The PURB–HOTAIR complex regulates p53-dependent promoter-specific transcriptional activation

Abstract

p53 executes its diverse functions through different transcriptional targets but the precise mechanism of promoter-specific regulation by p53 remains largely unknown. Through biochemical purification, we identify purine-rich element binding protein B (PURB), a dual DNA/RNA-binding protein, which acts as a transcriptional corepressor for p53 in a manner dependent on p53 acetylation status. PURB is overexpressed in human cancers, and its knockdown induces p53-dependent activation of p21 but has no effect on other major promoters such as PUMA and MDM2. In contrast to other p53 corepressors, PURB can recognize a unique DNA element at the p21 promoter, with the loss of this element not affecting p53-mediated transactivation but abrogating the ability of p53 to recruit PURB to the p21 promoter for repression. Mechanistically, PURB requires its sequence-specific binding with long noncoding RNA (lncRNA) HOX transcript antisense RNA (HOTAIR) to exert its repressive role. In turn, HOTAIR interacts directly with EZH2 and, bridged by the PURB–HOTAIR complex, p53 can recruit the EZH2 histone methyltransferase to target promoters for transcriptional repression. Further analysis of p53 targets reveals several promoters that may serve as targets for PURB binding, suggesting that this mechanism of PURB-dependent promoter-specific regulation may not be limited to p21. These data establish a mode of lncRNA-mediated regulation of p53 transcription in a sequence-specific manner and reveal a previously unanticipated mechanism for acetylation-mediated promoter-specific regulation through a cis-regulatory element recognized by the PURB–HOTAIR complex.

As a transcriptional factor, the p53 tumor suppressor governs diverse biological processes by regulating numerous cellular targets^1–5. The antitumor activity of p53 is mediated by different target genes participating in specific functions including cell-cycle arrest, senescence, apoptosis, metabolism, ferroptosis, DNA repair, immunity and others^6–9. p53 binds target promoters through a specific response element composed of two decameric repeats: RRRCWWGYYY (in which R denotes a purine; W denotes A or T; Y denotes a pyrimidine)^10–12. Hundreds of direct p53 transcriptional targets have been identified, each containing at least one p53 response element in their promoters¹¹. However, under specific conditions, only a subset of these target genes are activated or repressed by p53 (refs. 5,13). The promoter-specific transcriptional regulation leads to context-dependent p53 response that converges on several specific downstream pathways to mediate a tailored response to different levels and types of stimuli.

p53 undergoes various post-translational modifications including ubiquitination, phosphorylation, acetylation, methylation, SUMOylation and NEDDylation¹⁴. These modifications tightly regulate p53 activity. Both p53 post-translational modifications and promoter architecture of the target genes influence p53 interaction with its target genes. For instance, phosphorylated-p53 (at S46) stimulates apoptosis, whereas protein arginine methyltransferase 5-methylated p53 preferentially activates p21 (ref. 15). It was also proposed that the sequence specificity of p53 response element at the promoter of each target gene may be crucial for promoter-specific regulation¹⁶ and that interaction with cofactors may also have a role². We and others showed that the transcriptional activity of p53 is tightly regulated by acetylation¹⁷. Key cofactors in the acetylation pathway including acetylases (for example, CBP–p300 and TIP60–hMOF) and deacetylases (for example, histone deacetylase family and Sirtuin family) are directly involved in functional regulation of p53. Moreover, site-specific acetylation has a critical role in promoter selectivity. For example, we and others found that acetylation of p53 at K120 by TIP60 or hMOF selectively activates proapoptotic targets^18,19. PBRM1 reads p53 K382 acetylation and facilitates p21 activation²⁰. Brd4 association with p53, modulated by casein kinase 2-mediated phosphorylation, selectively activates p53 targets²¹. Nonetheless, the precise mechanisms of promoter-specific regulation remain to be fully elucidated.

Here, we identify the dual DNA/RNA-binding protein PURB^22–25 as a binding partner and transcriptional repressor of p53. PURB interacts with p53 both in vitro and in vivo and this interaction is abolished by C-terminal acetylation of p53 during transcriptional activation. Interestingly, PURB knockdown activates only a subset of p53 targets without altering p53 protein levels. PURB is specifically recruited by p53 to the p21 (also known as CDKN1A) promoter but not to other major p53 targets such as PUMA or MDM2. A unique cis-regulatory element at the p21 promoter is recognized by PURB; loss of this element does not affect p53-mediated transactivation but abrogates the ability of p53 to recruit PURB to the promoters for repression. Furthermore, PURB-mediated transcriptional repression requires its sequence-specific binding to HOX transcript antisense RNA (HOTAIR), one of the long noncoding RNAs (lncRNAs) tightly associated with PURB. HOTAIR interacts with EZH2 and, bridged by the PURB–HOTAIR complex, p53 is able to recruit EZH2 histone methyltransferase to the target promoters for transcriptional repression. Importantly, both PURB and HOTAIR are overexpressed in human tumors and knockdown of either induces tumor growth suppression in a p53-dependent manner. These data reveal a previously unrecognized mechanism for promoter-specific regulation through a unique cis-regulatory element recognized by the PURB–HOTAIR corepressor complex.

Results

Identification of PURB as a p53-binding partner

To discover potential p53 cofactors involved in the promoter-specific regulation, we performed a proteomic screen for p53-interacting proteins. Using double-affinity purification (streptavidin–agarose and anti-S protein–agarose) from a stable H1299 cell line expressing SFB (S protein–Flag–streptavidin-binding peptide)-tagged p53, we purified all potential p53-binding proteins (Fig. 1a). Liquid chromatography–tandem mass spectrometry (LC–MS/MS) analysis of the affinity-purified SFB–p53 associated proteins identified several known p53-binding proteins, such as p53BP1, USP7 and SET (refs. 26–28), along with seven unique peptides matching PURB (Fig. 1a and Extended Data Fig. 1a). PURB (also known as Purβ; encoded by PURB) is a multifunctional protein. PURB was initially characterized as a sequence-specific single-stranded DNA-binding protein preferentially binding the purine-rich element at numerous promoter-flanking regions and the origins of replication of human genomes^22–24. Interestingly, recent studies revealed that, through its intrinsic RNA-binding activity, PURB is able to directly interact with lncRNAs^{24,25,29–32}. Although PURB is a dual DNA/RNA-binding protein, its precise cellular function remains unclear.

Fig. 1 |. Identification of PURB as a binding partner of p53 C terminus.

a, Identification of PURB peptides by LC–MS/MS in the p53 complex purified by double immunoprecipitation (IP) from H1299 cells expressing SFB-tagged p53 protein. b, Western blot analysis of the interaction between overexpressed p53 (full-length (FL) or truncations) and endogenous PURB in H1299 cells. c, In vitro binding assay of GST-tagged p53 (full-length or truncations) and purified HA-tagged PURB protein. d, Western blot analysis of the interaction between endogenous p53 and PURB in indicated cell lines. e, Western blot analysis of the interaction between overexpressed p53 and PURB (full-length, truncations or deletion) in H1299 cells. f, Schematic diagram of the interaction between p53 and PURB revealed in e. The experiments b–e were repeated three times with similar results and representative results are shown.

Next, we examined the interactions between PURB and p53 by expression of SFB-tagged full-length p53, p53 (1–300) or p53 (301–393) in p53-null H1299 cells. Co-immunoprecipitation revealed that endogenous PURB associates with full-length p53 and p53 (301–393), but not with p53 (1–300) (Fig. 1b), indicating that PURB interacts with the C-terminal domain (CTD; 301–393) of p53. In vitro glutathione S-transferases (GST) pulldown assays further confirmed direct binding between purified hemagglutinin (HA)-tagged PURB and GST–p53 (301–393) but not GST alone or GST fusions containing the N terminus (1–99) or DNA-binding domain (100–300) of p53 (Fig. 1c). Further mapping showed that the extreme CTD (364–393) of p53 was sufficient to bind PURB (Fig. 1c). Furthermore, the in vivo interaction of PURB and p53 was validated by co-immunoprecipitation assays using endogenous PURB and p53 from human colorectal cancer HCT116 cells, human osteosarcoma U2OS cells, human melanoma A375 cells and human breast cancer MCF-7 cells (Fig. 1d). Domain mapping further revealed that amino acids 175–210 of PURB are essential for p53 binding (Fig. 1e,f). These results demonstrate a specific interaction between PURB and p53 both in vitro and in vivo.

The PURB–p53 interaction is tightly regulated by acetylation

Because the extreme CTD of p53 (364–393) is responsible for binding with PURB (Fig. 1c), we tested whether acetylation of this region affects the PURB–p53 interaction. As shown in Fig. 2a, both wild-type full-length p53 and the CTD acetylation-defective KR mutant (Lys-to-Arg at 370, 372, 373, 381, 382 and 386; Fig. 2b) were able to interact with endogenous PURB protein but the acetylation-mimicking CTD KQ mutant (Lys-to-Gln at 370, 372, 373, 381, 382 and 386; Fig. 2b) completely lost the ability to interact with PURB. To directly examine whether PURB binds p53 in a manner that is dependent on the CTD acetylation status, we synthesized biotin-conjugated CTD peptides (amino acids 364–393) in either unacetylated or fully acetylated forms (Extended Data Fig. 1b) and assessed PURB binding using immobilized peptidesb. As shown in Fig. 2c, PURB specifically interacted with the unacetylated but not the acetylated CTD peptide. To further examine the specificity of this regulation, we tested whether methylation of the same lysine residues affects the interaction. Both methylated and unmethylated CTD peptides bound PURB comparably (Fig. 2c), indicating that methylation has no notable effect on PURB binding. Furthermore, coexpression of CREB-binding protein (CBP), the acetyltransferase responsible for CTD acetylation, abolished the formation of the PURB–p53 complex with wild-type p53 but not with the p53KR mutant, confirming that CTD acetylation is critical for modulating the PURB–p53 interaction in cells (Fig. 2d).

Fig. 2 |. PURB represses p53 activity in a manner dependent on p53 acetylation status.

a, Western blot analysis of interaction between overexpressed wild-type (WT) p53, or mutants p53KR or p53KQ and endogenous PURB. b, Schematic diagram of the interaction between p53 and PURB revealed in a. c, In vitro binding assay of biotin-conjugated p53 DNA-binding domain peptide or C-terminal unmodified, acetylated (Ac) or methylated p53 peptides and purified HA-tagged PURB protein. d, Western blot analysis of the interaction between p53 and PURB in H1299 cells. e, Western blot analysis of expression levels of p53, p21, MDM2 and PUMA in H1299 cells transfected with indicated plasmids. f, Western blot analysis of the endogenous interaction between p53 and PURB in HCT116 cells upon doxorubicin (0.2 μg ml⁻¹) plus TSA or NAM treatment for 24 h. g, ChIP analysis of p53 occupancy at p21 promoter in HCT116 cells upon doxorubicin (0.2 μg ml⁻¹) plus TSA or NAM treatment for 24 h (n = 3). h, ChIP analysis of PURB occupancy at p21 promoter in HCT116 cells upon doxorubicin (0.2 μg ml⁻¹) plus TSA or NAM treatment for 24 h (n = 3). Data in g,h are presented as the mean ± s.d. of n = 3 independent biological repeats. P values were calculated using an unpaired, two-tailed Student’s t-test. The experiments in a,c–f were repeated three times with similar results and representative results are shown.

Our previous study showed that SET protein binds p53 in an acetylation-dependent manner, mediated by charge neutralization of the lysine-rich CTD²⁸. The p53 CTD is a lysine-rich domain with a positive charge while the acidic domain of SET is rich with acidic amino acids harboring a highly negative charge. Acetylation of p53 CTD neutralizes the positive charge of the lysine residues within the p53 CTD, thus disrupting SET binding. Similarly, PURB–p53 binding appears to follow the same charge-based mechanism. To validate this, we generated a charge-altering mutant p53KL (Lys-to-Leu at 370, 372, 373, 381, 382 and 386; Extended Data Fig. 1c,d). Like the KQ mutant, the KL mutant also failed to bind PURB. Moreover, analysis of the PURB region critical for p53 binding (amino acids 175–210) revealed the presence of six acidic residues (aspartic acid and glutamic acid), suggesting potential electrostatic interactions with the positively charged p53 CTD. Substitution of these residues to alanine (Asp/Glu-to-Ala at 191, 192, 196, 197 and 198) significantly reduced the ability of PURB to bind p53 (Extended Data Fig. 1e,f). These data indicate that the acetylation-mediated effect on charge neutralization is critical for modulating the p53–PURB interaction.

We next assessed whether PURB modulates p53 transcriptional activity and whether this regulation depends on p53 CTD acetylation. p53-null H1299 cells were cotransfected with expression vectors encoding p53 alone or together with PURB. As expected, p53 alone activated transcription of the target genes including p21, PUMA and MDM2 (lanes 1–4, Fig. 2e). However, coexpression of PURB strongly repressed p21 transactivation, with no obvious effect on PUMA or MDM2 expression. Notably, PURB-mediated repression of p21 was abolished when the p53 mutant lacking the C-terminal 30 amino acids (ΔCTD) or the acetylation-mimicking KQ mutant was expressed (lanes 5–10, Fig. 2e), suggesting that PURB-mediated transcription repression is both promoter specific and regulated by p53 CTD acetylation.

Given that p53 acetylation is dramatically induced upon stress, we examined the impact of DNA damage on the PURB–p53 interaction. As expected, the levels of p53 acetylation were markedly induced by doxorubicin treatment (Fig. 2f). Trichostatin A (TSA) and nicotinamide (NAM) are well-known histone deacetylase inhibitors to preserve the acetylation status of p53 (refs. 33,34). Notably, despite increased p53 protein levels upon DNA damage, p53–PURB binding was largely abolished during the DNA damage response as p53 acetylation levels were upregulated (Fig. 2f). By contrast, when p53 acetylation level was low, the interaction between PURB and p53 was readily detected (Fig. 2f). Moreover, we conducted chromatin immunoprecipitation (ChIP) assays to measure formation of the p53–PURB complex formation on the p21 promoter in HCT116 cells under both unstressed and DNA damage conditions. Under unstressed conditions, both p53 and PURB were recruited to the p21 promoter (Fig. 2g,h). However, upon DNA damage, PURB occupancy on the p21 promoter was dramatically reduced while p53 occupancy increased. Together, these data indicate that PURB functions as a critical repressor of p53 transactivation at specific promoters and this repression is abolished upon CTD acetylation.

PURB is overexpressed in human cancers and knockdown of PURB induces p53-dependent tumor growth suppression

To explore the physiological consequences of the PURB–p53 interaction, we performed small interfering RNA (siRNA)-mediated knockdown of PURB in multiple human cancer cell lines. In melanoma A375 cells (Fig. 3a), osteosarcoma U2OS cells (Fig. 3b) and breast cancer MCF-7 cells (Fig. 3c), depletion of PURB significantly upregulated p21 expression, with no notable changes in PUMA or MDM2 levels. Importantly, this effect was p53 dependent, as PURB depletion failed to induce p21 in isogeneic p53-null cells (Fig. 3a). Because PURB loss did not affect p53 protein levels or induce other p53 targets, PURB likely functions as a promoter-specific repressor. Interestingly, PURB was found to be overexpressed in several human cancers, including breast invasive carcinoma (BRCA), esophageal carcinoma (ESCA), glioblastoma multiforme (GBM), pancreatic adenocarcinoma (PAAD) and thymoma (THYM) (Fig. 3d), which is consistent with a potential oncogenic role of PURB.

Fig. 3 |. Loss of PURB activates p53-dependent p21 induction and tumor suppression.

a–c, Western blot analysis of protein levels in A375 (a), U2OS (b) and MCF-7 (c) cells transfected with control siRNA (si-Ctrl) or siRNA against PURB (si-PURB). d, High expression of PURB in BRCA, cholangiocarcinoma (CHOL), diffuse large B cell lymphoma (DLBC), ESCA, GBM, PAAD, stomach adenocarcinoma (STAD) and THYM tumor data matched with normal data pipelined by GEPIA2 web server. The q values were calculated using ANOVA. *q < 0.05 and **q < 0.01. e, Western blot analysis of protein levels in indicated HCT116 cells. f, Colony numbers shown as the mean ± s.e.m. of n = 3 independent plates of indicated HCT116 cells. g, Ratio of BrdU-positive HCT116 cells shown as the mean ± s.e.m. of n = 5 independent fields. h, Xenograft model of PURB-mediated effect on tumor growth using HCT116 cells. i, Tumor weights shown as the mean ± s.e.m. of n = 6 independent tumor samples. j, Xenograft model of PURB-mediated effect on tumor growth using A375 cells. k, Tumor weights shown as the mean ± s.e.m. of n = 7 independent tumor samples. Data are presented as the mean ± s.e.m. of independent biological repeats. P values were calculated using an unpaired, two-tailed Student’s t-test. The experiments in a–c,e were repeated three times with similar results and representative results are shown.

We next generated PURB-null cells using CRISPR–Cas9 and examined cancer cell growth. In HCT116 cells, loss of PURB significantly upregulated p21 expression without affecting PUMA or MDM2 and had no effect in isogenic p53^−/− (also known as TP53^−/−) cells (Fig. 3e). Colony formation assays and bromodeoxyuridine (BrdU) incorporation assays demonstrated that PURB knockout suppressed cancer cell growth in a p53-dependent manner (Fig. 3f,g and Extended Data Fig. 2a,b). To examine the role of PURB in regulating cell growth in vivo, we assessed whether loss of PURB affects p53-mediated tumor growth suppression in xenograft tumor models. In xenograft models, PURB deletion significantly impaired tumor growth of HCT116 and A375 cells but had no effect on tumors derived from p53-null cells (Fig. 3h–k and Extended Data Fig. 2c). Importantly, re-expression of PURB in PURB-knockout A375 cells restored p21 repression without affecting p53 levels (Extended Data Fig. 2d), confirming that the p21 upregulation observed in PURB-knockout cells was because of PURB loss. Together, these findings indicate that PURB is overexpressed in human cancers and its depletion induces p53-dependent tumor growth suppression through activation of p21 expression.

HOTAIR is critical for PURB-mediated repression on p53

Previous studies have shown that PURB is a bona fide RNA-binding protein^{24,25,29–32}. To investigate the mechanism by which PURB regulates p53 function, we tested whether its effects are mediated through lncRNAs. Two key assays are commonly used to examine the protein–lncRNA interactions: nondenaturing RNA immunoprecipitation (RIP) assays, which preserve indirect interactions through protein–protein contacts³⁵, and ultraviolet (UV) crosslinking and immunoprecipitation (CLIP) assays, which identify direct interactions under denaturing conditions^36,37.

To identify lncRNAs that directly bind PURB, we first generated a stable H1299 cell line expressing SFB-tagged PURB (Extended Data Fig. 3a). After UV crosslinking, cell extracts were subjected to double-affinity purification under high-stringency conditions, followed by next-generation sequencing (CLIP-seq) (Extended Data Fig. 3b). Indeed, sequencing analysis revealed several lncRNAs, including HOTAIR, linc-ROR, NEAT1, SNHG1, GAS5 and MALAT1, as direct binding partners of PURB (Extended Data Fig. 3c and Supplementary Table 1). NEAT1 and MALAT1, known to bind PURB directly³⁰, served as positive controls for RNA enrichment. However, HOTAIR, despite its lower abundance, was retrieved at a substantial level and ranked among the top hits (Supplementary Table 1), suggesting a specific interaction with PURB.

To validate the interaction between PURB and these lncRNAs, we performed CLIP assays in p53-null H1299 cells expressing either SFB–PURB or SFB–p53. PCR analyses confirmed that all six lncRNAs were detected in SFB–PURB complexes but not in SFB–p53 complexes (Extended Data Fig. 3d). These data indicate that PURB but not p53 directly binds these lncRNAs, suggesting that PURB may serve as a molecular adaptor linking p53 to lncRNAs.

We next investigated whether these lncRNAs contribute to PURB-mediated repression of p53 activity. Specific antisense oligonucleotides (ASOs) were used to deplete each of the six lncRNAs in U2OS cells (Extended Data Fig. 4a). Several controls were included, such as a scrambled ASO (ASO-Scr) and an unrelated ASO targeting GAS5. Knockdown of HOTAIR (ASO-HOTAIR) resulted in significant activation of p53 target gene p21, whereas ASO-Scr and ASO-GAS5 did not produce similar effects (Fig. 4a). Importantly, ASO-HOTAIR-mediated activation was promoter specific and did not alter p53 protein levels; although p21 expression was markedly upregulated, no significant changes were observed in other p53 targets such as MDM2 or PUMA (Fig. 4b,c and Extended Data Fig. 4b). Similar results were obtained in HCT116 cells (Extended Data Fig. 4c–g). These results suggest that HOTAIR is selectively involved in PURB-mediated modulation of p53 function.

Fig. 4 |. The repressive function of PURB on p53 requires HOTAIR.

a, Western blot analysis of p53 and p21 protein levels in U2OS cells transfected with ASOs targeting different lncRNAs. Scr., scrambled control. b, Western blot analysis of p53, p21, PUMA and MDM2 protein levels upon depletion of HOTAIR by three sets of ASOs in U2OS cells. c, Quantitative PCR (qPCR) analysis of p21 mRNA expression (n = 3) upon depletion of HOTAIR by three sets of ASOs in U2OS cells. d, Diagram showing HOTAIR consisting of F1, F2, F3 and F4. e, In vitro binding of purified PURB protein and biotin-labeled HOTAIR RNA. f, Diagram showing F1 fragment of HOTAIR. F1-Δ: deletion of PURB-binding sequence; F1-Mut: mutant that abolishes PURB binding. g, EMSA showing PURB–HOTAIR RNA (F1) complex formation in vitro and cold probe competition. h, Luciferase reporter assay of p21 promoter at 24 h after transfection of the indicated plasmids in the indicated HCT116 cells (n = 3). i, Western blot analysis of protein levels in control, p53^−/− or PURB^−/− U2OS cells transfected with ASOs targeting HOTAIR. Data are presented as the mean ± s.d. of independent biological repeats. P values were calculated using an unpaired, two-tailed Student’s t-test. The experiments in a,b,e,g,i were repeated three times with similar results and representative results are shown.

HOTAIR is known for roles in transcriptional regulation, epigenetic modification and chromatin dynamics^38,39 and has been implicated in promoting tumorigenesis^39–42. To further investigate its role, we mapped the PURB-binding region within HOTAIR. The FL HOTAIR transcript (~2,148 nt) was divided into four fragments (F1–F4; Fig. 4d). In vitro binding assays with purified PURB protein and biotin-labeled HOTAIR RNA demonstrated that PURB binds specifically to F1 (Fig. 4e). PURB was originally identified as a purine-rich DNA/RNA-binding protein²². Notably, a purine-rich motif (AGAGAAAAGG) within F1 was critical for binding; a mutant HOTAIR lacking this motif (Mut-P) failed to bind PURB (Fig. 4e), indicating a sequence-specific interaction. Electrophoretic mobility shift assays (EMSAs) further confirmed the direct and specific interaction between PURB and HOTAIR (Fig. 4f,g). A strong PURB–HOTAIR complex was detected when radiolabeled F1 RNA was incubated with PURB but not with p53 (lanes 2 and 3, Fig. 4g). The complex was supershifted by anti-PURB antibody (lane 4) and competition assays showed that binding was abolished by excess wild-type F1 cold probe but not by mutant F1 cold probes lacking the purine-rich element (lanes 5–7, Fig. 4g). To examine these interactions under physiological conditions, we conducted nondenaturing RIP assays in U2OS cells. HOTAIR was detected in p53 immunoprecipitates from wild-type cells but not from p53-deficient or PURB-deficient cells (Extended Data Fig. 4h,i), indicating that PURB is required to bridge p53 and HOTAIR. In contrast, HOTAIR was consistently associated with PURB even in the absence of p53, consistent with direct PURB–HOTAIR binding.

Next, we assessed whether HOTAIR expression could suppress p53-mediated transcription. p53-null HCT116 cells were cotransfected with p53 and increasing amounts of HOTAIR along with a p21 promoter luciferase reporter. Although p53 alone robustly activated the p21 promoter, coexpression of HOTAIR significantly repressed p21 activation in a dose-dependent manner (Fig. 4h). However, in p53/PURB double-null cells, HOTAIR coexpression failed to suppress p21 activation, suggesting that HOTAIR-mediated repression of p53 transactivation is PURB dependent. At the endogenous level, HOTAIR knockdown in U2OS cells upregulated p21 expression but this effect was abolished in p53-null or PURB-null cells (Fig. 4i), suggesting that the HOTAIR-mediated effect on p21 regulation is dependent on both p53 and PURB. Notably, MDM2 and PUMA were unaffected by HOTAIR knockdown, supporting the notion of promoter-specific regulation. These observations were corroborated in another wild-type p53 cancer cell line (A375) (Extended Data Fig. 4j).

Together, these data indicate that HOTAIR specifically binds PURB in a sequence-dependent manner and the PURB–HOTAIR complex has a crucial role in promoter-specific repression of p53 target genes.

Disruption of the HOTAIR–PURB interaction inhibited tumor growth in p53-dependent manner

Having identified the region of HOTAIR (nucleotides 371–380) necessary for PURB binding, we next examined whether this interaction is critical for regulating p53 function. Using CRISPR–Cas9, we introduced a partial deletion on the HOTAIR gene to express a mutant form of HOTAIR lncRNA (HOTAIR-Δ365–406) defective for PURB binding (Fig. 5a and Extended Data Fig. 5a). As confirmed by CLIP assays, the HOTAIR–PURB interaction was completely abolished in A375 cells expressing single-guide RNA (sgRNA) targeting HOTAIR (sg-HOTAIR) (Fig. 5b). Correspondingly, p21 mRNA expression was increased in sg-HOTAIR cells whereas MDM2 and PUMA expression remained unchanged (Fig. 5c), mirroring the promoter-specific effects previously observed for PURB (Fig. 3). Moreover, PURB knockdown increased p21 expression in control A375 cells but not in sg-HOTAIR cells (Fig. 5d), indicating that HOTAIR is required for PURB-mediated p21 regulation.

Fig. 5 |. Sequence-specific binding of HOTAIR by PURB is critical for the function of PURB–HOTAIR complex.

a, Diagram showing CRISPR–Cas9-mediated deletion of nucleotides 365–406 of HOTAIR at the genomic level. b, qPCR analysis of HOTAIR bound with endogenous PURB protein purified by CLIP in sg-HOTAIR A375 cells versus control cells (n = 3). c, qPCR analysis of gene expression in sg-HOTAIR A375 cells versus control cells (n = 3). d, Western blot analysis of protein levels in sg-HOTAIR A375 cells or control cells transfected with control siRNA or siRNA against PURB. e, Xenograft tumors derived from indicated A375 cells. f, Tumor weights shown as the mean ± s.e.m. of n = 7 independent tumor samples. g, Xenograft tumors derived from indicated A375 cells. h, Tumor weights shown as the mean ± s.e.m. of n = 7 independent tumor samples. Data in b,c are presented as the mean ± s.d. of independent biological repeats. P values were calculated using an unpaired, two-tailed Student’s t-test. The experiment in d was repeated three times with similar results and representative results are shown.

To evaluate the functional importance of HOTAIR–PURB binding in vivo, we performed xenograft tumor assays. Tumor growth was significantly suppressed in HOTAIR-Δ365–406 wild-type p53 cells but not in p53-null counterparts (Fig. 5e,f), with similar results observed in HCT116 xenografts (Extended Data Fig. 5b,c). Notably, loss of PURB expression in HOTAIR-Δ365–406 wild-type p53 cells had the same tumor growth-repressive effect as in PURB-null cell or HOTAIR-Δ365–406 cells (Fig. 5g,h). These findings demonstrate that the HOTAIR–PURB interaction is critical for modulating p53-mediated tumor growth suppression.

Mechanistic insights into HOTAIR-mediated transcriptional repression

Previous studies showed that HOTAIR represses gene transcription by recruiting the EZH2-containing Polycomb repressive complex 2 to induce histone H3 lysine 27 trimethylation (H3K27me3)^39,40. Using CLIP assays, we confirmed the direct interaction between HOTAIR and EZH2 (Fig. 6a). To investigate whether HOTAIR regulates p53-dependent transcription through EZH2, we assessed whether HOTAIR mediates the interaction between p53 and EZH2. In vitro pulldown assays revealed that p53 retrieved EZH2 from HeLa nuclear extracts (lane 3, Fig. 6b); however, this interaction was abolished when the nuclear exact was treated with RNase (lane 4, Fig. 6b) or the cells were transfected with ASOs targeting HOTAIR (lane 6, Fig. 6b), indicating that HOTAIR is required for p53–EZH2 association. Moreover, co-immunoprecipitation further showed that endogenous EZH2 coprecipitated with p53 in A375 cells but not in sg-HOTAIR or sg-PURB cells (Fig. 6c), suggesting that the HOTAIR–PURB complex bridges p53 and EZH2.

Fig. 6 |. HOTAIR recruits EZH2 to the p53 target promoter for transcriptional repression.

a, qPCR analysis of HOTAIR RNA bound with endogenous p53 or EZH2 protein in CLIP assay (n = 3). b, Western blot analysis of the endogenous EZH2 protein retrieved by purified p53 protein from HeLa cell nuclear extract. c, Western blot analysis of the endogenous EZH2 and PURB proteins and reverse transcription PCR (RT–PCR) analysis of HOTAIR lncRNA retrieved by endogenous p53 protein in co-immunoprecipitation assay in indicated A375 cells. d, In vitro binding analysis of biotin-labeled DNA containing p21 promoter region (160 bp), purified proteins (p53, PURB and EZH2) and in vitro transcribed HOTAIR RNA. e, ChIP analysis of p53–PURB–HOTAIR-mediated recruitment of EZH2 to p21 promoter in A375 cells (n = 3). f, ChIP analysis of p53–PURB–HOTAIR-mediated alteration of H3K27me3 occupancy on p21 promoter in A375 cells (n = 3). g, ChIP analysis of EZH2 occupancy at p21 promoter in sg-HOTAIR or sg-PURB in A375 cells (n = 3). h, ChIP analysis of H3K27me3 occupancy at p21 promoter in sg-HOTAIR or sg-PURB in A375 cells (n = 3). Data in a,e–h are presented as the mean ± s.d. of n = 3 independent biological repeats. P values were calculated using an unpaired, two-tailed Student’s t-test. The experiments in b–d were repeated three times with similar results and representative results are shown.

To evaluate the role of EZH2 in PURB-mediated repression, we knocked down PURB, EZH2 or both in A375 cells. Depletion of either PURB or EZH2 increased p21 expression and combined knockdown did not further enhance p21 activation (Extended Data Fig. 6a), indicating that EZH2 is critical for the PURB-mediated repression of p21. We next investigated whether p53, PURB, HOTAIR and EZH2 can form a quaternary complex on the p21 promoter. To this end, we first performed the in vitro binding assay by incubating the biotin-labeled DNA fragments derived from the p21 promoter with purified p53, PURB, HOTAIR and EZH2. The results showed that EZH2 was recruited to the p21 promoter by p53 only in the presence of both PURB and HOTAIR (lane 8, Fig. 6d). ChIP analysis confirmed that p53-dependent recruitment of EZH2 to the p21 promoter was significantly enhanced by coexpression of PURB and HOTAIR (Fig. 6e), along with increased H3K27me3 levels (Fig. 6f). No similar effects were observed at the MDM2 or PUMA promoters (Extended Data Fig. 6b–e). Additionally, endogenous EZH2 occupancy and H3K27me3 levels at the p21 promoter were markedly reduced in sg-HOTAIR or sg-PURB cells but not at MDM2 or PUMA promoters (Fig. 6g,h and Extended Data Fig. 6f–i), confirming the promoter-specific action of the HOTAIR–PURB complex. Collectively, these results indicated that the PURB–HOTAIR complex exerts repression on p53-mediated transcription through recruiting EZH2 and upregulating H3K27me3 levels.

Given that HOTAIR adopts a complex structure comprising four distinct fragments (F1–F4)^43,44 (Supplementary Fig. 1a) and is able to bridge EZH2 and PURB through different regions, we further explored how HOTAIR structure affects p53–PURB–EZH2 interactions. The 5′ region (nucleotides 1–300) within F1 of HOTAIR mediates EZH2 binding³⁸ (Supplementary Fig. 1a,b). Notably, the PURB recognition sequence (nucleotides 371–380), which we identified, is also situated within F1, specifically on the loop region of the stem-loop structure highlighted in Supplementary Fig. 1b. To examine whether the secondary structure of HOTAIR affects its interaction with the EZH2 or the PURB–p53 complex, we generated several HOTAIR structural mutants (Supplementary Fig. 2). Structural mutant analyses revealed that HOTAIRΔ1–300 and Δ1–150 mutants lost EZH2 binding, while their binding to PURB remained fully intact (Extended Data Fig. 7a,b), indicating that EZH2 and PURB interact with HOTAIR independently through different sites. Conversely, the HOTAIR Mut-P, where the residues required for PURB binding are substituted, was completely defective for PURB binding (Extended Data Fig. 7b). Notably, the interactions between PURB and HOTAIR were also significantly affected by HOTAIR Mut-LS and HOTAIR-Δ381–410 mutant, both of which retained the binding sequence of PURB but altered the secondary structure near the binding sequence (Extended Data Fig. 7b). Collectively, these findings demonstrate that PURB and EZH2 bind independently to distinct regions of HOTAIR and that both the primary binding elements and the secondary structure of HOTAIR are critical for assembling the PURB–HOTAIR–EZH2 complex (Extended Data Fig. 7d).

We also tested whether these mutants affect the ability of HOTAIR to form a complex with PURB–p53. As shown in Extended Data Fig. 7c, the Δ1–150 mutant retained the ability to bind the PURB–p53 complex; by contrast, the HOTAIR Mut-LS, which lost PURB binding, also failed to bind p53. Together, these data demonstrate that PURB and EZH2 interacts independently with HOTAIR at different sites and the PURB–HOTAIR–EZH2 complex is regulated by both the binding elements and the secondary structure of HOTAIR (Extended Data Fig. 7d).

A unique cis-regulatory DNA element at the p21 promoter is critical for the recruitment of PURB by p53 in a promoter-specific manner

To dissect the precise mechanism by which PURB modulates p53-mediated transcription in a promoter-specific manner, we transfected H1299 cells with a p53 expression vector in the presence or absence of a vector encoding HA-tagged PURB. ChIP assays revealed that expression of PURB did not significantly affect the DNA-binding activity of p53 on the p21 promoter (Fig. 7a). Interestingly, PURB alone was not sufficient to bind the p21 promoter; however, upon coexpression with p53, PURB was specifically recruited to the p21 promoter but not the promoters of either MDM2 or PUMA (Fig. 7b). Moreover, the ChIP analysis by the anti-PURB antibody in HCT116 cells revealed that the recruitment of endogenous PURB, specifically, to the p21 promoter occurred in native HCT116 cells but not in p53-null cells (Fig. 7c). Conversely, PURB was not recruited to the PUMA or MDM2 promoters, even in the presence of p53 (Extended Data Fig. 8a). These data demonstrate that PURB is recruited by p53 specifically to the p21 promoter but not to the promoters of PUMA or MDM2.

Fig. 7 |. A unique cis-regulatory element at the p21 promoter is critical for the recruitment of PURB by p53 in a promoter-specific manner.

a, ChIP analysis of p53 occupancy at p21 promoter in H1299 cells transfected with p53 alone or p53 together with PURB (n = 3). b, ChIP analysis of PURB occupancy at indicated promoters in H1299 cells transfected with PURB alone or PURB together with p53 (n = 3). c, ChIP analysis of PURB occupancy at p21 promoter in wild-type or p53-null HCT116 cells (n = 3). d, In vitro binding of purified PURB protein and biotinlabeled DNA containing p21 or MDM2 promoter region (160 bp) with or without p53 protein. RE, response element. e, In vitro binding of purified proteins (p53 and PURB) and biotin-labeled DNA containing p21 promoter region (160 bp) with or without PMS. f, Luciferase reporter assay of the wild-type p21 promoter (with PMS) at 24 h after transfection of the indicated plasmids in H1299 cells (n = 3). g, Luciferase reporter assay of the mutated p21 promoter (without PMS) at 24 h after transfection of the indicated plasmids in H1299 cells (n = 3). P values were calculated using a one-way ANOVA. h, ChIP analysis of p53 or PURB occupancy on the wild-type p21 promoter (with PMS) in H1299 cells transfected with the wild-type p21 promoter construct together with other indicated plasmids (n = 3). i, ChIP analysis of p53 or PURB occupancy on the mutated p21 promoter (without PMS) in H1299 cells transfected with the mutated p21 promoter construct together with other indicated plasmids (n = 3). Data in a–c,f,h,i are presented as the mean ± s.d. of n = 3 independent biological repeats. P values were calculated using an unpaired, two-tailed Student’s t-test. The experiments in d,e were repeated three times with similar results and representative results are shown.

Next, we tested whether this specific recruitment can be recapitulated in vitro using a purified system. Flag-tagged p53 and HA-tagged PURB proteins were purified from human cells under stringent conditions to remove associated factors. Biotin-labeled DNA fragments (~160 bp) containing the p53-binding site and flanking sequences from the p21 or MDM2 promoter were used as DNA probes for the protein–DNA complex formation assays. Western blot analysis of protein–DNA complexes precipitated by streptavidin beads showed that p53 bound strongly to both the p21 and MDM2 promoters (lanes 4 and 8, Fig. 7d). By contrast, PURB alone failed to bind either promoter effectively (lanes 3 and 7, Fig. 7d). However, upon coincubation with p53, PURB was robustly recruited to the p21 promoter but not to the MDM2 promoter (lanes 4 and 8, Fig. 7d), consistent with the ChIP data. Interestingly, upon long exposure, a weak binding of PURB alone was visible on the p21 promoter probe but not on the MDM2 promoter probe (lanes 3 and 7, Fig. 7d). As no other proteins were present in this system, we speculate that a unique DNA sequence present at the p21 promoter but absent at the MDM2 promoter facilitates this weak binding by PURB.

We next analyzed the flanking regions of p53-binding sites on p21, MDM2 and PUMA promoters using motif-based sequence analysis tools (MEME website)⁴⁵. Indeed, we identified two unique DNA sequences present on the p21 promoter but not on the MDM2 or PUMA promoters (Extended Data Fig. 8b). We named this sequence the PURB-mediated promoter-specific sequence (PMS). To investigate whether the PMS is required for PURB binding, we performed in vitro binding assays using biotin-labeled DNA probes containing either the wild-type p21 promoter sequence or a version with the PMS deleted. Loss of the PMS had no effect on p53 binding but completely abolished PURB recruitment by p53 (Fig. 7e). Furthermore, the weak binding of PURB alone observed in vitro was also diminished when the PMS was deleted, suggesting that PURB can directly recognize the PMS.

We next examined the functional role of PMS in PURB-mediated regulation of p21 expression in cells. H1299 cells were cotransfected with p53 expressing vector, varying amounts of PURB expressing vector and a luciferase reporter driven by either the wild-type p21 promoter or a PMS-deleted version. As expected, p53 induced activation of the wild-type p21 reporter and coexpression of PURB significantly repressed p21 expression in a dose-dependent manner (Fig. 7f). However, when the PMS was deleted, PURB-mediated repression was completely abolished (Fig. 7g). Lastly, ChIP assays performed on the reporter constructs in transfected H1299 cells showed that p53 binding to the promoter remained unaffected by PMS deletion but recruitment of PURB was completely lost when PMS was deleted (Fig. 7h,i). These experiments, using transiently transfected constructs, directly demonstrated the critical role of PMS in PURB recruitment.

Together, these data reveal that a unique cis-regulatory element, PMS, located at the p21 promoter, is specifically recognized by PURB. Loss of PMS does not impair p53 binding but abolishes the ability of p53 to recruit PURB to the p21 promoter, thereby preventing PURB-mediated transcriptional repression.

The mechanism of PURB-dependent promoter-specific regulation is widespread in nature

To further determine the role of PURB in regulating p53-mediated transactivation in a promoter-specific manner, we investigated whether the same mechanism applies to additional p53 target promoters. To this end, we sought to identify other p53 targets regulated by PURB. We first performed RNA sequencing (RNA-seq) analysis in the PURB-knockdown cells and indeed, in addition to p21, several known p53 targets were identified, including GADD45, BTG2, GLS2, TIGAR, RRAD and GDF15 (Fig. 8a). To validate whether these newly identified targets are indeed regulated by PURB, we examined their expression levels following PURB knockdown. Consistently, mRNA expression of p21 and all six additional targets was significantly upregulated upon PURB knockdown (Fig. 8b). By contrast, the expression of several other p53 targets—including apoptosis-related genes (PUMA, TNFRSF10B, SCN3B, TP53I3 and TRAF4), DNA repair-related genes (RRM2B, XPC and DDB2), and feedback loop regulators (MDM2 and PPM1D)—was not significantly affected under the same conditions (Extended Data Fig. 8c).

Fig. 8 |. The mechanism of PURB-dependent promoter-specific regulation is widespread in nature.

a, Top, heat map of gene expression profile in U2OS cells with PURB inducible knockdown for 72 h versus control U2OS cells, revealed by RNA-seq. Bottom, western blot analysis of PURB protein levels in PURB inducible knockdown U2OS cells or control U2OS cells. b, qPCR analysis of gene expression with PURB inducible knockdown for 72 h versus control U2OS cells (n = 3). Left, cell-cycle control genes; right, metabolism-related genes. c, The 8-nt consensus DNA sequences identified by MEME in the flanking region of p53 response element (RE) on the promoters of PURB-regulated genes. d, ChIP analysis of PURB occupancy at indicated promoters in H1299 cells transfected with PURB alone or PURB together with p53 (n = 3). e, Western blot analysis of p53 and PURB proteins in ChIP assay in d. Data in b,d are presented as the mean ± s.d. of n = 3 independent biological repeats. P values were calculated using an unpaired, two-tailed Student’s t-test. The experiments in a,e were repeated three times with similar results and representative results are shown.

Given the promoter-specific recruitment mechanism observed at the p21 promoter, we next asked whether similar cis-regulatory PMS elements are present at the promoters of these newly identified PURB-regulated p53 targets. Using MEME, we identified sequences highly similar to the PMS motif in the flanking regions of the p53-binding sites at these target promoters (Fig. 8c and Supplementary Fig. 3a,b). The raw sequences and nucleotide distance between the PMS and p53 response element sequences in each of these promoters are shown in Supplementary Fig. 3b. Furthermore, ChIP analysis confirmed that PURB was recruited to the promoters of all six newly identified targets (Fig. 8d,e), whereas no PURB recruitment was detected at promoters of p53 targets unresponsive to PURB knockdown (Supplementary Fig. 3c).

Together, these findings demonstrate that PURB is recruited specifically to p53 target promoters containing a unique cis-regulatory PMS sequence, such as the p21 promoter, to mediate transcriptional repression. Moreover, our analysis reveals that the mechanism of PURB-dependent promoter-specific regulation is widespread in nature and extends to a broader subset of p53-regulated genes.

Discussion

Promoter-specific transcriptional regulation has a crucial role in orchestrating the expression of distinct subsets of target genes, allowing cells to respond efficiently to diverse stresses. It has long been recognized that p53-mediated transcriptional regulation operates in a promoter-specific manner^7,46, although the underlying molecular mechanisms have remained largely unknown. In this study, we identify PURB as a binding partner and promoter-specific corepressor of p53. PURB directly binds to the CTD of p53; however, this interaction alone is insufficient for PURB co-occupancy on p53 target promoters. We demonstrate that PURB is recruited to specific p53 target promoters in a gene-specific manner, determined by a unique a DNA sequence flanking the p53 response elements on promoters. Although this unique DNA sequence is dispensable for p53 DNA binding, they are critical for the recruitment of PURB. Notably, in the absence of p53, no recruitment of PURB is detected in vivo, indicating that the recruitment of PURB to the promoter requires both the cis-regulatory DNA sequence and p53 binding.

C-terminal acetylation of p53 has been shown to regulate p53 activity through multiple mechanisms, including conformational changes, modulation of cofactor interactions and crosstalk with other modifications¹⁷. Here, we demonstrated that PURB preferentially binds to unacetylated p53 C-terminal tails and that p53 acetylation abolishes this interaction, resulting in the transcriptional activation of specific target genes. This finding reveals a model wherein acetylation dynamically regulates p53 activity by modulating cofactor binding in a promoter-specific manner.

Furthermore, we uncover that PURB, a known RNA-binding protein²⁴, functions as a molecular bridge connecting p53 with the lncRNA HOTAIR. Although PURB can potentially interact with several lncRNAs, our data highlight HOTAIR as a key mediator of PURB-dependent repression of p53 function. The functional form of the PURB repressor is known to be a homodimer, allowing it to simultaneously bind a cis-regulatory DNA element and HOTAIR through distinct domains. Notably, HOTAIR does not bind p53 directly but interacts with PURB in a sequence-specific manner. Depletion of either HOTAIR or PURB results in p53 activation and tumor growth suppression, emphasizing the functional importance of this axis. Using chromatin isolation by RNA purification (ChIRP) combined with qPCR, we confirmed that HOTAIR is selectively recruited to p53 target promoters, including p21, GADD45A, BTG2, GLS2 and TIGAR, but not to PUMA, TNFRSF10B or MDM2 promoters (Extended Data Fig. 9), mirroring the pattern of PURB occupancy on these promoters. HOTAIR has been implicated in regulating chromatin dynamics and transcription through various mechanisms^38–40,47. In our study, we show that the PURB–HOTAIR complex recruits EZH2 to these promoters, promoting H3K27me3 and establishing a transcriptionally repressive chromatin state (Extended Data Fig. 10).

Beyond HOTAIR, our data suggest that PURB may interact with other lncRNAs, including linc-ROR, NEAT1, SNHG1, GAS5 and MALAT1. Modest activation of p21 was observed upon knockdown of MALAT1 and linc-ROR (Extended Data Fig. 4d), suggesting that additional lncRNAs may participate in PURB-mediated regulation in certain circumstances. However, further studies are needed to determine whether these lncRNAs also interact with PURB in a sequence-specific manner. Interestingly, previous studies showed that MEG3, another lncRNA, induces p53-dependent cell-cycle arrest without promoting apoptosis in HCT116, although the underlying mechanism remains unknown⁴⁸. It is possible that different lncRNAs regulate p53-mediated transcription in a promoter-specific manner by recruiting unique cofactors. Lastly, NEAT1—a critical component of paraspeckles⁴⁹—is known to be transcriptionally activated by p53 and to have important roles in tumor development^49,50. It would be intriguing to investigate whether PURB is also a component of paraspeckles and whether the oncogenic role of PURB also acts through paraspeckles independent of transcription regulation.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41594-025-01597-3.

Methods

The study complied with relevant ethical regulations and all experimental protocols were approved by the Institutional Animal Care and Use Committee (IACUC) of Columbia University.

Cell culture, stable lines and reagent treatment

H1299 (p53-null) (CRL-5803), U2OS (p53-wild-type) (HTB-96), A375 (p53-wild-type) (CRL-1619), HCT116 (CCL-247) and MCF-7 (p53-wild-type) (CRL-3435) cell lines were obtained from the American Type Culture Collection (ATCC). Cells were cultured in DMEM supplemented with 10% FBS (Gibco), 100 U per ml penicillin and 100 μg ml⁻¹ streptomycin at 37 °C in a humidified 5% CO₂ incubator. Human primary dermal fibroblasts (PCS-201-012, ATCC) were cultured in fibroblast growth medium (Sigma-Aldrich, C-23010). All cell lines were verified as Mycoplasma-negative and not listed in the International Cell Line Authentication Committee database. Freshly thawed seed cells were cultured for no longer than 2 months, with weekly morphology checks against ATCC images to avoid cross-contamination. U2OS, HCT116 and A375 p53-null cells were previously generated^28,51. H1299 cells stably expressing SFB–p53 (100–393) or SFB–PURB were generated by transfecting expression plasmid into cells followed by puromycin selection (2.0 μg ml⁻¹). PURB inducible knockdown cells were generated by lentivirus-based transduction of TRIPZ inducible short hairpin RNA (shRNA) (Dharmacon; PURB shRNA1, RHS4696-200693392; PURB shRNA2, RHS4696-200776551; PURB shRNA3, RHS4696-200776791). Knockdown of PURB was induced by adding doxycycline (5.0 μg ml⁻¹) to the medium. DNA damage inducer doxorubicin (Sigma-Aldrich, D1515; CAS: 25316-40-9) was used at 0.2 μg ml⁻¹. Deacetylase inhibitor TSA or NAM was used for 6 h at 1 μM or 5 mM, respectively.

Animals

Mice were housed in temperature-controlled rooms (18–24 °C (65–75 °F), 40–60% humidity, 12-h light–dark cycle). All procedures were approved by the IACUC of Columbia University (AAAW1456) under the supervision of the Institute of Comparative Medicine. The 6-week-old female nude mice (NU/NU, Charles River; ~24.7 ± 1.77 g) were subcutaneously injected with 2.0 × 106 A375 or HCT116 cells mixed 1:1 with Matrigel (BD Biosciences). Mice were killed 3–4 weeks after injection and tumor weights were recorded. Tumor size did not exceed ethical limits (10% body weight or 2-cm diameter).

Plasmids, siRNAs, ASOs and transfection

Full-length p53 plasmid was previously generated²⁸. Full-length PURB cDNA was amplified from H1299 cDNA. LZRS-HOTAIR (26110) and pcDNA3.1–3×Flag-EZH2 (173717) plasmids were obtained from Addgene. HA–CBP was constructed previously²⁸. Constructs were subcloned into pCDNA3.1-topo, pRK5-MYC or SFB vectors. Mutants and deletions were generated using the QuickChange XL site-directed mutagenesis kit (Agilent, 200516). Transient knockdowns of PURB and EZH2 were achieved using SMARTpool siRNAs (Dharmacon; PURB, L-012136-00-0005; EZH2, L-004218-00-0005). lncRNA knockdowns, including HOTAIR, were performed using ASOs (Integrated DNA Technologies; sequences in Supplementary Table 2). Transfections were conducted using Lipofectamine 3000 (Invitrogen, L3000150) following the manufacturer’s protocol.

Complex purification and MS analysis

Cells were lysed in harvest buffer (10 mM HEPES pH 8.0, 50 mM NaCl, 0.5 M sucrose, 0.1 mM EDTA and 0.25% Triton X-100) on ice and nuclear fractions were isolated by centrifugation. Nuclei were lysed in BC100 buffer (50 mM Tris-HCl pH 7.3, 100 mM NaCl, 0.1 mM EDTA, 0.4% NP-40, 10% glycerol and protease inhibitors) for 1 h at 4 °C. After centrifugation and filtration (0.45 μm), lysates were sequentially immunoprecipitated with streptavidin beads (Cytiva, GE17-5113-01) and (Millipore, 69704), with biotin (2 mg ml⁻¹) and 100 mM glycine-HCl (pH 2.6) for elution. Samples were resolved on 4–20% SDS–PAGE and stained with GelCode Blue (Thermo Fisher, 24592). Gel bands were subjected to LC–MS/MS at the Harvard Center for MS. Proteomics data were deposited to ProteomeXchange through the PRIDE database (accession code: PXD042681)⁵².

Co-immunoprecipitation assay

Cells were washed with ice-cold PBS, lysed in BC100 buffer with protease inhibitors, 1 mM DTT and 1 mM PMSF and briefly sonicated. Supernatants were collected by centrifugation at 15,000g for 15 min. For cytosolic and nuclear fractionation, cells were first incubated with Harvest buffer, then nuclei were washed with buffer A (10 mM HEPES pH 8.0, 10 mM KCl, 0.1 mM EDTA and 0.1 mM EGTA) and lysed with BC100 buffer. Lysates were quantified by the Bradford method (Bio-Rad) using BSA as a standard. Then, 2 μg of antibody or 20 μl of streptavidin beads were incubated with lysates for 4 h at 4 °C. If using antibody, 20 μl protein A/G agarose beads was added to the lysates after antibody incubation. Complexes were washed with BC100 and eluted with Laemmli buffer, 100 mM glycine-HCl (pH 2.6) or biotin (2 mg ml⁻¹) for western blot.

Protein purification

H1299 cells were transfected with Flag-tagged p53, HA-tagged PURB or Flag-tagged EZH2 plasmids and harvested after 24 h. Cells were lysed in Flag lysis buffer (50 mM Tris-HCl pH 8.0, 137 mM NaCl, 1% Triton X-100, 0.2% Sarkosyl, 1 mM NaF, 1 mM Na₃VO₄, 10% glycerol, protease inhibitors, 1 mM DTT and 1 mM PMSF). HA affinity beads (Sigma-Aldrich, A2095) or M2 Flag beads (Sigma-Aldrich, A2220) were incubated with lysates overnight. After five washes with BC200 buffer (20 mM Tris-HCl pH 7.9, 200 mM NaCl, 10% glycerol, 0.2 mM EDTA and 0.1% Triton X-100), proteins were eluted with Flag (Sigma-Aldrich, F3290) or HA peptide (Sigma-Aldrich, I2149).

Western blot and antibodies

Cells were lysed in Flag lysis buffer, protein concentration was measured and equal amounts of protein from different experimental groups were used for western blot. Protein separation was performed using 4–20% precast SDS–PAGE gels (Invitrogen, XP0420) and transferred to a membrane. Primary antibodies used included anti-Vinculin (Sigma-Aldrich, V9131; 1:10,000), anti-PURB (Bethyl Laboratories, A303–650A; 1:1,000), anti-PURB (Proteintech, 18128–1-AP; immunoprecipitation, ChIP or 1:1,000 for western blot), anti-p53 (DO-1; Santa Cruz Biotechnology, sc-126; 1:1,000), anti-p53 (FL393; Bioss, bs-8687R; immunoprecipitatio or ChIP), anti-Ac p53 CTD (generated previously²⁸), anti-p21 (12D1; Cell Signaling Technology, 2947; 1:1,000), anti-PUMA (H-136; Santa Cruz Biotechnology, sc-28226; 1:500), anti-MDM2 (Ab-5; Millipore, OP-145; 1:500), anti-Flag (Sigma-Aldrich, F3165; ChIP or 1:5,000 for western blot), anti-HA (3F10; Roche, 11867423001; ChIP, nondenaturing RIP or 1:2,000 for western blot), anti-MYC (9E10; Santa Cruz Biotechnology, sc-40; 1:1,000), anti-H3K27me3 (Abcam, ab6002; ChIP) and anti-EZH2 (Cell Signaling Technology, 3147; CLIP, ChIP or 1:1,000 for western blot). Secondary antibodies used were horseradish-peroxidase-conjugated anti-mouse (SouthernBiotech, 1031-05; 1:5,000), anti-rabbit (SouthernBiotech, 4050-05; 1:5,000) and anti-rat (SouthernBiotech, 3030-05; 1:5,000).

Nondenaturing RIP

Nondenaturing RIP was performed according to previously described protocols⁵³ with modifications. Cells were collected, washed with PBS and lysed in nondenaturing RIP lysis buffer (150 mM KCl, 25 mM Tris pH 7.4, 5 mM EDTA, 0.5 mM DTT, 0.5% NP-40, 100 U per ml RNase inhibitor (SUPERase•in, added fresh each time) and protease inhibitors) for 30 min at 4 °C. Lysates were centrifuged at 20,000g for 30 min and the cleared supernatant was collected. After protein quantification and normalization, a fraction was saved as input. For pulldown, streptavidin beads presaturated with BSA and yeast tRNA (Invitrogen, AM7119) were incubated with lysates for 4 h at 4 °C. For antibody-based RIP, 2 μg of the indicated antibody was added to the lysates and incubated overnight at 4 °C, followed by 2-h incubation with 20 μl of BSA/tRNA-saturated protein A/G agarose beads. Beads were washed twice with RIP wash buffer (RIP lysis buffer, 0.05% NP-40 and 1 mM MgCl₂). DNA contaminants were removed using Turbo DNase (Invitrogen, AM2238). Small aliquots were saved for western blot analysis. Bound RNA was extracted with TRIzol and isopropanol precipitation, followed by a second DNase I treatment. RNA was reverse-transcribed and analyzed by qPCR. Primers for nRIP–qPCR are listed in Supplementary Table 3.

UV CLIP

CLIP assays were conducted according to established methods⁵⁴ with modifications. Cells were UV-irradiated (254 nm, 400 mJ cm⁻²) and lysed in CLIP lysis buffer (50 mM Tris-HCl pH 7.4, 100 mM NaCl, 1% NP-40, 0.1% SDS, 0.5% sodium deoxycholate, 100 U per ml RNase inhibitor and protease inhibitors) on ice for 20 min. Lysates were sonicated, digested with Turbo DNase and centrifuged at 22,000g for 20 min at 4 °C. The cleared supernatant was collected and normalized; then, a fraction was saved as input. For immunoprecipitation, lysates were incubated with BSA/tRNA-presaturated streptavidin beads for 2 h or with the indicated antibody overnight at 4 °C, followed by 2-h incubation with protein A/G agarose beads. Beads were washed once with CLIP lysis buffer and three times with high-salt wash buffer (50 mM Tris-HCl pH 7.4, 1 M NaCl, 1 mM EDTA, 1% NP-40, 0.1% SDS and 0.5% sodium deoxycholate). Proteins were eluted with biotin solution (2 mg ml⁻¹) or elution buffer (1% SDS and 100 mM NaHCO₃). Following Turbo DNase and proteinase K digestion in PK urea buffer (100 mM Tris-HCl pH 7.4, 50 mM NaCl, 10 mM EDTA and 7 M urea), RNA was extracted using TRIzol. RNA was either reverse-transcribed for qPCR analysis or used for library preparation (Swift Biosciences, R1024) followed by next-generation sequencing. Sequencing reads were processed with Cutadapt (adaptor removal and trimming reads < 10 nt), mapped to hg19 using STAR (version 2.7.11b) and filtered with SAMtools (version 1.21) (mapping quality ≥ 40). Read counts were obtained using ‘featureCounts’ with the ‘fraction = T’ option and normalization to fragments per kilobase of transcript per million mapped reads was performed. CLIP-seq data from H1299 cells were deposited to the Gene Expression Omnibus under accession code GSE295618. Primers for CLIP–qPCR are listed in Supplementary Table 3.

ChIP

Cells were fixed in 1% formaldehyde for 10–30 min at room temperature and quenched with 0.125 M glycine for 10 min. Cells were harvested and lysed sequentially in ChIP lysis buffer (10 mM Tris-HCl pH 8.0, 5 mM EDTA, 150 mM NaCl and 0.5% NP-40) and then radioimmunoprecipitation assay (RIPA) lysis buffer (10 mM Tris-HCl pH 8.0, 5 mM EDTA, 150 mM NaCl, 0.1% SDS, 1% Triton X-100 and 0.5% deoxycholate, supplemented with protease inhibitors). Lysates were sonicated and centrifuged at 8,000g for 10 min at 4 °C. Supernatants were precleared with salmon sperm DNA-saturated protein A agarose (Millipore, 16–157) for 1 h at 4 °C, followed by overnight incubation with the indicated antibody or IgG control. Protein A agarose beads were added and incubated for an additional 4 h at 4 °C. Beads were sequentially washed with RIPA buffer, high-salt wash buffer (20 mM Tris-HCl pH 8.0, 500 mM NaCl, 5 mM EDTA, 0.1% SDS and 1% Triton X-100), LiCl wash buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 250 mM LiCl, 1% DOC and 1% NP-40) and TE buffer (10 mM Tris-HCl pH 8.0 and 1 mM EDTA). Elution was performed twice with elution buffer (1% SDS and 100 mM NaHCO₃) and reverse crosslinking was carried out at 65 °C overnight. DNA was purified using UltraPure phenol–chloroform–isoamyl alcohol (25:24:1; Invitrogen, 15593) and analyzed by qPCR. Primers for ChIP–qPCR are listed in Supplementary Table 4.

ChIRP

ChIRP was performed according to the published protocol⁵⁵ with minor modifications. Cells were crosslinked (3% formaldehyde, 20 min, room temperature), quenched (0.125 M glycine, 5 min) and lysed in lysis buffer (50 mM Tris-HCl pH 7.5, 10 mM EDTA, 1% SDS and protease and RNase inhibitors). Lysates were sonicated, centrifuged and precleared with Dynabeads MyOne Streptavidin C1 (Thermo Fisher Scientific, 65001) for 45 min at 37 °C. Supernatants were incubated with biotinylated ASOs (Sigma-Aldrich, 03–312 M) against lncRNA HOTAIR or lacZ in hybridization buffer (50 mM Tris-HCl pH 7.5, 1 mM EDTA, 1% SDS, 750 mM NaCl, 15% formamide and protease and RNase inhibitors) at 37 °C for 4 h with shaking. Complexes were pulled down using streptavidin magnetic beads and washed five times in wash buffer (2× SSC and 0.5% SDS) at 37 °C. Beads were split for RNA, DNA and protein recovery. RNA was digested in proteinase K buffer (100 mM NaCl, 10 mM Tris-HCl pH 7.0, 1 mM EDTA, 0.5% SDS and 5% proteinase K Ambion (AM2546 20 mg ml⁻¹)) and reverse-crosslinked (95 °C, 10 min). RNA was extracted (using Trizol), reverse-transcribed and qPCR-quantified. Primers for ChIP–qPCR of RNA are listed in Supplementary Table 3. DNA was eluted in elution buffer (50 mM NaHCO₃, 1% SDS and RNase A/H) at 37 °C for 30 min, digested in proteinase K buffer and extracted using UltraPure phenol–chloroform–isoamyl alcohol (25:24:1 (v/v); Invitrogen, 15593) for qPCR. Primers for ChIP–qPCR of DNA are listed in Supplementary Table 4. Protein was eluted and reverse-crosslinked by boiling in SDS sample buffer (95 °C, 10 min) for western blot.

In vitro binding assay

For GST pulldown, GST or GST–p53 (0.3 μg) purified with GST bind resin (Novagen, 70541) was incubated with HA–PURB (0.3 μg) purified with HA affinity gel in BC100 buffer for 1 h at 4 °C, washed with BC200 buffer five times and eluted by boiling with 1× Laemmli buffer for western blot.

For RNA–protein binding, an equal amount of indicated purified HA–PURB or Flag–EZH2 protein (0.3 μg) was incubated with biotin-labeled HOTAIR RNA (50 pmol) in BC100 buffer for 1 h at 4 °C. Then, 20 μl of streptavidin beads were added, followed by incubation for 1 h at 4 °C. After washing beads with BC200 buffer five times, the binding components were eluted by boiling with 1× Laemmli buffer for western blot.

For DNA–protein (RNA) binding, biotinylated DNA fragments (160 bp, 50 pmol) labeled with Pierce biotin 3′ end DNA labeling kit (Thermo Fisher, 89818) were incubated with Flag–p53, HA–PURB or Flag–EZH2 (0.3 μg) with or without in vitro transcribed HOTAIR RNA (20 pmol) in BC100 buffer for 1 h at 4 °C. Then, 20 μl of streptavidin beads were added for 1 h at 4 °C. Beads were washed with BC200 buffer five times. Proteins were eluted by boiling with 1× Laemmli buffer for western blot. For RNA detection, RNA was eluted in biotin solution (2 mg ml⁻¹), purified by phenol–chloroform extraction and alcohol precipitation, reverse-transcribed and PCR-amplified. PCR products were electrophoresed on a 2% agarose gel. Primers are listed in Supplementary Table 3.

EMSA

Purified PURB or p53 protein (0.1 μg) was incubated with the in vitro transcribed, 32P-labeled HOTAIR lncRNA (F1) probe (1 pmol) in 1× binding buffer (10 mM HEPES pH 7.6, 40 mM NaCl, 50 μM EDTA, 6.25% glycerol; 1 mM MgCl₂, 1 mM spermidine, 1 mM DTT, 50 ng μl⁻¹ BSA and 5 ng μl⁻¹ yeast tRNA) for 20 min at room temperature. The complex was analyzed by 4% native TBE–PAGE and visualized by autoradiography. For the supershift assay, anti-PURB antibody (0.1 μg) was preincubated with purified PURB in the reaction system without probe for 30 min at room temperature and then the probe was added before incubating for a further 20 min. For the competition assay, 25× cold probe (with or without mutations) was added.

In vitro transcription and RNA labeling

HOTAIR RNA was transcribed using the MEGAshortscript T7 transcription kit (Invitrogen, AM1354) from PCR-amplified templates with T7 promoters. PCR products were purified using QIAquick PCR and gel cleanup kit (28506). RNA was purified by phenol–chloroform extraction and alcohol precipitation. To label the RNA with biotin or 32P, Bio-11-UTP (Thermo Fisher, AM8450) or α-32P UTP (Perkin Elmer, BLU007H250UC) was added to the in vitro transcription system.

Real-time qPCR

Total RNA was extracted using TRIzol reagent. cDNA was reverse-transcribed using SuperScript IV VILO master mix (Invitrogen). Gene expression was normalized to GAPDH. Primers used for RT–qPCR are listed in Supplementary Table 3.

CRISPR–Cas9 system-mediated gene ablation

PURB-knockout cells were generated by cotransfection of cells with TrueCut Cas9 protein v2 (Invitrogen, A36498) along with scrambled nonspecific guide RNA (Invitrogen, A35526) or gene-specific TrueGuide synthetic guide RNAs (CRISPR760434_SGM, 5′-ATAGAACACTACGCGCAGCT-3′; CRISPR760433_SGM, 5′-CGCGCTCAAGAGCGAATTCT-3′) for 48 h, followed by cell dilution and split-cell and single-cell culture. The silencing of designated genes in the CRISPR pool cells or single-cell-derived clones was validated by western blot. A375 HOTAIRΔ365–406/Δ365–406 cells were generated by cotransfection of the indicated cells with TrueCut Cas9 protein v2 (Invitrogen, A36498) along with two gene-specific TrueGuide synthetic guide RNAs that cleave HOTAIR genomic DNA at two sites for 48 h, followed by cell dilution and split-cell and single-cell culture. The target sequences of the two sgRNAs of HOTAIR were as follows: HOTAIR sgRNA1, 5′-GCCTTTTCTCTGCCAGGACG-3′; HOTAIR sgRNA2, 5′-AGCATCGTACCTTATAAGGA-3′. The deletion of designated region was validated by Sanger sequencing of the genomic DNA fragment containing the HOTAIR locus.

Luciferase assay

The TP53 genomic region (−2.3 kb to +100 bp) was subcloned into pGL3-Basic (E1751, Promega) for pGL3-p21 promoter-Luc. pGL3-p21-promopter(ΔPMS)-Luc was constructed using the QuickChange XL site-directed mutagenesis kit (Agilent, 200516). Cells were cotransfected with 10 ng of firefly reporter, 10 ng of Renilla control and indicated constructs (20 ng of p53 or empty vector; total: 1,200 ng) for 24 h. Luciferase activity was measured by dual-luciferase assay (Promega, E1910) using a GloMax Discover microplate reader.

Colony formation assay

A total of 500 cells were seeded in 6-cm dishes with three biological replicates and cultured as indicated. Cells were fixed with 4% paraformaldehyde, stained with 0.2% crystal violet solution and photographed. Colony numbers were counted.

Immunofluorescence for BrdU staining

Cells on coverslips were pulse-labeled with 10 μM BrdU (BD Biosciences, 550891) for 30 min, fixed (4% paraformaldehyde, 30 min, room temperature), permeabilized, blocked and incubated with anti-BrdU (Thermo Fisher Scientific, B35128) (1:500) overnight at 4 °C. Cells were then incubated with Alexa Fluor 488 secondary antibody (Invitrogen, A27023; 1:1,000) for 1 h, followed by DAPI (1:100) for 10 min. Images were visualized by microscopy (Olympus IX51).

Statistics and reproducibility

Statistical analysis was performed using GraphPad Prism 8.0 software or Microsoft Excel. Data in the figures are presented as the mean ± s.d. unless otherwise stated. P values were determined using a two-tailed Student’s t-test or one-way analysis of variance (ANOVA), as stated in the figure legends. P < 0.05 was considered statistically significant between groups. P > 0.05 was considered not statistically significant between groups. Data were graphed using GraphPad Prism 8.0 software. All experiments were independently repeated three times with similar results, as stated in the figure legends. The experimental sample size is indicated in the text and figure legends. No data were excluded from the analysis; data distribution was assumed normal but this was not formally tested. Mice for the xenograft experiments were allocated randomly to each experimental group. In vitro experiments were not randomized and investigators were not blinded during data collection and analysis.

Extended Data

Extended Data Fig. 1 |. PURB specifically binds unacetylated C-terminal domain of p53.

(a) The 7 unique PURB peptides identified by LC-MS/MS are highlighted with 44.55% coverage of the whole PURB protein. (b) Diagram of biotin-conjugated C-terminal unmodified, acetylated, or methylated p53 peptides. (c) Western blot analysis of interaction between overexpressed p53 (wild-type, KQ or KL mutant) and endogenous PURB in H1299 cells. (d) Schematic diagram of the interaction between p53 and PURB revealed in (c). (e) Western blot analysis of the interaction between overexpressed p53 and PURB (wild-type or DE-A mutant) in H1299 cells. (f) Schematic diagram of the interaction between p53 and PURB revealed in (e). The experiments (c, e) were repeated three times with similar results and representative results are shown. Source data are provided in the Source data file.

Extended Data Fig. 2 |. PURB inhibits p53-dependent p21 induction and tumor suppression.

(a) Colony formation experiment using indicated HCT116 cells. (b) Bromodeoxyuridine (BrdU) staining of indicated HCT116 cells visualized by immunofluorescence and microscopy. Scale bars, 100 μm. (c) Xenograft tumors derived from indicated HCT116 cells. (d) Western blot analysis of protein levels in indicated A375 cells transfected with or without PURB expression plasmids. The experiments (b, d) were repeated three times with similar results and representative results are shown. Source data are provided in the Source data file.

Extended Data Fig. 3 |. PURB directly binds LncRNAs.

(a) Schematic diagram of the construct expressing SFB-tagged PURB. (b) Workflow of UV crosslinking and immunoprecipitation (CLIP) combined with next-generation sequencing. (c) RNA transcripts identified by CLIP-seq from H1299 cells expressing SFB-tagged PURB protein. (d) qPCR analysis of LncRNAs bound with PURB or p53 protein in CLIP assay in H1299 cells transfected with SFB-PURB or SFB-p53 (n = 3). Data are presented as mean ± SD of n = 3 independent biological repeats. p values were calculated using unpaired, two-tailed Student’s t test. Source data are provided in the Source data file.

Extended Data Fig. 4 |. PURB regulate p53-dependent transcription through HOTAIR.

(a) qPCR analysis of knockdown efficiency of different lncRNAs by antisense oligonucleotides in U2OS cells (n = 3). (b) qPCR analysis HOTAIR RNA levels upon depletion of HOTAIR by three sets of antisense oligonucleotides in U2OS cells (n = 3). (c) Western blot analysis of p53 and p21 protein levels in HCT116 cells transfected with antisense oligonucleotides targeting different LncRNAs. (d) qPCR analysis of p21 mRNA in HCT116 cells transfected with antisense oligonucleotides targeting different LncRNAs (n = 3). (e) Western blot analysis of p53, p21, PUMA, and MDM2 protein levels upon depletion of HOTAIR by three sets of antisense oligonucleotides in HCT116 cells. (f) qPCR analysis HOTAIR RNA levels upon depletion of HOTAIR by three sets of antisense oligonucleotides in HCT116 cells (n = 3). (g) qPCR analysis of p21 mRNA levels upon depletion of HOTAIR by three sets of antisense oligonucleotides in HCT116 cells (n = 3). (h) qPCR analysis of HOTAIR RNA associated with endogenous p53 purified by non-denaturing RIP in indicated U2OS cells (n = 3). (i) qPCR analysis of HOTAIR RNA associated with endogenous PURB purified by non-denaturing RIP in indicated U2OS cells (n = 3). (j) Western blot analysis of protein levels in A375 cells transfected with antisense oligonucleotides targeting HOTAIR. Data (a, b, d, f-i) are presented as mean ± SD of n = 3 independent biological repeats. p values were calculated using unpaired, two-tailed Student’s t test. The experiments (c, e, j) were repeated three times with similar results and representative results are shown. Source data are provided in the Source data file.

Extended Data Fig. 5 |. Sequence-specific binding of HOTAIR by PURB is critical for the function of PURB-HOTAIR complex.

(a) Sequencing validation of the deletion of 365–406 nucleotides of HOTAIR at genomic DNA level. (b) Xenograft tumors derived from indicated HCT116 cells. (c) Quantitation of tumor weight (n = 3) in (b). Data (c) are presented as mean ± SEM of n = 3 independent biological repeats. p values were calculated using unpaired, two-tailed Student’s t test. Source data are provided in the Source data file.

Extended Data Fig. 6 |. PURB-HOTAIR represses p53 transactivation through recruiting EZH2 and upregulating histone H3K27 tri-methylation.

(a) Western blot analysis of protein levels in A375 cells transfected with siRNA against EZH2 and/or siRNA against PURB. (b-e) ChIP analysis of p53/PURB/HOTAIR-mediated effects on EZH2 (b, d) and H3K27Me3 (c, e) occupancy at MDM2 (b, c) or PUMA (d, e) promoter in A375 cells (n = 3). (f-i) ChIP analysis of endogenous EZH2 (f, h) and H3K27Me3 (g, i) occupancy at MDM2 (f, g) or PUMA (h, i) promoter in sg-HOTAIR or sg-PURB A375 cells (n = 3). Data (b-i) are presented as mean ± SD of n = 3 independent biological repeats. p values were calculated using One-Way ANOVA. The experiment (a) was repeated three times with similar results and representative results are shown. Source data are provided in the Source data file.

Extended Data Fig. 7 |. Interacting dynamics between HOTAIR and EZH2 or PURB-p53 complex.

(a) In vitro binding of purified EZH2 protein and biotin labelled HOTAIR RNA (wild-type or mutant). (b) In vitro binding of purified PURB protein and biotin labelled HOTAIR RNA (wild-type or mutant). (c) In vitro binding of purified PURB or p53 protein and biotin labelled HOTAIR RNA (wild-type or mutant). (d) Schematic diagram showing the interaction between HOTAIR and EZH2 or PURB. The experiments (a-c) were repeated three times with similar results and representative results are shown. Source data are provided in the Source data file.

Extended Data Fig. 8 |. A unique cis-regulatory element at the p21 promoter is critical for the recruitment of PURB by p53 in a promoter-specific manner.

(a) ChIP analysis of PURB occupancy at MDM2 or PUMA promoter in wild type HCT116 cells (n = 3). (b) The eight-nucleotide PMS (PURB Mediated promoter-specific Sequence) was identified by MEME (motif-based sequence analysis tools) in the flanking region of p53 responsive element (p53 RE) on p21 promoter, but not on MDMS or PUMA promoter. (c) qPCR analysis of gene expression in U2OS cells with inducible PURB knockdown for 72 hours vs. control cells (n = 3). Data (a, c) are presented as mean ± SD of n = 3 independent biological repeats. p values were calculated using unpaired, two-tailed Student’s t test. Source data are provided in the Source data file.

Extended Data Fig. 9 |. HOTAIR is recruited to chromatin in a promoter-specific manner that mirrors the pattern of PURB.

(a) qPCR analysis of RNA retrieved by ChIRP using ASO-HOTAIR probes (n = 3). (b) Western blot analysis of proteins bound with HOTAIR RNA in the ChIRP assay. (c-d) qPCR analysis of recruitment of HOTAIR at the promoters of indicated genes (n = 3). Data (a, c, d) are presented as mean ± SD of n = 3 independent biological repeats. p values were calculated using unpaired, two-tailed Student’s t test. Source data are provided in the Source data file.

Extended Data Fig. 10 |. Model of promoter-specific, acetylation-regulated modulation of p53-dependent transcription by PURB-HOTAIR-EZH2 complex.

PURB directly binds p53 C-terminal tail and acts as a transcriptional co-repressor for p53 in a manner dependent on p53 acetylation status. PURB modulates p53-dependent activation of p21, GADD45A, etc., but has no effect on other major promoters such as PUMA and Mdm2. Interestingly, PURB is able to recognize a unique DNA element (PMS) at the p21 promoter; loss of this element does not affect p53-mediated transactivation but abrogates the ability of p53 o recruit PURB to the p21 promoter for repression. Mechanistically, the ability of PURB in transcriptional repression requires its sequence-specific binding with HOTAIR. HOTAIR interacts directly with EZH2 and, bridging by the PURB/HOTAIR complex, p53 is able to recruit EZH2 histone methyltransferase to the target promoters for transcriptional repression. Further analysis of the p53 targets has identified a number of promoters to serve as a target for PURB-binding and this mechanism of PURB-dependent promoter-specific regulation is widespread in nature.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41594-025-01597-3.

Data availability

Proteomics data were deposited to ProteomeXchange through the PRIDE database under accession number PXD042681. RNA-seq data from U2OS cells and CLIP-seq data from H1299 cells were deposited to the Gene Expression Omnibus (GEO) under accession numbers GSE295588 and GSE295618, respectively. The ‘expression analysis’ module of the GEPIA2 web server (http://gepia2.cancer-pku.cn/#analysis) was used to obtain differential expression levels of PURB between tumor and normal tissues. All data needed to evaluate the conclusions in this study are available in the main text and Supplementary Information. Source data are provided with this paper.

Code availability

The study did not develop new code. Standard procedures for mathematical and statistical analysis of data are described in the Methods.