Targeting codon 158 p53-mutant cancers via the induction of p53 acetylation

Gain of function (GOF) DNA binding domain (DBD) mutations of TP53 upregulate chromatin regulatory genes that promote genome-wide histone methylation and acetylation. Here, we therapeutically exploit the oncogenic GOF mechanisms of p53 codon 158 (Arg158) mutation, a DBD mutant found to be prevalent in lung carcinomas. Using high throughput compound screening and combination analyses, we uncover that acetylating mutp53R158G could render cancers susceptible to cisplatin-induced DNA stress. Acetylation of mutp53R158G alters DNA binding motifs and upregulates TRAIP, a RING domain-containing E3 ubiquitin ligase which dephosphorylates IĸB and impedes nuclear translocation of RelA (p65), thus repressing oncogenic nuclear factor kappa-B (NF-ĸB) signaling and inducing apoptosis. Given that this mechanism of cytotoxic vulnerability appears inapt in p53 wild-type (WT) or other hotspot GOF mutp53 cells, our work provides a therapeutic opportunity specific to Arg158-mutp53 tumors utilizing a regimen consisting of DNA-damaging agents and mutp53 acetylators, which is currently being pursued clinically. Codon 158 gain-of-function mutant p53 (158-mutp53) promotes tumourigenesis in lung cancer. Here, the authors show that 158-mutp53 render cancers sensitive to cisplatin and p53 acetylation agents through a mechanism where acetylated mutant p53 upregulates TRAIP and inhibits NF-ĸB signaling.

T P53 missense mutations are among the most common genetic lesions in tumors 1 , which often coincide with the earlier onset of oncogenesis than patients with p53 loss 2 . A single nucleotide substitution at the DNA-binding domain (DBD) renders the protein defective in DNA-binding, loss of tumor suppressive properties and concomitantly prevents the negative feedback regulation through MDM2 3,4 , leading to massive accumulation of full length mutant p53 (mutp53). Growing evidence from recent studies suggest that cells with prevalent mutp53 acquire additional oncogenic gain-of-function (GOF) based on their unique structural modifications [5][6][7][8] . Depletion of mutp53 or inhibition of its co-activator have demonstrated strong cytotoxicity in tumor cells 6,9,10 .
In this study, we uncover a mechanism of activating mutp53dependent apoptotic function in cancer cells through p53 R158G acetylation, and demonstrate that TRAIP regulation of NFĸB is the main molecular driver underpinning this observed sensitivity. We further show in a high-throughput screen that acetylation of p53 R158G can be achieved with several pharmacologic agents, providing a cogent basis for further clinical development.
Acetylation of mutp53 R158G induced pro-apoptotic phenotype. As HDAC inhibitors can acetylate histones and other cellular proteins, possible mechanisms that modulate mutp53 function include Lysine acetylation of histones affecting DNA configuration favoring mutp53 access to DNA-binding elements, or direct modification of mutp53 itself leading to alternative DNA binding. To evaluate this, we first demonstrated that a low, sub-cytotoxic concentration of belinostat (0.1 µM) was sufficient to acetylate both H3 and H4 (Fig. 3c) in H2170 cells, and was synergistic with cisplatin in apoptotic assays (Fig. 3d, f-h; Supplementary Fig. 3C, D). The transcriptional regulation of p53 targets (CDKN1A, PUMA, PMAIP1, BAX) and Bcl-2 family members (BCL2, BAD, BAK1) strongly implicated p53 transcriptional regulation in this observed synergy with DNA-damaging cisplatin (Fig. 3j, k). The dependence on mutp53 R158G , functioning through active transcription regulation, was further substantiated by the positive p53-reporter-luciferase activity of combination drug-treatment ( Supplementary Fig. 4A). In addition, DNA-binding affinity of mutp53 was validated with gel retardation assay, with wtp53 and mutp53 showing comparable binding affinity to a 31-bp p53 response element sequence from CDKN1A ( Supplementary  Fig. 4B, C). Consistently, depletion of p53 from H2170 parental cells with different short hairpin (shRNA) constructs or small interfering RNA (siRNA) reduced PARP and caspase 3 cleavage   Mutation at Arg 158 is a GOF p53 isoform. a-d Cell invasion assays were performed on isogenic Calu-1 cells (p53 −/− , p53 wt and mutp53 R158G ) and H2170 cells. Cells seeded in Matrigel invasion chambers were fixed and stained at the indicated time point. Representative images were shown for Calu-1 clones (n = 3 independent experiments) (a) and H2170 cells with Scr of TP53 siRNA knockdown (n = 2 independent experiments) (c). Immunoblots verifying p53 knockdown are shown on the right. The number of cells in four random microscopic fields ×4 were quantified for Calu-1 (b) and H2170 (d) cells. Data are expressed as mean ± SD. 50 nM of siRNA used per transfection. Two tailed Student's t-test; *P < 0.05. Scale bar, 200 µm. e, f Anchorageindependent colony forming assays were performed on isogenic Calu-1 cells (p53 −/− and five independent mutp53 R158G clones). Representative images of colony formation for each cell type, stained with MTT at assay endpoint (e). The number of colonies were quantified (f) and expressed as mean of triplicates ± SD (n = 3 independent experiments). Two tailed Student's t-test; *P < 0.05. g, h In vivo growth rates of p53 −/− , p53 wt , p53 R158G cells were measured upon day of implantation and monitored over time (g). Images shown are representative tumor sizes at assay endpoint (h). Tumor sizes are presented as mean ± SEM (n = 5 animals). Two-way ANOVA with Bonferroni correction; *P < 0.05; **P < 0.01; ***P < 0.001.  4D). Apoptosis induced by the combination treatment was significantly abrogated in the shp53 cells (Supplementary Fig. 4E). Logically, destabilization and depletion of GOF mutp53 explains potent anti-tumor cytotoxicity 9,10 , but paradoxically, we observed no reduction of mutp53 R158G in H2170 cells despite robust apoptosis (Fig. 3e). Rather, analyses of the post-translational modifications (PTM) of p53 showed that while cisplatin dose-dependently increased p53 phosphorylation and acetylation, combination with belinostat potentiated only its acetylation (Fig. 3e).
To establish the role of p53 R158G in promoting apoptosis, we conducted synonymous experiments using isogenic Calu-1 celllines. Cisplatin alone and in combination with belinostat triggered comparable S-phase arrest regardless of the p53 status, suggesting that the synergy is independent of its cytostatic effect (Figs. 3i, 4c). Cisplatin alone induced pronounced apoptotic response in both p53 wt and p53 R158G cells, however, enhanced cytotoxicity and p53 acetylation by belinostat were restricted to p53 R158G cells (Fig. 4a, b, d), which is concordant with the IC 50 values (Fig. 2d). Transcriptional activation of MDM2 was only observed in wtp53 but not mutp53 or null cells (Fig. 4e); whereas, CDKN1A, PUMA, and BAX were induced post-treatment in mutp53 but not null cells (Fig. 4f-i), implicating p53 R158G in the synergistic pro-apoptotic effect.
To investigate the role of p53 acetylation in mediating the proapoptotic function of mutp53, we first constructed p53 R158G mutants by substituting the Lysine residues at the DBD (p53 R158G(DBD, K-A) ), C-terminal domain (p53 R158G(CT, K-A) ) or the whole protein (p53 R158G(K20A) ). Expectedly, these modifications were unable to overcome cell death induced by single agent cisplatin, which could occur via the p53-independent extrinsic apoptotic pathway 35 . Nonetheless, belinostat-induced synergy was abrogated in p53 R158G(DBD, K-A) and p53 R158G(K20A) cells ( Fig. 4a, b, d), both cell types with impaired Lysine acetylation at the p53 DBD. In contrast, the synergistic effect was maintained in p53 R158G (CT,K-A) clones (Fig. 4b), suggesting that C-terminal acetylation is not involved in mutp53-effected cytotoxicity.
To compare the impact of acetylation and phosphorylation on p53 R158G -dependent apoptosis, we conducted high-content immunofluorescence imaging to detail the post-translational events in individual nuclei (Fig. 4j), and two key observations were made. Firstly, single treatment of belinostat and cisplatin increased the acetylated-p53 population (Q4), with the combination treatment further augmented cells in Q4 ( Supplementary  Fig. 5A, C). Secondly, cisplatin but not belinostat treatment strongly increased the phosphorylated-p53-labelled cell population (Q3), and combination treatment had no effect on cells in Q3 ( Supplementary Fig. 5B, D). This data demonstrated the positive correlation of acetylated p53 R158G with combination treatment, which significantly triggered cellular apoptosis. To further validate the importance of p53 R158G acetylation, isogenic p53 cell-lines were treated with tenovin-6, a potent sirtuin inhibitor and known acetylator of p53 36 . While p53-expressing cells were more sensitive to tenovin-6 ( Fig. 4k), acetylation of p53 R158G increased cleavage   of PARP and caspase 3 (Fig. 4l). Collectively, acetylation of the DBD is crucial to effecting apoptosis by mutp53 R158G .
Mutp53 R158G binds to distinct DNA sequence motifs. We postulated that mutp53 R158G binds to and activates a distinct spectrum of chromatin regulatory genes from wtp53, which may be significantly altered by drug-induced acetylation. Firstly, stateof-the-art molecular modelling and simulation studies were performed to compare the structural dynamics of wild-type p53, mutp53 R158G and acetylated-mutp53 R158G using monomers of the respective DBD. The loss of the R158 sidechain creates a large impaired in the acetylated R158G mutant, it is compensated for by the interactions between acetylated K201 and the minor groove ( Supplementary Fig. 6D). The most intriguing alteration in the acetylated R158G mutant is an increase in the length of the dimerization helix H1 from R181 to D186-about 1.5 extra turns of a helix-that will promote the formation of a dimer that is more stable than the WT and is also associated with the introduction of two salt bridges across the dimer interface (E180 of each monomer forms a salt bridge with R181 of the second monomer; Supplementary Fig. 6D). We speculate that this may increase DNA binding and result in the enhanced transcription activity seen in the acetylated mutp53 R158G . The simulations also highlighted the formation of key stabilizing interactions between the polar groups introduced by the acetylated K101 and K164 that stabilize the overall conformation of the mutp53. This supports our earlier observations that conversions of these two Lysines to Alanines in the DBD (p53 R158G(DBD,K-A) ) resulted in attenuation of the proapoptotic activity seen in acetylated R158G ( Fig. 4a, b, d).
Next, we compared the DNA-binding sequences and transcriptional targets of vehicle-and drug-treated p53 wt and p53 R158G cells through p53 chromatin immunoprecipitation (ChIP) followed by sequencing (ChIP-Seq). In untreated cells, both wtp53 and mutp53 bound significantly to the proximal regions (within 1 kb) of the transcription start sites (TSS) (Fig. 5a, Supplementary Fig. 7A). Drug treatment led to marked increase in genomic binding in both p53 R158G and p53 wt cells, with higher signal intensity around the TSS-proximal regions (Fig. 5a, Supplementary Fig. 7A), suggestive of DNA-binding capability of both p53 isoforms. Importantly, predicted motif analysis for TSS-proximal peaks suggested that p53 consensus motifs were enriched not just in p53 wt but also in p53 R158G (motif similarity P = 0.029 and 0.039, TOMTOM match statistic; P = 1 × 10 −16 and 1 × 10 −265 , HOMER statistic) (Fig. 5a, Supplementary  Fig. 7A). However, by comparing the binding patterns of wtp53 and mutp53 to TSS, we found that the mutp53 R158G peaks were highly dissimilar from the peaks of wtp53 ( Supplementary Fig. 8, column 1 row 3). In addition, we aligned the ChIP-seq data from the cells treated with the belinostat/cisplatin combination, and showed that TSS-proximal peaks post-treatment were distinct from the vehicle control in both wtp53 and mutp53. Interestingly, peaks enriched in drug-treated mutp53 R158G cells resembled that of drug-treated wtp53 ( Supplementary Fig. 8, column 2 row 4). These data collectively suggest that mutp53 R158G partially retains wtp53 activity when cells are exposed to stress signal, which supports the induction of wtp53 response genes by the druginduced transactivation of p53 R158G in H2170 cells (Supplementary Fig. 4A), as well as the increased DNA-binding affinity of mutp53 upon acetylation ( Supplementary Fig. 4B, C).
To identify the genes activated by acetylated p53 R158G which are associated with the induction of apoptosis, we compared the treatment-induced perturbations of transcriptomic profiles (by Ampliseq RNA sequencing) between wild-type and mutant cells. These gene panels were subjected to KEGG ontogeny enrichment analysis. Expectedly, genes mediating p53 signaling were elevated in p53 wt cells after treatment (P-value= 7.87 × 10 −11 ); in p53 R158G , genes associated with cell cycle, DNA replication and DNA damage repair pathways were prioritized ( Supplementary  Fig. 7B). We further analyzed gene sets obtained from both ChIPseq and AmpliSeq for concordance of significantly perturbed genes pre-and post-treatment (log 2 > 1.5), also found to be bound by p53 within the proximal regions (<10 kb from TSS), as direct p53 targets (Fig. 5b, Supplementary Table 4). Selected genes were validated with qPCR, and consistently, direct p53 wt binding increased expression of canonical p53 downstream genes (MDM2, GADD45A, PMAIP1) (Supplementary Fig. 7C-E, I-K). Among the mutp53-bound target genes, TRAIP and RAD51 were markedly induced (log 2 -fold = 2.72 and 2.767, respectively) after drug treatment relative to wild-type cells (Supplementary Table 4, Fig. 5c, d, Supplementary Fig. 7F, L). Work on other p53 hotspot mutants have described part of the mechanistic basis for mutp53's GOF as direct interaction with E26 transformation-specific (ETS) motifs 37 . Two chromatin regulatory genes downstream of ETS2, KMT2D, and KAT6A 22 , reported to be positively associated with hotspot GOF mutp53 (R175H, R248Q, R248W, R249S, or R273H), were neither transactivated nor upregulated in p53 R158G (Fig. 5f, g, Supplementary Fig. 7G, H), thereby distinguishing DNA binding by Arg 158 mutants from other mutated p53 GOF isoforms.
Acetylated p53 R158G suppresses NFĸB signaling through TRAIP. To prove this hypothesis, we first determined the changes in chromatin binding affinity of acetylated mutp53 in the context of TRAIP transactivation. Using ChIP-quantitative PCR (ChIP-qPCR), we validated the increased binding of p53 R158G after belinostat/cisplatin treatment to the promoter region of the TRAIP gene ( Supplementary Fig. 11A-C). In parallel experiments involving p53 wt and p53 R158G(K20A) cells, there was no apparent enrichment in term of wtp53 binding to TRAIP, and more importantly, the ChIP-qPCR signal of mutp53 binding was attenuated in the p53 R158G(K20A) variant, which cannot be acetylated ( Supplementary Fig. 11B, C). Collectively, these findings strengthened the concept that acetylation of p53 R158G leads to the transactivation of TRAIP (Fig. 5d).
We next sought to investigate the relevance of TRAIP transactivation on NFĸB signaling in the context of differential p53 alterations. NFĸB stimulation renders resistance to programmed cell death through nuclear translocation of its activated heterodimeric complex, DNA interaction and induction of  survival signals, which is regulated by an auto feedback loop involving its negative regulator IĸB 47 . Activation of NFĸB is determined either by the canonical NFĸB1/RelA complex or the non-canonical NFĸB2/RelB complex. While wtp53 and NFĸB are often coordinated in an opposing manner due to their contrasting cellular functions 48,49 , mutp53 proteins have been shown to promote NFĸB activation 11 . In the context of LUSC, we first showed using DNA-binding ELISAs that cisplatin treatment increased p65 (RelA, NFĸB p65 subunit) transactivation in p53 wt cells (Fig. 6a). In contrast, a reciprocal downregulation was observed in p53 R158G cells, with a further reduction observed when combined with belinostat; whereas NFĸB was unaffected in p53 deficient cells (Fig. 6a). To rule out the possible alternative involvement of non-canonical NFĸB signaling, we investigated  the transactivation of p52 (NFĸB2 subunit) and showed that DNA-binding was marginally affected in all cell types (Fig. 6b) 50 . Cytoplasm to nuclear shuttling of p65 is critical for the initiation of the NFĸB pathway 51 . In p53 R158G cells, drug treatment reduced nuclear accumulation of p65; conversely, pronounced p65 nuclear translocation was observed in wild-type cells, in concordance with the increase in DNA-binding assays (Fig. 6c-f). The wtp53-p65 relationship was illustrated experimentally in colon cancer cells HCT116 p53 wild-type/null pairs 52 , where drug combination induced p65 nuclear translocation in wild-type (HCT116 +/+ ) but not in null (HCT116 −/− ) cells ( Supplementary  Fig. 12A-D). In addition, we observed that basal nuclear NFĸB is kept at levels lower in p53 wt than in p53 R158G or p53 null cells (Fig. 6c-f, Supplementary Fig. 12A, B), likely through higher TRAIP expression in wtp53 cells (Fig. 5e). Together, our findings support the contention that increased transcription of TRAIP by acetylated p53 R158G leads to inhibition of NFĸB through blocking IĸB phosphorylation and nuclear import of p65.
To elucidate the mechanisms relating TRAIP to NFĸB inhibition, we overexpressed TRAIP in both p53 −/− and p53 wt cells. Surprisingly, cisplatin alone or in combination downregulated exogenous TRAIP in p53 −/− cells without affecting the downstream targets (Fig. 6g, left). In contrast, ectopic TRAIP expression was sustained in p53 wt cells, leading to the downregulation of TRAF2, c-IAP1, and c-IAP2, pro-survival components of the TNF receptor family 53 , potentially through the ubiquitin-proteasomal degradation pathway (Fig. 6g, right). These data are consistent with previous work demonstrating that TRAF2 and c-IAP are downstream effectors of TRAIP 42,45 . As previously described 54 , ablation of TRAF2 reduced phosphorylation and degradation of IĸB (Fig. 6g, right), therefore facilitating IĸB-mediated impediment of NFĸB transactivation (Fig. 6h). This negative NFĸB regulation by exogenous TRAIP sensitized p53expressing cells to drug-induced apoptosis, as indicated by increased Annexin V positive population and PARP cleavage (Fig. 6i-j). Importantly, exogenous TRAIP recapitulated p53 R158G response to drug treatment in p53 wt cells. Consistently, acetylation of p53 R158G is associated with p21 upregulation, p-IĸB suppression and elevated TRAIP transcript in H2170 cells ( Supplementary Fig. 13A-C). Of note, these modulations were consistent in other cell types expressing Arg 158 mutp53 (H441, H661), but were not found in other GOF mutp53 cells (H1417, H1975, HCC70, MDA-MD-468, and SK-BR-3 cells) (Supplementary Fig. 13A, B), thereby highlighting the mechanistic distinctiveness of the Arg 158 mutp53. Effective silencing of TRAIP markedly rescued cellular apoptosis in H2170 cells, as demonstrated by the reduced PARP and caspase 3 cleavage, by sustaining p-IĸB without affecting p53 acetylation (Supplementary Fig. 13D). It is interesting to note that TRAIP expression was consistently detected in the DMSO-treated cells upon siRNA interference, and we reasoned that the presence of a compensatory mechanism could potentially stabilize TRAIP and counterbalance the knockdown effect. We postulate that this mechanism could be impaired under cellular stress, thus allowing knockdown of TRAIP to be possible upon drug treatment. However, the detailed mechanism remains elusive at this moment.
To further investigate the mechanisms of the TRAIP/TRAF2/ NFĸB axis, dominant negative IĸBα mutants (IĸBα-ΔN, IĸBα-S32A) were introduced to shut off NFĸB activity. As expected, dephosphorylation of IĸB in wtp53 cells reduced p-p65, p-TRAF2, and c-IAP2 upon drug treatment, recapitulating the effects of TRAIP overexpression (Supplementary Fig. 16A). Surprisingly, transfection of IĸBα mutants has little effect in p53 −/− cells, and conversely, p65 was highly phosphorylated. Functionally, NFĸB inhibition also induced pronounced apoptosis in wtp53 and mutp53 cells, but to a lesser extent in null cells ( Supplementary Fig. 16B). Collectively, these data strongly suggest a direct role of TRAIP, expressed upon acetylation of mutp53 R158G , in facilitating apoptosis through disruption of NFĸB pathway signaling.
Potent p53 acetylators reduce mutp53 tumor growth. Further analyses of compounds found active against p53 R158G cells revealed that eight out of the 10 compounds induced p53 R158G acetylation ( Supplementary Fig. 15A), which was correlated to increased p21, elevated CDKN1A and TRAIP transcripts, suppression of TRAF2 and p-IĸB, as well as augmented PARP cleavage ( Supplementary Fig. 15A, D, E), thus supporting our earlier observations. On the contrary, exposure to these acetylators induced stabilization of wtp53 and upregulated p21, increased MDM2 and CDKN1A transcripts, but had minimal impact on p-IĸB and TRAIP mRNA ( Supplementary Fig. 15B-E), thereby highlighting the specificity of these single agents in targeting mutp53 R158G cells.
When applied in vivo, single agent cisplatin was effective in reducing tumor growth of p53 R158G xenograft as compared to vehicle, with no efficacy observed in p53 −/− and p53 wt tumors ( Supplementary Fig. 16A-C). Of note, efficacy of cisplatin in the p53 R158G(K20A) tumors with defective acetylating mechanism was evidently reduced (Supplementary Fig. 16A-C). Mechanistically, acetylation of mutp53 R158G was accompanied by elevated TRAIP (Supplementary Fig. 16E), p65 inhibition ( Supplementary Fig. 16F) and reduced Ki67 staining, a marker for cellular proliferation (Supplementary Fig. 16G); in contrast with increased p65 nuclear accumulation and lack of efficacy in the wtp53 tumors. Scoring of TRAIP transcripts demonstrated a reciprocal inverse correlation with NFĸB activity, for instance, a high basal score was associated Fig. 6 Drug-induced TRAIP expression negatively regulates NFĸB signaling in p53 R158G cells. a, b Quantification of active nuclear NFĸB1 (p65) (a) and NFĸB2 (p52) (b) in Calu-1 (p53 −/− , p53 wt , p53 R158G ) cells 48 h after vehicle or drug treatment. Data are presented as mean ± SD (n = 3 independent experiments). c, d Distribution of p65 (in cytoplasm or nucleus) was determined by Western blot (c) after nuclear-cytosolic fractionation. TATA-box binding protein (TBP) (nuclear) and α-tubulin (cytoplasmic) were used as loading controls. Nuclear p65 expression was quantified with densitometry after normalizing to TBP (d). Data are presented as mean ± SD (n = 3 independent experiments). Two tailed Student's t-test; **P < 0.01. e, f Immunofluorescence staining was performed to determine the localization of p65 (Alexa Fluor-488) in p53 wt and p53 R158G cells 48 h post-treatment. At least five independent fields were taken for each condition with a minimum of 30 nuclei per group (n = 3 independent experiments). Representative confocal images shown at 63× magnification (e). Scale bar, 10 µm. Merged images displayed with blue indicates DAPI, green indicates p65. Integrated density quantifying nuclear NFĸB (p65) in p53 wt and p53 R158G cells (f). Data are presented as mean ± SD (n = 3 independent experiments). Two tailed Student's t-test; *P < 0.05. g Western blot showing effects of TRAIP overexpression on the indicated proteins in Calu-1 isogenic cells (p53 −/− , p53 wt , p53 R158G ) 48 h after vehicle or drug treatment. β-actin shown as loading control (n = 3 independent experiments). h Quantification of active nuclear NFĸB1 (p65) in Calu-1 (p53 −/− , p53 wt , p53 R158G ) cells after TRAIP overexpression with the indicated drug treatment. Data are presented as mean ± SD (n = 3 independent experiments). i, j Annexin V staining (i) and Western blotting (j) in TRAIP overexpressed cells 48 h post-treatment. Data are presented as mean ± SD (n = 3 independent experiment). Two tailed Student's t-test; *P < 0.05, **P < 0.01.
with low p65 accumulation in p53 wt tumors ( Supplementary  Fig. 16H). Further analyses of vehicle-and cisplatin-treated xenograft for drug-activity showed elevated TRAIP mRNA copies in p53 R158G tumors, but decrease in wild-type tumors, validating the in vitro data (Supplementary Fig. 16H). We expanded our in vivo studies to screen for efficacy of other potent acetylating agents, as the clinical application of HDACi has been challenging in solid cancers, possibly due to their poor therapeutic index 55 . Whereas volasertib demonstrated high toxicity in mice (died within days of administration), tolerable doses of JQ1 and topotecan were effective in suppressing growth of p53 R158G tumors compared to the p53 wt and p53 −/− counterparts ( Supplementary Fig. 16I-K), providing direction for future development of p53-acetylating agents in p53 R158G -positive tumors.
Patient-derived xenograft (PDX) tumors of gastric cancer (GC; p53 R158C ) and hepatocellular carcinoma (HCC; p53 R158H ) harboring Arg 158 mutp53 were selected for evaluation with p53-acetylators: cisplatin, JQ1 and topotecan. All three compounds significantly reduced tumor growth in both PDX models (Fig. 7a, b, d, e) at tolerable doses (Fig. 7c, f). Tumor specimens from GC-PDX showed that cisplatin and topotecan significantly induced p53 acetylation and phosphorylation, with TRAIP upregulation as well as PARP cleavage observed in most of the treated tumors (Fig. 7g). Moreover, we demonstrated using RNA in situ hybridization the transcriptional induction of TRAIP mRNA in the drug-treated tumors (Fig. 7h, i), which was concordant with the protein expression in the matched tumor samples (Fig. 7g). These findings suggest that, while the Arg 158 p53 mutation is more prevalent in lung carcinomas, the mechanism conferring tumor cytotoxicity could be applied across cancer types harboring the same alteration. We further explored the potential of combining a DNA-damaging agent (cisplatin) with an effective p53-acetylating agent (topotecan) in both PDX models. Compared with single agents, combination of cisplatin/ topotecan rapidly and profoundly suppressed tumor growth with no significant weight loss observed in all animals (Fig. 7j), with tumor regression observed in multiple cases (Fig. 7k).

Discussion
Missense mutations in the DNA-binding domain constitute more than 70% of tumor-associated p53 mutations. These mutations alter DNA-binding capability of mutp53, concomitantly leading to a variable loss of p53 tumor suppressive functions, while mediating oncogenic GOF through transcriptional aberrations involving chromatin remodeling and interaction with transcription cofactors such as SREBP, ETS2, or NRF2 6,22,37,56 . Therapeutic interventions that have been proposed to target mutp53 cells include attempts to exploit the loss of checkpoint function 27,28 , inhibit the GOF pathways 6,29 , and direct destabilization of the mutated protein; 9,10 but these have yet to gain clinical traction. Whether these strategies can be applied globally to all mutp53 remains debatable, as only the six most common "hotspot" mutants have been studied in depth. We studied a GOF Arg 158 mutp53 in the DBD found to be relatively frequent in lung carcinomas, and elucidated a unique pathway of mutp53 that can be exploited therapeutically.
The complex crosstalk between p53 and NFĸB has been described previously, with the suppression of p65 in wild-type cells a result of competitive interaction with p300 or glucocorticoid receptor; 48,49 conversely mutp53 prolongs and enhances NFĸB activation 11,12,57,58 . We propose here a mechanism for this divergent effect through acetylation of Arg 158 mutp53, which alters its DNA-binding spectrum and upregulates TRAIP as key target gene, leading to NFĸB suppression through TRAF2 degradation and culminating in cell death (Fig. 8). This p53 isoform appears to carry distinctive GOF mechanisms that does not involve the chromatin regulatory genes as observed in other DBD hotspot mutants; other hotspot p53 mutants, such as R175H and R273H, did not upregulate TRAIP in response to belinostat and cisplatin treatment. Given that a point mutation is sufficient to affect p53 transactivation 59 , and that protein misfolding regulated by zinc-binding could affect p53 transcriptional functions 60 , it is possible that Arg 158 mutp53 modulates sitespecific DNA binding through aberrant Zinc2+ interaction which should be explored further. One possible caveat of the study is the lack of a universal cellular system to distinguish the functions of the various GOF mutp53. The inability to establish a robust CRISPR-Cas9 model has prevented us from studying this.
Our expression analysis and ChIP-Seq findings concordantly demonstrated the retention of transcriptional activity of mutp53 R158G , which has been described in other DBD variants. Exposure to cellular stress, such as cisplatin, could trigger a partial wild-type activity in mutp53 R158G that induces p53-dependent pro-apoptotic signals, which is not seen in p53 null cells. However, the DNA-binding activity of mutp53 R158G is distinct from p53 wt , manifesting in GOF activity. Paradoxically, this DNAbinding ability is prerequisite for the induction of apoptosis as initiated through mutp53 R158G acetylation, a key posttranslational event closely associated with its transcriptional activity and stability 61,62 . Loss of acetylation, particularly at the DBD, significantly impacted cell cytotoxicity, indicating that this post-translational event is indispensable for mutp53 activation. We studied this using in silico modeling of the expected molecular disruptions effected by the acetylation of DBD, and our simulation suggested strongly that the distorted mutp53 conformation could be restored through acetylation within the DBD, which appears to act as a stabilizing factor to facilitate the formation of a functional and stable mutp53 dimers with strong DNA-binding affinity. These further refined the commonly held credence that targeting acetylation and eventual depletion of mutp53 is a plausible anti-tumor approach 63 . Such findings are made therapeutically relevant by positive screen hits for compounds with pharmacological activity of p53 R158G acetylation. The selective cytotoxicity of these compounds in Arg 158 mutp53 tumors, compared to wild-type and null tumors, is promising as a companion biomarker for selecting patients with greater potential to respond to the p53-acetylating agents. From the therapeutic standpoint, dosages of acetylating agents used in the xenograft experiments were well-tolerated, further heightening the possibility of clinical translation. For instance, the bromodomain (BRD) inhibitor JQ1 is currently in clinical development to target a rare genomic BRD-NUT fusion event in NUT midline carcinoma 64 , and our data extends potential evaluation in Arg 158 mutp53 tumors. Curated data from public databases (IARC TP53, COSMICs, ICGC, and cBioportal) reveal the presence of Arg 158 mutp53 in multiple carcinomas; and the effectiveness of mutp53 acetylators in the PDX models of gastric and liver cancers further extends the applicability of these findings to other cancer types.
In conclusion, we provide evidence of a unique mechanism of p53 activation that is specific to Arg 158 mutations, exposing a previously unrecognized therapeutic vulnerability and facilitating a biomarker-implemented approach directed against this GOF mutant. This is unsurprising, as it has been recently recognized that different p53 mutations are attributable to distinctive phenotypes and cellular effects 65,66 , and our work adds to this. If validated in the clinic, this represents an advance in therapeutic treatment of tumors with mutated p53, particularly that of LUSC, in which Arg 158 mutant is found to be prevalent.

Methods
Genomic analyses of human cancers. The pan-cancer mutation annotation file were downloaded from GDAC (version 2016_01_28 25 ). Mutations found on TP53 genes and the corresponding protein change were extracted for analysis.
Transwell invasion assay. Matrigel invasion assay was carried out with Corning® BioCoat™ Matrigel Invasion Chambers (#354480). In brief, 25,000 (for Calu-1) or 50,000 (H2170) cells were plated in the transwell chamber with serum-free medium. For gene silencing experiments, cells were incubated with siRNA overnight k ARTICLE prior to plating. Tumor cells were allowed to invade for 24-36 hours before the transwell inserts were fixed with 4% paraformaldehyde, stained with gentian violet and imaged for analysis. Number of invaded cells was quantified with ImageJ Software (NIH), and expressed as mean ± SD.
Anchorage-independent soft agar assay. The colony forming assay comprises three layers of soft agar/medium matrix in 24-well plate: a bottom layer of 0.6% agar matrix; a middle layer containing 0.36% agar matrix with the indicated cells (10,000 cells/well); and a top layer comprising complete medium. Culture medium were changed weekly. For gene silencing experiments, cells were incubated with siRNA overnight prior to culturing in soft agar. After 4 weeks of culture, colonies were stained with MTT stain (Promega, #G4000) for 2 h and imaged. Colonies were counted with ImageJ Software, and expressed as mean ± SD.
PCR and sequencing. Total RNA was isolated from cells using RNeasy kit (Qiagen, #74106) and reverse-transcribed to cDNA using iScript RT Supermix Fig. 7 Cisplatin, JQ1 and topotecan reduce tumor growth in PDX models harboring Arg 158 p53 mutations. a-f Efficacy of cisplatin, JQ1 and topotecan were investigated in PDX model of gastric cancer (GC) and hepatocellular (HCC) cancer. Growth curve analysis of GC-PDX (a) and HCC-PDX (d) treated with vehicle, cisplatin (CDDP; 4 mg kg −1 ), JQ1 (30 mg kg −1 ), or topotecan (1 mg kg −1 ). Tumor sizes are presented as mean ± SEM. Two-way ANOVA with Bonferroni correction; *P < 0.05; **P < 0.01; ***P < 0.001. Waterfall plots showing change in tumor volume (relative to initial tumor volume at treatmentstart day) for each individual GC-PDX (b) and HCC-PDX (e) tumor in respective treatment group. Bodyweight of mice with GC-PDX (c) and HCC-PDX (f) at assay endpoint was tabulated as mean ± SD. g Western blots demonstrating changes of the indicated proteins in tumors of respective treatment (n = 3 independent tumors). β-actin shown as loading control. Densitometric quantification of TRAIP expression was tabulated on the right. Relative fold change is normalized to β-actin, relative to vehicle control tumors. Data presented as mean ± SD. Two tailed Student's t-test; *P < 0.05, **P < 0.01. h-i RNA in situ hybridization (RNAscope) showing TRAIP expression in respective tumors. Representative images showing TRAIP mRNA signal are shown at ×20 magnification (h). Scale bar, 50 µm. Semi-quantitative scoring (0-4) of TRAIP mRNA signal (dots/cell) in p53 wt and p53 R158G , respectively, was tabulated (i). Data are represented as scattered dot plot ± SD (n = 5 independent tumors). Two tailed Student's t-test, *P < 0.05. j-k Efficacy of cisplatin and topotecan combination were investigated in PDX models of gastric cancer and hepatocellular cancer. j Growth curve analyses of GC-PDX and HCC-PDX treated with vehicle, cisplatin (CDDP; 4 mg kg −1 ), topotecan (1 mg kg −1 ) or CDDP/topotecan combination. Tumor sizes are presented as mean ± SEM. Two-way ANOVA with Bonferroni correction; *P < 0.05; **P < 0.01; ***P < 0.001. k Change in tumor volume (relative to initial tumor volume at treatmentstart day) for each individual tumor in respective treatment group were calculated and tabulated into four groups (>500%; >200% <500%; >100% <200%; <100%). Tumor volume below 100% indicates tumor regression.  Fig. 8 Proposed mechanism of Arg 158 mutp53 in response to treatment with acetylating agents. Tumor cells harboring mutp53 expressed inherently high level of NFĸB activity to confer cellular survival. We showed that in mutp53 R158G cells, NFĸB/p65 signaling could be suppressed through negative regulation of the TRAF2/c-IAP/IĸB-axis via the upregulation of TRAIP, which is induced transcriptionally by monotherapy or combination treatment of compounds shown to induce mutp53 acetylation. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15608-y ARTICLE NATURE COMMUNICATIONS | (2020) 11:2086 | https://doi.org/10.1038/s41467-020-15608-y | www.nature.com/naturecommunications (Bio-Rad, #1708841). Sanger sequencing were carried out on amplified DNA to sequence region of interest according to standard protocol. Real-time qPCR was performed in triplicates using standard procedures on a 7500HT Fast-Real-Time PCR platform (ABI), using probe-based assays. Gene expressions were normalized to the housekeeping gene GAPDH. ChIP was performed using Active Motif ChIP-IT ChIP-seq kit according to manufacturer's protocol with slight modifications. In brief, cells were crosslinked with 1% formaldehyde, glycine quenched, lysed with cell lysis buffer (Sigma, #2978) supplemented with 1 mM PMSF, and subjected to pulse sonication (20 s with 30 s internal, five cycles). The supernatant was then immunoprecipitated with the p53 and rabbit IgG antibodies (Cell Signaling, #9282 and #2729) at 4°C overnight and incubated with Magna ChIP Protein G magnetic beads (Millipore, #16-662). The beads were washed before DNA was reversecrosslinked and purified. For ChIP-Seq, sequencing libraries were prepared using TruSeq DNA LT Sample Prep kits, and pair-end sequencing was performed on Illumina Hi-Seq platform. ChIP-qPCR validation was performed in triplicates using standard procedures on a 7500HT Fast-Real-Time PCR platform (ABI), using primer-based assays. For Ion AmpliSeq Transcriptome Human Gene Expression (Thermo Fisher), 10 ng RNA was used for barcoded cDNA library preparation using PCR-based targeted-amplicon amplification according to manufacturer's protocol. Pooled libraries were amplified with emulsion PCR on Ion Torrent OneTouch2 instruments (OT2) and sequenced on Ion Torrent Proton sequencing system. Both Hi-Seq and AmpliSeq were performed at AITbiotech (Singapore), Sanger sequencing at Axil Scientific (Singapore). All primer sequences used for RT-qPCR and ChIP-qPCR in this study are tabulated in Supplementary  Table 5.
Drug screening bioinformatics, Bliss Independent index. Drug response of the Calu-1 cell-lines were quantified using the Inhibition score (I-score). For each compound c, the I-score was determined as follows: where the denominator is the median of the luminescence emitted in presence of Dimethy sulfoxide (DMSO). The inhibition heatmap of selected drugs (filtered at I-score >50 in at least one of the cell-lines) between the Calu-1 wild-type, mutant and null cell-lines were hierarchically clustered using complete linkage clustering and Euclidean distance measure. Combination Index (CI) was calculated based on the Bliss Independence model, by considering all dose combinations tested (observed combination response against the predicted combination response). By assuming that there is not drug-drug interaction, the combination effect is declared synergistic if the observed response is greater than the predicted response (Bliss < 0).
Caspase activity, apoptosis and cell cycle assays. Caspase 3/7 enzyme activities were measured by using Caspase-Glo 3/7 Assay Systems (Promega, #G8090) according to manufacturer's protocol. Cells were incubated with indicated compounds for 24 h, and the luciferase activities were measured after substrate incubation with a TECAN Infinite plate reader. Apoptosis assay was carried out using annexin V-APC (Thermo Fisher, #A35110) and propidium iodide (PI, BD Biosciences, #556463) according to manufacturer's protocol. In brief, harvested cells (0.2 × 10 6 ) were washed with PBS followed by annexin V buffer (Thermo Fisher, #V13246), and stained with Annexin V/PI mixture for 15 min at 25°C. Cell cycle fractions were determined through PI nuclear staining. In brief, harvested cells (0.2 × 10 6 ) were fixed with 70% cold ethanol, and stained with PI staining buffer (0.03 mg ml −1 PI, 0.1 mg ml −1 RNAse A, 0.1% Triton-X in PBS) for 30 min at 25°C. The staining profile was acquired (≥10,000 cell events) using LSR II Flow Cytometer (BD Biosciences). The acquisition was analyzed using the FlowJo 8.1.1 software (Tree Star).
Reporter-luciferase assay. The transcriptional activity of p53 was assessed by reporter-luciferase assay. Cells were co-transfected with a combination of Cignal reporter plasmids (Qiagen, # CCS-004L), including a mixture of inducible p53responsive firefly luciferase construct or non-inducible firefly luciferase construct, with constitutively expressing Renilla luciferase construct. The cells were then drug treated for the indicated durations, and luciferase assay was performed with Dual Luciferase Reporter Assay System (Promega, #E1910) according to manufacturer's instructions and detected by Lumat LB 9507 (Berthold Technologies). The ratio between Renilla luciferase across samples was used to normalize the transfection efficiency.
In vitro translation (IVT), acetylation, and EMSA. IVT was performed according to manufacturer's protocol (Thermo Scientific, #88881). Briefly, 1 µg each of pcDNA3.1-p53wt and pcDNA3.1-p53R158G plasmids were incubated with the Hela extract and reaction mix for 6 h at 30°C. The in vitro protein acetylation (IVA) assay was performed, with a standard reaction included half of the sample from the IVT assay, 0.5 µg of histone acetyltransferase domain of p300/CBP protein (Upstate, #14-418), 4 µg of Acetyl-CoA (Sigma, #A2056) and diluted in assay buffer (final concentration of 50 mM Tris-HCl, pH 8.0, 10% glycerol, 0.1 mM EDTA, 1 mM dithiothreitol). The reaction mix was incubated for 6 h at 30°C in a shaking incubator. The expression and acetylation of the respective p53 proteins were validated with immunoblotting.
Electrophoretic mobility shift assay (EMSA) was conducted according to manufacturer's protocol (Thermo Scientific, E33075). 31-mer oligo containing p53 response element of CDKN1A gene (5'-CAGCATGCTCCAGGTAGAAGGAAA CAGGCCC-3') was ordered from IDT. Briefly, 50 ng of the synthesized oligos was incubated with 2 μl of the reaction containing the respective p53 isoforms, and incubated for 2 h at 30°C. The samples were then separated by PAGE using 6% non-denaturing gel. DNA and protein were stained simultaneously with SYBR ® Green nucleic acid gel stain and SYPRO ® Ruby protein gel stain, respectively. The images were then visualized and captured using UV transilluminator.
Molecular dynamics simulations of p53. The dimeric structure of the p53 DNA-Binding Domain (DBD) complexed to a small fragment of DNA (PDB ID: 2AHI) was taken from the RCSB Protein Data Bank. Molecular dynamics (MD) simulations of the p53-DBD WT , p53-DBD R158G , and acetylated Lysine p53-DBD R158G were carried out with the pemed.CUDA module of the program Amber16 with the Amber 14SB force field (ff14SB). Force field parameters for acetylated Lysine were taken as described previously 68 . The Xleap module was used to prepare the system for the MD simulations. All the simulation systems were neutralized with appropriate numbers of counterions. Each neutralized system was solvated in an octahedral box with TIP3P water molecules, leaving at least 10 Å between the solute atoms and the borders of the box. All MD simulations were carried out in explicit solvent at 300 K. During the simulations, the long-range electrostatic interactions were treated with the particle mesh Ewald method using a real space cutoff distance of 9 Å. The Settle algorithm was used to constrain bond vibrations involving hydrogen atoms, which allowed a time step of 2 fs during the simulations. Solvent molecules and counterions were initially relaxed using energy minimization with restraints on the protein and inhibitor atoms. This was followed by unrestrained energy minimization to remove any steric clashes. Subsequently the system was gradually heated from 0 to 300 K using MD simulations with positional restraints (force constant: 50 kcal mol −1 Å −2 ) on the protein over a period of 0.25 ns allowing water molecules and ions to move freely. During an additional 0.25 ns, the positional restraints were gradually reduced followed by a 2 ns unrestrained MD simulation to equilibrate all the atoms. Production simulations were carried out for 250 ns each in triplicates. Enhanced conformational sampling was carried out by subjecting the systems to accelerated MD (aMD) implemented in AMBER 16, using the "dual-boost" version. Conventional MD simulations carried out earlier were used to derive the aMD parameters (EthreshP, alphaP, EthreshD, alphaD). aMD simulations were carried out for 250 ns each. Simulation trajectories were visualized using VMD and figures were generated using Pymol.
Chromatin immunoprecipitation (ChIP)-seq, AmpliSeq analyses. At least 16 million 51-bp long reads were mapped to hg19 using bowtie v2.1.0 with parameters -N 1 -sensitive -p 2 -no-unal. Peaks were identified by MACS 2.0.9 using a maximum of 2 reads per unique position and otherwise default parameters. Only highly enriched (enrichment over background ≥5-fold, pileup ≥25) and highly significant (q-value < 0.01) peaks were shortlisted from the analysis. de novo motif search was conducted by MEME-ChIP54 using the TSS proximal peaks center extended by 75 bp in both directions to determine the distribution of wtp53 and mutp53-binding sites post-treatment with reference to hg19 annotations. MEME was programmed to search for the top 10 motifs with significant matches between query and target database using HOMER (Hypergeometric Optimization of Motif EnRichment) 55 v4.5. The heatmaps of p53 enrichments (wtp53 and mutp53 R158G ) across 3 kb region from peak centre were plotted using ngsplot, v2.61. For AmpliSeq, fastq files were analyzed with the ampliSeqRNA plugin available for Ion Torrent sequencing platforms that used the Torrent Mapping Alignment Program (TMAP). TMAP first identify a list of Candidate Mapping Locations (CMLs) and aligned with Smith Waterman algorithm for specific and sensitive mapping. DEG analysis was performed using R/Bioconductor package DESeq2, with read count normalized to produce differential gene expressions as determined by p-value and the log 2 fold change. GEO superSeries: GSE129027.
Comet assay. Comet assay was performed using OxiSelect 96-well kit (Cell Biolabs, #STA-355) according to manufacturer's recommendation. Briefly, treated cells were trypsinized and mixed with low melting-point agarose at 37°C. Cells/agarose mixture was quickly placed onto pre-coated 96-well Comet slide, and cooled at 4°C to solidify the agarose layer. The ready slide was treated with hypertonic lysis buffer, and followed by incubation with alkaline buffer to unwind DNA strands and expose the alkali labile sites (comet tails), which were imaged by fluorescence microscopy at ×20 magnification (Olympus IX71/DP71). Olive Tail Moment (OTM) was quantified using ImageJ software (NIH, 1.51n) with OpenComet plugin, using the formula OTM = (mean Tail-Intensitymean Head-Intensity) X Tail%DNA/100. The mean ± S.D. of OTM from at least 50 cells in each treatment group was quantified.
Nuclear NFĸB activity assay. Activation of nuclear NFĸB p65 subunit was quantified in nuclear protein extracts using the ELISA-based TransAM p65 NF-kB Kit (Active Motif, #40596) according to manufacturer's recommendation. Cellular fractionation was performed as described, and absorbance measurements (450 nm) on nuclear extracts (5 µg) in triplicates were performed using TECAN Infinite plate reader.
Tumor xenograft assay. All in vivo xenograft studies adhered to the SingHealth Animal Care and Use Committee (IACUC) guidelines on animal use and handling. Usage of patient-derived materials have been approved by SingHealth Institutional Review Board (IRB). All tumor xenografts were established in male C.B-17 SCID mice aged 9-10 weeks and weighed 23-25 g (InVivos Pte. Ltd., Singapore). Mice were provided with sterilized food and water ad libitum, and housed in negative pressure isolators, which were set at 23°C and 43% humidity, with 12-h light/dark cycles.
In brief, 10 × 10 6 isogenic Calu-1 cells (empty vector control, wtp53, mutp53) in 100 µL of PBS or 1-2 mm 3 dices of PDX tumors (GC-PDX and HCC-PDX) were injected subcutaneously into the flanks of each mice. Tumor size was monitored and measured three times weekly with Vernier calipers (tumor volume = length × width 2 × 3.14159/6). For inhibitor studies, mice were assigned into stratified groups based on average tumor volume (n = 5 animals/10 tumors per group). Treatment began when the tumor sizes reached~200 mm 3 . Cisplatin was provided in 0.9% NaCl and given at 4 mg kg −1 by intraperitoneal (ip), once weekly for 3 weeks. JQ1 and topotecan were formulated in 30% (w/v) Captisol and 30 % (w/v) PEG300 at 30 mg kg −1 and 1 mg kg −1 , respectively, and given via PO (orally), 7 days a week. Bodyweight at sacrifice and tumor samples were collected 3 weeks after treatment commenced.
Immunohistochemistry (IHC), RNA in situ hybridization (ISH). Xenograft tumors from control and compounds administered mice were fixed in formalin after surgical excision overnight, and embedded in paraffin and sectioned (4 µm). Tumor sections were deparaffinized using standard histologic procedures prior to antigen retrieval. For IHC, slides were incubated with primary antibodies [NF-κB p65 (1:1000, Cell Signaling, #8242), cleaved caspase 3 (1:1000, Cell Signaling, #9661), Ki67 (1:1000, Abcam, #ab15580) overnight at 4°C. Color development was performed using EnVision+ System-HRP (DAB) kit (Dako, Agilent Technologies) according to the manufacturer's protocol. Single molecule mRNA ISH (RNAscope, ACD Biotech) was performed using the Advanced Cell Diagnostics RNAscope 2.5 HD Detection kit (#322360) according to manufacturer's recommendation. TRAIP probe was customized. For both IHC and RNAscope, cell nuclei were counterstained with Hematoxylin (Thermo Scientific, #72604). Slide images were captured under optical imaging microscope at ×20 magnification (Olympus IX71/DP71). Images were acquired from five tumor samples each for every treatment group, and analyzed with ImageJ software using an image processing workflow described previously 67 . Briefly, Hematoxylin and DAB/Fast Red color-separated images were derived from the original image using ImageJ Color Deconvolution plugin. Object segmentation using the Particle Analyzer plugin was then performed to identify the individual cellular regions in each image, and the signal intensity (DAB) or count (Fast Red) in the image was then tabulated. IHC staining was analyzed as the percentage of DAB-positive cells/100 cells in each slide image. RNA-ISH staining was scored based on the number of positive signal in each cell (0: no staining or <1 dot/ 10 cells; 1: 1-3 dots/cell; 2: 4-9 dots/cell; 3: 10-15 dots/cell; 4: >15 dots/cell). At least 20 images were analyzed for each tumor sample.
Statistics, data reporting, and reproducibility. No statistical method was applied to predetermine sample size. All experiments were conducted three times unless stated otherwise. Data that could be reproduced in biological replicates were reported. Statistical analysis for the comparison between two groups was conducted using Student's t-test, while comparisons between multiple groups was conducted using two-way ANOVA post-hoc tests.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
The ChIP-seq and AmpliSeq datasets generated during the current study have been deposited in GEO with the Accession number GSE129027 and could be accessed at this link: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE129027. All the other data supporting the findings of this study are available within the article and its supplementary information files and from the corresponding author upon reasonable request. A reporting summary for this article is available as a Supplementary Information file.