TEAD1 condensates are transcriptionally inactive storage sites on the pericentromeric heterochromatin in cancer cells Authors: Yiran Wang, Jindayi Liang, Kimberly S Lange, Justin Demmerle, Eleanor A Liu, Ethan Black, Britney Jiayu He, Christopher J Ricketts, Shawn R Yoshida, Shasha Chong, W Marston Linehan, Jennifer M Kavran, Chongzhi Zang, Danfeng Cai Published: 2026 ABSTRACT AbstractTEA domain transcription factor 1 (TEAD1), a Hippo pathway transcription factor important in cellular homeostasis and development, is increasingly implicated in cancer biology. Here we reveal an unexpected role for TEAD1 in organizing nuclear condensates, independent of active transcription. Using high-resolution imaging, ChIP–seq, RNA sequencing and proximity-based proteomics, we demonstrate that in patient-derived renal cell carcinoma cells, TEAD1 forms micrometre-sized condensates by binding to heterochromatic pericentromeric regions using its DNA-binding domain. TEAD-specific MCAT motifs selectively enrich and cluster in the pericentromeric region to specifically seed TEAD1 condensates. TEAD1 condensates do not activate transcription but instead serve as depots for excess TEAD1, and disruption of TEAD1 condensates leads to increases in YAP and TEAD target gene expression. This organization of TEAD1 contrasts with that observed in other genomic regions of both renal cell carcinoma and normal kidney cells, in which TEAD1 associates with markers of active transcription. Our findings provide a mechanistic framework for the dual regulatory roles of TEAD1 and offer new insights into its contribution to transcriptional dysregulation and tumour progression.Subject terms: Nuclear organization, Centromeres Main Key nuclear processes such as transcription, RNA splicing and DNA repair occur in membraneless compartments known as biomolecular condensates1–3. These condensates are broadly defined as dynamic assemblies formed through weak interactions between proteins and nucleic acids1,2,4. Specifically, transcriptional condensates are formed by transcription-related factors5. Although they have garnered increased attention, their mechanisms of formation and function remain poorly understood. Liquid–liquid phase separation is the dominant model explaining their assembly6,7. However, alternative mechanisms—such as low-valency interactions with spatially clustered DNA-binding sites—have also been proposed8,9. Moreover, although transcriptional condensates are generally associated with promoting gene expression10–12, some have been shown to repress transcription13,14. Thus, a deeper understanding of how transcriptional condensates form and function in biologically relevant contexts is essential, particularly for the development of strategies to target them in diseases like neurodegeneration and cancer15,16. Renal cell carcinoma (RCC) is a common type of kidney cancer. Although there have been considerable improvements in disease-free survival among patients with RCC, effective management of metastatic disease continues to be a substantial challenge in clinical practice. Moreover, a clear understanding of the molecular mechanism of RCC progression is still much needed17. Genetic alterations in RCC frequently occur in the Hippo pathway18, a deeply conserved signalling network that regulates cell growth19. The Hippo pathway includes key proteins such as STE20-like kinase 1 (MST1) and MST2, large tumour suppressor kinase 1 (LATS1) and LATS2, scaffold proteins such as Salvador homologue 1 (SAV1) and neurofibromatosis type 2 (NF2), which can be activated through cell–cell contacts20. The main role of the Hippo pathway is to phosphorylate and repress its downstream transcriptional coactivators Yes-associated protein (YAP) or its paralogue transcriptional coactivator with PDZ-binding motif (TAZ) by sequestrating them in the cytoplasm21–23. Therefore, when the Hippo pathway is ‘off’ or mutated such as in RCC (most mutations occur in NF2 or SAV1)18,24, YAP and TAZ translocate into the nucleus, bind to the transcription factors TEAD1–TEAD4 and activate genes that promote cell proliferation and survival25. However, where and how components of the Hippo pathway are activated inside the cell is not well understood. Recently, many constituents of the Hippo pathway have been found to form biomolecular condensates. The two transcription coactivators YAP and TAZ were first to be discovered to form condensates that activate transcription4,10,26. Subsequently, upstream Hippo pathway kinases and adaptor proteins were also observed to form condensates, and the effects of these condensates on signal transduction varied27,28. Although the transcription factors TEAD1–TEAD4 are mostly considered to be client proteins that are incorporated into existing YAP and TAZ condensates4,10,26, a recent study reported that TEAD condensates could turn into active hubs after YAP or TAZ overexpression29. However, the overexpression system used to activate transcription limited our understanding of the physiological roles of TEAD condensates. Furthermore, the function of condensates that make up the Hippo pathway components in various cancers is also poorly understood. In this study, we focus on transcription condensates formed by TEAD1 in Hippo-mutated RCC. We found that different from most other transcriptional condensates, TEAD1 condensates localize to heterochromatin in pericentromeric regions and do not mediate active transcription. Our combined chromatin immunoprecipitation followed by sequencing (ChIP–seq) and RNA sequencing (RNA-seq) analyses and proteomics data suggest that TEAD1 condensates may be storage sites that buffer excess TEAD1, such as in Hippo-mutated kidney cancer cells. These results reveal an interesting alternative function for transcriptional condensates and new ways that cells can harness condensates in homeostasis and disease. Results TEAD1 forms large condensates in cancer cells We aimed to determine whether endogenous YAP and TEAD form biomolecular condensates in cancer cells, especially in those with high YAP or TEAD expression. Therefore, we examined a previously established diverse panel of patient-derived RCC cells: the UOK cell lines30. Among these cell lines, UOK121, UOK122, UOK139 and UOK140 are clear-cell RCC lines, whereas UOK275 and UOK342 are papillary RCC lines. A non-cancerous kidney tubule cell line, HK2, was used as a control. As phase separation and condensate formation are tightly coupled to protein expression levels11, we first measured the levels of YAP and TEAD1 in these cell lines. HK2, UOK121, UOK342, UOK122 and UOK140 expressed high levels of YAP (Fig. 1a,b), among which, UOK121 and UOK342 also expressed high levels of TEAD1 (Fig. 1a,b). Examination of these cell lines for YAP–TEAD1 condensate formation, we found that although YAP was diffusely localized to the nucleus and formed small condensates in different cell lines (Extended Data Fig. 1h), TEAD1 formed prominent micrometre-sized condensates only in UOK121 and UOK342 cells, which express high levels of TEAD1 (Fig. 1c). We observed at least one, and up to five, TEAD1 condensates per nucleus in the majority of UOK121 and UOK342 cells (Fig. 1d). Notably, most of these TEAD1 condensates were either close to the nuclear lamina or the nucleolus (Fig. 1e,f). These condensates were not artefacts of immunofluorescence, as TEAD1 knockdown by RNA interference (Extended Data Fig. 1a,b) eliminated the formation of TEAD1 condensates (Extended Data Fig. 1c). They were also not aggregates, as they did not stain positive for the amyloid marker AmyTracker26 (Extended Data Fig. 1d). Interestingly, TEAD1 condensates seemed to be a general feature of cancer cells. In a panel of melanoma, breast cancer, cervical cancer and osteosarcoma cell lines, TEAD1 condensates were readily observed in most of them, whereas in the respective noncancerous cells, none were found (Extended Data Fig. 2a–e). Among other members of the TEAD family, TEAD2 and TEAD3 did not form condensates, whereas TEAD4 formed condensates, although less prominently, in UOK121 and UOK342 cell lines (Extended Data Fig. 1e). These results may be explained by the lower expression of TEAD2 and TEAD3 than TEAD1 and TEAD4 in different cell lines (Extended Data Fig. 1f,g). We have previously found that TEAD1 cannot form condensates by itself but can incorporate into YAP condensates and stabilize YAP condensate formation10,26. Interestingly, we here found that TEAD1 formed condensates that do not enrich endogenous YAP (Extended Data Fig. 1h,i). We were intrigued by these TEAD1 condensates and went on to examine how they form and function. TEAD1 forms condensates in various RCC lines. a, Representative immunoblots showing TEAD1 and YAP protein levels in LTC-positive versus LTC-negative cell lines. b, Quantification of TEAD1 and YAP expression bands from a, normalized to the loading control. Data are the mean ± s.e.m. (n = 4 biological replicates and 3 biological replicates for YAP). P values were determined by one-way analysis of variance (ANOVA) multiple-comparison tests, with exact adjusted P < 0.05 values shown. P values of each comparison are provided in the ‘Statistics and reproducibility’ section in the Methods. c, Representative immunofluorescence images of TEAD1 condensate formation in kidney tubule and RCC lines. Arrows indicate TEAD1 condensates. Nuclear boundaries are indicated by dotted lines. Scale bar, 5 μm. d, Quantification of LTCs in UOK cell lines using 3D analysis and compared with HK2 cells. Data are presented as a bar graph with mean ± s.e.m. (n = 73, 90, 72 and 77 cells for HK2, UOK121, UOK122 and UOK342, respectively). Significance was determined by one-way ANOVA multiple-comparison tests. P values are provided on the graph. e, Representative maximum-intensity projections from 3D images of TEAD1 (green) and nucleolin (red) co-immunofluorescence in UOK121 and UOK342 cells. Arrows indicate TEAD1 condensates. Nuclear boundaries are indicated by dotted lines. f, Quantification of TEAD1 condensate nucleolar and perinuclear localization (n = 94 and 98 cells for UOK121 and UOK342 cells, respectively). g, Full-bleach FRAP and half-bleach FRAP over 10 s of UOK342 cells stably expressing eGFP–TEAD1. Scale bar, 2 μm h, Quantification of recovered fluorescence intensity of fully bleached TEAD1 condensates (left), and of half-bleached and unbleached regions (right). Shaded areas indicate the s.e.m. for each measurement (n = 10 technical replicates). i, Representative live-cell images of UOK342 TEAD1–HaloTag knock-in cells after treatment with 1% DMSO control or 1,6HD. HaloTag fluorescence was labelled with JFX554 HaloTag ligand. Nuclear boundaries are indicated by dotted lines. Scale bar, 5 μm. j. Quantification of LTC counts per cell after DMSO or 1,6HD treatment shown in i (n = 32 and 30 cells for DMSO and 1,6HD, respectively). Source data TEAD1 condensates are not protein aggregates or artifacts. a-b, Immunoblotting of efficient TEAD1 knockdown in UOK121 and UOK342 cells in (a), Mock: no transfection. Quantification of TEAD1 expression between TEAD1 siRNA treatment and negative control (NC) in UOK121 and UOK342 in (b), data are shown as mean ± SEM (n = 3 biological replicates). Statistical significance was determined by one-way ANOVA. c, Immunofluorescence images of TEAD1 (green) in UOK121 and UOK342 cells upon treatment with TEAD1 knockdown siRNA (TEAD1) or control siRNA (NC). Scale bars represent 5 μm. Nuclear boundaries are indicated by dotted lines. d, Representative 2D immunofluorescence images showing AmyTrakcer680 staining do not overlap with TEAD1 condensates visualized by TEAD1 antibody, but overlaps with protein aggregates formed by expressing FUS R495X mutant (detected by Myc antibody) in U-2 OS cells. Scale bars are 5μm. Insets show magnified images of boxed regions. Nuclear boundaries are indicated by dotted lines. e, Representative 2D immunofluorescence images showing TEAD2-4 staining in HK2, UOK121, UOK342, and UOK122 cells. Scale bars are 5μm. Nuclear boundaries are indicated by dotted lines. f-g, Immunoblotting of TEAD2-4 in LTC-positive and LTC-negative cell lines in (f). Quantification of TEAD1-4 expression band intensity cross-compared in HK2 and UOK cells in (g), data normalized to β-actin. Data are presented as mean ± SEM. Statistical significance was determined by one-way ANOVA, only p-values < 0.05 are shown, exact p values of each comparison provided in Statistics and Reproducibility section. h, Representative immunofluorescence images of TEAD1 condensates and their spatial relationship with YAP in UOK121 and UOK342. Insets show magnified images of boxed regions. Scale bars are 5μm. Nuclear boundaries are indicated by dotted lines. i, Quantitative analysis and statistical evaluation of TEAD1 condensates co-localization with YAP. Co-localization was assessed using Manders’ coefficient within defined regions surrounding TEAD1 condensates, with control measurements taken from random nucleoplasmic regions (n indicated in the figure equals the number of regions of interest). On the box plot, centre line shows the median, the bounds of the box correspond to interquartile (25th–75th) percentile, and whiskers extend to furthest data points within 1.5x the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. Statistical significance was determined by one-way ANOVA. Source data TEAD1 forms condensates in other cancer cell lines. a, Representative immunofluorescence images of TEAD1 staining in different cancer (Melanoma: 1205Lu, WM983B, FS4; Breast cancer: T47D, MCF-7, MDA-MB-231; Cervical cancer: HeLa; Osteosarcoma: U-2 OS) and non-cancer cell lines (Fibroblast: AG13004; mammary gland epithelial cells: MCF-10A; embryonic kidney cells: HEK293). Scale bars = 5 µm. b, Quantification of percent LTC-positive cells in various cell lines shown in (a). Data are presented as mean ± SEM (n = 3 independent experiments, with >75 cells of 1205Lu, >54 cells of WM983B, >36 cells of FS4, >27 cells of AG13004, and >100 cells of other cell lines imaged per experiment). Statistical significance was determined by one-way ANOVA, only p-values < 0.05 are shown, exact p values of each comparison provided in Statistics and Reproducibility section. c-d, Immunoblot of TEAD1 expression in cell lines shown in (a), quantification and statistical analysis of the expression bands in (d). Data normalized to β-actin and TEAD1 expression in HK2. Data are presented as mean ± SD (n = 3 biological replicates). e, Scatter plot of TEAD1 protein expression by percent LTC-positive cells in cancer cell lines (black) and non-cancer cell lines (pink). Each data point shown as mean ± SD of percent LTC-positive cells (b) by averaged TEAD1 protein expression level of each cell line from (d). (n = 3 biological replicates) f, Representative immunofluorescent images of TEAD1 (green) and H3K9me3 (Red) in 1205Lu, Hela, and U-2 OS cell lines. Scale bars = 5 µm. Insets show magnified images of boxed regions. Nuclear boundaries are indicated by dotted lines. Source data To probe the biophysical properties of TEAD1 condensates, we used fluorescence recovery after photobleaching (FRAP). Specifically, we used full-condensate FRAP to examine the diffusion of TEAD1 molecules between TEAD1 condensates and the nucleoplasm, and half-condensate FRAP to examine the dynamics of TEAD1 inside the condensates. TEAD1 condensates recovered fluorescence rapidly after full-condensate FRAP (t1/2 = 3.44 ± 0.067 s, mobile fraction = 0.838), which indicated that TEAD1 exchanges quickly between the nucleoplasm and the condensates (Fig. 1g,h (left)). In the half-condensate FRAP experiments, the bleached area also recovered fluorescence rapidly (Fig. 1g,h (right, black line)), which was accompanied by a 25% dip in the unbleached area immediately after bleaching (Fig. 1h, right, green line). These findings indicate that TEAD1 molecules preferentially mix within TEAD1 condensates, a result consistent with a mechanism driven by liquid–liquid phase separation8. Notably, treatment with 1,6-hexanediol (1,6HD), an aliphatic alcohol1,2,31, did not dissolve the TEAD1 condensates in the nucleus nor did it alter their number (Fig. 1i,j). Together, these results suggest that TEAD1 condensates are liquid-like compartments stabilized by interactions that are not primarily hydrophobic. TEAD1 condensates do not mediate active transcription The TEAD family of transcription factors usually mediate active gene expression. To investigate whether TEAD1 condensates also mediate active transcription, we examined their association with the following markers: RNA polymerase II phosphor-serine2 (RNA PolII-pSer2), a marker of transcription elongation32; histone 3 lysine 27 acetylation (H3K27ac), a histone modification denoting active enhancers33; and bromodomain-containing protein 4 (BRD4), a transcriptional coactivator found at super-enhancers34,35. To our surprise, TEAD1 condensates did not colocalize with any of these active transcription markers but instead excluded them (Fig. 2a,b). Consistently, there was no active transcription from large TEAD1 condensates (LTCs), as evidenced by a lack of 5-ethynyl uridine staining in nascent RNA labelling experiments, unlike what was seen for the nucleolus, an active ribosomal RNA transcription centre (Fig. 2c,d and Extended Data Fig. 3a). Furthermore, TEAD1 condensates did not colocalize with the canonical YAP–TEAD target gene CYR61 (also known as CCN1)36,37 when visualized by simultaneous immunofluorescence against TEAD1 and nascent RNA fluorescence in situ hybridization (FISH) against CYR61 (Fig. 2e–g). These findings indicate that TEAD1 condensates do not mediate active transcription at the YAP–TEAD1 canonical target gene. Moreover, TEAD1 condensates did not colocalize with other known nuclear membraneless organelles such as Cajal bodies, nuclear speckles or PML nuclear bodies (Extended Data Fig. 3b–e). TEAD1 condensates therefore represent a unique kind of condensate that contains a transcription factor yet does not mediate active transcription. Interestingly, TEAD1 not only formed LTCs but also formed many small TEAD1 condensates (STCs). Compared with LTCs (1.31 ± 0.48 µm3 in size and around 3 per cell), STCs were only 0.030 ± 0.015 µm3 in size and were present at an average of around 169 per cell (Fig. 2h). The STCs resembled canonical transcriptional condensates. That is, they colocalized with the coactivator YAP and mediated active transcription, as evidenced by their colocalization with the YAP–TEAD1 target gene transcript CYR61 (Fig. 2i,j) and their much shorter distance to CYR61 than the LTCs (Fig. 2k,l). Therefore, TEAD1 can form two types of condensates that differ in size and function. We therefore focused on characterizing the LTCs, as they differed from the canonical transcription condensates. From here on, we refer to them as TEAD1 condensates for simplicity. LTCs are transcriptionally inactive. a, Representative immunofluorescence images of TEAD condensates (green) and the active transcription markers (red) RNA PolII-pSer2, H3K27ac and BRD4 in UOK121 and UOK342 cells. Scale bars, 5 µm. Insets show magnified boxed regions of merged channels with adjusted intensity. Scale bars, 2 μm. Nuclear boundaries are indicated by dashed lines. b, Quantitative assessment and statistical evaluation of colocalization of TEAD1 with active transcription markers around condensates compared with control regions in the nucleoplasm (n indicated in the figure equals the number of regions of interest). P values were determined by one-way ANOVA multiple-comparison tests, with exact P values provided in the ‘Statistics and reproducibility’ section in the Methods. c, Representative immunofluorescence images of TEAD1 and nucleolin in UOK342 cells treated with DMSO control or 1 mM 5-ethynyl uridine (EU) for 10 min. Nascent RNA incorporated with EU are visualized by Click-iT RNA Alexa Fluor 594 azide (Alexa594 azide). Insets show magnified boxed regions. Scale bars, 5 µm d, Quantification of colocalization (assessed using Manders’ coefficient) of TEAD1 with EU in UOK121 and UOK342 cells treated with DMSO control or EU for 10 min. Measurements were made around LTCs and compared with control regions in the nucleoplasm (n indicated in the figure equals the number of regions of interest). P values were determined by one-way ANOVA multiple-comparison tests. e,f, Representative images of LTCs and FISH of nascent CYR61 mRNA in UOK121 (e) and UOK342 (f) cells. Arrows indicating CYR61 nascent mRNA signals and TEAD1 condensates are highlighted in the boxed regions. Nuclear boundaries are indicated by dotted lines. Insets show averaged images centred on a LTC, TEAD1 (grey) and CYR61 (red) in UOK121 cells (n = 35 images) and UOK342 cells (n = 20 images). Scale bars, 1 µm. g, Line plots of LTC and CYR61 average intensities (in arbitrary units (a.u.)) along the dashed lines in e and f for UOK121 (left) and UOK342 (right) cells. h, Representative images of STC (blue) and LTC (pink) 3D object surface identification using Imaris. Scale bars, 3 µm. The table on the right shows the average size and count of the two types of condensates in UOK121 and UOK342 cells (n = 62 cells). i, Averaged images centred on nascent CYR61 mRNA (magenta), YAP (red) and TEAD1 (green) in UOK342 cells (n = 107 images). Scale bars, 0.5 µm. j, Line plots of CYR61, YAP and TEAD1 average intensities along the dashed lines in i. k, Representative images of 3D object surface classification processed using Imaris (v.10.2.0, Bitplane). Scale bar, 5 µm l, Quantification of the minimum distance from nascent CYR61 mRNA to STCs, YAP and LTCs. Data are the mean ± s.e.m. (n = 30, 22 and 17 cells for the distance from CYR61 to STCs, YAP and LTCs, respectively). P values were determined by one-way ANOVA multiple-comparison tests. For box plots in b and d, the centre line shows the median, the bounds of the box correspond to the interquartile (25th–75th) percentile and the whiskers extend to the farthest data points within 1.5× the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. Source data TEAD1 condensates do not colocalize with other nuclear bodies. a, Quantification of colocalization of nucleolin with EU in UOK121 and UOK342 treated with DMSO control or EU for 10 min, measurement made around nucleolin condensates with comparable size of LTC and control regions, as shown in Fig. 2c (n indicated in the figure equals the number of regions of interest). p values were determined by one-way ANOVA multiple comparison tests. b-d, Representative UOK342 immunofluorescence images of TEAD1 condensates (green) with various Nuclear Bodies (red): Cajal bodies (b), nuclear speckles (c), and PML bodies (d). Scale bars are 5μm. Nuclear boundaries are indicated by dotted lines. e, Quantification of co-localization of immunofluorescence imaging of TEAD1 with indicated MLO markers in UOK121 and UOK342 around LTC regions and randomly selected nuclear regions (control) in (b-d) (n indicated in the figure equals the number of regions of interest). p values were determined by one-way ANOVA. For box plots in (a) and (e), centre line shows the median, the bounds of the box correspond to interquartile (25th–75th) percentile, and whiskers extend to furthest data points within 1.5x the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. Source data TEAD1 condensates localize to the pericentromeric heterochromatin region TEAD1 condensates do not associate with markers of active transcription (Fig. 2a,b) and are close to the nuclear lamina and the nucleolus (Fig. 1e,f), regions typically enriched in heterochromatin. Therefore, we went on to test whether TEAD1 condensates localize to heterochromatin. Histone 3 lysine 9 trimethylation (H3K9me3) is an epigenetic marker characteristic of heterochromatin38. Staining of TEAD1 and H3K9me3 showed that TEAD1 condensates colocalized with a H3K9me3 domain in condensate-positive kidney cancer and other cancer cell lines (Fig. 3a–d and Extended Data Fig. 2f). No co-occupancy between TEAD1 and H3K9me3 was observed in condensate-negative UOK122 and HK2 cells (Extended Data Fig. 4a–d). Furthermore, in UOK121 cells, TEAD1 condensates colocalized with heterochromatin protein 1α (HP1α; Extended Data Fig. 4e–h), a protein responsible for heterochromatin assembly39. Polycomb bodies organize heterochromatin regions to repress gene expression40,41. TEAD1 condensates did not colocalize with markers of Polycomb bodies (Extended Data Fig. 4i,j), which indicated that TEAD1 condensates localize at unique heterochromatin regions distinct from Polycomb bodies. Instead, TEAD1 condensates were always adjacent to the centromere and showed a bimodal distribution around the centromere together with H3K9me3 (Fig. 3e–h and Extended Data Fig. 4k–n). This result suggests that TEAD1 condensates specifically bind to pericentromeric heterochromatin, a region crucial for centromere integrity and proper segregation of chromosomes during mitosis42,43. This finding was further supported by the much closer distance of TEAD1 condensates (0.2 µm) to the centromere histone marker centromere-specific protein A (CENP-A)42 than to the telomere marker telomeric repeat binding factor 2 (TRF2)44 (1 µm; Fig. 3i–l). These results suggest that TEAD1 condensates specifically localize to pericentromeric heterochromatin. TEAD1 condensates colocalize with heterochromatin markers in different RCC cell lines. a–d, Representative immunofluorescence images of TEAD1 (green) and H3K9me3 (red) in UOK121 (a) and UOK342 (b) cells. Scale bars, 5 μm. Insets show magnified images of the boxed regions. Colocalization of TEAD1 condensates and H3K9me3 (assessed using Manders’ coefficient) in UOK121 (c) and UOK342 (d) cells compared with the control regions, where TEAD1 is diffused in the nucleoplasm (n indicated in the figure equals the number of regions of interest). The centre line in the box plot shows the median, the bounds of the box correspond to the interquartile (25–75th) percentile and whiskers extend to farthest data points within 1.5× the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. Significance was determined by unpaired t-tests, two-tailed. e–h, Representative immunofluorescence images illustrate the colocalization of TEAD1 (red) with H3K9me3 (magenta) and adjacent to centromere DNA FISH (green) in UOK121 (e) and UOK342 (g) cells. Insets show magnified images of the boxed regions. Scale bars, 5 µm. Signal intensity profiles along the lines indicated on the merged-channel images for UOK121 (f) and UOK342 (h) cells. i–l, Representative immunofluorescence images and quantification illustrate the colocalization of TEAD1 (green) with TRF2 (magenta) and CENP-A markers (red) in UOK121 (i) and UOK342 (j) cells. Insets show magnified images of the boxed regions. Quantification used 3D nearest neighbour analysis in UOK121 (k) and UOK342 (l) cells, which demonstrated a tighter distribution of CENP-A at a smaller minimal distance. Data are the mean ± s.e.m. (n indicated in the figure equals the number of regions of interest). Significance was determined by unpaired t-tests, two-tailed. Source data TEAD1 condensates colocalize with heterochromatin markers but not the polycomb bodies. a-d, Representative immunofluorescence images of TEAD1 (green) and H3K9me3 (red) in HK2 (a) and UOK122 (b) cells, scale bars = 5 μm. Insets show magnified images of boxed regions. Co-localization analysis between TEAD1 and H3K9me3 around LTC and control nuclear regions (control) in HK2 (c) and UOK122 (d) cells (n indicated in the figure equals the number of regions of interest). Statistical significance determined by unpaired t-test, two-tailed. e-h, Representative immunofluorescence images illustrate the co-localization of TEAD1 (green) with HP1 (red) in UOK121 (e) and UOK342 (f) cells. Insets show magnified images of boxed regions. Co-localization analysis between TEAD1 and HP1 around LTC and control nuclear regions (control) in UOK121 (g) and UOK342 (h) cells (n indicated in the figure equals the number of regions of interest). Statistical significance determined by unpaired t-test, two tailed. i-j, Representative immunofluorescence images illustrate the co-localization of TEAD1 (green) with Polycomb Repressive Complex 1 (PRC1) marker RING1B or Polycomb Repressive Complex 2 (PRC2) marker EED (Red) in UOK121 and UOK342. Insets show magnified images of boxed regions. Co-localization analysis between TEAD1 and Polycombs around LTC and control nuclear regions (control) in UOK121 and UOK342 in (j) (n indicated in the figure equals the number of regions of interest). Statistical significance determined by one-way ANOVA. For box plots in (c), (d), (g), (h) and (j), centre line shows the median, the bounds of the box correspond to interquartile (25th–75th) percentile, and whiskers extend to furthest data points within 1.5x the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. k-n, Average intensities of images centered at the centromere in UOK121 (k) and UOK342 (l). UOK121 n = 23 condensates, UOK342 n = 23 condensates. Line plots of H3K9me3 (magenta line), TEAD1 (red line), and centromere DNA-FISH signal (green line) in UOK121 (m) and UOK342 (n). Source data The DNA-binding domain of TEAD1 is essential for TEAD1 condensate formation The localization of LTCs to the pericentromeric region suggests that TEAD1 condensate formation may be seeded by DNA binding. To test this possibility, we generated variants of TEAD1 with impaired DNA binding. The DNA-binding domain of TEAD1 (the TEA domain) is composed of a highly conserved homeodomain region containing three α-helices: helix 1 (H1), helix 2 (H2) and helix 3 (H3)45 (Fig. 4a). To disrupt DNA binding by TEAD1, we constructed TEAD1 variants that either lacked the DNA-binding domain (Δ55–122) or bore previously validated site-specific substitutions in the DNA-binding domain: either F43P and L47P (TEAD1(H1Mut)) or Q89P and S92P (TEAD1(H3Mut))46 (Fig. 4a). When transiently expressed in both LTC-positive and LTC-negative cells, compared with wild-type (WT) TEAD1 labelled with eGFP (eGFP–TEAD1(WT)), variants with impaired DNA binding formed fewer condensates (Fig. 4b–e and Extended Data Fig. 5a,b) under similar or even higher expression levels (Extended Data Fig. 5c–g). Furthermore, the eGFP signals of these mutants did not colocalize with heterochromatin (Fig. 4f–k and Extended Data Fig. 5h–m). By contrast, alteration of other TEAD1 regions, such as the amino-terminal intrinsically disordered region (IDR), the YAP-binding site (Y406)47 or the protein kinase A phosphorylation site (S102)48, did not disrupt LTCs nor their association with heterochromatin (Fig. 4a–e,l–p). These data indicate that the DNA-binding domain of TEAD1 is essential for TEAD1 condensate formation on heterochromatin. TEAD1 condensate formation is mediated by the TEAD1 DNA-binding domain. a, Illustration of the expression constructs of eGFP-tagged TEAD1(WT) (WT) and the following DNA-binding domain (DNA BD) mutants of TEAD1: Δ55–122, H1Mut and H3Mut. The underlined amino acid sequence represents the NLS of TEAD1, and the red amino acids are mutated in the respective mutants. The PKA phosphorylation site (S102) is labelled purple. The triangle indicates the site responsible for YAP binding (Y406). b–d, Representative images of UOK342 cells expressing eGFP-tagged WT TEAD1 or TEAD1 mutants (b). Scale bar, 20 µm. Quantification of the percentage of LTC-positive cells after transfection of TEAD1 variants in UOK121 (c) and UOK342 (d) cells. Data are the mean ± s.e.m. (n = 3 independent experiments, with at least 100 cells of each cell line imaged per experiment). e, Quantification of LTC area of UOK342 cells expressing LTC-positive mutants compared with eGFP–TEAD1(WT) shown in b–d (n = 29, 30, 34, 31 and 48 LTCs for WT, ΔIDR, Y406H, S102A and S102D, respectively). P values were determined by one-way ANOVA Dunnett’s multiple-comparisons tests. f–i, Representative images of UOK121 and UOK342 cells expressing eGFP-tagged variants of TEAD1—WT (f), Δ55–122 (g), H1Mut (h) or H3Mut (i)—with immunofluorescence signals of H3K9me3 (red). Insets show magnified images of the boxed regions. Scale bars, 5 μm. j,k, Quantification of colocalization between H3K9me3 and eGFP-tagged TEAD1 variants shown in f–i (n indicated in the figure equals the number of regions centred at H3K9me3 foci). l–o, Representative images of UOK342 cells expressing eGFP-tagged variants of TEAD1—ΔIDR (l), S102A (m), Y406H (n) or S102D (o)—with immunofluorescence signals of H3K9me3 (red). Insets show magnified images of the boxed regions. Scale bars, 5 μm. p, Quantification of colocalization between H3K9me3 and eGFP-tagged TEAD1 variants compared with WT shown in l–o (n indicated in the figure equals the number of regions centred at H3K9me3 foci). P values were determined by one-way ANOVA Dunnett’s multiple-comparisons tests. For the box plots in e, j, k and p, the centre line shows the median, the bounds of the box correspond to the interquartile (25–75th) percentile and whiskers extend to the farthest data points within 1.5× the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. Source data TEAD1 condensates formation is mediated by its DNA-binding domain. a-b, Quantification of percentage LTC-positive cells upon transfection in UOK122 (a), and HK2 (b). Data are presented as mean ± SEM (n = 3 independent experiments, with >100 cells of each cell line imaged per experiment). c, Immunoblot shows EGFP expression level after transfection in HK2 and UOK cell lines. d-g, Quantification of immunoblot proteins expression levels in (a) with GFP antibodies across different indicated cell lines. Data are presented as mean ± SD (n = 1, 1, 2, 2 biological replicates for TEAD1WT, ∆55-122, H1Mut, H3Mut for each cell line transfected respectively). h-k, Representative images of UOK122 and HK2 cells expressing EGFP-TEAD1-WT (h), EGFP-TEAD1-DNA BD deletion (i), EGFP-TEAD1-H1Mut (j), and EGFP-TEAD1-H3Mut (k), with immunofluorescence signal of H3K9me3 (red). Insets show magnified images of boxed regions. Scale bars = 5μm. l-m, Quantification of co-localization between H3K9me3 and EGFP-TEAD1 variants shown in (h-k) (n indicated in the figure equals the number of H3K9me3 foci). On the box plot, centre line shows the median, the bounds of the box correspond to interquartile (25th–75th) percentile, and whiskers extend to furthest data points within 1.5x the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. Statistical significance determined by one-way ANOVA. Source data TEAD1 and YAP co-occupy distal enhancer sites to upregulate gene expression We wanted to understand the functions of TEAD1 in regions both outside and inside TEAD1 condensates. Therefore, we set out to determine the binding sites of TEAD1 on the genome and investigate their relationship to gene expression. To achieve this, we performed RNA-seq and ChIP–seq using antibodies against endogenous TEAD1, YAP and H3K27ac in TEAD1 condensate-positive (UOK121 and UOK342) and condensate-negative (UOK122 and HK2) cell lines. TEAD1 and YAP are known to co-occupy distal enhancer sites to upregulate gene expression in various stem cells and cancer cells49–51. Consistent with these results, in all cell lines, most TEAD1–YAP co-occupied peaks mapped to either distal intergenic (41.9%) or intronic regions (46.7%), whereas few (5.2%) were found at promoters (Extended Data Fig. 6a). Specifically, over 50% of TEAD1–YAP-bound regions were 50-kb away from transcription start sites (TSSs) in UOK121 and UOK342 cells (Extended Data Fig. 6b,c). Alignment of these distal TEAD1-bound sites with H3K27ac peaks revealed a bimodal distribution of H3K27ac centred on TEAD1–YAP peaks (Fig. 5a). These findings indicate that most TEAD1 binds to distal enhancer sites. These regions were also enriched for gene ontology (GO) terms such as actin cytoskeleton organization and cell junction assembly (Extended Data Fig. 6d), which are typical TEAD1–YAP-regulated pathways. As expected, TEAD1 and YAP peaks colocalized at canonical TEAD1–YAP target genes such as CTGF and CYR61 (Extended Data Fig. 6e,f). When we combined ChIP–seq and RNA-seq data and performed binding and expression target analysis (BETA)52, the results showed that most YAP–TEAD1 co-bound regions exhibited upregulated gene expression (Extended Data Fig. 6g,h). As TEAD1 is a sequence-dependent transcription factor, we performed MEME–ChIP to examine its binding motif. As expected, the TEAD-family-specific MCAT motif53 was the most enriched binding motif in regions bound by TEAD1 (Fig. 5b). TEAD1 binds to both enhancers and heterochromatin regions in the genome. a, TEAD1 peaks were mapped back to various regions of the genome. The percentage of the TEAD1 peaks found within each region of the genome is shown in the pie chart. b-c, Bar graph from GREAT analysis showing the percent distribution of TEAD1/YAP peaks in relation to the transcription start site (TSS) of genes in UOK121 (b) and UOK342 (c) cells. Distance to TSS is shown in kilobases (kb). d, Gene ontology analysis of TEAD1-bound peaks using DAVID under default parameter. GO enrichment P values were calculated in DAVID using the EASE score, a modified one-tailed Fisher exact test. Adjusted P values were calculated using the Benjamini-Hochberg method. e-f, Gene tracks showing TEAD1/YAP1 binding at regulated genes CTGF (e) and Cyr61 (f), with their corresponding mRNA expression profiles in HK2 and UOK121 cells. Black tracks are TEAD1 MACS2 peak-called regions with FDR < 0.01. g-h, Line graph showing correlation of TEAD1 ChIP-seq binding data with differential gene expression in UOK121 (g) and UOK342 (h) using Binding and Expression Target Analysis (BETA). Genes were ranked by regulatory potential score (x-axis) based on ChIP-seq peaks that are co-bound by TEAD1 and YAP. Cumulative fractions of upregulated (red), downregulated (purple), and static (black) genes were plotted (y-axis). P values were calculated using the BETA one-sided Kolmogorov Smirnov test, comparing the cumulative regulatory potential distributions of upregulated or downregulated genes with nondifferentially expressed background genes. Source data TEAD1 condensates are seeded by the repeated MCAT motifs. a, Heatmaps showing merged TEAD1 and H3K27ac ChIP–seq signals (merged data from HK2). Active enhancer regions are shown as regions that have H3K27ac peaks. b, MEME–ChIP result showing that the MCAT motif is present according to TEAD1 ChIP data from a. c, Schematic of a chromosome structure. The pericentric region is defined as 4 Mb that flank the centromere region. d, Top, gene tracks centred around the centromeric and pericentromeric regions on chromosome 4 of the hg38 reference genome. MCAT and AP-1 motif-count BED files were converted into BigWig density tracks with a 5,000 bp bin size. Bottom, a small region (chromosome 4: 49,094,512 –49,094,600 bp) is zoomed in to examine multiple MCAT sequences. Three MCAT motifs are highlighted in red. e,f, Quantification of the number of MCAT (e) and AP-1 (f) motifs per megabase in the pericentric and non-pericentric regions. Data are presented as scatterplots with mean ± s.e.m. (n = 2,827 and 277 bins for non-pericentromeric and pericentromeric regions, respectively). P values were determined by Mann–Whitney tests. g,h, ECDF (empirical cumulative distribution function) plot of the distance (kb) between each MCAT (g) and AP-1 (h) motif. i, Gene tracks showing multimapped TEAD1 (green) and H3K9me3 (red) co-binding at the MCAT-dense (grey) region of a near centromere in UOK342 cells. Blue tracks are TEAD1 SICER2 peak-called regions with FDR < 0.05. j–l, A volume of 5 µM purified, fluorescently tagged TEAD1 was mixed with the indicated concentrations of dsDNAs with scrambled sequence (pink), one MCAT repeat (lavender) or four MCAT repeats (purple). j, Representative fluorescent images of TEAD1 condensates. Scale bar, 5 mm. k, The number of condensates was quantified in each condition. Data are plotted as a superplot of 3 biological replicates, indicated by different shapes, with each data point representing a single image (n = 19, 17 and 18 images for conditions with 0, 1 and 4 MCAT repeats, respectively). Bars represent the mean ± s.d. P values were determined by one-way ANOVA. l, Schematic of TEAD1 phase separation in the presence of different DNA sequences illustrating the factors that drive condensate formation in each. TEAD1 is in green. dsDNA is coloured as in k. Source data TEAD1 condensates localize to clustered MCAT motifs at pericentromeric heterochromatin Having established that TEAD1 binds MCAT motifs to activate canonical YAP–TEAD1 target genes, we next examined where TEAD1 condensates form in the genome. We found a striking enrichment of MCAT motifs in pericentromeric regions, where these motifs were also highly clustered. This finding was consistent with our imaging data showing that TEAD1 condensates assemble at pericentromeric heterochromatin in a DNA-binding–dependent manner (Fig. 5c–e,g). By contrast, a length-matched AP-1 motif showed neither enrichment nor clustering in these regions (Fig. 5d,f,h), which indicated that this pattern is specific to TEAD1-binding sites. These results suggest that clustered MCAT motifs at pericentromeric heterochromatin provide a genomic scaffold for the nucleation of TEAD1 condensates. Consistent with this model, additional ChIP–seq analyses of H3K9me3 marks revealed that 25–35% of the total TEAD1-binding sites intersected with H3K9me3 regions. Moreover, these TEAD1–H3K9me3 co-bound regions coincided with MCAT-rich pericentromeric regions (Fig. 5i and Extended Data Fig. 7a–d). These findings indicate that the binding of TEAD1 to clustered MCAT motifs in heterochromatin drives the formation of TEAD1 condensates. ChIP-seq reveals universal binding sites of TEAD1 condensates. a, Venn diagram representing uniquely mapped TEAD1 peaks in HK2 and UOK cell lines, overlapped regions of H3K9me3. b, Venn diagram representing shared TEAD1 overlapped regions of H3K9me3 across cell lines. c-d, Gene tracks showing TEAD1/H3K9me3 in UOK121 and UOK342 cells at clusterd MCAT motif at pericentromeric region in chromosome 1 (c) and chromosome 4 (d). The grey regions are showing intersected SICER peak-called TEAD1 and H3K9me3 region. TEAD1 is mapped with multimap parameter and then normalized with counts per million (CPM). TEAD1 tracks are shown in green and H3K9me3 tracks are in red. Two biological repeats are shown. e, Purified TEAD1 for in-vitro droplet assay shown in Fig. 5j. Coomassie stained SDS-PAGE gel of final purified TEAD1 samples. Lane 1 is unlabeled TEAD1, lane 2 is Alexa Fluor 488-labeled TEAD1, and lane 3 is a 10:1 mixture of unlabeled: labeled TEAD1. f-g, MA plots showing differential gene expression from RNA-seq analysis comparing UOK121 (f) and UOK342 (g) to HK2 cells. Gene expression is plotted as log2 fold change (y-axis) versus log10 mean expression (x-axis). DESeq2 P values were calculated using a two-tailed Wald test from a negative binomial generalized linear model. Adjusted P values were calculated using the Benjamini-Hochberg method. Red dots represent significantly upregulated genes and blue represent downregulated genes (p ≤ 0.01). Non-significant genes are shown in gray. h-i, Bar graph from GREAT analysis showing the percent distribution of TEAD1/H3K9me3 peaks in relation to the transcription start site (TSS) of genes in UOK121 (h) and UOK342 (i) cells. Distance to TSS is shown in kilobases (kb).j-k, Line graph showing correlation of TEAD1 ChIP-seq binding data with differential gene expression in UOK121 (j) and UOK342 (k) using BETA. Genes were ranked by regulatory potential score (x-axis) based on ChIP-seq peaks that are co-bound by TEAD1 and H3K9me3. Cumulative fractions of upregulated (red), downregulated (purple), and static (black) genes were plotted (y-axis). P values were calculated using the BETA one-sided Kolmogorov Smirnov test, comparing the cumulative regulatory potential distributions of upregulated or downregulated genes with nondifferentially expressed background genes. Source data To define the mechanism of TEAD1 condensate formation, we investigated whether and how TEAD1 interactions with DNA altered TEAD1 solubility and condensate formation. We monitored condensate formation of full-length, purified TEAD1 (Extended Data Fig. 7e) incubated with a panel of double-stranded DNA (dsDNA) containing differing numbers of MCAT motifs: 0, 1 or 4 tandem repeats (denoted 0×MCAT, 1×MCAT and 4×MCAT, respectively) (Fig. 5j,k). Although TEAD1 formed condensates in the presence of all the DNA variants screened, the number of condensates differed, which suggested that interactions with DNA influence TEAD1 solubility. TEAD1 formed the most condensates in the presence of 4×MCAT DNA (for example, 3.4-fold higher than 0×MCAT and 13.7-fold higher than 1×MCAT). In the presence of a single MCAT motif, TEAD1 formed fewer condensates than either 4×MCAT or 0×MCAT. This altered solubility profile supports a model in which TEAD1 condensation is less likely in conditions that support weak or no DNA binding (for example, 0×MCAT or 1×MCAT), driven mostly by weak TEAD1 self-associations. And in conditions with high-affinity DNA binding (for example, 4×MCAT or pericentromeric regions), condensation is further enhanced by both multivalent interactions (both protein–protein and protein–DNA) and an increase in the effective local concentration of TEAD1 (refs. 45,54) (Fig. 5l). These cooperative effects drive phase separation of TEAD1–dsDNA complexes beyond the intrinsic phase separation capacity of TEAD1. The effect of multivalent DNA interactions provides a biophysical rationale for the formation of LTCs in pericentromeric regions. TEAD1 condensates serve as depots to buffer excess TEAD1 To further evaluate whether TEAD1 condensates are associated with transcriptional regulation, we performed differential expression analysis of RNA-seq data and compared condensate-positive kidney cancer cell lines (UOK121 and UOK342) with the condensate-negative HK2 kidney cell line55 (Extended Data Fig. 7f,g). We next used BETA to determine whether differentially expressed genes were associated with TEAD1–H3K9me3 co-bound genomic regions. TEAD1–H3K9me3 co-binding sites were over 500-kb away from the nearest TSS (Extended Data Fig. 7h,i), and genes proximal to these sites did not show significant enrichment among either upregulated or downregulated transcripts (Extended Data Fig. 7j,k). Consistent with this lack of transcriptional association, increasing the number of TEAD1 condensates from zero to three did not alter overall nuclear TEAD1 abundance or expression of the canonical TEAD target gene CYR61 (Fig. 6a–d). Together with the observation that LTCs are spatially separated from active transcription markers and transcription sites (Fig. 2), these data suggest that pericentromeric TEAD1 condensates are unlikely to directly drive transcription and may instead function as nuclear depots for TEAD1 in kidney cancer cells. TEAD1 condensates are depots for buffering target gene expression. a–d, Quantification of CYR61 nascent RNA FISH with TEAD1 immunofluorescence 3D image analysis. Cells expressing similar levels of TEAD1 form 1–3 LTCs (a). CYR61 RNA FISH signal count (b), mean intensity (c) and sum intensity (d) in the LTC = 1, 2 or 3 group are compared with the LTC = 0 group. Data are the mean ± s.e.m. (n = 8, 7, 9 and 6 cells for LTC = 0, 1, 2 and 3, respectively). P values were determined by one-way ANOVA using Dunnett’s multiple-comparisons tests. e, Representative images of UOK342 HaloTag-TEAD1 knock-in cells labelled with JFX554 HaloTag ligand for single-particle tracking. The whole nucleus was imaged using epifluorescence microscopy to generate a mask of the TEAD1 condensate (top). HILO microscopy was used for single-particle tracking of individual HaloTag-TEAD1 molecules (bottom). Scale bars, 5 μm. f, Trajectories of individual TEAD1 molecules were categorized as inside TEAD1 condensates (top) and/or in the nucleoplasm (bottom). To obtain trajectories located in the nucleoplasm, a binary mask of the TEAD1 condensate generated from e was randomly rotated and placed elsewhere in the nucleus. Individual trajectories present either in the condensate mask or in the masked area placed randomly in the nucleoplasm are shown, with each colour representing a different trajectory. Scale bars, 5 μm. g,h, Quantification of dissociation constants (g) and mean residence time (h) of individual TEAD1 molecules inside hubs and those outside hubs in the nucleoplasm. Points represent trajectories grouped by replicate (n = 4 independent experiments with 8 cells per replicate). Data are the mean ± s.e.m. Significance was determined by paired t-tests, one-tailed. i, Representative live-cell images of UOK342 HaloTag-TEAD1 knock-in cells labelled with JFX554 HaloTag ligand and treated with sorbitol 0.2 M, NaCl 0.2 M, 50% water or control. Scale bars, 5 μm. j, Quantification of LTC count per cell after treatment. Data are the mean ± s.e.m. (n = 21, 20, 26 and 20 cells for control, 0.2 M, NaCl 0.2 M and 50% water treatment, respectively). k, Quantification of CYR61 expression after 0.2 M sorbitol treatment for 3 h in UOK342 HaloTag-TEAD1 knock-in cells compared with control. Data are presented as scatter plots with mean ± s.d. (n = 3 biological replicates). Significance was determined by unpaired t-tests, two-tailed. l, Quantification of CYR61 expression after 0.2 M sorbitol treatment for 3 h in LTC-negative cell lines (HK2 and UOK122) and in LTC-positive cell lines (UOK121 and UOK342). Data are the mean ± s.e.m. (n = 3 biological replicates). Significance was determined by multiple unpaired t-tests. For multiple comparisons, the FDR was calculated using two-stage step-up (Benjamini, Krieger and Yekutieli) with the desired FDR (q) set at 1.00%; exact q values are provided on the graph. m, Per cent increased CYR61 expression after sorbitol treatment relative to control of each cell line in l. Data are the mean ± s.e.m. (n = 3 biological replicates). Significance was determined by one-way ANOVA. Source data In Hippo-mutated cancer cells, depots may buffer TEAD1 overexpression so that downstream gene expression remains stable. Indeed, single-particle tracking experiments showed that TEAD1 dissociated later and stayed longer in LTCs than in the nucleoplasm (Fig. 6e–h). This result suggests that LTCs trap TEAD1 proteins within its structure. To definitively show that LTCs serve as depots, we sought for ways to disrupt TEAD1 condensates without mutating the TEAD1 DNA-binding domain and measured the consequent effects on TEAD1 target gene expression. Hyperosmotic stress mediated by sorbitol or NaCl rapidly dissociated LTCs, whereas hypo-osmotic stress did not (Fig. 6i,j). Consistent with LTCs serving as depots in TEAD1 condensate-positive cells (UOK121 and UOK342), their dissociation and thereby the release of TEAD1 to the nucleoplasm under hyperosmotic stress were accompanied by an increase in TEAD1 target gene expression. By contrast, this effect was not observed in condensate-negative cells (HK2 and UOK122; Fig. 6k–m). In summary, LTCs are seeded specifically in the pericentromeric region by the clustered MCAT sequences, which serve important roles to buffer downstream gene expression. TEAD1 condensates selectively incorporate protein components To further understand the functions of TEAD1 condensates, we set out to identify protein components of TEAD1 condensates using proximity labelling mass spectrometry56,57 with UOK342 cells stably expressing miniTurbo-TEAD1-HaloTag (WT hereafter; Fig. 7a). In this system, miniTurbo is a compact biotin ligase that rapidly labels TEAD1-surrounding proteins with biotin56, and HaloTag is used for visualizing TEAD1 localization with Halo dyes. We used two additional cell lines as controls: UOK342 cells expressing miniTurbo-TEAD1(H3Mut)-HaloTag (H3Mut hereafter), which does not form condensates; and UOK342 cells expressing miniTurbo-HaloTag-NLS (NLS hereafter), which localizes diffusely to the nucleus using a nuclear localization sequence (Fig. 7a,b). We first used imaging to confirm efficient biotinylation with miniTurbo after the addition of exogenous biotin in all three cell lines (Fig. 7b). We then enriched the biotinylated proteins from whole-cell lysate using streptavidin beads and confirmed that protein biotinylation increased after the addition of biotin and that all three constructs were expressed at similar levels (Extended Data Fig. 8a,b). Using mass spectrometry with tandem mass tag (TMT) labelling, we identified a total of 7,580 proteins in the WT, H3Mut and NLS samples combined (false discovery rate (FDR) < 0.01; Supplementary Table. 4). Comparing the combined dataset of WT samples and H3Mut samples with the NLS control, we first identified nonspecific nuclear proteins (Fig. 7c, blue). Then, we compared the differential enrichment of proteins between the WT and H3Mut groups and identified 186 proteins enriched in the WT samples and 409 proteins enriched in the H3Mut samples (Fig. 7d). Among the 186 proteins enriched in the WT samples, 0% were nonspecific nuclear proteins identified in Fig. 7c, whereas this percentage increased to 56% in proteins enriched in the H3Mut samples. These results demonstrate the specificity of our mass spectrometry approach for detecting TEAD1 condensate components (Fig. 7e). GO terms of the unique proteins in the WT samples revealed nuclear proteins associated with DNA binding and transcriptional regulation, consistent with the nuclear functions of TEAD1 (Fig. 7f and Extended Data Fig. 8c). Using co-immunofluorescence, we confirmed the colocalization of endogenous replication protein A2 (RPA2, part of the trimeric RPA complex for maintaining genome stability) with TEAD1 condensates in UOK121 and UOK342 cells (Fig. 7g,h). RPA2 is not essential for TEAD1 condensate formation, as knocking it down did not affect TEAD1 condensate number or size (Fig. 7i,j and Extended Data Fig. 8d–g). Instead, RPA2 may be a client that partitions into TEAD1 condensates, as the fluorescence of RPA2 recovered much faster than TEAD1 after photobleaching58,59 (Fig. 7k). These findings indicate that in addition to storing TEAD1, TEAD1 condensates may selectively incorporate proteins to maintain chromatin structure. Identification of protein components of TEAD1 condensates by proximity-based mass spectrometry. a, Schematic overview of the proximity labelling proteomic workflow used to identify TEAD1-associated proteins. Cells expressing miniTurbo-TEAD1-HaloTag (WT), miniTurbo-TEAD(H3Mut)-HaloTag (H3Mut) or miniTurbo-NLS-HaloTag (NLS) constructs were incubated with biotin for 10 min to induce biotinylation, followed by protein extraction, streptavidin-based enrichment, on-bead digestion and TMT labelling and pooling. The multiplexed samples were fractionated and analysed by liquid chromatography and tandem mass spectrometry (LC–MS/MS) to compare specific TEAD1-dependent and nonspecific biotinylation profiles. b, Representative immunofluorescence images of UOK342 stably expressing the WT, H3Mut or NLS construct with or without biotin treatment showing biotinylation (red) in relation to HaloTag (magenta), TEAD1 (green) and nuclei (blue). Scale bars, 5 μm. c, Volcano plot of the biotinylated proteins identified from cells expressing the WT or H3Mut construct, together with the NLS control representing nonspecific biotinylation. The x axis represents the fold change comparing the combined WT and H3Mut conditions against the NLS control on a log2 scale, and the y axis represents the P value of the statistical analysis on a –log10 scale (n = 3 biological replicates). The curved line indicates the boundary corresponding to a q value of 0.05. Proteins with q < 0.05 and upregulated in the NLS condition are shown in blue. d, Volcano plot comparing biotinylated proteins between cells expressing the WT or H3Mut construct. The x axis represents the log2 fold change (WT versus H3Mut), and the y axis represents the –log10[P]. The curved line indicates the threshold corresponding to a q value of 0.05. Red symbols (two-sided t-test, q < 0.05) denote proteins significantly enriched in the WT condition, whereas blue symbols indicate proteins belonging to the nonspecific biotinylation group identified in c. e, Evaluation of nonspecific protein biotinylation in the H3Mut and WT conditions. Among the significantly enriched proteins with a q value less than 0.05, nonspecific binding resulting from nonspecific protein biotinylation was evaluated. Donut charts show the number and proportion of proteins upregulated in the H3Mut and WT condition, respectively. Proteins were categorized as specific or nonspecific based on nonspecific biotinylation, and inner labels indicate the total number of proteins and the percentage of each category. f, GO enrichment analysis of proteins significantly enriched under the WT condition with a q value less than 0.05, as shown in red in c. Enriched GO terms were categorized into molecular function (MF, red), biological process (BP, green) and cellular component (CC, yellow). The top five significantly enriched GO terms for each category are displayed based on P values, and the circle size represents the number of proteins enriched in each term. g, Representative Immunofluorescence images of endogenous RPA2 and TEAD1 in UOK121 and UOK342 cells. Scale bars, 5 μm. Insets show magnified images of the boxed regions. h, Quantification of colocalization between RPA2 and TEAD1 at LTCs compared with control regions in the nucleoplasm shown in g (n indicated in the figure equals the number of regions centred at LTCs). Significance was determined by one-way ANOVA. i,j, Quantification of the percentage of LTC-positive cells (i) and LTC area (j) after RPA2 knockdown via short-interfering RNA (siRNA) in UOK121 and UOK342 cells compared with control (NC) (n = 3 biological replicates for each cell line). Significance was determined by one-way ANOVA. k, Quantification of recovered fluorescence intensity of TEAD1 and RPA2 at LTCs in UOK342 HaloTag-TEAD1 knock-in cells. Shaded areas indicate the s.e.m. for each measurement (n = 11 technical replicates). Significance was determined by one-way ANOVA. mCer, mCerulean. For the box plots in h and j, the centre line shows the median, the bounds of the box correspond to the interquartile (25–75th) percentile, and whiskers extend to the farthest data points within 1.5× the interquartile range from the hinges; the dots beyond the whiskers show Tukey’s fences outliers. Schematic in a created in BioRender; Na, C. https://biorender.com/gbl7pb4 (2026). Source data Identification and validation of protein components in TEAD1 condensate. a, Western blot showing enriched biotinylated proteins (Streptavidin IRDye680) with or without biotin treatment of UOK342 cells expressing miniTurbo-TEAD1-HaloTag. b, Western blot comparing biotinylated protein enrichments of miniTurbo-TEAD1 WT-, H3Mut-, and NLS-HaloTag stably expressed in UOK342 cells upon biotin treatment. c, Interactome analysis of proteins significantly enriched in the TEAD1 WT condition (red group) using STRING protein–protein interaction (PPI) analysis. The analysis was performed with high confidence (interaction score threshold of 0.7). The resulting network contained 56 nodes and 105 edges, with an average node degree of 1.47 and an average local clustering coefficient of 0.381. The expected number of edges was 51, and the observed network showed significantly more interactions than expected (PPI enrichment p < 1.0 × 10⁻16). Protein–protein interaction analysis was conducted using STRING based on the evidence types of text mining (yellow edges), experiments (purple edges), databases (blue edges), and co-expression (black edges). Table on right summarized top two functional enrichments in the network for each gene ontology term Biological Process (BP), Cellular Component (CC), Molecular Function (MF). False discovery rate shown are p values corrected for multiple testing within each category using the Benjamini-Hochberg procedure. d-f, Immunoblotting of efficient RPA2 knockdown in UOK121 and UOK342 cells in (d). Quantification of RPA2 expression (e) and TEAD1 expression (f) between TEAD1 siRNA treatment and negative control (NC) in UOK121 and UOK342, data are shown as mean ± SEM (n = 3 biological replicates). Statistical significance determined by one-way ANOVA. g, Immunofluorescence images of TEAD1 (green) and RPA2 (red) in UOK121 and UOK342 cells upon treatment with RPA2 knockdown siRNA (RPA2) or control siRNA (NC). Scale bars = 5 μm. Source data Discussion In this study, we made an interesting discovery that TEAD1 forms two types of condensates that differ in both size and function: (1) small, transcriptionally active TEAD1 condensates that are the canonical condensates that colocalize with YAP; and (2) large, transcriptionally silent TEAD1 condensates that localize to heterochromatin and function as depots to store excess TEAD1 protein at a relatively harmless genomic location. Cells need to maintain optimal concentrations of proteins for homeostasis. Too little protein may delay entry into mitosis60,61, whereas too much may promote protein aggregation and/or hyperactivation of downstream signalling62,63. Moreover, aberrant activation of oncogenes in non-transformed cells often induces oncogene-induced senescence involving cell cycle arrest64–66. Multiple cellular processes contribute to protein homeostasis, including regulation of transcription, translation and protein degradation systems. We propose that protein condensation of TEAD1 represents another way to regulate TEAD1 concentrations and is used to both buffer and limit the damaging effects caused by excess TEAD1 that may confer survival advantages to RCC cells. In RCC cells, YAP–TAZ and TEAD1 can activate genes that mediate ferroptosis, an iron-dependent cell death pathway67,68. Preventing excessive TEAD1 activity by forming inactive TEAD1 condensates could represent a previously unknown way to protect cancer cells from ferroptosis. However, not all phenomena in cancer cells promote cancer. TEAD1 condensates may instead represent a futile attempt of these cells to prevent tumorigenesis by decreasing YAP–TEAD-mediated cell proliferation. Although we successfully used hyperosmotic stress to disrupt TEAD1 condensates, future studies will investigate physiological modulation of TEAD1 condensates to examine their roles in cellular fitness and cancer progression. Intriguingly, LTCs do not form at random locations in the nucleus but instead localize to pericentromeric heterochromatin regions. These areas have long been considered transcriptionally inactive domains but are critical for maintaining centromere integrity and proper segregation of chromosomes during mitosis42,43. Although most transcription factors are absent from those areas, a few can be found there. Heat shock transcription factor 1 (HSF1) binds to the pericentromeric region of human chromosome 9 during heat shock, transcribes satellite III repeats and forms nuclear stress granules containing HSF1 (refs. 69,70). These HSF1 granules are thought to sequester active transcription complexes, such as mammalian CREB-binding protein, to shut down global transcription during heat shock71. Another transcription factor that localizes to the pericentromeric region is GAGA factor, which binds to GA/CT repeats and establishes chromatin boundaries in cooperation with Polycomb proteins and the inner nuclear membrane72,73. Ikaros is a protein that localizes to the centromeric heterochromatin of lymphocytes, forms condensates and represses nearby genes74. Here we propose a novel role of the pericentromeric region, which is to serve as a reservoir for excess TEAD1 in a DNA-binding-dependent manner. Indeed, when we overexpressed TEAD1 in other condensate-negative cell lines, we observed that TEAD1 condensates formed at heterochromatin and were likely to be pericentromeric (Extended Data Fig. 5h,l,m), which indicates that this is a general mechanism used by cells to store TEAD1. This result raises the question of whether other transcription factors can localize specifically to pericentromeric or other heterochromatic regions and form condensates. If so, the functions of these condensates will be of interest. Recently, condensate-targeting therapies have emerged as a promising way to treat previously untreatable diseases such as neurodegeneration and cancer15,75–77. Our data on TEAD1 condensates in RCC reveal that not all the condensates observed in disease may drive that disease. Furthermore, a single transcription factor can also form condensates of different sizes and functions in different cell types15. Therefore, careful studies of condensate morphology and function are paramount for designing precise condensate-targeting treatments. Methods Cell culture, transfection and siRNA The patient-derived immortalized cell lines UOK121, UOK139, UOK140 and UOK275 cells were a gift from W. Marston Linehan.The HK2 line is from the American Type Culture Collection (ATCC, CRL-2190). Cells were cultured at 37 °C and 5% CO2 in DMEM supplemented with 10% fetal bovine serum (FBS; Gibco, 26140079), 100 U ml−1 (1%) penicillin–streptomycin (Gibco, 15140122), 2 mM (1%) GlutaMAX-l (Gibco, 35050061) and 1× NEAA (Gibco, 11140050). The cell lines 1205Lu, WM983B, FS4 (ref. 78) and AG13004 (Coriell Institute) were a gift from A. Weeraratna. 1205Lu, WM983B and AG13004 cells were cultured in DMEM (Gibco, 12430-047) supplemented with 10% FBS. FS4 cells were cultured in RPMI (Gibco, 11875-085) and supplemented with 10% FBS and 1% GlutaMAX (Gibco, 35050-061). T47D, MCF-7, MDA-MB-231 and MCF-10A cells were a gift from B. Jenkins-Lord. T47D cells were cultured in RPMI-1640 (ATCC, 30-2001) with 10% FBS (Gibco, A52558-01) and 10 µg ml−1 insulin (Lonza, BE02-033E20). MCF-7 cells were cultured in EMEM (ATCC, 30-2003) supplemented with 10% FBS and 10 µg ml−1 insulin, and MDA-MB-231 cells were cultured in DMEM (ATCC, 30-2002) supplemented with 10% FBS. MCF-10A cells were cultured in DMEM-F12 (ATCC, 30-2006) supplemented with 5% horse serum (Millipore, H1138), 20 ng ml−1 EGF (Rockland, 009-001-C26), 0.5 mg ml−1 hydrocortisone (VWR, 1315004), 100 ng ml−1 cholera toxin (Millipore, 9012-63-9), 10 µg ml−1 insulin and 1% penicillin–streptomycin (VWR, 10220-718). MCF-7 cells are on the frequently misidentified cell list but are necessary for TEAD1 validation experiments in other cancer cell lines. They were authenticated via STR by the ATCC. The UOK and HK2 cell lines used in this study have been authenticated through the Johns Hopkins University GRCF facility (GeneSifter). 1205Lu and WM983B cells gifted from the Weeraratna Laboratory were authenticated at the Johns Hopkins University GRCF. Jenkins–Lord Laboratory cells were authenticated via STR by the ATCC. For overexpression experiments, UOK and HK2 cells were transfected with eGFP–TEAD1(WT), DNA-binding mutants and other eGFP–TEAD1 variants using Lipofectamine 3000 transfection reagent (Thermo Fisher, L3000015) for 16 h. For silencing of TEAD1 experiments, TEAD1-targeting siRNA (Thermo Fisher, Silencer Select s13962), RPA2-targeting siRNA (Horizon Discovery, J-017058-09-0002) or scrambled negative control siRNA (Thermo Fisher, AM4611) was used at a final concentration of 10 nM and was transfected into cells using Lipofectamine RNAiMAX transfection reagent (Thermo Fisher, 13778075) for 48 h, after which cells were replated for western blotting and imaging. Cell line generation UOK342 HaloTag-TEAD1 knock-in cell line generation Endogenously tagged UOK342 Halo-Tag knock-in cells were engineered using the CRISPR–Cas9 system. pTEAD1-gDNA targeting the N terminus of TEAD1 was assembled into a PX458 template plasmid (Addgene, 48138). pTEAD1-donor was assembled into pUC57mini (Addgene, 88845) using a gBlock containing a HaloTag sequence and 300 bp homology arms of the gDNA target site. UOK342 cells were transfected with Lipofectamine 3000 (L3000015, Thermo Fisher) following the manufacture’s protocol. The cells were cultured for >14 days. Cells were stained for 30 min with JF549 HaloTag ligand (100 nM, Janelia Materials) and sorted by fluorescence-activated cell sorting (FACS) and expanded. For cell line validation, the gDNA from the selected clone was extracted using a GeneJET Genomic DNA Purification kit (K0721, Thermo Scientific). The PCR product of the edited region was amplified using Q5 High-Fidelity 2×901 master mix (M0492, NEB) and verified using Sanger sequencing (see Supplementary Table 1 for the sequence). miniTurbo plasmid construct and stable cell line generation miniTurbo-TEAD1-HaloTag, miniTurbo-TEAD1(H3Mut)-HaloTag and miniTurbo-HaloTag-NLS expression plasmids were generated by high-fidelity of the PCR product amplified from the eGFP–TEAD1 vector and 3×HA-miniTurbo-NLS_pCDNA3 (Addgene, Plasmid 107172) (see Supplementary Table 1 for sequences). Plasmids were validated by whole-plasmid sequencing (Plasmidsaurus). For cell line generation, miniTurbo-TEAD1-HaloTag and miniTurbo-TEAD1(H3Mut)-HaloTag plasmids were transfected in UOK342 cells using Lipofectamine 3000 transfection reagent (L3000015, Thermo Fisher) and passaged for >14 days. Cells were stained for 30 min with JF549 HaloTag ligand (100 nM, Janelia Materials). Cells with stable expression were sorted by FACS and expanded. Expression of the fusion proteins was validated by western blotting. Immunofluorescence staining After transfection, UOK and HK2 cells were plated on coverslips pre-coated with fibronectin (7.5 μg ml−1; Millipore, FC010). Cells were grown for 16 h and fixed with 4% formaldehyde (EMS), permeabilized with 0.1% Triton X-100 and blocked with 3% BSA in 1× PBS. The cells were then incubated overnight with primary antibodies in 1% BSA at 4 °C and incubated with Alexa Fluor-conjugated secondary antibodies in 1% BSA for 1 h at room temperature (Supplementary Table 2). For imaging and quantification, at least 20 fields of view per coverslip were randomly chosen on the basis of Hoechst nuclear staining and imaged using a Zeiss LSM900 Airyscan microscope, followed by Airyscan processing (2D, default settings). Domain deletion and site-directed mutagenesis The plasmid for eGFP–TEAD1 was first digested using BamHI and ApaI to create the backbone. This step was followed by the amplification of two specific DNA fragments from the parental eGFP–TEAD1 plasmid. To generate eGFP–TEAD1(Δ55–122), primer set TEAD1-1 was used for the first fragment and primer set TEAD1-2 was used for the second fragment (see Supplementary Table 1 for sequences). To generate the eGFP–TEAD1(H3Mut) variant, the process began with BamHI digestion of eGFP–TEAD1 followed by PCR to generate two segments using primer sets TEAD1-3 and TEAD1-4. To generate eGFP–TEAD1(H1mut), PCR amplification was performed to produce three distinct fragments. The first fragment used primers TEAD1-3f and TEAD1-5r, the second used TEAD1-5f and TEAD1-4r and the third fragment was a previously prepared backbone. These PCR fragments were subsequently assembled into the final construct through the HIFI assembly method. For the creation of eGFP–TEAD(1ΔIDR), PCR amplification was performed using primer pairs, TEAD1(ΔIDR) 5′ of the PCR product was phosphorylated using T4 polynucleotide kinase (NEB, M0201S), parental plasmids were digested using DpnI restriction enzyme (NEB, R0176S), then the PCR product was ligated into plasmids using T4 DNA ligase (NEB, M0202S). For generation of the eGFP–TEAD1(Y406H), eGFP–TEAD1(S102A) and eGFP–TEAD1(S102D) mutant constructs, PCR amplification was performed using the parental eGFP–TEAD1 plasmids and the primer pairs for Y406H, S102A and S102D, respectively (Supplementary Table 1). DNA FISH and immunofluorescence staining of TEAD1 condensates UOK and HK2 cells were plated on no. 1.5H high-performance coverslips (Marienfeld Superior) pre-coated with fibronectin (7.5 μg ml−1; Millipore, FC010). Cells were grown overnight, washed with 1× PBS and fixed with 4% methanol-free formaldehyde (Cell Signaling, 12606) with 1× PBS for 10 min. Coverslips were washed in 1× PBS and 0.002% Tween (PBS-T) 5 times using a previously described transfer method79, and permeabilized with 0.2% Triton-X-100 for 10 min before blocking in MaxBlock (Active Motif, 15252) for 1 h. Coverslips were incubated overnight in primary antibody (H3K9me3, Active Motif, 39161, 1:500; TEAD1, BD, 610922, 1:200) in MaxBlock and washed 5 times in PBS-T. Secondary antibodies (AF647 1:500 and AF594 1:1,000) were incubated in MaxBlock for 1 h in the dark and washed 5 times in PBS-T before post-fixation in 4% methanol-free formaldehyde for 10 min. Subsequently, FISH hybridization for centromere probes was performed as previously described80. AF488-labelled probes (gifts from A. Meeker) suspended in hybridization buffer were denatured at 84 °C for 5 min before hybridization at room temperature for 2 h. Coverslips were then washed twice in PNA buffer for 15 min at room temperature, then washed twice in PBS-T and counterstained with Hoescht 33342 at 1:5,000, before washing twice in PBS-T then once in ddH2O. Pre-equilibration and mounting in Vectashield were performed before proceeding to Airyscan imaging. Intron RNA FISH combined with immunofluorescence Human CYR61 intron with Quasar 570 dye (Biosearch Technologies, ISMF-2066-5), human ACTB intron with Quasar 570 dye (Biosearch Technologies, ISMF-2002-5), Stellaris RNA FISH hybridization buffer (Biosearch Technologies, SMF-HB1-10), Stellaris RNA FISH wash buffer A (Biosearch Technologies, SMF-WA1-60) and wash buffer B (Biosearch Technologies, SMF-WB1-20) were purchased from Biosearch Technologies. We followed the protocol for sequential immunofluorescence and FISH in adherent cells listed on the Biosearch Technologies website under Stellaris RNA FISH protocols. Nascent RNA labelling assay UOK121 and UOK342 cells were cultured on coverslips and grown for 24 h. Cells were treated with 1 mM EU or DMSO vehicle control for 10 min. Nascent RNA labelling assays were conducted using a Click-It RNA Alexa Fluor 594 Imaging kit (Invitrogen, C10330) following the manufacturer’s protocol. Immunofluorescence experiments against TEAD1 and nucleolin were conducted after cells were washed with Click-It rinse buffer. Images were captured with a Zeiss LSM900 Airyscan microscope and colocalization analysis was quantified using ImageJ software. Image analyses TEAD1 condensate nuclear localization quantification Immunofluorescence staining of UOK121 and UOK342 cell lines was performed. 3D images were acquired using a Zeiss LSM 900 with an Airyscan 2 detector. Z stacks were captured at 0.15 µm intervals and 3D Airyscan processing was applied. TEAD1 condensate localization was analysed in 3D using Zeiss Zen software. Nucleoli were identified on the basis of nucleolin staining, and perinuclear regions were defined as areas within 1 µm of the nuclear envelope, as delineated by Hoechst staining. TEAD1 condensates were considered colocalized with the nucleolus if they overlapped with nucleolin-positive regions in 3D space. Similarly, TEAD1 condensates were classified as perinuclear if they were located within 1 µm of the nuclear envelope. Colocalization was visually confirmed in all three dimensions. The percentage of TEAD1 condensates localized to the nucleolus, perinuclear regions, both or neither was calculated for each cell line. For representative images, maximum projections of the z stack were generated in Fiji. TEAD1 condensate quantification Images were taken of fixed cells stained with Hoechst dye and TEAD antibody. Cellprofiler (v.4.2.8) was used to identify both cell counts and the appearance of TEAD condensates. Using Hoechst as a nuclear marker, we used Cellprofiler to identify the nuclei as objects. Using the TEAD antibody, Cellprofiler thresholded the TEAD condensates on the basis of their size and fluorescence intensity. The objects were manually screened for false positives, and any object that had been thresholded inaccurately was removed from the dataset before final quantification. To determine the appearance rate of TEAD condensates, we divided the number of cells that positively generated at least one TEAD condensate by the total number of nuclei detected. Colocalization analysis A square region measuring 3 μm by 3 μm was delineated around the focal point of TEAD1 condensates. A negative control square was also randomly selected in the nucleoplasm where TEAD1 staining appeared diffused. The colocalization analysis was conducted using Fiji software81 with the BIOP-JACoP plugin82. Statistical evaluation was performed using unpaired t-tests in GraphPad prism. To calculate the enrichment of TEAD1 and YAP signals at CYR61 intron RNA FISH images, we used a 2.12 × 2.12 μm region of interest (ROI) box to centre around RNA FISH signals in 2D immunofluorescence images using ImageJ and duplicated TEAD, YAP and CYR61 channels. The duplicated smaller images were then combined into stacks of each channel, and the averaged images are derived using the Z-project-> Average function. Enrichment of averaged signals across the centre point was generated using the Plot Profile function in ImageJ. Similar analyses were done except for centring at the LTC or centromeric DNA FISH signals. 3D distance analysis Immunofluorescence staining of HK2 and UOK cell lines was performed. The imaging was conducted using Airyscan z stacks with a z interval of 0.13 μm. 3D distance analysis of TEAD1 condensates with CENP-A or TRF2 was performed using the ImageJ plugin Distance Analysis (DiAna)83. Specific thresholds were applied to different immunofluorescence stains in the DiAna-segment module: a threshold of 5,347 with a size larger than 1 pixel for CENP-A; 2,250 with a size larger than 1,000 pixels for TEAD1 condensates; and 8,405 with a size larger than 30 pixels for TRF2. Using the DiAna-analysis module, the minimal distance of each TEAD1 condensate to CENP-A or TRF2 was determined. Distance histograms, visualizations and statistical analyses were conducted using GraphPad Prism. Imaris 3D object surface classification and analysis Nucleus, STC, LTC, YAP condensate (YAP) and CYR61 intron RNA FISH signals were identified using the surface detection module using Imaris (v.10.2.0, Bitplane) with the absolute threshold of different channels manually set the same across images. STC and YAP surfaces were filtered on the basis of the number of voxels being 35–1,000, and LTC surfaces were filtered on the basis of the number of voxels being above 3,000. Surface-to-surface distance analysis was performed by enabling the shortest distance algorithm (Imaris). TEAD1 nuclear intensity was measured using a nuclear 3D mask generated with the Hoechst channel. CYR61 RNA FISH count and intensity were collected from averaged and specific values of build-in surface statistics. Live-cell imaging, drug treatment and FRAP UOK342 HaloTag-TEAD1 CRISPR knock-in cells were plated into a 24-well glass-bottom plate (Cellvis, P96-1.5H-N) for drug treatment and imaging the following day. Before drug treatment, the cells were labelled with 10 nM JFX554 HaloTag ligand for 30 min. The medium was replaced with FluoroBrite DMEM complete medium (Gibco, A1896701) supplemented with 10% FBS (Gibco, 26140079), 2 mM GlutaMAX-l (Gibco, 35050061) and 1× NEAA (Gibco, 11140050). Live-cell images of individual cells were taken before and after treatment with 1% 1,6HD (Sigma-Aldrich, 88571) or DMSO (Corning, 01023017) for 30 min with a 2-min interval. UOK342 cells stably expressing eGFP–TEAD1 were plated into an 8-well chambered cover glass plate (Cellvis, C8-1.5H-N) for FRAP experiments. Fluorescence bleaching was done on nuclear TEAD1 condensates using bleach mode in Zen software. The ROI was selected on either the entire condensate or part of it using a rectangular box of approximately 1 × 1 μm in half-FRAP and 2 × 2 μm in full-FRAP. At least 20 iterations of bleaching were done with a 488 nm Argon laser at 100% power. Around 140 rounds of imaging were performed before and after bleaching until the fluorescence signal plateaued, with an interval of 410 ms. FRAP analysis was performed as described in previous studies84. In brief, for each time point, raw fluorescence intensity (I) values from the bleached ROI, an unbleached ROI, a background ROI and a reference region were measured. For full-bleach FRAP, bleached intensities (FRAPFull bleach) were first computed by standard normalization: FRAPFull bleach = (I − IBackground)/(IReference − IBackground). For half-bleach FRAP, bleach (IB) and non-bleached (INB) intensities were further normalized in 4 steps sequentially (I-IV) to correct for an artefactual intensity drop in the non-bleached trace immediately after bleaching. Both traces were offset by a constant equal to the difference between the first two post-bleach non-bleach points (II·B and II·NB), which produced adjusted signals. Signals were then scaled by the relative ROI areas to account for differing region sizes, III⋅B/NB=II⋅B/NBArea1Area1+Area2. Next, traces were normalized to the pre-bleach and immediate post-bleach levels to obtain IIII·B/NB. The immobile fraction was estimated from the terminal plateau by averaging the last three time points and comparing the non-bleached and bleached signals, and the recovery curve was corrected accordingly to generate the final normalized non-bleach trace and bleached trace (IIV·B/NB). The t1/2 was calculated using frapplot85, with fit formula for one diffusion component. Hyperosmotic and hypo-osmotic stress treatments UOK342 HaloTag-TEAD1 knock-in cells were plated into an 8-well chambered cover glass plate 48 h before treatments and replaced with FluoroBrite DMEM complete medium (Gibco, A1896701) with no FBS 1 h before imaging. Cells were exposed to hyperosmotic stress by adding 10× filter-sterilized 2 M D-sorbitol (Sigma, S1876) and NaCl (RPI, 7647-14-5) dissolved in FluoroBrite DMEM medium to a final concentration or 0.2 M or vehicle control. Hypo-osmotic stress was induced by adding 50% v/v filter-sterilized molecular-biology-grade water (Corning, 46-000-cv). Cells were imaged before and after treatment for 30 min with a 2-min interval. To collect RNA from UOK342 TEAD1–HaloTag CRISPR knock-in cells, parental UOK cell lines and HK2 cells, cells were cultured similarly and collected 3 h after 0.2 M sorbitol treatment or vehicle control. Western blotting Cells were seeded on 6-well culture plates (Falcon, 353046), and plasmid transfection using Lipofectamine 3000 was done the day before cell lysis. After 20 h of transfection, cells were washed in ice-cold PBS, then 100 µl lysis buffer (50 mM Tris-HCl pH 8, 100 nM NaCl, 1% Igepal and 0.5% sodium deoxycholate) with 1× protease inhibitor (complete, mini, EDTA-free protease inhibitor cocktail) was added. Cells were lysed on ice for 20 min and supernatant was collected after centrifugation at 21,000g for 30 min at 4 °C. Cell lysate was later added 1× loading buffer (10% glycerol, 2% SDS, 0.05 M Tris-HCl pH 6.4, 0.01 M 2-mercaptoethanol, 0.6% bromophenol blue), and denatured at 90 °C for 10 mins. A 10 µl mixture was then added to SurePAGETM 4–12% Bis-Tris gels (GenScript, M00654). Protein was transferred to a nitrocellulose membrane (Licor, P/N 926-31090) by wet transfer and blocked with 5% non-fat milk in TBS buffer. Membranes were incubated overnight at 4 °C with 5% non-fat milk in TBS-T with primary antibodies (Supplementary Table 2). After washing 3 times in TBS-T, membranes were incubated at room temperature for 1 h with secondary antibodies. Protein amounts were detected using Licor Odyssey M Imagers. Bands were quantified using Fiji. RT–qPCR and RNA-seq Total RNA was isolated from HK2 and UOK cells using a Direct-zol RNA MiniPrep kit (Zymo Research, R2052) and converted to complementary DNA using a High-Capacity RNA-to-cDNA reverse transcription kit (Thermo Fisher, 4387406). RT–qPCR was carried out on a QuantStudio 3 Real-Time PCR instrument using PowerUp SYBR Green master mix (Thermo Fisher, A25742). Primers were purchased from Integrated DNA Technologies (IDT), and their sequences are detailed in Supplementary Table 1. The mRNA levels of each gene were normalized to those of GAPDH. RNA extraction for RNA-seq was performed in a similar manner. RNA-seq was performed as a service by Novogene. In brief, mRNA samples were enriched using oligo(dT) beads and then used to prepare strand-specific cDNA libraries. Sequencing was done on an Illumina Novaseq PE150. Individual reads are mapped to human genome hg38 by Hisat 2 (v.2.0.5). Featurecount in Rsubread was used to quantify gene expression levels, and DESeq2Rpackage was used for differential expression analyses. GO enrichment analysis of differentially expressed genes was performed using DAVID86. ChIP–seq Chromatin immunoprecipitation was performed as per previous studies87. In general, 108 cells were used per each ChIP–seq experiment. HK2 and UOK cells were first treated with 1% formaldehyde in UOK medium, followed by the addition of 0.125 M glycine to halt crosslinking. The cells, washed in ice-cold PBS, were then lysed in cell lysis buffer, with subsequent nutation at 4 °C. Cells were treated by Dounce homogenization on ice, followed by centrifugation at 500g for 10 min. The supernatant was discarded, and the pellets were resuspended using nuclear lysis buffer. Nuclear lysates were then sheared using a Covaris E220 to achieve an average chromatin length of 300–400 bp. The protein–chromatin complexes were captured using antibodies specific to TEAD1, H3K9me3, YAP1, H3K27ac and mouse IgG, and the mixtures were incubated overnight on a nutator at 4 °C. The antibody–chromatin complexes were later captured using magnetic protein G beads, washed and eluted using elution buffer. After reversal of formaldehyde crosslinks, DNA was purified with ChIP DNA Clean & Concentrator (Zymo Research, D5205) and library preparation was performed using NEBNext Multiplex Oligos for Illumima (NEB, 10129566). ChIP–seq data analysis ChIP–seq data were processed using the following pipeline. Raw sequencing reads in FASTQ format were quality-checked using FastQC. For identifying TEAD1–YAP co-occupied regions, YAP and H3K27ac reads were aligned to the human reference genome (hg38) using Bowtie2 (v.2.3.5.3) with default parameters. The resulting BAM files were used to generate coverage tracks using bamCoverage from the deepTools suite88, with normalization of counts per million (CPM). Gene tracks were visualized using IGV (v.2.16.2). Peaks for YAP and H3K27ac ChIP–seq were identified using MACS2, with default narrow peak settings and corresponding input controls (HK2 and UOK121 1% input) used to reduce background signal. A FDR cutoff of q < 0.01 was applied for peak selection. To identify putative co-regulatory TEAD1–YAP regions, overlapping peaks between TEAD1 and YAP datasets were identified using the bedintersect function in BEDTools. The resulting intersected peak regions were used for motif enrichment analysis via MEME–ChIP from the MEME Suite. GO enrichment analysis of intersected peak regions was done using DAVID86. To investigate the genomic context of TEAD1 condensates, sequenced reads were mapped with Bowtie2 using repetitive mapping (-k 50). SICER2 (ref. 89) was used for peak calling to identify broad enrichment regions characteristic of H3K9me3-marked heterochromatin. Peaks for TEAD1 and H3K9me3 ChIP–seq were identified using SICER2 with broad peak parameters (-w 1000 rt 1 –fdr 0.05 -g 3000 –e 1000), and input controls with overlapped mapping (-k 50) ChIP–seq peaks were selected. To identify TEAD1–H3K9me3 regions, overlapping peaks between the two were identified using the bedintersect function in BEDTools, and centromere regions were eliminated using the bedsubtract function. To compare these co-bound regions across cell lines, we used computeMatrix90 to quantify signal intensities at TEAD1-binding regions across all four cell lines. Heatmaps of TEAD1 signals were generated using plotHeatmap to visualize TEAD1-binding profiles in each cell line in TEAD1–H3K9me3 co-bound regions in condensate-positive cell lines (UOK121 and UOK342). Motif discovery and spacing calculation To quantify whether motif occurrences are enriched near centromeres, we analysed genome-wide motif hit coordinates. BED files for MCAT (a TEA transcription factor binding motif) and AP-1 (a JUN–FOS transcription factor binding motif) were pre-built using FIMO91. Only primary hg38 chromosomes (chromosome 1–chromosome 22, chromosome X and, chromosome Y) were retained, and each motif hit was presented by its genomic midpoint. Centromere annotations were collapsed per chromosome by taking the minimum start and maximum end. For each chromosome, we defined a pericentromeric interval as the centromere region expanded by ±4 Mb (bounded by chromosome ends). We then partitioned each chromosome into consecutive 1 Mb bins and counted motif midpoints per bin using a histogram-based approach; counts were converted to densities (hits per Mb) for each bin. Motif density distributions between pericentromeric and non-pericentromeric bins were compared using a one-tailed Welch’s t-test. To calculate the ECDF in each chromosome, motif midpoints were de-duplicated to remove identical positions, sorted and the distances between consecutive midpoints were computed (bp) using the first difference. Each distance was assigned a genomic position at the midpoint of the corresponding gap. BETA To evaluate the regulatory function of TEAD1, we performed BETA using the BETA basic module52. ChIP–seq peak regions for TEAD1 were used as input along with RNA-seq differential expression results from UOK121 and UOK342 cells. Differentially expressed genes were identified on the basis of log2 fold change thresholds and were categorized as upregulated, downregulated or unchanged. The distance between TEAD1 peaks and target genes was defined using regions identified by the intersection of TEAD1 and H3K9me3 ChIP–seq peaks (via BEDtools), which reflected chromatin context-specific binding. BETA computes regulatory potential scores for each gene on the basis of binding site proximity and density and assesses whether TEAD1 functions primarily as an activator or repressor using a Kolmogorov–Smirnov test to compare cumulative distributions of gene categories. TEAD1 expression, purification and labelling T7Express cells (NEB) were transformed with the pSAT vector encoding His-6, SUMO and TEAD1 (residues 1–426). Cells were grown in TB with 200 mg ml−1 ampicillin. Expression was induced with 0.5 mM isopropyl-β-D-1-thiogalactopyranoside at OD600 = 2.0, and cells were left shaking overnight at 20 °C. The cell pellet was resuspended in 100 ml lysis buffer (50 mM Tris pH 7, 2 M NaCl, 5 mM beta-mercaptoethanol (b-ME), 5% glycerol and 10 mM imidazole, supplemented with EDTA-free protease inhibitor cocktail (Sigma) and 1 mM phenylmethylsulfonyl fluoride) and lysed with a microfluidizer. Clarified lysates were incubated with Ni-charged Profinity IMAC resin (Bio-Rad) for 30 min at 4 °C. Resin was collected and then washed with 100 ml lysis buffer, followed by 100 ml lysis buffer supplemented with 15 mM MgCl2 and 5 mM ATP, and then 100 ml of 50 mM Tris pH 7, 500 mM NaCl, 5 mM b-ME, 5% glycerol and 20 mM imidazole. Protein was eluted with 250 mM imidazole and dialysed overnight at 4 °C into 50 mM Tris pH 7, 400 mM NaCl, 5 mM b-ME and 5% glycerol. The H6-SUMO tag was removed by incubation with SENP (made in-house). Protein was pelleted with 40% NH4SO4, resuspended in 5 ml dialysis buffer and further purified by size exclusion chromatography using a Superdex 75 16/600 column (GE) equilibrated in dialysis buffer. The protein was concentrated to 65 µM and flash-frozen (Extended Data Fig. 7e). Labelled TEAD1 was made as above except that the dialysis buffer contained 10 mM HEPES pH 7.4, was tagged with Alexa Fluor 488 TFP ester (Invitrogen) according to the manufacturer’s instructions after gel-filtration and then flash frozen. The final degree of labelling was determined to be 0.5 based on absorbance, as defined in the manual. Phase separation of recombinant TEAD1 DNA oligonucleotides were synthesized by IDT (sequences listed in Supplementary Table 3). Equimolar amounts of complimentary oligonucleotides were resuspended at 1 mM in 10 mM Tris pH 8.0, 100 mM NaCl and incubated at 95 °C for 5 min in a heat block that was then cooled to room temperature overnight. Unlabelled and labelled TEAD1 were mixed to a final ratio of 10:1. Next, 5 µM of mixed TEAD1 was added to annealed dsDNA containing the indicated concentrations and sequences in 20 mM Tris pH 7, 150 mM NaCl and 5 mM MgCl2. Glass slides and coverslips were cleaned with 5% Hellmanex (Hellma Analytics), rinsed 3 times with water, then coated with 1% BSA solution for 1 h at room temperature, rinsed with water again and left to dry. SecureSeal imaging spacers (Grace Bio-Labs) were placed on each coverslip. Next, 3 µl of TEAD1–DNA solution was added to each well. Wells were sealed and imaged using a Zeiss LSM 900 at 488 nm. TEAD1 condensates were identified on the basis of fluorescence intensity, size and circularity using Fiji81. Six random locations were analysed per condition, and each condition was replicated three times. Data were plotted and one-way ANOVA test was performed using GraphPad Prism (v.10.6.0 for Windows, GraphPad Software). Proximity-based proteomic experiments Immunofluorescence confirmation of biotinylation of TEAD1 condensate components Cells were plated on glass coverslips in a 24-well plate and transfected the next day at 70% confluency using Lipofectamine 3000 transfection reagent (L3000015, Thermo Fisher), followed by 24 h of incubation. To image nuclear localization of the construct miniTurbo-TEAD1-HaloTag, cells were stained with JF646 HaloTag ligand (100 nM; Janelia Materials) for 30 min before fixation. Cells were treated with 500 µM biotin in fresh growth medium for 10 min at 37 °C. Immediately after incubation, cell were washed with ice-cold DPBS for 3 times and fixed with 4% paraformaldehyde in PBS for 15 min on ice, permeabilized with ice-cold methanol at –20 °C for 5 min and blocked with 1% w/v BSA in DPBS at 4° C for 30 min. Primary antibodies against TEAD1 (610923, BD Biosciences) were incubated overnight at 4 °C. Cells were then incubated with secondary antibody Alexa Fluor 488 goat anti-mouse IgG (H+L) (a11001, Thermo Fisher). Biotinylated proteins were stained using streptavidin-Alexa Fluor 568 (S11226, Thermo Fisher) for 1 h at room temperature. The cells were then incubated with Hoechst 33342 at room temperature for 5 min and mounted on glass slides with Vectashield and sealed with nail polish. Images were Airyscan-processed with 1071 Zeiss Zen software (standard, 2D processing). Confirmation of protein biotinylation by western blotting Cells were plated on 10 cm Petri dishes and transfected the next day at 70% confluency using Lipofectamine 3000 transfection reagent (L3000015, Thermo Fisher), followed by 24 h of incubation. Cells were treated with 500 µM biotin in fresh growth medium for 10 min at 37 °C, washed with ice-cold DPBS for 3 times and collected using cell scrapers and centrifuged at 300g, 4 °C for 5 min. Whole-cell lysates were collected in 400 µl RIPA buffer containing 50 mM Tris pH 8 (T60050-1000.0, Research Products International), 140 mM NaCl (S23025-3000.0, Research Products International), 1% Igepal CA-630 (56741-50ML-F, Sigma Aldrich), 0.1% sodium deoxycholate (30970-25G, Sigma Aldrich), 0.1% w/v SDS (L3771, Sigma-Aldrich) and complete mini protease inhibitor cocktail (11836170001, Sigma Aldrich) for 20 min on ice and clarified by centrifugation at 13,000g, 4 °C for 10 min. Protein concentrations were measured by BCA assay. In brief, 175 µg protein from the lysates was enriched for biotinylation using Pierce streptavidin magnetic beads (88816, Thermo Fisher) and incubated overnight at 4 °C on a rotator. The bead slurry was washed twice with RIPA buffer, once with 1.0 M KCL, once with 50 mM triethylammonium bicarbonate (TEAB) buffer and twice with RIPA buffer. About 10 µg of protein from input, post-immunoprecipitation supernatant and pull-down samples were loaded onto a gel for immunoblotting. Biotinylated proteins were stained using streptavitin-IRDye680 (926-68079, Licor). Other antibodies used were anti-TEAD1 antibody (610923, DB Biosciences) and anti-β-actin antibodies (AC004, ABClonal). Sample preparation for proteomic analysis with miniTurbo UOK342 cells stably expressing miniTurbo-TEAD1-HaloTag, miniTurbo-TEAD1(H3Mut)-HaloTag and miniTurbo-HaloTag-NLS were cultured overnight in 15 cm Petri dishes and treated with 500 µM biotin for 10 min at 37 °C. Cells were collected and the nuclear protein extract was collected using a Nuclear Extract kit (Active Motif, 40010). Around 600 µg of nuclear protein lysate was collected. Next, 60 µl magnetic beads (washed twice with RIPA buffer) was added. The mixture was topped up with additional RIPA buffer to 1 ml. Samples were incubated on a rotator overnight at 4 °C. After overnight incubation, streptavidin beads were collected on the side of the 1.5 ml tube using a magnet. The lysate solution was transferred into new tubes, and the beads were washed twice with 1 ml RIPA lysis buffer, once with 1 ml of 1.0 M KCl, once with 1 ml of 50 mM TEAB buffer and twice with Tris-HCl pH 7.5. TCA–acetone precipitation Nuclear lysates were buffered to pH 8.0 with 100 mM TEAB buffer in 1.5 ml centrifuge tubes, then reduced using 25 µl of 50 mM dithiothreitol for 1 h at 60 °C. Samples were chilled on ice then alkylated using 25 µl of 50 mM iodoacetamide for 15 min in the dark. Samples were then diluted on ice in 8× volume of 10% trichloroacetic acid in acetone w/v stored at –20 °C. Samples were vortexed and incubated for at least 4 h at –20 °C. Samples were pelleted by centrifugation at 16,000g for 10 min at 0 °C. The supernatant was carefully removed to leave the pelleted protein on the walls of the tube. A volume of 500 µl of neat acetone (stored at –20 °C) was added to the pelleted protein and vortexed. This was then incubated for at least 10 min at –20 °C before centrifugation at 16,000g once again and the supernatant carefully removed. The remaining acetone was allowed to evaporate for 10 min before the addition of trypsin (Pierce) in 100 mM TEAB at a ratio of 1:50 enzyme to protein and allowed to digest overnight at 37 °C. Digested peptides were lyophilized and reconstituted in 500 µl RIPA buffer and incubated with 50 µl Pierce streptavidin magnetic beads overnight at 4 °C on a rotator. The bead slurry was washed as described in the western blotting section, followed by two washes with ultrapure water. The washed beads were air-dried at 4 °C for 6 h and stored at –20 °C. Eluting peptides from streptavidin linked magnetic beads Biotinylated peptides bound to streptavidin linked magnetic beads were separated from the final wash buffer by placing the Eppendorf tubes on a magnetic holder. After removing and setting aside the final wash buffer, the following steps were performed in a fume hood. Neat HFIP (200 µl Millipore/Sigma, 105228) was added to each tube and put on a shaker at 1,000 rpm for 10 min. The bead slurry was then placed on a magnetic holder to recover the HFIP-eluted peptides from the magnetic beads. HFIP extraction was performed twice and both peptide elutions were combined in a new Eppendorf tube. The combined extractions were evaporated to dryness in a Speedvac and resuspended in 0.1% trifluoroacetic acid (TFA). Samples were buffer exchanged using an Oasis HLB (Waters) with two washes of 0.1% TFA and eluted with 0.1% TFA, 60% acetonitrile into new tubes and dried by vacuum centrifugation before mass spectrometry analysis. Mass spectrometry analysis Tryptic peptides were analysed by reverse-phase chromatography tandem mass spectrometry on a Vantage NEO UPLC interfaced with an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific). Peptides were separated using a 2–90% acetonitrile in 0.1% formic acid gradient over 90 min at 300 nl min−1 using an in-house packed 75-µm ID by 15 cm column packed with ReproSIL-Pur-120-C18-AQ (2.4 µm, 120 Å bulk phase, Dr Maisch). Survey scans of precursor ions were acquired from 375 to 1,500 m/z at a resolution setting of 120,000. Precursor ions were isolated in data-dependent mode with a 3 s cycle time using a 1.5 dalton isolation width and a 45 s dynamic exclusion. Peptides were fragmented at a normalized HCD collision energy of 30 with a resolution setting of 30,000. Data analysis and candidate protein selection Mass spectrometry data for all six files were processed together using Proteome Discoverer (v.3.1.1.93, Thermo Fisher Scientific) with Sequest with Match Between Runs enabled against the UniProt Homo sapiens database. The TEAD constructs were inserted at the end of the Human FASTA file. Search criteria included trypsin as the enzyme with two allowed missed cleavages, a 10 ppm precursor mass tolerance and a 0.02 Da fragment mass tolerance. Modifications included fixed carbamidomethylation on C, oxidation on M, deamidation on N or Q and biotin on K as variable modifications. Peptide identifications were filtered at the 1% FDR level using the Percolator Node in Proteome Discoverer. Abundance ratio and adjusted P values were generated and visualized by volcano plots showing differentially enriched proteins. String Analysis was generated based on the pulled candidate list. Protein–protein interaction networks and GO enrichment were rendered with settings at high confidence (0.700). Clusters composed of fewer than four proteins were removed. See Supplementary Table 4 for proteomic source data. The mass spectrometry proteomic data have been deposited into the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers PXD072858 and PXD072858. Single-particle tracking Single-particle tracking (SPT) of TEAD1-Halo molecules was performed on a Nikon Ti2-Eclipse Motorized TIRF system with a ×100/NA 1.49 oil-immersion objective, a Hamamatsu ORCA Fusion BT CMOS camera system and a Perfect Focus System to combat axial focus drift. A Tokai Hit system was used to keep the humidified incubation chamber at 37 °C with 5% CO2. Nikon NIS-Elements was used to control the microscope hardware. UOK342 TEAD1-Halo cells were plated in an 8-well chambered cover glass plate (Cellvis, C8-1.5H-N) 48 h before imaging. On the day of imaging, cells were stained with 10 nM JFX554 Halo ligand for 30 min in FluoroBrite DMEM complete medium (Gibco, A1896701) supplemented with 10% (Gibco, 26140079), 2 mM GlutaMAX-l (Gibco, 35050061) and 1× NEAA (Gibco, 11140050). Images used for SPT were obtained at 16-bit (no binning), 300 ms exposure and 25% 561 nm laser (LUN-F XL) for 2,000 frames at an image size of 512 × 512 pixels (0.065 μm per pixel resolution). Before imaging, a snapshot of the same field of view was taken using epifluorescence (10 ms of exposure, 25% 550 nm channel D-LEDI) for segmenting the nucleus and thresholding a mask of TEAD1 condensates during analysis. Four independent replicates were performed with 8 cells imaged per replicate. To correct for photobleaching of the fluorophore during subsequent data analysis, individual H2B molecules were also imaged as they are stably bound on chromatin. UOK342 cells were plated and transiently transfected as described above with PB-EF1-Halo-H2B 16 h before staining with JFX554 Halo ligand and imaging. At least 5 cells were imaged per replicate under the same parameters. SPT analysis A previously published protocol92 was used for SPT analysis to determine the dissociation kinetics and residence time of long-binding TEAD1 molecules inside and outside TEAD1 condensates. The steps and parameters used are briefly described here. The nuclear region in each image stack was first segmented using the epifluorescence snapshot as a mask. We used a custom-written ImageJ macro using Trainable Weka Segmentation to create a binary mask of the nucleus from the snapshot, which was then manually verified. The segmented image stacks were then converted to TIFF format and used in subsequent analyses. Localization and tracking of single molecules were performed using SLIMfast, a GUI implementation of a previously described MTT algorithm93, with the following parameters: localization error = 10-6, deflation loops = 0, detection box = 9 pixels, lag time = 300 ms, max. expected diffusion coefficient (μm2 s−1) = 1, max. no. competitors = 1. ImageJ macros and Matlab code that have previously been reported94 were used to threshold and create a binary mask of TEAD1 condensates from the same nuclear snapshot and to subsequently categorize trajectories as being inside a condensate, which was defined as spending over 50% of its lifetime in the masked area. A custom-written ImageJ macro was also used to rotate and place the condensate mask randomly in the nucleus so that it remained in the same area as the condensate but did not overlap with it. Trajectories that were inside the original condensate or inside the randomly placed mask in the nucleoplasm were then compiled by replicate and fit to a bi-exponential decay model to extract the measured dissociation rate constant (kobserved) and residence time (1/k). To correct for photobleaching, the same analysis was performed for Halo-H2B molecules in the whole nucleus (without categorization into a condensate), which are largely immobile. Because they were transiently transfected, we observed a proportion of H2B molecules that behaved unbound and were highly dynamic. As a result, we filtered out tracks that were fewer than 4 frames long and used the remaining trajectories to calculate k of the H2B molecules (kpb). We subtracted kpb from kobserved to get the true koff of TEAD1 molecules. Photobleaching correction was done per replicate, whereby each replicate was corrected by the kpb of Halo-H2B movies taken on the same day. The corrected residence time was calculated by 1/koff. The koff and residence times were averaged and GraphPad Prism was used to perform two-tailed paired t-tests between in-condensate and masked nucleoplasm trajectories. Statistics and reproducibility All experiments were repeated as stated in the figures, legends and Methods. Statistical details are presented in the figure legends and as detailed below. Comparisons were performed in GraphPad Prism (v.10.1.1). No statistical method was used to predetermine sample sizes but our sample sizes were similar to those reported in previous publications26,95. Data distribution was assumed to be normal, but this was not formally tested. For experiments quantifying the per cent of condensate-positive cells, fields of view were randomly selected using the Hoechst channel. Spatial analysis including RNA FISH, 3D image quantification, colocalization analysis and SPT were not randomized, and cells with similar intensity and with LTCs were selected and imaged. For in vitro condensate experiment data analyses, clear outliers were removed from the dataset. Of the remaining data points per condition, the first six images were analysed and plotted to ensure equal representation of each condition. For drug treatment assays, time-lapse images of cells moving out of focus during the imaging period were carefully removed. The investigators were not blinded to allocation during experiments and outcome assessment. Data collection and analyses were not performed blinded to the conditions of the experiments. Exact P values were as follows: Fig. 1b, P(HK2 vs UOK121) = 0.0146, P(HK2 vs UOK342) = 0.8588, P(HK2 vs UOK122) = 0.9998, P(HK2 vs UOK140) = 0.1004, P(HK2 vs UOK275) = 0.6848; Fig. 1d, P(HK2 vs UOK121) < 0.0001, P(HK2 vs UOK342) < 0.0001, P(HK2 vs UOK122) = 0.3106. Fig. 2b, P(control vs LTC) = 0.0411 for UOK121 RNA PolII-pSer2, P(control vs LTC) < 0.0001 for comparisons in UOK342 RNA PolII-pSer2, UOK121 and UOK342 H3K27ac, UOK121 and UOK342 BRD4; Fig. 2l, P(STC vs LTC) < 0.0001, P(STC vs YAP) = 0.9322, P(LTC vs YAP) < 0.0001. Fig. 3c, P(control vs LTC) < 0.0001; Fig. 3d, P(control vs LTC) < 0.0001; Fig. 3k, P(TRF2 vs CENP-A) < 0.0001; Fig.3l, P(TRF2 vs CENP-A) < 0.0001. Fig. 4c, P < 0.0001 for WT vs Δ55–122, WT vs H1Mut, WT vs H3Mut, P(WT vs ΔIDR) = 0.4902, P(WT vs Y406H) = 0.9997, P(WT vs S102A) = 0.2866, P(WT vs S102D) = 0.6823; Fig. 4d, P < 0.0001 for WT vs Δ55–122, WT vs H1Mut, WT vs H3Mut, P(WT vs ΔIDR) = 0.721, P(WT vs Y406H) = 0.7609, P(WT vs S102A) = 0.4948, P(WT vs S102D) = 0.9814; Fig. 4j, P < 0.0001 for WT vs Δ55–122, WT vs H1Mut, WT vs H3Mut; Fig. 4k, P(WT vs Δ55–122) < 0.0001, P(WT vs H1Mut) = 0.0018, P(WT vs H3Mut) = 0.001. Fig. 5k, P < 0.000001 for MCAT = 0 vs MCAT = 4 and MCAT = 1 vs MCAT = 4, P(MCAT=0 vs MCAT=1) = 0.008821. Extended Data Fig. 1g, P < 0.0001 for TEAD1 vs TEAD2, TEAD1 vs TEAD3, TEAD1 vs TEAD4; Extended Data Fig. 2b, P(HK2 vs 1205Lu) < 0.0001, P(HK2 vs WM983B) = 0.9632, P(HK2 vs FS4) = 0.6661, P(HK2, vs AG13004) = 0.4256, P(HK2 vs T47D) and P(HK2 vs MCF-7) = 0.9999, P(HK2 vs MDA-MB-231) = 0.7929, P(HK2 vs MCF-10A) = 0.9953, P(HK2 vs HeLa) = 0.0003, P(HK2 vs U-2 OS) = 0.7969, P(HK2 vs HEK293) = 0.9234. Extended Data Fig. 2d, P(HK2 vs 1205Lu) = 0.2788, P(HK2 vs WM983B) = 0.4889, P(HK2 vs FS4) = 0.0988, P(HK2, vs AG13004) = 0.9006, P(HK2 vs T47D) = 0.9985, P(HK2 vs MCF-7) = 0.9192, P(HK2 vs MDA-MB-231) = 0.4077, P(HK2 vs MCF-10A) = 0.1474, P(HK2 vs HeLa) < 0.0001, P(HK2 vs U-2 OS) = 0.3941, P(HK2 vs HEK293) < 0.0001. Extended Data Fig. 3a, P(UOK121 DMSO control vs foci) = 0.9792, P(UOK121 EU control vs foci) < 0.0001, P(UOK342 DMSO control vs foci) = 0.9988, P(UOK342 EU control vs foci) < 0.0001. Extended Data Fig. 4g, P(control vs LTC) < 0.0001; Extended Data Fig. 4h, P(control vs LTC) < 0.0001. Extended Data Fig. 5a, P(WT vs Δ55–122) < 0.0001, P(WT vs H1Mut) = 0.0004, P(WT vs H3Mut) = 0.0006; Extended Data Fig. 5b, P(WT vs ∆55–122) < 0.0001, P(WT vs H1Mut) < 0.0001, P(WT vs H3Mut) = 0.0001; Extended Data Fig. 5l, P < 0.0001 for WT vs Δ55–122, WT vs H1Mut, WT vs H3Mut; Extended Data Fig. 5m, P < 0.0001 for WT vs Δ55–122, WT vs H1Mut, WT vs H3Mut. Detailed statistical information, including 95.00% CI of difference and degrees of freedom for each multiple comparison test is provided in the source data of the respective figure and Extended Data figure. The imaging experiments in Figs. 1c, 3e,g and 7b were performed twice independently with similar results. The imaging experiments in Extended Data Fig. 2f were performed three times independently with similar results, The SPT experiments in Fig. 6e were performed four times with similar results. The western blot experiments in Extended Data Fig. 5c–g were done with one to two replicates to confirm protein integrity after transfections. The western blot experiments in Extended Data Fig. 8a were performed three times independently with similar results. The western blot experiment for validating gene knockdown in Extended Data Fig. 8b was performed twice independently with similar results. The Coomassie total protein gel in Fig. 7e was performed once to confirm the purity and stability of labelled and unlabelled TEAD1 used for in vitro experiments. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Online content Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41556-026-01985-x. Supplementary information Reporting Summary Primers used in this study. Antibodies used in this study. DNA oligonucleotides used in this study. Protein identification and statistical analyses. Light microscopy reporting table. Source data Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Unprocessed western blots. Unprocessed western blots. Unprocessed western blots. Unprocessed western blots. Unprocessed western blots. Unprocessed protein gel. Unprocessed western blots. Acknowledgements We appreciate the discussions and feedback from members of the Cai Laboratory, S. Botterbusch and A. Leung; the centromere DNA FISH probe gifted by A. Meeker; and assistance from staff at the Johns Hopkins School of Medicine Single Cell & Transcriptomics Core and Center for Proteomics Discovery. Extended data Author contributions D.C. was responsible for conceptualization. Y.W., J.L., J.D., E.B. and B.J.H. conducted imaging and analyses. K.S.L. conducted in vitro TEAD1 assays under the supervision of J.M.K. E.A.L. conducted SPT assays and analyses, consulting with S.R.Y. and S.C. J.D. performed ChIP–seq experiments. J.L. analysed the ChIP–seq data under the guidance of C.Z. C.J.R. and W.M.L. provided the RNA-seq data and the UOK cell lines. D.C., Y.W. and J.L. drafted the manuscript with the input from all authors. Funding acquisition: D.C., J.M.K., C.Z., W.M.L. and S.C. Peer review Peer review information Nature Cell Biology thanks Francois-Michel Boisvert, Zhijie Liu and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Funding This work is supported by the Department of Defense Kidney Cancer Idea Development Award (W81XWH2210900 to D.C.), the National Institutes of Health (T32GM080189 to K.S.L.; T32CA009110 to J.D., E.L. and E.B.; R35GM156573 to J.M.K.; R35GM142837 to D.C.; R35GM133712 to C.Z.; and an intramural grant to W.M.L.), a Johns Hopkins University Catalyst Award (to D.C.), the National Science Foundation Graduate Research Fellowship Program (DGE-1745301 to S.R.Y), the Pew-Stewart Scholar for Cancer Research Award (to S.C.), the Searle Scholar Award (to S.C.), the Merkin Innovation Seed Grant (to S.C.), the Mallinckrodt Research Grant (to S.C.), the Margaret E. Early Medical Research Trust Grants (to S.C.), and the Alex’s Lemonade Stand Foundation Innovation Grant (1260879 to S.C.). Data availability The RNA-seq data and ChIP–seq data (FASTQ and BAM files) reported in this paper have been deposited into the Sequence Read Archive (PRJNA1268000), and the processed data (BIGWIG) have been deposited into the Gene Expression Omnibus (GSE298316). The mass spectrometry proteomic data have been deposited into the ProteomeXchange Consortium via the PRIDE96 partner repository with the dataset identifier PXD072858. Source imaging data have been published and available from the Johns Hopkins Research Data Repository (10.7281/T1LFLQL2). All other data supporting the findings of this study are available within the paper and its Supplementary Information. Source data are provided with this paper. Code availability All codes associated with this manuscript have been deposited into GitHub (https://github.com/BMBCaiLab/TEAD1-NCB-paper.git). Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Yiran Wang, Jindayi Liang. Extended data is available for this paper at 10.1038/s41556-026-01985-x. Supplementary information The online version contains supplementary material available at 10.1038/s41556-026-01985-x. References Associated Data Supplementary Materials Reporting Summary Primers used in this study. Antibodies used in this study. DNA oligonucleotides used in this study. Protein identification and statistical analyses. Light microscopy reporting table. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Statistical source data. Unprocessed western blots. Unprocessed western blots. Unprocessed western blots. Unprocessed western blots. Unprocessed western blots. Unprocessed protein gel. Unprocessed western blots. Data Availability Statement The RNA-seq data and ChIP–seq data (FASTQ and BAM files) reported in this paper have been deposited into the Sequence Read Archive (PRJNA1268000), and the processed data (BIGWIG) have been deposited into the Gene Expression Omnibus (GSE298316). The mass spectrometry proteomic data have been deposited into the ProteomeXchange Consortium via the PRIDE96 partner repository with the dataset identifier PXD072858. Source imaging data have been published and available from the Johns Hopkins Research Data Repository (10.7281/T1LFLQL2). All other data supporting the findings of this study are available within the paper and its Supplementary Information. Source data are provided with this paper. All codes associated with this manuscript have been deposited into GitHub (https://github.com/BMBCaiLab/TEAD1-NCB-paper.git).