Molecular Targeting of the Most Functionally Complex Gene in Precision Oncology: p53

Simple Summary Precision medicine is a powerful treatment modality for controlling tumor growth. However, given the substantial variability between cancer patients, identifying suitable targets is no easy task. Tumor protein p53 represents one of the most promising anti-cancer drug targets due to its high mutation rate and critical role in tumorigenesis. Despite this, current treatment strategies targeting p53 signaling have only seen modest clinical success presumably due to the complex signaling network surrounding p53 enabling cancer cells to have the ability to adapt to each treatment strategy. This review will focus on the multiple approaches to target p53 and will touch on some of the limitations that may be hindering their clinical success. Abstract While chemotherapy is a key treatment strategy for many solid tumors, it is rarely curative, and most tumor cells eventually become resistant. Because of this, there is an unmet need to develop systemic treatments that capitalize on the unique mutational landscape of each patient’s tumor. The most frequently mutated protein in cancer, p53, has a role in nearly all cancer subtypes and tumorigenesis stages and therefore is one of the most promising molecular targets for cancer treatment. Unfortunately, drugs targeting p53 have seen little clinical success despite promising preclinical data. Most of these drug compounds target specific aspects of p53 inactivation, such as through inhibiting negative regulation by the mouse double minute (MDM) family of proteins. These treatment strategies fail to address cancer cells’ adaptation mechanisms and ignore the impact that p53 loss has on the entire p53 network. However, recent gene therapy successes show that targeting the p53 network and cellular dysfunction caused by p53 inactivation is now possible and may soon translate into successful clinical responses. In this review, we discuss p53 signaling complexities in cancer that have hindered the development and use of p53-targeted drugs. We also describe several current therapeutics reporting promising preclinical and clinical results.


Introduction
Despite decades of research and rapidly advancing treatment modalities, cancer remains the second leading cause of premature death worldwide [1,2]. Cells are constantly exposed to exogenous stressors that can cause DNA mutations, such as ionizing radiation (UV, X-rays, Gamma) and carcinogenic chemicals. Additionally, replication errors and defective DNA repair mechanisms can gradually accumulate mutations and cause cancer over time [3,4]. Multicellular organisms have evolved redundant tumor-suppressive functions to prevent malignant transformation, such as cellular senescence and apoptosis. However, cancerous cells can avoid, overcome, or override these suppressive mechanisms [5,6]. The Hanahan and Weinberg hallmarks of cancer illustrated some of these abilities, such as resistance to apoptotic stimuli, limitless proliferation, immune avoidance, persistent mutation accumulation, metastasis, and many more [7][8][9][10][11][12]. These hallmarks are likely driven by natural selection, as cells developing these characteristics show a survival and Following detection of single stranded or double stranded DNA breaks, the kinases ATM and ATR become activated. Both ATM and ATR phosphorylate p53 at Ser15. ATM also phosphorylates Chk2, which phosphorylates p53 at Ser20, while ATR phosphorylates Chk1. Phosphorylation of p53 at Ser15 and Ser20 causes it to be released from its negative regulator, MDM2. Additionally, Chk2 and Chk1 both inhibit MDMX, which functions to enhance the E3 ligase activity of MDM2 and inhibit p53. With p53 now released from MDM2, it functions as a transcription factor resulting in the expression of various tumor suppressive genes. For example, Bax and NOXA are proapoptotic proteins, while p21 functions to inhibit cell cycle progression. P53 also plays a role in regulating its own activity by facilitating the expression of more MDM2, while p21 acts to inhibit further DDR signaling by blocking the activity of Chk1. (B) Cell cycle progression also leads to increased activity of p53. As cells prepare to enter the S phase of the cell cycle, the kinases CDK4 and CDK6 become activated. CDK4/CDK6 phosphorylate the cell cycle inhibitor protein Rb, which causes it to dissociate from E2F. E2F is now free from inhibition and functions as a transcription factor resulting in the expression of genes that facilitate cell cycle progression. E2F also facilitates the expression of p53 as well as ARF, which functions as an MDM2 inhibitor-therefore engaging p53 signaling. E2F can also facilitate nuclear accumulation of p53 by blocking its nuclear export signal. Legend: Arrows represent activation pathways, hammers represent inhibition, faded hammers represent inhibitory pathways that are being repressed by the signal pathway. Abbreviations: ATM, ataxia-telangiectasia mutated kinase; ATR, ataxia-telangiectasia, and Rad3-related kinase; Chk1/Chk2, checkpoint kinases 1&2; MDM2, mouse double minute 2 homolog; MDMX mouse double minute X; CDK4/CDK6, cyclin-dependent kinases 4&6; Rb, retinoblastoma protein; E2F, E2 factor; ARF, alternative reading frame (p14).
Cell cycle progression indirectly leads to activation of p53, as transition through the G1 phase to S phase results in the activation of the cyclin-dependent kinases, CDK4 and Following detection of single stranded or double stranded DNA breaks, the kinases ATM and ATR become activated. Both ATM and ATR phosphorylate p53 at Ser15. ATM also phosphorylates Chk2, which phosphorylates p53 at Ser20, while ATR phosphorylates Chk1. Phosphorylation of p53 at Ser15 and Ser20 causes it to be released from its negative regulator, MDM2. Additionally, Chk2 and Chk1 both inhibit MDMX, which functions to enhance the E3 ligase activity of MDM2 and inhibit p53. With p53 now released from MDM2, it functions as a transcription factor resulting in the expression of various tumor suppressive genes. For example, Bax and NOXA are proapoptotic proteins, while p21 functions to inhibit cell cycle progression. p53 also plays a role in regulating its own activity by facilitating the expression of more MDM2, while p21 acts to inhibit further DDR signaling by blocking the activity of Chk1. (B) Cell cycle progression also leads to increased activity of p53. As cells prepare to enter the S phase of the cell cycle, the kinases CDK4 and CDK6 become activated. CDK4/CDK6 phosphorylate the cell cycle inhibitor protein Rb, which causes it to dissociate from E2F. E2F is now free from inhibition and functions as a transcription factor resulting in the expression of genes that facilitate cell cycle progression. E2F also facilitates the expression of p53 as well as ARF, which functions as an MDM2 inhibitor-therefore engaging p53 signaling. E2F can also facilitate nuclear accumulation of p53 by blocking its nuclear export signal. Legend: Arrows represent activation pathways, hammers represent inhibition, faded hammers represent inhibitory pathways that are being repressed by the signal pathway. Abbreviations: ATM, ataxia-telangiectasia mutated kinase; ATR, ataxia-telangiectasia, and Rad3-related kinase; Chk1/Chk2, checkpoint kinases 1&2; MDM2, mouse double minute 2 homolog; MDMX mouse double minute X; CDK4/CDK6, cyclin-dependent kinases 4&6; Rb, retinoblastoma protein; E2F, E2 factor; ARF, alternative reading frame (p14).
Cell cycle progression indirectly leads to activation of p53, as transition through the G1 phase to S phase results in the activation of the cyclin-dependent kinases, CDK4 and CDK6, which phosphorylate retinoblastoma protein (pRb) causing dissociation of the pRb-E2F complex. E2F, now free from inhibition, activates the transcription of multiple genes involved in the cell cycle progression from G1 phase to S phase. Interestingly, E2F also facilitates the transcription of ARF (p14 ARF in humans and p19 ARF in mice) [58], which inhibits the ubiquitin ligase activity of MDM2, thereby resulting in p53 accumulation [59]. Additionally, E2F can directly bind to p53 and block a nuclear export signal causing it to accumulate in the nucleus ( Figure 1B) [60]. p53 functioning as a transcription factor facilitates the expression of p21, which inhibits the activity of CDK2, CDK4, and CDK6, thus causing G1, G2, or S phase arrest and therefore halts the cell cycle [61].
Activated p53 oligomerizes to form tetramers, which stabilize the binding interaction with p53-response elements in DNA and enable the expression of p53-regulated genes to initiate various cellular programs [62]. p53 regulates thousands of genes involved in cell cycle arrest, senescence, apoptosis, autophagy, metabolism, and DNA repair [63]. Cell fate decisions depend on the degree of p53-inducing stress. More importantly, the decision whether to facilitate growth arrest or apoptosis in response to DNA damage appears to depend on overall p53 levels, as p53 simultaneously expresses growth arrest proteins p21 and polo-like kinase 3 (PLK3) as well as proapoptotic proteins PIG3 and APAF1. Cell cycle arrest is initiated immediately following stimulation with low p53 levels, however, prolonged high levels of p53 are required for PIG3 and APAF1 to reach the levels required to overcome the apoptotic threshold and commit cell to cell death [64,65]. In the early stages of growth arrest, p53 also facilitates the expression of DNA repair enzymes, these include the nucleotide excision repair factors p48 and XPC, as well as the base excision repair enzymes 8-oxoguanine glycosylase and MUTYH [66]. Growth arrest works cooperatively with DNA repair to first prevent the damage from propagating but then give the cell an opportunity to mend the damage. If these two functions are unsuccessful in halting the DDR, p53 levels will continue to rise and therefore cause increased expression of proapoptotic proteins until they reach critical levels required for cell death [64]. Persistent DNA damage can also result in p53-dependent activation of cellular senescence to cause irreversible cell cycle arrest without apoptosis induction. As mentioned previously, p53 facilitates the expression of p21, which functions as a CDK inhibitor, therefore halting cell cycle progression [67]. Interestingly, p21 may be partially responsible for regulating the decision to undergo senescence or apoptosis as p21 has been found to negatively regulate caspase and Jun N-terminal kinase (JNK) activity in response to DNA damage, thus preventing apoptosis induction [68]. Autophagy represents another cell fate decision regulated by p53 in response to DNA damage [69]. In the context of p53 signaling, autophagy primarily functions as a tumor suppressor by facilitating the degradation of damaged cellular components to regulate nutrient cycling under stressful metabolic conditions [70]. p53 facilitates the expression of tuberous sclerosis complex 2 (TSC2) and AMP-activated protein kinase (AMPK), which inhibit the activity of mammalian target of rapamycin (mTOR), a key negative regulator of autophagy [71]. A direct link between p53 and autophagy exists, whereby p53 induces the expression of the autophagy promoting genes, damage-regulated autophagy modulator (DRAM) [72], the autophagy-related genes, Atg4a, Atg4c, and Atg7, as well as Ulk1 and Ulk2 [73]. Autophagy exhibits indirect negative feedback on p53 signaling, presumably by reducing oxidative stress that may be inducing p53 activity [74].
Since its critical role as a tumor suppressor was uncovered [75][76][77][78][79], there has been a significant research effort being directed to understanding the complexities of p53 signaling in cancer and in normal cell biology [33]. Most of this research has been focused on understanding p53 signaling at the protein level and not on the transcriptional regulation of p53 itself [34]. However, a few transcription factors have been found to directly bind to and regulate the p53 promoter. Positive regulators such as Myc [80] and HoxA5 [81] are interesting as oncogenic Myc signaling is intricately linked to cancer progression [82] and HoxA5 appears to be an important tumor suppressor [83,84]. Understanding the regulation of p53 transcription in the context of cancer is not a straightforward topic due to the complex interplay between transcription factors, epigenetic mechanisms, and miRNAs, as well as the impact mutant p53 (mup53) has on its own regulation [34].

p53 and Cancer
p53 is the most mutated protein in cancer, with nearly 50% of all tumor cells displaying some form of p53 mutation [31], with mutation being used to characterize cancer progression [85][86][87]. Even in the absence of a detectable mutation, some estimates predict that 80% of tumors have impaired p53 function, caused by epigenetic silencing or overexpression of negative regulators such as MDM2 [88,89]. Most p53 mutations are missense mutations that typically lead to the accumulation of the mutant protein [90]. Interestingly, these missense mutations tend to cluster into 'hotspots,' with~25% of all mutations occurring at 1 of 6 codons within the p53 DNA-binding domain [31,88]. Missense mutations are classified as either DNA contact mutants (R248, R273) or p53 conformation mutants (R175, G245, R249, R282) [31,91]. The type of p53 mutation can have vastly different consequences on the cell, with these hotspot mutations being associated with mup53 oncogenic gain of function (GOF) tumor-promoting capabilities. If only one p53 allele is mutated, a dominant negative (DN) effect may occur, where the mup53 can bind and sequester wildtype (WT) p53 and block normal function [90]. Additionally, the conformation mutation mup53 R175H has been found to bind and sequester the tumor suppressors p73 and p63 [92], both of which facilitate the expression of genes similar to p53 [93]. GOF mup53 plays a pivotal role in tumorigenesis and cancer progression. Mice possessing germline GOF mutations in one p53 allele develop unique tumors with high metastatic capacity when compared to p53 +/− and p53 −/− mice [94][95][96]. GOF mup53 has been demonstrated to affect multiple oncogenic signaling pathways, upregulating various stem cell markers in colorectal cancer cells [97] and increasing human epidermal growth factor receptor 2 (HER2) expression in breast cancer cells [98].
p53 is implicated in virtually all the Hanahan and Weinberg "hallmarks of cancer." Activation of p53 in response to DNA damage, proliferation, and oxidative stress leads to cell cycle arrest and activation of apoptotic pathways [29]. p53 is intimately linked to cellular metabolism and functions to regulate nutrient availability and glycolysis [99,100]. Additionally, WT p53 functions to prevent metastasis in part due to the expression of miRNAs that regulate the epithelial to mesenchymal transition [101,102]. This could possibly explain why most advanced cancers present with a mutation in p53 [86,87]. p53 is activated in response to proliferative signals brought on by inflammation, and proinflammatory cytokines possess the ability to directly regulate p53 expression and accumulation [103,104]. The master regulator of inflammatory signals, NF-κB, is activated by various cytokines and facilitates the transcription of anti-apoptotic proteins, cytokines, and growth factors; all of which lead to enhanced tumorigenesis [105,106]. Despite the opposing functions of NF-κB and p53 on tumor progression, NF-κB has been found to directly bind to the p53 promoter and facilitate its expression, establishing a potential mechanism for negative regulation of tumorigenic inflammatory signals [107]. Clearly, the importance of p53 dysfunction in tumorigenesis cannot be overlooked. As such, a handful of novel treatments are currently under development with the intention of targeting p53 and p53 signaling.

Restoring Endogenous p53 Signaling
Tumor suppressors evolved to prevent malignant transformation, with oncogenic stimuli leading to their activation. As such, cancer cells are constantly activating the pathways that evolved to trigger p53 signaling. However, with p53 function lost, they exist in a state of perpetual p53 activation. This led to the hypothesis that if normal p53 function were to be restored, cancer cells would be primed to arrest their growth and undergo apoptosis. Support for this method of treatment was strengthened following reports that replacement of WT p53 into p53-null cell lines resulted in arrested growth, increased apoptosis induction, and caused tumor regression in vivo [108,109]. p53 gene therapy was born in 1994 after Fujiwara, T. et al. [110] demonstrated that intratracheal administration of a retroviral vector expressing WT p53 suppressed the growth of established lung tumors in vivo. A small clinical trial was conducted in nine non-small cell lung cancer (NSCLC) patients and found that bronchoscopic administration of the retroviral-WT p53-vector suppressed tumor growth and increased the presence of the TUNEL apoptosis marker at follow-up [111]. A major limitation of this study was the use of the retroviral vector, which has significant safety concerns, mainly, the risk of insertional mutagenesis [112]. To overcome the safety concerns, studies examining the use of a safer adenoviral WT p53 expression vector (Ad-p53) were conducted in multiple cancer subtypes and demonstrated similar effects-apoptosis induction and arrested growth in vitro and in vivo ( Figure 2A), with genotoxic combination therapy resulting in a synergistic effect [113][114][115]. Clinical trials with this treatment have produced mixed results. Intratumoral injection of Ad-p53 in NSCLC patients resulted in a clinical response in 72% of patients [116], and intratumoral injection into patients with head and neck cancer resulted in a clinical response in 47% of patients (Table 1) [117]. These response rates could be strengthened by combining treatment with radiation therapy [118,119]. Because of the encouraging response in patients with head and neck cancer, in 2003, China granted regulatory approval for the Ad-p53 treatment, making it the first gene therapy product ever approved [120]. Thousands of cancer patients have received some form of restorative p53 gene therapy with varying success rates. It appears that restoring endogenous p53 activity via genetic means represents a safe means for controlling tumor growth, however, some form of genotoxic combination therapy seems to be necessary to produce meaningful response rates. Unfortunately, the addition of Ad-p53 often does not produce a survival benefit when compared to the genotoxic treatment alone, which raises concerns about its potential use in the future [121]. The variation in response to treatment could potentially be due to the presence of different p53 mutations. Wang et al. [122] found that restoring endogenous p53 function in mice harboring the p53 R172H mutation (analogous to the R175H hot-spot mutation in humans) did not lead to tumor regression as it does in p53-null mice. Mutant p53 R172H was demonstrated to bind to the restored WT p53 and exert a dominant negative effect that prevented the expression of pro-apoptotic proteins but not proteins involved in cell cycle arrest [123]. Cancer cells with WT p53 would also be largely unaffected by p53 restoration, as the p53 signaling network is still relatively intact [124,125]. Unfortunately, it appears that the intricacies surrounding p53 inactivation are too complex to be overcome by simply replacing WT p53. However, with the recent success of immunotherapies in controlling tumor growth, there has been slightly renewed interest in Ad-p53 treatments as it pertains to combination therapy. Adenoviruses are highly immunogenic [126,127], and though this is regarded as a negative aspect of treatment, as humoral immune responses to the viral vector may hinder repeat dosing, immune stimulation may be advantageous for cancer therapy. Dendritic cells have been demonstrated to uptake apoptotic bodies and illicit potent anti-tumor responses [128,129]. A potential synergy exists between the immune stimulation of the adenoviral vector and the generation of immunogenic apoptotic bodies caused by the WT p53 gene. Ad-p53 treatment has been demonstrated to influence various immune signaling pathways, resulting in an increased CD8 T cell response that correlates with enhanced immune checkpoint inhibitor efficacy [130,131]. Clinical trials are being conducted to evaluate the feasibility of combining Ad-p53 therapy with anti-PD-1 or anti-PD-L1 checkpoint inhibitors [NCT03544723]. Compounds that stabilize the mutant protein so that the WT conformation of the DNA binding domain can be restored are able to facilitate the expression of WT p53 target genes resulting in growth arrest and apoptosis. These include PRIMA-1/APR-246, which following conversion into methylene quinuclidinone (MQ) can covalently attach to cysteine residues, which stabilizes the binding of mup53 to the WT proteins DNA response elements. Additionally, mutations can create novel binding domains on the mup53 protein. The Y220C mutation generates a binding pocket on mup53 that when bound by the small molecule PK7088 stabilizes the mutant protein in its WT conformation enabling binding to DNA response elements. These compounds enable restoration of WT p53 transcriptional functions resulting in cell cycle arrest and apoptosis. (D) Heat shock protein 90 (Hsp90) functions to stabilize mup53, preventing its degradation. Addition of the Hsp90 inhibitor, ganetespib, blocks this interaction and facilitates degradation of mup53. This can also be achieved by inhibiting histone deacetylase activity with suberoylanilide hydroxamic acid (SAHA), which causes Hsp90 to remain acetylated, which blocks its chaperone activity. Legend: Arrows represent activation pathways, hammers represent inhibition, and faded hammers represent inhibitory pathways that are being repressed by the signal pathway.
Despite the limitations surrounding restorative p53 gene therapy, it does indicate Compounds that stabilize the mutant protein so that the WT conformation of the DNA binding domain can be restored are able to facilitate the expression of WT p53 target genes resulting in growth arrest and apoptosis. These include PRIMA-1/APR-246, which following conversion into methylene quinuclidinone (MQ) can covalently attach to cysteine residues, which stabilizes the binding of mup53 to the WT proteins DNA response elements. Additionally, mutations can create novel binding domains on the mup53 protein.
The Y220C mutation generates a binding pocket on mup53 that when bound by the small molecule PK7088 stabilizes the mutant protein in its WT conformation enabling binding to DNA response elements. These compounds enable restoration of WT p53 transcriptional functions resulting in cell cycle arrest and apoptosis. (D) Heat shock protein 90 (Hsp90) functions to stabilize mup53, preventing its degradation. Addition of the Hsp90 inhibitor, ganetespib, blocks this interaction and facilitates degradation of mup53. This can also be achieved by inhibiting histone deacetylase activity with suberoylanilide hydroxamic acid (SAHA), which causes Hsp90 to remain acetylated, which blocks its chaperone activity. Legend: Arrows represent activation pathways, hammers represent inhibition, and faded hammers represent inhibitory pathways that are being repressed by the signal pathway. Despite the limitations surrounding restorative p53 gene therapy, it does indicate that there is potential that p53 reactivation could result in more meaningful clinical responses if a targeted approach is taken. In the case of cancer cells with WT p53 that present with overexpression of a negative regulator, there are several treatment modalities being explored for each specific indication. Perhaps the most promising and extensively studied are inhibitors of the MDM2-p53 interaction. The discovery that the small molecule nutlin-3a could bind to MDM2 and release p53 from its inhibition resulting in apoptosis and hindered tumor growth in vivo [132] spurred the development of multiple small molecule drugs that can also inhibit this interaction [133][134][135][136][137]. In patients with liposarcoma, the MDM2 antagonist, RG7112, was able to facilitate p53 accumulation and control tumor growth ( Figure 2B and Table 1). However, 40% of patients presented with serious adverse events, primarily neutropenia and thrombocytopenia [138], presumably due to increased apoptosis in hematopoietic progenitors in which the MDM2-p53 axis plays an essential role in their differentiation [139]. In addition to safety concerns, MDM2 inhibitors are also subject to resistance mechanisms such as overexpression of efflux pumps and mutation to p53, which further hinders their clinical success [140]. Of the MDM2 antagonists currently in clinical trials, most compounds display similar efficacy with newer generation drugs displaying slightly improved safety profiles [138]. Due to the early clinical stages of these compounds, few have evaluated clinical responses to treatment and overall survival. One compound, MK-8242, tested in patients with liposarcoma resulted in a median progressionfree survival of 237 days (Table 1). Unfortunately, due to the small cohort size, a proper control was not included in this study and the authors compared their median progressionfree survival to that of another study that resulted in a median progression-free survival of 126 days in patients with liposarcoma treated with a CDK4 inhibitor [141]. Evidently, larger clinical studies with better controls are necessary to determine the validity of this treatment approach.
The variability in therapeutic responses to MDM2 antagonists can be partially explained due to the engagement of multiple p53-dependent pathways apart from apoptosis. Efeyan, A., et al. [142] demonstrated that nutlin-3a induced cell cycle arrest and senescence in fibrosarcoma cells. The authors also showed that senescence was induced in primary fibroblasts as well. This is important as senescent cells possess novel tumor-promoting capabilities due to the acquisition of the senescence-associated secretory phenotype (SASP)-a collection of soluble and insoluble markers that influence the tumor microenvironment with factors that are proinflammatory (cytokines/chemokines), matrix remodeling (MMP), or promote cell growth (growth factors) [143]. Nutlin-3a has also been demonstrated to facilitate p53-dependent autophagy by stimulating the expression of AMPK and DRAM [144]. Though autophagy possesses tumor suppressive functions in healthy cells, in cancer cells autophagy is tumorigenic by enabling cells to endure stressful conditions such as nutrient deprivation and oxidative damage [70]. It is possible that MDM2 antagonists produce poor clinical responses due to these indirect mechanisms of action. Interestingly, combination treatment with nutlin-3a and a Wip1 inhibitor significantly increases apoptosis induction when compared to monotherapy with either compound [145]. As nutlin-3a is a reversible MDM2 antagonist [146], Wip1 inhibition may prevent negative regulation by MDM2 at lower drug concentrations and enable prolonged p53 signaling, which is required to express sufficient amounts of proapoptotic proteins to overcome the apoptotic threshold [64,65].
Targeting the MDMX-p53 interaction represents an alternative strategy for restoring p53 activity as it may result in fewer side effects than MDM2 antagonists. Hematopoietic stem cells recover from short-term p53 restoration in mice lacking MDMX, presumably due to functional MDM2 retaining the ability to regulate p53 activity [147]. MDMX targeted treatments are still in preclinical stages of development, though a handful of promising compounds have been identified [148]. Small molecule antagonists targeting the p53 binding site of MDMX, such as SJ-172550 and CTX1, typically lead to accumulation of the p53 protein and expression of p53-regulated genes resulting in similar effects to MDM2 antagonists: growth arrest and apoptosis ( Figure 2B) [148]. Though reducing toxicity associated with MDM2 inhibition was the primary driving force behind the development of MDMX antagonists, these compounds display potent synergism when combined [149,150]. One of the most exciting MDMX inhibitors is an antisense oligonucleotide (ASO) directed against full-length MDMX. Dewaele et al. [151] discovered that adult tissue lacked fulllength MDMX primarily due to inefficient splicing of Mdmx mRNA at exon 6, which does not occur in cancer cells. Therefore, they designed a cancer-selective MDMX-ASO that targets the exon-intron boundary of exon 6 and leads to a reduction in MDMX protein levels, impaired in vitro growth, and increased apoptosis induction as well as attenuated the growth of both melanoma and large B cell lymphoma patient-derived xenografts ( Figure 2B). Clinical data will be necessary to indicate if MDMX inhibitors provide any advantage over targeting MDM2. Unfortunately, it is likely that MDMX inhibitors will be subject to similar resistance mechanisms as MDM2 inhibitors and may have unique safety concerns yet to be uncovered.
Cancer cells possessing mutant forms of p53 can be targeted using compounds that reverse the conformational changes acquired by mutation, thus restoring p53 function. The most clinically advanced compound, PRIMA-1, can preferentially inhibit the growth of cancer cell lines with the R273H DNA contact mutant or the R175H conformation mutant [152]. Upon entry into cancer cells, PRIMA-1 is converted to methylene quinuclidinone (MQ) that covalently binds to the p53 core via alkylation of cysteine residues. This may prevent the formation of disulfide bonds that fix mup53 in an unfolded confirmation [153]. Additionally, MQ alkylation was recently demonstrated to stabilize mup53 binding to DNA, therefore enabling expression of WT p53 genes ( Figure 2C) [154]. A recent phase II clinical trial examined the efficacy of the PRIMA-1 derivative, APR-246 (eprenetapopt), in combination with the first-line therapy, azacytidine, in patients with acute myeloid leukemia or myelodysplastic syndromes with at least one p53 mutation (Table 1). APR-246 was well tolerated and displayed favorable response rates with a potentially increased survival when compared to previously published data on monotherapy with azacytidine alone [155]. Unfortunately, a phase III study with APR-246 in this patient population failed to display significant improvements in complete remission rate when compared to monotherapy with azacytidine (33.3% vs. 22.4%, respectively; p = 0.13) [156]. Neither of these studies were specific in their p53 mutation inclusion criteria, and therefore it is possible that targeting specific p53 mutants (R273 and R175) may improve treatment responses. Unfortunately, as the inclusion criteria for clinical studies becomes more specific for each p53 mutation, our ability to perform well-powered trials is reduced, which will further delay the entry of these compounds into the clinic.
Other mup53-selective compounds are in preclinical development. PK7088 targets cancer cells harboring a Y220C p53 mutation, the ninth most common p53 mutation. PK7088 binds mup53 in a pocket created by the Y220C mutation and slows its denaturation [157]. This results in a restoration of the WT p53 conformation, nuclear accumulation of p53, and the expression of p53-dependent genes, which facilitate cell cycle arrest and apoptosis induction ( Figure 2C) [158]. Other indirect methods of restoring the WT p53 conformation have also been discovered. The small molecule chetomin was identified in a high-throughput screen and preferentially facilitated p53 gene expression and apoptosis induction in cell lines and tumors containing the R175H mutation. Interestingly, chetomin does not bind mup53 and instead increases the binding affinity of Hsp40 for mup53 R175H . Hsp40 can then bind to mup53 R175H and stabilize the unfolded protein, restoring its function [159]. The promising preclinical results from these compounds are encouraging and indicate that we may soon have multiple p53 restorative therapeutics entering the clinic, provided they produce meaningful clinical responses [160].

Inhibiting Oncogenic Gain of Function p53 Signaling
As if understanding endogenous p53 signaling was not complicated enough, even understanding p53 mutation is not a straightforward topic. Contrary to most tumor suppressor genes that undergo a loss-of-function following mutation, p53 can lose its endogenous function while simultaneously gaining novel tumor-promoting capabilities. DNA contact mutants and conformational mutants alter the DNA binding domain, which can alter the DNA elements recognized by p53 and result in the expression of tumorigenic genes involved in apoptosis resistance, proliferation, angiogenesis, and metastasis [90]. Additionally, mup53 may associate with other oncogenic proteins and facilitate activation of their tumorigenic pathways [161].
In some cases, cancer cells possessing a GOF mutation in p53 are dependent on it for their survival, and knockdown of mup53 abrogates tumorigenic potential in vitro and in vivo [162,163]. In a proof-of-concept experiment, Alexandrova et al. [164] demonstrated that genetically ablating mup53 R248Q slowed tumor growth and prolonged survival. The authors also demonstrated that Hsp90 stabilized mup53, and adding the Hsp90 inhibitor, ganetespib, increased median survival in tumor-bearing mice possessing either the R248Q or the R172H (R175H in humans) p53 mutants by 59% and 48%, respectively, with no effect on p53-null mice. Ganetespib-mediated inhibition of Hsp90 results in selective degradation of mup53 that is accompanied by apoptosis induction, thus providing evidence that elimination of GOF mup53 represents a viable treatment option ( Figure 2D). Histone deacetylase inhibitors (HDACi) represent an alternative strategy to indirectly enhance mup53 degradation. The HDACi, suberoylanilide hydroxamic acid (SAHA), was found to inhibit HDAC6 and as a result, Hsp90 remains acetylated at K294, which inhibits its chaperone activity. In turn, mup53 is released from Hsp90 and degraded resulting in apoptosis induction selectively in cancer cells harboring mup53 (IC50 values are >100-fold higher in cells with WT p53) ( Figure 2D) [164,165]. Unfortunately, clinical trials with both ganetespib and SAHA have not been very promising (Table 1). Patients in these trials were not selected based on their p53 status and therefore it is entirely possible that the efficacy of these two compounds can be improved by altering the inclusion criteria [166,167].
Perhaps one of the more obvious strategies to inhibit GOF mup53 would be via knockdown with siRNA. However, there is potential that siRNA may still recognize WT p53 in healthy tissue leading to off-target effects. Martinez et al. [168] demonstrated that siRNAs can be designed to preferentially recognize single nucleotide changes in p53 mRNA by ensuring that the ninth nucleotide in the siRNA contained the mutated base pair. Recently, Ubby et al. [169] generated a panel of siRNAs targeting each GOF mup53 and used lipid nanoparticles to deliver them to subcutaneous tumors, where they abrogated tumor growth. Though clearly still in its infancy, the preclinical data have indicated that eliminating mup53 can have potent anti-tumor effects.

Exploiting Dysfunctional p53 Signaling
Rather than trying to correct the problems brought on by p53 mutation, an alternative approach is to capitalize on it and utilize the dysfunction for treatment options. The earliest example of this is ONYX-015, a replication-competent adenovirus that selectively infects and lyses cancer cells lacking functional p53. By removing the E1B gene that normally functions to inactivate p53 in infected cells [170], the adenovirus would only be able to replicate in cells lacking p53 as cells with a functional copy of p53 would block viral replication by arresting their growth or undergoing apoptosis [171]. However, subsequent reports found that viral replication was not restricted to cells lacking p53, instead, it showed a slight preference for replicating in cancer cells with mup53 [172][173][174][175]. Unfortunately, ONYX-015 failed clinical trials as a monotherapy, because although it had a favorable safety profile, it demonstrated minimal antitumor efficacy [176,177]. However, ONYX-015 was able to sensitize cancer cells to chemotherapy [177][178][179], which reinvigorated interest in the treatment and eventually led to regulatory approval of a similar oncolytic virus (H101) in China [180].
GOF mutations in p53 can result in unique gene expression signatures relative to WT p53 possessing counterparts [90]. Because of this, mup53 may facilitate the expression of druggable genes offering a novel approach for selectively targeting cancer. For example, mup53 can bind and sequester p73, which enables NF-Y to be released from p73 and facilitate the expression of platelet-derived growth factor receptor β (PDGFRβ). Consequentially, PDGFRβ then stimulates invasion and metastasis. However, this also enables the process to be blocked by adding PDGFRβ inhibitors such as imatinib and crenolanib [181]. Mutant p53 has also been demonstrated to directly bind to NF-Y and recruit p300 in response to DNA damage. This results in the expression of genes involved in cell cycle progression that would normally be inhibited in response to DNA damage [182]. P300 is a histone acetyltransferase, therefore making it a druggable target. In fact, a number of p300 (and CREB-binding protein) inhibitors are currently under development and clinical investigation for the treatment of multiple cancer subtypes [183][184][185]. Mutation of p53 can also lead to substantial immune dysfunction, resulting in the secretion of tumorigenic cytokines that support cell growth and cancer progression [186]. Missense mutations in p53 have recently been demonstrated to increase the expression of programmed cell death ligand 1 (PD-L1) and can therefore be utilized to predict treatment responses to the anti-PD-L1 immune checkpoint inhibitor, nivolumab [187][188][189][190].
An alternative approach for targeting p53 dysfunction that is of particular interest to our group is targeting p53 transcriptional activation [24,191]. Typically, p53 activation is thought of as an accumulation of p53 protein due to decreased turnover by MDM2, without activation of p53 transcription [30,50,51,59]. Unsurprisingly, p53 transcription is also induced by oncogenic stimuli. For example, DNA-damaging chemotherapeutic agents and ionizing radiation have been found to increase p53 transcription [43,44,47]. The DDR pathway activates Chk1 and Chk2 [48], both of which lead to the accumulation of p73 and E2F1 [192]. These two proteins can directly bind to the p53 promoter and facilitate p53 transcription [43,44]. p53 then negatively regulates DDR signaling by inducing the transcription of p21, which facilitates the downregulation of Chk1 ( Figure 1A) [193].
Various oncogenes have been demonstrated to directly bind to the p53 promoter and facilitate its expression, for example, c-Myc [80] and NF-κB [107]. In most cases, p53 acts as a negative regulator of these oncogenes, inhibiting their activity as well as their transcriptional activation of p53 following induction [194,195]. As p53 is a master regulator of its own transcription, it is enticing to speculate about the impact that a loss of p53 function would have on the transcriptional activation state of p53. Perhaps one of the most interesting possibilities is cooperative regulation by c-Myc and RAS; both of which are frequently mutated and activated in many cancers [82,196,197]. RAS and c-Myc are intricately linked, with aberrant RAS signaling resulting in increased c-Myc protein stability [198]. c-Myc directly binds to and positively regulates the p53 promoter [80]. Conversely, p53 can bind to the c-Myc promoter and inhibit its expression [199]. Additionally, the p53 target, miR-145, can silence c-Myc [194]. It is possible that a loss of p53 can directly increase c-Myc activity, therefore resulting in upregulation of p53 promoter activity. This has been partially demonstrated before, as p53 loss was found to increase c-Myc expression and the expression of c-Myc regulated genes in breast cancer stem cells [200]. RAS and p53 also share a link, but it is far less direct than connection between p53 and c-Myc. Wild-type p53 has been demonstrated to facilitate expression of B-cell translocation gene 2 (BTG2), which functions as an inhibitor of aberrantly activate RAS [201]. Therefore, a loss of p53 could result in BTG2 loss, which causes RAS activity to increase, leading to increased stability of c-Myc and transcriptional activation of c-Myc regulated genes, including p53. RAS has also been demonstrated to activate NF-κB, which in addition to increasing c-Myc expres-sion [198] also directly binds to the p53 promoter and induces its expression ( Figure 3A). Oncogenic signaling through these proteins would remain active in the event of p53 loss, resulting in persistent activation of the p53 promoter. Additionally, p53 loss would mean that the cell loses the ability to halt the DDR pathway, causing p73 and E2F1 to activate the p53 promoter as well. We hypothesize that loss of p53 in cancer cells would lead to a non-productive upregulation of p53 transcription due to a loss of p53-mediated negative feedback ( Figure 3B). The examples above were just a handful of potential networks that can become impaired following p53 loss [35,202]. The challenge then becomes developing a treatment strategy that can capitalize on this dysfunctional transcription network and generate meaningful clinical responses in patients. With the recent successes of novel genetic medicines, we may soon be able to modulate p53 signaling at a transcriptional level. RAS signaling leads to the activation of NF-κB as well as c-Myc, both of which directly bind to the p53 promoter and facilitate its expression. NF-kB also binds to the c-Myc promoter and induces its expression. The p53 protein then regulates these pathways, by stimulating the production of BTG2 and miRNA-145, which inhibit RAS and c-Myc function, respectively. Additionally, p53 protein can directly bind to the c-Myc promoter and repress its transcription. (B) Proposed effect that p53 loss can have on p53 promoter activity. Loss of p53 function can result in persistent RAS signaling through NF-κB and c-Myc, both of which will directly bind to and stimulate the p53 promoter. In cells with constitutively active mutant RAS and/or c-Myc, this pathway may be active even in the absence of p53 mutation/silencing. Legend: Arrows represent activation pathways (purple for NF-κB, pink for c-Myc, black for RAS and p53, green for phosphorylation event by ERK) and hammers represent inhibition.

Future Directions
Rather than targeting specific aspects of p53 inactivation, we propose taking a multifaceted approach to target the p53 network. The Hsp90 inhibitors mentioned above give a great example of this. SAHA and ganetespib facilitate degradation of mup53 protein that abrogates tumorigenic potential in animal models of cancer but failed to perform in the clinic [164][165][166][167]. When combined with MDM2 and MDMX inhibitors, the efficacy of treatment was enhanced [149,150], indicating that combining modalities may offer some treatment benefits as long as there is no overlapping toxicity (Table 1). Therefore, it may be possible to improve treatment responses by combining these Hsp90 inhibitors with therapeutics such as Ad-p53 [113][114][115], which would activate WT p53 signaling while simultaneously blocking the effects of mup53-preventing a dominant negative effect from occurring-while also thwarting GOF mup53 signaling pathways [122]. Of course, this is purely speculative and proper experimentation is required to determine the feasibility of multimodal p53 targeting.
An alternative strategy for targeting the p53 network relies on the use of gene therapy. The WT p53 adenovirus expression vectors have arguably demonstrated the most clinical success of any p53 targeted therapy thus far, but they still face many limitations, primarily low treatment responses and cancer cell adaptation to the WT protein [203]. An additional limitation relates to the use of viral vectors as a gene delivery platform. Adaptive immune responses to the viral vector itself prevents repeat administration and therefore also long-term tumor control [204][205][206][207][208][209][210]. Non-viral gene delivery platforms, such as lipid nanoparticles, represent a non-immunogenic alternative to viral platforms and have demonstrated remarkable success in the delivery of nucleic acids [211][212][213][214][215]. Given the poor clinical responses to current p53 targeting approaches, it is reasonable to assume that tumors can overcome restoration of p53 activity. Therefore, it is necessary to examine the impact that p53 loss can have on the entire p53 network. Because p53 has such an integral role in regulating its own expression, loss of p53 function may result in an upregulation in the activity of the p53 promoter due to loss of negative feedback ( Figure 3B) [35,202]. A simple approach to capitalize on the transcriptional dysfunction of the p53 network would be to design a gene therapy vector where the expression of a therapeutic protein is controlled by the p53 promoter and deliver this vector using a non-viral nucleic acid delivery platform [191]. The encoded therapeutic protein could directly target tumor cells, or indirectly target tumor cells by targeting the immune system or the cell cycle. Direct tumor-targeting therapies using proapoptotic proteins (Bax, NOXA, PIG3, APAF1) [216] or tumor suppressors (miR-7, 15-PGDH) are in preclinical or clinical studies [217]. Research into targeting the immune system using immunogenic proteins (cytokines, co-stimulatory proteins, tumor neoantigens) as a therapy for solid tumor cancers has rapidly increased recently which has translated into significant progress in the field of cancer vaccines [218]. Targeting the cell cycle using cytostatic proteins also shows promise as a cancer therapy strategy since tumor cells show aberrant activity from numerous cell cycle proteins [219]. Although for the treatment of solid tumor cancers, therapies that utilize a combination of these therapeutic proteins may be the most effective strategy due to synergistic anticancer activity, overcoming clonal heterogeneity, reducing regimen toxicity and drug resistance, and removing adaptive resistance by the tumor cells [220]. Combination therapies also allow treatments to be modified to maximize clinical responses.

Conclusions
Despite the complexities of p53 signaling, precision treatment using p53 remains a rapidly growing field in cancer therapy because so many aspects of tumorigenesis are influenced by p53. Unfortunately, despite promising preclinical data, most therapeutic strategies directed toward the modulation of p53 signaling have had low to moderate clinical success. There are a few compounds that show promise though, such as SAHA and ganetespib (Table 1), that were examined without considering p53 status in the inclusion criteria, and therefore treatment responses can likely be improved [166,167]. As we understand how each form of p53 inactivation impacts tumor biology, our ability to identify patient populations that can benefit from treatment grows larger. Precision medicine is undoubtedly the future of cancer treatment [221][222][223], and as such, molecular targets that function in nearly all subtypes of cancer will certainly carve out an essential role in this new era of treatment.

Conflicts of Interest:
The authors declare no conflict of interest.