Salvation of the fallen angel: Reactivating mutant p53

Abstract

The transcription factor p53 is known as the guardian of the genome for its powerful anti‐tumour capacity. However, mutations of p53 that undermine their protein structure, resulting in loss of tumour suppressor function and gain of oncogenic function, have been implicated in more than half of human cancers. The crucial role of mutant forms of p53 in cancer makes it an attractive therapeutic target. A large number of candidates, including low MW compounds, peptides, and nucleic acids, have been identified or designed to rescue p53 mutants and reactivate their anti‐tumour capacity through a variety of mechanisms. In this review, we summarize the progress made in the reactivation of mutant forms of p53, focusing on the pharmacological mechanisms of the reactivators of p53 mutants.

Abbreviations

DBD

DNA‐binding domain

GOF

gain of function

mutp53

mutant p53

OD

oligomerization domain

TAD

transactivation domain

1. INTRODUCTION

The transcription factor p53 is one of the most widely studied tumour suppressors and has been referred to as the “guardian of the genome” due to its pivotal role in the anticancer signalling pathway (Bieging, Mello, & Attardi, 2014; Lane, 1992). Clinically, p53 is the most frequently mutated gene across a broad spectrum of cancer types, with an estimated mutation rate of over 50% (Lawrence et al., 2014). p53 controls and regulates a series of cell processes, including apoptosis, senescence, differentiation, autophagy, and metabolism, as well as contributing to cell fate: survival or apoptosis (Figure 1; Kruiswijk, Labuschagne, & Vousden, 2015). The function of p53 relies on its transcription factor activity regulating many downstream target genes (Menendez, Inga, & Resnick, 2009) and its interaction with other proteins in the cytoplasm (Green & Kroemer, 2009). In cytoplasm, p53 binds with Bcl‐2 family proteins to induce mitochondria apoptosis (Aubrey, Kelly, Janic, Herold, & Strasser, 2018; Sola, Morgado, & Rodrigues, 2013) and interacts with AMPK and mTOR to inhibit autophagy (Sanli, Steinberg, Singh, & Tsakiridis, 2014; Yu, Li, et al., 2018).

Figure 1.

The crucial role of p53 signalling. p53 is activated by cell stress, such as DNA damage, physical or chemical perturbation, hypoxia, and nutrient fluctuation, which induce phosphorylation and acetylation of p53 through protein kinases. Then, p53 translocates into the nucleus, assembles into a tetramer (Protein Data Bank code 2AC0), binds to target DNA, and initiates transcription of many genes that regulate a series of cell processes, including cell cycle, apoptosis, metabolism, and DNA repair. The level of p53 is largely controlled by negative feedback regulation of MDM2, a downstream E3 ubiquitin ligase of p53 that inhibits p53 activity by mediating p53 degradation and blocking the phosphorylation site in the transcriptional activation domain

p53 is activated following cell stress such as DNA damage, physical or chemical perturbation, hypoxia, and nutrient fluctuation, which induce phosphorylation and acetylation of p53 through protein kinases (Humpton & Vousden, 2016). A number of protein kinases are involved in p53 phosphorylation, for example, ATR that responds to persistent single‐stranded DNA and ATM that responds to double strand breaks in DNA or chromatin disruption (Figure 1; Blackford & Jackson, 2017; Sancar, Lindsey‐Boltz, Unsal‐Kacmaz, & Linn, 2004). Phosphorylation of p53 releases it from MDM2, a negative regulator of p53 (Blackford & Jackson, 2017), and allows p53 to bind to p300 and PCAF (Figure 1), the acetyltransferases that acetylate p53, leading to exposure of the DNA‐binding domain (DBD; Jin, Zeng, Dai, Yang, & Lu, 2002; L. Liu et al., 1999). After phosphorylation and acetylation, p53 is activated and translocates into the nucleus, assembles into a tetramer through the oligomerization domain (OD), and binds to the target DNA sequence through the DBD (Figure 1; Demir, Ieong, & Amaro, 2017; Friedman, Chen, Bargonetti, & Prives, 1993; Kitayner et al., 2006). The protein–protein interaction within the p53 tetramer not only stabilizes the structure of DBD but also supports the p53–DNA interaction, which “locks” the target DNA sequence (Kitayner et al., 2006). The p53–DNA interaction initiates transcription of many genes (Riley, Sontag, Chen, & Levine, 2008), such as p21 that leads to cell cycle arrest (Abbas & Dutta, 2009) and GADD45 that inhibits cell growth (Tamura et al., 2012). Moreover, the DNA–p53 interaction promotes the release of cytochrome c from mitochondria by initiating BAX transcription (Geng et al., 2010), and also up‐regulates PUMA (Yu & Zhang, 2008), which then frees Bax and/or Bak via interacting with anti‐apoptotic Bcl‐2 family members and triggers apoptosis of mitochondria. Moreover, the apoptosis regulator NOXA (Oda et al., 2000), as well as metabolism‐related genes, such as GLS2 that catalyses the hydrolysis of glutamine (Suzuki et al., 2010) and TIGAR that regulates glucose breakdown in human cells (Bensaad et al., 2006), are also initiated by DNA–p53 interaction. Also, p53 induces mitochondrial membrane permeabilization by directly interacting with multiple Bcl‐2 family members, resulting in transcription‐independent cell death (Vaseva & Moll, 2009). The level of p53 is largely controlled by negative feedback regulation of MDM2 (Wade, Li, & Wahl, 2013), a downstream E3 ubiquitin ligase of p53 that inhibits p53 activity in two ways: (a) binding to p53 and ubiquitinating its C‐terminal lysine residues, mediating p53 degradation by the proteasomes (Kubbutat, Jones, & Vousden, 1997) and (b) inhibiting p53 activation by blocking the phosphorylation site in the transactivation domain (TAD; S. Wang, Zhao, Aguilar, Bernard, & Yang, 2017). MDMX (also known as MDM4), a binding partner of MDM2, is also a negative regulator of p53 by binding to p53 and blocking its TAD (Karni‐Schmidt, Lokshin, & Prives, 2016). Overall, a wide range of post‐translational modifications and protein–protein interactions orchestrate the regulation of p53 activity (Bode & Dong, 2004).

The p53 protein is composed of 393 amino acid residues. Beginning at the amino terminus, p53 contains a TAD, followed by a proline‐rich region, DBD, OD, and carboxyl terminus domain that has a self‐suppressive effect on p53 (Figure 2a; Joerger & Fersht, 2010). With the exception of DBD and OD, which are in ordered structures, the other regions are highly flexible and intrinsically disordered (Figure 2c; Wells et al., 2008). The structured DBD, which is crucial for the conformational stability of p53 (Cho, Gorina, Jeffrey, & Pavletich, 1994), comprises a β‐sandwich scaffold embraced by three unstable loops, and the loops are stabilized by a tetrahedrally coordinated zinc ion through Cys¹⁷⁶, His¹⁷⁹, Cys²³⁸, and Cys²⁴² (Figure 2c). Human p53 is thermodynamically unstable with a melting point of approximately 44 °C. It is marginally stable at body temperature and may rapidly unfold with a half‐life of 9 min (Bullock et al., 1997). This low intrinsic stability is necessary for p53 to conduct its versatile biological activities, making p53 flexible enough to facilitate binding to many different partners (Lavin & Gueven, 2006; Okorokov et al., 2006). On the other hand, the low intrinsic stability makes p53 sensitive to missense mutations, accounting for over 70% of the oncogenic p53 mutations (Soussi & Wiman, 2015). Approximately 95% of the cancer‐related p53 mutations are located in the DBD and several hot spots, with high mutation frequency, have been identified in the DBD, such as Arg¹⁷⁵, Arg²⁴⁸, and Arg²⁷³ (Figure 2b; data from the IARC TP53 database; http://p53.iarc.fr). These mutations can be divided into two types: mutations that weaken the affinity between p53 and its cognate DNA‐binding site (e.g., Arg²⁴⁸ and Arg²⁷³) and mutations that destabilize the local (e.g., Gly²⁴⁵ and Arg²⁴⁹) or even the global conformation (e.g., Arg¹⁷⁵ and Arg²⁸²; Figure 2b; Joerger & Fersht, 2007). Both types of mutations decrease the thermodynamic stability of mutant p53 (mutp53), leading to alteration of p53 structure and loss of tumour suppressive function (Soussi, Kato, Levy, & Ishioka, 2005). Notably, mono‐allelic p53 mutations may show a dominant‐negative effect by driving co‐expressed wild‐type (WT) p53 into the misfolded conformation (Billant et al., 2016; Dridi et al., 2006; Milner & Medcalf, 1991). Mutp53 may also promote tumourigenesis through gain of function (GOF), which has been shown to profoundly influence the development of cancers (Brosh & Rotter, 2009). GOF phenotypes of mutp53 have been widely observed and include enhanced cell proliferation, drug resistance, angiogenesis, invasion and metastasis, deregulated genomic stability, metabolism, inflammation, and micro‐environment (Brosh & Rotter, 2009). Mutp53 may inappropriately interact with many cytoplasmic proteins, including HSP90 and DAB2IP, or transcription factors, such as NF‐Y, E2F, and the p53 paralogous tumour suppressors p63 and p73 (Brosh & Rotter, 2009). These aberrant interactions all represent important molecular mechanisms of the GOF by mutp53 (Brosh & Rotter, 2009).

Figure 2.

p53 mutations and structure analysis. (a) The domain architecture of the p53 protein. (b) Distribution of somatic p53 mutations; hot spots of p53 mutation favour the DNA‐binding domain (data derived from the IARC TP53 mutation database, version R18, April 2016). (c) Structure of p53 DNA‐binding domain (DBD; Protein Data Bank code: 1TSR). Mutation hot spots are coloured purple; the most aggregation‐prone segment S9 is coloured red; the zinc is coloured black, and the zinc chelated His¹⁷⁹ is coloured blue; cysteines (including the zinc chelated Cys¹⁷⁶, Cys²³⁸, and Cys²⁴²) in p53 DBD are coloured yellow; p53 213–217 (recognition epitopes of unfolded p53 antibody pAb240) are coloured orange; while Arg¹⁵⁶, Leu²⁰⁶, Arg²⁰⁹, and Gln/Asn²¹⁰ (recognition epitopes of conformation specific p53 antibody pAb1620) are coloured green. (d) Molecular docking of methylene quinuclidinone (the active form of PRIMA‐1 and PRIMA‐1^Met) and p53 R175H; methylene quinuclidinone occupies the transient pocket of p53 mutants and can covalently bind to the Cys¹²⁴. (e) Crystal structure of DBD in the p53 Y220C mutant, in complex with PhiKan083 (Protein Data Bank code: 2VUK). The tyrosine‐to‐cysteine mutation creates a unique druggable surface crevice. PhiKan083 occupies this crevice and stabilizes the structure of the Y220C mutant

The altered functions of mutp53 are intertwined with the amyloidogenic process of p53 aggregation. As previously demonstrated by Wang and Fersht (2012, 2015, 2017) and Wilcken, Wang, Boeckler, and Fersht (2012), p53 misfolds or unfolds into an aggregation‐prone stage that loses its DNA‐binding capacity. After the transition stage, p53 rapidly aggregates, by itself or with other proteins, into oligomers and subsequently into fibrils (Figure 3), and finally the aggregates accumulate and become resistant to degradation (Wang & Fersht, 2012). p53 aggregates also exhibit prion‐like behaviour, which converts normally folded proteins to the abnormal forms, propagating catalytically (Rangel, Costa, Vieira, & Silva, 2014; Silva, De Moura Gallo, Costa, & Rangel, 2014). Similarly, misfolded p53 may convert WT p53 to a misfolded form and accelerate p53 aggregation (Forget, Tremblay, & Roucou, 2013). In addition to the prion‐like behaviour of p53 aggregates, they can enter cells through micropinocytosis and subsequently co‐aggregate with endogenous p53, which enables cell‐to‐cell transmission of p53 aggregates (Ghosh et al., 2017). This chain reaction may be the mechanism underlying the dominant‐negative effect of mutp53 (Forget et al., 2013).

Figure 3.

Schematic representation of mutp53 aggregation and reactivation by mutp53 reactivators. The amyloidogenic property of p53 not only demonstrates its significant role in cancer but also provides the basis for a prevention/reversal strategy. Mutp53 are more easily unfolded and misfolded than wild‐type (WT) p53 due to their highly unstable nature, promoting self‐aggregation or aggregation with other proteins. Unfolded and misfolded p53 lack tumour suppressor activity. The aggregation process is involved in the mechanisms of GOF. Two strategies that interfere with this process are described: (a) inhibition of mutp53 unfolding/misfolding and (b) inhibition of misfolded p53 aggregation. Reactivators of p53 guided by these strategies can restore mutp53 function or block the gain of function (GOF) pathway that tumour cells are adapted to, causing cell death and proliferation arrest, thus leading to tumour regression

Several domains of p53 have shown a propensity to aggregate, but the major segment driving amyloid formation is the central hydrophobic β‐sandwich scaffold in the DBD, including residues 251–257 (Figure 2c; Xu et al., 2011). The amyloidogenic properties of proteins are known to associate with many factors, such as amino acid sequences (Ma et al., 2018), aromatic residues (Zhang et al., 2011), disulfide bonds (Li et al., 2012; Li, Yan, Zhang, & Huang, 2013), and exposure of hydrophobic sequences (Li, Huang, et al., 2013). Oncogenic mutations of p53 in the DBD increase the exposure of these hydrophobic sequences, which may initiate the aggregation of p53 and induce the aberrant interaction with other proteins resulting in the GOF (Cino, Soares, Pedrote, de Oliveira, & Silva, 2016). The aggregated mutp53 lose their DNA‐binding capacity (Wang & Fersht, 2015; Wilcken et al., 2012) and may even convert WT p53 to misfolded forms (Silva et al., 2014). The propensity of p53 to aggregate makes p53‐driven cancers potentially amyloid diseases, sharing similarities with Type 2 diabetes caused by amylin (also known as hIAPP), Parkinson's disease caused by α‐synuclein, and Alzheimer's diseases caused by amyloid‐β, that is, protein aggregation contributing to a pathological process (Gong et al., 2015; Huang, Liu, Cheng, & Huang, 2015). Inhibiting aggregation of these amyloidogenic proteins with inhibitors, such as low MW compounds (Cheng et al., 2012; Gong et al., 2014; Guo et al., 2015; Y. Zhang et al., 2018), metal chelators (Spinello, Bonsignore, Barone, Keppler, & Terenzi, 2016), peptides (Cheng et al., 2018), and antibodies (Cheng et al., 2013; Giorgetti, Greco, Tortora, & Aprile, 2018), has been regarded as an important strategy to treat these diseases (Mullane & Williams, 2018a, 2018b). Thus, therapeutic strategies for amyloidogenic disease may be relevant for targeting mutp53‐driven cancers.

In the battle against cancer, p53 constitutes one of the most important strategic targets, as once p53 is activated, the cancer cell is suppressed (Ventura et al., 2007). The inactivation of p53 is an essential step in almost all cancer development, and mutations of p53 usually provide an unfavourable prognosis (Petitjean, Achatz, Borresen‐Dale, Hainaut, & Olivier, 2007). Thus, cancer therapies have been designed and tested that target p53, such as restoring p53 conformation (Bullock & Fersht, 2001), delivering p53 to cancer cells (Lane, Cheok, & Lain, 2010), and targeting p53 regulators (Gomes, Ramos, Soares, & Saraiva, 2018; Schulz‐Heddergott & Moll, 2018). In roughly half of the instances, p53 inactivation is caused by dysfunction of p53 regulators, for example, MDM2 and MDMX, and their inhibitors that can restore p53 function have long been in clinical trials (Gembarska et al., 2012; Wang et al., 2017). ALRN‐6924, a peptide that strongly binds to MDM2 and MDMX (Carvajal et al., 2018), showed a disease control rate of 59% for advanced solid tumours and lymphomas in Phase I trial and has entered Phase IIA trial for peripheral T‐cell lymphoma (nos NCT02264613 and NCT02909972; www.clinicaltrials.gov). On the other hand, the inactivation induced by mutp53‐is much more difficult to reverse and drug development targeting this aspect of mutp53 has been relatively slow due to several challenges. First, a detailed and dynamic structure of full length p53 is still not available due to its highly flexible nature. Second, structurally, p53 and most of its mutants (except for Y220C) lack crevices or allosteric sites that are commonly used for rational drug design. Third, it is almost impossible to target the over 2,000 known p53 mutants, all with different conformations and biological functions, using a single compound. Therefore, mutp53 have traditionally been regarded as “undruggable” (Weissmueller et al., 2014). However, modern techniques, including structural biology, protein engineering, and computational simulation, are gradually revealing the conformation of p53 (Demir et al., 2017). Recent attempts at reactivation of p53 that combined virtual screening, rational drug design, and cell‐based large‐scale drug screening have produced a number of mutp53 reactivators (Ahire, Das, Mishra, Kulkarni, & Ackland, 2016; Parrales & Iwakuma, 2015).

2. MUTP53 REACTIVATORS: MULTIPLE STRUCTURES WITH TWO MECHANISMS

Both low MW compounds and macromolecules show potential for reactivating mutp53 (Table 1) through two primary mechanisms: inhibiting mutp53 misfolding and interfering with aggregation of misfolded p53 (Figure 3). The ideal mutp53 reactivator should meet the following criteria: demonstrated mutp53‐dependent tumour suppressor effect and up‐regulation of downstream p53 signalling including p21, PUMA, and MDM2; adequate anti‐tumour activity in vivo with good tolerance; increased thermal stability of mutp53 and inhibition of mutp53 misfolding, rehabilitating native conformation as demonstrated by a native conformation specific antibody pAb1620 recognition and decreased denatured/aggregate specific antibody pAb240 recognition; and substantially decreased mutp53 deposition, or reversing GOF.

Table 1.

Summary of p53 reactivators

Mechanism	Name	Structure	Method	Mutants saving	Research stage
Inhibition of protein misfolding
Covalent binding	CP31398 (Foster, Coffey, Morin, & Rastinejad, 1999)		pAb1620 affinity screening	R273H	Mouse model
	STIMA‐1 (Zache et al., 2017)		pAb1620 affinity screening	R175H, R273H	Mouse model
	PRIMA‐1 (Bykov et al., 2002)		Cell‐based screening	R175H, R273H	Mouse model
	PRIMA‐1Met (Bykov et al., 2002)		Cell‐based screening	R175H, R273H	Phase I/II clinical trial
	MIRA‐1 (Bykov et al., 2005)		Cell‐based screening	R175H, R273H	Cell model
	PK11007 (Bauer, Joerger, & Fersht, 2016)		Protein folding assay	Y220C	Cell model
Non‐covalent binding	SCH529074 (Demma et al., 2010)		DNA‐binding assay	R273H, R249S	Cell model
	P53R3 (Weinmann et al., 2008)		DNA‐binding assay	R273H, R249S	Cell model
	Stitic acid (Wassman et al., 2013)		In silico screening	R175H, G245S	Cell model
	PK7088 ( Liu et al., 2013)		Protein folding assay	Y220C	Cell model
	PhiKan083 (Boeckler et al., 2008)		In silico screening, protein aggregation	Y220C	Cell model
	R‐GON (Punganuru, Madala, Arutla, & Srivenugopal, 2018)		Cell‐based screening	R175H	Cell model
	MB725 (Baud et al., 2018)		Binding assay	Y220C	Cell model
Zinc ion chelating properties	NSC319726 (ZMC1; X. Yu, Vazquez, Levine, & Carpizo, 2012)		Cell‐based screening	R175H	Mouse model
	COTI‐2 (Salim, Maleki Vareki, Danter, & Koropatnick, 2016)		Machine learning in silico screen	Multiple mutations	Phase I clinical trial
	PEITC (Aggarwal et al., 2016)		Cell‐based screening	R175H	Mouse model
Chaperone/aptamer effect	pCAP‐250a (Tal et al., 2016)	Myr‐RRHSTPHPD	Phage display screening	R175H, R249S	Mouse model
	CDB3a (Friedler et al., 2002)	REDEDEIEW	In vitro binding assay	I195T, R249S, G245S	Cell model
	P53R175H‐APTb (Chen et al., 2015)	ATTAGCGCATTTTAACATAGGGTGC	SELEX	R175H	Mouse model
Interference with misfolded p53 aggregation
Inhibit p53 aggregation	ReACp53a (Soragni et al., 2016)	RRRRRRRRRLTRITLE	Rational design derived from P53 251–257	R175H, R248Q, I251S, T234C, Y326L	Mouse model
Release of p73 from mutant p53	RETRA (Kravchenko et al., 2008)		Cell‐based screening	R273H, R248W, R280K, G266E	Mouse model
Inhibit HDAC, decrease HSP90, promote mutant p53 degradation	SAHA ( Li, Marchenko, & Moll, 2011)		Cell‐based screening	R280K, R273K, L194F, R175H, R273H, P309S, P223L, V274F,	FDA approved
Binds to HSP40, increase chaperone effect	Chetomin (Hiraki et al., 2015)		Cell‐based screening	R175H	Mouse model
Mechanism unclear	KSS‐9 (Punganuru et al., 2016)		Cell‐based screening	R175H	Cell model
	SLMP53‐1 (Soares et al., 2016)		Yeast‐based screening assay	R280K	Mouse model

Note. FDA: Food and Drug Administration; SELEX: systematic evolution of ligand exponential enrichment.

^a This molecule is a peptide.

^b This molecule is an RNA.

2.1. Mechanism 1: Inhibiting protein misfolding

Most mutp53 are temperature sensitive and thus are prone to misfolding at normal body temperature. Drugs may inhibit misfolding of mutp53 and restore the WT‐like structure and function through decreasing the Gibbs free energy of mutp53 for unfolding, raising its melting point. There are a number of agents that utilize this mechanism that can be grouped into the following categories.

2.1.1. Covalent binding to mutp53

Several soft electrophiles bind to the cysteine residues in the mutp53 DBD to stabilize mutp53 conformation, thus restoring their transcriptional activities (Zache et al., 2017). Among them, PRIMA‐1 and its methylated derivative PRIMA‐1^Met (APR‐246) are well‐characterized reactivators that covalently bind to mutp53 (Perdrix et al., 2017; Wesierska‐Gadek, 2018). PRIMA‐1 was first identified through screening a library of 2,000 compounds in a saos‐2 cell model expressing the R273H mutp53 (Bykov et al., 2002). Its derivative, PRIMA‐1^Met, was later found to be more effective, probably due to its higher lipophilicity, and is currently in Phase Ib/II clinical trials (Bykov et al., 2016). Physiologically, both compounds can be converted into active methylene quinuclidinone, which covalently binds to cysteine residues in p53 to stabilize p53 (Figure 2d; Lambert et al., 2009). PRIMA‐1^Met also demonstrated strong synergistic effects with other anticancer drugs that employed different pharmacological mechanisms, including the DNA damaging agents doxorubicin and cisplatin, the proteasome inhibitor carfizomib, the MDM2 inhibitor nutlin‐3, the histone methyltransferase EZH2 inhibitor DZNep, the PARP inhibitor olaparib, and the BRAF inhibitor vemurafinib (Bykov, Eriksson, Bianchi, & Wiman, 2018).

Additional compounds that use a covalent binding mechanism to reactivate mutp53 are in an experimental or preclinical stage of development. Through in vitro structure‐based screening, CP‐31398 was identified as the first mutp53 reactivator to restore native WT‐like p53 conformation from denatured forms of mutp53, up‐regulate genes downstream of p53 including MDM2 and p21 in multiple cancer cell lines, and subsequently induce apoptosis (Foster et al., 1999). STIMA‐1 is a CP‐31398 derivative, which also induced suppression of mutp53‐dependent tumour growth (Zache et al., 2017). Based on the same screening approach used to identify PRIMA‐1, a maleimide‐derived molecule, MIRA‐1, was observed to exert mutp53 reactivating activity via covalent binding (Bykov et al., 2005).

The 2‐sulfonylpyrimidine, PK11007, designed to stabilize the Y220C mutant of p53, turns out to be an alkylator of p53 cysteines (Bauer et al., 2016). Alkylation of cysteine residues in p53 may prevent disulfide bond formation that induces protein misfolding, creating new DNA contacts, which allow more efficient DNA binding promoting correct protein folding (Lambert et al., 2009). Computational simulation identified a transiently open pocket between loop1 and sheet3 of the p53 DBD as a critical target for reactivating mutp53 (Wassman et al., 2013). Cys¹²⁴, Cys¹³⁵, or Cys¹⁴¹ in the pocket are thought to be the binding sites of methylene quinuclidinone, STIMA‐1, and MIRA‐1 through the Michael addition reaction (Wassman et al., 2013). Crystal structure reveals that PK11007 alkylates Cys¹⁸² or Cys²⁷⁷ through nucleophilic aromatic substitution (Bauer et al., 2016). However, the detailed mechanisms by which these compounds act to stabilize mutp53 remain incompletely understood.

2.1.2. Non‐covalent binding to mutp53

As an alternative to covalent binding, chaperones can non‐covalently stabilize mutp53 structures to preclude their misfolding. Chaperones, such as low MW compounds, peptides, and nucleic acids, show important roles in regulating p53 stability and aggregation (Kovachev et al., 2017). In addition, macromolecules usually have relatively high binding specificity compared to the low MW compounds involved in covalent binding.

Using a screening assay based on p53 DNA binding ability, SCH529074, a low MW reactivator of the mutants R273H and R249S was identified (Demma et al., 2010). SCH529074 is thought to non‐covalently bind to the DBD of p53, restore tumour suppressor function, and interrupt HDM2‐mediated ubiquitination of mutp53 (Demma et al., 2010). The same screening strategy also identified P53R3, which shares a similar structure with SCH529074 (Weinmann et al., 2008). P53R3 induces p53‐dependent antiproliferative effects, and it shows much higher p53 binding specificity than PRIMA‐1 in p53 null glioma cell lines that overexpress several exogenous p53 mutants (Weinmann et al., 2008).

Using molecular docking to screen for chemical compounds that interact with mutp53 has provided several effective compounds. Virtual screening based on the transient pocket formed by mutp53 using a computer simulation suggests that stictic acid is a potential mutp53 reactivator, and this compound exhibited strong induction of p21 in human osteosarcoma cells bearing the mutp53 R175H (Wassman et al., 2013).

The Y220C mutation in p53, the ninth most frequent p53 cancer mutation, dramatically destabilizes the p53 DBD. However, the Y220C mutation creates a unique druggable crevice on the surface of p53 making it a classic model for mutp53 rational drug design. The Fersht group discovered a series of compounds with potential for reactivation of the Y220C mutant, such as PhiKan083 (Boeckler et al., 2008) and PK7088 (X. Liu et al., 2013). These molecules occupy the Y220C druggable pocket, stabilize protein structure, and raise the melting point of the Y220C mutp53 (Figure 2e). Fersht also described CDB3, a nine‐residue peptide derived from a p53‐binding protein, 53BP2, that could bind to the p53 DBD to stabilize it (Friedler et al., 2002). Further studies showed that CDB3 interacts and partly overlaps the edge of the p53 DNA‐binding site and, in vitro, this peptide was able to restore the conformation of p53 hot spot mutants such as G245S, R249S, and I195T (Friedler et al., 2002).

A cell‐based high‐throughput screen identified the fungal extract chetomin as a specific reactivator of mutp53 R175H. However, chetomin rescues mutp53 indirectly as it interacts with HSP40 and increases the chaperone effect of HSP40 on the R175H mutant, inducing a conformational change of this mutant to a WT‐like p53, thus activating downstream p53 signalling (Hiraki et al., 2015). This indirect effect reveals that there may be additional mechanisms based on the complex p53‐chaperone interacting network that could be used to rescue mutp53.

Tal et al. (2016) adopted phage display technology to select mutp53 reactivating peptides. By allowing interactions between random peptide libraries presented on phages and mutp53, peptide sequences with the ability to stabilize correctly folded mutp53 were enriched. In this way, a large pool of potential reactivating peptides was identified. In mouse xenograft models, lead peptides, called pCAPs, caused dramatic tumour regression (Tal et al., 2016). Perhaps not surprisingly, many of the lead peptides share sequence similarity with known p53‐binding proteins (Tal et al., 2016).

Using systematic evolution of ligand exponential enrichment process, an RNA aptamer p53R175H‐APT that targeted the R175H p53 mutant was isolated. This RNA aptamer has a stronger affinity for the R175H mutant than for the WT p53, leading to the correction of the R175H mutant to WT p53 conformation (Chen et al., 2015).

The structure‐guided design of a novel class of aminobenzothiazole derivatives, MB710 and MB725, which bind tightly to the Y220C pocket and stabilizes this mutp53 in vitro; and MB725, an ethylamide analogue of MB710, induced selective reduction of viability in several Y220C cancer cell lines while being well tolerated in control cell lines (Baud et al., 2018). Another candidate is goniothalamin, which was toxic to coronary artery smooth muscle cells by activating the p53 pathway (Chan et al., 2010). Later, it was found to induce apoptosis in the p53‐positive hepatocellular carcinoma‐derived cells (Kuo et al., 2011). Recently, R‐goniothalamin, the R‐enantiomer of goniothalamin, was reported to selectively kill human breast cancer cells by markedly reactivating the mutp53 R175H (Punganuru et al., 2018).

2.1.3. Zinc ion chelators

Zinc ion is required for the stabilization of WT p53 structure and some p53 hot spot mutants (e.g., R175H) fail to coordinate zinc (Butler & Loh, 2003). Moreover, the concentration of zinc affects the conformation and aggregation of p53 (Butler & Loh, 2003; Meplan, Richard, & Hainaut, 2000). Zinc ions not only increase the tumour suppressor ability of mutp53 by restoring their active conformations (Puca et al., 2011) but also release WT p53 from mutp53 binding by accelerating the clearance of misfolded mutp53 through autophagy (Garufi et al., 2014). Some metal chelators may specifically rescue zinc‐relevant p53 mutations.

NSC319726 (ZMC1), a compound of the thiosemicarbazone family, was identified through analyzing the anticancer drug screening data of the National Cancer Institute ( Yu et al., 2012). ZMC1 restores the mutp53 R175H to a WT‐like structure and function by providing an optimal zinc concentration and promoting correct folding (Yu et al., 2012). It also rescues several other p53 missense mutants with impaired zinc binding, such as th eC242S, C242F, and C176F mutants (Yu et al., 2014). Moreover, two derivatives of ZMC1, which also belongs to the thiosemicarbazone family, behave similarly to ZMC1 in biophysical and cell‐based assays (Yu et al., 2017). Further studies indicated that ZMC1 is sufficient to significantly improve survival in mice expressing a zinc‐deficient allele (p53‐R172H) while having no effect in mice expressing a non‐zinc‐deficient allele (p53‐R270H; Yu, Kogan, et al., 2018).

COTI‐2 was identified using a computational platform CHEMSAS, which is an in silico machine learning system that predicts target biological activity from molecular structures (Salim et al., 2016). COTI‐2 inhibits the growth of a wide range of cancer cell lines, especially those bearing p53 mutations (Salim et al., 2016). COTI‐2 appears to work by both reactivating mutp53 and inhibiting the PI3K/Akt pathway (Salim et al., 2016). As COTI‐2 has a thiosemicarbazone structure similar to ZMC1, it is likely that COTI‐2 functions as a mutp53 reactivator by zinc chelation (Salim et al., 2016; Yu et al., 2012).

A low MW compound, phenethyl isothiocyanate, derived from edible cruciferous plants, was shown to reactivate mutp53, especially the R175H mutant, with implications for dietary cancer prevention and therapy (Aggarwal et al., 2016). The authors claimed phenethyl isothiocyanate functions through chelating zinc, and it is now in clinical trials to treat cancer, by a variety of effects, including inhibition of cytochrome P450s, induction of Phase II detoxifying enzymes, cell cycle arrest, and apoptosis (Aggarwal et al., 2016).

2.2. Mechanism 2: Inhibiting aggregation of misfolded p53

The GOF of p53 mutations make p53 a proto‐oncogene. There is evidence that blocking the mutp53 GOF pathway in cancer cells bearing p53 mutations leads to tumour suppression (Brosh & Rotter, 2009). As mentioned above, the amyloidogenic nature of p53 plays a critical role in p53 GOF, so targeting p53 aggregation should provide another promising strategy for reactivation of mutp53.

2.2.1. Targeting amyloidogenic sequences in p53

A peptide, ReACp53, constructed to inhibit p53 amyloid formation rescues p53 function, which reduced in vivo xenograft growth and metastasis in p53 mutant, high‐grade serous ovarian carcinoma models (Soragni et al., 2016). ReACp53 was designed based on the protein sequence with the highest aggregation propensity in the p53 DBD (residues 251–257; ILTIITL) predicted using the ZipperDB algorithm (Ghosh et al., 2014). To target this amyloidogenic sequence, peptide mimics were designed retaining the original p53 sequence but introducing an aggregation‐inhibitory arginine in a central position with a poly‐arginine cell‐penetrating peptide fused to the N‐terminal. ReACp53 showed astonishing tumour suppression in ovarian carcinomas bearing many p53 mutations and was well tolerated in vivo (Soragni et al., 2016). Recently, additional peptides targeting p53 aggregation‐prone sequences have been tested in vitro but failed to inhibit mutp53 aggregation (Wang & Fersht, 2017). However, these peptides were undoubtedly cytotoxic towards cancer cells and decreased mutp53 deposit. This puzzling phenomenon, up to now, could only be explained by the hypothesis that physiological and experimental protein aggregation processes may be different. Physiologically, many molecules may participate in p53 aggregation, and peptides that interfere with these complex interactions are sufficient to block mutp53 GOF ( Wang & Fersht, 2017). The hypothesis is supported by a recent study showing that ReACp53 blocks the GOF of mutp53 R282W in cisplatin resistance, by inhibiting mutp53 aggregation ( Zhang et al., 2017).

2.2.2. Targeting the p53–p73 axis

As mentioned previously, the interaction of p53 with p73 represents a crucial axis of the p53 GOF. Blocking the mutp53–p73 interaction may free p73 from the trapping by mutp53, which reactivates the tumour suppressive function of p73. Using a reporter assay on an epidermal carcinoma cell line A431 bearing the mutp53 R273H, a compound RETRA (2‐(4,5‐dihydro‐1,3‐thiazol‐2‐ylthio)‐1‐(3,4‐dihydroxyphenyl)ethanone hydrobromide) was identified as a mutp53 reactivator, and its treatment of mutp53‐expressing cancer cells leads to a substantial increase in p73, which yields tumour suppressive effects, similar to mutp53 reactivation (Kravchenko et al., 2008).

2.2.3. Targeting p53–HSP90 interaction

HSP90 is a specific chaperone of mutp53 and a contributor to mutp53 aggregation. HSP90 forms a stable complex with mutp53, thus inhibiting mutp53 degradation mediated by p53 E3 ubiquitin ligases MDM2 and CHIP, and aggravates the mutp53 GOF pathway. Vorinostat (also known as SAHA) is a histone deacetylase inhibitor approved by the Food and Drug Administration. Recent research indicated that SAHA exhibits preferential cytotoxicity to mutant, rather than WT and null p53 cancer cells. A mechanistic study suggested that SAHA inhibits histone deacetylase 6, a critical positive regulator of HSP90, thus releasing mutp53 from the mutp53–HSP90 complex facilitating degradation of mutp53 (Li et al., 2011). Moreover, SAHA also sensitizes mutp53 cancer cells to chemotherapy ( Li et al., 2011).

2.3. Compounds with undefined mechanisms

Compounds that can reactivate mutp53 through various screening methods are easy to identify; however, their mechanism of action may not be as easy to elucidate. Extensive investigation is necessary to gain a comprehensive understanding of their mechanism of action. For instance, reconstruction of natural products with anticancer activity may produce more powerful anticancer products. An interesting example is KSS‐9, a hybrid molecule composed of two natural compounds: piperlongumine, an alkaloid that can produce ROS in cancer cells (Adams et al., 2012), and a well‐known antimicrotubule agent CA4 (Pettit et al., 1989). This hybrid compound exhibited robust tumour‐suppressive activities in breast cancer cells, particularly those harbouring an R175H mutation of p53. Moreover, KSS‐9 restored the conformation and activity of mutp53, while neither CA4 nor piperlongumine could do this individually (Punganuru et al., 2016). As a Michael acceptor, KSS9 may also bind to p53 cysteines, although direct proof of this is lacking (Punganuru et al., 2016).

SLMP53‐1 was identified as a novel reactivator of mutp53 R280K using yeast‐based screening, and it also increased the inhibitory effect of WT p53 on yeast growth (Soares et al., 2016). In both human WT p53 and mutp53 R280K expressing tumour cells, SLMP53‐1 showed anti‐proliferative activity, triggered p53 transcription‐mediated and mitochondrial apoptotic pathways, and inhibited migration. Moreover, it shows promising synergistic effects with conventional chemotherapeutics in a p53‐dependent manner (Soares et al., 2016). The mechanism by which SLMP53‐1 reactivates mutp53 R280K remains unclear.

3. GAPS FROM BENCH TO BEDSIDE

Although candidates with mutp53 reactivating ability seem abundant, most of them have remained in the preclinical stage for a variety of reasons. The first defect of the mutp53 rescuers that prevents them from reaching the clinic is their lack of specificity. Almost all low MW mutp53 reactivators have broad side effects (Bykov et al., 2018). For example, PRIMA‐1^Met inhibited proliferation of colorectal cancer cells, independent of p53 status (Li et al., 2015). Meanwhile, MEK and thioredoxin reductase 1 are also reported as targets of PRIMA‐1^Met (Lu et al., 2016; Peng et al., 2013). Following initial studies, CP‐31398 was found to destroy DNA, and MIRA‐1 is toxic to normal cell lines that were hoped to be insensitive to the mutp53 reactivator (Bou‐Hanna et al., 2015; Rippin et al., 2002). The ion chelator ZMC1 binds not only to Zn²⁺ but also to other transition metals, such as Fe²⁺ and Cu²⁺, and subsequently generates ROS and oxidative stress (Yu et al., 2012). Although they contribute to tumour suppression, further clinical development of these compounds may not be possible due to their poor selectivity.

The polypeptides and nucleic acid aptamers with Mutp53 reactivating properties show reduced side effects due to their high binding specificity to target (Chen et al., 2015; Tal et al., 2016), but the poor solubility or stability of these macromolecules makes efficient drug delivery another challenge. Polypeptides usually need to be decorated or fused with a cell‐penetrating peptide to modulate their amphiphilicity (Copolovici, Langel, Eriste, & Langel, 2014; Ma et al., 2017). RNA needs to be enclosed in nanoparticles (Fan et al., 2018). The instability of these macromolecular mutp53 activators remains a serious limitation to their application.

Blocking the p53 GOF has a broad spectrum of advantages for it provides solutions to p53 mutations that are too difficult to restore to their natural conformation (Kravchenko et al., 2008). However, the exact mechanism of how these compounds block p53 GOF is still elusive. Moreover, some compounds have diverse effects on many other cellular pathways. Hence, clinical application of these agents awaits thorough investigations.

4. CONCLUDING REMARKS

The development of mutp53 reactivators is promising. Theoretically, at least half of all cancer patients would benefit from strategies targeting p53. Also, the combination of mutp53 reactivators and traditional anticancer drugs may exhibit synergetic effects, as conventional anticancer drugs usually depend on an intact p53 pathway (Deben et al., 2016; Fransson et al., 2016), which expands the application of mutp53 reactivators. Encouragingly, some of the mutp53 reactivators are in clinical trials; PRIMA‐1^Met (APR‐246) and COTI‐2 showed positive results (Deneberg et al., 2016). Both low MW compounds and macromolecules show potential for p53 reactivation through a variety of mechanisms, including covalent binding, chaperone effect, ion chelation, and preventing protein aggregation. Although many questions remain to be answered, this master tumour suppressor now appears to be “druggable”, since its discovery nearly four decades ago.

4.1. Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander, Fabbro, et al., 2017; Alexander, Kelly, et al., 2017).

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

Li Y, Wang Z, Chen Y, Petersen RB, Zheng L, Huang K. Salvation of the fallen angel: Reactivating mutant p53. Br J Pharmacol. 2019;176:817–831. 10.1111/bph.14572