Phase separation of phospho-HDAC6 drives aberrant chromatin architecture in triple-negative breast cancer

Lu, Bing; Qiu, Ru; Wei, Jiatian; Wang, Li; Zhang, Qinkai; Li, Mingsen; Zhan, Xiudan; Chen, Jian; Hsieh, I-Yun; Yang, Ciqiu; Zhang, Jing; Sun, Zicheng; Zhu, Yifan; Jiang, Tao; Zhu, Han; Li, Jie; Zhao, Wei

doi:10.1038/s43018-024-00816-y

Article
Published: 28 August 2024

Phase separation of phospho-HDAC6 drives aberrant chromatin architecture in triple-negative breast cancer

Bing Lu¹^na1,
Ru Qiu²^na1,
Jiatian Wei²^na1,
Li Wang³^na1,
Qinkai Zhang²,
Mingsen Li²,
Xiudan Zhan²,
Jian Chen²,
I-Yun Hsieh⁴,
Ciqiu Yang⁵,
Jing Zhang⁶,
Zicheng Sun⁶,
Yifan Zhu²,
Tao Jiang²,
Han Zhu²,
Jie Li ORCID: orcid.org/0000-0003-1232-3248^6,7 &
…
Wei Zhao ORCID: orcid.org/0000-0002-0774-2571^1,2

Nature Cancer (2024)Cite this article

1960 Accesses
9 Altmetric
Metrics details

Subjects

Abstract

How dysregulated liquid–liquid phase separation (LLPS) contributes to the oncogenesis of female triple-negative breast cancer (TNBC) remains unknown. Here we demonstrate that phosphorylated histone deacetylase 6 (phospho-HDAC6) forms LLPS condensates in the nuclei of TNBC cells that are essential for establishing aberrant chromatin architecture. The disordered N-terminal domain and phosphorylated residue of HDAC6 facilitate effective LLPS, whereas nuclear export regions exert a negative dominant effect. Through phase-separation-based screening, we identified Nexturastat A as a specific disruptor of phospho-HDAC6 condensates, which effectively suppresses tumor growth. Mechanistically, importin-β interacts with phospho-HDAC6, promoting its translocation to the nucleus, where 14-3-3θ mediates the condensate formation. Disruption of phospho-HDAC6 LLPS re-established chromatin compartments and topologically associating domain boundaries, leading to disturbed chromatin loops. The phospho-HDAC6-induced aberrant chromatin architecture affects chromatin accessibility, histone acetylation, RNA polymerase II elongation and transcriptional profiles in TNBC. This study demonstrates phospho-HDAC6 LLPS as an emerging mechanism underlying the dysregulation of chromatin architecture in TNBC.

You have full access to this article via your institution.

Download PDF

Prion-like domain mediated phase separation of ARID1A promotes oncogenic potential of Ewing’s sarcoma

Article Open access 03 August 2024

Comparative characterization of 3D chromatin organization in triple-negative breast cancers

Article Open access 05 May 2022

Phase separation drives aberrant chromatin looping and cancer development

Article 23 June 2021

Main

Triple-negative breast cancer (TNBC) is a highly heterogeneous group of diseases that exhibit the most aggressive cancer phenotypes among persons with breast cancer¹. Because of the absence of estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2), conventional endocrine therapy or chemotherapy is ineffective in treating this type of cancer². Moreover, the presence of invasive tumor heterogeneity further complicates treatment strategies. Previous studies have focused on exploring the genomic, transcriptomic and proteomic heterogeneity of TNBC cells, identifying four TNBC subtypes on the basis of gene expression profiling^3,4,5. However, the role of these dysregulated features in perturbing high-order chromatin organization in TNBC remains poorly understood.

Recent studies have demonstrated that aberrant forms of liquid–liquid phase separation (LLPS) have a crucial role in driving multiple hallmarks of cancer⁶. Investigating the mechanism of abnormal LLPS condensate formation in cancer may provide opportunities to elucidate the role of chromatin structure in gene expression regulation. One such example is the LLPS of nuclear pore complex (NPC) protein 98 (NUP98) with homeobox A9 (HOXA9), which contributes to the formation of a broad superenhancer (SE)-like binding pattern, thereby promoting transcriptional activation of leukemogenic genes⁷. Additionally, our previous work revealed that the LLPS formed by HOXB8 and Fos-related antigen 1 at SE loci of chromatin established deregulated transcription in osteosarcoma⁸. Despite these exciting discoveries, efforts to elucidate the molecular function of LLPS condensates in breast cancer, particularly TNBC, are still in the early stages.

Biomolecular LLPS is driven by multivalent weak interactions that rapidly form, break and reform⁹. Proteins achieve multivalency through tandem binding modules, often containing intrinsically disordered regions (IDRs)¹⁰. Although still challenging, progress has been made in the discovery of small-molecule inhibitors that directly disrupt phase separation by targeting IDRs. For instance, ralaniten (EPI-002) is the first drug to be tested in a clinical trial that directly binds to the IDR of androgen¹¹. However, it is worth noting that IDRs are widely distributed in various cell types, indicating that drugs targeting IDRs may lack specificity. Recent studies revealed that small changes in the local abundance of core components, such as post-translational modifications (PTMs), can significantly regulate the properties of phase separation¹². Therefore, one potential strategy for effective perturbation of LLPS will be to develop drugs tailored toward specific PTMs.

Histone deacetylase 6 (HDAC6) is a well-characterized class II HDAC that is responsible for deacetylating nonhistone and histone substrates¹³. HDAC6 has been shown to be a critical driver of tumorigenesis in various contexts, including breast cancer¹⁴. Recent studies have demonstrated that approximately 30% of human breast cancers could potentially benefit from HDAC6 inhibitor therapy¹⁵. Inhibition of HDAC6 using a compound, BAS-2, was shown to reduce glycolytic metabolism in TNBCs¹⁶. Although HDAC6 mainly localizes in the cytoplasm¹⁷, where it has a crucial role in misfolded protein aggregate formation^18,19, it remains unknown whether HDAC6 is involved in aberrant LLPS in breast cancer.

In this study, we report the discovery that phosphorylated HDAC6 (phospho-HDAC6) forms LLPS condensates in the transcriptionally active regions of chromatin within the nuclei of TNBCs but to a lesser extent in non-TNBCs. Further investigation revealed that exposure to epidermal growth factor (EGF) induces phosphorylation of S22 in the N-terminal IDR domain of HDAC6 by GRK2. Subsequently, importin-β binds to phospho-HDAC6 and facilitates its nuclear translocation, ultimately leading to condensate formation with 14-3-3θ. Remarkably, we identified Nexturastat A as a potential inhibitor of phospho-HDAC6 LLPS, which effectively suppressed in vivo tumor growth of TNBCs. Disruption of phospho-HDAC6 LLPS by Nexturastat A resulted in a reshaped chromatin architecture and altered transcriptional profiling. Our findings suggest that pharmacological modulation of phosphorylation-triggered HDAC6 phase separation may offer a promising avenue for developing treatments for refractory TNBC.

Results

Discovery of phospho-HDAC6 undergoing LLPS in TNBC

We aimed to investigate the potential involvement of HDAC6 in TNBC tumorigenicity and its association with aberrant LLPS in this cancer. Our findings revealed a significant increase in nuclear HDAC6 expression in TNBCs when compared to non-TNBCs and paracancerous tissues (Extended Data Fig. 1a). Moreover, we noted that HDAC6 in the nucleus of TNBCs was predominantly found in a phosphorylated form (Extended Data Fig. 1b). To further explore this, we systematically substituted the potential phosphorylation sites of HDAC6 one by one (Extended Data Fig. 1c) and found that the substitution of S22 significantly impeded the nuclear localization of HDAC6 (Extended Data Fig. 1d,e). Remarkably, we observed a significant increase in nuclear expression of phospho-HDAC6 at S22 in TNBC samples when compared to non-TNBC and paracancerous tissues (Extended Data Fig. 1f). Upon analysis of data from persons with TNBC along with their follow-up information, we unveiled that elevated levels of nuclear phospho-HDAC6 exhibited a pronounced positive correlation with tumor stage (P < 0.0001) and metastasis (P = 0.043) (Extended Data Fig. 1g). Furthermore, persons with higher levels of phospho-HDAC6 exhibited shorter overall survival periods compared to those with lower levels of phospho-HDAC6 (Extended Data Fig. 1h).

Previous studies linked phospho-HDAC6 to pathological protein aggregates in Parkinson’s disease¹⁹. Therefore, we used a super-resolution microscope to examine the phospho-HDAC6 LLPS in TNBC specimens and cell lines. Phospho-HDAC6 exhibited a significantly elevated level of condensate formation within the nuclei of TNBC specimens and cell lines (BT-549 and MDA-MB231) when contrasted with paracancerous tissue, mammary epithelial cells (MCF-10A) and non-TNBC specimens and cells (MCF-7) (Fig. 1a,b). In contrast, unmodified HDAC6 remained localized in the cytoplasm, displaying a granular staining pattern in breast cancer cells (Extended Data Fig. 2a,b). Notably, the assembly of phospho-HDAC6 condensates was significantly disrupted by 5% 1,6-hexanediol, an aliphatic alcohol known to weaken hydrophobic interactions (Fig. 1c). Fluorescence recovery after photobleaching (FRAP) analysis demonstrated the rapid recovery of condensates of the HDAC6 S22E substitution, which mimics phosphorylation of HDAC6, on a timescale of seconds (Fig. 1d). Our single-molecule imaging studies revealed that HDAC6-S22E exhibited high mobility, as evidenced by the observation of track displacement (Extended Data Fig. 2c). These results strongly support the notion that the condensates of phospho-HDAC6 in nuclei of TNBCs exhibit features of LLPS.

**Fig. 1: Immunofluorescence analysis of phospho-HDAC6 in TNBC cells and clinical samples.**

We reconstructed three-dimensional (3D) images of BT-549 cells stained for phospho-HDAC6 showed that the condensates were exclusively distributed in euchromatin regions, outside the 4′,6-diamidino-2-phenylindole (DAPI)-stained heterochromatin areas (Fig. 1e). Active transcription markers, such as RNA polymerase (Pol) II phosphorylated at S5 (C-terminal domain) and H3K27ac (acetylation of histone H3 at K27)^20,21, were detected within the phospho-HDAC6 condensates (Fig. 1f). We induced acetylation of C27 within the H3K27C C110A framework using thioacetic acid to generate H3K27ac proteins. These proteins were subsequently combined with H2a, H2b, H4 and DNA containing 6× 177-bp tandem repeats of the Widom 601 sequence to create H3K27ac nucleosome chromatin. Remarkably, the resulting H3K27ac nucleosome chromatin formed spherical puncta in conjunction with HDAC6 (Fig. 1g,h). It is worth noting that there was no colocalization of phospho-HDAC6 condensates with the heterochromatin marker HP1α or the inactive transcriptional epigenetic marker H3K9me3 (trimethylation of histone H3 at K9)²² (Extended Data Fig. 2d).

Prior research showed that a significant proportion of TNBCs exhibit elevated expression of the EGF receptor (EGFR)²³. In our investigation, we observed that exposure to EGF led to an increase in levels of phospho-HDAC6 protein and concomitant formation of condensates in breast cancer cell lines of different subtypes (Fig. 1i and Extended Data Fig. 2e,f). Conversely, inhibition of G-protein-coupled receptor kinase 2 (GRK2), a substrate of EGFR and a kinase enzyme²⁴, resulted in a reduction in HDAC6 phosphorylation and the subsequent formation of LLPS condensates (Fig. 1j and Extended Data Fig. 2g,h). Additionally, we noted that knockdown of GRK2 impeded the activation of HDAC6 phosphorylation and LLPS in response to EGF stimulation (Fig. 1k,l and Extended Data Fig. 2i).

Phosphorylation residue and nuclear export signal (NES) regions in HDAC6 phase separation

As depicted in Fig. 2a, HDAC6 contains two major IDRs with high PONDR (Predictor of Natural Disordered Regions) scores (>0.5), indicating a high probability of disorder. Importantly, we observed a reduction in phospho-HDAC6 phase separation upon deletion of IDR1 (ΔIDR1) compared to the full-length protein in HDAC6-knockout (KO) TNBCs (Fig. 2b and Extended Data Fig. 3a,b). Conversely, deletion of IDR2 (ΔIDR2) had no effect on phospho-HDAC6 phase separation. The IDR2 region contains the NES and its depletion might lead to an increased extent of nuclear localization of HDAC6, which may compensate for the potential decrease in ΔIDR2 phase separation ability.

**Fig. 2: The roles of IDRs, phosphorylation residue and conserved domains in the phase separation of phospho-HDAC6.**

To investigate whether the phosphorylation status of HDAC6 at S22 within IDR1 affects its phase separation ability, we expressed and purified recombinant proteins of IDR1 wild type (WT), IDR1 S22F (mimicking the nonphosphorylated state) and IDR1 S22E (mimicking the hyperphosphorylated state). Compared to the IDR1 WT control, IDR1 S22F exhibited reduced phase separation ability, while the hyperphosphorylated counterpart (IDR S22E) enhanced IDR1 phase separation ability (Fig. 2c–f). All the IDR1 droplets (WT, S22E and S22F) were sensitive to 1,6-hexanediol and could be disrupted (Extended Data Fig. 3c). Furthermore, the overexpression of HDAC6-S22E significantly enhanced the proliferation, sphere formation and invasion of TNBCs in vitro and promoted tumor growth in vivo (Extended Data Fig. 3d–g).

To investigate how conserved domains may regulate HDAC6 nuclear import, we generated a series of deletion mutants based on the deacetylase catalytic activity domain²⁵. We observed that deletion of the BUZ domain (ΔBUZ), DD1 domain (ΔDD1), DD2 domain (ΔDD2) or DD domain (ΔDD) did not impact the nuclear localization and phase separation of HDAC6 compared to the full-length control (Fig. 2g and Extended Data Fig. 3h). In contrast, deletion of both NES domains (ΔNES) resulted in a significant increase in nuclear localization and phase separation ability (Fig. 2h and Extended Data Fig. 3i). These findings indicate that NES has a dominant negative role in regulating nuclear HDAC6 protein levels and phase separation.

We further adopted functional assays to explore whether phospho-HDAC6 LLPS promotes tumorigenicity. TNBC cells overexpressing ΔNES had significantly stronger sphere-forming ability than those overexpressing full-length HDAC6 (Extended Data Fig. 3j,k). Moreover, we observed a further acceleration of in situ tumor growth in the ΔNES overexpression group compared to the marginal acceleration observed with full-length HDAC6 overexpression (Fig. 2i). In addition, tumors overexpressing ΔNES exhibited higher levels of nuclear phospho-HDAC6 phase separation, which positively correlated with the tumor growth marker Ki67 (Fig. 2j).

Disruption of HDAC6 LLPS inhibits TNBC growth through Nexturastat A

By quantifying the number of condensates within the nucleus, we identified Nexturastat A, a selective HDAC6 inhibitor²⁶, which effectively inhibited phospho-HDAC6 condensate formation (Fig. 3a,b). Whereas HPOB, another HDAC6 inhibitor²⁷, failed to inhibit LLPS formation (Fig. 3a,b). Phase separation droplet assays further demonstrated that HDAC6 IDR1 S22E puncta were sensitive to Nexturastat A but not to HPOB (Fig. 3c). Moreover, increasing concentrations of Nexturastat A led to a decrease in the level of phospho-HDAC6 protein, while total HDAC6 protein levels remained unaffected (Fig. 3d). This suppressive effect was absent in TNBCs where PPA1 (also known as PP1), a phosphatase that interacts with HDAC6, was depleted (Extended Data Fig. 4a–c). Likewise, cells exposed to 1,6-hexanediol displayed no such inhibition (Extended Data Fig. 4d). These findings suggest that Nexturastat A-induced HDAC6 dephosphorylation relies on the presence of PP1 and the context of LLPS. Consistent with this, condensate formation was substantially inhibited with increasing concentrations of Nexturastat A (Fig. 3e). Notably, Nexturastat A treatment swiftly disassembled the intranuclear phospho-HDAC6 condensates within a brief exposure period (Fig. 3f).

**Fig. 3: Identification of Nexturastat A as a selective inhibitor for phospho-HDAC6 LLPS.**

Nexturastat A exhibited potent anti-TNBC activity with minimal toxicity to noncarcinoma cells (Extended Data Fig. 4e). Dose-dependent assays demonstrated that Nexturastat A effectively inhibited TNBC cell proliferation (Fig. 3g), 3D sphere-forming ability (Fig. 3h), cell invasion (Extended Data Fig. 4f) and promoted apoptosis (Extended Data Fig. 4g,h). However, Nexturastat A exhibited no impact on apoptosis in HDAC6-KO cells that were overexpressing HDAC6 S22F (Extended Data Fig. 4i). Moreover, Nexturastat A treatment induced cell-cycle arrest in the G2/M phase in a concentration-dependent manner (Extended Data Fig. 4j). In both subcutaneous cell-derived xenograft (CDX) models and orthotopic patient-derived xenograft (PDX) models, Nexturastat A exhibited a significant inhibition of TNBC tumor growth (Fig. 3i,j and Extended Data Fig. 4k). This inhibition correlated with a decrease in the number of phospho-HDAC6 condensates within the nuclei of TNBC PDX tumor cells (Fig. 3k and Extended Data Fig. 4l). Notably, Nexturastat A did not affect animal body weight, indicating its high safety profile (Extended Data Fig. 4m).

HDAC6 knockdown suppressed TNBC proliferation (Extended Data Fig. 4n) and abolished the anti-TNBC effects of Nexturastat A on sphere-forming ability and cell invasion (Extended Data Fig. 4o,p). Accordingly, HDAC6 depletion did not further increase apoptosis in TNBC cells treated with Nexturastat A (Extended Data Fig. 4q). The specificity of Nexturastat A was confirmed by an in situ TNBC tumor model, where interference with HDAC6, the target of Nexturastat A, did not further enhance tumor growth inhibition (Fig. 3l). These findings were consistent with a reduced propensity of phospho-HDAC6 to undergo phase separation (Fig. 3m).

Mechanisms of phospho-HDAC6 nuclear translocation and LLPS

To gain further insight into the components of phospho-HDAC6 condensates, we used mass spectrometry (MS) analysis to investigate HDAC6-interacting proteins and their interaction networks in TNBC cells with or without treatment with Nexturastat A (Extended Data Fig. 5a). Our findings unveiled a notable increase in the interaction between HDAC6 and α-tubulin (Extended Data Fig. 5b,c), coupled with a reduction in HDAC6 interactions with members of the 14-3-3 protein family, particularly 14-3-3θ, subsequent to treatment with Nexturastat A (Fig. 4a–c and Extended Data Fig. 5d). The 14-3-3 proteins are predicted to be potential regulators of LLPS²⁸ and their silencing affected the phase separation formation of phospho-HDAC6 in TNBCs (Fig. 4d and Extended Data Fig. 5e). We also found that the purified recombinantly expressed mCherry–14-3-3θ fusion protein mixed with HDAC6-S22E proteins formed phase-separated droplets (Fig. 4e,f), whereas mixing 14-3-3θ with either HDAC6 WT or HDAC6-S22F resulted in much lower droplet formation (Extended Data Fig. 5f–h). Furthermore, Nexturastat A effectively inhibited the condensate of mixed proteins and blocked the colocalization of 14-3-3θ (Fig. 4g) and phospho-HDAC6 in the nucleus (Fig. 4h). Thus, we conclude that 14-3-3θ facilitates the phospho-HDAC6 LLPS condensates in the nucleus of TNBCs.

**Fig. 4: Components of phospho-HDAC6 LLPS condensates.**

We also detected an interaction between HDAC6 and importin-β, a nucleocytoplasmic transporter²⁹. Remarkably, Nexturastat A treatment led to a significant decrease in the interaction between phospho-HDAC6 and importin-β (Fig. 4i). This interaction between HDAC6 and importin-β relies on the S22 residue, which is located within the nuclear localization signal region (Fig. 4j). Importantly, we found that knockdown of importin-β restricts the translocation of phospho-HDAC6 into the nucleus while having no effect on the nonphosphorylated form of HDAC6 (Fig. 4k and Extended Data Fig. 5i,j). Through short hairpin RNA (shRNA) screening, we identified NUP153 as a critical nuclear porin that mediates the nuclear entry of HDAC6 (Fig. 4l and Extended Data Fig. 5k).

Disruption of phospho-HDAC6 LLPS alters chromatin structure

Given the significant alterations in chromatin architecture observed in MDA-MB231 cells following Nexturastat A treatment (Extended Data Fig. 6a), we aimed to explore the distinct roles of phospho-HDAC6 condensates in chromatin organization. To accomplish this, we performed Hi-C analyses on TNBC cells treated with either dimethyl sulfoxide (DMSO) or Nexturastat A to visualize genome-wide chromatin interaction sites as chromosome heat maps at a resolution of 10 kb (Fig. 5a,b and Extended Data Fig. 6b). Principal component analysis (PCA) of the Hi-C contact matrices revealed active A and inactive B compartments in both groups (Extended Data Fig. 6c). We also observed that compartment A had a higher G+C content than compartment B (Extended Data Fig. 6d). Disrupting phospho-HDAC6 LLPS led to significant changes in the distribution of open and closed compartments, with some regions exhibiting a change in genomic compartmentalization from type A to B (5%) and vice versa (9%) (Fig. 5c,d and Extended Data Fig. 6e). Our gene ontology (GO) pathway analysis revealed that closed compartments were strongly associated with the lipoprotein metabolic process (Fig. 5e), while open compartments were closely associated with adaptive immune responses (Fig. 5f). Furthermore, our investigation found that treatment of TNBC cells with Nexturastat A resulted in the accumulation of lipids (Extended Data Fig. 6f). Subsequent targeted lipidomics analysis was conducted to quantitatively assess the levels of specific lipids in the Nexturastat A-treated samples (Fig. 5g). In immunocompetent TNBC mouse models, the administration of Nexturastat A led to a reduction in tumor growth (Fig. 5h), which coincided with a decrease in phospho-HDAC6 condensates (Fig. 5i). Importantly, an increase in the number of CD68⁺ macrophages and PD1⁺ cells was observed within the Nexturastat A-treated groups (Fig. 5j and Extended Data Fig. 6g).

**Fig. 5: Genome-wide chromatin interaction changes in response to Nexturastat A treatment in TNBC cells.**

Disruption of phospho-HDAC6 LLPS by Nexturastat A treatment led to the identification of 4,350 newly formed topologically associating domain (TAD) boundaries (Extended Data Fig. 6h). These newly formed TAD boundaries were classified into four categories: stable, merge, split and rearrangement (Extended Data Fig. 6i,j). To investigate the critical transcription factors (TFs) involved in the formation of TAD boundaries, we analyzed the top ten motifs for TFs in both the DMSO-treated and the Nexturastat A-treated groups. Our analysis revealed that the Nexturastat A-treated group was enriched with motifs for TFs such as TEAD2 and TBX21 (Extended Data Fig. 6k). Notably, both groups contained CTCF motifs, which is consistent with the critical role of CTCF as a chromatin structural protein mediating domain-boundary insulation³⁰.

Nexturastat A treatment resulted in the identification of 33,517 newly formed chromatin loops (Fig. 5k,l) that were enriched in Pol II-specific DNA binding and immune system process pathways (Extended Data Fig. 6l). Through differential analysis, we identified 2,003 loops exclusive to the DMSO control and 4,627 specific to the Nexturastat A-treated samples (Fig. 5m). Notably, enhancer regions encompassing SEs and typical enhancers (TEs), characterized by a high occupancy of H3K27ac, exhibited heightened interaction frequencies. Intriguingly, these interactions were perturbed by the administration of Nexturastat A (Fig. 5n). As illustrated in Fig. 5o, we observed the formation of new chromatin interactions and disappearance of loops at specific chromatin sites. To assess whether the changes in large-scale chromosomal interactions and genomic compartments observed between Nexturastat A and control genomes impacted chromatin accessibility and histone modification, we performed an assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq) and chromatin immunoprecipitation followed by sequencing (ChIP-seq), including H3 pan-acetylation (pan-acetyl), HDAC6 and RNA Pol II. Our analysis showed an increase in loop anchor overlapping with H3 pan-acetyl, ATAC peaks and Pol II binding sites upon Nexturastat A treatment, while a decrease was observed in loop anchors overlapping with HDAC6 binding sites (Fig. 5p). We also found that genes with newly formed loop anchors following Nexturastat A treatment displayed increased expression compared to control (Fig. 5q and Extended Data Fig. 6m).

Transcriptional reprogramming induced by Nexturastat A treatment

To investigate the effects of phospho-HDAC6 LLPS on transcription in TNBC, we conducted an analysis of ATAC-seq and Pol II binding patterns in TNBC cells treated with or without Nexturastat A (Fig. 6a and Extended Data Fig. 7a). We observed that disruption of phospho-HDAC6 LLPS mainly led to increased chromatin accessibility, which was correlated with increased Pol II binding (Fig. 6b and Extended Data Fig. 7b). Specifically, we identified 13,048 genes associated with overlapping increased peaks (designated as gene set 1) and 12,173 genes associated with overlapping decreased peaks (gene set 2) upon Nexturastat A treatment compared with mock treatment (Fig. 6c). Notably, the gene set 1 peaks were predominantly localized in gene regulatory regions, including introns (35.9%), promoters (34.4%) and distal intergenic regions (17%) (Fig. 6d). It is worth mentioning that Nexturastat A treatment resulted in transcription-pausing release events during the transcription of genes in gene set 1 (Fig. 6e). In addition, we conducted ATAC-seq and RNA Pol II ChIP-seq experiments in HDAC6-depleted cells and integrated these datasets with those from cells treated with Nexturastat A, revealing a similar pattern (Extended Data Fig. 7c).

**Fig. 6: Nexturastat A treatment reprograms chromatin accessibility and transcriptional activity in TNBC cells.**

To further elucidate the role of phospho-HDAC6 LLPS in transcriptional regulation, we performed RNA sequencing (RNA-seq) experiments in TNBC cells treated with Nexturastat A, HPOB, TSA or HDAC6 knockdown. The transcriptome analysis unveiled a relatively modest number of differential genes following HPOB or Tubastatin A treatment, contrasting the extensive changes observed subsequent to Nexturastat A intervention (Extended Data Fig. 7d,e). Our results revealed upregulation of tumor suppressor genes, downregulation of proto-oncogenes and activation of immunomodulators expressed in cancer cells upon disruption of phospho-HDAC6 LLPS (Extended Data Fig. 7f). Additionally, gene set enrichment analysis (GSEA) revealed the activation of the antigen processing and presentation pathway, along with the tumor necrosis factor-α signaling pathway through nuclear factor κB, in both the Nexturastat A treatment and HDAC6 knockdown groups (Extended Data Fig. 7g,h).

By integrating data from ATAC-seq, Pol II ChIP-seq and RNA-seq (Fig. 6f), we were able to identify upregulation of a set of tumor suppressor genes downstream of TP53 (Fig. 6g), including ATF3 (Fig. 6h). Conversely, a set of proto-oncogenes were downregulated (Fig. 6g). Moreover, we observed upregulation of several immunomodulators, such as MHC-I and CD274 (also known as PD-L1) (Fig. 6g). Notably, we detected reshaped chromatin interaction in proto-oncogenes (for example, CAV1 and FOXQ1)^31,32 (Fig. 6i), tumor suppressor genes (for example, CDKN1A)³³ and immunomodulator genes (for example, PD-L1)³⁴ (Fig. 6j) after Nexturastat A treatment.

ATF3 as one of the downstream targets of phospho-HDAC6 LLPS

To further elucidate the role of phospho-HDAC6 condensates in modulating histone modification, we performed a comparative analysis of HDAC6 binding and H3 pan-acetyl modification profiles in TNBC cells treated with or without Nexturastat A (Extended Data Fig. 8a). Our results revealed a strong correlation between the HDAC6 binding profile and H3 pan-acetyl modification profile (Fig. 7a), suggesting that the alterations in HDAC6 binding profile induced by disrupting phospho-HDAC6 LLPS may directly trigger changes in H3 pan-acetyl modification profile. Notably, Nexturastat A treatment induced significant changes in the HDAC6 binding and H3 pan-acetyl modification profiles primarily around the transcription start site (TSS) (Fig. 7b and Extended Data Fig. 8b) and most of the HDAC6 and H3 pan-acetyl binding sites were annotated to the promoter (Extended Data Fig. 8c).

**Fig. 7: *ATF3* as a downstream target of the phospho-HDAC6 LLPS condensates in TNBCs.**

Next, we aimed to identify genes that are directly affected by phospho-HDAC6 LLPS. For this purpose, we identified 9,806 genes that exhibited consistent changes in both HDAC6 binding and H3 pan-acetyl upregulation upon treatment with Nexturastat A (Fig. 7c). Further analysis of these genes revealed their enrichment in the endoplasmic reticulum stress pathway and the TP53 pathway (Fig. 7d), which includes the tumor suppressor gene ATF3 (Fig. 7e). Additionally, the typical tumor suppressor gene NDRG1, which is also a downstream effector of TP53 (ref. ³⁵), demonstrated consistent changes in HDAC6 binding and H3 pan-acetyl modification profiles (Extended Data Fig. 8d). Analysis of data from The Cancer Genome Atlas (TCGA) and the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) databases revealed that ATF3 was expressed at higher levels in persons with TNBC than in those with non-TNBC (Extended Data Fig. 8e). Furthermore, ATF3 expression decreased as TNBC stage increased (Extended Data Fig. 8f) and higher expression of ATF3 correlated with longer survival in persons with TNBC (Fig. 7f).

To ascertain whether ATF3 is a specific downstream gene of HDAC6, we performed an ATF3 knockdown experiment in MDA-MB231 cells (Extended Data Fig. 8g). The results showed that the ATF3 knockdown group exhibited compromised responses to Nexturastat A treatment in terms of cell viability, apoptosis, invasion and sphere formation (Fig. 7g–j and Extended Data Fig. 8h–j). We further conducted an in vivo orthotopic transplantation of BT-549 cells with or without ATF3 knockdown (Extended Data Fig. 8k). Interestingly, Nexturastat A significantly disrupted the phase formation of phospho-HDAC6 in both xenograft tumors with and without ATF3 knockdown (Fig. 7k). However, knockdown of ATF3 led to increased malignancy and reduced sensitivity to Nexturastat A treatment (Fig. 7l,m).

Discussion

The impact of histone acetylation on chromatin dynamics has been extensively investigated³⁶, yet the contribution of dysregulated histone acetylation modifiers to aberrant chromatin structure in TNBC remains poorly understood. Herein, we report the identification of unexpected phase-separated condensates of phospho-HDAC6 within transcriptionally active regions of chromatin in TNBC. Our study demonstrates that phospho-HDAC6 cocondensates with 14-3-3θ, driving aberrant chromatin architecture, distinctive histone acetylation modifications and reprogrammed transcriptional profiles, ultimately resulting in malignant phenotypes in TNBC (Fig. 7n).

Although HDAC6 is a well-established deacetylase with diverse functions in both the cytoplasm and the nucleus, the role of its phosphorylated version remains unclear. A previous study found phospho-HDAC6 in Papp–Lantos bodies in persons with atypical parkinsonisms, suggesting a potential involvement of phospho-HDAC6 in mediating the formation of aggresomes in Parkinson disease¹⁹. Our investigation revealed that phospho-HDAC6 is abundantly distributed in the nucleus of TNBCs. Furthermore, we observed prominent condensate formation of phospho-HDAC6 in the nuclei of TNBC cell lines but much less in luminal carcinoma cell lines and little in breast epithelial cells, suggesting a crucial threshold of HDAC6 phosphorylation for the initiation and progression of LLPS. This LLPS formation of phospho-HDAC6 in the nucleus of TNBC cells may enable prolonged deacetylase activity of HDAC6, thus sustaining oncogenic chromatin looping states and promoting tumorigenesis.

Additionally, previous research showed that acetylation is an important PTM that regulates HDAC6 nuclear import³⁷. Our investigation unveiled that phosphorylation at the S22 residue can also act as a switch for HDAC6 translocation into the nucleus of TNBC in an importin-β-dependent and NUP158-dependent manner. Nexturastat A blocks the interaction between HDAC6 and importin-β by preventing the phosphorylation of HDAC6 at S22. This disruption in interaction subsequently leads to the failure of HDAC6 to pass through the NPC and enter the nucleus. Moreover, S22 phosphorylation proves to be a pivotal step for HDAC6 LLPS by involving 14-3-3θ. The action of Nexturastat A entails disrupting phospho-HDAC6 through the interference of the interaction between phospho-HDAC6 and 14-3-3θ. This, in turn, results in the release of phospho-HDAC6 molecules from the condensed phase into the dilute phase. Consequently, these released phospho-HDAC6 molecules can be subjected to dephosphorylation through the activity of PP1.

We elucidated the biochemical mechanism underlying the formation of phospho-HDAC6-mediated LLPS in a stepwise manner. Previous studies demonstrated that EGFR activation induces the translocation of GRK2 to the plasma membrane, where it can phosphorylate HDAC6 in breast cancer²⁴, although the specific phosphorylation sites remain to be defined. The EGF–GRK2-mediated pathway controls HDAC6 S22 phosphorylation in TNBCs. We propose that LLPS occurs spontaneously when the core components reach a threshold concentration. For instance, P granules dissolve in the perinuclear region and recondense in the cytoplasm, which is regulated by a concentration gradient³⁸. Likewise, the segregation of phospho-HDAC6 puncta in the TNBC nucleus may also be induced by the concentration gradient.

We deciphered the sequential mechanism underlying the regulation of chromatin architecture by phospho-HDAC6. Growing evidence supports the notion that histone modifications, particularly acetylation, can directly alter chromatin architecture³⁹. For instance, H3K56 acetylation promotes DNA unwrapping and regulates chromatin at higher-order levels⁴⁰. Our ChIP-seq analyses revealed that disruption of HDAC6 LLPS condensates directly affect pan-histone acetylation of HDAC6 target genes, which in turn may affect chromatin architecture. Intriguingly, previous studies showed that HDACs are enriched at heterochromatin and repressive genomic regions⁴¹. However, we demonstrated that HDAC6 LLPS condensates contain H3K27ac deposition and the active form of Pol II, suggesting that HDAC6 condensates not only suppress the transcriptional program of tumor suppressor genes but also enhance the assembly of the transcription machinery to activate oncogenic programs.

The role of ATF3 in cancer is characterized by its intricate and multifaceted nature. Within the context of non-TNBC, ATF3 has been documented to display oncogenic tendencies, driving metastasis as validated by pertinent literature. For instance, ATF3 enhances cell proliferation, mobility and invasiveness in the context of malignant MCF10CA1a breast cancer cells⁴². Conversely, the role of ATF3 in TNBC has yielded conflicting outcomes across earlier reports. These divergent findings can be ascribed to the distinct cellular context, where ATF3’s functional dynamics are profoundly shaped by its transcriptional targets and interacting partners. As a result, comprehending ATF3’s role within the framework of phospho-HDAC6-mediated tumorigenesis in TNBC necessitates a more thorough investigative approach.

The discovery of aberrant phase separation in cancers has the potential to impact drug discovery and create clinical opportunities. In this study, we report the identification of a highly specific HDAC6 inhibitor, Nexturastat A, which selectively disrupts the formation of phospho-HDAC6 condensates and subsequently suppresses the growth of TNBC. While pan-HDAC inhibitors have been tested in clinical settings, their nonselective nature and adverse effects, such as cardiac toxicity, have limited their use in combination therapies⁴³. The specificity of Nexturastat A as a molecularly targeted agent for TNBC is likely achieved because there is little phosphorylation and LLPS of HDAC6 in non-TNBC and nontransformed cells. Multiple mechanisms may contribute to the TNBC inhibition events of priming by Nexturastat A, such as upregulation of ATF3-induced apoptosis, suppression of oncogenic pathways and modulation of immune components. Studies have shown that Nexturastat A has minimal cytotoxic effects and can improve anti-PD1 antitumor immune responses¹⁴. Moreover, a phase 1b clinical trial (NCT02632071) demonstrated that combining an HDAC6 inhibitor (ricolinostat) with other chemotherapeutic drugs can safely treat HR⁺HER2⁻ breast cancer¹⁵. These findings suggest that combining Nexturastat A with chemomodulatory and immunomodulatory agents could be a promising and powerful approach to treating TNBC.

Methods

Our research complies with all relevant ethical regulations. Research involving animal experiments with all relevant ethical regulations was approved by the Ethics Committee of Guangdong Provincial People’s Hospital (Guangdong Academy of Medical Sciences) (KY2023-1038-02). Research involving human samples with all relevant ethical regulations was approved by the Ethics Committee of Guangzhou Women and Children’s Medical Center (2023, no. 046A01). All participants gave written informed consent.

Cell culture and clinical samples

BT-549 (American Type Culture Collection (ATCC), HTB-122) and MDA-MB231 (ATCC, HTB-26) cell lines were cultured in DMEM supplemented with 10% FBS and 1% penicillin–streptomycin (P/S). MCF-10A (ATCC, CRL-10317) cells were maintained in DMEM/F12 medium supplemented with 5% horse serum, 20 ng ml⁻¹ EGF, 0.5 μg ml⁻¹ hydrocortisone, 100 ng ml⁻¹ cholera toxin, 10 μg ml⁻¹ insulin and 1% P/S. MCF-7 (ATCC, HTB-22) cells were cultured in RPMI-1640 medium supplemented with 10% FBS and 1% P/S. The 293T cells (Cell Bank/Stem Cell Bank, Chinese Academy of Sciences, GNHu17) were maintained in DMEM supplemented with 10% FBS and 1% P/S. All cell lines were incubated at 37 °C in a humidified atmosphere containing 5% CO₂ and subcultured every 2–3 days. All cells were routinely checked for Mycoplasma contamination (MycoAlert, Lonza). Clinical specimens were collected (in February 2023) from female participants at the Department of Breast and Thyroid Surgery, Guangzhou Women and Children’s Medical Center, after obtaining informed consent (Ethics Committee, 2023, no. 046A01).

Constructs of shRNA and cell transfection

To generate shRNA vectors for silencing the target genes, we used an online tool (http://rnaidesigner.thermofisher.com/rnaiexpress/) to design shRNA sequences that target specific regions of the genes of interest. Oligonucleotides with the appropriate restriction sites were synthesized and annealed to form double-stranded DNA, which was then ligated into the pLKO-puro lentiviral vector (Addgene, 8453). The shRNA vectors were verified by restriction digestion and DNA sequencing.

To produce lentiviral particles, we cotransfected the shRNA expression plasmids and overexpression plasmid with the packaging plasmids pVSV-G and PSPAX2 into 293T cells using Lipofectamine 2000 (Invitrogen, cat. no. 11668019). After 48 h, we collected the viral supernatants and used them to infect the cells of interest. The infected cells were then treated with culture medium containing 2 mg l⁻¹ puromycin (Solarbio, cat. no. P8230) for 3 days to select for cells that stably expressed the shRNA.

The shRNA sequences and qPCR primers used in this study are listed in Supplementary Tables 2 and 3.

PDX models of TNBC

Fresh tumor tissues were procured from persons with breast cancer who had undergone surgical resection at Guangzhou Women and Children’s Medical Center. Written informed consent was obtained from the participants and the study received approval from the Ethics Committee of Guangzhou Women and Children’s Medical Center (Ethics Committee, 2023, no. 046A01). These tumor tissues were sectioned into small fragments (2–3 mm³) and subsequently subcutaneously implanted into the mammary fat pads of female B-NDG mice (4–8 weeks old) procured from Zhuhai BesTest Bio-Tech. All mice were housed under specific pathogen-free conditions with normal mouse chow in housing rooms at 22 °C and 40–70% humidity in 12-h cycles of light–darkness.

Maintained within a specific pathogen-free environment, the mice were regularly monitored for tumor progression, assessing growth twice weekly. The maximum tumor diameter of 20 mm approved by the Laboratory Animal Care Committee of Guangdong Provincial People’s Hospital (Guangdong Academy of Medical Sciences) was not exceeded. When the tumor volume reached a range of 500–1,000 mm³, the mice were humanely killed and the tumors were isolated. The isolated tumors were either rapidly frozen using liquid nitrogen for molecular analyses or were further transplanted into new mice to facilitate subsequent drug treatments. At the end of the treatment period, the mice were humanely killed and their tumors were collected for subsequent histological examination.

Animal studies

Female BALB/c nude mice (18 ± 2 g, 4–6 weeks old) purchased from GemPharmatech were used in the study. All animal experiments were approved by the Institutional Animal Care and Use Committee of Guangdong Provincial People’s Hospital (Guangdong Academy of Medical Sciences) (KY2023-1038-02) and were performed in accordance with the relevant guidelines and regulations. BT-549 cells, stably expressing different genes, were established by lentiviral infection and selection with puromycin. To establish the tumor model, 5 × 10⁶ cells in 100 μl of PBS were injected into the mammary fat pad of each mouse. The mice were randomly assigned to different experimental groups (n = 6 per group). In the medication groups, Nexturastat A was administered by intraperitoneal injection every 2 days. Tumor growth was monitored daily by measuring the width (W) and length (L) with a caliper and tumor volume (V) was calculated as V = (W² × L)/2.

The maximum tumor size permitted by the ethics committee is 2,000 mm³, with no single dimension exceeding 20 mm. In this study, all animal tumor-bearing experiments adhered to these guidelines, ensuring that the tumors did not exceed the specified limits.

LLPS assay of proteins

To induce LLPS, purified proteins were added to a buffer containing 50 mM Tris-HCl pH 7.4, 10% glycerol, 1 mM DTT and 5% PEG-8000. The concentration of NaCl was adjusted to the indicated concentrations. The protein solution was incubated for 5 min at room temperature, followed by imaging with a Zeiss LSM 800 microscope. Specifically, a droplet of the protein solution was placed onto a clean glass slide and covered with a coverslip. The slide was then mounted onto the microscope stage for imaging. The resulting droplets were imaged using a ×63 oil-immersion objective with excitation wavelengths of 488 nm for mEGFP and 561 nm for mCherry.

FRAP

FRAP experiments were conducted using an LSM880 Airyscan microscope with a 488-nm laser. A bleach spot was positioned at the center of a cluster and images were captured with a 1-s interval for 100 s to monitor the fluorescence recovery within the cluster. The integrated intensity of the cluster was quantified over time by subtracting the background intensity from an adjacent region of equal size, correcting for overall photobleaching using a reference region within the same cell and normalizing the prebleach fluorescence intensity to 1.

Generation of phase diagram

Phase diagrams were generated by mixing HDAC6-S22F-mEGFP and 14-3-3θ proteins separately or together at concentrations ranging from 5 to 320 μM in a phase separation buffer with varying sodium chloride concentrations from 0 to 500 mM. Droplets were detected using a Zeiss LSM 800.

In situ Hi-C library construction and Hi-C read processing

In situ Hi-C library construction was performed following a previously published protocol⁴⁴. We processed the Hi-C reads using a standard pipeline. First, we assessed the quality of the raw reads using FastQC (version 0.11.9) and removed low-quality reads and adaptors using Trim Galore (version 0.6.7) with default settings. The clean Hi-C reads were then iteratively mapped to the human genome using HiC-Pro (version 3.1.0). We identified chromatin A/B compartments using PCA of the normalized Hi-C matrices at 100-kb resolution, with the PC1 value of each bin defining the A (positive score) and B (negative score) compartments. We determined the location and number of TADs and identified TAD borders using an insulation score algorithm based on normalized contact matrices at 40-kb resolution. We transformed the valid pair files to .hic files, which were used for further analysis with Juicer (version 1.11.04). We used Ay’s Fit-Hi-C software (version 1.0.1) to determine the intrachromosome and interchromosome interactions at 10-kb resolution, with interactions having a P value < 0.01 with a false discovery rate < 0.01 and a contact count > 2 considered significant. Finally, we visualized the results using Juicebox (version 1.11.08).

Preparation of the nucleosome chromatin with H3K27ac modification

To produce H3K27ac, we induced acetylation of C27 within the H3K27C C110A framework using thioacetic acid. The resulting proteins were expressed in Escherichia coli and subsequently purified following the procedures outlined in Supplementary Methods. The presence of acetylated K27 on H3K27 was confirmed through immunoblot assays using an anti-histone H3K27ac antibody. For the preparation of the 6× 177-bp 601 DNA template, we followed established methods using a plasmid containing DNA with 6× 177-bp tandem repeats of the Widom 601 sequence. Histone octamers comprising H2A, H2B, H3K27ac and H4 were assembled through successive dialysis steps. The DNA templates, containing 6× 177-bp tandem repeats of the Widom 601 sequence, were combined with histone octamers at a 6:1 molar ratio and the resulting reconstituted nucleosome chromatin underwent further dialysis⁴⁵.

DNaseI terminal deoxynucleotidyl transferase deoxyuridine triphosphate (dUTP) nick end labeling (TUNEL) assay for chromatin architecture analysis

To evaluate alterations in chromatin architecture induced by Nexturastat A treatment in MDA-MB231 cells, we conducted the DNaseI TUNEL assay. This approach combines DNA fragmentation through DNase I action and subsequent labeling using TdT, representing a modified version of the TUNEL assay. Briefly, MDA-MB231 cells were exposed to Nexturastat A (10 μM) or the vehicle control, DMSO, for a duration of 24 h. Following treatment, cells were fixed with 4% paraformaldehyde for 15 min at room temperature and then permeabilized using 0.1% Triton X-100 for 10 min on ice. After two PBS washes, cells were subjected to DNase I (2 U per ml) in DNase I reaction buffer (40 mM Tris pH 8, 10 mM MgSO₄ and 1 mM CaCl₂) for 10 min at room temperature. Subsequently, TdT enzyme (300 U per ml) and FITC–dUTP (0.5 μM) were introduced into the cells in TdT reaction buffer (30 mM Tris pH 7.2, 140 mM sodium cacodylate and 1 mM CoCl₂), followed by a 60-min incubation at 37 °C in darkness. The reaction was quelled by introducing 2× SSC buffer (300 mM NaCl and 30 mM sodium citrate; pH 7) and two PBS washes were performed. Nuclei were counterstained with DAPI (1 μg ml⁻¹) for 5 min at room temperature and the coverslips were mounted onto glass slides using antifade mounting medium. Fluorescence signals were captured using a confocal microscope, with subsequent image analysis carried out using ImageJ software.

RNA-seq and data analysis

RNA-seq was performed following previously established protocols⁸. RNA-seq reads were processed using standard procedures. Quality assessment was performed using FastQC (version 0.11.9) and low-quality reads and adaptors were removed using Trim Galore (version 0.6.7) with default settings. The filtered reads were then aligned to the human reference genome (GRCh38) using STAR (version 2.7.9a) with default settings. Gene-level read counts were obtained using featureCounts (version 2.0.3). Differential expression analysis was performed using DESeq2 (version 1.30.1) with a significance threshold of adjusted P value < 0.05 and |log₂ fold change (FC)| > 1. GSEA (version 4.1.0) was performed using the Molecular Signatures Database (MSigDB) with default parameters. Results were visualized using Integrative Genomics Viewer (IGV; version 4.1.0), GSEA (version 4.1.0) and ggplot2.

ChIP-seq and ATAC-seq

ChIP-seq and ATAC-seq were performed following previously established protocols⁸. ChIP-seq reads were processed as follows: Raw reads were first evaluated for quality using FastQC (version 0.11.9) and subsequently filtered for low-quality reads and adapters using Trim Galore (version 0.6.7) with default parameters. The filtered reads were then aligned to the human reference genome (GRCh38) using Bowtie 2 (version 2.4.4) with default settings. The resulting BAM files were sorted and indexed using SAMtools (version 1.13) and subsequently converted to bigWig files using deepTools (version 3.5.1) for visualization in IGV (version 4.1.0). Peaks were called using MACS3 (version 3.0.0a6) with a q-value cutoff of 0.01. The resulting peak files were annotated with gene information using ChIPseeker (version 1.28.2) and used for downstream analysis, including GSEA. The results were visualized using R packages, including ggplot2 and deepTools.

Tumor suppressor genes and proto-oncogenes sets

The tumor suppressor gene set of breast cancer was downloaded from the tumor suppressor gene database (https://bioinfo.uth.edu/TSGene/) and the proto-oncogene set was downloaded from the oncogene database (https://bioinfo-minzhao.org/index.html).

Image analysis with HALO image system

We used the HALO image analysis software (version 2.3.2089.34, IndicaLabs) for local slide image analysis. This involved differentiating between the paracancerous and tumor regions in persons with TNBC or non-TNBC. The intensity of HDAC6 and phospho-HDAC6 was quantified using the software, which allowed for the separation of the nucleus from the whole cell by detecting the stained nuclei. Additionally, the degree of intensity was defined.

Single-molecule tracking

Single-molecule tracking was conducted using HIS-SIM to track the movement of HDAC6-S22E–mEGFP droplets within the nucleus of cells. Cells expressing HDAC6-S22E–mEGFP were imaged under the same lattice illumination parameters for a total exposure of 100 s. The mean molecular speeds of HDAC6-S22E–mEGFP droplets within a single cell were calculated using Imaris, a microscopy image analysis software.

Cell proliferation assay

Cells were seeded at a density of 5 × 10³ cells per well in 96-well plates and treated with nine different concentrations of Nexturastat A (ranging from 0.20 μM to 50 μM) or DMSO for 48 h. Cell proliferation was assessed using the CellTiter-Glo assay (Promega, cat. no. G7571) according to the manufacturer’s instructions. Briefly, CellTiter-Glo reagent was added to each well and incubated at room temperature for 10 min. The luminescence signal was then measured using an automated microplate reader (BioTek). Cell viability was calculated as a percentage of the DMSO-treated control group.

Cell invasion assay

Cells were seeded at a density of 3 × 10⁴ cells per well in the top chamber of Transwell (Falcon) inserts coated with Matrigel (Corning, cat. no. 354234) and treated with Nexturastat A (8 μM) or DMSO control for 48 h. The lower chamber was filled with medium containing 10% FBS (ExCell, cat. no. FSP500) as a chemoattractant. After 24 h of incubation, noninvading cells on the upper surface of the membrane were removed with a cotton swab and the inserts were fixed with 4% paraformaldehyde and stained with 0.5% crystal violet. The number of invaded cells was counted by taking images of the inserts under a microscope and quantifying the cells using image analysis software.

3D spheroid formation assay

Cells were seeded at a density of 2 × 10³ cells per well in ultralow-attachment 96-well plates (Corning, cat. no. 3474) and treated with Nexturastat A (5 μM) or DMSO control for 48 h. After 5 days, the spheroids were visualized using an automated microscope or microplate reader (BioTek).

Immunoblot analysis

Total protein was extracted from cells by lysing them in radioimmunoprecipitation assay buffer (Thermo Scientific) supplemented with protease inhibitor cocktail (TOPSCIENCE) and phosphatase inhibitor tablets (Roche). For nucleoprotein extraction, cells were lysed in the nuclear and cytoplasmic protein extraction kit (Beyotime) with protease inhibitor cocktail (TOPSCIENCE) and phosphatase inhibitor tablets (Roche). The lysates were then separated by SDS–PAGE and transferred onto nitrocellulose membranes (Bio-Rad). The membranes were blocked for 1 h at room temperature in TBS-T buffer supplemented with 5% nonfat dried milk (Cell Signaling Technology) and incubated overnight at 4 °C with primary antibodies diluted in universal antibody diluent (New Cell and Molecular Biotech). After three washes with TBS-T, the membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies diluted in universal antibody diluent for 1 h at room temperature. Following another three washes with TBS-T, the membranes were developed using an ultrasensitive enhanced chemiluminescence detection kit (New Cell and Molecular Biotech) and imaged with ChemiDoc XRS+ (Bio-Rad). Antibodies used in this study included anti-phospho-HDAC6 (S22) (Affinity, cat. no. AF3485), anti-HDAC6 (D2E5) (Cell Signaling Technology, cat. no. 7558), anti-histone H3 (D1H2) (Cell Signaling Technology, cat. no. 4499) and anti-β-actin monoclonal antibody (Proteintech, cat. no. 66009-1-Ig) as the loading control. Anti-rabbit IgG, HRP-linked antibody (Cell Signaling Technology, cat. no. 7074) and anti-mouse IgG, HRP-linked antibody (Cell Signaling Technology, cat. no. 7076) were used as secondary antibodies.

qPCR with reverse transcription (RT–qPCR)

Total RNA was extracted from cells using TRIzol reagent (Ambion, cat. no. 15596018) according to the manufacturer’s protocol. One microgram of purified total RNA was reverse-transcribed into complementary DNA (cDNA) using the HiScript II Q RT SuperMix for qPCR (+gDNA wiper) (Vazyme, cat. no. R223). The levels of specific RNAs were quantified using a Bio-Rad real-time PCR machine and ChamQ Universal SYBR qPCR master mix (Vazyme, cat. no. Q711-02) following the manufacturer’s instructions. The primer sequences used are listed in Supplementary Table 3. The 2^−ΔΔCt method was used to normalize gene expression to the GAPDH expression level for cell lines.

Flow cytometry analysis of cell apoptosis and cell cycle

Cells were seeded at 30% confluency in six-well plates and treated with Nexturastat A (8 µM) or DMSO for an additional 48 h. After treatment, cells were harvested, washed with cold PBS and stained using an annexin V-APC/7-AAD apoptosis kit (MULTI SCIENCES, cat. no. AP105-100) as per the manufacturer’s protocol. For cell cycle, cells were collected and subjected to two washes with precooled PBS. The residual small volume of PBS was combined with the cell precipitate, followed by the addition of precooled 70% ethanol at −20 °C. The mixture was then incubated in a −20 °C refrigerator for a minimum duration of 4 h. Centrifugation was performed at 350g for 5 min. The cells underwent two additional washes with 2 ml of PBS. Then, 500 μl of PBS containing 50 μg ml⁻¹ propidium iodide and 100 μg ml⁻¹ RNase A (Thermo Scientific, cat. no. EN0531) was added and the mixture was incubated in the dark at 4 °C for 15–30 min. The stained cells were then analyzed using a CytoFLEX flow cytometer (Beckman Coulter) and the data were analyzed using CytExpert software (Beckman Coulter). The percentage of apoptotic cells was calculated as the early apoptotic stage (annexin V⁺/7-AAD⁻).

Immunofluorescence staining assay

Immunofluorescence staining was performed according to established protocols. Briefly, breast cancer cells were cultured in confocal dishes under 5% CO₂ and 37 °C and seeded at a density of 10,000 cells per dish for 24 h. The cells were then fixed with 4% paraformaldehyde in PBS at room temperature for 15 min, washed three times with cold PBS, permeabilized with 0.3% Triton X-100 in PBS for 15 min at room temperature and washed three times with PBS. The cells were blocked with goat serum for 60 min at room temperature and then incubated overnight at 4 °C in a humidified chamber with diluted primary antibodies: anti-HDAC6 (D2E5) rabbit monoclonal antibody (Cell Signaling, cat. no. 7558), anti-phospho-HDAC6 (S22) antibody (Affinity, cat. no. AF3485) and anti-14-3-3 (pan) antibody (Cell Signaling, cat. no. 8312). The corresponding secondary antibodies were Alexa Fluor 488-conjugated goat anti-mouse IgG (Thermo Scientific, cat. no. A11001), Alexa Fluor 594-conjugated goat anti-rabbit IgG (Thermo Scientific, cat. no. A11012) and Alexa Fluor 647-conjugated goat anti-rat IgG (Thermo Scientific, cat. no. A21247). After three washes with PBST (PBS containing 0.1% Tween-20) for 3 min each, the cells were incubated with the appropriate fluorescently labeled secondary antibodies for 2 h at room temperature and then washed three times with PBST for 3 min each. Finally, the nuclei were counterstained with DAPI (Thermo Scientific, cat. no. P36935).

Tumor samples were fixed and paraffin-embedded using standard methods. Tissue sections were deparaffinized, rehydrated and subjected to antigen retrieval using sodium citrate buffer (pH 6.0) at 95 °C for 20 min. After quenching endogenous peroxidase activity with 3% hydrogen peroxide in methanol, sections were blocked with 10% goat serum in PBS for 60 min at room temperature. Primary antibodies were then applied and incubated overnight at 4 °C. Following washing with PBST, sections were incubated with fluorescently labeled secondary antibodies for 2 h at room temperature and washed again with PBST.

Recombinant protein purification

To purify the proteins of interest, we constructed fusion proteins with mEGFP or mCherry and a His6 tag at the N terminus of each protein. The fusion proteins were His-mEGFP–HDAC6, His-mCherry–14-3-3, His-mEGFP–HDAC6-IDR1, His-mEGFP–HDAC6-IDR2 and His-mCherry–H3K27ac. The cDNA sequences of the fusion proteins were cloned into the pET28a vector (Novagen) using the In-Fusion HD cloning kit (Clontech) according to the manufacturer’s instructions. The fusion protein plasmids were verified by restriction digestion using NcoI and XhoI (New England Biolabs) and DNA sequencing by Sangon Biotech (Shanghai).

For protein expression, we transformed the fusion protein plasmids into E. coli BL21(DE3) cells (Novagen) and selected for kanamycin-resistant colonies. The colonies were inoculated into Luria–Bertani (LB) medium containing kanamycin (50 μg ml⁻¹) and grown at 37 °C with shaking until the optical density at 600 nm (OD₆₀₀) reached 0.6. The expression of the fusion proteins was induced by adding IPTG to a final concentration of 0.5 mM and incubating the cultures at 16 °C for 16 h. The cells were harvested by centrifugation at 4,000g for 15 min at 4 °C and resuspended in lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl and 10 mM imidazole; pH 8.0) supplemented with protease inhibitors (Roche). The cells were lysed by sonication on ice and the cell debris was removed by centrifugation at 12,000g for 30 min at 4 °C. The supernatants were collected and loaded onto Ni-NTA agarose columns (Qiagen) pre-equilibrated with the lysis buffer. The columns were washed with wash buffer (50 mM NaH₂PO₄, 300 mM NaCl and 20 mM imidazole; pH 8.0) and the fusion proteins were eluted with elution buffer (50 mM NaH₂PO₄, 300 mM NaCl and 250 mM imidazole; pH 8.0). The eluted fractions were analyzed by SDS–PAGE. The purified fusion proteins were dialyzed against PBS and stored at −80 °C until use.

Plasmid construction

To express HDAC6 and 14-3-3θ, we cloned these genes from human cDNA samples using primers with appropriate restriction sites and performed PCR amplification. The amplified fragments were then recombined into the PLVX-puro lentiviral vector using recombinase. To create tagged versions of these proteins for imaging studies, we replaced the original tags with mEGFP or mCherry using primers containing the appropriate sequences by overlap-PCR. For HDAC6, we also added a 3×HA (hemagglutinin) tag at the C terminus using primers containing the HA sequence by PCR, while for 14-3-3θ, we added a 3×Flag tag at the C terminus using primers containing the Flag sequence by PCR.

We also designed primers based on the nucleotide sequences of the HDAC6 gene to construct truncated mutants (IDR1 and IDR2) using the PCR products cloned into the PLVX-puro lentiviral vector or pet28a vector for mammalian or bacterial expression, respectively. Additionally, we constructed truncated mutants of HDAC6 based on the catalytic domain and nuclear localization signal by designing primers with the desired mutations and performing overlap-PCR for splicing. The resulting PCR products were cloned into the PLVX-puro lentiviral vector for stable expression in mammalian cells.

To simulate phosphorylation and nonphosphorylation point mutants of HDAC6 plasmids, we used primer design and overlap-PCR for site-directed mutagenesis. All plasmids were verified by DNA sequencing before use.

BODIPY 493/503 staining for neutral lipids

For the visualization of neutral lipids, oil and other nonpolar lipids, we used BODIPY 493/503 staining. The TNBC cells were treated with Nexturastat A (10 μM) or DMSO (vehicle control) for a duration of 24 h.

Following treatment, cells were fixed using 4% paraformaldehyde for 15 min at room temperature and then subjected to BODIPY 493/503 staining (5 μg ml⁻¹) for 30 min in darkness at room temperature. Nuclei were counterstained with DAPI (1 μg ml⁻¹) for 5 min at room temperature and coverslips were mounted on glass slides using antifade mounting medium. The visualization of fluorescence signals was achieved using a confocal microscope, while the images were subsequently analyzed using ImageJ software.

Targeted lipidomics analysis for specific lipids

To quantitatively evaluate the levels of specific lipids within Nexturastat A-treated samples, targeted lipidomics analysis was conducted. Cultured TNBC cells were treated with Nexturastat A (10 μM) or DMSO (vehicle control) for 24 h. Upon treatment completion, cells were harvested and lipid extraction was carried out using a modified Bligh and Dyer method. Before extraction, internal standards from Avanti Polar Lipids were added to each sample. Lipid extracts were dried using nitrogen gas and subsequently reconstituted in methanol. Liquid chromatography (LC)–MS analysis was performed using an Agilent 1290 Infinity II LC system coupled with an Agilent 6545 quadrupole time-of-flight MS instrument (Agilent Technologies). Separation was accomplished using an Agilent Zorbax Eclipse Plus C18 column (2.1 × 100 mm, 1.8 μm).

The mobile phases comprised water with 10 mM ammonium acetate and 0.1% acetic acid (A) and acetonitrile–isopropanol (9:1, v/v) with 10 mM ammonium acetate and 0.1% acetic acid (B). The gradient elution sequence was as follows: 0–2 min, 40% B; 2–12 min, 40–99% B; 12–15 min, 99% B; 15–16 min, 99–40% B; 16–20 min, 40% B. The flow rate was set at 0.4 ml min⁻¹, with an injection volume of 5 μL. Operating in positive ion mode with electrospray ionization, the MS instrument was configured with a capillary voltage of 4 kV and a nozzle voltage of 500 V. Gas parameters included a gas temperature of 250 °C, gas flow of 11 L min⁻¹, nebulizer pressure of 35 psi and sheath gas temperature of 350 °C with a sheath gas flow of 11 L min⁻¹. Data acquisition was conducted across a mass range of m/z 100 to m/z 1,700, using both full scan mode and targeted MS/MS mode with dynamic multiple reaction monitoring.

Data acquisition and analysis were performed using MassHunter Workstation software (Agilent Technologies). Identification and quantification of lipids relied on factors such as retention time, accurate mass and MS/MS spectra, facilitated by LipidSearch software (Thermo Scientific).

Statistics and reproducibility

No statistical method was used to predetermine sample size but our sample sizes are similar to those reported in previous publications^7,46,47,48. Data distribution was assumed to be normal but this was not formally tested. The statistical tests used are indicated in the accompanying figure legends. All comparisons between two groups were made using a two-tailed unpaired t-test, unless otherwise stated. P values less than 0.05 were defined as statistically significant. Exact P values are indicated within the figures. For the animal studies, mice were randomized from different cages and allocated to control and treatment groups. Immunohistochemistry and immunofluorescence images were acquired and analyzed in a blinded fashion. For other experiments, blinding was not used. Biological replicates for each experiment are noted in figure legends, with similar results. No data were excluded from the analyses. Plots and graphs were generated and analyzed with HALO image analysis software (version 2.3.2089.34, IndicaLabs), Operetta CLS high-content analysis system, GraphPad prism 8, Image J (version 1.51), Imaris (version 8.4), R (version 3.5.1), SPSS (version 22.0), FlowJo (version 10.8.1) and Microsoft Excel (version 16.0).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Hi-C, ChIP-seq, ATAC-seq and RNA-seq data that support the findings of this study were deposited to the Gene Expression Omnibus under accession codes GSE254213, GSE253743 and GSE253744. The human breast cancer data were derived from TCGA Research Network (http://cancergenome.nih.gov/) and METABRIC dataset. All other data supporting the findings of this study are available from the corresponding author on reasonable request. Source data are provided with this paper.

Code availability

No unique code was developed for this study.

References

Shi, Y., Jin, J., Ji, W. & Guan, X. Therapeutic landscape in mutational triple negative breast cancer. Mol. Cancer 17, 99 (2018).
Article PubMed PubMed Central Google Scholar
Garrido-Castro, A. C., Lin, N. U. & Polyak, K. Insights into molecular classifications of triple-negative breast cancer: improving patient selection for treatment. Cancer Discov. 9, 176–198 (2019).
Article CAS PubMed PubMed Central Google Scholar
Derakhshan, F. & Reis-Filho, J. S. Pathogenesis of triple-negative breast cancer. Annu. Rev. Pathol. 17, 181–204 (2022).
Article CAS PubMed PubMed Central Google Scholar
Kim, C. et al. Chemoresistance evolution in triple-negative breast cancer delineated by single-cell sequencing. Cell 173, 879–893 (2018).
Article CAS PubMed PubMed Central Google Scholar
Shu, S. et al. Response and resistance to BET bromodomain inhibitors in triple-negative breast cancer. Nature 529, 413–417 (2016).
Article CAS PubMed PubMed Central Google Scholar
Mehta, S. & Zhang, J. Liquid–liquid phase separation drives cellular function and dysfunction in cancer. Nat. Rev. Cancer 22, 239–252 (2022).
Article CAS PubMed PubMed Central Google Scholar
Ahn, J. H. et al. Phase separation drives aberrant chromatin looping and cancer development. Nature 595, 591–595 (2021).
Article CAS PubMed PubMed Central Google Scholar
Lu, B. et al. Pharmacological inhibition of core regulatory circuitry liquid–liquid phase separation suppresses metastasis and chemoresistance in osteosarcoma. Adv. Sci. 8, e2101895 (2021).
Article Google Scholar
Chiu, Y.-P. et al. Liquid–liquid phase separation and extracellular multivalent interactions in the tale of galectin-3. Nat. Commun. 11, 1229 (2020).
Article CAS PubMed PubMed Central Google Scholar
Shi, B. et al. Phase separation of Ddx3xb helicase regulates maternal-to-zygotic transition in zebrafish. Cell Res. 32, 715–728 (2022).
Article CAS PubMed PubMed Central Google Scholar
Sadar, M. D. Discovery of drugs that directly target the intrinsically disordered region of the androgen receptor. Expert Opin. Drug Discov. 15, 551–560 (2020).
Article CAS PubMed PubMed Central Google Scholar
Yasuda, S. et al. Stress- and ubiquitylation-dependent phase separation of the proteasome. Nature 578, 296–300 (2020).
Article CAS PubMed Google Scholar
Valenzuela-Fernández, A., Cabrero, J. R., Serrador, J. M. & Sánchez-Madrid, F. HDAC6: a key regulator of cytoskeleton, cell migration and cell–cell interactions. Trends Cell Biol. 18, 291–297 (2008).
Article PubMed Google Scholar
Banik, D. et al. HDAC6 plays a noncanonical role in the regulation of antitumor immune responses, dissemination, and invasiveness of breast cancer. Cancer Res. 80, 3649–3662 (2020).
Article CAS PubMed PubMed Central Google Scholar
Zeleke, T. Z. et al. Network-based assessment of HDAC6 activity predicts preclinical and clinical responses to the HDAC6 inhibitor ricolinostat in breast cancer. Nat. Cancer 4, 257–275 (2023).
Article CAS PubMed Google Scholar
Dowling, C. M. et al. Multiple screening approaches reveal HDAC6 as a novel regulator of glycolytic metabolism in triple-negative breast cancer. Sci. Adv. 7, eabc4897 (2021).
Article CAS PubMed PubMed Central Google Scholar
Pulya, S. et al. HDAC6 as privileged target in drug discovery: a perspective. Pharmacol. Res. 163, 105274 (2021).
Article CAS PubMed Google Scholar
Kawaguchi, Y. et al. The deacetylase HDAC6 regulates aggresome formation and cell viability in response to misfolded protein stress. Cell 115, 727–738 (2003).
Article CAS PubMed Google Scholar
Mazzetti, S. et al. Phospho-HDAC6 gathers into protein aggregates in Parkinson’s disease and atypical parkinsonisms. Front. Neurosci. 14, 624 (2020).
Article PubMed PubMed Central Google Scholar
Tellier, M. et al. CDK12 globally stimulates RNA polymerase II transcription elongation and carboxyl-terminal domain phosphorylation. Nucleic Acids Res. 48, 7712–7727 (2020).
Article CAS PubMed PubMed Central Google Scholar
Nativio, R. et al. An integrated multi-omics approach identifies epigenetic alterations associated with Alzheimer’s disease. Nat. Genet. 52, 1024–1035 (2020).
Article CAS PubMed PubMed Central Google Scholar
Padeken, J., Methot, S. P. & Gasser, S. M. Establishment of H3K9-methylated heterochromatin and its functions in tissue differentiation and maintenance. Nat. Rev. Mol. Cell Biol. 23, 623–640 (2022).
Article CAS PubMed PubMed Central Google Scholar
Shen, M. et al. Tinagl1 suppresses triple-negative breast cancer progression and metastasis by simultaneously inhibiting integrin/FAK and EGFR signaling. Cancer Cell 35, 64–80 (2019).
Article CAS PubMed Google Scholar
Nogués, L. et al. G protein-coupled receptor kinase 2 (GRK2) promotes breast tumorigenesis through a HDAC6–Pin1 axis. EBioMedicine 13, 132–145 (2016).
Article PubMed PubMed Central Google Scholar
Bhat-Nakshatri, P. et al. Breast cancer cell detection and characterization from breast milk-derived cells. Cancer Res. 80, 4828–4839 (2020).
Article CAS PubMed PubMed Central Google Scholar
Bergman, J. A. et al. Selective histone deacetylase 6 inhibitors bearing substituted urea linkers inhibit melanoma cell growth. J. Med. Chem. 55, 9891–9899 (2012).
Article CAS PubMed PubMed Central Google Scholar
Lee, J.-H. et al. Development of a histone deacetylase 6 inhibitor and its biological effects. Proc. Natl Acad. Sci. USA 110, 15704–15709 (2013).
Article CAS PubMed PubMed Central Google Scholar
Huang, X. et al. 14-3-3 proteins are potential regulators of liquid–liquid phase separation. Cell Biochem. Biophys. 80, 277–293 (2022).
Article CAS PubMed PubMed Central Google Scholar
Kuersten, S., Ohno, M. & Mattaj, I. W. Nucleocytoplasmic transport: Ran, beta and beyond. Trends Cell Biol. 11, 497–503 (2001).
Article CAS PubMed Google Scholar
Li, Y. et al. The structural basis for cohesin–CTCF-anchored loops. Nature 578, 472–476 (2020).
Article CAS PubMed PubMed Central Google Scholar
Nwosu, Z. C., Ebert, M. P., Dooley, S. & Meyer, C. Caveolin-1 in the regulation of cell metabolism: a cancer perspective. Mol. Cancer 15, 71 (2016).
Article PubMed PubMed Central Google Scholar
Schlesinger, Y. et al. Single-cell transcriptomes of pancreatic preinvasive lesions and cancer reveal acinar metaplastic cells’ heterogeneity. Nat. Commun. 11, 4516 (2020).
Article CAS PubMed PubMed Central Google Scholar
Price, J. G. et al. CDKN1A regulates Langerhans cell survival and promotes Treg cell generation upon exposure to ionizing irradiation. Nat. Immunol. 16, 1060–1068 (2015).
Article CAS PubMed PubMed Central Google Scholar
Zhang, Y. et al. Single-cell analyses reveal key immune cell subsets associated with response to PD-L1 blockade in triple-negative breast cancer. Cancer Cell 39, 1578–1593 (2021).
Article CAS PubMed Google Scholar
Croessmann, S. et al. NDRG1 links p53 with proliferation-mediated centrosome homeostasis and genome stability. Proc. Natl Acad. Sci. USA 112, 11583–11588 (2015).
Article CAS PubMed PubMed Central Google Scholar
Gibson, B. A. et al. Organization of chromatin by intrinsic and regulated phase separation. Cell 179, 470–484 (2019).
Article CAS PubMed PubMed Central Google Scholar
Liu, Y., Peng, L., Seto, E., Huang, S. & Qiu, Y. Modulation of histone deacetylase 6 (HDAC6) nuclear import and tubulin deacetylase activity through acetylation. J. Biol. Chem. 287, 29168–29174 (2012).
Article CAS PubMed PubMed Central Google Scholar
Folkmann, A. W., Putnam, A., Lee, C. F. & Seydoux, G. Regulation of biomolecular condensates by interfacial protein clusters. Science 373, 1218–1224 (2021).
Article CAS PubMed PubMed Central Google Scholar
Tessarz, P. & Kouzarides, T. Histone core modifications regulating nucleosome structure and dynamics. Nat. Rev. Mol. Cell Biol. 15, 703–708 (2014).
Article CAS PubMed Google Scholar
Bernier, M. et al. Linker histone H1 and H3K56 acetylation are antagonistic regulators of nucleosome dynamics. Nat. Commun. 6, 10152 (2015).
Article CAS PubMed Google Scholar
Peixoto, P. et al. HDAC5 is required for maintenance of pericentric heterochromatin, and controls cell-cycle progression and survival of human cancer cells. Cell Death Differ. 19, 1239–1252 (2012).
Article CAS PubMed PubMed Central Google Scholar
Yin, X. et al. ATF3, an adaptive-response gene, enhances TGFβ signaling and cancer-initiating cell features in breast cancer cells. J. Cell Sci. 123, 3558–3565 (2010).
Article CAS PubMed PubMed Central Google Scholar
Zheng, Y.-C. et al. Curriculum vitae of HDAC6 in solid tumors. Int. J. Biol. Macromol. 230, 123219 (2023).
Article CAS PubMed Google Scholar
Li, M. et al. Comprehensive 3D epigenomic maps define limbal stem/progenitor cell function and identity. Nat. Commun. 13, 1293 (2022).
Article CAS PubMed PubMed Central Google Scholar
Simon, M. D. et al. The site-specific installation of methyl-lysine analogs into recombinant histones. Cell 128, 1003–1012 (2007).
Article CAS PubMed PubMed Central Google Scholar
Parida, P. K. et al. Limiting mitochondrial plasticity by targeting DRP1 induces metabolic reprogramming and reduces breast cancer brain metastases. Nat. Cancer 4, 893–907 (2023).
Article CAS PubMed PubMed Central Google Scholar
Sun, X. et al. Loss of the receptors ER, PR and HER2 promotes USP15-dependent stabilization of PARP1 in triple-negative breast cancer. Nat. Cancer 4, 716–733 (2023).
Article CAS PubMed Google Scholar
Xie, J. et al. Targeting androgen receptor phase separation to overcome antiandrogen resistance. Nat. Chem. Biol. 18, 1341–1350 (2022).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

This work was supported by the National Natural Science Foundation of China (82372857 to B.L.; 81972651 and 82172698 to W.Z.), Natural Science Foundation of Guangdong Province (2017A030312009 to W.Z.) and the High-Level Hospital Construction Project (DFJHBF202102 to W.Z.). We extend our gratitude to Guangzhou Raybio Medical Technology Co., Ltd. for generously providing the HALO image analysis platform.

Author information

These authors contributed equally: Bing Lu, Ru Qiu, Jiatian Wei, Li Wang.

Authors and Affiliations

Medical Research Institute, Guangdong Provincial People’s Hospital (Guangdong Academy of Medical Sciences), Southern Medical University, Guangzhou, China
Bing Lu & Wei Zhao
Center for Stem Cell Biology and Tissue Engineering, Key Laboratory for Stem Cells and Tissue Engineering, Ministry of Education, Sun Yat-sen University, Guangzhou, China
Ru Qiu, Jiatian Wei, Qinkai Zhang, Mingsen Li, Xiudan Zhan, Jian Chen, Yifan Zhu, Tao Jiang, Han Zhu & Wei Zhao
Department of Anesthesiology, Affiliated Hospital of Guilin Medical University, Guilin, China
Li Wang
Shunde Hospital (The First People’s Hospital of Shunde), Southern Medical University, Foshan, China
I-Yun Hsieh
Department of Breast Cancer, Cancer Center, Guangdong Provincial People’s Hospital (Guangdong Academy of Medical Sciences), Southern Medical University, Guangzhou, China
Ciqiu Yang
Department of Breast and Thyroid Surgery, Guangzhou Women and Children’s Medical Center, Guangzhou, China
Jing Zhang, Zicheng Sun & Jie Li
Department of Thyroid Surgery, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, China
Jie Li

Authors

Bing Lu
View author publications
You can also search for this author in PubMed Google Scholar
Ru Qiu
View author publications
You can also search for this author in PubMed Google Scholar
Jiatian Wei
View author publications
You can also search for this author in PubMed Google Scholar
Li Wang
View author publications
You can also search for this author in PubMed Google Scholar
Qinkai Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Mingsen Li
View author publications
You can also search for this author in PubMed Google Scholar
Xiudan Zhan
View author publications
You can also search for this author in PubMed Google Scholar
Jian Chen
View author publications
You can also search for this author in PubMed Google Scholar
I-Yun Hsieh
View author publications
You can also search for this author in PubMed Google Scholar
Ciqiu Yang
View author publications
You can also search for this author in PubMed Google Scholar
Jing Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Zicheng Sun
View author publications
You can also search for this author in PubMed Google Scholar
Yifan Zhu
View author publications
You can also search for this author in PubMed Google Scholar
Tao Jiang
View author publications
You can also search for this author in PubMed Google Scholar
Han Zhu
View author publications
You can also search for this author in PubMed Google Scholar
Jie Li
View author publications
You can also search for this author in PubMed Google Scholar
Wei Zhao
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

B.L. performed the phase separation experiments and wrote the paper. R.Q. discovered the phospho-HDAC6 LLPS, conducted ATAC-seq, ChIP-seq and RNA-seq analyses and helped in revising the paper. L.W. and J.W. performed the animal experiments and in vitro assays. Q.Z. performed the ChIP-seq and DNaseI TUNEL assays. M.L. analyzed the Hi-C seq data and provided technical support. X.Z. and J.C. performed the metabolic assays. I.-Y.H. and C.Y. performed the drug screening assays. J.Z. and Z.S. generated the PDX model. Y.Z., T.J. and H.Z. prepared the plasmids and shRNA lentiviral vectors. J.L. designed the experiments, interpreted the data, revised the paper and provided supervision. W.Z. conceptualized and designed the experiments, interpreted the data, wrote and revised the paper and provided supervision.

Corresponding authors

Correspondence to Jie Li or Wei Zhao.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Cancer thanks the anonymous reviewers for their contribution to the peer review of this work.

Additional information

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

Extended data

Extended Data Fig. 1 Phosphorylated HDAC6 (phospho-HDAC6) in the nuclei of triple-negative breast cancer (TNBC).

Related to Fig. 1. a. HALO^® image analysis platform was used to identify tumor cells and measure the staining intensity of HDAC6 in breast cancer samples (n = 13 for para-cancerous, n = 19 for non-TNBC, and n = 3 for TNBC). Solid lines represent the median, and dashed lines represent the upper and lower quartiles. P values were calculated by one-way ANOVA with multiple comparisons. Scale bar, 50 μm. b. MDA-MB231 cells were transduced with HA-tagged HDAC6 or empty vector control. Subsequently, nuclear cell lysates underwent a pulldown process utilizing the HA antibody, followed by immunoblotting with the phosphorylation antibody. c. Sanger sequencing results of HDAC6 mutant variants. d. Immunoblot analysis was conducted using the phospho-HDAC6 antibody on BT-549 cells transfected with either HDAC6 wild-type or mutant constructs. e. Representative immunofluorescence images of BT-549 cells transfected with HDAC6 wide-type (WT) or mutations. Scale bar, 5 μm. f. Quantification of phospho-HDAC6 staining intensity in breast cancer samples using the HALO^® image analysis platform. The staining intensity was separated into nucleus and cytoplasm in breast cancer specimens (n = 6 for para-cancerous, n = 53 for non-TNBC, and n = 9 for TNBC). Solid lines represent the median, and dashed lines represent the upper and lower quartiles. P values were calculated by one-way ANOVA with multiple comparisons. Scale bars, 100 μm. g. Statistical analysis table showing the correlation between phospho-HDAC6 staining intensity and clinical data from 62 breast cancer patients (non-TNBC, n = 53; TNBC, n = 9). Significance was determined using a two-sided chi-squared test. h. Kaplan-Meier overall survival curve of 62 breast cancer patients based on their nuclear phospho-HDAC6 levels. N = 53 for non-TNBC, n = 9 for TNBC. Statistical significance was determined using log-rank test. Experiments were repeated three times independently with similar results; representative images are shown (a-b, d-f).

Source data

Extended Data Fig. 2 Liquid-liquid phase separation (LLPS) of phospho-HDAC6 in TNBCs.

Related to Fig. 1. a. Representative immunofluorescence images of HDAC6 in breast cancer specimens were obtained using a Super-resolution microscope (50 cells in each sample are used for statistical analysis). Nuclei were counterstained with DAPI. Scale bars, 25 μm. b. High-resolution images of HDAC6 immunofluorescence staining in mammary epithelial cell line (MCF-10A), luminal breast cancer cell line (MCF-7), and TNBC cell lines (BT-549 and MDA-MB231). N = 15 cells per group. Scale bar, 5 μm. c. Single-particle tracks for mean speed of simulated phosphorylation HDAC6 (HDAC6-S22E) single molecules within the temporally registered reference frame binned into 1 s intervals. HDAC6-S22E was fused with monomeric enhanced green fluorescent protein (mEGFP). d. BT-549 cells were subjected to staining using anti-phospho-HDAC6, anti-HP1α (heterochromatin marker), and anti-H3K9me3 antibodies. Subsequently, the correlation of fluorescence intensity between phospho-HDAC6 and HP1α or H3K9me3 was quantified. Scale bar, 5 μm. e. MCF-7 and MCF-10A cells were treated with EGF (200 ng/ml) for 30 mins, and the statistical results of the number of condensates in the control and EGF treatment group were shown, n = 15 cells per group. f. Immunoblot analysis of HDAC6 and phospho-HDAC6 proteins in BT-549 cells treated with or without EGF (200 ng/ml) for 30 mins. Scale bar, 5 μm. g. Immunofluorescence imaging of phospho-HDAC6 condensates in MCF-10A and MCF-7 cells upon GRK2 inhibitor GSK180736A (2 μM) treatment for 24 h, with statistical results of condensates number in control and GRK2 inhibitor treatment group shown. N = 15 cells per group, Scale bar, 5 μm. h. Immunoblot analysis of HDAC6 and phospho-HDAC6 proteins in BT-549 cells treated with or without GRK2 inhibitor GSK180736A (2 μM) for 48 h. i. Bar chart showing the relative expression levels of GRK2 in shCtrl and shGRK2 BT-549 cells, as determined by real-time polymerase chain reaction (qPCR) analysis. N = 3 per group. In a-b, e, g the boxplots represent the median values and quartiles, and the whiskers represent the maximum and minimum values. Data in c and i are shown as the mean ± s.d.; P values in a-b, i were calculated by one-way ANOVA with multiple comparisons. P values in e, g were calculated by two-way ANOVA with multiple comparisons. Experiments were repeated three times independently with similar results; representative images are shown in a-b and d-h.

Source data

Extended Data Fig. 3 Role of phosphorylation residue, catalytic domain, and nuclear export sequence (NES) regions in modulating HDAC6 phase separation.

Related to Fig. 2. a. Immunoblotting analysis of HDAC6 in WT and HDAC6 knockout (KO) MDA-MB231 and BT-549 cells. b. Immunofluorescence images of phosphorylated form of HDAC6 (or HDAC6 truncates) when overexpressing HDAC6 deletion segments(ΔIDR1/2) in the cell nuclei of HDAC6 KO MDA-MB231 cells. N = 15 cells per group. Scale bar, 5 μm. c. Analysis of droplet formation in the indicated protein mixture (160 μM each) before and after 5% 1,6-hexanediol treatment. WT refers to the wide-type HDAC6 IDR1 fused with mEGFP. S22F, HDAC6 IDR1 Ser 22 to Phe; S22E, HDAC6 IDR1 Ser 22 to Glu. N = 3 per group. Scale bar, 5 μm. Each point in the statistical chart represents the average diameter of all droplets in each experiment. d. Cell viability assays were performed using CellTiter-Glo in BT-549 and MDA-MB321 cells transduced with HDAC6 wild-type (WT) or mutations (S22E or S22F). N = 3 per group, statistical analysis is conducted after a 72-hour period. e. Sphere formation assay was conducted in BT-549 and MDA-MB321 cells transduced with HDAC6 WT or mutations, and the results were compared with the empty vector control. N = 3 per group, the number of spheres formed was counted on day 7. f. Invasion assays were performed in BT-549 and MDA-MB321 cells transduced with HDAC6 WT or mutations, respectively, n = 3 per group. g. Representative images of orthotopic tumor tissue (day 14) derived from MDA-MB231 cells transfected with HDAC6 WT, S22E, or S22F mutations (6 mice in each group). The tumor volume and weight data (day 14) are also presented. h. Immunofluorescence images of phosphorylated form of HDAC6 (or HDAC6 truncates) when overexpressing HDAC6 deletion segments (ΔBUZ, ΔDD1, ΔDD2, ΔDD) in the cell nuclei of HDAC6 KO MDA-MB231 cells. N = 15 cells per group. Scale bar, 5 μm. i. Immunofluorescence images of phosphorylated form of HDAC6 (or HDAC6 truncates) when overexpressing HDAC6 deletion segments (ΔNES1, ΔSE14, ΔNES2, ΔNES) in the cell nuclei of HDAC6 KO MDA-MB231 cells. N = 15 cells per group. Scale bar, 5 μm. j. Sphere formation assay of BT-549 cells overexpressing the empty vector (EV), full-length (FL), or NES-truncated HDAC6 variants. k. Extreme limiting dilution assay of BT-549 cells overexpressing the empty vector (EV), full-length (FL), or NES-truncated HDAC6 variants (ΔNES). (Left) An extreme limiting dilution algorithm was used to calculate the frequency of cancer stem cells, showing significant differences between EV and ΔNES (P = 0.0492). (Right) The table displays the estimated stem cell frequency and 95% confidence interval (CI). In b, h-i the boxplots represent the median values and quartiles, and the whiskers represent the maximum and minimum values. Data in c-g, j are shown as the mean ± s.d.; P values in b, e-j were calculated by one-way ANOVA with multiple comparisons. P values in c-d were calculated by two-way ANOVA with multiple comparisons. P values in k were calculated by chi-square test. Experiments were repeated three times independently with similar results; representative images are shown in a-c and g-j.

Source data

Extended Data Fig. 4 Inhibition of HDAC6 LLPS condensates by Nexturastat A suppresses TNBC growth in vitro and in vivo.

Related to Fig. 3. a. MDA-MB231 cells were transduced with HA tagged HDAC6 WT or S22F. Co-immunoprecipitation (Co-IP) analysis showing the interaction between HDAC6 and PPP1CA in MDA-MB231 cells. b. Immunoblotting analysis of HDAC6 and phosphorylated HDAC6 levels in MDA-MB231 cells transduced with shPPP1CA and control shRNA. c. Immunoblotting analysis of HDAC6 and phosphorylated HDAC6 in MDA-MB231 cells transduced with shPPP1CA or control shRNA, followed by treatment with 5 μM Nexturastat A or 5 mM HPOB (a reference compound) for 24 h. d. Immunoblotting analysis of HDAC6 and phosphorylated HDAC6 in MDA-MB231 cells treated with or without 5 mM Nexturastat A or 1.5% 1,6-hexanediol. e. Cell viability assays were performed in chondrocytes (as a normal control), MCF-7 cells treated with Nexturastat A for 48 h, n = 3 per group. f. Transwell assay was conducted in BT-549 cells treated with DMSO, HPOB, or Nexturastat A for 24 h, respectively. g. Flow cytometry analysis of apoptotic cells in BT-549 and MDA-MB231 cells treated with DMSO, HPOB, or Nexturastat A. h. Flow cytometry analysis of apoptotic cells in MDA-MB231, BT-549, and MCF-7 cells treated with Nexturastat A at the indicated concentrations. i. Flow cytometry analysis of apoptotic cells in MDA-MB231 cells stably transduced with HDAC6 WT or S22F mutation, with or without treatment of Nexturastat A (5 μM). j. Cell cycle analysis of BT-549 cells treated with Nexturastat A at the indicated concentrations for 48 h, n = 3 per group. k. Representative images of IHC staining and quantification of Ki67-positive cells in MDA-MB231-derived subcutaneous tumors treated with Nexturastat A at the indicated concentration, n = 3 per group. l. Representative immunohistochemistry (IHC) staining for ER, PR, and HER2 in TNBC patient-derived xenografts. m. Weight curve of mice bearing MDA-MB231-derived subcutaneous tumors after treatment with Nexturastat A at the indicated concentration, n = 5 per group. n. Cell viability assays were conducted in MDA-MB231 and BT-549 cells stably transduced with non-targeting scrambled control shRNA (shCtrl) or two HDAC6 shRNAs (shHDAC6-1 and shHDAC6-2), respectively. Related cell viability values are normalized to the control (day 0), n = 3 per group. o. Sphere formation assays were conducted in MDA-MB231 and BT-549 cells stably transduced with shCtrl, shHDAC6-1, or shHDAC6-2, respectively, with or without Nexturastat A (5 μM). N = 3 per group, the number of spheroids was counted on day 7. p. Invasion assays were conducted in MDA-MB231 cells stably transduced with shCtrl, shHDAC6-1, or shHDAC6-2, respectively, with treatment of DMSO or Nexturastat A (5 μM), n = 3 per group. q. Flow cytometry analysis of apoptotic cells in MDA-MB231 and BT-549 cells stably transduced with shCtrl, shHDAC6-1, or shHDAC6-2, respectively, with or without Nexturastat A (5 μM), n = 3 per group. Data in e-k and m-q are shown as the mean ± s.d.; P values in f-h, k were calculated by one-way ANOVA with multiple comparisons. P values in i, n-q were calculated by two-way ANOVA with multiple comparisons. Experiments were repeated three times independently with similar results; representative images are shown in a-d and k-l.

Source data

Extended Data Fig. 5 14-3-3θ promotes phospho-HDAC6 LLPS in TNBCs.

Related to Fig. 4. a. Schematic diagram depicting the immunoprecipitation-mass spectrometry (IP-MS) experiment performed to identify components in phospho-HDAC6 LLPS condensates. b. The HDAC6 up-regulated interacting proteins upon 8 μM Nexturastat A treatment. c. Co-IP analysis in MDA-MB231 cells transduced with HA-tagged HDAC6 and treated with or without 10 μM Nexturastat A, showing the interaction between HDAC6 and a-Tubulin. EV, empty vector. d. Mass spectrometry images showing the 14-3-3θ interacted with HDAC6. e. qPCR showing the knockdown efficiency of 14-3-3θ in BT-549 cells. f-g. In vitro droplet formation assays demonstrating that HDAC6-WT droplets but not HDAC6-S22F droplets incorporate 14-3-3θ protein. Scale bar = 50 μm. h. Statistical analysis of the diameter distribution of mixed protein droplets formed by HDAC6-WT, HDAC6-S22F, and HDAC6-S22E with 14-3-3θ protein in vitro. i. Quantitative PCR (qPCR) demonstrating the knockdown efficiency of Importin β in BT-549 cells j. Representative super-resolution microscope images showing total HDAC6 localization in BT-549 cells with or without Importin β KD (n = 15). Nuclei were counterstained with DAPI. Scale bars, 5 μm. k. qPCR demonstrating the knockdown efficiency of NUP153, NUP214, NUP88, and NUP98 in BT-549 cells. In j the boxplots represent the median values and quartiles, and the whiskers represent the maximum and minimum values. Data in e and h-k are shown as the mean ± s.d.; P values in e, i-k were calculated by one-way ANOVA with multiple comparisons. P values in h were calculated by two-way ANOVA with multiple comparisons. Experiments were repeated three times independently with similar results; representative images are shown in c, f-g and j.

Source data

Extended Data Fig. 6 Disruption of phospho-HDAC6 LLPS re-establish chromatin structure.

Related to Fig. 5. a. DNase I-treated TUNEL assay was performed to analyze chromatin accessibility in MDA-MB231 cells treated with either DMSO control (Ctrl) or Nexturastat A. Nuclei were counterstained with DAPI. b. Heatmaps showing chromatin interactions of chromosome 2 at different resolutions: 100-kb, 50-kb, and 10-kb. c. Chromatin compartments and Pearson correlation heatmap of cis-interactions on chromosome 1. Positive first principal component (PC1) values represent compartment A (blue), and negative PC1 values represent compartment B (red). d. Boxplots showing the GC content of compartments A and B in the Ctrl and Nexturastat A-treated cells. e. Comparison of chromatin compartments and Pearson correlation heatmap of Ctrl and Nexturastat A-treated cells on chromosome 6, chromosome 10, and chromosome 17. The shaded area indicates the switching region between compartments A and B. f. BODIPY 493/503 staining of BT-549 and MDA-MB231 cells treated with or without 10 mM Nexturastat A for 48 h. BODIPY 493/503 is used to visualize lipid droplets, indicating changes in lipid metabolism upon treatment. g. Representative images of immunohistochemistry (IHC) staining of CD4, CD8, PD-1, and CD86 in 4T1 xenografts. h. Venn diagram shows the overlap of TAD boundaries between control and Nexturastat A treatment groups. i. The number of differential TAD in four classification groups after Nexturastat A treatment. j. Representative images of merge, split and rearrangement in TAD. k. Top ten TF-binding motifs found in control and Nexturastat A treatment group TAD boundaries, separately. l. GO enrichment of genes in Nexturastat A specific loops. Gene Ontology (GO) enrichment analysis was performed using the hypergeometric test, a two-tailed non-parametric method. m. IGV Genome Browser visualization depicting histone pan-acetylation, HDAC, and RNA Polymerase II (Pol II) Chromatin Immunoprecipitation followed by DNA Sequencing (ChIP-seq) data, RNA-seq results, and interactions identified from Hi-C in MDA-MB231 cells treated with or without Nexturastat A for PCDHA1, UPB1, ATP1A2, and GALNT9 loci. In a, d the boxplots represent the median values and quartiles, and the whiskers represent the maximum and minimum values. Data in f is shown as the mean ± s.d.; P values in a, d, and f were calculated by two-tailed student’s t-test. Experiments were repeated three times independently with similar results; representative images are shown in a, f-g.

Source data

Extended Data Fig. 7 Phospho-HDAC6 LLPS condensates disruption induces transcriptional reprogramming in TNBCs.

Related to Fig. 6. a. Genomic binding patterns of Assay for Transposase-Accessible Chromatin using Sequencing (ATAC-seq) and Pol II ChIP-seq with or without Nexturastat A treatment, centered around peak regions and 1 kb upstream and downstream of lost, common and new peaks. b. Box plots displaying the consistency of ATAC-seq and Pol II ChIP-seq signal. X-axis is grade of log₂ (fold change) of ATAC signal and Y-axis is log₂ (fold change) of pol II occupancy. The boxplots represent the median values and quartiles, and the whiskers represent the maximum and minimum values. P values were calculated by one-way ANOVA with multiple comparisons. c. Heatmap showing binding pattern of ATAC-seq and Pol II ChIP-seq surrounding new peaks. d-e. Volcano plot displaying the changes in gene expression following Nexturastat A, HPOB, Tubastatin A treatment or knockdown of HDAC6 in MDA-MB231 cells. Gene expression differences were assessed using DESeq2, which models read counts with a negative binomial distribution and uses two-tailed Wald tests to identify differentially expressed genes between conditions. f. Heatmap displaying the expression levels of differentially expressed genes, including tumor suppressor genes, proto-oncogenes, and immunomodulators, in MDA-MB-231 cells with or without Nexturastat A treatment. g-h. Gene set enrichment analysis (GSEA) enrichment plot showing the pathway of Antigen Processing and Presentation (GO:0019882) and TNFα Signaling via NF-κB for Nexturastat A treatment versus control and shHDAC6 versus shCtrl. GSEA was performed using a two-tailed non-parametric permutation-based method. Genes were first ranked by their correlation with the phenotype. P values in d-e were calculated by two-tailed Wald tests and g-h was performed using a two-tailed non-parametric permutation-based method. N = 3 biologically independent experiments.

Source data

Extended Data Fig. 8 Identification of direct targets of phospho-HDAC6 LLPS condensates in TNBCs.

Related to Fig. 7. a. Genomic binding patterns of HDAC6 ChIP-seq and H3 Pan-acetylation (Pan-acetyl) ChIP-seq with and without Nexturastat A treatment, centered around the transcription start site (TSS) and 3 kb upstream and downstream of differentially expressed genes in HDAC6 ChIP-seq. HDAC6 ChIP-seq replicate 1 (Rep1) utilized HDAC6 antibody from Cell Signaling Technology, while replicate 2 (Rep2) employed HDAC6 antibody from Novusbio. b. Metagene plot showing HDAC6 ChIP-seq and H3 Pan-acetyl ChIP-seq occupancy profiles across the transcription start site (TSS) and transcription end site (TES) regions with and without Nexturastat A treatment. c. Pie chart illustrating the percentage of genomic occupancy for decreased binding regions in HDAC6 ChIP-seq and increased binding regions in H3 Pan-acetyl ChIP-seq. d. Genomic tracks of HDAC6 ChIP-seq and H3 Pan-acetyl ChIP-seq with and without Nexturastat A treatment at the tumor suppressor gene NDRG1 and immunomodulator PD-L1 loci. HDAC6 ChIP-seq Rep1 utilized HDAC6 antibody from Cell Signaling Technology, while Rep2 employed HDAC6 antibody from Novusbio. e. Boxplot showing the expression levels of ATF3 in TNBC and non-TNBC patients from TCGA. Statistical significance was determined using Student’s two-tailed t-test. f. Boxplot showing the expression levels of ATF3 in different stages (I-IV) of TNBC patients. Statistical significance was determined using one-way ANOVA followed by post-hoc tests with Bonferroni correction for multiple comparisons. g. Bar chart showing the relative expression levels of ATF3 in shCtrl and shATF3 MDA-MB231 cells as determined by qPCR analysis. P values were calculated by one-way ANOVA with multiple comparisons. h. Representative images showing the number of invasive cells in the control group (shCtrl) and ATF3 knockdown group (shATF3) with or without Nexturastat A treatment. i. Representative images showing the number of spheres in the control group (shCtrl) and ATF3 knockdown group (shATF3) with or without Nexturastat A treatment. j. Flow cytometry plots showing the percentage of apoptotic cells in the control group (shCtrl) and ATF3 knockdown group (shATF3) treated with or without Nexturastat A treatment. k. Schematic representation of the orthotopic tumor model. BT-549 cells with ATF3 knockdown or control were injected into nude mice, respectively. Nexturastat A or 0.9% NaCl was administered every 2 days for 11 days. In e-f the boxplots represent the median values and quartiles, and the whiskers represent the maximum and minimum values. Data in g are shown as the mean ± s.d.; P values in e, g were calculated by two-tailed student’s t-test, f was calculated by one-way ANOVA. Experiments were repeated three times independently with similar results; representative images are shown in h-j.

Source data

Supplementary information

Supplementary Information

Supplementary Fig. 1: gating strategies.

Reporting Summary

Supplementary Tables 1–6

Supplementary Table 1: HDAC inhibitors information in Fig. 3. Supplementary Table 2: shRNAs used in this study. Supplementary Table 3: qPCR primers used in this study. Supplementary Table 4: Antibodies used in this study. Supplementary Table 5: The list of abbreviations for lipid metabolites in Fig. 5g. Supplementary Table 6: Gene_exp_loop in Fig. 5q.

Source data

Source Data Fig. 1

Statistical source data.

Source Data Fig. 2

Statistical source data.

Source Data Fig. 3

Statistical source data.

Source Data Fig. 4

Statistical source data.

Source Data Fig. 5

Statistical source data.

Source Data Fig. 6

Statistical source data.

Source Data Fig. 7

Statistical source data.

Imaging source data

Unprocessed western blots and/or gels.

Source Data Extended Data Fig. 1

Statistical source data.

Source Data Extended Data Fig. 2

Statistical source data.

Source Data Extended Data Fig. 3

Statistical source data.

Source Data Extended Data Fig. 4

Statistical source data.

Source Data Extended Data Fig. 5

Statistical source data.

Source Data Extended Data Fig. 6

Statistical source data.

Source Data Extended Data Fig. 7

Statistical source data.

Source Data Extended Data Fig. 8

Statistical source data.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Cite this article

Lu, B., Qiu, R., Wei, J. et al. Phase separation of phospho-HDAC6 drives aberrant chromatin architecture in triple-negative breast cancer. Nat Cancer (2024). https://doi.org/10.1038/s43018-024-00816-y

Download citation

Received: 02 April 2023
Accepted: 30 July 2024
Published: 28 August 2024
DOI: https://doi.org/10.1038/s43018-024-00816-y

Subjects

Abstract

Similar content being viewed by others

Main

Results

Discovery of phospho-HDAC6 undergoing LLPS in TNBC

Phosphorylation residue and nuclear export signal (NES) regions in HDAC6 phase separation

Disruption of HDAC6 LLPS inhibits TNBC growth through Nexturastat A

Mechanisms of phospho-HDAC6 nuclear translocation and LLPS

Disruption of phospho-HDAC6 LLPS alters chromatin structure

Transcriptional reprogramming induced by Nexturastat A treatment

ATF3 as one of the downstream targets of phospho-HDAC6 LLPS

Discussion

Methods

Cell culture and clinical samples

Constructs of shRNA and cell transfection

PDX models of TNBC

Animal studies

LLPS assay of proteins

FRAP

Generation of phase diagram

In situ Hi-C library construction and Hi-C read processing

Preparation of the nucleosome chromatin with H3K27ac modification

DNaseI terminal deoxynucleotidyl transferase deoxyuridine triphosphate (dUTP) nick end labeling (TUNEL) assay for chromatin architecture analysis

RNA-seq and data analysis

ChIP-seq and ATAC-seq

Tumor suppressor genes and proto-oncogenes sets

Image analysis with HALO image system

Single-molecule tracking

Cell proliferation assay

Cell invasion assay

3D spheroid formation assay

Immunoblot analysis

qPCR with reverse transcription (RT–qPCR)

Flow cytometry analysis of cell apoptosis and cell cycle

Immunofluorescence staining assay

Recombinant protein purification

Plasmid construction

BODIPY 493/503 staining for neutral lipids

Targeted lipidomics analysis for specific lipids

Statistics and reproducibility

Reporting summary

Data availability

Code availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Extended data

Supplementary information

Source data

Rights and permissions

About this article

Cite this article

Share this article

Search

Quick links