loops are more likely to change in the same direction as the gene (i.e. increased contact frequency with distal regions associated with increased gene expression) (Fig. 4A-B). The fold-change of diYerential expressed genes which also showed diYerential loping were significantly higher than a random sampling of diYerentially expressed genes ( Fig. 4A). For example, of 3,261 genes that were diYerentially upregulated between MCF10A and MCF10CA1a, loops were significantly strengthened at 98 of these genes and significantly weakened at 31. Similarly, of 3,088 downregulated genes, 65 genes overlap weakened loop anchors and 41 genes overlap strengthened loop anchors . In contrast, the number of expected genes at strengthened or weakened loops for a random sampling of genes this size is 38 and 32, respectively (Supp. Fig. 5B-C). We also find a subset of chromatin loops and genes changed in opposite direction s (Fig. 4B). The genes whose changes in expression correlate with changes in looping are enriched for several cancer -relevant pathways, such as morphogenesis, diYerentiation, and proliferation (Fig. 4C) (81). In total, we identified 1 27 unique genes upregulated in areas that experience increased chromatin contacts, either at loop anchors or within existing structures. As an example, the promoter of the COL12A1 gene, which encodes a collagen protein known to be associated with aggressive and mesenchymal phenotypes, overlaps a loop boundary that is very weak in MCF10A cells where COL12A1 is not expressed (82–84). As COL12A1 gene expression is upregulated during progression, contacts increase at a distal region 310 kb away, and H3K27ac and accessibility also increase at these likely distal regulatory regions (Fig. 4D). Similarly, we observe 1 23 unique genes that are downregulated during oncogenic progression. One example is WNT5A which encodes for an important signaling protein and is downregulated in breast cancer and across MCF10A progression (85–87). Similar to COL12A1, the promoter of WNT5A is involved in many diYerential distal regulatory contacts, ranging in distance from 240 to 640 kb ( Fig. 4E). Unlike COL12A1, these contacts are strongest in MCF10A cells and severely weaken in MCF10AT1 and MCF10CA1a cells. Accompanying these changes in contact, the distal regulatory regions that appear to support WNT5A in MCF10A cells become deacetylated and decrease in accessibility. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 19 Figure 4. DiVerential loops are enriched for cancer -relevant diVerentially expressed genes (A) Log2 fold -change of diYerentially expressed genes at the anchors of gained (blue), weakened (green), or static (grey) loops. Boxplots show the median (middle line), 25 th and 75th quartiles (box perimeters), and range excluding outliers (dashed line whiskers). Outliers are defined as values that are over 1.5 times the interquartile range beyond the box bounds and are excluded from these plots. P-values represent T-tests comparing the mean of each set to 0. (B) Bar plot showing the number of diYerentially expressed genes at strengthened or weakened loop anchors. Bar segments are colored by whether the gene is changing the same (blue for upregulated in strengthened loops, green for d ownregulated in weakened .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 20 loops) or opposite (grey) direction as the loop. P -value represents a Fisher’s Exact Test for whether the odds ratio (OR) is greater than 1. (C) GO term enrichments for genes upregulated in MCF10A, MCF10AT1, or MCF10CA1a. Size indicates p-value. Terms are color- coded based on gene type; morphogenesis (purple), proliferation (orange), and cell adhesion (teal). (D) An example of an upregulated gene (COL12A1) with a promoter that overlaps a strengthened loop with distal enhancers. Black boxes show loop annota tions, while red boxes indicate diYerential loops. Red compartment tracks indicate A compartments, while blue indicates B compartments. In CTCF signal tracks, red highlights indicate diYerential CTCF peaks. In H3K27ac and ATAC -Seq signal tracks, red high lights indicate diYerential enhancers as determined by changes in H3K27ac. Genes highlighted in black are diYerentially expressed. (E) An example of a downregulated gene (WNT5A) with a promoter that overlaps with several weakened loops containing distal enhancers that lose H3K27ac. Plots are annotated as in (A). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 21 Genome reorganization occurs at cancer -relevant genes and is accompanied by proximal and distal changes in histone marks Finally, we aimed to comprehensively explore the genes that are diYerentially regulated in areas that also have strong changes in chromatin looping, to better understand the potential regulatory mechanisms at play. A locus -by-locus view of gene and loop fold -change allows us to view the relationship between changes in expression and structure among each pairwise comparison of cells (Fig. 5A-C). While we see that a majority of genes have a corresponding change in looping (i.e. up- regulated genes overlapping strengthened loops), we observe that the percentage of corresponding changes increases in the later stages of cancer progression. For example, the percentage of diYerential loop -gene pairs where the gene overlaps at least one gained loop is 47.5% and 69.9% among genes up-regulated in MCF10A compared to MCF10AT1 and MCF10CA1a, respectively, 66.7% and 79.6% among genes up -regulated in MCF10AT1 compared to MCF10A and MCF10AT1, respectively, and 70.7% and 58.0% among genes up- regulated in MCF10CA1a compared to MCF10A and MCF10AT1 , respectively . This may indicate that the regulatory impacts of changes in chromatin looping occur over longer timescales, or that genes impacted by changes in chromatin structure may be more common in metastatic cells. We do not, however, find any correlation between the magnitude of loop fold-change and gene fold-change. To better understand how changes in looping are related to gene expression, we compared patterns of gene expression, promoter H3K27 acetylation, and distal enhancer H3K27 acetylation or trimethylation at looped genes that change in either the same or opposite direction as the loops they overlap ( Fig. 5D -E, Supp. Fig. 5D-E). These two histone H3 modifications are mutually exclusive and have opposite eYects on gene expression, marking enhancers and silencers, respectively (88). Both modifications are able to impact distal gene expression via chromatin interactions (89). Up-regulated genes between MCF10A and MCF10CA1a show similar epigenetic signatures at both proximal and distal regions, regardless of whether the loop they overlap gets stronger or weaker (Fig. 5D). Promoter regions gain H3K27ac, while distal regions gain H3K27ac and lose H3K27me3. There is however a significant diYerence in the extent of distal changes depending on the loop direction, with strengthened loops exhibiting a significantly higher increase in distal H3K27ac and a decrease in H3K37me3 marks. This behavior further supports the notion that these distally looped regulatory regions are important functional elements. Down-regulated genes show distinctly diYerent signatures (Fig. 5E). Genes that overlap with weakened loop anchors show decreased promoter H3K27ac and distal H3K27ac and increased distal H3K27me3, consistent with signatures typically associated with reduced gene expression (88). Interestingly, genes that overlap strengthened loop anchors show diYerent patterns , with a gain in promoter H3K27ac and loss of distal H3K27me3 repressive marks. We conclude that expression of a subset of progression - associated genes is strongly correlated with loop formation. These trends are similar but weaker for genes that change between diYerent cell types (Supp. Fig. 5D-E). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 22 Taken together, our genome-wide analysis of structural and regulatory changes during MCF10A cancer progression has highlighted hundreds of restructured regions where cancer- relevant genes are diYerentially regulated. These findings suggest that, while relatively rare, changes in chromatin looping may be associated with regulatory changes that drive expression of hundreds of oncogenes during cancer progression. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 23 Figure 5. Progression-associated diVerentially expressed genes exhibit local and distal epigenetic changes at diVerential loops. (A-C) Log2 fold-change of genes (colored dots) and the diYerential loops they overlap with (black/grey dots) for genes and loops that change between (A) MCF10A and MCF10AT1, (B) MCF10AT1 and MCF10CA1a, and (C) MCF10A and MCF10CA1a. Gene labels are below. (D) Log2 fold -change between MCF10A and MCF10CA1a of distal H3K27me3 (grey), distal .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 24 H3K27ac (red), promoter H3K27ac (orange), gene expression (yellow), and loops (blue) among upregulated genes that overlap with gained loops (darker colors) or lost loops (lighter colors). Boxplots are defined as in (A). P-values represent T-tests comparing the mean values of the features at loops that change in the same and those that change in opposite directions from the diYerential genes at their anchors. Non-significant (n.s.) p-values are any p- values above 0.01. (E) Log2 fold-change of distal H3K27me3 (grey), distal H3K27ac (red), promoter H3K27ac (orange), gene expression (yellow), and loops (blue) among downregulated genes that overlap with gained loops (darker colors) or lost loops (lighter colors). P -values represent T-tests comparing the mean values of the features at loops that change in the same and those that change in opposite directions from the diYerential genes at their anchors. Non-significant (n.s.) p-values are any p-values above 0.01. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 25

Multiple levels of chromatin organization and integration with epigenetic parameters contribute to regulation of gene expression. We have identified dynamic chromatin organizational changes on multiple scales across breast cancer progression using the MCF10A model system. By comparing both a pre-malignant and metastatic cell line to non- cancerous epithelial cells we were able to detect both early- and late-stage changes as well as similarities in genome structure and relate them to gene expression changes, including many cancer-relevant genes. We found that c ompartmental shifts occur more often in early stages of cancer development. This behavior is consistent with previous studies that have shown intermixing of chromatin A and B compartments in cancer cells (42,50). The general shift towards more active compartments during cancer progression may reflect a broader epigenetic landscape and the greater heterogeneity in gene expression within cancer cell populations (6,90–92). On a finer scale, structural changes in TADs occur more often during later stages of metastasis. We also find an abundance of weakened TAD boundaries in MCF10 breast cancer progression compared with boundaries that gain insulation. This finding is in line with previous studies which have shown TAD weakening in triple-negative breast cancer (49), but is contrary to other observations which have suggested that prostate cancer is associated with the formation of many additional TADs (46), or findings of minimal changes in TAD structure in colorectal or breast cancer (42,50). The variety of observations on TAD structure in cancer may reflect the diYerences in the types of cancer samples analyzed or the analysis

used. In the MCF10 model, we were able to detect subtle changes in insulation score at TAD boundaries, weakening of which could again support a more heterogeneous structure in more aggressive cancer stages. Chromatin loops functionally connect gene promoters to distal regulatory elements. In the MCF10 model, many genes diYerentially regulated during cancer progression are associated with chromatin loops shared between all three cell lines , but show changes in distal enhancer H3K27ac and H3K27me3 . These trends were stronger with up -regulated genes, suggesting that we may need to explore diYerent epigenetic signatures to better understand how chromatin structure may influence repression . Activity-by-contact analysis showed that l ooped enhancer -promoter pairs also exhibit greater correlation between distal enhancer H3K27ac and gene expression than non-looped enhancers due to their increased contact frequency . These findings suggest that persistent chromatin loops that do not change during cancer progression nevertheless have functional relevance and that they do so by bridging enhancers to target gene promoters. Activity-by-contact analysis also revealed that subtle changes in chromatin contact can contribute to the rewiring of enhancer -promoter regulatory connections . Enhancer- promoter pairs that exhibit c hanges in contact correlated with stronger changes in target gene expression than those with only changes in activity, in line with the concept that contact .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 26 with distal regulatory elements is an important component of gene regulation (93). We also found that changes in chromatin contact are associated with more modest changes in activity, and vice-versa. This correlation between enhancer activity and enhancer-promoter contact further points to a functional link between the two . Together, these results suggest that both contact frequency and activity contribute to enhancer-promoter connectivity. Strong changes in contact at chromatin loops are relatively rare across cancer progression, but many clear examples do exist, and they are notably enriched for diYerentially expressed genes. Interestingly, only a small portion of diYerential loops can be explained by changes in CTCF binding at anchors, suggesting other forces may be influencing their contact frequency. Loss of H3K27ac can be seen at the anchors of lost loops, consistent with the idea that active enhancers can help recruit cohesin to chromatin (69,70). Therefore, some weakened loops might be explained by a loss of H3K27ac which leads to a loss of cohesin at that region. We also found that genes at diYerential loops are more likely to be diYerentially regulated in the same direction as the change in loop strength. This supports the notion that a subset of genes may be regulated in a structure-dependent manner. This interpretation is in line with previous observations which have shown a subset of genes to be sensitive to cohesin or NIPBL depletion which disrupts chromatin loops (11,17,20,25,32,94–96). Importantly, these findings suggest this subset of structure -sensitive genes may include many with direct relevance to breast cancer progression. Epigenetic signatures at gene promoters and distal regions diYered based on the direction of gene change. Up -regulated genes consistently showed a gain of active -associated H3K27ac marks at both promoters and distal regions, and a loss of distal repressive H3K27me3 marks. These changes were shared between static, strengthened, and weakened loops, although up -regulated genes at strengthened loops had stronger distal changes. These findings are consistent with the loops functionally supporting interactions with distal enhancers via increase contact . However, down-regulated genes showed more complex patterns. Fewer down -regulated genes could be explained by changes in H3K27ac or H3K27me3 at the promoter or distal regions, with over half of down-regulated genes having no clear epigenetic driver, compared to only 15% of up -regulated genes. This suggests that these histone marks are not suYicient to explain down -regulated genes as well as they can explain up -regulated genes . Down-regulated genes at diYerential loops also showed opposite patterns based on the direction of loop change; weaken ed loops showed loss of distal H3K27ac and gain of H3K27me3 consistent with an inactivated enhancer, while strengthened loops showed the opposite . Additional studies will be required to fully understand the repressive mechanisms in this system and how they relate to chromatin structure. Our study has several limitations. Based on its largely correlative nature, we are unable to determine the causal relationships of chromatin structure changes at the loci of diYerentially expressed genes. Follow-up studies exploring any of the hundreds of gene loci .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 27 identified here as possibly being contact -dependent could help elucidate these relationships. Additionally, as a population -level assay, Micro -C is only able to provide aggregate data across an entire population of cells. To address how cell -to-cell heterogeneity contributes to some of the functional relationships we observe, and whether that heterogeneity is occurring at a cellular or population level, we would need to apply single-cell sequencing or imaging-based approaches. These questions will be the su bjects of future studies. Taken together, using the powerful MCF10 breast cancer model, we have generated a rich genomic dataset of structural and functional changes in the genome during breast cancer progression. Our data uncovers new insights into the structure function relationship in gene regulation and into the role of genome organization during malignant breast cancer progression. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 28

Cell culture MCF10A cells were obtained from AATC (CRL-10317). MCF10AT1 (MCF-10AneoT) and MCF10CA1a (MCF10CA1a.cl1) cells were obtained from the Karmanos Cancer Institute. All cells were received frozen , thawed in media and grown to a confluence of 80% after 2-5 passages. Stock solutions were frozen down to be used for all experiments. MCF10A and MCF10AT1 cells were cultured in standard high calcium medium with growth factors, consisting of DMEM/F12 (Invitrogen, 21041025) supplemented with 1.05 mM CaCl2, 10 mM HEPES, 10 ug/ml insulin (Sigma, I1882), 20 ng/ml EGF (Peprotech, AF -100-15), 0.5 ug/ml Hydrocortisone (Sigma, H0135), 100 ng/ml cholera toxin (Sigma, C8052), and 5% horse serum (Invitrogen, 16050122). MCF10CA1a cells were cultured in standard high calcium medium without growth factors, consisting of DMEM/F12, 1.05 mM CaCl2, 10 mM HEPES, and 5% horse serum. Karyotypic analysis Metaphase chromosomes were prepared by incubating cells with 100 ng/ml Colcemid (Roche, Brighton, MA) for two hours, followed by mitotic shake -oY. The mitotic cells were then treated with a hypotonic solution (0.075M KCl) for 20 minutes at 37°C. After this treatment, the cells were centrifuged, the supernatant was extracted, and the cells were initially fixed with a methanol/acetic acid solution (3:1). This step was repeated three times. Finally, the cells were fixed onto a slide using a humidity -controlled Thermotron (Thermotron, Holland MI). SKY probes were purchased from Applied Spectral Imaging (Carlsbad, CA) and hybridized onto slides that were aged at 37°C for 3 days. Detection was then carried out according to the protocol provided by Applied Spectral Imaging, using the following purchase d antibodies: Avidin Cy5 (Rockland Immunochemicals, Limerick, PA), Mouse Anti -Digoxin antibody (Sigma-Aldrich), and a goat anti -mouse antibody conjugated to Cy5.5 (Rockland Immunochemicals). Slides were then mounted and counterstained with DAPI (Vector Laboratories, Newark, CA). Hybridization occurred over a period of two days at 37°C. Approximately 20-25 metaphases were imaged and karyotyped using the ASI GenASIS 8.2.2 software on an Olympus BX63 microscope (Evident, Tokyo, Japan) equipped with a Spectral Cube (Applied Spectral Imaging, Carlsbad, CA). Micro-C library preparation Cells were froze n and pellets of 1M cells were used for Micro -C library preparation. The Micro-C library was prepared using the Dovetail® Micro -C Kit according to the manufacturer’s protocol. Briefly, the chromatin was fixed with 0.3M disuccinimidyl glutarate (DSG) and 37% formaldehyde in the nucleus. The cross-linked chromatin was then digested in situ with micrococcal nuclease (MNase). Following digestion, cells were lysed with 20% SDS to extract the chromatin fragments and the chromatin fragments were bound to Chromatin Capture Beads. Next, the chromatin ends were repaired and ligated to a .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 29 biotinylated bridge adapter followed by proximity ligation of adapter -containing ends. After proximity ligation, the crosslinks were reversed, the associated proteins were degraded, and the DNA was purified then converted into a sequencing library using Il lumina-compatible adaptors. Biotin-containing fragments were isolated using streptavidin beads prior to PCR amplification. Each library was sequenced on an Illumina Novaseq platform to generate 240- 340 million 2 x 150 bp read pairs (average depth of 280M r eads). Mico-C libraries were prepared by Dovetail Genomics (Scotts Valley, CA). RNA-Seq library preparation Total RNA was isolated from cells using Trizol (Life Technologies) and purified using the Direct-zol RNA kit (Zymo Research, Irvine, CA, USA: R2050). RNA quality and quantity were assessed using the RNA 6000 Nano Kit with the Agilent 2100 Bioanalyzer (Agil ent Technologies, Santa Clara, CA). RNA quantity was further assessed using a Nanodrop2000 (Thermo Scientific, Lafayette, CO) and Qubit HS RNA assay (Thermo Fisher Scientific). Total RNA was depleted of ribosomal RNA (New England Biolabs, NEB #7400), reverse transcribed using a random hexamer strategy, and strand -specific adapters were added following the NEBNext Ultra II RNA -seq library prep kit (New England Biolabs, E7770). Paired -End sequencing was used to generate high depth coverage ranging from 80 to 120 million paired- end reads. ChIP-Seq library preparation ChIPseq for CTCF (Cell Signaling Technology, catalog number 3418) and histone marks H3K27ac (Abcam, ab4729) and H3K27Me3 (Abcam ab6002). Independent biological replicates of each cell line (MCF10A, MCF10AT1, MCF10CA1a) were performed as described, including the optional step of snap freezing of chromatin and excluding the optional third washing step (97). Additionally, the chromatin was precleared with Pierce protein A/G beads for 2 hours at 4°C prior to incubation with antibodies. For CTCF we used 20ul of antibody (Cell Signaling Technologies, 3418S) and 150ug of chromatin for each sample. For H3K27ac 10ug of chromatin was used and 2ug of antibody, while for H3K27me3 15ug of chromatin was used with 4ug of antibody. ATAC-Seq library preparation The OMNI -ATACseq protocol was followed as previously describe d, with an optimized 5 minutes of nuclear extraction rather than 3 minutes (98,99). Micro-C processing Micro-C data from each technical replicate (library) was processed from raw reads into contact maps using guidelines outlined by Dovetail Genomics (100). Paired reads were aligned to the hg38 human genome assembly (NCBI GRCh38) using BWA mem (version 0.7.17; settings: -5SP -T0) (101). Pairtools (version 0.3.0) was then used to parse ligations, sort reads, and remove PCR duplicates using the parse (settings: --min-mapq 40 --walks- policy 5unique --max-inter-align-gap 30) , sort, and dedup (settings: --mark-dups) commands (102). Alignment files were generated using pairtools split to create .pairs and .sam files, and samtools view ( version 1.17; settings: -bS) to create .bam files (103). Final .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 30 .bam alignment files were sorted and indexed using samtools sort and index commands. The .bam files were used to extract QC metrics using a script from Dovetail Genomics, and calculate complexity using preseq lc_extrap (settings: -bam -pe -extrap 2.1e9 -step 1e7 - seg_len 1000000000). Pairs files were used to generate contact maps using juicer_tools pre (version 1.22.01), and normalized using SCALE (104). For merged contact maps , we first merged pairs files using pairtools merge, and then ran juicer_tools pre command on the resulting .pairs files. In total, we generated contact maps for each library (technical replicate s), each sample (biological replicates), each cell type, and we created one fully merged map including all reads for a precise map of shared features. Micro-C copy number variation analysis Structural variant and copy number variations were detected from contact maps using NeoLoopFinder calculate-cnv at 1Mb resolution (-e uniform) (105). Denylists of regions to ignore for feature calling were generated based on regions where SCALE normalization factors were unable to be calculated at 5kb or 10kb , ignoring single bins and merging continuous areas within 100 kb. Activity-by-contact analysis also combined this denylist with the ENCODE denylist for hg38 (106). Micro-C compartment calling Compartments were called using CALDER (version 2.0; R version 4.2.1) at 10kb resolution and SCALE normalization (107). We also used FAN -C (version 0.9.21) to calculate eigenvectors at 10 0kb using SCALE normalization to build saddle plots, and oriented eigenvectors manually based on overlap with active genes (108). Micro-C topologically associated domain calling Topologically associated domains (TADs) were called using SpectralTAD (version 1.18.0) at 10kb resolution with SCALE normalization, a window size of 300, and a level of 3, with a quality filter applied (109). We then created a unified TAD list by merging TADs that overlap at both ends between cell types, with a maximum gap of 10kb (1 pixel). We then combined this analysis with continuous insulation score (IS) calculations from FAN -C insulation command at 10kb resolution with SCALE normalization . We then used the FAN-C boundaries command to detect IS boundaries and only kept TADs that also overlapped with an IS boundary. Micro-C chromatin loop calling Chromatin loops were called using SIP (version 1.6.2), run at 5kb and 10kb and then merged to 10kb (-g 2 -fdr 0.05). Loops were called in each cell type map using SCALE normalization, in addition to the combined map, and then merged similarly to TADs to create a unified loop list (110). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 31 RNA-Seq processing All RNA-seq data was analyzed using the nf-core/rnaseq pipeline (111). Adapter and quality trimming was performed with Trim Galore (112). Reads were aligned to the hg38 reference genome using STAR (113) and gene expression was quantified using Salmon (114). Differentially expressed genes were called using DESeq2 (115). An adjusted p-value of 0.05 and a log2fold change of 1 were used as thresholds to select significant differential expression. ChIP-Seq processing and peak calling Adapters were cut (cutadapt v1.11) and low quality reads trimmed (Galaxy FASTQ Quality Trimmer 1.0.0; window 10, step 1, minimum quality 20). Reads were mapped to the human genome (hg38 canonical) using STAR version 2.4 (113) with splicing disabled ( – alignIntronMax 1). Enriched regions (narrowPeak calls) for each replicate were generated using MACS2 (Feng et al., 2012) and replicates were then evaluated using deepTools (116) to correlate alignments and IDR (117) to evaluate peak call reproducibility. After pooling replicates, MACS2 (118) was used on H3K4me3 to call narrowPeak at high stringency (P - value <10e-5), these peaks were further filtered according to IDR cutoffs. FE wiggle tracks were generated using MACS2’s bdgcmp and UCSC’s bedGraphToBigwig utility.ChIP -seq datasets have been deposited in the Gene Expressions Omnibus (GEO) under accession code GSE98551 for CTCF and XYZ for H3K27ac and H3K27me3. ATAC-Seq processing and peak calling Read trimming and quality filtering was performed using FastQC (119) and TrimGalore (112). The filtered fastq were then downsampled to approximately 50 million reads per replicate. Reads were aligned to the hg38 reference genome using Bowtie2 (120). Mitochondrial, Multi-mapped, and low quality reads, and duplicates were removed using samtools (103). MACS2 (118) was used to call narrowPeaks, followed by IDR (117) to generate sample level peak sets. Enhancer and promoter definitions Gene promoters were defined as regions between 2000 bp upstream and 500 bp downstream of gene transcription start sites. Potential enhancer regions were identified based on regions that contained both an ATAC - Seq and H3K27ac ChIP -Seq peak . For activity -by-contact analysis, potential enhancers were defined as 150,000 ATAC-Seq peaks with the highest levels of H3K27ac signal, but were subset for regions with H3K27ac peaks after running (see below). Compartmental saddle plots Saddle plots were made manually in R. We selected three chromosomes that had no major karyotypic diYerences between cell lines and had high correlation of eigenvectors between replicates with the same signage (chr2, chr12, chr17). For each chromosome and cell type, we sorted eigenvectors into 20 bins . We then calculated the mean observed/expected .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 32 values (using SCALE normalization) between regions belonging to diYerent bins, and plotted it as a heatmap. DiVerential loop, TAD, gene, and peak analysis DiYerential genes were calculated using DESeq2 (version 1.42.0) (115). Each cell type had 3 replicates, and a design of ~cell Type was used. No fold -change cutoY was applied . Genes with an adjusted p-value below 0.01 were considered significant. DiYerential H3K27ac within ATAC-Seq peaks were calculated using a similar design, but with a p-value cutoY of 0.05. DiYerential loops were also identified using DESeq2, but with additional adjustments. Raw, un-normalized loop counts were pulled from each technical replicate map, resulting in 8 values per cell type (4 technical replicates for each of 2 biological replicates per cell type). An LRT design was used of ~technicalRep + biologicalRep + cellType, with ~technicalRep + biologicalRep as the comparison. Size factors were also provided manually based on NeoLoopFinder output (Supp. Table). A fold -change cutoY of 1.5 was applied, as was an adjusted p-value cutoY of 0.025. DiYerential TAD boundaries were detected using an alternative method. Insulation scores were pulled from all TAD boundaries at the technical replicate level (8 values per cell type) . A T-test was then applied for each pairwise comparison of cell types, and p -values were adjusted using FDR. Boundaries with a n adjusted p-value below 0.01 were considered significant. DiYerential loops, TAD boundaries, and genes were clustered based on their patterns of change across the three cell types using k -means clustering of centered and scaled normalized counts. Micro-C feature overlap and aggregate analysis Micro-C feature overlap and analysis was conducted in R primarily using the GenomicRanges, InteractionSet, and mariner packages (121–123). Aggregate matrices of loop pixels and TAD boundaries were generated using mariner and visualized using plotgardener (123,124). Activity-by-contact The activity-by-contact was used based on Fulco et al. 2019 with slight adjustments (125). To allow for direct comparison of all enhancer-promoter pairs across cell lines, we manually defined potential enhancer regions and used the same input for each cell type. These potential enhancer regions were defined as they typically are in the pipeline, by finding 150,000 ATAC-Seq peaks with the highest H3K27ac levels . The output was filtered using a suggested ABC score cutoY of 0.025 to identify likely enhancer-promoter pairs. To allow for direct comparison with our other enhancer analysis, we filter ed the output based on the .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 33 enhancer regions also overlapping H3K27ac, which still represented a majority of the valid pairs identified. Matched enhancer-promoter sets Covariate-matched subset selection among non -looped enhancer -promoter pairs was performed using the matchRanges function from the nullranges package (126,127). Enhancer-promoter pair distance or mean Micro -C contact frequency were used as covariates. Matching was done with the stratified matching method without replacement. Gene ontology and gene set enrichment analysis Gene ontology (GO) term enrichment was performed in R using the gprofiler2 package (127). Gene set enrichment analysis (GSEA) was performed with the GSEA software, using size factor normalized RNA-seq counts as input (128) and the Hallmark H1 gene set. ATAC-Seq motif analysis Motif analysis of ATAC -Seq peaks within strengthened and weakened loop anchors was conducted using the HOMER suite findMotifsGenome.pl script with size given (129). ATAC- Seq peaks within the anchors of static loop anchors were used as background. H3K27ac peak pileup analysis H3K27ac ChIP-Seq analysis in the anchors of gained, lost, and static loops was conducted using deeptools (116). Alignment files were normalized using RPGC with the bamCoverage function, and adjusted using scale factors generated from edgeR TMM normalization factors of counts from overlapping H3K27ac and ATAC-Seq peaks (130). Aggregate profile plots were then created using the plotProfile command. Genomic data visualization Micro-C contact frequency maps, aggregate analysis plots, gene annotations, and genomic signal tracks (RNA -Seq, ChIP -Seq, ATAC-Seq) were visualized and plotted in R using the plotgardener package (124). .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 34 DECLARATIONS Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Availability of data and materials The Micro-C datasets generated and analyzed during the current study are available in the GEO repository, accession GSE254045. CTCF ChIP-Seq data was previously published under GEO repository GSE98551. H3K27ac ChIP-Seq data was previously published under GEO repository GSE229295. The code used to generate figures from the current study are available on Github at the following repository: https://github.com/ksmetz/MCF10-MicroC. Competing interests The authors declare they have no competing interests. Funding This work was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research (grant no. ZIABC010309-24 to T.M.), the Northern New England Clinical Translation Network ( grant no. GM115516 to G.S. ), funding from the National Cancer Institute (grant no. P01CA240685 to G.S., J.S., and S.F .), and funding from the Charlotte Perelman and Arthur Jason Perelman Fund for Cancer Research (G.S. and J.S.). Authors’ contributions KSMR, TM, AF, HG, JS, and GS conceptualized the project, designed the experiments, and interpreted the results. TM, SF , GS, and JS obtained funding. KSMR and AF executed experiments. KSMR and HG performed computational processing and analysis. SF conducted ChIP-Seq processing. KHH coordinated SKY karyotyping. KSMR and TM drafted the manuscript. KSMR, AF, HG, KHH, JS, GS, and TM participated in reviewing and editing the manuscript.

We thank Kathleen Quinn and Joseph Boyd for their work building the ChIP-Seq libraries and processing the ChIP -Seq data , respectively. We tha nk Darawalee Wangsa and Danny Wangsa in the CCR Genetics Branch and OMICS Technology Facility at the NCI for their expert SKY analysis. We thank Alquassem Abuorquob for his assistance in ChIP-Seq library preparation. STR analyses and Next Generation Sequencing was in part done with the assistance of the Vermont Integrated Genomics Resource (VIGR) at the Vermont Cancer Center, University of Vermont and the Genomics Sequencing Facility (GSF) at Greehey Children's Cancer Research Institute UT Health San Antonio. We thank the members of the .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 35 Misteli and Stein labs for feedback . We thank Jordan Zhang, Misha Gattengo, and Sierra Wilson at Dovetail Genomics for coordinating Micro-C library preparation and preliminary analysis. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 36 ADDITIONAL FILES Additional File 1: TableS1-microcMetrics.xls Supplementary Table 1. Micro-C library quality metrics for individual technical replicates, biological replicates, and cell types. Additional File 2: TableS2-loops.xls Supplementary Table 2. Chromatin loop summary table for total loop set (n = 29,205). Columns include loop anchor coordinates (A-J), loop call status by cell type (K-N), SIP AP score (O), loop name (P), loop span length (Q), average and maximum un-normalized counts (R-S), log2 fold-change (T-V), adjusted p-value (W), diYerential status by pairwise comparison (X-Z), un-normalized counts by technical replicate (AA-AX), maximum log2 fold- change (AY), pairwise comparison with greatest fold-change (AZ), diYerential cluster (BA), Z- score normalized counts (BB-BE), and variance stabilized counts (BF-BH). Additional File 3: TableS3-TADbounds.xls Supplementary Table 3. TAD boundary summary table for total boundary set (n = 17,097). Columns include boundary coordinates (A-C), insulation scores by technical replicate (D- AA), adjusted p-values, diYerence in insulation score, and diYerential status for each pairwise comparison (AB-AJ), diYerential status across all comparisons (AK), diYerential cluster (AL), average insulation score per cell type (AM-AO), and Z-score normalized insulation scores (AP-AS). Additional File 4: TableS4-TADs.xls Supplementary Table 4. TAD summary table for total set of TADs (n = 13,231). Columns include TAD boundary coordinates (A-J), and whether the TAD was called in each cell type (K-M). Additional File 5: MCF10_suppFigs.pdf Supplementary figures and legends. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 37 1. Dekker J, Mirny L. The 3D genome as moderator of chromosomal communication. Cell [Internet]. 2016 Mar 10 [cited 2025 Mar 8];164(6):1110. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC4788811/ 2. Jerkovic´ I, Cavalli G. Understanding 3D genome organization by multidisciplinary methods. Nat Rev Mol Cell Biol [Internet]. 2021 May 5 [cited 2025 Mar 8];22(8):511– 28. Available from: https://hal.science/hal-03269159 3. Misteli T. The self-organizing genome: Principles of genome architecture and function. Cell [Internet]. 2020 Oct 1 [cited 2025 Mar 8];183(1):28. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC7541718/ 4. Pombo A, Dillon N. Three-dimensional genome architecture: players and mechanisms. Nature Reviews Molecular Cell Biology 2015 16:4 [Internet]. 2015 Mar 11 [cited 2025 Mar 8];16(4):245–57. Available from: https://www.nature.com/articles/nrm3965 5. Rowley MJ, Corces VG. Organizational principles of 3D genome architecture. Nature Reviews Genetics 2018 19:12 [Internet]. 2018 Oct 26 [cited 2025 Mar 8];19(12):789– 800. Available from: https://www.nature.com/articles/s41576-018-0060-8 6. Corces R, Corces VG, Croce L Di, Shilatifard A. The three-dimensional cancer genome. Curr Opin Genet Dev [Internet]. 2016 [cited 2025 Mar 8];36:1–7. Available from: http://dx.doi.org/10.1016/j.gde.2016.01.002 7. Peng A, Peng W, Wang R, Zhao H, Yu X, Sun Y . Regulation of 3D Organization and Its Role in Cancer Biology. Front Cell Dev Biol [Internet]. 2022 Jun 8 [cited 2025 Mar 8];10:879465. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC9213882/ 8. Zhang L, Liu F , Wu L, Fu S, Xing L, Zhang L, et al. Disorder of three-dimensional chromosome structure plays a role in carcinogenesis. Clinical and Translational Discovery [Internet]. 2021 Oct 1 [cited 2025 Mar 8];1(1):e17. Available from: https://onlinelibrary.wiley.com/doi/full/10.1002/ctd2.17 9. Amodeo ME, Eyler CE, Johnstone SE. Rewiring cancer: 3D genome determinants of cancer hallmarks. Curr Opin Genet Dev [Internet]. 2025 Apr 1 [cited 2025 Apr 1];91. Available from: https://pubmed.ncbi.nlm.nih.gov/39862605/ 10. Davidson IF , Peters JM. Genome folding through loop extrusion by SMC complexes. Nature Reviews Molecular Cell Biology 2021 22:7 [Internet]. 2021 Mar 25 [cited 2025 Mar 8];22(7):445–64. Available from: https://www.nature.com/articles/s41580-021- 00349-7 11. de Wit E, Nora EP . New insights into genome folding by loop extrusion from inducible degron technologies. Nature Reviews Genetics 2022 24:2 [Internet]. 2022 Sep 30 [cited 2025 Mar 8];24(2):73–85. Available from: https://www.nature.com/articles/s41576-022-00530-4 12. Sanborn AL, Rao SSP , Huang SC, Durand NC, Huntley MH, Jewett AI, et al. Chromatin extrusion explains key features of loop and domain formation in wild-type and engineered genomes. Proc Natl Acad Sci U S A [Internet]. 2015 Nov 24 [cited 2025 Mar 8];112(47):E6456–65. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.1518552112 13. Goel VY , Huseyin MK, Hansen AS. Region Capture Micro-C reveals coalescence of enhancers and promoters into nested microcompartments. Nature Genetics 2023 .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 38 55:6 [Internet]. 2023 May 8 [cited 2025 Mar 8];55(6):1048–56. Available from: https://www.nature.com/articles/s41588-023-01391-1 14. Hsieh THS, Cattoglio C, Slobodyanyuk E, Hansen AS, Rando OJ, Tjian R, et al. Resolving the 3D Landscape of Transcription-Linked Mammalian Chromatin Folding. Mol Cell [Internet]. 2020;78(3):539-553.e8. Available from: http://dx.doi.org/10.1016/j.molcel.2020.03.002 15. Oberbeckmann E, Quililan K, Cramer P , Oudelaar AM. In vitro reconstitution of chromatin domains shows a role for nucleosome positioning in 3D genome organization. Nature Genetics 2024 56:3 [Internet]. 2024 Jan 30 [cited 2025 Mar 19];56(3):483–92. Available from: https://www.nature.com/articles/s41588-023- 01649-8 16. Rowley MJ, Corces VG. Organizational principles of 3D genome architecture. Nat Rev Genet [Internet]. 2018;19(12):789–800. Available from: http://dx.doi.org/10.1038/s41576-018-0060-8 17. Chakraborty S, Wenzlitschke N, Anderson MJ, Eraso A, Baudic M, Thompson JJ, et al. Structural perturbation of chromatin domains with multiple developmental regulators can severely impact gene regulation and development. bioRxiv [Internet]. 2024 Aug 3 [cited 2025 Mar 8];2024.08.03.606480. Available from: https://www.biorxiv.org/content/10.1101/2024.08.03.606480v1 18. Chakraborty S, Wenzlitschke N, Anderson MJ, Eraso A, Baudic M, Thompson JJ, et al. Deletion of a single CTCF motif at the boundary of a chromatin domain with three FGF genes disrupts gene expression and embryonic development. Dev Cell [Internet]. 2025 Feb [cited 2025 Mar 8];0(0). Available from: https://www.cell.com/action/showFullText?pii=S1534580725000644 19. da Costa-Nunes JA, Noordermeer D. TADs: Dynamic structures to create stable regulatory functions. Curr Opin Struct Biol. 2023 Aug 1;81:102622. 20. Xiao J, Hafner A, Boettiger AN. How subtle changes in 3D structure can create large changes in transcription. Elife [Internet]. 2021 Jul 1 [cited 2025 Mar 8];10:e64320. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC8352591/ 21. Sood V , Misteli T. The stochastic nature of genome organization and function. Curr Opin Genet Dev [Internet]. 2022 Feb 1 [cited 2025 Mar 17];72:45–52. Available from: https://pubmed.ncbi.nlm.nih.gov/34808408/ 22. D’Ippolito AM, McDowell IC, Barrera A, Hong LK, Leichter SM, Bartelt LC, et al. Pre- established Chromatin Interactions Mediate the Genomic Response to Glucocorticoids. Cell Syst [Internet]. 2018 Aug 22 [cited 2025 Mar 8];7(2):146-160.e7. Available from: https://pubmed.ncbi.nlm.nih.gov/30031775/ 23. Kane L, Williamson I, Flyamer IM, Kumar Y , Hill RE, Lettice LA, et al. Cohesin is required for long-range enhancer action at the Shh locus. Nature Structural & Molecular Biology 2022 29:9 [Internet]. 2022 Sep 12 [cited 2025 Mar 8];29(9):891–7. Available from: https://www.nature.com/articles/s41594-022-00821-8 24. Schoenfelder S, Fraser P . Long-range enhancer-promoter contacts in gene expression control. Nat Rev Genet [Internet]. 2019 Aug 1 [cited 2025 Mar 8];20(8):437–55. Available from: https://pubmed.ncbi.nlm.nih.gov/31086298/ .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 39 25. Stolper RJ, Tsang FH, Georgiades E, Hansen LLP , Downes DJ, Harrold CL, et al. Loop extrusion by cohesin plays a key role in enhancer-activated gene expression during diYerentiation. bioRxiv [Internet]. 2023 Sep 7 [cited 2025 Mar 8];2023.09.07.556660. Available from: https://www.biorxiv.org/content/10.1101/2023.09.07.556660v1 26. Popay TM, Dixon JR. Coming full circle: On the origin and evolution of the looping model for enhancer–promoter communication. J Biol Chem [Internet]. 2022 Aug 1 [cited 2025 Mar 24];298(8):102117. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC9283939/ 27. Benabdallah NS, Williamson I, Illingworth RS, Kane L, Boyle S, Sengupta D, et al. Decreased Enhancer-Promoter Proximity Accompanying Enhancer Activation. Mol Cell [Internet]. 2019 Nov 7 [cited 2025 Mar 8];76(3):473-484.e7. Available from: https://pubmed.ncbi.nlm.nih.gov/31494034/ 28. Despang A, Schöpflin R, Franke M, Ali S, Jerković I, Paliou C, et al. Functional dissection of the Sox9–Kcnj2 locus identifies nonessential and instructive roles of TAD architecture. Nature Genetics 2019 51:8 [Internet]. 2019 Jul 29 [cited 2025 Mar 8];51(8):1263–71. Available from: https://www.nature.com/articles/s41588-019- 0466-z 29. Hansen KL, Adachi AS, Braccioli L, Kadvani S, Boileau RM, Shah R, et al. Synergy between cis-regulatory elements can render cohesin dispensable for distal enhancer function. bioRxiv [Internet]. 2024 [cited 2025 Mar 8]; Available from: https://doi.org/10.1101/2024.10.04.615095 30. Hsieh THS, Cattoglio C, Slobodyanyuk E, Hansen AS, Darzacq X, Tjian R. Enhancer- promoter interactions and transcription are largely maintained upon acute loss of CTCF , cohesin, WAPL or YY1. Nat Genet [Internet]. 2022 Dec 1 [cited 2025 Mar 8];54(12):1919–32. Available from: https://pubmed.ncbi.nlm.nih.gov/36471071/ 31. Rao SSP , Huang SC, Glenn St Hilaire B, Engreitz JM, Perez EM, KieYer-Kwon KR, et al. Cohesin Loss Eliminates All Loop Domains. Cell. 2017 Oct 5;171(2):305-320.e24. 32. Cuartero S, Weiss FD, Dharmalingam G, Guo Y , Ing-Simmons E, Masella S, et al. Control of inducible gene expression links cohesin to hematopoietic progenitor self- renewal and diYerentiation. Nat Immunol [Internet]. 2018 Sep 1 [cited 2025 Mar 8];19(9):932–41. Available from: https://pubmed.ncbi.nlm.nih.gov/30127433/ 33. Popay TM, Pant A, Munting F , Black ME, Haghani N, Dixon JR. Genome-wide in vivo dynamics of cohesin-mediated loop extrusion and its role in transcription activation. bioRxiv [Internet]. 2024 Oct 2 [cited 2025 Mar 8];2024.10.02.616323. Available from: https://www.biorxiv.org/content/10.1101/2024.10.02.616323v1 34. Wang M, Sunkel BD, Ray WC, Stanton BZ. Chromatin structure in cancer. BMC Mol Cell Biol [Internet]. 2022 Dec 1 [cited 2025 Mar 8];23(1):1–10. Available from: https://bmcmolcellbiol.biomedcentral.com/articles/10.1186/s12860-022-00433-6 35. Guo YA, Chang MM, Huang W, Ooi WF , Xing M, Tan P , et al. Mutation hotspots at CTCF binding sites coupled to chromosomal instability in gastrointestinal cancers. Nat Commun [Internet]. 2018 Dec 1 [cited 2025 Mar 17];9(1). Available from: https://pubmed.ncbi.nlm.nih.gov/29670109/ 36. Katainen R, Dave K, Pitkänen E, Palin K, Kivioja T, Välimäki N, et al. CTCF/cohesin- binding sites are frequently mutated in cancer. Nat Genet [Internet]. 2015 Jun 26 .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 40 [cited 2025 Mar 17];47(7):818–21. Available from: https://pubmed.ncbi.nlm.nih.gov/26053496/ 37. Cosenza MR, Rodriguez-Martin B, Korbel JO. Structural Variation in Cancer: Role, Prevalence, and Mechanisms. Annu Rev Genomics Hum Genet [Internet]. 2022 Aug 31 [cited 2025 Mar 8];23:123–52. Available from: https://pubmed.ncbi.nlm.nih.gov/35655332/ 38. Mitelman F , Johansson B, Mertens F . The impact of translocations and gene fusions on cancer causation. Nat Rev Cancer [Internet]. 2007 Apr [cited 2025 Mar 8];7(4):233–45. Available from: https://pubmed.ncbi.nlm.nih.gov/17361217/ 39. Della Chiara G, Jiménez C, Virdi M, Crosetto N, Bienko M. Enhancers dysfunction in the 3D genome of cancer cells. Front Cell Dev Biol. 2023 Nov 6;11:1303862. 40. Barutcu AR, Lajoie BR, McCord RP , Tye CE, Hong D, Messier TL, et al. Chromatin interaction analysis reveals changes in small chromosome and telomere clustering between epithelial and breast cancer cells. Genome Biol [Internet]. 2015 Sep 28 [cited 2025 Mar 8];16(1). Available from: https://pubmed.ncbi.nlm.nih.gov/26415882/ 41. Fritz AJ, Ghule PN, Boyd JR, Tye CE, Page NA, Hong D, et al. Intranuclear and higher- order chromatin organization of the major histone gene cluster in breast cancer. J Cell Physiol [Internet]. 2017 Feb 1 [cited 2025 Mar 8];233(2):1278. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC5705002/ 42. Johnstone SE, Reyes A, Qi Y , Adriaens C, Hegazi E, Pelka K, et al. Large-Scale Topological Changes Restrain Malignant Progression in Colorectal Cancer. Cell [Internet]. 2020 Sep 17 [cited 2025 Mar 8];182(6):1474. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC7575124/ 43. Li J, Fang K, Choppavarapu L, Yang K, Yang Y , Wang J, et al. Hi-C profiling of cancer spheroids identifies 3D-growth-specific chromatin interactions in breast cancer endocrine resistance. Clin Epigenetics [Internet]. 2021 Dec 1 [cited 2025 Mar 8];13(1):1–13. Available from: https://clinicalepigeneticsjournal.biomedcentral.com/articles/10.1186/s13148-021- 01167-6 44. Mourad R, Hsu PY , Juan L, Shen C, Koneru P , Lin H, et al. Estrogen Induces Global Reorganization of Chromatin Structure in Human Breast Cancer Cells. PLoS One [Internet]. 2014 Dec 3 [cited 2025 Mar 8];9(12):e113354. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC4255042/ 45. Rickman DS, Soong TD, Moss B, Mosquera JM, Dlabal J, Terry S, et al. Oncogene- mediated alterations in chromatin conformation. Proc Natl Acad Sci U S A [Internet]. 2012 Jun 5 [cited 2025 Mar 8];109(23):9083–8. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC3384175/ 46. Taberlay PC, Achinger-Kawecka J, Lun ATL, Buske FA, Sabir K, Gould CM, et al. Three- dimensional disorganization of the cancer genome occurs coincident with long- range genetic and epigenetic alterations. Genome Res [Internet]. 2016 Jun 1 [cited 2025 Mar 8];26(6):719–31. Available from: https://genome.cshlp.org/content/26/6/719.full .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 41 47. Yang Y , Choppavarapu L, Fang K, Naeini AS, Nosirov B, Li J, et al. The 3D genomic landscape of diYerential response to EGFR/HER2 inhibition in endocrine-resistant breast cancer cells. Biochimica et Biophysica Acta (BBA) - Gene Regulatory Mechanisms. 2020 Nov 1;1863(11):194631. 48. Zhou Y , Gerrard DL, Wang J, Li T, Yang Y , Fritz AJ, et al. Temporal dynamic reorganization of 3D chromatin architecture in hormone-induced breast cancer and endocrine resistance. Nature Communications 2019 10:1 [Internet]. 2019 Apr 3 [cited 2025 Mar 8];10(1):1–14. Available from: https://www.nature.com/articles/s41467-019-09320-9 49. Kim T, Han S, Chun Y , Yang H, Min H, Jeon SY , et al. Comparative characterization of 3D chromatin organization in triple-negative breast cancers. Experimental & Molecular Medicine 2022 54:5 [Internet]. 2022 May 5 [cited 2025 Mar 8];54(5):585– 600. Available from: https://www.nature.com/articles/s12276-022-00768-2 50. Rossini R, Oshaghi M, Nekrasov M, Bellanger A, Domaschenz R, Dijkwel Y , et al. Loss of multi-level 3D genome organization during breast cancer progression. bioRxiv [Internet]. 2024 Nov 27 [cited 2025 Mar 8];2023.11.26.568711. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC10705249/ 51. Mortenson KL, Dawes C, Wilson ER, Patchen NE, Johnson HE, Gertz J, et al. 3D genomic analysis reveals novel enhancer-hijacking caused by complex structural alterations that drive oncogene overexpression. Nature Communications 2024 15:1 [Internet]. 2024 Jul 20 [cited 2025 Mar 8];15(1):1–15. Available from: https://www.nature.com/articles/s41467-024-50387-w 52. Wang X, Xu J, Zhang B, Hou Y , Song F , Lyu H, et al. Genome-wide detection of enhancer-hijacking events from chromatin interaction data in rearranged genomes. Nature Methods 2021 18:6 [Internet]. 2021 Jun 3 [cited 2025 Mar 8];18(6):661–8. Available from: https://www.nature.com/articles/s41592-021-01164-w 53. Puleo J, Polyak K. The MCF10 Model of Breast Tumor Progression. Cancer Res [Internet]. 2021 Aug 15 [cited 2025 Mar 8];81(16):4183–5. Available from: /cancerres/article/81/16/4183/670266/The-MCF10-Model-of-Breast-Tumor- ProgressionThe 54. Soule HD, Maloney TM, Wolman SR, Brenz R, Russo J, Pauley RJ, et al. Isolation and characterization of a spontaneously immortalized human breast epithelial cell line, MCF-10. Cancer Res [Internet]. 1990 Sep 1 [cited 2025 Mar 8];50(18):6075–86. Available from: https://europepmc.org/article/med/1975513 55. Dawson PJ, Wolman SR, Tait L, Heppner GH, Miller FR. MCF10AT: a model for the evolution of cancer from proliferative breast disease. Am J Pathol [Internet]. 1996 Jan [cited 2025 Mar 8];148(1):313. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC1861604/ 56. Santner SJ, Dawson PJ, Tait L, Soule HD, Eliason J, Mohamed AN, et al. Malignant MCF10CA1 cell lines derived from premalignant human breast epithelial MCF10AT cells. Breast Cancer Res Treat [Internet]. 2001 [cited 2025 Mar 8];65(2):101–10. Available from: https://pubmed.ncbi.nlm.nih.gov/11261825/ 57. Bhardwaj A, Singh H, Rajapakshe K, Tachibana K, Ganesan N, Pan Y , et al. Regulation of miRNA-29c and its downstream pathways in preneoplastic progression of triple- .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 42 negative breast cancer. Oncotarget [Internet]. 2017 [cited 2025 Mar 8];8(12):19645. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC5386711/ 58. Buchanan NS, Zhao J, Zhu K, Patwa TH, Miller FR, Lubman DM. DiYerential expression of acidic proteins with progression in the MCF10 model of human breast disease. Int J Oncol [Internet]. 2007 Oct 1 [cited 2025 Mar 8];31(4):941–9. Available from: http://www.spandidos-publications.com/10.3892/ijo.31.4.941/abstract 59. Dorgham MG, Elliott BA, Holley CL, Mansfield KD. m6A regulates breast cancer proliferation and migration through stage-dependent changes in Epithelial to Mesenchymal Transition gene expression. Front Oncol [Internet]. 2023 [cited 2025 Mar 8];13:1268977. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC10661887/ 60. Kadota M, Yang HH, Gomez B, Sato M, CliYord RJ, Meerzaman D, et al. Delineating Genetic Alterations for Tumor Progression in the MCF10A Series of Breast Cancer Cell Lines. PLoS One [Internet]. 2010 Feb 15 [cited 2025 Mar 8];5(2):e9201. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC2821407/ 61. Lin HH, Qraitem M, Lian Y , Taylor SR, Farkas ME. Analyses of BMAL1 and PER2 Oscillations in a Model of Breast Cancer Progression Reveal Changes With Malignancy. Integr Cancer Ther [Internet]. 2019 Apr 1 [cited 2025 Mar 8];18:1534735419836494. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC6449806/ 62. Maguire SL, Peck B, Wai PT, Campbell J, Barker H, Gulati A, et al. Three-dimensional modelling identifies novel genetic dependencies associated with breast cancer progression in the isogenic MCF10 model. J Pathol [Internet]. 2016 Nov 1 [cited 2025 Mar 8];240(3):315. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC5082563/ 63. Marella N V ., Malyavantham KS, Wang J, Matsui SI, Liang P , Berezney R. Cytogenetic and cDNA Microarray Expression Analysis of MCF10A Human Breast Cancer Progression Cell Lines. Cancer Res [Internet]. 2009 Jul 15 [cited 2025 Mar 8];69(14):5946. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC2826242/ 64. Rhee DK, Park SH, Jang YK. Molecular signatures associated with transformation and progression to breast cancer in the isogenic MCF10 model. Genomics. 2008 Dec 1;92(6):419–28. 65. So JY , Lee HJ, Kramata P , Minden A, Suh N. DiYerential Expression of Key Signaling Proteins in MCF10 Cell Lines, a Human Breast Cancer Progression Model. Mol Cell Pharmacol [Internet]. 2012 [cited 2025 Mar 8];4(1):31. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC3928091/ 66. Sen A, Xu Z, Tyndale ST, Yasis J, Cho CY , Bump R, et al. Heterogeneity in chromatin structure drives core regulatory pathways in B-cell Acute Lymphoblastic Leukemia. bioRxiv [Internet]. 2024 Oct 5 [cited 2025 Mar 8];2024.10.04.616668. Available from: https://www.biorxiv.org/content/10.1101/2024.10.04.616668v1 67. Bose A, Datta S, Mandal R, Ray U, Dhar R. Increased heterogeneity in expression of genes associated with cancer progression and drug resistance. Transl Oncol [Internet]. 2024 Mar 1 [cited 2025 Mar 8];41:101879. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC10832509/ .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 43 68. Dagogo-Jack I, Shaw AT. Tumour heterogeneity and resistance to cancer therapies. Nature Reviews Clinical Oncology 2017 15:2 [Internet]. 2017 Nov 8 [cited 2025 Mar 8];15(2):81–94. Available from: https://www.nature.com/articles/nrclinonc.2017.166 69. Georgiades E, Harrold CL, Roberts N, Kassouf M, Riva SG, Francis HS, et al. Active regulatory elements recruit cohesin to establish cell-1 specific chromatin domains. [cited 2025 Mar 8]; Available from: https://doi.org/10.1101/2023.10.13.562171 70. Rinzema NJ, Sofiadis K, Tjalsma SJD, Verstegen MJAM, Oz Y , Valdes-Quezada C, et al. Building regulatory landscapes reveals that an enhancer can recruit cohesin to create contact domains, engage CTCF sites and activate distant genes. Nature Structural & Molecular Biology 2022 29:6 [Internet]. 2022 Jun 16 [cited 2025 Mar 8];29(6):563–74. Available from: https://www.nature.com/articles/s41594-022- 00787-7 71. Creyghton MP , Cheng AW, Welstead GG, Kooistra T, Carey BW, Steine EJ, et al. Histone H3K27ac separates active from poised enhancers and predicts developmental state. Proc Natl Acad Sci U S A [Internet]. 2010 Dec 14 [cited 2025 Mar 8];107(50):21931–6. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.1016071107 72. Glossary – ENCODE [Internet]. [cited 2025 Mar 8]. Available from: https://www.encodeproject.org/glossary/ 73. Heintzman ND, Stuart RK, Hon G, Fu Y , Ching CW, Hawkins RD, et al. Distinct and predictive chromatin signatures of transcriptional promoters and enhancers in the human genome. Nature Genetics 2007 39:3 [Internet]. 2007 Feb 4 [cited 2025 Mar 8];39(3):311–8. Available from: https://www.nature.com/articles/ng1966 74. Ferlier T, Coulouarn C. Regulation of Gene Expression in Cancer—An Overview. Cells [Internet]. 2022 Dec 1 [cited 2025 Mar 8];11(24):4058. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC9776464/ 75. Gavish A, Tyler M, Greenwald AC, HoeYlin R, Simkin D, Tschernichovsky R, et al. Hallmarks of transcriptional intratumour heterogeneity across a thousand tumours. Nature [Internet]. 2023 Jun 15 [cited 2025 Mar 8];618(7965):598–606. Available from: https://pubmed.ncbi.nlm.nih.gov/37258682/ 76. Liberzon A, Birger C, Thorvaldsdóttir H, Ghandi M, Mesirov JP , Tamayo P . The Molecular Signatures Database (MSigDB) hallmark gene set collection. Cell Syst [Internet]. 2015 Dec 23 [cited 2025 Mar 8];1(6):417–25. Available from: https://pubmed.ncbi.nlm.nih.gov/26771021/ 77. Ma XJ, Salunga R, Tuggle JT, Gaudet J, Enright E, McQuary P , et al. Gene expression profiles of human breast cancer progression. Proc Natl Acad Sci U S A [Internet]. 2003 May 13 [cited 2025 Mar 8];100(10):5974–9. Available from: https://www.pnas.org/doi/abs/10.1073/pnas.0931261100 78. He Q, Jing H, Liaw L, Gower L, Vary C, Hua S, et al. Suppression of Spry1 inhibits triple-negative breast cancer malignancy by decreasing EGF/EGFR mediated mesenchymal phenotype. Sci Rep [Internet]. 2016 Mar 15 [cited 2025 Mar 9];6:23216. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC4791662/ .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 44 79. Yang Y , Zhong X, Lan Y , Liang P , Huang Y , Wang Y , et al. Aberrant inactivation of SCNN1G promotes the motility of head and neck squamous cell carcinoma. Pathol Res Pract. 2022 Dec 1;240:154175. 80. Fulco CP , Nasser J, Jones TR, Munson G, Bergman DT, Subramanian V , et al. Activity- by-contact model of enhancer–promoter regulation from thousands of CRISPR perturbations. Nature Genetics 2019 51:12 [Internet]. 2019 Nov 29 [cited 2025 Mar 8];51(12):1664–9. Available from: https://www.nature.com/articles/s41588-019- 0538-0 81. Hanahan D, Weinberg RA. Hallmarks of cancer: The next generation. Cell [Internet]. 2011 Mar 4 [cited 2025 Mar 24];144(5):646–74. Available from: https://www.cell.com/action/showFullText?pii=S0092867411001279 82. Hu Y , Huang X, Lu X, Lin S. COL12A1 as a prognostic biomarker in HER2-enriched breast cancer and its association with immune infiltration. Eur J Gynaecol Oncol [Internet]. 2022 [cited 2025 Mar 8];43(5):85–90. Available from: https://www.ejgo.net/ 83. Papanicolaou M, Parker AL, Yam M, Filipe EC, Wu SZ, Chitty JL, et al. Temporal profiling of the breast tumour microenvironment reveals collagen XII as a driver of metastasis. Nature Communications 2022 13:1 [Internet]. 2022 Aug 6 [cited 2025 Mar 8];13(1):1–21. Available from: https://www.nature.com/articles/s41467-022- 32255-7 84. Song Y , Wang L, Wang K, Lu Y , Zhou P . COL12A1 Acts as a Novel Prognosis Biomarker and Activates Cancer-Associated Fibroblasts in Pancreatic Cancer through Bioinformatics and Experimental Validation. Cancers (Basel) [Internet]. 2023 Mar 1 [cited 2025 Mar 8];15(5):1480. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC10000532/ 85. Borcherding N, Kusner D, Kolb R, Xie Q, Li W, Yuan F , et al. Paracrine WNT5A signaling inhibits expansion of tumor-initiating cells. Cancer Res [Internet]. 2015 May 15 [cited 2025 Mar 8];75(10):1972. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC4433621/ 86. Roarty K, Baxley SE, Crowley MR, Frost AR, Serra R. Loss of TGF-β or Wnt5a results in an increase in Wnt/β-catenin activity and redirects mammary tumour phenotype. Breast Cancer Res [Internet]. 2009 Apr 3 [cited 2025 Mar 8];11(2):R19. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC2688948/ 87. Zeng R, Huang J, Zhong MZ, Li L, Yang G, Liu L, et al. Multiple Roles of WNT5A in Breast Cancer. Med Sci Monit [Internet]. 2016 Dec 22 [cited 2025 Mar 8];22:5058. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC5201118/ 88. Pasini D, Malatesta M, Jung HR, Walfridsson J, Willer A, Olsson L, et al. Characterization of an antagonistic switch between histone H3 lysine 27 methylation and acetylation in the transcriptional regulation of Polycomb group target genes. Nucleic Acids Res [Internet]. 2010 Apr 12 [cited 2025 Mar 8];38(15):4958. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC2926606/ 89. Cai Y , Zhang Y , Loh YP , Tng JQ, Lim MC, Cao Z, et al. H3K27me3-rich genomic regions can function as silencers to repress gene expression via chromatin interactions. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 45 Nature Communications 2021 12:1 [Internet]. 2021 Jan 29 [cited 2025 Mar 8];12(1):1–22. Available from: https://www.nature.com/articles/s41467-021-20940-y 90. Guo M, Peng Y , Gao A, Du C, Herman JG. Epigenetic heterogeneity in cancer. Biomarker Research 2019 7:1 [Internet]. 2019 Oct 31 [cited 2025 Mar 24];7(1):1–19. Available from: https://biomarkerres.biomedcentral.com/articles/10.1186/s40364- 019-0174-y 91. Carter B, Zhao K. The epigenetic basis of cellular heterogeneity. Nat Rev Genet [Internet]. 2020 Apr 1 [cited 2025 Mar 24];22(4):235. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC10880028/ 92. Flavahan WA, Gaskell E, Bernstein BE. Epigenetic plasticity and the hallmarks of cancer. Science [Internet]. 2017 Jul 21 [cited 2025 Mar 30];357(6348). Available from: https://pubmed.ncbi.nlm.nih.gov/28729483/ 93. Yang JH, Hansen AS. Enhancer selectivity in space and time: from enhancer– promoter interactions to promoter activation. Nat Rev Mol Cell Biol [Internet]. 2024 Jul 1 [cited 2025 Mar 20];25(7):574. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11574175/ 94. Guckelberger P , Doughty BR, Munson G, Rao SSP , Tan Y , Cai XS, et al. Cohesin- mediated 3D contacts tune enhancer-promoter regulation. bioRxiv [Internet]. 2024 Jul 22 [cited 2025 Mar 8];2024.07.12.603288. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC11257546/ 95. Kubo N, Ishii H, Xiong X, Bianco S, Meitinger F , Hu R, et al. Promoter-proximal CTCF binding promotes distal enhancer-dependent gene activation. Nature Structural & Molecular Biology 2021 28:2 [Internet]. 2021 Jan 4 [cited 2025 Mar 8];28(2):152–61. Available from: https://www.nature.com/articles/s41594-020-00539-5 96. Thiecke MJ, Wutz G, Muhar M, Tang W, Bevan S, Malysheva V , et al. Cohesin- Dependent and -Independent Mechanisms Mediate Chromosomal Contacts between Promoters and Enhancers. Cell Rep [Internet]. 2020 Jul 21 [cited 2025 Mar 8];32(3). Available from: https://pubmed.ncbi.nlm.nih.gov/32698000/ 97. O’Geen H, Frietze S, Farnham PJ. Using ChIP-seq Technology to Identify Targets of Zinc Finger Transcription Factors. Methods Mol Biol [Internet]. 2010 [cited 2025 May 8];649:437. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC4151297/ 98. Maor-Nof M, Shipony Z, Marinov GK, Greenleaf WJ, Gitler AD. An optimized ATAC-seq protocol for genome-wide mapping of active regulatory elements in primary mouse cortical neurons. STAR Protoc [Internet]. 2021 Dec 17 [cited 2025 May 8];2(4):100854. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC8496302/ 99. Corces MR, Trevino AE, Hamilton EG, Greenside PG, Sinnott-Armstrong NA, Vesuna S, et al. An improved ATAC-seq protocol reduces background and enables interrogation of frozen tissues. Nat Methods [Internet]. 2017;14(10):959–62. Available from: http://dx.doi.org/10.1038/nmeth.4396 100. Welcome to Micro-C documentation — Micro-C 0.1 documentation [Internet]. [cited 2025 Mar 27]. Available from: https://micro-c.readthedocs.io/en/latest/index.html 101. Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler transform. Bioinformatics [Internet]. 2009 Jul 15 [cited 2025 Mar 27];25(14):1754–60. Available from: https://dx.doi.org/10.1093/bioinformatics/btp324 .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 46 102. Open2C, Abdennur N, Fudenberg G, Flyamer IM, Galitsyna AA, Goloborodko A, et al. Pairtools: from sequencing data to chromosome contacts. bioRxiv [Internet]. 2023 Feb 15 [cited 2025 Mar 27];2023.02.13.528389. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC9949071/ 103. Danecek P , Bonfield JK, Liddle J, Marshall J, Ohan V , Pollard MO, et al. Twelve years of SAMtools and BCFtools. Gigascience [Internet]. 2021 Jan 29 [cited 2025 Mar 27];10(2):1–4. Available from: https://dx.doi.org/10.1093/gigascience/giab008 104. Durand NC, Shamim MS, Machol I, Rao SSP , Huntley MH, Lander ES, et al. Juicer Provides a One-Click System for Analyzing Loop-Resolution Hi-C Experiments. Cell Syst [Internet]. 2016 Jul 27 [cited 2025 Mar 27];3(1):95–8. Available from: https://pubmed.ncbi.nlm.nih.gov/27467249/ 105. Wang X, Xu J, Zhang B, Hou Y , Song F , Lyu H, et al. Genome-wide detection of enhancer-hijacking events from chromatin interaction data in rearranged genomes. Nature Methods 2021 18:6 [Internet]. 2021 Jun 3 [cited 2025 Mar 27];18(6):661–8. Available from: https://www.nature.com/articles/s41592-021-01164-w 106. Amemiya HM, Kundaje A, Boyle AP . The ENCODE Blacklist: Identification of Problematic Regions of the Genome. Scientific Reports 2019 9:1 [Internet]. 2019 Jun 27 [cited 2025 Mar 27];9(1):1–5. Available from: https://www.nature.com/articles/s41598-019-45839-z 107. Liu Y , Nanni L, Sungalee S, ZuYerey M, Tavernari D, Mina M, et al. Systematic inference and comparison of multi-scale chromatin sub-compartments connects spatial organization to cell phenotypes. Nature Communications 2021 12:1 [Internet]. 2021 May 10 [cited 2025 Mar 27];12(1):1–11. Available from: https://www.nature.com/articles/s41467-021-22666-3 108. Kruse K, Hug CB, Vaquerizas JM. FAN-C: a feature-rich framework for the analysis and visualisation of chromosome conformation capture data. Genome Biol [Internet]. 2020 Dec 1 [cited 2025 Mar 27];21(1):1–19. Available from: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-020-02215-9 109. Cresswell KG, Stansfield JC, Dozmorov MG. SpectralTAD: An R package for defining a hierarchy of topologically associated domains using spectral clustering. BMC Bioinformatics [Internet]. 2020 Jul 20 [cited 2025 Mar 27];21(1):1–19. Available from: https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-020-03652- w 110. Jordan Rowley M, Poulet A, Nichols MH, Bixler BJ, Sanborn AL, Brouhard EA, et al. Analysis of Hi-C data using SIP eYectively identifies loops in organisms from C. Elegans to mammals. Genome Res [Internet]. 2020 [cited 2025 Mar 27];30(3):447– 58. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC7111518/ 111. Patel H, Ewels P , Peltzer A, Hammarén R, Botvinnik O, Sturm G, et al. nf-core/rnaseq: nf-core/rnaseq v3.0 - Silver Shark. [cited 2025 May 7]; Available from: https://zenodo.org/records/4323183 112. Babraham Bioinformatics. Trim Galore! [Internet]. [cited 2017 Feb 15]. Available from: http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/ .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 47 113. Dobin A, Davis CA, Schlesinger F , Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics [Internet]. 2013 Jan 1 [cited 2018 Nov 17];29(1):15–21. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23104886 114. Patro R, Duggal G, Love MI, Irizarry RA, Kingsford C. Salmon provides fast and bias- aware quantification of transcript expression. Nat Methods [Internet]. 2017 Apr 6 [cited 2018 Nov 17];14(4):417–9. Available from: http://www.nature.com/articles/nmeth.4197 115. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol [Internet]. 2014 Dec 5 [cited 2025 Mar 27];15(12):1–21. Available from: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-014-0550-8 116. Ramírez F , Ryan DP , Grüning B, Bhardwaj V , Kilpert F , Richter AS, et al. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res [Internet]. 2016;44(W1):W160-5. Available from: http://dx.doi.org/10.1093/nar/gkw257 117. Li Q, Brown JB, Huang H, Bickel PJ. Measuring reproducibility of high-throughput experiments. https://doi.org/101214/11-AOAS466 [Internet]. 2011 Sep 1 [cited 2025 May 7];5(3):1752–79. Available from: https://projecteuclid.org/journals/annals-of- applied-statistics/volume-5/issue-3/Measuring-reproducibility-of-high-throughput- experiments/10.1214/11-AOAS466.full 118. Zhang Y , Liu T, Meyer CA, Eeckhoute J, Johnson DS, Bernstein BE, et al. Model-based Analysis of ChIP-Seq (MACS). Genome Biol [Internet]. 2008 Sep 17 [cited 2018 Nov 17];9(9):R137. Available from: http://genomebiology.biomedcentral.com/articles/10.1186/gb-2008-9-9-r137 119. Babraham Bioinformatics. FastQC: A Quality Control tool for High Throughput Sequence Data [Internet]. [cited 2017 Feb 15]. Available from: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/ 120. Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods [Internet]. 2012 Mar 4 [cited 2017 Feb 16];9(4):357–9. Available from: http://www.nature.com/doifinder/10.1038/nmeth.1923 121. Lawrence M, Huber W, Pagès H, Aboyoun P , Carlson M, Gentleman R, et al. Software for Computing and Annotating Genomic Ranges. PLoS Comput Biol [Internet]. 2013 [cited 2025 May 7];9(8):e1003118. Available from: https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1003118 122. Lun ATL, Perry M, Ing-Simmons E. Infrastructure for genomic interactions: Bioconductor classes for Hi-C, ChIA-PET and related experiments. F1000Res [Internet]. 2016 [cited 2025 May 7];5:950. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC4890298/ 123. Davis ES, Parker SM, Kramer NE, Flores JP , Kiran M, Phanstiel DH. Mariner: explore the Hi-Cs. Bioinformatics [Internet]. 2024 Jun 3 [cited 2025 Mar 27];40(6). Available from: https://dx.doi.org/10.1093/bioinformatics/btae352 124. Kramer NE, Davis ES, Wenger CD, Deoudes EM, Parker SM, Love MI, et al. Plotgardener: cultivating precise multi-panel figures in R. Bioinformatics [Internet]. .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint 48 2022 Mar 28 [cited 2025 Mar 27];38(7):2042–5. Available from: https://dx.doi.org/10.1093/bioinformatics/btac057 125. Fulco CP , Nasser J, Jones TR, Munson G, Bergman DT, Subramanian V , et al. Activity- by-contact model of enhancer–promoter regulation from thousands of CRISPR perturbations. Nature Genetics 2019 51:12 [Internet]. 2019 Nov 29 [cited 2025 Mar 27];51(12):1664–9. Available from: https://www.nature.com/articles/s41588-019- 0538-0 126. Love M Davis E, Phanstiel D, Lee S MW. nullranges: Generation of null ranges via bootstrapping or covariate matching [Internet]. 2022. Available from: https://nullranges.github.io/nullranges, https://github.com/nullranges/nullranges LB - 4LTs 127. Peterson H, Kolberg L, Raudvere U, Kuzmin I, Vilo J. gprofiler2 -- an R package for gene list functional enrichment analysis and namespace conversion toolset g:Profiler. F1000Res [Internet]. 2020 [cited 2025 May 7];9:ELIXIR-709. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC7859841/ 128. Subramanian A, Tamayo P , Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A [Internet]. 2005 Oct 25 [cited 2025 May 7];102(43):15545–50. Available from: /doi/pdf/10.1073/pnas.0506580102?download=true 129. Heinz S, Benner C, Spann N, Bertolino E, Lin YC, Laslo P , et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol Cell [Internet]. 2010 May 28 [cited 2018 Nov 17];38(4):576–89. Available from: http://www.ncbi.nlm.nih.gov/pubmed/20513432 130. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for diYerential expression analysis of digital gene expression data. Bioinformatics [Internet]. 2010 Jan 1 [cited 2017 Feb 15];26(1):139–40. Available from: http://www.ncbi.nlm.nih.gov/pubmed/19910308 .CC-BY-NC-ND 4.0 International licenseavailable under a 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 preprint (whichthis version posted May 17, 2025. ; https://doi.org/10.1101/2025.05.14.654144doi: bioRxiv preprint