CREPT is required for the metastasis of triple-negative breast cancer through a co-operational-chromatin loop-based gene regulation

Li, Jianghua; Xu, Lu; Wang, Jiayu; Wang, Xuning; Lin, Yuting; Zou, Alex Yutian; Ren, Fangli; Wang, Yinyin; Li, Jun; Chang, Zhijie

doi:10.1186/s12943-025-02361-3

Research
Open access
Published: 10 June 2025

CREPT is required for the metastasis of triple-negative breast cancer through a co-operational-chromatin loop-based gene regulation

Molecular Cancer volume 24, Article number: 170 (2025) Cite this article

2267 Accesses
2 Citations
14 Altmetric
Metrics details

Abstract

Triple-negative breast cancer (TNBC) is recognized for its aggressiveness, yet the mechanism underlying metastasis remains unclear. Here, we report that CREPT/RPRD1B, which exhibits somatic gene copy-number amplifications and elevated expression, correlates with poor patient survival and drives TNBC metastasis. We demonstrate that CREPT alters three-dimensional genome structures in topologically-associating domain (TAD) status and chromatin loops via occupying promoters and enhancers. Specifically, CREPT mediates 1082 co-operational chromatin loops configured by enhancer-promoter and promoter-termination loops, which are validated by HiChIP analyses and visualized by Tn5-FISH experiments. These loops orchestrate RNAPII loading and recycling to enhance the metastatic gene expression. Disruption of these co-operational loops using CRISPR-dCas9 suppresses TNBC metastasis in vivo. Furthermore, depletion of CREPT using an AAV-based shRNA blocks TNBC metastasis in both preventative and therapeutic mouse models. We propose that targeting CREPT to disrupt the co-operational chromatin loop structures represents a promising therapeutic strategy for metastatic TNBC.

Graphical Abstract

Highlights

• Somatic CREPT gene copy-number amplifications with elevated expression occur in the metastatic triple-negative breast cancer (TNBC) and correlate with poor patient survival

• CREPT mediates 1082 co-operational chromatin loops configured by enhancer-promoter and promoter-termination loops

• CREPT-mediated co-operational chromatin loops regulate metastatic genes during the metastasis of TNBC

• Depleting CREPT by adeno-associated virus (AAV) blocks TNBC metastasis in preventative and therapeutic mouse models

Introduction

Triple-negative breastcancer (TNBC) is a highly malignant disease characterized by the absence of estrogen receptor (ER) and progesterone receptor (PR) with no amplification of human epithelial growth factor receptor 2 (HER2) [1, 2]. The disease often progresses aggressively to distant organs, making metastasis a major cause of mortality [3, 4]. Extensive research has been conducted to uncover the factors driving TNBC metastasis [5,6,7]. In particular, gene mutations such as TP53, AKT1, PTEN, KEAP1, STK11, EGFR were identified during the tumorigenesis and progression of breast cancers including TNBC [8,9,10]. Interestingly, somatic gene copy-number variations (SCNVs) are reported to be critical factors affecting the tumor growth and responses to therapies in different cancers [11,12,13]. A recent single-cell analysis revealed a strong correlation between gene expression heterogeneity and SCNVs in TNBC patients [14]. Accumulating evidence suggests that abnormal SCNVs may underlie the metastatic characteristics of TNBC [15,16,17,18], as documented in other cancers [19,20,21,22]. However, the mechanisms of gene expression regulation mediated by SCNVs during metastasis remains unclear.

Studies have revealed that gene expression is tightly regulated by enhancers and super-enhancers [23,24,25]. Enhancers typically recruit various transcription factors (TFs) [8], which trigger the recruitment of chromatin-remodeling enzymes to facilitate chromatin accessibility and decorate distinct functional patterns of covalent histone modifications on adjacent nucleosomes [26, 27]. Typically, enhancers exist in primed, poised, and active states. These states are marked by different histone modifications, such as H3 lysine 4 mono-methylation (H3K4me1) for primed enhancers, H3K4me1 and H3 lysine 27 tri-methylation (H3K27me3) for poised enhancers, and H3K4me1 and H3 lysine 27 acetylation (H3K27ac) for active enhancers [28, 29]. A unique class of enhances, termed super-enhancers [30], consists of multiple active enhancers in close genomic proximity [30, 31]. Super-enhancers are characterized by extraordinarily broad and high levels of H3K27ac and obviously strong peaks of mediator complexes [32, 33]. To date, super-enhancers have been reported to be abnormally activated in various cancers, playing critical roles in processes ranging from tumorigenesis to drug resistance and metastasis [32, 34].

The activity of enhancers and super-enhancers has been widely attributed to the chromatin structure [35]. The genome-wide chromosome conformation capture (Hi-C) techniques have revealed key features of the hierarchical chromatin structure, including compartmentalization, topologically-associating domains (TADs), and various forms of chromatin loops [36, 37]. Compartmentalization is represented as a checkerboard structure on the megabase scale, dividing the genome into A- and B-compartments containing active and repressed regions, respectively [38]. At the sub-megabase level, chromatin is further folded into TADs, which appear as triangles on Hi-C maps with sharp boundaries. TADs are thought to be structural and functional domains of the genome that facilitate or restrict interactions between promoters and distant regulatory elements [39]. The TAD structure is considered to regulate the activity of enhancers and super-enhancers. At the kilobase level, chromatin loops were widely observed for their critical roles in regulating the activity of enhancers and super-enhancers [40, 41]. Recently, several types of chromatin loops have been identified at the genome level, including enhancer–silencer loops, insulator–insulator loops, polycomb-mediated loops, architectural loops and enhancer–promoter loops [42]. Among these, enhancer-promoter loops are the most powerful and proximal for gene expression regulation [43, 44]. Enhancer-promoter loops regulate gene expression in an RNA polymerase II (RNAPII)-dependent manner, acting over short or long distances on a genome wide scale [43, 45]. Although numerous transcriptional regulators have been implicated in the formation of chromatin loops, how enhancer-promoter loops regulate transcription to promote tumorigenesis and metastasis remains incompletely understood [46,47,48,49]. In the recent years, studies have showed that the enhancer-promoter loops constitute a major component of chromatin topology [50, 51]. Importantly, alterations in chromatin topology have been linked to tumorigenesis. The altered chromatin topology may change the accessibility of chromatin to activate the oncogenic genes that were silenced in normal cells or to inhibit the tumor suppressor gene expression [52]. Additionally, this alteration may also alter the epigenetic modifications of histones to regulate gene expression and influence the DNA breakage and recombination during tumorigenesis [53, 54]. To date, the chromatin topology alterations in metastasis have gained considerable attention for their role in understanding the mechanisms of cancer development.

CREPT (cell-cycle related and expression-elevated protein in tumors [55], also named RPRD1B) is a homologue of the yeast transcription termination factor Rtt103 [56, 57]. CREPT is recognized as an RNAPII interacting protein due to its RNAPII CTD-interacting domain (CID) and its role in regulating phosphorylation of the RNAPII C-terminal domain (CTD) [57, 58]. Accumulating evidence indicates that CREPT acts as an oncoprotein to promote tumorigenesis in multiple cancer types [57, 59,60,61]. Interestingly, our initial study demonstrated that CREPT could prevent RNAPII from “reading through” the termination site and promote the formation of a promoter-termination loop in the CCND1gene [55]. Another study from our group showed that CREPT occupied both promoters and termination regions of Cd44, Alcam, and EphB1genes in mouse intestinal stem cells [62]. Collectively, these studies suggest that CREPT functions as a regulator for chromatin loop configuration. Here, we report that CREPT, with gene copy-number amplifications, regulates enhancer activities and directly participates in the formation of co-operational chromatin loops configured by enhancer-promoter loops and promoter-termination loops. Remarkably, disruption of CREPT-mediated co-operational loops dramatically blocked the metastasis of TNBC cells in vitro and in vivo. We propose that targeting CREPT and its associated loop structures represents a promising therapeutic strategy for TNBC patients.

Results

CREPT amplification is highly related to the metastasis of TNBC

To search for genes that drive TNBC metastasis, we analyzed somatic gene copy-number variations (SCNVs) between primary and lymph node metastatic (LNM) TNBC tumors using whole-exome sequencing database from The Cancer Genome Atlas (TCGA). We identified 5 amplification regions (20q11.23, 1q21.3, 12p13.33, 12q12, and 8p11.21) that were enriched with significantly elevated positive selection degrees in the LNM TNBC (Fig. 1A). Notably, 20q11.23 was the most significantly amplified region (Fig. 1A, red dot in the right top). Analyses using an expression dataset from TCGA revealed that 117 genes within these amplified regions were upregulated in LNM TNBC patients compared to those in primary TNBC patients (Fig. 1B, Supplementary Table 1). Interestingly, we observed that CREPT was at the top position with its expression among the 117 genes (Fig. 1B). We confirmed that the SCNVs of CREPT were much higher in the 20q11.23 region in LNM TNBC patients than in primary TNBC patients (Fig. 1C). Further analyses revealed that the amplification frequency of CREPT was significantly higher in the LNM TNBC cohort than in the primary TNBC cohort (Figure S1A). Simultaneously, CREPT gene copy-number amplification was strongly correlated with its mRNA levels in TNBC patients (R² = 0.5568, P < 0.001, Figure S1B), suggesting a causal linkage between CREPT gene copy-number amplification and increased gene expression. The elevated CREPT expression in the LNM TNBC was further supported by data from the Gene Expression Omnibus (GEO) and the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) databases (Figures S1C-S1D). Moreover, elevated CREPT levels were significantly correlated with worse overall survival in TNBC patients (Figures S1E-S1F). In line with this, TNBC patients with CREPT copy number amplification exhibited worse prognosis (Figure S1G). Together, these findings demonstrate that CREPT is amplified with increased expression and associated with the metastatic status of TNBC patients.

To examine CREPT expression in intact tumor tissues, we analyzed a single cell database obtained from primary and LNM TNBC patients (GSE118390). The result showed that CREPT was significantly elevated in a subset of cancer cells (Fig. 1D). Notably, all CREPT-high cancer cells were distributed in two patients (PT084 and PT126) with high metastatic potential, whereas the patients without lymph-node metastasis exhibited low CREPT expression (Fig. 1E). Consistently, CREPT gene copy-number was higher in patients with high metastasis than in primary TNBC patients (Fig. 1F). Gene set variation analysis (GSVA) showed that the gene expression profiles of patients PT084 and PT126, which exhibited high CREPT expression, were enriched for metastasis-related processes, including inflammation response, angiogenesis, extracellular matrix secretion, and cell adhesion (Figure S1H). Interestingly, a differential gene expression pattern was observed in tumor cells from patients with high or low CREPT expression (Figure S1I, Supplementary Table 2). Genes upregulated in CREPT-high cells were enriched in biological processes related to metastasis, including focal adhesion, extracellular matrix (ECM)-receptor interaction, collagen fibril organization, and extracellular matrix organization (Figure S1J). In particular, CREPT-high cancer cells co-expressed known metastasis-associated genes, such as IGFBP4, BCL2 and TMPRSS4 (Figure S1K). These results suggest that cancer cells with high CREPT expression represent a highly metastatic subpopulation.

To experimentally validate the bioinformatics results, we performed an immunohistochemistry (IHC) experiment using a tissue microarray comprising 56 primary TNBC cases, 72 LNM TNBC cases, and 16 normal epithelial tissues (Supplementary Table 3). The quantitative analysis using H-score showed that CREPT was nearly undetectable in normal tissues but detected in primary TNBC cases, and was highly expressed in the LNM TNBC cases (Fig. 1G-H). Kaplan Meier analysis showed that high CREPT expression was significantly; associated with poor overall survival (P = 0.021; Fig. 1I), consistent with our previous database analyses (Figures S1E-S1F). Together, these findings suggest that CREPT amplification and expression is critical for TNBC metastasis.

CREPT promotes TNBC metastasis

To validate the role of CREPT in metastatic TNBC, we initially examined the protein level in several TNBC cell lines with varying metastatic abilities in comparison with normal mammary epithelial cell lines. The result showed that CREPT was significantly upregulated in the highly metastatic TNBC cells, including MDA-MB-231, LM2, and 4T1, compared with the low metastatic cells, including MDA-MB-436 and 4TO7 (Figures S2A-S2B). Of note, CREPT remained at a basal level in normal breast cell lines such as human MCF10 A and mouse NMUMG cells (Figures S2A-S2B, left bands). Next, we investigated the influence of CREPT on the invasion and migration abilities of cells, which were stably transfected by a luciferase reporter (4T1-Luc and LM2-Luc). Matrigel invasion and transwell migration assays showed that depletion of CREPT by short hairpin RNAs (shRNAs) (Fig. 2A and S2C) significantly decreased the invasion and migration abilities of 4T1-Luc (Fig. 2B-D) and LM2-Luc (Figures S2D-S2F) cells. Reciprocally, stable overexpression of CREPT using a doxycycline (Dox)-inducible system (Fig. 2E) increased the invasion and migration abilities of 4TO7-Luc cells in a dose-dependent manner (Fig. 2F-H). In addition, CREPT deletion abolished (Figures S2G-S2I), and its depletion impaired (Figures S2J-S2K), the ability of 4T1 cells at forming tumor spheres, whereas overexpression of CREPT enhanced the tumor sphere formation of 4TO7 cells in a dose-dependent manner (Figures S2L-S2M). To distinguish the role of CREPT on the invasion and migration from its function of promoting cell proliferation, we performed a CCK8 and colony formation assays. The results showed that deletion or overexpression of indeed affected the cell proliferation feature (Figures S2N-S2Q), consistent with our previous studies. Together, these results suggest that CREPT promotes not only cell proliferation but also invasion and migration of TNBC cells.

To further assess the function of CREPT on metastasis in vivo, mice implanted with tumor cells via the tail vein were analyzed by bioluminescent imaging (BLI). The result showed that depletion of CREPT in 4T1-Luc (Fig. 2I-J) and LM2-Luc (Figures S2R-S2S) markedly decreased metastasis at multiple organs and extended metastasis-free survival of the mice (Fig. 2K and S2T). Reciprocally, overexpression of CREPT in the low metastatic cell 4TO7 significantly exacerbated the lung-metastatic burden after intravenous inoculation of cancer cells into the mice (Fig. 2L-M) and caused the death of the animals at different times (Fig. 2N). Of note, the metastatic tumors emerged at week 3 after CREPT was induced but the tumor appearance time was delayed to week 6 when CREPT overexpression was delayed (Fig. 2L, comparing Dox-driven expression after week 1 and week 2). These results suggest that the metastatic tumors emerge just when CREPT is induced. Histological analyses showed that the metastatic tumors in the lung were dramatically inhibited or promoted dependent on CREPT depletion in LM2 (Figures S2U-S2V) and 4T1 (Figures S2W-S2X) cells or its overexpression in 4TO7 cells (Figures S2Y-S2Z). To rule out the possibility that the relatively small metastatic tumor burden resulted from slowed tumor cell proliferation, we conducted an in situ metastatic experiment by injecting tumor cells into the fat pads and allowing the tumors to grow to the same size. The result showed that the in situ tumors formed in the fat pads by mock 4T1-Luc cells generated significant lung metastasis, whereas tumors formed by CREPT-depletion cells failed to induce any lung metastasis at the same in situ tumor size with an extended growth for 2 more weeks (Fig. 2O-Q). Simultaneously, histological analyses showed no metastatic tumors in the lung of the mice inoculated with 4T1-Luc cells under CREPT depletion even when inoculated in situ tumors remained at the same size (Fig. 2R-S). Taken together, all these results suggest that CREPT is necessary and sufficient for the metastasis of TNBC.

CREPT upregulates the expression of metastasis-associated genes

To identify metastasis-associated genes that CREPT upregulated, we analyzed both pre- and mature mRNAs in 4T1 cells. A global run-on sequencing (GRO-seq) experiment showed that 3654 genes were significantly downregulated (FDR = < 0.05 & Log2(Fold change(sh-CREPT vs WT) > = 2) at the pre-mRNA level when CREPT was depleted (Fig. 3A, Supplementary Table 4). An RNA sequencing (RNA-seq) experiment showed that 3465 transcripts were downregulated in CREPT depletion cells (FDR = < 0.05 & Log2(Fold change(sh-CREPT vs WT) > = 2) (Fig. 3B, Supplementary Table 5). A gene set enrichment analysis (GSEA) of the transcriptome showed that the hallmark metastasis-associated gene sets were highly enriched in wild-type (WT) but not in CREPT-depletion (sh-CREPT) cells, concordantly at both pre- and mature mRNA levels (Fig. 3C). To confirm the altered gene expression, we compared the alteration in transcriptomes and identified 2574 genes downregulated as mutually detected by GRO-seq and RNA-seq (Fig. 3D, left panels) analyses. A gene ontology (GO) analysis of these downregulated genes revealed that the top clusters were associated with multiple metastatic events (Fig. 3D, right panel), while the up-regulated genes seemingly appeared in diverse bioprocesses (Figure S3B). All these results suggest that CREPT is critical for the expression of metastasis-related genes during TNBC metastasis.

To confirm the protein alteration of the downregulated genes, we performed a Tandem Mass Tag (TMT)-based quantitative proteomics experiment. We correlated the protein abundance to the level of mature mRNA and marked known metastasis-associated proteins. The result showed that 1924 proteins were decreased when CREPT was depleted (Fig. 3E, Supplementary Table 6), in correspondence to decreased mRNAs as identified by our GRO-seq and RNA-seq analyses (see Fig. 3D, 2574 genes). This result suggests that 75% (1924/2574) of altered genes correspond to changed proteins under CREPT depletion, although 25% of genes showed no regulation on protein level, which might be an indirect regulation. Of note, 15.7% (302/1924) of decreased proteins such as Mmp13, Vcam1, Ccl2 and Itga3 had been reported as metastasis-related factors (Fig. 3E, blue dots) [63], as denoted by the Human Cancer Metastasis Database (HCMDB) (http://hcmdb.i-sanger.com/). This result suggests that depletion of CREPT preferably influences proteins functioning during metastasis.

Then, we performed RT-qPCR and nuclear run-on experiments to examine changes of mRNA and nascent transcripts for critical metastatic genes. The result showed that the expression of genes involved in cell adhesion (Itgb3, Itga2, Nectin4, Vcam1, and Itgb1), inflammation (Cxcl1, Ccl2, Cxcl12, Cxcl13, and Cxcr4), tumor angiogenesis (Macc1, Angptl4, Vegfc, and Tgf-a), and extracellular matrix remodeling (Col1a1, Myof, Postn, Serpine2, Mmp2, Mmp9, and Mmp12) was significantly decreased when CREPT was depleted by two shRNAs (Fig. 3F). We further confirmed the alterations of these genes at pre-mRNA levels (Figure S3A). Collectively, our data suggest that CREPT is critical for the regulation of genes involved in metastasis.

CREPT regulates the activity of enhancers in a p300-dependent manner

To decipher how CREPT regulates gene expression, we determined to identify CREPT-interacting proteins occupying the chromatin by a chromatin immunoprecipitation followed by mass spectrometry (ChIP–MS) analysis using an antibody against endogenous CREPT [64]. We ultimately obtained 107 transcription-related regulatory proteins that interacted with CREPT and occupied the chromatin (Supplementary Table 7). We then clustered these proteins into five groups according to their functions in transcriptional regulation (Fig. 4A, left panels). Notably, among the CREPT-interacting proteins, 15 transcription factors, 42 regulators of transcription initiation, elongation and termination, and 25 factors involved in the epigenetic modification of the chromatin emerged. Indeed, several proteins, including Wrd43, RPRD2, Pcf11, RPRD1a, Xrn2, and Scaf4/8, were reported to interact with CREPT by other groups [65, 66]. Another analysis based on epigenetic regulation revealed that most of these CREPT-interacting proteins mainly function by regulating both enhancer activities and chromatin structure configurations (Fig. 4A, right panel). All these findings indicate that CREPT may regulate enhancer activities and maintain chromatin structures of tumor-associated genes.

To illustrate the role of CREPT in chromatin configurations, we performed a ChIP-seq experiment and characterized its chromatin occupancy pattern in TNBC cells. The result showed that the majority of CREPT occupancy sites were located at promoter regions and intergenic regions in 4T1 (Fig. 4B, 33.71% and 36.56%) and LM2 (Figure S4A, 43.69% and 21.64%) cells. We used antibodies against H3K4me1, H3K4me3, and H3K27ac in ChIP-seq experiments to find active promoters and enhancers that CREPT binds. In total, 10415 promoters where CREPT, H3K4me3, and H3K27ac were concomitantly enriched (Figures S4B-S4C) and 8618 enhancers where CREPT, H3K4me1, and H3K27ac co-existed (Fig. 4C-D) were identified in 4T1 cells. Of note, CREPT-occupied active promoters accounted for 29.0% (10415/35973) of CREPT-occupied sites (35973) and 77.2% (10415/13488) of active promoters (10415 + 3073 = 13488) (Figure S4C). Coincidentally, CREPT-occupied active enhancers accounted for 24.0% (8618/35973) of CREPT-occupied sites (35973) and 48.9% (8618/17641) of active enhancers (8618 + 9023 = 17641) (Fig. 4C). Similarly, we identified 6147 active enhancers where CREPT occupied in LM2 cells (Figures S4D-S4E). These results suggest that CREPT occupies the chromatin at active promoters and enhancers. To examine whether CREPT regulates the activity of active promoters and enhancers, we performed ChIP-seq experiments in both mouse and human cells. The results showed that deletion of CREPT significantly reduced H3K4me3 and H3K27ac signals at the promoter regions (Figure S4 F) and H3K4me1 and H3K27ac signals at the enhancer regions (Fig. 4E for mice, Figure S4G for human). All these results suggest that CREPT regulates the activation of both promoters and enhancers.

As CREPT-occupied active promoters and enhancers appeared comparable in their numbers (10415 vs. 8618) and percentages (29.0% vs. 24.0%), we determined to examine the density of CREPT occupancy between active promoters and enhancers. The result demonstrated that CREPT was significantly more abundant in active enhancer regions than in active promoter regions (Figure S4H). This result suggests that CREPT may function preferably by occupying active enhancers. We then analyzed factors in CREPT-occupied active promoters and enhancers and listed top 10 transcriptional factors (Figure S4I). Intriguingly, these top 10 transcription factors at the enhancers (CTCF, NRF1, FOS, YY1, RUNX1, STAT1/3, TEAD1, KLF4, ELF3, and FOXC1) were denoted to relate to tumor malignancy, whereas the top transcription factors binding to the promoters (E2F3, SOX15, HOXA5, Pitx1, ZNF740, ETS1, DUX4, TCF12, GATA5, and HMX2) functioned preferably on cell proliferation [68]. These findings indicate that CREPT regulates metastasis by activating enhancers.

To investigate whether CREPT affects the super-enhancer configuration, we examined the coverage size of enhancers and the level of histone modifications, including H3K4me1 and H3K27ac. In total, we identified 1058 (4T1 cells) and 643 (LM2 cells) super-enhancers from 8618 (4T1 cells) and 6147 (LM2 cells) CREPT-mediated active enhancers respectively (Fig. 4F-G for mouse 4T1 cells; Figures S5A-S5B for human LM2 cells). Interestingly, we observed that the CREPT occupancy density was much stronger at super-enhancers than at typical enhancers (Figure S5C). We found that when CREPT was deleted, the average size and density of H3K4me1 (Fig. 4F and S5A) and H3K27ac (Fig. 4G and S5B) at super-enhancer regions became much smaller. This result suggests that CREPT controls the configuration of super-enhancers. As an example, we analyzed the feature of an enhancer that regulates Vegfc and a super-enhancer that regulates Slc12a5. Markedly, the binding densities of the enhancer and super-enhancer markers were much stronger at the super-enhancer than at the typical enhancer (Figure S5D, left vs. right in WT), and deletion of CREPT dramatically decreased activities of both enhancers and super-enhancers (Figure S5D, WT vs. CREPT KO). Taken together, all these results suggest that CREPT regulates the enhancer configuration, with a stronger effect on the formation of super-enhancers than typical enhancers.

To unfold how CREPT regulates activities of typical enhancers and super-enhancers, we focused on the histone acetyltransferase p300, a well-known regulator of enhancers [69, 70]. Our aforementioned ChIP–MS analyses showed that both p300 and CREPT were connected to chromatin concurrently (see Fig. 4A). Here, we show that CREPT helps p300 remain longer in the enhancer regions. A reciprocal co-immunoprecipitation experiment showed that a strong interaction occurred between endogenous CREPT and p300 proteins in 4T1 cells (Figure S5E), which is consistent with our previous report [71]. A ChIP-seq experiment revealed that distributions of CREPT and p300 were very similar, both centered at an average unit across active enhancers and super-enhancers (Figure S5F). A Venn diagram analysis showed that 6523 enhancers were overlapped by CREPT- and p300-binding peaks (Figure S5G). Intriguingly, we observed that deletion of CREPT reduced p300 recruitment at enhancers (Figure S5H). This result is consistent with our aforementioned observation that deletion of CREPT reduced the level of H3K27ac (see Fig. 4E, right panel). We also profiled the effect of CREPT on physical chromatin accessibility using an assay for transposase-accessible chromatin followed by sequencing (ATAC-seq). The result showed that CREPT deletion resulted in a significant loss of chromatin accessibility at CREPT-bound enhancer regions in 4T1 cells (Figure S5I). All these findings suggest that CREPT interacts with and promotes p300 loading onto enhancer regions thereafter to activate enhancers and super enhancers.

CREPT influences the large-scale chromatin structure

To investigate the consequence of CREPT deletion on the three-dimensional (3D) chromatin structure at the genome scale, we performed a tethered chromatin conformation capture experiment (Hi-C) in 4T1 cells. Two libraries from wild-type and CREPT-deletion cells were sequenced to a total depth of over 12 billion reads. The high-quality deep-sequencing data were validated with a high cis-interaction rate and the analysis reached 1 kb resolution. The result showed that deletion of CREPT decreased the interaction of chromatins at the whole genome level (Figure S6A). Notably, deletion of CREPT dramatically reduced strong intra-interactions of chromatins of chromosomes 3, 8, 11, 15, 17, and X but had no effect on weak extra-interactions between other chromatins (Figure S6A, WT vs. CREPT KO). On the other hand, deletion of CREPT appeared barely to alter the configuration of chromatin A/B compartments (Figure S6B). Overall, we observed more switches at the chromatin of chromosome X but less switches of A/B compartments in autosomal chromatins (Figures S6C-S6D). A quantitative analysis for the genome sizes with compartment A/B and B/A switches indicated significant difference between the chromatin of chromosome X and autosome chromatins (Figure S6E). However, we observed no significant gene expression changes within these regions (Figure S6F). These results indicate that the role of CREPT in regulating gene expression may not be deciphered at the level of A/B compartments.

To unveil the alteration of chromatin interactions at a high resolution level, we focused on the configuration of TADs, which are defined by a pair of strong boundaries measured by insulation scores and typically span a distance from approximately 100 kb to 1 Mb in murine and human genomes [72]. Using Arrowhead Software [40], we detected 3387 TADs in wild-type cells and 2901 TADs in CREPT deletion cells (Fig. 5A). Of note, the numbers of TADs with larger sizes (over 300 kb) were decreased in CREPT deletion cells (Fig. 5A, compare the columns with over 300 kb). Interestingly, we observed a dramatic reduction in the strengths and peaks of TADs when CREPT was deleted (Fig. 5B), indicating that CREPT may potentially influence the configuration of TADs. Indeed, we found that 17.95% of TADs were split or disappeared (Fig. 5C, top), and 4.16% were merged (Fig. 5C, middle) although 77.88% remained unchanged (Fig. 5C, bottom) when CREPT was deleted. We observed ubiquitous split patterns of TADs in different chromatins (Figs. 5D and S7A). Taken together, these results demonstrate that deletion of CREPT dramatically alters the TAD configuration in tumor cells.

To address why TADs were split under CREPT deletion, we examined their boundaries using insulation scores [73]. The result showed that deletion of CREPT dramatically altered distributions of the insulation scores, with a decrease in the center and an increase at two side-margins along with TAD boundaries, compared with wild-type cells (Fig. 5E), indicating that CREPT might maintain TAD boundaries. To examine whether CREPT occupies TAD boundaries, we analyzed its occupancy density on chromatins. Indeed, the CREPT occupancy peak appeared at TAD boundaries (Fig. 5F), suggesting that CREPT physically regulates the TAD boundary establishment. For example, we illustrated the alteration of TAD boundaries in the chromatin of chromosome 5. The result showed that CREPT deletion induced splits of large TADs and generated small TADs (Fig. 5G, top), which led to the generation of a new TAD boundary and the disappearance of cis-element interactions of several genes (Fig. 5G, middle). Similarly, the alteration of TADs was observed in the chromatin of chromosome 1 as another example (Figure S7B). These results implied that CREPT-altered TADs might regulate the boundary and cis-element interactions. We speculate that the genes with the boundary alteration and cis-element interaction changes may be affected when CREPT is deleted. Indeed, our RNA-seq results showed that depletion of CREPT decreased the expression of genes (Cxcl1 and Cxcl2) with boundary and cis-element alterations but had no effect on genes (Gm30664, Pf4, Cxcl5, Ppbp, and Cxcl3) without boundary and cis-element alteration in CREPT-altered TADs (Fig. 5H). We also observed that genes (Wdr75 and Sk40a1) of boundary and cis-element alterations in the chromatin of chromosome 1 were downregulated by CREPT deletion (Figure S7C). Overall, we identified that 342 genes with boundary and cis-element alterations were regulated by CREPT-altered TADs in the whole genome (Fig. 5I and S7D, Supplementary Table 8). These results suggest that CREPT regulates the TAD configuration to maintain the boundary and cis-element interactions for gene expression (Figure S7E). Taken together, our results suggest that deletion of CREPT influences TAD configuration and related gene expression. However, the number of downregulated genes with boundary and cis-element alterations in CREPT-altered TADs (342 genes, Fig. 5I) was far less than the number of downregulated genes observed by RNA-seq (3456 genes, Fig. 3B). Therefore, we speculate that changes in other types of chromatin structures, possibly the chromatin loop configuration, could also be the cause of the gene expression alteration regulated by CREPT.

CREPT promotes the formation of chromatin loops

To determine whether CREPT influences chromatin loop formation, we analyzed our Hi-C data using a modified differential loop detection strategy [74]. The result showed that 27363 loops disappeared or decreased, while 5805 loops emerged or increased in CREPT deletion cells, based on 63911 loops detected in wild-type 4T1 cells (Figure S8A). Additionally, the overlap analysis of CREPT ChIP-seq peaks with chromatin loop anchors revealed that 26,952 peaks (74.67% of total peaks) co-localized with loop regions, significantly higher than random expectations (permutation test, P < 0.0001). Enriched peaks were predominantly located in gene promoter regions (5828 16.15%), enhancer regions (8143, 22.56%), transcription termination regions (1932, 5.35%), and other regions (11049, 30.61%), suggesting that CREPT may influence transcriptional activity by regulating chromatin loop structures (Figure S8B). We characterized CREPT-mediated loops as chromatin structures anchored at CREPT binding sites and proceeded to analyze their diverse configurations. Notably, we identified 38072 CREPT-mediated loops, accounting for 59.6% of total 63911 loops (Fig. 6A, first and second panels), suggesting that CREPT regulates the loop formation for the majority of genes in tumor cells. Among these loops, we observed a configuration of enhancer-promoter (E-P) loops where the anchor bridges the enhancer and promoter regions. Simultaneously, we found that the super-enhancer-associated interactions, assayed by PET (paired-end tag) counts, were much more than typical enhancer-associated interactions (Figure S8C), suggesting that super-enhancers primarily mediate this loop configuration. Overall, CREPT mediated 15472 E-P loops (Fig. 6A, the third panel).

Interestingly, we observed another type of chromatin loops, which we defined here as promoter-termination (P-T) loops where the anchor crosses promoter and termination regions (Fig. 6A, right panel). This echoes our previous findings that CREPT mediates a loop conformation between promoter and termination regions in the CCND1 gene and occupied both promoter and termination regions in Cd44, Alcam, and EphB1genes [55, 62]. In this study, we showed that CREPT occupancy peaks emerged at the promoter (see Figure S4F) and termination regions (Figure S8D), which were nearly abolished upon CREPT deletion (Figures S4F and S8D, CREPT KO panels). Overall, 2126 CREPT-mediated P-T loops (63.0% of total) were identified (Fig. 6A, right panel, Figure S8E). An aggregate peak analysis (APA) from the Hi-C results revealed a marked reduction in contact strength and/or frequency of these loops when CREPT was deleted in 4T1 cells (Fig. 6A, bottom). Furthermore, HiChIP experiments using a CREPT antibody [64] confirmed its direct involvement in 26412 chromatin loops, including 11214 E-P loops and 813 P-T loops (Fig. 6B), concordant with the Hi-C results. Collectively, these results demonstrate that CREPT is critical for the chromatin loop configuration, particularly for E-P and P-T loop establishment.

To determine whether CREPT-mediated loops regulate gene expression, we analyzed the number of genes associated with different loop types. We found that 10295 genes were regulated by genome-wide loops, while 6213, 4383, and 1652 genes were regulated by CREPT-mediated loops, E-P loops, and P-T loops, respectively (Fig. 6A, middle panels). Obviously, the gene numbers were much lower than the corresponding loop numbers, implying that one gene could be regulated by multiple loops. Notably, we identified a subset of genes contemporaneously regulated by both E-P loop and P-T loops, which we defined here as the co-operational loops (Fig. 6C, upper panel). In total, we identified 1082 genes with the co-operational loop structure (Fig. 6C, bottom panel).

By integrating RNA-seq and Hi-C data, we systematically analyzed gene expression patterns associated with different chromatin loop configurations. The results showed that the mRNA levels of the genes with CREPT-mediated E-P loops or P-T loops were significantly higher than those of the genes with non-CREPT-mediated loops (Fig. 6D, column 2 vs. 1, 5 vs. 4). Strikingly, we observed that the expression of genes with the co-operational loop was higher than that of genes with a single E-P or P-T loop (Fig. 6D, column 8 vs. 2 and 5), indicating that the synergistic interaction between E-P and P-T loops in co-operational configurations exerts a more potent regulatory influence on gene expression. Furthermore, the mRNA levels in both E-P and P-T loop-regulated genes were decreased when CREPT was deleted (Fig. 6D, columns 3 vs. 2, 6 vs. 5). Consistently, CREPT deletion caused a more pronounced reduction in expression for genes with co-operational loops compared to those with single E-P or P-T loops (Fig. 6D, column 9 vs. 7). These results collectively demonstrate that CREPT-mediated co-operational loops represent a sophisticated and highly effective mechanism for gene regulation through coordinated chromatin loop formation.

To elucidate the functional characteristics of genes regulated by CREPT-mediated chromatin loops, we conducted a comprehensive Gene Ontology (GO) analysis. Our investigation revealed distinct functional enrichment patterns among genes associated with different loop configurations, which were systematically categorized into metastasis-, development-, and proliferation-related biological processes (Figure S8F). Although some molecular pathways may overlap between these categories, we identified predominant functional associations for genes with E-P loops, P-T loops, and co-operational loops. In particular, 1082 genes with the co-operational chromatin loop were denoted to contribute to metastasis-associated events such as extracellular matrix organization, the integrin-mediated signaling pathway, vascular development, and focal adhesion (Figure S8 F, left panels). Of note, the top-ranked genes based on co-operational loop density, including Itga3, Mmp13, Tgf-a, Ccl2, Itga2, S100a4, Serpine2, Cxcl1, Vcam-1, and Mmp12, have been extensively documented in literatures as key regulators of metastatic processes (Fig. 6E) [4, 5, 63, 75]. We confirmed that CREPT deletion disarmed the co-operational loop configuration, as evidenced by the structural changes observed in Itga3 (Fig. 6F), Mmp13 (Figure S8G), and Ccl2 (Figure S8H) by Hi-C and HiChIP experiments. These findings establish that CREPT-mediated co-operational loops play a crucial role in regulating metastasis-associated gene expression, in concordance with our aforementioned observations from GRO-seq and RNA-seq experiments (see Fig. 3D).

To precisely characterize the spatial organization of co-operational loops at sub-kilobase resolution, we employed Tn5-based fluorescence in situ hybridization (Tn5-FISH [76]) to visualize the structure. We designed specific probes against the enhancer (labeled with TAMRA, red), promoter (labeled with Alexa Fluor 488, green), and termination region (labeled with Alexa Fluor 647, blue) of the Itga3 gene (Fig. 6G). High-resolution imaging revealed distinct nuclear foci (two specific dots emerged in the nucleus), demonstrating the specificity of the probes (Fig. 6H, top panels). In wild-type cells, we observed complete colocalization of all three fluorescent signals, as all the red, green, and blue dots were distributed at the same location, indicating that co-operational loop conformations were maintained (Fig. 6H, top panels, WT). However, CREPT deletion resulted in spatial separation of these signals, with separated distribution of differently-colored dots, indicating a collapse of the co-operational loop architecture (Fig. 6H, bottom panels, CREPT KO). This pattern of loop disruption was consistently observed in two additional metastasis-related genes: Mmp13 (Figures S9A-S9B) and Ccl2 (Figures S9C-S9D), where CREPT deletion similarly led to loss of three-way chromatin interactions. These complementary findings provide compelling visual evidence that CREPT is essential for establishing and maintaining co-operational loop configurations in tumor cells.

CREPT-mediated co-operational loops orchestrate the RNAPII loading and recycling

To decipher how the co-operational loop regulates gene expression, we employed CRISPR-dCas9-mediated targeted disruption of super-enhancer regions in the Itga3, Mmp13, and Ccl2 genes (Fig. 7A) [77, 78]. Tn5-FISH analyses revealed that CRISPR-dCas9-based sgRNAs against the Itga3 gene (dCas9-E-Itga3) effectively disrupted the co-operational loop architecture, as evidenced by spatial separation of the enhancer, promoter, and termination signals (Fig. 7B, dots with different colors). Consistently, 3C analyses showed that dCas9-E-Itga3 disrupted the E-P loop (Fig. 7C), accompanied by significant reduction in chromatin occupancy of CREPT, H3K4me1, H3K27ac, and RNAPII at the super-enhancer region (Fig. 7D). As a consequence, we observed that this disruption led to diminished RNAPII recruitment at the promoter region (Fig. 7E), suggesting that co-operational loops facilitate RNAPII loading at the Itga3 promoter through E-P loop formation. Simultaneously, we detected that the P-T loop was broken (Fig. 7F) when the super-enhancer was interrupted, indicating impaired RNAPII recycling efficiency. Finally, we observed that disrupting the co-operational loop by dCas9-E-Itga3 dramatically decreased the expression of Itga3 at both mRNA (Fig. 7G) and protein (Fig. 7H) levels.

Similar results were observed in Mmp13 (Figures S10A-S10G) and Ccl2 (Figures S10H-S10 N) genes. All these results suggest that the super-enhancer, which directs the E-P loop and then the P-T loop formation, maintains the CREPT-mediated co-operational loop structures. In conclusion, we propose that CREPT-mediated co-operational loops function as molecular platforms that sequentially orchestrate RNAPII loading through E-P loop and subsequent recycling via P-T loops, thereby optimizing transcriptional efficiency.

To demonstrate the significance of co-operational loop structures, we leveraged our Dox-inducible system to overexpress CREPT (see Fig. 2E) and then interrupted co-operational loops in different genes in 4TO7 cells, which exhibit moderate endogenous CREPT expression (see Figure S2B). The result showed that the addition of Dox promoted co-operational loop formation in the Itga3 locus (Fig. 7I-L). Notably, we observed that interrupting the super-enhancer by dCas9-E-Itga3 (Fig. 7I, red columns) impaired the occupancy of H3K4me1, H3K27ac, and RNAPII at the enhancer (Figs. 7J, red columns) and diminished RNAPII recruitment at the promoter (Fig. 7K, red columns). Consequently, P-T loop formation was compromised by dCas9-E-Itga3-mediated super-enhancer disruption even under CREPT overexpression conditions (Fig. 7L). Furthermore, co-operational loop disruption completely abrogated Itga3 gene expression despite elevated CREPT levels (Figs. 7M-N). This regulatory paradigm was consistently observed in Mmp13 (Figures S11A-S11F) and Ccl2 (Figures S11G-S11L) genes. Taken together, our Tn5-FISH and Hi-C analyses indicate that CREPT facilitates E-P loop formation, which boosts the activation of enhancers and brings RNAPII to the promoter region to generate P-T loops. These findings collectively establish that CREPT-mediated co-operational loops function as integrated regulatory platforms that coordinate RNAPII loading and recycling through spatially organized chromatin interactions, thereby optimizing transcriptional output.

To investigate the pathological relevance of genes with co-operational loops, we evaluated the metastatic potential of the top 10 genes that we identified to have strong CREPT-mediated co-operational loop densities (see Fig. 6E). Using pairs of guide RNAs (sgRNAs) targeting super-enhancers to disrupt co-operational loop structures (see Fig. 7A), we systematically screened all these genes based on their cancer cell invasion and migration capabilities. Strikingly, among all tested genes, we observed that the co-operational loop structure formed in Itga3, Mmp13, or Ccl2 emerged to have the most pronounced impact on promoting invasion and migration (Fig. 7O). Simultaneously, we found that disrupting the co-operational loop structure decreased the colony formation (Figure S12A), but the effect appeared less than that in the invasion and migration (comparing Figure S12A with Fig. 7O). Reminiscently, these 3 genes were widely acknowledged for their abilities in the metastasis of tumors [22, 63, 79, 80]. These results suggest that deletion of CREPT impairs metastatic gene expression by disrupting chromatin loop architecture and compromising enhancer/super-enhancers activity.

To evaluate the effect of CREPT-mediated co-operational loops on metastasis in vivo, we established a metastatic model by tail vein injection of 4T1 cells into mice. BLI analyses showed that the metastatic tumors at multiple organs were markedly decreased by CRISPR-dCas9 sgRNAs targeting CREPT-mediated co-operational loops at the metastatic genes Itga3, Mmp13, and Ccl2 (Figs. 7P-Q). Importantly, targeted disruption by all three CRISPR-dCas9 sgRNAs significantly extended the metastasis-free survival of the mice (Fig. 7R). To further address whether CREPT promotes metastasis via the co-operational loop structure, we disrupted the co-operational loop structures by CRISPR-dCas9 s against different genes in 4TO7 cells, where CREPT overexpression was driven by Dox. Strikingly, while CREPT overexpression enhanced metastatic potential, simultaneous disruption of co-operational loops at key genes attenuated metastatic tumor growth (Figures S12B-S12C) and prolonged the survival time of the mice (Figure S12D). Histological examinations confirmed that disruption of the co-operational loops effectively blocked lung metastasis even in the presence of Dox-induced CREPT overexpression (Figures S12E-S12F). These in vivo findings demonstrate that CREPT drives metastasis initiation through the establishment and maintenance of functional co-operational loop structures at critical metastasis-associated genes.

Targeting CREPT is a promising strategy for the therapy of metastatic TNBC

To investigate the functional relevance of CREPT-mediated co-operational loops in human metastatic TNBC, we analyzed super-enhancer activity in LM2 cells. The result showed that deletion of CREPT significantly decreased occupancy densities of H3K4me1 and H3K27ac, particularly in the super-enhancer regions of ITGA3, MMP13, and CCL2 (Figures S13A-S13C). Using a CRISPR-dCas9 system, we observed that the loop structure was consistently disrupted by targeting the supper-enhancers of ITGA3, MMP13, and CCL2 genes (Figures S13D-S13I), leading to a dramatic repression of their expression (Figures S13J-S13O). As a consequence, we observed that interrupting the co-operational loops in these genes dramatically repressed tumor metastasis and prolonged survival time of the mice that burdened human tumor cells (Figures S13P-S13T). These results suggest that CREPT-mediated co-operational loop structures are critical for governing metastatic gene expression and tumor progression in human TNBC.

To explore the potential clinical application, we developed an AAV-based gene therapy system for CREPT deletion (AAV-driven shRNA against CREPT, AAV-shCREPT). The engineered AAV-shCREPT vector demonstrated high transduction efficiency, achieving viral titers comparable to control AAV (Figure S14A), and effectively suppressed CREPT expression in human TNBC cells (Figure S14B). In a prevention model (Fig. 8A), systemic administration of AAV-shCREPT via tail vein injection significantly inhibited distal metastasis formation (Figs. 8B-C). Strikingly, we observed that while all control mice succumbed to metastatic disease, 100% of AAV-shCREPT-treated mice survived throughout the experimental period (Fig. 8D). A histology analysis demonstrated that lung metastatic nodules were significantly decreased in AAV-shCREPT-treated animals (Figs. 8E and S14C). Immunohistochemical staining confirmed significant downregulation of CREPT and its downstream targets ITGA3, MMP13, and CCL2 in lung tissues from treated mice (Fig. 8F). These preclinical findings demonstrate the therapeutic potential of AAV-mediated CREPT inhibition for suppressing metastatic progression in human TBNC.

To assess the therapeutic efficacy of AAV-shCREPT on metastasis, mice were allowed to develop metastatic tumors for 3 weeks before AAV treatment (Fig. 8G). The results demonstrated that AAV-shCREPT administration significantly inhibited metastatic progression, extended overall survival (Figs. 8H-J), and reduced metastatic protein expression in lung tissues (Figure S14D). To delineate the specific role of CREPT in metastasis versus tumor proliferation, we conducted an orthotopic tumor model by inoculating tumor cells into mammary fat pads (Fig. 8K). We compared lung metastasis at time points when the inoculated tumors reached the same volume. The result showed that primary in situ tumors formed by the mock LM2-Luc cells exhibited significant lung metastasis at week 5, whereas tumors formed by AAV-shCREPT-treated cells showed no lung metastasis at week 6, even when the inoculated tumors reached the same size as those in the mock group (Figs. 8K, S14E). Notably, histological analysis revealed no metastatic lesions in the lungs of mice bearing LM2-Luc cells after AAV-CREPT treatment at weeks 5 and 6 (Figs. 8M-N). Additionally, the expression levels of ITGA3, MMP13, and CCL2 were nearly undetectable (Figure S14F). These findings demonstrate that CREPT targeting preferably inhibits the metastatic cascade in TNBC with also an effect on primary tumor growth, highlighting its potential as a therapeutic strategy for metastatic diseases.

To more closely model clinical management of metastatic disease, we evaluated AAV-shCREPT efficacy in a post-surgical intervention setting following resection of primary orthotopic tumors (Fig. 8O). The results showed that control mice rapidly developed pulmonary metastases despite complete removal of primary mammary fat pad tumors (Fig. 8P, top panel). In striking contract, we observed that mice treated with AAV-shCREPT exhibited significantly reduced metastatic burden in the lungs (Figs. 8P-S), with concomitant suppression of metastasis-associated gene expression (Fig. 8T). These preclinical findings demonstrate that targeting CREPT effectively blocks the metastasis of human TNBC in both prevention and therapeutic settings.

Discussion

Metastasis represents the primary cause of mortality in triple-negative breast cancer (TNBC) patients, posing a significant clinical challenge despite advances in primary tumor management [1,2,3,4]. A great effort has been focused on identifying driver genes that orchestrate the initiation of metastatic TNBC to develop more effective therapeutic interventions for this aggressive disease [7, 19, 81]. In this study, we discovered that CREPT is a critical metastatic driver gene in TNBC. Our results demonstrated that CREPT ranked at the top among all other genes with copy-number alterations, with a strong correlation to its high expression and overall survival in LNM TNBC. Using both in vitro and in vivo models, we demonstrated that deletion or depletion of CREPT impeded the metastasis of TNBC. Mechanically, we elucidate that CREPT facilitates the formation of co-operational chromatin loop structures, integrating enhancer-promoter (E-P) and promoter-termination (P-T) interactions into a unified regulatory platform (Fig. 9). Intriguingly, we discovered that CREPT-mediated co-operational loops preferentially regulate a network of metastasis-associated genes, and targeted disruption of these loops significantly attenuates metastatic progression while improving survival outcomes in preclinical models. Concordantly, deletion of CREPT profoundly disrupts co-operational loop architectures and downregulates key metastatic regulators genes including Itga3, Mmp13, and Ccl2. These findings position CREPT as a promising therapeutic target for metastatic TNBC, offering potential for clinical translation in this challenging disease context.

Identifying genuine metastatic driver genes remains a formidable challenge [82]. Conventional strategies were based on the differential gene expression between primary and metastatic tumors [63]. However, since metastatic tumors migrated and implanted in the target organs have undergone a series of alterations at the genome level, including gene mutations, expressions, and epigenetic modifications [1, 2, 75, 82], the identified genes with changes could be from a final consequence after tumor evolution, and precise metastatic initiation events might be neglected. Therefore, novel mutations in somatic cells and germlines should be examined for the initiation of TNBS metastasis [15, 18, 20]. In this study, we sought to search for genes that drive the initiation of tumor metastasis. We decided to compare primary tumors with or without the ability of lymph node metastasis, as numerous studies have indicated that tumors with lymph node metastasis have the characteristics at the metastatic initiation before spreading into distal organs [15, 83]. We have identified five amplified regions in TNBC patients with lymph node metastasis. Notably, 20q11.23 was the most significantly altered chromatin region containing the CREPT gene. Intriguingly, we found that the SCNVs of CREPT were much higher in LNM TNBC patients than primary TNBC patients at 25 ~ 40 Mb genomic position. Although another set of SCNVs at 15 ~ 25 Mb genomic position (encoded TTI1 gene) in the primary TNBC patients appeared higher than the LNM TNBC patients, we consider that the highly elevated CREPT SCNVs might be selectively amplified during the metastatic process. Since TTI1 encodes a subunit of the mTOR complex that regulates the proliferation and metabolism of tumor cells [84], we speculate that these two different genes might be related to the different preference of tumor cells to their stages to adopt the new microenvironments including the balance of metabolic alteration and proliferation during their evolution. We then used a set of experiments to prove that the identified CREPT gene is responsible for the metastasis of TNBC. These experiments include deletion of CREPT in 4T1, a highly metastatic cell line with a high level of CREPT expression, and overexpression of CREPT in 4TO7, a non-metastatic cell line with a low level of CREPT expression. Importantly, we confirmed the ability of CREPT to induce metastasis in mice with in situ primary tumors in the mammary fat pads. Strikingly, we observed no metastasis when CREPT-mediated co-operational loop structures were disrupted. All these in vitro and in vivo results indicate that CREPT is a driver gene for the metastasis of TBNC.

Contemporary understanding of gene expression regulation has been revolutionized by the recognition of three-dimensional genome architecture, characterized by A/B compartments, TADs, and high resolution chromatin loops [36,37,38,39,40,41,42]. Many technologies have been developed to study 3D genome structures [38, 85, 86]. In this study, we employed different methods including Hi-C, HiChIP, and Tn5-FISH to elucidate the role of CREPT in shaping 3D genome organization. We observed that CREPT influenced the large-scale chromatin structure at the TAD configuration maintenance, seemingly not affecting the A/B compartments. At the high-resolution scale, we found that CREPT regulates chromatin loop formation. We propose that CREPT induces gene expression in metastasis mostly through co-operational-loop structures configured by E-P and P-T loops. This hypothesis was also based on our previous studies that CREPT occupied both the promoter and termination regions in the CCND1gene and other genes [55, 62]. Through comprehensive Hi-C and HiChIP analyses, we identified 1082 genes regulated by CREPT-mediated co-operational loops. Convincingly, we used Tn5-FISH, a recently developed method to visualize 3D genome structures [86], to demonstrate the occurrence or collapse of co-operational loops in the presence or absence of CREPT. Strikingly, targeting the co-operational loops using a CRISPR-dCas9 strategy almost abolished gene expression in regulating tumor metastasis, even under the CREPT overexpression condition. Our data provided compelling evidence that CREPT orchestrates gene expression programs through the establishment of co-operational loop chromatin structures, offering new insights into the genome structural basis of transcriptional regulation in cancer.

However, the molecular mechanisms underlying CREPT-mediated co-operational loop formation remains to be fully elucidated. Based on the analysis of CREPT interactomes, we propose several mechanisms that may collectively contribute to this process. First, CREPT may coordinate with transcriptional factors like Hmgb1/2, Dek, JunB, Kif4, Runx1, Ect1, Yy1, Set1, Yap1, and Stat1, as identified in our interaction studies (see Fig. 4). This hypothesis is supported by the consistent presence of corresponding transcription factor binding sites in CREPT-mediated promoters and enhancers (see Figure S4I), with Yy1, Runx1, and Stat1 being particularly enriched in CREPT-mediated enhancers. These findings are reminiscent of our previous observations that CREPT interacted with transcriptional factors such as STAT3 and β-catenin/TCF4 [59, 71]. Researchers have reported that YY1 mediates broad enhancer-promoter loops in both tumor and embryo [87], and Hmgb1/2/3 family proteins bend DNA to form loops in T cells and tumor cells [88]. Additionally, reports suggest that Yap1, JunB, and Runx1 regulate enhancer activities by recruiting costimulatory factors [89,90,91]. Therefore, it is possible that the interaction of CREPT with these transcriptional factors may mediate co-operational loop formation. A second plausible mechanism is that CREPT-driven loop formation might be through enhancer RNAs (eRNAs), which were widely reported to participate in chromatin loop formation primarily by interacting with chromosomal cyclization factors [92, 93]. We observed that several chromosomal cyclization factors, including Rad21, Smac1a, Smac2/4, Lamin A, Med12, and Chd4, appeared to interact with CREPT in our sets of interacting proteins (see Fig. 4A). We speculate that CREPT might associate with these chromosomal cyclization factors to recruit eRNAs for co-operational loop formation. In this context, it is important to confirm whether CREPT directly binds eRNAs, since CREPT is not a DNA binding protein [56]. A third potential way for CREPT-mediated loop structure might be through chromatin modification. Notably, CREPT interacts with components of the SWI/SNF chromatin remodeling complex. These components including Smarca5 and Smarca4, as well as p300 and Kmt2c, were identified in our MS experiment. We previously reported that CREPT interacted with p300 to promote gene transcription [71]. In this study, we showed that p300-mediated H3K27ac required CREPT (see Figures S5E-S5F). Therefore, it is possible that CREPT interacts with p300 to induce chromatin modification during the establishment of loop structures. Consistently, several studies have shown that Smarca4/5 regulate oncogene expression by promoting chromatin accessibility in enhancer regions [93]. As we observed that depletion of CREPT reduced the monomethylation modification of histone H3K4, which is mediated by the SWI/SNF complex, as well as lysine methyltransferase Kmt2c, a subunit of the compass complex, we speculate that CREPT may affect the modification of H3K4me1. Therefore, CREPT may orchestrate chromatin modifications, with increased H3K4me1 and H3K27ac, that facilitate co-operational loop formation. Last but not least, given that CREPT has a strong interaction with RNAPII through its CID domain [55], which aggregates to enhancers to mediate the formation of E-P loops, we speculate that CREPT might mediate the loop configuration through its interaction with RNAPII. Recently, researchers reported that super-enhancers activated gene expression through classical loops heavily involving RNAPII [44]. Additionally, reports indicated that RNAPII used CTCF-mediated structures to regulate the enhancer-promoter interaction [45, 94], and could be transferred in the P-T loop for the re-initiation of transcription [95]. All these studies imply that CREPT might coordinate RNAPII during the loop formation. Studying whether CREPT-mediated co-operational loops require RNAPII and CTCF, as resolved in our HiChIP experiment, is of interest. Overall, these proposed mechanisms are not mutually exclusive and may operate concurrently in metastatic progression. We hypothesize that CREPT serves as a molecular hub, integrating multiple regulatory pathways to establish co-operational loops that drive metastatic gene expression programs in cancer cells.

The role of co-operational loop chromatin structures in gene transcription remains to be determined. While the importance of enhancer-promoter (E-P) loops in gene transcription is well-established [30, 74], our study reveals a more complex regulatory paradigm. We demonstrate that super-enhancers play a pivotal role in establishing E-P loops, which in turn facilitate drive P-T loop formation. Indeed, when we disrupted the super-enhancer using CRISPR-dCas9 sgRNAs, both E-P loops and P-T loops collapsed. These findings support a sequential model where CREPT first mediates E-P loop formation, followed by P-T loop assembly.

To our surprise, when we disrupted the super-enhancers, overexpression of CREPT induced neither E-P loops nor P-T loops. This observation suggests that the co-operational loop configuration is required for CREPT in the regulation of gene expression. This could also be attributed to the fact that activated enhancers and super-enhancers facilitate RNAPII recruitment to initiate gene transcription. Indeed, we previously demonstrated that the P-T loop structure enhanced the efficiency of RNAPII during the recycling of transcription since deletion of CREPT led to RNAPII reading-through the termination site [55]. Typically, the P-T loop structure is considered a way for the transcriptional memory and the re-initiation of transcription [95,96,97]. Whether the CREPT-mediated co-operational loop configuration provides the base for transcriptional memory is to be determined. Nevertheless, our data reveal an intriguing dependency relationship, where P-T loop integrity is contingent upon intact E-P loops. This suggests a temporal hierarchy in loop formation: E-P loops are established prior to transcription initiation, while P-T loops form subsequently, dependent on RNAPII loading. Based on these observations, we propose a model wherein E-P loops promote initial RNAPII recruitment to the promoter, while P-T loops facilitate RNAPII recycling during subsequent transcription cycles. This dual regulatory mechanism would significantly enhance transcriptional efficiency, potentially explaining the robust gene activation observed in CREPT-mediated co-operational loop structures.

To date, it remains a big challenge for TNBC therapy due to its high heterogeneity [1, 2, 7]. Our study provided a strategy using AAV to deplete CREPT for metastatic TNBC therapy. We have employed various mouse models, including implanted metastasis, in situ tumor-derived metastasis, and metastasis following surgical tumor resection. All the results consistently demonstrated that targeting CREPT by the AAV-based shRNA blocked the progression of metastatic TNBC. We also demonstrated that targeting CREPT-driven loops had a strong ability to block metastatic tumors. We expect that this AAV system, or an equivalent strategy against CREPT-mediated co-operational loops [31, 78], could be useful for metastatic TNBC therapy in the near future.

Limitations of the study

While our study provides compelling evidence for CREPT-mediated co-operational loop structures, several limitations warrant consideration. Firstly, the precise molecular mechanisms by which CREPT orchestrates co-operational loop formation remain incompletely characterized. Although we established CREPT's essential role in loop formation, the specific coordination between CREPT and key architectural factors such as CTCF, cohesin complex, and RNAPII requires further investigation [45, 94, 98] In this context, the hint is that CREPT is associated with various transcription regulation factors (see Figs. 4A and S4I). Indeed, our previous studies revealed that CREPT associated with STAT3 and β-catenin/TCF4 to promote transcription [59, 71]. Others reported that CREPT functioned, as its yeast homology Rtt103, to regulate the termination of transcription [57, 58, 66, 96]. Recently, a genome-wide screening identified that CREPT served as a novel factor at the polyadenylation site to release RNAPII from the 3’end of genes [67]. How CREPT, in coordination with transcriptional factors, chromatin regulators, cohesin complex, and RNAPII, facilitates mRNA synthesis needs more efforts to characterize.

Second, the molecular basis for the specific association between co-operational loop-regulated genes and metastatic processes remains unclear. While GO analysis denoted the genes in the metastasis, we have yet to determine whether these genes share common structural or sequence features at their promoters and enhancers where CREPT occupies.

Third, the translational potential of our findings awaits validation in human clinical trials. Although AAV-mediated CREPT targeting demonstrated significant anti-metastatic efficacy across multiple mouse models, its therapeutic application in TNBC patients requires rigorous clinical evaluation. Future studies should also explore potential synergistic effects between CREPT targeting and existing therapeutic modalities. Given the promising clinical potential of AAV-based gene therapy in oncology [99], we anticipate that forthcoming clinical trials will provide valuable insights into the therapeutic efficacy of CREPT inhibition in metastatic TNBC.

Future outlooks

The identification of driver genes with somatic mutations remains a critical challenge in developing targeted therapies for aggressive malignancies. Our study establishes CREPT as a key driver gene in TNBC, characterized by somatic gene copy number amplifications and regulation of metastasis-associated gene expression via co-operational loop structures. This discovery opens new avenues for investigating similar three-dimensional genome regulatory mechanisms in other aggressive cancer types, potentially revealing conserved principles of oncogene regulation across malignancies. Importantly, our development of AAV-mediated CREPT depletion and CRISPR-dCas9 targeting of CREPT-dependent co-operational loops represents a promising therapeutic strategy that may be adaptable to multiple cancer types, potentially offering a novel approach to targeting oncogenic transcriptional regulation across malignancies.

STAR + methods

Key resources table

Reagent or Resource	Identifier	Source
Antibodies
Rabbit polyclonal, H3K4me1	Cat # ab176877	Abcam
Rabbit polyclonal, H3K4me3	Cat # ab213224	Abcam
Rabbit polyclonal, H3K27ac	Cat # ab177178	Abcam
Rabbit polyclonal, RNAPII	Cat # ab76123	Abcam
Rabbit polyclonal, p300	Cat # ab14984	Abcam
Mouse monoclonal, CREPT	3E10	Chang et al
Mouse monoclonal, Actin	Cat # A5316	Sigma

Bacterial and virus strains.

DH5α	D1031S	TIANGEN

Chemicals, peptides, and recombinant proteins.

SUPERase In	AM2694	Ambion
RNaseZAP	AM9780	Ambion
GlycoBlue	AM9515	Ambion
AMPure XP beads	A63881	Beckman Coulter,
Micro Bio-Spin™ P-30 Gel Columns	732–6250	Bio-Rad
Matrigel	356237	Corning Costar
Transwell chambers of 8-μm pore	3422	Corning Costar
Corning Costar spin-x column	8160	Corning Costar
CircLigase™ II ssDNA Ligase	CL9021 K	Epicentre
25 bp DNA Ladder	10597–011	Invitrogen
DNase I	18068015	Invitrogen
TRIzol	15596018 CN	Invitrogen
10% TBE-urea gel	EC6875Box	Invitrogen
biotin-14-dATP	19524–016	Life Technologies
Trizol LS	10296–028	Life Technologies
DMEM-Ham's F12	CM16405	Macgene
Insulin from bovine pancreas	CC101	Macgene
Cholera toxin	CC104	Macgene
Hydrocortisone	CC103	Macgene
biotin-16-UTP	HY-D1686	MCE
Q5® High-Fidelity DNA Polymerase	M0491S	NEB
Phusion® High-Fidelity DNA Polymerase	M0530S	NEB
Antarctic phosphatase	M0289 s	NEB
T4 PNK	M0201 s	NEB
poly(A)-polymerase buffer	B0276 s	NEB
E. coli poly(A)-polymerase	2761	NEB
APE 1	M0282S	NEB
Phusion Hot-start	M0530L	NEB
MboI restriction enzyme	R0147	NEB
Large (Klenow) Fragment	M0210	NEB
T4 DNA ligase buffer	B0202	NEB
T4 DNA Ligase	M0202	NEB
proteinase K	P8102	NEB
T4 DNA ligase buffer	B0202	NEB
T4 PNK	M0201	NEB
T4 DNA polymerase I	M0203	NEB
NEB Klenow exo minus	M0212	NEB
1X Quick ligation reaction buffer	B6058	NEB
DNA Quick ligase	M2200	NEB
EGF	AF-100–15	Peprotech
20 × SSC buffer	B548109	Sangon Biotech
Formamide Deionized	A600211	Sangon Biotech
Saponin	A604521	Sangon Biotech
Triton X-100	93443	Sigma
Sulfate Dextran	D8906-5G	Sigma
protease inhibitors	P8340	Sigma
Formaldehyde	F8775	Sigma
Takara PrimeStar HS Premix	R040 A	Takara
Superscript III RT	18080085	Thermofisher
Exonuclease I	EN0581	Thermofisher
Dynabeads MyOne Streptavidin T1 beads	65602	Thermofisher
Cot-1 DNA	18440016	Thermofisher
Salmon Sperm DNA	15632011	Thermofisher
GTP	18332015	Thermofisher
ATP	18330019	Thermofisher
CTP	18331017	Thermofisher
BrUPT	B21551	Thermofisher
GlycoBlue™ Coprecipitant	AM9515	Thermofisher
RNaseZAP	AM9780	Thermofisher
Novex™ TBE-Urea Gels, 10%, 10 well	EC6875BOX	Thermofisher
Trypan Blue Solution	15250061	Thermofisher
SuperScript™ III Reverse Transcriptase	18080044	Thermofisher
Exonuclease I	EN0581	Thermofisher
FBS	10099141 C	Thermofisher
Trypsin EDTA (0.25%)	25200056	Thermofisher
Penicillin–Streptomycin	15140122	Thermofisher
PBS,7.4	C10010500BT	Thermofisher
Dynabeads™ M-280	11205D	Thermofisher
dCTP	10119ES74	Yeasen
dGTP	10121ES74	Yeasen
dTTP	10120ES74	Yeasen
dATP	10118ES74	Yeasen
Hieff NGS® DNA Selection Beads	12601ES08	Yeasen

Critical commercial assays.

Quantscript RT Kit	4992783	TIANGEN Biotech
Talent qPCR PreMix (SYBR Green) Kit	FP209-01	TIANGEN Biotech
PureLink™ Genomic DNA Mini Kit	K182001	Thermofisher
QIAquick PCR Purification Kit	28104	QIAGEN
TruSeq Stranded Total RNA Gold sequencing kit	20020598	Illumina
Zymo clean & concentrate kit	D4003	Zymo Research
TruePrep DNA Library Prep Kit V2 for Illumina	TD501	Vazyme
TruePrep Index Kit V2 for Illumina	TD202	Vazyme

Deposited data.

RNA-seq		This paper
GRO-seq		This paper
ChIP-seq		This paper
ATAC-seq		This paper
HiC		This paper
HiChIP		This paper
scRNA	GSE118390	Leif W Ellisen et al., 2018

Experimental models: Cell lines.

MCF10 A	RRID: CVCL_059	ATCC
MCF10 A	CTCC-001–0045	Meisen
MDA-MB-436	CTCC-ZHYC-0030	Meisen
MDA-MB-231	CTCC-001–0019	Meisen
LM2	N/A	This paper
NMUMG	CTCC-001–0327	Meisen
4T1	CTCC-001–0649	Meisen
4TO7	N/A	This paper
LM2-shCREPT	N/A	This paper
LM2-KO-CREPT	N/A	This paper
4T1-shCREPT	N/A	This paper
4T1-CREPT	N/A	This paper
4T07-Teton-Flag-CREPT	N/A	This paper

Recombinant DNA.

pSpCas9(BB)−2A-GFP PX458	48138	Addgene
Lenti-dCas9-KRAB-blast	52962	Addgene
Lenti-dCas9-KRAB-blast	89567	Addgene
Lenti-sgRNA(MS2)-puro	61427	Addgene

Software and algorithms.

Prism	https://www.graphpad.com/	GraphPad
bowtie 1.2.2	https://bowtie-bio.sourceforge.net/index.shtml	Langmead et al
STAR 2.7.3a	https://github.com/alexdobin/STAR	Dobin et al.1986
R 3.6.1	www.r-project.org https://www.r-project.org/	RCoreTeam
cutadapt	https://cutadapt.readthedocs.io/en/stable/	Martin
Samtools 1.9	http://www.htslib.org/	Li et al
HOMER 4.10.3	http://homer.ucsd.edu/homer/	Heinz et al
Deseq2	https://www.bioconductor.org/packages/release/bioc/html/DESeq2.html	Loveetal., 2014
Deeptools	https://deeptools.readthedocs.io/en/develop/	Ramirezetal.1975
MACS2	http://liulab.dfci.harvard.edu/MACS/index.html	Zhangetal.,2008
HiC-Pro 2.8.0	https://github.com/nservant/HiC-Pro	Servant et al
Juicebox v1.11.08	https://github.com/aidenlab/Juicebox	Durand et al
WashU Epigenome Browser	http://epigenomegateway.wustl.edu/	Zhouetal., 2013
UCSC Genome Browser	http://genome.ucsc.edu/cgi-bin/hgGateway	Kentetal., 2002

Data availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Zhijie Chang (zhijiec@tsinghua.edu.cn).

Materials availability

Unique and stable reagents generated in this study are available upon request.

Data and code availability

All PRO-seq, ChIP-seq, ATAC-seq, and RNA-seq data have been deposited at SRA database and are publicly available as of the date of publication. Accession numbers are listed in the key resource table (PRJNA1224039). HiC data have been deposited at GSA database and its bioproject number is PRJCA040004.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

sgRNA for CRISPR/Cas9 mediated knock-out.

CREPT-KO#1	Target + PAM	AAGAACGAAGTGTGTACGGCGG
CREPT-KO#2	Target + PAM	GGCAAGAACGAAGTGTGTACGG

sgRNA for CRISPR-dCas9 mediated silencing.

CRISPR-dCas9-Mouse-scramble		ACGGAGGCTAAGCGTCGCAA
CRISPR-dCas9-Mouse-Itga3#1	Target + PAM	TTATTGAGGATATGTGGCTTGG
CRISPR-dCas9-Mouse-Itga3#2	Target + PAM	AGCACCTTGGTGTACCGATGGG
CRISPR-dCas9-Mouse-Mmp13#1	Target + PAM	TACCATCCTGCGACTCTTGCGG
CRISPR-dCas9-Mouse-Mmp13#2	Target + PAM	TCTCGGAGCCTGTCAACTGTGG
CRISPR-dCas9-Mouse-Ccl2#1	Target + PAM	GCTCTCCAGCCTACTCATTGGG
CRISPR-dCas9-Mouse-Ccl2#2	Target + PAM	ATGATCCCAATGAGTAGGCTGG
CRISPR-dCas9-Mouse-Tgfa#1	Target + PAM	CATGGAACCTGCCGGTTTTTGG
CRISPR-dCas9-Mouse-Tgfa#2	Target + PAM	TCCTGCACCAAAAACCGGCAGG
CRISPR-dCas9-Mouse-Mmp12#1	Target + PAM	CATCCTCACGCTTCATGTCCGG
CRISPR-dCas9-Mouse-Mmp12#2	Target + PAM	ACTCCGGACATGAAGCGTGAGG
CRISPR-dCas9-Mouse-Serpine2#1	Target + PAM	AATAAAGACATTGTGACCGTGG
CRISPR-dCas9-Mouse-Serpine2#2	Target + PAM	GGAAGTGCCTTTTGCAGTAAGG
CRISPR-dCas9-Mouse-Itga2#1	Target + PAM	AGCAGCTTACGAACCCACAAGG
CRISPR-dCas9-Mouse-Itga2#2	Target + PAM	ATTGTCTGGCGTATAATGTTGG
CRISPR-dCas9-Mouse-Cxcl1#1	Target + PAM	AGCGCAGCTCATTGGCGATAGG
CRISPR-dCas9-Mouse-Cxcl1#2	Target + PAM	CACTGACAGCGCAGCTCATTGG
CRISPR-dCas9-Mouse-S100a4#1	Target + PAM	GATGGAGAGGCCTGTTACTCGG
CRISPR-dCas9-Mouse-S100a4#2	Target + PAM	AACTGTAATGACTCGAAACTGG
CRISPR-dCas9-Mouse-Vcam1#1	Target + PAM	GACAGCCCACTAAACGCGAAGG
CRISPR-dCas9-Mouse- Vcam1#2	Target + PAM	CCCACTAAACGCGAAGGTGAGG
CRISPR-dCas9-Human-Itga3#1	Target + PAM	GATGTTTCGCCTGTAGTTGGGG
CRISPR-dCas9-Human-Itga3#2	Target + PAM	TTATCTGAGTATAGTTACAAGG
CRISPR-dCas9-Human-Mmp13#1	Target + PAM	ACCATCCTACAAATCTCGCGGG
CRISPR-dCas9- Human -Mmp13#2	Target + PAM	TCAGAATGAGTCATATCAGGGG
CRISPR-dCas9- Human -Ccl2#1	Target + PAM	TTCAAGACCATTGTGGCCAAGG
CRISPR-dCas9- Human -Ccl2#2	Target + PAM	CCCAAGCAGAAGTGGGTTCAGG

TN5-Fish.

AL488-F-5`Alexa Fluor 488	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG	33
AL488-R-5`Alexa Fluor 488	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG	34
TAMRA565-F-5`TAMRA	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG	33
TAMRA565-R-5`TAMRA	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG	34
AL647-F-5`Alexa Fluor 647	TCGTCGGCAGCGTCAGATGTGTATAAGAGACAG	33
AL647-R-5`Alexa Fluor 647	GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAG	34

Primers for probes of Tn5-Fish.

Ccl2-Promoter	Forward primer	GAGGCCAGTCTGGTCTACA
Ccl2-Promoter	Reverse primer	GTGGGGCGTTAACTGCATC
Ccl2-Termition region	Forward primer	ATATAAGCTTGGAGAGCAACACAG
Ccl2-Termition region	Reverse primer	CAGGGACCCATTGCAATAAATC
Ccl2-Enhancer	Forward primer	AGAGAAGTTGAAGAAGGCCC
Ccl2-Enhancer	Reverse primer	GAGTCATGAGGTAAAGGTTATGTAGTG
Ccl2-Enhancer	Forward primer	AGATGCAGTGTGCCAGCTG
Ccl2-Enhancer	Reverse primer	TGCTCCCCATGTCTCACTG
Mmp13-Promoter	Forward primer	ATATATTTGAAAAAAAGCCCAATATGACC
Mmp13-Promoter	Reverse primer	GTTAAGTTCGTTTGGGACC
Mmp13-Termition region	Forward primer	ACAAAGAATTTTAATAAAGAATTCAAATG
Mmp13-Termition region	Reverse primer	ATGGAATGATCTCCTATGC
Mmp13-Enhancer	Forward primer	GCTTCCAGTTTGAGAGCAAAGT
Mmp13-Enhancer	Reverse primer	CCTTGCTCTTGCATAGGACT
Mmp13-Enhancer	Forward primer	AGTACTCCTAACCACTAGG
Mmp13-Enhancer	Reverse primer	TTTGCTGGCTTCCTCTCGAT
Mmp13-Enhancer	Forward primer	ATACACAAGCCTTAACTGCTATAGTTATC
Mmp13-Enhancer	Reverse primer	GCAGACATGCCCACAGGC
Ccl2-Promoter	Forward primer	AAGCAATAATAGAGCCCTGGAGGGACC
Ccl2-Promoter	Reverse primer	CTAGGGACCGGAGCCTGC
Ccl2-Termition region	Forward primer	ACTTGCTCAGTATGCATGAAGC
Ccl2-Termition region	Reverse primer	CTCTGCAGCCGCTTAGTCT
Ccl2-Enhancer	Forward primer	CTTTTGCTTCAGCTCCTC
Ccl2-Enhancer	Reverse primer	ACAAGGCTAACGGTGAAG
Ccl2-Enhancer	Forward primer	AAAGACGGGAAGGGAGCTC
Ccl2-Enhancer	Reverse primer	TGGTGTTACCTTCTGCAAGC
Ccl2-Enhancer	Forward primer	TGTGTGTTTTCCAGGAGATGGCATTG
Ccl2-Enhancer	Reverse primer	TCCACCCAGCCTGCCTTC

Primers for nuclear run-on.

Itgb3_F	Forward primer	GAGGAATGACGCATCCCATTTG
Itgb3_R	Reverse primer	TCTCACCATGGTAGTGGAGGCA
Itgb1_F	Forward primer	CGCTGATTGGCTGGAGGAAT
Itgb1_R	Reverse primer	TACTTACATAGTAATGGCTCAT
Itga2_F	Forward primer	GTGGATGTTGTAGTTGTATG
Itga2_R	Reverse primer	TCCTTACCTGTGTCTTTTTAGGT
Nectin4_F	Forward primer	GCTTCATGTGAACCCTGCTTA
Nectin4_R	Reverse primer	CACTCACCCAGAACACGAAGA
Vcam1_F	Forward primer	GAAATGCCACCCTCACCTT
Vcam1_R	Reverse primer	TACTCACCTTGAACAACTAA
Cxcl1_F	Forward primer	AGCCACACTCAAGAATGGT
Cxcl1_R	Reverse primer	CAACTCACTTTAGCATCTT
Cxcl12_F	Forward primer	GTAAACCAGTCAGCCTGAGCT
Cxcl12_R	Reverse primer	TAGGACTTACACAATCTGA
Ccl2_F	Forward primer	ATGGGCCCAATGCATCCACA
Ccl2_R	Reverse primer	TGGGAAGCTGTTATGTAAG
Cxcl13_F	Forward primer	TATTCTGGAAGCCCATTACA
Cxcl13_R	Reverse primer	ACTTACACAACTTCAGTT
Cxcr4_F	Forward primer	TTGCAGATATACACTTCTGATA
Cxcr4_R	Reverse primer	AAGATCCTATTGAAATGGACG
Macc1_F	Forward primer	ATGCTAATCAGTGAAAGA
Macc1_R	Reverse primer	TTGTACCTGTAATATTTT
Angptl4_F	Forward primer	ATAGACCTCTTGGCCCCCAC
Angptl4_R	Reverse primer	TACTCACTGTGTAAGTGGGTGGC
Vegfc_F	Forward primer	CTACTGGAAAATGTACAAGTG
Vegfc_R	Reverse primer	CTTACTTTTCAGGATCTCTGT
Tgf-a_F	Forward primer	ACTCACCCGTGGCGGCTGCAG
Tgf-a _R	Reverse primer	ACTTACACACATGCTGGCTTC
Myof_F	Forward primer	CCTGAGAAACGAGACCGTGACA
Myof_R	Reverse primer	ACATACTCTGGGGGATGTCCTC
Postn _F	Forward primer	ATGACACTCTGAGCATGGA
Postn _R	Reverse primer	TCACCTACCGTCCGATACACA
Mmp2_F	Forward primer	ACTCTCAGGACCCTGGAGCTC
Mmp2_R	Reverse primer	GCTTACCATAGAGCTCCTGGA
Serpine2 _F	Forward primer	TTGTGGAAGTCTCGGTTTCAA
Serpine2 _R	Reverse primer	ACTCACCTGAGCGGAACACA
Mmp12_F	Forward primer	GGACATGAAGCGTGAGGATGT
Mmp12_R	Reverse primer	TTGTATACCTCCAAATGCAAAA
Mmp13_F	Forward primer	CTAGGTCTGGATCACTCCAA
Mmp13_R	Reverse primer	TACCATAAAGAAACTGAATTCCT
Gapdh_F	Forward primer	ATGCTGTGCTCATCTGTGTTT
Gapdh_R	Reverse primer	GAAGCCCACCTGCATGGCTTT

Primers for RT-PCR.

Itgb3_F	Forward primer	GAGCCCATTTTCTTCTCCCG
Itgb3_R	Reverse primer	GCAACACCATGAATCCATCCC
Itgb1_F	Forward primer	AAACTCATCATAGGCATTGGAGG
Itgb1_R	Reverse primer	GTCCTGGGGCTTGAAGAAGT
Itga2_F	Forward primer	TTCTTGGGTCCTAGTGCTGTT
Itga2_R	Reverse primer	CGCTTCCATGTTTGTCCTTATGA
Nectin4_F	Forward primer	TGGCATCGTTTACAGGCCAAT
Nectin4_R	Reverse primer	CCACTGTCACTACGTCAGAGG
Vcam1_F	Forward primer	AGTTGGGGATTCGGTTGTTCT
Vcam1_R	Reverse primer	CCCCTCATTCCTTACCACCC
Cxcl1_F	Forward primer	CAAGGCTGGTCCATGCTCC
Cxcl1_R	Reverse primer	TGCTATCACTTCCTTTCTGTTGC
Cxcl12_F	Forward primer	TGCATCAGTGACGGTAAACCA
Cxcl12_R	Reverse primer	TTCTTCAGCCGTGCAACAATC
Ccl2_F	Forward primer	ATTCTGTGACCATCCCCTCAT
Ccl2_R	Reverse primer	TGTATGTGCCTCTGAACCCAC
Cxcl13_F	Forward primer	GGCCACGGTATTCTGGAAGC
Cxcl13_R	Reverse primer	GGGCGTAACTTGAATCCGATCTA
Cxcr4_F	Forward primer	GACTGGCATAGTCGGCAATG
Cxcr4_R	Reverse primer	AGAAGGGGAGTGTGATGACAAA
Macc1_F	Forward primer	AGGTACAGTAAGAGCCATAGGAC
Macc1_R	Reverse primer	CTTGGTTGTCAAAATGCCATCAG
Angptl4_F	Forward primer	CATCCTGGGACGAGATGAACT
Angptl4_R	Reverse primer	TGACAAGCGTTACCACAGGC
Vegfc_F	Forward primer	GAGGTCAAGGCTTTTGAAGGC
Vegfc_R	Reverse primer	CTGTCCTGGTATTGAGGGTGG
Tgf-a_F	Forward primer	CACTCTGGGTACGTGGGTG
Tgf-a _R	Reverse primer	CACAGGTGATAATGAGGACAGC
Myof_F	Forward primer	ACCGCTTTCGGTGTGATCC
Myof_R	Reverse primer	GCCAGTAATGGTTTGGTGTCTTC
Postn _F	Forward primer	CCTGCCCTTATATGCTCTGCT
Postn _R	Reverse primer	AAACATGGTCAATAGGCATCACT
Serpine2 _F	Forward primer	CACATGGGATCGCGTCCATC
Serpine2 _R	Reverse primer	CAGCACTTTACCAACTCCGTTTA
Mmp2_F	Forward primer	CTGCTGGCTAGTTGCACCC
Mmp2_R	Reverse primer	CCTTTGGGCTAGGTATCTCTGA
Mmp12_F	Forward primer	CTGCTCCCATGAATGACAGTG
Mmp12_R	Reverse primer	AGTTGCTTCTAGCCCAAAGAAC
Mmp13_F	Forward primer	CTTCTTCTTGTTGAGCTGGACTC
Mmp13_R	Reverse primer	CTGTGGAGGTCACTGTAGACT
CREPT_F	Forward primer	TGTCCCTTTGGCTCATCCAC
CREPT_R	Reverse primer	CATCTGCCTCTCTGGCAACA
Actin_F	Forward primer	CATGTACGTTGCTATCCAGGC
Actin_R	Reverse primer	CTCCTTAATGTCACGCACGAT

Experimental model and study participant details

Cell culture

All cells (4T1; 4TO7; LM2; NMUMG; MDA-MB-436; MDA-MB-231) were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum (FBS). MCF-10 A cells were cultured in 5% horse serum-DMEM/F12 medium with 10 µg/mL insulin, 0.1 µg/mL cholera toxin, 0.5 µg/mL hydrocortisone and 20 ng/mL epidermal growth factor (EGF). All cell lines were maintained at 37 °C with 5% CO₂.

Method details

In vitro invasion and migration assays

The assay was performed using transwell chambers with polycarbonate filters of 8 μm pore size. The upper side of a polycarbonate filter was coated with or without matrigel. The cancer cells (1 × 10⁶) were re-suspended in 300 μL 0.1% serum-free medium and cultured in the upper chamber. The lower chamber was filled with 10% FBS medium (200 μL). After a 24 h incubation, the filters were fixed with methanol for 10 min and stained with 0.5% crystal violet (0.02% in PBS) for 20 min. The cells on the upper surfaces of the filters were removed with a cotton swab. The cells on the reverse sides were counted in three random fields using a microscope at a 100X magnification. Each assay was performed in triplicate. We used Image J to count the cell numbers in the in vitro invasion and migration assays. We took the pictures and performed counting on ImageJ. In brief, the cell images of the lower surface of the transwell chamber were obtained through a microscope and saved in TIF format. We then analysis the TIF files in ImageJ by setting up the parameters to set the threshold for distinguishing cells and background. Adjust threshold: Use the slider to ensure that all cells are selected and the background. The data were transferred into Graphpad Prism for statistical analyses.

Tumor sphere formation assay

A total of 1 × 10³ cells to be used for tumor spheres were trypsinized to generate a single cell suspension and plated to a final volume (1 mL) of serum free medium (DMEM/F12 with 20 ng/mL bFGF, 20 ng/mL EGF and penicillin) in the lower attachment plate. Fresh medium was added once every two days. The tumor spheres were cultured for 6 days before counting.

Immunohistochemical (IHC) assay

IHC was performed with the indicated antibodies. The staining intensity of CREPT protein in IHC images was quantitatively analyzed using ImageJ software, with the following scoring criteria: 1 point for weak staining (light brown), 2 points for moderate staining, 3 points for strong staining (dark brown). The percentage of positively stained area relative to the total tissue area was calculated by assigning scores as: 1 point (1–25% positive cells), 2 points (26–50% positive cells), 3 points (51–75% positive cells), 4 points (76–100% positive cells). The expression level of CREPT was comprehensively evaluated using the formula H-Score = Σ (staining intensity grade × corresponding percentage of positive cells), followed by a T test analysis.

Real-time quantitative PCR (RT-qPCR) and data analysis

RNA extraction with TRIzol involves homogenizing the sample in TRIzol reagent to lyse cells, followed by phase separation using chloroform. After centrifugation, the RNA-containing aqueous layer is collected, mixed with isopropyl alcohol to precipitate RNA, and centrifuged to pellet RNA. The pellet is washed with ethanol, air-dried, and dissolved in RNase-free water. RNA concentration is measured using a NanoDrop by absorbance at 260 nm, with the A260/A280 ratio indicating purity. RNA was reversely transcribed using a Tiangen Quantscript RT Kit. Simply, a reaction mixture was prepared by combining 2 μg of total RNA, 10 μL of 2 × RT Mix (containing dNTPs, reaction buffer, and random primers), and 1 μL of RT Enzyme Mix (reverse transcriptase and RNase inhibitor) in an RNase-free tube. After gentle mixing and brief centrifugation, the reaction underwent reverse transcription with the conditions of 5 min at 25 °C for primer annealing, 30 min at 42 °C for cDNA synthesis, and 5 min at 85 °C to inactivate the enzyme. The resulting cDNA was cooled to 4 °C and stored at −20 °C. RT-qPCR was performed using the Talent qPCR (quantitative polymerase chain reaction) PreMix (SYBR Green) Kit. The Talent qPCR PreMix (SYBR Green) protocol involves preparing a 20 μL reaction mix containing 10 μL of 2 × SYBR Green Premix, 0.4 μL each of forward and reverse primers (10 μM), 1–2 μL of template cDNA, and RNase-free water. After brief centrifugation, reactions were loaded into BioRad qPCR instrument. PCR reactions were performed on the following conditions: pre-denaturation at 95 °C for 3 min; denaturation at 95 °C for 5 s, annealing and extension at 60 °C for 15 s, 40 cycles. The relative transcript level of each gene was normalized to the Actin gene, and calculated according to the 2^−ΔΔCt formula. The primers used were listed in KEY RESOURCES TABLE.

Nuclear run-on

The cell pellet (a total of 2 × 10⁷ cells) was resuspended in 4 mL cell lysis buffer (10 mM Tris–HCl, pH 7.4, 3 mM MgCl₂, 10 mM NaCl, 150 mM sucrose and 0.5% NP40) for 10 min. Nuclei were then collected by centrifugation (4 °C, 170 × g) and re-suspended in 100 µL freezing buffer (50 mM Tris–HCl, 40% glycerol, 5 mM MgCl₂ and 0.1 mM EDTA). A total of 50 ul nuclear solution were mixed with the same volume of transcription buffer (200 mM KCl, 20 mM Tris–HCl, pH8.0, 5 mM MgCl₂, 4 mM dithiothreitol (DTT), 4 mM each of ATP, GTP and CTP, 200 mM sucrose and 20% glycerol) and 8 µL of 10 mM biotin-16-UTP, and incubated at 29℃ in water bath for 30 min. Reaction was stopped by adding 6 µL 250 mM CaCl₂, 6 µL RNase-free DNase I and incubating for 10 min at 29 °C. RNA was extracted with Trizol, and mixed with the same amount of pre balanced magnetic beads for 4 h at 4℃. Subsequently, the M280 magnetic beads were washed twice with 2X SSC containing 15% formamide and 2X SSC, and finally re-suspended with about 30 μL of DEPC water for RT-qPCR analyses. The primers used were listed in KEY RESOURCES TABLE.

Western blots

Cell lysates were prepared by treating cells with lysis buffer (2 M Tris–Cl, 0.5 M EDTA, 5 M NaCl, 1% NP-40,1% SDS and protease inhibitors PMSF) and denatured by heating at 100 °C for 10 min. Proteins were separated on a 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) gel and transferred to a nitrocellulose membrane. After transfer, the membrane was blocked in 5% non-fat milk for 1 h at room temperature and incubated with indicated antibodies in blocking buffer overnight at 4 °C. Subsequently, the membrane was incubated with a secondary antibody for 1 h at room temperature. Western blotting results were visualized with MiniChemi610 Imaging System.

Immunoprecipitation

For endogenous protein–protein interaction assays, cell lysates (with 1 × 10⁷ cells) were prepared using lysis buffer (50 mM Tris–Cl, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate and 1% SDS) with protease inhibitors. The lysates were incubated with appropriate antibodies at 4 °C overnight, followed by addition of protein G-agarose beads for 4 h. Beads were washed four times with cell lysis buffer. Cell lysates and immunoprecipitants were analyzed by western blots using the indicated antibodies.

Sample preparation for tandem mass tag (TMT)

The quantitative proteomics analysis was performed according to a routine protocol [100]. For TMT-labeling proteomic analysis, wild-type 4T1 cell line (control) and CREPT knockout cell line (4T1-KO-CREPT) were extracted three times independently for trypsin digestion. Then, TMT 6-plex reagent (90,061; Thermo Scientific) was used to label. Two sets of control samples were tagged with TMTs 126.1 and 127.1 and two sets of 4T1-KO-CREPT samples were tagged with TMTs 129.1 and 130.1 for the next LC–MS/MS analysis. Protein identifications, quantifications, and database searches were performed by Proteome Discoverer (RRID: SCR_014477). The raw files from the fractions were searched against a Homo sapiens Swiss-Prot UniProt protein database. The protein ratio was computed as 129 + 130 + 131 over 126 + 127 + 128. Differentially expressed proteins were selected with protein ratio > 1.5 or < 0.67 above the 95% confidence level in the comparison. Channel 126 was used for labeling the internal reference sample.

ChIP-MS

ChIP experiments were perform as previously described [71]. Precipitates were eluted in 7.5% SDS, 200 mM DTT and incubated at 37 °C for 30 min to revert crosslinks. After washing, the beads were boiled for 35 min and ChIP-MS samples were run on 10% precast SDS–PAGE gels. Gel lanes were sliced, in-gel digested with trypsine to yield peptides and proteins, which were identified on an LQT-Orbitrap mass spectrometer.

Chromosome conformation capture (3C)

Cells (1 × 10⁷) were crosslinked in 1% formaldehyde for 15 min at room temperature. Formaldehyde was quenched by adding 2.5 M glycine to a final concentration of 0.125 M for 5 min with rotations. After washing in PBS, cells were re-suspended in lysis buffer (1 ml; 10 mM Tris–HCl pH8.0, 10 mM NaCl, 5 mM EDTA, 0.5% NP 40, protease inhibitor). After spinning down at 2000 rpm for 10 min at 4 °C, nuclei were washed with 1X CutSmart buffer (NEB), suspended in NEB buffer 2 plus 0.1% SDS, and lysis at 65 °C for 30 min, then quenched by 1% Triton X-100. HindIII-HF was added into each tube (700 U per tube) for overnight digestion at 37 °C with shaking. The next day after confirming a digestion efficiency over 80%, 50 µg digested DNA were taken out into a new tube and the volume adjusted to 600 µl with nuclease free water. After adding 80 µl 10% SDS, samples were incubated at 65 °C for 25 min. Heat inactivated chromatin was added into ligation mix (5 ml 10X ligation buffer, 2.687 ml 20% Triton X-100) and the volume adjusted to 50 ml with nuclease free water to a final concentration of 1 ng/µl, followed by 1 h incubation at 37 °C with rotation at 30 RPM. After adding 990 U T4 DNA ligase, samples were incubated at 16 °C overnight with rotation. The next day, 132 µl of proteinase K were added into each sample, followed by 65 °C incubation overnight. The next day, after adding 132 µl of RNase A, samples were incubated at 37 °C for 1 h, followed by phenol: chloroform DNA purification and subject to quantitative PCR analysis.

CRISPR/Cas9 mediated knock-out

For CREPT deletion cell lines, 4T1 and LM2 cells were transfected by PX458 carrying sgRNAs against CREPT and screened by green fluorescence through flow cytometry. Gene targeting by CRISPR/Cas9 was achieved by transfection of pSpCas9(BB)−2A-GFP PX458 (Addgene, 48138) with guide RNA sequences cloned into the Cas9 vector (see KEY RESOURCES TABLE). The clones were randomly picked up and identified by Western blotting.

CRISPR-dCas9 mediated silencing

For cell lines with co-operational-loop disruption, 4T1 cells were transfected Lenti-dCas9-KRAB-blast, Lenti-KRAB, and Lenti-sgRNA(MS2)-puro-barcode carrying sgRNAs against targeted genes of super-enhancer and screened by blasticidin, puromycin and green fluorescence through flow cytometry. Guide RNA sequences cloned into the Lenti-sgRNA(MS2)-puro-barcode are shown in KEY RESOURCES TABLE. The final clones were randomly picked up and identified by Western blotting.

AAV construction and virus package

AAV9-sh-CREPT was constructed in this lab. The pZDonor_Seq1-U6-shRNA-hEF1a-EGFP-2 A-Puro was used as the expression vector. The CREPT target shRNA sequence was GCACGAAGATTAGGTGCATTT, while negative control shRNA sequence was AAACGTGACACGTTCGGAGAACGAA. We used the AAV2/2 package system: the plasmid pRV1 containing the AAV2 Rep and Cap sequences, the plasmid pFdelta6 was Adenovirus-helper plasmid, and two AAV plasmid (AAV-shNC and AAV-shRNA) were constructed above containing the recombinant expression cassette flanked by AAV2 packaging signals (inverted terminal repeats, ITRs).

Animal studies

All procedures and experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Tsinghua University. For the metastasis assay, female Balb/c mice at week 6 were anaesthetized and injected with indicated tumor cells via tail vein. Tumor cells were suspended in cold PBS (100 µL) before injection. At the experimental endpoint, mice were euthanized and lung tissues were dissected out and fixed for metastatic nodule counting. Balb/c nude mice were used for LM2 cells. For primary tumor growth and spontaneous lung metastasis experiments, female Balb/c mice at week 6 were anaesthetized and injected orthotopically in fat pad. Primary tumor growth was measured and tumor volume was calculated as volume (mm³) = (width² x length)/2. At the experimental endpoint, mice were sacrificed and lung samples were sectioned and stained with hematoxylin and eosin (H&E), and lung metastasis nodules were counted. Survival analyses were performed using Cox proportional hazards regression models, and p-values were obtained from log-rank tests.

Bioluminescent imaging

We used cell lines with the expression of luciferase. Then, we implanted the cells into the mouse for tumorigenesis. We intraperitoneally injected fluorescein substrate (150 mg/kg) into the mouse and waited for 10 min to ensure even distribution. After anesthesia with isoflurane, the mouse was placed on the PerkinElmer IVIS Lumina III imaging platform, with right position to ensure obtaining a complete image. The image was obtained by setting the exposure time to 1 s. The image was saved and bioluminescence signals were recorded for statistical analysis.

RNA-seq library preparation and analysis

Total RNA was extracted from the cells using TRIzol reagent according to the manufacturer’s instructions. The paired-end RNA-seq sequencing library was sequenced with the Illumina HiSeq X-ten/NovaSeq 6000 sequencer (2 × 150 bp read length). The library construction and sequencing were performed by Shanghai Majorbio Bio-pharm Technology Co., Ltd, China. RNA-seq was carried out with two biological replicates.

The RNA-seq results were mapped to the mm10 genome sequence from UCSC using Bowtie2 with default parameters, and then the gene expression level was calculated for each sample with RSEM. DEseq2 algorithms were used to identify the differential expression genes. All transcriptomic heatmaps were generated using the ComplexHeatmap R package. Gene set enrichment analyses (GSEA) were performed with the GSEA platform of the Broad Institute using the pre-ranked lists of gene fold changes against the public gene sets of gene signature.

GRO-Seq library preparation and analysis

Briefly, 10 million nuclei per sample were used for GRO-seq. Purified run-on RNA was subjected to polyA tailing by Poly(A)-polymerase for 12 min at 37 °C. PolyA-tailed RNA was subjected to immunopurification by using anti-BrdU antibody-conjugated beads. Reverse transcription was then performed using SuperScript III Reverse Transcriptase with RT primers for 2 h at 48 °C. Extra RT primers were removed by Exonuclease I for 2 h at 37 °C. cDNAs were fragmented with basic hydrolysis and size-selected (130–500 nucleotides) in a 6–8% polyacrylamide TBE-urea gel. Purified cDNAs were circularized using CircLigase for 2 h at 60 °C and re-linearized with Ape 1 for 2 h at 37 °C. The re-linearized single-stranded DNA was used as template, subjected to PCR amplification by using barcoded primers for Illumina TrueSeq small RNA sample and Phusion High-Fidelity DNA Polymerase. Subsequently, PCR products were size-selected in 6% polyacrylamide TBE gel (175–400 bp) and purified. The final libraries were sequenced using Illumina NEXTSEQ 500 following the manufacturer’s instructions.

Low quality and adapter sequences of the raw FASTQ reads were trimmed using Trim_galore with default parameters. Reads were aligned to the mm10 annotation using STAR. Transcripts were called using “findPeaks” of the HOMER package underuse of the “–minReadDepth 200” parameter. Samtools and the HOMER package were used to make visualization tracks and TPM calculations. RNA expression values were calculated by gene accurately using the Homer package.

Chromatin immunoprecipitation sequencing (ChIP-seq) experiment.

Chromatin isolation and sonication

Cells (1 × 10⁷) were crosslinked with 1% formaldehyde for 10 min at 37 °C. Crosslinking was quenched by the addition of glycine to a final concentration of 0.125 M. Cells were collected and washed with ice-cold 1X PBS before being resuspended in Sonication Buffer at a concentration of 1 × 10⁷ cells per mL. The chromatin was sheared to a length of approximately 100–300 bp by sonicating the samples with 20 pulses using a QSonica sonicator.

Immunoprecipitation (IP)

The chromatin after sonication was subject to centrifugation for 10 min with12000 rpm at 4 °C. The chromatin on the supernatant was collected and diluted with dilution buffer (20 mM Tris pH 8.0, 2 mM EDTA, 0.5% Triton X-100, 150 mM NaCl, 10% Glycerol, 5% BSA) for clearing. Cleared samples were collected and combined with primary antibody before overnight incubation at 4 °C with rotation. For CREPT-ChIP samples, 30 ug CREPT antibody was used. For H3K4me1, H3K27ac, P300 and RNAPII treated samples, 10 ug of the respective antibody was used. Subsequently, 50 mL Protein A beads were added to each IP reaction, and samples were rotated for 3 h at 4 °C. Samples were washed once with Low-Salt Buffer (20 mM Tris pH 8.0, 2 mM EDTA, 1% Triton X-100, 150 mM NaCl, 0.1% SDS), three times with High-Salt Buffer (20 mM Tris pH 8.0, 2 mM EDTA, 1% Triton X-100, 500 mM NaCl, 0.1% SDS), once with Lithium Chloride Buffer (20 mM Tris pH 8.0, 2 mM EDTA, 250 mM LiCl, 1% IGEPAL CA-630, 1% sodium deoxycholate), and twice with TE Buffer (10 mM Tris–HCl, 1 mM EDTA). Each wash was performed by rotating samples for 3 min with 1 mL volume of ice-cold wash solution. Two elutions were performed by resuspending beads in Elution Buffer (100 mM NaHCO3, 1% SDS) and rotating for 15 min at room temperature. The combined eluate was supplemented with 200 mM NaCl and incubated overnight in a 65 °C water bath. Samples were treated with Proteinase K, extracted with phenol:chloroform:isoamyl alcohol, and resuspended in 20 μl ddH₂O.

ChIP-qPCR

For ChIP-quantitative PCR, the immunoprecipitated DNA was analyzed together with input DNA. To validate ChIP-seq results, qPCR was performed using experimental primers targeting a panel of candidate genes and enhancer regions. For each sample, Cq values of experimental primers were normalized to the average Cq value across all background primers for that same sample, allowing differences in accessibility between conditions to be expressed in terms of “normalized accessibility”. The background primers were as follows: Human GAPDH (Chr12: 6,534,543–6,534,720) Forward primer: 5-GAAGGTGAAGGTCGGAGT-3 and Reverse primer: 5-GAAGATGGTGATGGGATTTC-3; Mouse Actin (Chr5: 142,903,112–142,903,265) Forward primer: 5-GGCTGTATTCCCCTCCATCG-3 and Reverse primer: 5-CCAGTTGGTAACAATGCCATGT-3.

Library preparation

Libraries were prepared using the NEB Next Ultra II DNA Library Prep Kit for Illumina according to the manufacturer’s instructions. Libraries were sequenced on an Illumina NovaSeq using an S1 flow cell and a paired-end 100-bp cycle run with sequencing performed by The Bauer Core Facility in the Novogene Bioinformatics Institute, Beijing, China.

Data processing and mapping

Paired-end fastq files were trimmed using cutadapt 1.14. Reads were first aligned to the mm10 genome sequence from UCSC using Bowtie2 with default parameters. Bowtie2 SAM output files were converted to sorted BAM files using SAMtools. Multiple mapped reads were discarded. When multiple reads mapped to the same genomic loci, only one read was retained. Figures of selected genomic intervals were created in the IGV browser. To assign ChIP-seq enriched regions to genes, a complete set of Refseq genes was obtained from the UCSC Table Browser. The ChIP-seq density profiles were normalized by deeptools with parameters –normalizeUsingRPKM. Heatmaps and pileups were also generated by using deeptools.

ATAC-seq experiment

Cell preparation and transposition

A total of 10,000 cells were washed in cold PBS and re-suspended in lysis buffer. This single-cell suspension was incubated on ice for 5 min with gentle mixing by pipetting every 2 min. The lysate was centrifuged at 1,300 g for 5 min at 4 °C. Nuclei were re-suspended in 50 μl TD buffer, and then incubated with 2 μl Tn5 enzyme for 30 min at 37 °C.

Library preparation

Samples were immediately purified and PCR-amplified with the NEB next high-fidelity 2X PCR master Mixture. The amplified library was further purified by Qiagen minElute column and SPRI beads. Custom primers were used to incorporate Illumina adaptors and index sequences into sample fragments. Sequencing was performed on the Illumin HiSeq 2500 platform and 150 bp paired-end reads were generated in Novogene Bioinformatics Institute, Beijing, China.

ATAC-seq data processing and mapping

Paired-end fastq files were trimmed to remove adaptor sequences and low-quality reads using Cutadapt. All reads that failed to align to the spike genome were subsequently aligned to the M. musculus genome (mm10) using bowtie. The markdup tool was used to flag duplicate reads, which were then discarded. Fragments were filtered to retain unique reads between 10 and 150 bp, representing regions of accessible chromatin, which were then converted to bedGraph format. Reads mapped to mitochondrial or duplicated reads were removed by Samtools. Filtered bam files were merged for downstream analysis. MACS2 was used to call ATAC-seq peaks. The coverage tracks were generated using Bam2wig, and then bigwig files were visualized using the IGV pen-source genome browser.

In situ Hi-C and analysis

In situ Hi-C was performed as described previously [40]. For each template, roughly 10 million cells were permeabilized in 0.5% SDS at 37 °C, quenched with Triton X-100 and chromatin was digested with 100- 200U MboI at 37 °C overnight. Samples were treated with proteinase K at 55 °C for 30 min and cross-links were reversed at 68 °C overnight. DNA was ethanol precipitated and sheared on a Covaris LE220. DNA was cleaned up via AMPure XP beads (Beckman Coulter, A63881) and quantified by Qubit dsDNA High Sensitivity Assay. Samples were bound to Dynabeads MyOne Streptavidin T1 beads and washed. End repair, dATP attachment and adaptor ligation were performed. Final PCR amplification was performed using barcoded sequencing primers and PCR.

To analyze mouse Hi-C data, raw paired-end FASTQ reads were aligned to the reference genome (mm10) using the Juicer pipeline (juicer.sh) with BWA-MEM, and uniquely mapped non-duplicate reads were obtained. A Knight-Ruiz (KR)-normalized contact matrix (1 kb resolution) was generated and exported as a.hic file. For chromatin loop identification, the software HiCCUPS (java -jar juicer_tools.jar hiccups -r 1000 -k KR -f 0.1 input.hic loops) was run with the parameter selection at 1-kb resolution. For identification of differential loops between wildtype and CREPT knockout cells, the software diffHic was used to calculate the loop coordinates and interaction frequencies and negative binomial models were used to calculate adjusted p-values (FDR < 0.1). Two-sample Kolmogorov–Smirnov test (P < 0.05) was used to compare the genome-wide contact probability distributions between control and CREPT-knockout cells to confirm a significant shift toward lower interaction frequencies. The.hic file was converted to.cool format using the software hic2cool and genome compartment analysis was performed with the software cooltools. In brief, genome-wide eigenvectors calculated by setting the parameter as eigs-cis at 100-kb bins. For TAD boundary detection, the software cooltools was used by setting the parameter of “cooltools insulation” at a 40-kb window to identify boundaries. The softwares BEDTools and diffTAD were used to analyze differential TADs by comparing boundary strengths/positions between wildtype and CREPT-knockout cells. The final results of the differential TADs were statistically tested by Wilcoxon rank-sum test (p < 0.05). The related results were visualized for presentation with the software HiGlass or Juicebox.

HiChIP library preparation and data analysis

HiChIP was performed as described previously [101]. Briefly, crosslinked cells (5 × 10⁶) were lysed in Hi-C lysis buffer and permeabilized in 0.5% SDS at 63 °C for 10 min. Chromatin was digested with MboI for 2 h at 37 °C. Overhangs were filled in and marked with Biotin-dATP, and ends were ligated with T4 DNA Ligase for 4 h at room temperature. Nuclei were pelleted and lysed and chromatin was sheared on the Covaris E220 with the following conditions: Fill level 5, Duty Cycle 5, PIP 140, Cycles/burst 200, Time 4 min. Samples were clarified, diluted in ChIP dilution buffer and precleared with Protein G beads for 1 h at 4 °C. Samples were cleared on magnet and supernatant was incubated with appropriate antibodies at 4 °C overnight with rotation. Protein G beads were added and incubated for 2 h at 4 °C with rotation. Beads were washed with low salt, high salt and LiCl buffers. Sample was eluted in ChIP elution buffer, treated with Proteinase K and crosslinks were reversed. DNA was purified using the Zymo clean & concentrate kit. Streptavidin M280 beads were washed and resuspended in 2X biotin binding buffer and DNA was bound for 15 min at room temperature. Beads were washed and ChIP DNA was tagged and amplified with KAPA Hifi DNA polymerase for 8–12 PCR cycles. Libraries were amplified using the Nextera DNA Library Prep kit. The DNA libraries were size selected (300–600 bp) using Hieff NGS® DNA Selection Beads and quantified by Qubit. Libraries were sequenced on the HiSeq2500.

Raw fastq files were aligned using BWA mem with the −5SP options with an index containing only the main chromosome from the genome release mm10 (available from the UCSC genome). The aligned paired reads were annotated with Pairtools with the following options–min-mapq 40–walks-policy 5unique–max-inter-align-gap 30 and the–chroms-path file corresponding to the size of the chromosome used for the alignment index. The paired reads were further processed to remove duplicated reads, sorted with unaligned reads, removed with the pairtools sort and the pairtools dedup tools with the basic option to produce an alignment file in the bam format as well as the location of the valid pair. The valid pairs were finally converted to the cool format using the cooler cload and cooler zoomify tools and to the hic format using the juicer tool. For the generation of the aggregate peak analyses (APA) plots, we used the HiCExplorer tools and the hicAggregateContacts command with–range 50000:100000–number Of Bins 30. Plots for all chromosomes were individually computed and summated to generate the global APA plots. For the HiChIP contact heatmap, hic files were uploaded to the WashU Epigenome Browser, and screenshots from gene loci of interest were downloaded using the default viewing conditions.

Tn5-FISH experiment

Tn5 FISH probe preparation

Tn5-FISH probes were constructed as follows. The sequences of interacting sites were downloaded from UCSC genome browser with repeats masked as N. DNA fragments used for probe library generation were amplified by PCR and recovered by DNA Cleanup kit. The amplified probe library served as templates for Tn5-FISH probes. The Tn5-FISH probes were obtained by a second PCR amplification with fluorescence-tagged primers. Probe library was constructed with 50 ng DNA fragments treated by V50 Tn5 enzyme provided in Vazyme TruePrep DNA Library Prep kit at 55 °C for 10 min, recovered and immediately amplified by Q5 enzyme with amplification primers as instructed by DNA Library Prep kit. The amplified probe library could serve as templates for Tn5-FISH probes. The Tn5-FISH probes were obtained by a second PCR amplification with fluorescence-tagged primers. The probes were designed to the regions in the enhancers, promoters and termination sites according the genome structures of each gene. Probes for the enhancers were labeled with TAMRA (red), promoters with Alexa Fluor 488 (green), and termination regions with Alexa Fluor 647 (blue). After recovered by DNA Cleanup kit, Salmon sperm DNA was added into the Tn5-FISH probes (50 μg Salmon sperm DNA per 1 μg Tn5-FISH probes), ethanol precipitated, and dissolved in DNA FISH buffer (50% deionized formamide, 10% Dextran Sulfate, 2X SSC, all from Sigma, Germany) at a concentration of 20 ng Tn5-FISH probes per microliter.

Fish processing

Briefly, cells were fixed by 4% paraformaldehyde and permeated by 0.1% Triton-X 100, then incubated with 20% glycerol for 30 min. After quick freeze–thaw by liquid nitrogen for 3 times, cells were further permeated with 0.5% Triton-X 100 and treated with 0.1 M HCl. Cells were incubated with pre-hybridization buffer (50% deionized formamide, 2X SSC) after thorough washing with PBS. A total of 10 ng Tn5-FISH probes for each color were mixed with DNA FISH buffer and applied to cells. The FISH hybridization program was set to 75 °C for 5 min, then 37 °C overnight. Cells were washed with FISH-washing buffer (0.2% CA-630, 2X SSC) for 3 time and then counterstained with 2 ng/mL DAPI. The slides were sealed with Mowiol mounting medium and imaged by Olympus FV3000. The images were obtained by sequential illumination at 405 nm, 488 nm, 594 nm and 647 nm lasers. For the details of all primers used in KEY RESOURCES TABLE.

Single cell analysis

The single-cell RNA analysis was performed using a published code in R (version 3.4.3). We used the dataset and the code source (https://github.com/Michorlab/tnbc_scrnaseq) cited from a published literature [14]. In brief, FASTQ files were converted to TPM values using RSEM. Low-quality cells (with small library size, few expressed genes, or low total mRNA) and genes not expressed (log2(TPM + 1) < 0.1 in ≥ 95% cells across all patients) were removed. A total of 13280 genes remained after filtering. To normalize single-cell RNA-seq data, a three step strategy was employed: (1) to transform the TPM values into relative counts with the Census algorithm; (2) to normalize the Census counts with the deconvolution strategy implemented in the R package scran57; (3) to remove additional sources of unwanted variation in the scran-normalized Census counts with RUVSeq. After normalization, 1189 cells remained for downstream analyses. A two-step combination approach to identify the different cell types (1) Markers: (Epithelial cell: EPCAM/EGFR/CDH1; Stroma cell: FAP,COL1 A1/COL3 A1/COL5 A1/ACTA2/TAGLN/LUM/FBLN1/COL6 A3/COL1 A2/COL6 A1/COL6 A2; endothelial cell: PECAM1/VWF/CDH5/SELE; T cell: CD2/CD3D/CD3E/CD3G/CD8 A/CD8B; B cell: MS4 A1/CD79 A/CD79B/BLNK CE; macrophage: CD14/CD68/CD163/CSF1R). (2) A t-SNE-based clustering using Monocle on a projection of the cells into a lower dimensional space was performed. The number of clusters was automatically chosen by Monocle. By running the code program, we obtained tSNE and generated our results.

Copy-number analysis

Whole-genome sequencing BAM files were download from The Cancer Genome Atlas (TCGA). Normalization against baseline coverage from diploid normal samples was performed using LOESS regression to correct for GC content bias and technical noise, yielding log2 copy-number ratios for tumor samples. These ratios are segmented into genomic intervals via DNAcopy algorithms (thresholds: log2 ratio > 0.3 for amplifications, < −0.3 for deletions) to identify significant CNV events. Allele-specific copy-number analysis was conducted by integrating heterozygous germline SNPs to detect allelic imbalances and loss of heterozygosity. The ABSOLUTE software was employed to estimate tumor purity and ploidy, enabling conversion of relative ratios into absolute copy numbers. Recurrent CNV events across the cohort were identified using GISTIC2, which statistically evaluates recurrence peaks (q-value < 0.25). Finally, copy-number profiles were visualized through Xena platform, and integrated with transcriptomic (RNA-seq) and clinical data to assess functional and prognostic implications of CNVs in cancer biology. For CREPT copy number analysis, specific coverage counts (exons padded by 250 bp) were extracted and the copy-numbers were calculated by GISTIC2. GISTIC score was used to present the density of copy numbers.

TCGA-RNAseq analysis

RNA seq count data were downloaded from TCGA and filtered for low expression genes (count > 10 in at least half of the samples), followed by normalization using DESeq2 to correct sequencing depth and gene length bias. By constructing a design matrix (tumor and normal group) to fit a generalized linear model (GLM), Wald test or likelihood ratio test (LRT) was used to calculate the significance of gene differential expression, and the Benjamin Hochberg method was applied to correct multiple hypothesis tests to obtain FDR values. The screening threshold is often set to |log2 fold change (FC)|> 1 and FDR < 0.05 for differentially expressed genes. The results are visualized through volcano plots.

Data availability

Sequence data that support the findings of this study is prepared to upload toSRA database. Accession numbers are listed in the key resource table (PRJNA1224039). HiC data have been deposited at GSA database and its bioproject number is PRJCA040004.

References

Nolan E, Lindeman GJ, Visvader JE. Deciphering breast cancer: from biology to the clinic. Cell. 2023;186:1708–28. https://doi.org/10.1016/j.cell.2023.01.040.
Article PubMed CAS Google Scholar
Popovic, L. S. et al. Targeting triple-negative breast cancer: A clinical perspective. Oncol Res. 31, 221–238 (2023). https://doi.org/10.32604/or.2023.028525.
Jin J, et al. Incidence, pattern and prognosis of brain metastases in patients with metastatic triple negative breast cancer. BMC Cancer. 2018;18:446. https://doi.org/10.1186/s12885-018-4371-0.
Article PubMed PubMed Central Google Scholar
Kennecke H, et al. Metastatic behavior of breast cancer subtypes. J Clin Oncol. 2010;28:3271–7. https://doi.org/10.1200/JCO.2009.25.9820.
Article PubMed Google Scholar
Valastyan S, Weinberg RA. Tumor metastasis: molecular insights and evolving paradigms. Cell. 2011;147:275–92. https://doi.org/10.1016/j.cell.2011.09.024.
Article PubMed PubMed Central CAS Google Scholar
Vogelstein B, et al. Cancer genome landscapes. Science. 2013;339:1546–58. https://doi.org/10.1126/science.1235122.
Article PubMed PubMed Central CAS Google Scholar
Weiss F, Lauffenburger D, Friedl P. Towards targeting of shared mechanisms of cancer metastasis and therapy resistance. Nat Rev Cancer. 2022;22:157–73. https://doi.org/10.1038/s41568-021-00427-0.
Article PubMed PubMed Central CAS Google Scholar
Duan, H. et al. Integrated analyses of multi-omic data derived from paired primary lung cancer and brain metastasis reveal the metabolic vulnerability as a novel therapeutic target. Genome Med. 16 (2024);138 https://doi.org/10.1186/s13073-024-01410-8
Conley-LaComb, M. K. et al. PTEN loss mediated Akt activation promotes prostate tumor growth and metastasis via CXCL12/CXCR4 signaling. Mol Cancer. 12 (2013)85 https://doi.org/10.1186/1476-4598-12-85
Boiarsky D, et al. Molecular markers of metastatic disease in KRAS-mutant lung adenocarcinoma. Ann Oncol. 2023;34:589–604. https://doi.org/10.1016/j.annonc.2023.04.514.
Article PubMed CAS Google Scholar
Jiang, Y. Z. et al. Genomic and Transcriptomic Landscape of Triple-Negative Breast Cancers: Subtypes and Treatment Strategies. Cancer Cell. 35, 428-+ (2019). https://doi.org/10.1016/j.ccell.2019.02.001.
Kumar, A. et al. Substantial interindividual and limited intraindividual genomic diversity among tumors from men with metastatic prostate cancer. Nat Med. 22, 369-+ (2016). https://doi.org/10.1038/nm.4053.
Sweet-Cordero, E. A. & Biegel, J. A. The genomic landscape of pediatric cancers: Implications for diagnosis and treatment. Science. 363, 1170-+ (2019). https://doi.org/10.1126/science.aaw3535.
Karaayvaz M, et al. Unravelling subclonal heterogeneity and aggressive disease states in TNBC through single-cell RNA-seq. Nat Commun. 2018;9:3588. https://doi.org/10.1038/s41467-018-06052-0.
Article PubMed PubMed Central CAS Google Scholar
Pariyar M, Johns A, Thorne RF, Scott RJ, Avery-Kiejda KA. Copy number variation in triple negative breast cancer samples associated with lymph node metastasis. Neoplasia. 2021;23:743–53. https://doi.org/10.1016/j.neo.2021.05.016.
Article PubMed PubMed Central CAS Google Scholar
Bao L, et al. Coexisting genomic aberrations associated with lymph node metastasis in breast cancer. J Clin Invest. 2018;128:2310–24. https://doi.org/10.1172/jci97449.
Article PubMed PubMed Central Google Scholar
Ronowicz A, et al. Concurrent DNA Copy-Number Alterations and Mutations in Genes Related to Maintenance of Genome Stability in Uninvolved Mammary Glandular Tissue from Breast Cancer Patients. Hum Mutat. 2015;36:1088–99. https://doi.org/10.1002/humu.22845.
Article PubMed CAS Google Scholar
Nik-Zainal S, et al. Landscape of somatic mutations in 560 breast cancer whole-genome sequences. Nature. 2016;534:47–54. https://doi.org/10.1038/nature17676.
Article PubMed PubMed Central CAS Google Scholar
Yu CC, et al. Mutant allele specific imbalance in oncogenes with copy number alterations: Occurrence, mechanisms, and potential clinical implications. Cancer Lett. 2017;384:86–93. https://doi.org/10.1016/j.canlet.2016.10.013.
Article PubMed CAS Google Scholar
Gerstung M, et al. The evolutionary history of 2,658 cancers. Nature. 2020;578:122–8. https://doi.org/10.1038/s41586-019-1907-7.
Article PubMed PubMed Central CAS Google Scholar
Priestley P, et al. Pan-cancer whole-genome analyses of metastatic solid tumours. Nature. 2019;575:210–6. https://doi.org/10.1038/s41586-019-1689-y.
Article PubMed PubMed Central CAS Google Scholar
Shih DJH, et al. Genomic characterization of human brain metastases identifies drivers of metastatic lung adenocarcinoma. Nat Genet. 2020;52:371–7. https://doi.org/10.1038/s41588-020-0592-7.
Article PubMed PubMed Central CAS Google Scholar
Shlyueva D, Stampfel G, Stark A. Transcriptional enhancers: from properties to genome-wide predictions. Nat Rev Genet. 2014;15:272–86. https://doi.org/10.1038/nrg3682.
Article PubMed CAS Google Scholar
Debek S, Juszczynski P. Super enhancers as master gene regulators in the pathogenesis of hematologic malignancies. Biochim Biophys Acta Rev Cancer. 2022;1877: 188697. https://doi.org/10.1016/j.bbcan.2022.188697.
Article PubMed CAS Google Scholar
Schoenfelder S, Fraser P. Long-range enhancer-promoter contacts in gene expression control. Nat Rev Genet. 2019;20:437–55. https://doi.org/10.1038/s41576-019-0128-0.
Article PubMed CAS Google Scholar
Calo E, Wysocka J. Modification of enhancer chromatin: what, how, and why? Mol Cell. 2013;49:825–37. https://doi.org/10.1016/j.molcel.2013.01.038.
Article PubMed CAS Google Scholar
Andersson R, Sandelin A. Determinants of enhancer and promoter activities of regulatory elements. Nat Rev Genet. 2020;21:71–87. https://doi.org/10.1038/s41576-019-0173-8.
Article PubMed CAS Google Scholar
Buecker C, et al. Reorganization of enhancer patterns in transition from naive to primed pluripotency. Cell Stem Cell. 2014;14:838–53. https://doi.org/10.1016/j.stem.2014.04.003.
Article PubMed PubMed Central CAS Google Scholar
Lee JE, et al. H3K4 mono- and di-methyltransferase MLL4 is required for enhancer activation during cell differentiation. Elife. 2013;2: e01503. https://doi.org/10.7554/eLife.01503.
Article PubMed PubMed Central Google Scholar
Whyte WA, et al. Master transcription factors and mediator establish super-enhancers at key cell identity genes. Cell. 2013;153:307–19. https://doi.org/10.1016/j.cell.2013.03.035.
Article PubMed PubMed Central CAS Google Scholar
Vasileva, A. V. et al. Super-Enhancers and Their Parts: From Prediction Efforts to Pathognomonic Status. Int J Mol Sci, 25 (2024). https://doi.org/10.3390/ijms25063103.
Sengupta S, George RE. Super-Enhancer-Driven Transcriptional Dependencies in Cancer. Trends Cancer. 2017;3:269–81. https://doi.org/10.1016/j.trecan.2017.03.006.
Article PubMed PubMed Central CAS Google Scholar
Tang F, Yang Z, Tan Y, Li Y. Super-enhancer function and its application in cancer targeted therapy. NPJ Precis Oncol. 2020;4:2. https://doi.org/10.1038/s41698-020-0108-z.
Article PubMed PubMed Central Google Scholar
Jia Q, Chen S, Tan Y, Li Y, Tang F. Oncogenic super-enhancer formation in tumorigenesis and its molecular mechanisms. Exp Mol Med. 2020;52:713–23. https://doi.org/10.1038/s12276-020-0428-7.
Article PubMed PubMed Central CAS Google Scholar
Oudelaar AM, Higgs DR. The relationship between genome structure and function. Nat Rev Genet. 2021;22:154–68. https://doi.org/10.1038/s41576-020-00303-x.
Article PubMed CAS Google Scholar
Razin SV, Ulianov SV. Gene functioning and storage within a folded genome. Cell Mol Biol Lett. 2017;22:18. https://doi.org/10.1186/s11658-017-0050-4.
Article PubMed PubMed Central CAS Google Scholar
Rowley, M. J. et al. Evolutionarily Conserved Principles Predict 3D Chromatin Organization. Mol Cell. 67, 837–852 e837 (2017). https://doi.org/10.1016/j.molcel.2017.07.022.
Quinodoz, S. A. et al. Higher-Order Inter-chromosomal Hubs Shape 3D Genome Organization in the Nucleus. Cell. 174, 744–757 e724 (2018). https://doi.org/10.1016/j.cell.2018.05.024.
Symmons O, et al. Functional and topological characteristics of mammalian regulatory domains. Genome Res. 2014;24:390–400. https://doi.org/10.1101/gr.163519.113.
Article PubMed PubMed Central CAS Google Scholar
Rao SS, et al. A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell. 2014;159:1665–80. https://doi.org/10.1016/j.cell.2014.11.021.
Article PubMed PubMed Central CAS Google Scholar
Li Q, Barkess G, Qian H. Chromatin looping and the probability of transcription. Trends Genet. 2006;22:197–202. https://doi.org/10.1016/j.tig.2006.02.004.
Article PubMed CAS Google Scholar
Hansen AS, et al. Distinct Classes of Chromatin Loops Revealed by Deletion of an RNA-Binding Region in CTCF. Mol Cell. 2019;76:395-411.e313. https://doi.org/10.1016/j.molcel.2019.07.039.
Article PubMed PubMed Central CAS Google Scholar
Robson MI, Ringel AR, Mundlos S. Regulatory Landscaping: How Enhancer-Promoter Communication Is Sculpted in 3D. Mol Cell. 2019;74:1110–22. https://doi.org/10.1016/j.molcel.2019.05.032.
Article PubMed CAS Google Scholar
Blayney, J. W. et al. Super-enhancers include classical enhancers and facilitators to fully activate gene expression. Cell. 186, 5826–5839 e5818 (2023). https://doi.org/10.1016/j.cell.2023.11.030.
Zhang S, Ubelmesser N, Barbieri M, Papantonis A. Enhancer-promoter contact formation requires RNAPII and antagonizes loop extrusion. Nat Genet. 2023;55:832–40. https://doi.org/10.1038/s41588-023-01364-4.
Article PubMed CAS Google Scholar
Perlman BS, et al. Enhancer-promoter hubs organize transcriptional networks promoting oncogenesis and drug resistance. Nat Commun. 2024;15:8070. https://doi.org/10.1038/s41467-024-52375-6.
Article PubMed PubMed Central CAS Google Scholar
Flavahan WA, et al. Altered chromosomal topology drives oncogenic programs in SDH-deficient GISTs. Nature. 2019;575:229–33. https://doi.org/10.1038/s41586-019-1668-3.
Article PubMed PubMed Central CAS Google Scholar
Hnisz D, et al. Activation of proto-oncogenes by disruption of chromosome neighborhoods. Science. 2016;351:1454–8. https://doi.org/10.1126/science.aad9024.
Article PubMed PubMed Central CAS Google Scholar
Flavahan WA, et al. Insulator dysfunction and oncogene activation in IDH mutant gliomas. Nature. 2016;529:110–4. https://doi.org/10.1038/nature16490.
Article PubMed CAS Google Scholar
Llinas-Arias, P. et al. Chromatin insulation orchestrates matrix metalloproteinase gene cluster expression reprogramming in aggressive breast cancer tumors. Mol Cancer. 22 (2023). 190. https://doi.org/10.1186/s12943-023-01906-8
Sanalkumar, R. et al. Highly connected 3D chromatin networks established by an oncogenic fusion protein shape tumor cell identity. Sci Adv. 9 (2023). eabo3789. https://doi.org/10.1126/sciadv.abo3789
Yusufova, N. et al. Histone H1 loss drives lymphoma by disrupting 3D chromatin architecture. Nature. 589, 299-+ (2021). https://doi.org/10.1038/s41586-020-3017-y.
Lu, B. et al. Phase separation of phospho-HDAC6 drives aberrant chromatin architecture in triple-negative breast cancer. Nature Cancer. 5 (2024).https://doi.org/10.1038/s43018-024-00816-y.
Huang, J. Z. et al. SLFN5-mediated chromatin dynamics sculpt higher-order DNA repair topology. Molecular Cell. 83, 1043-+ (2023). https://doi.org/10.1016/j.molcel.2023.02.004.
Lu D, et al. CREPT accelerates tumorigenesis by regulating the transcription of cell-cycle-related genes. Cancer Cell. 2012;21:92–104. https://doi.org/10.1016/j.ccr.2011.12.016.
Article PubMed CAS Google Scholar
Srividya I, Tirupataiah S, Mishra K. Yeast transcription termination factor Rtt103 functions in DNA damage response. PLoS ONE. 2012;7: e31288. https://doi.org/10.1371/journal.pone.0031288.
Article PubMed PubMed Central CAS Google Scholar
Li M, Ma D, Chang Z. Current understanding of CREPT and p15RS, carboxy-terminal domain (CTD)-interacting proteins, in human cancers. Oncogene. 2021;40:705–16. https://doi.org/10.1038/s41388-020-01544-0.
Article PubMed CAS Google Scholar
Ni Z, et al. RPRD1A and RPRD1B are human RNA polymerase II C-terminal domain scaffolds for Ser5 dephosphorylation. Nat Struct Mol Biol. 2014;21:686–95. https://doi.org/10.1038/nsmb.2853.
Article PubMed PubMed Central CAS Google Scholar
Zhai W, et al. CREPT/RPRD1B promotes tumorigenesis through STAT3-driven gene transcription in a p300-dependent manner. Br J Cancer. 2021;124:1437–48. https://doi.org/10.1038/s41416-021-01269-1.
Article PubMed PubMed Central CAS Google Scholar
Lin Y, et al. Microenvironment-induced CREPT expression by cancer-derived small extracellular vesicles primes field cancerization. Theranostics. 2024;14:662–80. https://doi.org/10.7150/thno.87344.
Article PubMed PubMed Central CAS Google Scholar
Ding L, et al. CREPT/RPRD1B associates with Aurora B to regulate Cyclin B1 expression for accelerating the G2/M transition in gastric cancer. Cell Death Dis. 2018;9:1172. https://doi.org/10.1038/s41419-018-1211-8.
Article PubMed PubMed Central CAS Google Scholar
Yang L, et al. CREPT is required for murine stem cell maintenance during intestinal regeneration. Nat Commun. 2021;12:270. https://doi.org/10.1038/s41467-020-20636-9.
Article PubMed PubMed Central CAS Google Scholar
Minn AJ, et al. Genes that mediate breast cancer metastasis to lung. Nature. 2005;436:518–24. https://doi.org/10.1038/nature03799.
Article PubMed PubMed Central CAS Google Scholar
Ren F, et al. Characterization of a monoclonal antibody against CREPT, a novel protein highly expressed in tumors. Monoclon Antib Immunodiagn Immunother. 2014;33:401–8. https://doi.org/10.1089/mab.2014.0043.
Article PubMed PubMed Central CAS Google Scholar
Gregersen, L. H. et al. SCAF4 and SCAF8, mRNA Anti-Terminator Proteins. Cell. 177, 1797–1813 e1718 (2019). https://doi.org/10.1016/j.cell.2019.04.038.
Morales JC, et al. Kub5-Hera, the human Rtt103 homolog, plays dual functional roles in transcription termination and DNA repair. Nucleic Acids Res. 2014;42:4996–5006. https://doi.org/10.1093/nar/gku160.
Article PubMed PubMed Central CAS Google Scholar
Ni Z, et al. Identifying human pre-mRNA cleavage and polyadenylation factors by genome-wide CRISPR screens using a dual fluorescence readthrough reporter. Nucleic Acids Res. 2024;52:4483–501. https://doi.org/10.1093/nar/gkae240.
Article PubMed PubMed Central CAS Google Scholar
Zheng R, et al. Cistrome Data Browser: expanded datasets and new tools for gene regulatory analysis. Nucleic Acids Res. 2019;47:D729–35. https://doi.org/10.1093/nar/gky1094.
Article PubMed CAS Google Scholar
Narita, T. et al. Enhancers are activated by p300/CBP activity-dependent PIC assembly, RNAPII recruitment, and pause release. Mol Cell. 81, 2166–2182 e2166 (2021). https://doi.org/10.1016/j.molcel.2021.03.008.
Wang F, Marshall CB, Ikura M. Transcriptional/epigenetic regulator CBP/p300 in tumorigenesis: structural and functional versatility in target recognition. Cell Mol Life Sci. 2013;70:3989–4008. https://doi.org/10.1007/s00018-012-1254-4.
Article PubMed PubMed Central CAS Google Scholar
Zhang Y, et al. CREPT facilitates colorectal cancer growth through inducing Wnt/beta-catenin pathway by enhancing p300-mediated beta-catenin acetylation. Oncogene. 2018;37:3485–500. https://doi.org/10.1038/s41388-018-0161-z.
Article PubMed PubMed Central CAS Google Scholar
Dixon JR, et al. Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature. 2012;485:376–80. https://doi.org/10.1038/nature11082.
Article PubMed PubMed Central CAS Google Scholar
Crane E, et al. Condensin-driven remodelling of X chromosome topology during dosage compensation. Nature. 2015;523:240–4. https://doi.org/10.1038/nature14450.
Article PubMed PubMed Central CAS Google Scholar
Phanstiel, D. H. et al. Static and Dynamic DNA Loops form AP-1-Bound Activation Hubs during Macrophage Development. Mol Cell. 67, 1037–1048 e1036 (2017). https://doi.org/10.1016/j.molcel.2017.08.006.
Vanharanta S, Massague J. Origins of metastatic traits. Cancer Cell. 2013;24:410–21. https://doi.org/10.1016/j.ccr.2013.09.007.
Article PubMed PubMed Central CAS Google Scholar
Niu J, et al. A novel cytogenetic method to image chromatin interactions at subkilobase resolution: Tn5 transposase-based fluorescence in situ hybridization. J Genet Genomics. 2020;47:727–35. https://doi.org/10.1016/j.jgg.2020.04.008.
Article PubMed Google Scholar
Klann TS, et al. CRISPR-Cas9 epigenome editing enables high-throughput screening for functional regulatory elements in the human genome. Nat Biotechnol. 2017;35:561–8. https://doi.org/10.1038/nbt.3853.
Article PubMed PubMed Central CAS Google Scholar
Zhu, T. et al. Integrated enhancer regulatory network by enhancer-promoter looping in gastric cancer. NAR Cancer. 6, zcae020 (2024). https://doi.org/10.1093/narcan/zcae020.
Xie M, et al. FGF19/FGFR4-mediated elevation of ETV4 facilitates hepatocellular carcinoma metastasis by upregulating PD-L1 and CCL2. J Hepatol. 2023;79:109–25. https://doi.org/10.1016/j.jhep.2023.02.036.
Article PubMed CAS Google Scholar
Liu, M. et al. Zinc-Dependent Regulation of ZEB1 and YAP1 Coactivation Promotes Epithelial-Mesenchymal Transition Plasticity and Metastasis in Pancreatic Cancer. Gastroenterology. 160, 1771–1783 e1771 (2021). https://doi.org/10.1053/j.gastro.2020.12.077.
Jiang, Y. Z. et al. Genomic and Transcriptomic Landscape of Triple-Negative Breast Cancers: Subtypes and Treatment Strategies. Cancer Cell. 35, 428–440 e425 (2019). https://doi.org/10.1016/j.ccell.2019.02.001.
Turajlic S, Swanton C. Metastasis as an evolutionary process. Science. 2016;352:169–75. https://doi.org/10.1126/science.aaf2784.
Article PubMed CAS Google Scholar
Reticker-Flynn, N. E. et al. Lymph node colonization induces tumor-immune tolerance to promote distant metastasis. Cell. 185, 1924–1942 e1923 (2022). https://doi.org/10.1016/j.cell.2022.04.019426" Text="" -->
Jebali, A. & Dumaz, N. The role of RICTOR downstream of receptor tyrosine kinase in cancers. Mol Cancer. 17(2018). 39. https://doi.org/10.1186/s12943-018-0794-0
Mateo LJ, et al. Visualizing DNA folding and RNA in embryos at single-cell resolution. Nature. 2019;568:49–54. https://doi.org/10.1038/s41586-019-1035-4.
Article PubMed PubMed Central CAS Google Scholar
Yang, L. et al. (Tn5-)FISH-based imaging in the era of 3D/spatial genomics. Biophys Rep. 9, 15–25 (2023). https://doi.org/10.52601/bpr.2023.220025.
Weintraub, A. S. et al. YY1 Is a Structural Regulator of Enhancer-Promoter Loops. Cell. 171, 1573–1588 e1528 (2017). https://doi.org/10.1016/j.cell.2017.11.008.
Sofiadis, K. et al. HMGB1 coordinates SASP-related chromatin folding and RNA homeostasis on the path to senescence. Mol Syst Biol. 17, e9760 (2021). https://doi.org/10.15252/msb.20209760.
Hu S, et al. NOTCH-YAP1/TEAD-DNMT1 Axis Drives Hepatocyte Reprogramming Into Intrahepatic Cholangiocarcinoma. Gastroenterology. 2022;163:449–65. https://doi.org/10.1053/j.gastro.2022.05.007.
Article PubMed CAS Google Scholar
Tu M, et al. TNF-alpha-producing macrophages determine subtype identity and prognosis via AP1 enhancer reprogramming in pancreatic cancer. Nat Cancer. 2021;2:1185–203. https://doi.org/10.1038/s43018-021-00258-w.
Article PubMed CAS Google Scholar
Owens DDG, et al. Dynamic Runx1 chromatin boundaries affect gene expression in hematopoietic development. Nat Commun. 2022;13:773. https://doi.org/10.1038/s41467-022-28376-8.
Article PubMed PubMed Central CAS Google Scholar
Adhikary S, Roy S, Chacon J, Gadad SS, Das C. Implications of Enhancer Transcription and eRNAs in Cancer. Cancer Res. 2021;81:4174–82. https://doi.org/10.1158/0008-5472.CAN-20-4010.
Article PubMed CAS Google Scholar
Liu J, et al. N (6)-methyladenosine of chromosome-associated regulatory RNA regulates chromatin state and transcription. Science. 2020;367:580–6. https://doi.org/10.1126/science.aay6018.
Article PubMed PubMed Central CAS Google Scholar
Barshad G, et al. RNA polymerase II dynamics shape enhancer-promoter interactions. Nat Genet. 2023;55:1370–80. https://doi.org/10.1038/s41588-023-01442-7.
Article PubMed PubMed Central CAS Google Scholar
Allepuz-Fuster P, et al. RNA polymerase II plays an active role in the formation of gene loops through the Rpb4 subunit. Nucleic Acids Res. 2019;47:8975–87. https://doi.org/10.1093/nar/gkz597.
Article PubMed PubMed Central CAS Google Scholar
Barilla D, Lee BA, Proudfoot NJ. Cleavage/polyadenylation factor IA associates with the carboxyl-terminal domain of RNA polymerase II in Saccharomyces cerevisiae. Proc Natl Acad Sci U S A. 2001;98:445–50. https://doi.org/10.1073/pnas.98.2.445.
Article PubMed PubMed Central CAS Google Scholar
O’Brien MJ, Ansari A. Beyond the canonical role of TFIIB in eukaryotic transcription. Curr Genet. 2022;68:61–7. https://doi.org/10.1007/s00294-021-01223-x.
Article PubMed CAS Google Scholar
Mattingly M, et al. Mediator recruits the cohesin loader Scc2 to RNA Pol II-transcribed genes and promotes sister chromatid cohesion. Curr Biol. 2022;32:2884-2896.e2886. https://doi.org/10.1016/j.cub.2022.05.019.
Article PubMed PubMed Central CAS Google Scholar
Shchaslyvyi, A. Y., Antonenko, S. V., Tesliuk, M. G. & Telegeev, G. D. Current State of Human Gene Therapy: Approved Products and Vectors. Pharmaceuticals (Basel). 16 (2023). https://doi.org/10.3390/ph16101416.
Li S, et al. Proteomic Landscape of Exosomes Reveals the Functional Contributions of CD151 in Triple-Negative Breast Cancer. Mol Cell Proteomics. 2021;20: 100121. https://doi.org/10.1016/j.mcpro.2021.100121.
Article PubMed PubMed Central CAS Google Scholar
Mumbach MR, et al. HiChIP: efficient and sensitive analysis of protein-directed genome architecture. Nat Methods. 2016;13:919–22. https://doi.org/10.1038/nmeth.3999.
Article PubMed PubMed Central CAS Google Scholar
Vences-Catalán, F. et al. Targeting the tetraspanin CD81 reduces cancer invasion and metastasis. Proc Natl Acad Sci USA. 118 (2021). https://doi.org/10.1073/pnas.2018961118.
Gao H, et al. The BMP inhibitor Coco reactivates breast cancer cells at lung metastatic sites. Cell. 2012;150:764–79. https://doi.org/10.1016/j.cell.2012.06.035.
Article PubMed PubMed Central CAS Google Scholar

Download references

Acknowledgements

This work was supported by grants from the Chinese National Major Scientific Research Program (2016YFA0500301), Beijing Scientific Projects (7172213), Jinfeng Lab (JFLKYXM202303 AZ-203), and the National Natural Science Foundation of China (81230044, 81872249 and 81830092). The finding body agreed our design, data collection methods and analysis processes.

Funding

This work was supported by grants from the Chinese National Major Scientific Research Program (2016YFA0500301), Beijing Scientific Projects (7172213), Jinfeng Lab (JFLKYXM202303 AZ-203), and the National Natural Science Foundation of China (81230044, 81872249 and 81830092). The finding body agreed our design, data collection methods and analysis processes.

Author information

Authors and Affiliations

State Key Laboratory of Membrane Biology, School of Basic Medical Sciences, School of Medicine, Tsinghua University, Beijing, 100084, China
Jianghua Li, Lu Xu, Jiayu Wang, Yuting Lin, Fangli Ren, Yinyin Wang & Zhijie Chang
Department of Hematology, The First Affiliated Hospital, Hainan Medical University, Haikou, Hainan, 570102, China
Lu Xu
Department of General Surgery, General Hospital of Northern Theater Command, No. 83, Wenhua Road, Shenhe District, Shenyang, 110016, China
Xuning Wang
College of Letters & Science, University of California Berkeley, Berkeley, CA, 94720, USA
Alex Yutian Zou
Jinfeng Laboratory, No. 313, Jinyue Road, High-Tech Zone, Chongqing, 401329, China
Jun Li

Authors

Jianghua Li
View author publications
Search author on:PubMed Google Scholar
Lu Xu
View author publications
Search author on:PubMed Google Scholar
Jiayu Wang
View author publications
Search author on:PubMed Google Scholar
Xuning Wang
View author publications
Search author on:PubMed Google Scholar
Yuting Lin
View author publications
Search author on:PubMed Google Scholar
Alex Yutian Zou
View author publications
Search author on:PubMed Google Scholar
Fangli Ren
View author publications
Search author on:PubMed Google Scholar
Yinyin Wang
View author publications
Search author on:PubMed Google Scholar
Jun Li
View author publications
Search author on:PubMed Google Scholar
Zhijie Chang
View author publications
Search author on:PubMed Google Scholar

Contributions

Jianghua Li, Yinyin Wang, Jun Li and Zhijie Zhang jointly designed the entire experimental design of the article and wrote the main manuscript text. Jianghua Li completed most of the experiments in the main manuscript. Lu Xu prepared Fig. 2I-2S. Jiayu Wang prepared Fig. 8L-8 T and animal experiments. Xuning Wang analyzed the data in supplementary figure S1 A-S1 F. Yuting Lin prepared Fig. 2 A-2H. Alex Yutian Zou analyzed the data in supplementary figure S1H-S1G. Yinyin Wang and Fangli Ren provided technical support on AAV preparation.

Corresponding authors

Correspondence to Yinyin Wang, Jun Li or Zhijie Chang.

Ethics declarations

Ethics approval and consent to participate

All animal experiments were conducted in strict accordance with the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, 8 th Edition, 2011) and were approved under protocol number 23-CZJ4. The study received approval from the Ethics Committee of the Institutional Animal Care and Use Committee (IACUC) of Tsinghua University. Informed consent was obtained from all participants before their inclusion in the research.

Consent for publication

All authors have provided their consent for the publication of this manuscript.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Supplementary Material 1.

Supplementary Material 2.

Supplementary Material 3.

Supplementary Material 4.

Supplementary Material 5.

Supplementary Material 6.

Supplementary Material 7.

Supplementary Material 8.

12943_2025_2361_MOESM9_ESM.tif

Supplementary Material 9: Figure S1 CREPT is highly expressed in TNBC with lymph-node metastasis. (A) Frequencies of CREPT amplifications. The error bars denote the 80% confidence intervals. Significance was assessed by weighted logistic regression. (B) The correlation between the CREPT copy-number amplifications and its mRNA expression levels (R² = 0.5568, P < 0.0001). (C-D) CREPT is highly expressed in TNBCs in different stages. Boxplots show the mRNA level of CREPT in normal and tumor tissues. Data were obtained from GEO (GSE235640 and GSE33926) (C), METABRIC (D), and GTEx database and analyzed using an unpaired two-tailed t-test (**: P < 0.01;***: P < 0.001). (E-F) An association between CREPT expression and overall survival of TNBC patients by Kaplan-Meier analysis. Data were obtained from TCGA (E) and METABRIC (F) and analyzed using the Kaplan–Meier method (log-rank test). (G) An association between CREPT copy number and overall survival of TNBC patients by Kaplan-Meier analysis. Data were obtained from TCGA. (H) An enrichment analysis comparing the expression of metastasis-associated events in six patients. Enrichment scores were calculated at the individual sample level using gene set variation analysis (GSVA). (I) A volcano plot showing genes with differential expression levels in CREPT-high or CREPT-low cancer cells. (J) A GO analysis of upregulated genes in CREPT-high cells versus CREPT-low cells. (K) T-SNE plot shows the relative expression of metastasis-associated genes, including IGFBP4,BCL2, TMPRSS4, for cancer cells as indicated in blue (right). Data were obtained from (GSE118390) from (I-K).

12943_2025_2361_MOESM10_ESM.tif

Supplementary Material 10: Figure S2 CREPT promotes tumorsphere growth and metastasis for TNBC. (A-B) The expression of CREPT in different cell lines. Western blots were performed in different human (A) or mouse (B) TNBC cell lines with different metastatic potential using an antibody against CREPT. Beta-actin was used as a loading control. The metastatic capability is demonstrated according to literatures arbitrarily [102, 103]. (C) Establishment of stable cell lines with CREPT depletion using three shRNAs against CREPT in LM2 cells. The protein level of CREPT was detected by Western blot. (D-F) Depletion of CREPT inhibits the migration and invasion abilities of LM2 cells. The invasion and migration cells were stained with crystal violet (D) and quantified (E-F). (G) Deletion of CREPT by CRISPR-dCas9 in 4T1 cells (CREPT KO #1, #2). A Western blot showed the completed deletion of CREPT. (H-K) Representative images and quantitation of the tumorspheres formed by 4T1 cells upon CREPT deletion (H-I) and depletion (J-K). (L-M) Overexpression of CREPT promotes tumor sphere formation. CREPT was induced in 4TO7 cells by Dox at different dosages. Representative images and quantitation of tumorspheres in 4TO7 cells are presented. (N-O) A CCK8 proliferation assay was performed. The absorbance at 450 nm was used as an indicator of cell numbers. (P-Q) Colony formation assays were performed, along with the corresponding quantification results. (R-T) Depletion of CREPT represses metastasis of LM2 cells. A total of 5 x 105 LM2 cells with CREPT depletions were injected into Balb/c nude mice (n = 6 mice per group) via the tail-vein. Metastatic tumors were measured by BLI at the indicated time points (R). A quantitative analysis of the tumor sized is shown (S) and the metastasis-free survival of the mice is demonstrated (T). (U-Z) The lung surfaces, histological images and numbers of metastatic nodules are shown. Mice were implanted with LM2 (U-V) and in 4T1 (W-X) cells upon CREPT depletion, and 4TO7 cells where CREPT was induced by Dox (Y-Z). Data were shown as the mean ± SD and analyzed using a T-test. (N-O,S), Kaplan–Meier analysis with log-rank test (T) and an unpaired two-tailed t-test (E-F,I,K,M,P-Q,V,X,Z). *: P < 0.05;**: P < 0.01;***: P< 0.001.

12943_2025_2361_MOESM11_ESM.tif

Supplementary Material 11: Figure S3 CREPT upregulates metastasis-associated genes at the transcriptional level. (A) Pre-mRNAs of metastasis-associated genes were determined by nuclear run-on experiment in 4T1 cells. Genes were selected according the category of metastasis events including cell adhesion, inflammatory factors, tumor angiogenesis, and extracellular matrix remodeling. (B) Significantly enriched hallmark terms for CREPT-regulated upregulated genes in both pre-mRNA and mRNA were denoted. Data were shown as the mean ± SD and analyzed using an unpaired two-tailed t-test. *: P < 0.05;**: P < 0.01;***: P < 0.001.

12943_2025_2361_MOESM12_ESM.tif

Supplementary Material 12: Figure S4The characteristics of CREPT-regulated enhancers and promoters of genome-wide binding patterns. (A) Genomic distribution of the CREPT occupancy in LM2 cells. Different regions of the genome were colored according to the CREPT occupancy density. (B) Average density of CREPT, H3K4 me3, and H3K27ac by ChIP-seq signals at promoter regions in 4T1 cells. (C) Venn diagram analysis of ChIP-seq peaks of CREPT, H3K4me3, and H3K27ac in 4T1 cells. (D) Average density of CREPT, H3K4me1, and H3K27ac by ChIP-seq signals at enhancer regions in LM2 cells. (E) Venn diagram analysis of ChIP-seq peaks of CREPT, H3K4me1 and H3K27ac at enhancers in LM2 cells. (F) Heatmaps (top panels) and average intensity curves (bottom panels) of ChIP-seq signals for CREPT, H3K4me3 and H3K27ac at promoter regions. Promoters are shown in a 2-kb window (centered on the middle of the TSS, transcription start site) in 4T1 cells under CREPT deletion (CREPT KO). (G) Heatmaps (top panels) and average intensity curves (bottom panels) of ChIP-seq signals for CREPT, H3K4 me1, and H3K27ac at enhancer regions. Enhancers are shown in a 2-kb window (centered on the middle of the enhancer) in CREPT-deletion (KO) or control LM2 cells. (H) Box plot of the CREPT occupancy density by ChIP–seq signals between the promoters and enhancer regions in 4T1 cells. Data were shown as the mean ± SD and analyzed using a one-tailed Mann–Whitney U-test. ****P< 0.001. (I) Top ten de novo motifs (ranked by p-value) enriched at CREPT-occupied promoter regions and enhancer regions in 4T1 cells.

12943_2025_2361_MOESM13_ESM.tif

Supplementary Material 13: Figure S5 CREPT regulates the activity of enhancers in p300-dependent manner. (A-B) Heatmaps (top panels) and average intensity curves (bottom panels) of ChIP-seq signals for H3K4 me1 (A) and H3 K27ac (B) at super-enhancer regions in human LM2 cells. Super-enhancers are shown in a 10-kb window (centered on the middle of the enhancer) in LM2 cells under CREPT deletion (CREPT KO). Wilcoxon rank-sum test was used for the statistical analysis of (A) and (B). (C) Metagene analyses showing the occupancy of CREPT at typical enhancer and super-enhancer constituents in 4T1 (top) and LM2 (bottom) cells. ChIP-seq profiles are shown in mean reads per million per base pair for elements of each class in a region of ± 2 kb centered on the middle of the enhancer. (D) Examples of a typical enhancer near the Vegfc gene (left) and a super-enhancer near the Slc12a5 gene (right) in 4T1 cells with or without CREPT deletion (CREPT KO). The tracks show CREPT, H3K4 me1, H3K27ac, p300, RNAPII, and ATAC-seq signals. (E) Reciprocal immunoprecipitations show the interaction of endogenous CREPT and p300 in 4T1 cells. Antibodies against CREPT and p300 were used for the precipitation and Western blot. IgG was used as a negative control. (F) The average densities of CREPT and p300 by ChIP-seq signals at enhancer regions in 4T1 cells. (G) A Venn diagram analysis shows the overlap between p300- and CREPT-mediated active enhancers defined by H3K4 me1 and H3K27ac. (H) Heatmaps (left panel) and average intensity curves (right panel) of p300 at enhancer regions in 4T1 cells. Wilcoxon rank-sum test was used for the statistical analysis. (I) Heatmaps (left panel) and average intensity curves (right panel) of ATAC-seq signals at enhancer regions in 4T1 cells. Wilcoxon rank-sum test was used for the statistical analysis.

12943_2025_2361_MOESM14_ESM.tif

Supplementary Material 14: Figure S6 CREPT is indispensable for A/B compartment formation. (A) Hi-C maps showing normalized Hi-C interaction frequencies in 4T1 cells. Note that CREPT deletion (CREPT KO) decreased the intra-interactions of chromatins. (B) Demonstration of A/B compartments in sampled chromatins. The maps were constructed from normalized Hi-C interaction frequencies and calculated in an Observed/Expected and Observed/Pearson manner. Sampled chromatins were demonstrated. (C) Switches of A/B compartments. A PCA analysis for the first principle component representing compartment A and B on the whole genome was performed at 100 kb resolution. The density plot describes a global correlation (R²=0.6868) between the eigenvector value from CREPT deletion (KO) and wild-type (WT) 4T1 cells. The X-axis represents the eigenvector value in wild-type 4T1 cells on each genome locus, while the Y-axis represents the value in CREPT-deletion cells in the same locus. (D) Bar plots showing the switches of A/B compartments in the chromatins between WT and CREPT KO cells. Less alterations were observed in the A/B compartment switches in autosomal chromatins, exampled by Chr 17. Switches of the A/B compartments were shown in Chr X between WT and CREPT KO cells. (E) A statistical analysis for the genome sizes with compartment A/B and B/A switches. (F) Relative mRNA levels of genes located in the regions with A/B compartment switched under CREPT depletion. The mRNA levels were calculated from the RNA-seq results corresponding to the genes with A/B compartment switch. In total, 36 genes located in the regions with A/B compartment switch and 71 genes located in the regions with B/A compartment switch were calculated. No significant difference (ns) was observed. Data were shown as the mean ± SD and analyzed using an unpaired two-tailed t-test. ns: no significant; ***: P < 0.001.

12943_2025_2361_MOESM15_ESM.tif

Supplementary Material 15: Figure S7 Analyses of the TAD boundary alteration by CREPT deletion in different chromatins (A) Representative examples of a split TAD in chromosome 3, 10 and 16 following CREPT deletion. Hi-C interaction heatmaps are shown. (B) A representative example of a split TAD boundary with the chromatin loops (upper). The loops were visualized by arcs (middle). Genome-browser snapshots of the regions corresponding to the Hi-C contact matrix displaying the ChIP-seq signals of CREPT, H3K4 me1 and H3 K27ac binding profiles in the chromatin of chromosome 1 are shown (bottom). (C) Relative mRNA levels of genes regulated by TAD alteration in chromosome 1 was calculated according to CREPT-altered TAD- and non-CREPT-altered TAD genes. Data were analyzed using an unpaired two-tailed t-test. ns: no significant: *: P < 0.05; **: P < 0.01. (D) Fold changes in all genes regulated by TAD alteration genome-wide. Fold changes were calculated according the RNA-seq results corresponding to the genes (342) regulated by TAD and genes (248) with non-TAD regulation. Data were shown as the mean ± SD and analyzed using an unpaired two-tailed t-test. ns: no significant; ***: P< 0.001. (E) A schematic model illustrates the split TAD on gene regulation. The split of a large TAD (right) into two small TADs results in the emergence of a boundary, which interrupts the cis-element interaction, leading to the inactivation of gene expression. If the TAD is not split (left), the cis-element interaction persists, sustaining the gene transcription.

12943_2025_2361_MOESM16_ESM.tif

Supplementary Material 16: Figure S8 CREPT mediates the formation of chromatin loops. (A) A density plot illustrating the altered interaction frequency resulting from the deletion of CREPT. The plots were mapped according to the changes in interaction frequency after CREPT deletion and the normalized interaction frequency for each gene. (B) Number of CREPT peaks in different CREPT-mediated chromatin loops. (C) Heatmaps displaying CREPT-mediated chromatin interaction numbers of paired-end tags (PETs) with high-confidence. Typical enhancer associated interactions were defined as those with at least one anchor overlapping a super-enhancer (left). Super-enhancer associated interactions were defined as those with no ends of overlapping a super-enhancer (right). Each row represents an interaction and the color intensity represents the PET count for that interaction. (D) Heatmaps (left) and average intensity curves (right) of ChIP-seq signals for CREPT at TTS (transcription termination site) regions. Note that the CREPT-occupied ChIP signals disappeared in CREPT-deletion (CREPT KO) 4T1 cells. (E) Diagraphs illustrating loop percentages. CREPT-mediated loops were marked with green color and non-CREPT mediated loops were marked with blue color. The loop numbers and percentages are labeled. (F) A GO analysis for the biological process of genes regulated by co-operational loop structures. The categories were defined by gene functions, including the metastasis, development and proliferation. Detailed events were listed with their emergence frequency (denoted as count). (G-H) Epigenetic characteristics of the Mmp13 gene (G) and the Ccl2 gene (H) in 4T1 cells with or without CREPT deletion. Genome-browser snapshots present Hi-C interaction frequencies. The balanced Hi-C two-dimensional contact matrix is plotted (top). The color intensity presents the interaction frequency. Black boxes indicate the E-P loops, and blue boxes indicate the P-T loops. CREPT-mediated chromatin loops by HiChIP are presented in arcs (middle). Normalized ChIP-seq signals of CREPT, H3K4 me1, H3 K27ac, p300, and RNAPII binding profiles and ATAC-seq signals are presented (bottom). The maximum Y-axis values of ChIP-seq signal were set as indicated.

12943_2025_2361_MOESM17_ESM.tif

Supplementary Material 17: Figure S9 CREPT regulates the co-operational loop structures in genes associated with metastasis (A-D) Results of Tn5-FISH show the co-operational loop structure alteration by CREPT deletion. (A and C) Diagrams depict the linear chromatin structure with chromatin interactions (denoted by curves) (A for gene Mmp13, C for geneCcl2). Boxes represent the regions of enhancer, promoter and termination of the gene. Colored boxes indicate the locations of the Tn5 probes. The chromatin regions were marked. (B and D) Images of the co-operational loop structures for Mmp13 (B) and Ccl2 (D) detected by the Tn5 probes. A nucleus is circled and demonstrated in bright field (Gray). Super-enhancers are labeled with probes in red, promoters in green and termination regions in blue. The E-P loops (yellow) are the merged colors with red and green and the P-T loops (cyan) are the merged colors with green and blue. Co-operational loops are the merges of red, green and blue. Note that no merged dots were observed when CREPT was deleted (CREPT KO).

12943_2025_2361_MOESM18_ESM.tif

Supplementary Material 18: Figure S10 Targeting super-enhancers against Mmp13 and Ccl2 disturbs their co-operational loop formation and expression. The effects of interrupting the co-operational loop structures in Mmp13 (A-G) and Ccl2 (H-N) were demonstrated by examining the co-operational loop structure using Tn5-FISH probes (A, H), The E-P loop interaction intensity (B, I), the enrichment of CREPT, H3K4me1, H3K27ac, and RNAPII at the super-enhancer (C, J), the RNAPII occupancy density at the promoter (D, K), the P-T loop interaction intensity (E, L), the expression of Mmp13 or Ccl2 at mRNA (F, M) and protein (G, N) levels were examined in 4T1 cells. The cells were transfected with dCas9-Ctrl or with dCas9-E-Mmp13, or dCas9-E-Ccl2, respectively. Data were shown as the mean ± SD and analyzed using an unpaired two-tailed t-test (B-F,I-M). *: P < 0.05;**: P < 0.01;***: P < 0.001.

12943_2025_2361_MOESM19_ESM.tif

Supplementary Material 19: Figure S11 CREPT activates Mmp13 and Ccl2 in a co-operational loop dependent manner.CREPT was overexpressed by a Dox-inducible system in 4TO7 cells. The cells were transfected with dCas9-Ctrl, dCas9-E-Mmp13 or dCas9-E-Ccl2. The E-P loop interaction intensity (A, G), the enrichment of CREPT, H3K4 me1, H3K27ac, and RNAPII (B, H), the RNAPII occupancy density at the promoter regions (C, I), the P-T loop interaction intensity (D, J), the expression of Mmp13 and Ccl2 at mRNA (E, K) and protein (F, L) levels were examined. Note that Dox-inducible expression of CREPT failed to promote the gene expression when the loop structures were interrupted by using dCas9-E-Mmp13 or dCas9-E-Ccl2.Data were shown as the mean ± SD and analyzed using an unpaired two-tailed t-test (A-E,G-K). *: P< 0.05;**: P < 0.01;***: P < 0.001.

12943_2025_2361_MOESM20_ESM.tif

Supplementary Material 20: Figure S12 Targeting co-operational loop structures dramatically inhibits metastasis in the presence of CREPT overexpression. (A) Evaluating the inhibitory efficiency of co-operational loops targeting different genes on cell proliferation through colony formation assays. along with the corresponding quantification results (B) A total of 5 x 10⁵ 4TO7 cells transfected with dCas9-Ctrl or with dCas9-E-Itga3, dCas9-E-Mmp13, or dCas9-E-Ccl2 were injected into Balb/c mice (n = 6 mice per group) via the tail vein under an induction of CREPT overexpression by Dox. The systemic metastasis was examined by BLI at the indicated time points. (C) The metastatic tumors in the lung were quantified according to the luminescent density (BLI signal). (D) The metastasis-free survival of the mice was analyzed. (E) The metastatic nodules were demonstrated by the lung surface nodules and H&E staining. (F) The number of the metastatic nodules was counted and statistically analyzed. Data were shown as the mean ± SD and analyzed using T-tests (C), Kaplan–Meier analysis with log-rank test (D) and an unpaired two-tailed t-test (A,F) ns: no significant; *: P < 0.05; **: P < 0.01; ***: P < 0.001.

12943_2025_2361_MOESM21_ESM.tif

Supplementary Material 21: Figure S13 CREPT mediates the co-operational loop structure to promote metastasis in the human TNBC cell. (A-C) Deletion of CREPT resulted in disappearance of super-enhancers. CREPT was deleted in human LM2 cells. The densities of H3K4 me1 and H3K27ac indicating the super-enhancers are marked for genes ITGA3 (A), MMP13 (B) and CCL2 (C). (D-F) The E-P loops represented by the interaction frequency for genes ITGA3 (D), MMP13 (E) and CCL2 (F) are shown in the wild-type (blue) and CREPT deletion (green) LM2 cells. (G-I) The P-T loops for genes ITGA3 (G), MMP13 (H) and CCL2 (I) are shown in the wild-type (blue) and CREPT deletion (green) LM2 cells. (J-O) RT-PCR and Western blot analyses showthe gene expression at mRNA (J-L) and protein (M-O) levels in ML2 cells. (P-T) Interrupting the co-operational loops by dCas9-E-ITGA3, dCas9-E-MMP13, or dCas9-E-CCL2 inhibited the metastasis of human LM2 cells in vivo. A total of 5 x 10⁵ of LM2 cells transfected by dCas9-E-ITGA3, dCas9-E-MMP13, or dCas9-E-CCL2 were tail-vein injected into Balb/c mice (n = 6 mice per group). Systemic metastasis was measured by BLI at the indicated time (P), with a quantitative analysis (Q). Metastasis-free survival of the mice (R) and metastatic nodules (S-T) were also examined. Data were shown as the mean ± SD and analyzed using T-tests (Q), Kaplan–Meier analysis with log-rank test (R) and an unpaired two-tailed t-test (D-L,T) *: P < 0.05; **: P < 0.01; ***: P < 0.001.

12943_2025_2361_MOESM22_ESM.tif

Supplementary Material 22: Figure S14 Therapeutic application of AAV-shCREPT on TNBC metastasis.(A) GFP density demonstrates the titers of the AAV-Ctrl and AAV-shCREPT viruses. (B) A Western blot demonstrates the efficiency of AAV-shCREPT on the expression of CREPT. (C) A quantitative presentation of lung surface nodules in mice in a prevention treatment model. (D) Immunostaining for the expression of proteins in the lung of mice with metastasis from veil tail injection of LM2 cells. (E) Tumor weights were monitored during the experiment. (F) Immunostaining for the expression of proteins in the lung of mice from in situ tumor-initiated metastasis. Data were shown as the mean ± SD and analyzed using T-tests (E) and an unpaired two-tailed t-test (C) ns: no significant; ***: P < 0.001.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Li, J., Xu, L., Wang, J. et al. CREPT is required for the metastasis of triple-negative breast cancer through a co-operational-chromatin loop-based gene regulation. Mol Cancer 24, 170 (2025). https://doi.org/10.1186/s12943-025-02361-3

Download citation

Received: 24 December 2024
Accepted: 20 May 2025
Published: 10 June 2025
Version of record: 10 June 2025
DOI: https://doi.org/10.1186/s12943-025-02361-3

CREPT is required for the metastasis of triple-negative breast cancer through a co-operational-chromatin loop-based gene regulation

Abstract

Graphical Abstract

Highlights

Introduction

Results

CREPT amplification is highly related to the metastasis of TNBC

CREPT promotes TNBC metastasis

CREPT upregulates the expression of metastasis-associated genes

CREPT regulates the activity of enhancers in a p300-dependent manner

CREPT influences the large-scale chromatin structure

CREPT promotes the formation of chromatin loops

CREPT-mediated co-operational loops orchestrate the RNAPII loading and recycling

Targeting CREPT is a promising strategy for the therapy of metastatic TNBC

Discussion

Limitations of the study

Future outlooks

STAR + methods

Key resources table

Data availability

Lead contact

Materials availability

Data and code availability

Experimental model and study participant details

Cell culture

Method details

In vitro invasion and migration assays

Tumor sphere formation assay

Immunohistochemical (IHC) assay

Real-time quantitative PCR (RT-qPCR) and data analysis

Nuclear run-on

Western blots

Immunoprecipitation

Sample preparation for tandem mass tag (TMT)

ChIP-MS

Chromosome conformation capture (3C)

CRISPR/Cas9 mediated knock-out

CRISPR-dCas9 mediated silencing

AAV construction and virus package

Animal studies

Bioluminescent imaging

RNA-seq library preparation and analysis

GRO-Seq library preparation and analysis

Chromatin immunoprecipitation sequencing (ChIP-seq) experiment.

Chromatin isolation and sonication

Immunoprecipitation (IP)

ChIP-qPCR

Library preparation

Data processing and mapping

ATAC-seq experiment

Cell preparation and transposition

Library preparation

ATAC-seq data processing and mapping

In situ Hi-C and analysis

HiChIP library preparation and data analysis

Tn5-FISH experiment

Tn5 FISH probe preparation

Fish processing

Single cell analysis

Copy-number analysis

TCGA-RNAseq analysis

Data availability

References

Acknowledgements

Funding

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Ethics approval and consent to participate

Consent for publication

Competing interests

Additional information

Publisher’s Note

Supplementary Information

Rights and permissions

About this article

Cite this article

Share this article