Mitochondrially Targeted p53-Bad* Fusion Gene Therapy Promotes Apoptosis in Hepatocellular Carcinoma by Pan-Bcl-2 Inhibition

Abstract

Presenting a considerable disease burden and global health threat, hepatocellular carcinoma (HCC) is in desperate need of potent and effective therapies. p53-Bad* is designed to provide mitochondrial targeting, enhance binding interactions with antiapoptotic Bcl-2 family members, and improve apoptotic activity in various cancers. While we have shown that fusion of p53 and Bad* improves mitochondrial localization and increases apoptosis in HCC, critically, the mechanism of action and nature of this heightened apoptotic function have not been delineated. Here, the functional activity and apoptotic mechanism of action of our p53-Bad* construct was explored, demonstrating that fusion of pro-apoptotic Bad* to p53 does indeed enhance interactions with antiapoptotic Bcl-2 family members Mcl-1, Bcl-2, and Bcl-xL relative to p53-WT. We confirm that p53-Bad* acts as a pro-apoptotic agent specifically at the mitochondria by inhibition and blockade of downstream Bak/Bax oligomerization events, rescuing cells from p53-Bad*-induced cytotoxicity and apoptosis. Using mutant p53-Bad* variants designed to disrupt binding interactions with antiapoptotic targets, a direct relationship between functional binding interactions and the apoptotic activity of p53-Bad* is established. These findings present strong evidence for the use of p53-Bad* as a pan-Bcl-2 inhibitor and synergistic pro-apoptotic agent in cancers with upregulation of multiple antiapoptotic and prognostic markers.

Graphical Abstract

INTRODUCTION

As of 2022, primary liver cancer is positioned as the third highest cause of cancer-related deaths and the second highest cause of premature cancer-related deaths globally.¹ Of these liver cancer cases, hepatocellular carcinoma (HCC) comprises approximately 75–85%, representing a considerable disease burden and health threat globally.^1,2 Approximately 758,000 people succumb to HCC each year, and 865,000 more are diagnosed globally. With rates of incidence and mortality in such close agreement, the especially poor prognosis of this disease is apparent.³ Despite a reduction in HCC in previously endemic countries due to improved rates of HBV vaccination, incident rates continue to rise or have plateaued in many formerly low-risk regions including the United States, New Zealand, Australia, and most European countries,^1,4 in part due to the increasing number of cases associated with alcohol-associated liver disease (ALD) and metabolic dysfunction-associated steatotic liver disease (MASLD).^3–9 Until recently, the standard of care and front-line therapy for HCC consisted of tyrosine kinase inhibitors (TKIs) including sorafenib and lenvatinib, which have now largely been superseded by combination immunotherapy with atezolizumab and bevacizumab due to severe adverse events.^10,11 However, approximately 15–20% of patients will remain on TKI treatment due to underlying autoimmune disease,^10,11 and incident rates of grade 3 adverse events, including fatalities, are comparable or higher relative to sorafenib (0.8% of sorafenib vs 4.6% atezolizumab/bevacizumab).¹² The enduring and desperate need for safer and more tolerable treatments for advanced HCC is thus apparent.

Tumor protein 53, or p53, is the most frequently mutated gene in human cancer, with mutations of various degrees of significance observed in approximately 50% of human primary tumors,¹³ varying from low (10% in hematopoietic malignancies) to near total (nearly 100% in high-grade serous ovarian carcinoma) rates of incidence.^14,15 In HCC specifically, p53 is mutated in more than 45% of HBV related cases, 13% of HCV related cases, and 50% of aflatoxin B₁-induced HCC cases,^16,17 suggesting potential for p53 gene therapy treatment strategies. Moreover, HCC tumors are uniquely vascularized by the hepatic artery, whereas the healthy liver is dominantly vascularized by the portal vein—providing a natural mechanism for tumor specificity and targeted gene delivery via intraarterial injection in addition to intratumoral injection.^18,19 However, wild-type p53 gene therapy has met limited success to date, with recombinant adenoviral TP53 (rAd-p53, Gendicine; China Shenzhen SiBiono GeneTech Co., Ltd.) yielding variable outcomes between different cancers.²⁰ While Gendicine has been approved for the treatment of head and neck squamous cell carcinoma in China since 2003, p53 gene therapy has thus far failed to establish itself in the Western commercial market or for use in HCC. The limited success of p53 gene therapy is generally attributed to (i) dominant-negative inhibition by mutant p53 and p63/p73 splice variants, (ii) mutational evasion and/or resistance elsewhere in p53 pathways, and (iii) immune reactions to early generation adenovirus following repeat administration of the rAd-p53 injection.^21,22

Wild-type p53 gene therapies such as Gendicine rely primarily upon the nuclear programs of p53, where p53 tetramerizes and then binds to DNA and acts as a transcription factor to activate hundreds of genes. In response to cellular stress signals, nuclear p53 classically regulates cellular signaling, metabolism, and life cycle in an attempt to repair damaged DNA and prevent its propagation, ultimately initiating cell death via the extrinsic apoptotic pathway if necessary.^23–25 In addition to its nuclear activity, p53 regulates apoptosis in a transcriptionally independent way at the mitochondria through the intrinsic apoptotic pathway.^26–29 Upon death signal stimulation, (multi)-monoubiquitinated p53 is liberated from MDM2 and transported to the mitochondria³⁰ where, as a monomer, it binds and neutralizes B-cell lymphoma 2 (Bcl-2) family antiapoptotic factors Bcl-2, Bcl-xL, Bcl-w, and Mcl-1. Additionally, it activates pro-apoptotic effectors Bak and Bax^25,31 (Figure 1). Notably, this mechanism bypasses inactivation by the dominant-negative effect and is insensitive to genetic aberrations of the tumor environment beyond p53 itself. However, the mitochondrial activity of p53 is transient and has not proven strong enough for therapeutic application, likely influenced by its lack of a dedicated mitochondrial targeting signal (MTS) and reliance upon complex signaling cascades for transport, although the exact mechanism of translocation is not well understood.^32–34 We have previously shown that these limitations can be overcome by fusion of p53 to a mitochondrial targeting signal (p53-MTS) to enhance mitochondrial localization and induce apoptosis in breast and ovarian cancers.^28,35 However, the apoptotic effect of p53-MTS is relatively short-lived (unpublished data), necessitating the pursuit of longer acting p53 constructs. Toward this end, we explored fusion of p53 to a Bcl-2 associated agonist of cell death (Bad).

Figure 1.

p53-Bad* and the intrinsic apoptotic pathway. p53-Bad* is proposed to induce apoptosis in a transcription-independent way at the mitochondria via (A) direct activation of Bak/Bax, (B) binding and neutralization of antiapoptotic Bcl-2 family proteins such as Bcl-xL, Bcl-2, and Mcl-1, and (C) liberation of pro-apoptotic BH3-only protein tBID (which can also activate Bak/Bax), resulting in (D) mitochondrial outer membrane permeabilization (MOMP) via Bak/Bax homo-oligomer pore formation and release of cytochrome C. (E) Cytochrome C binds to cytosolic APAF1, facilitating apoptosome formation and (F) activation of caspases, which dismantle the cell. Created in BioRender. Lim, C. (2025) https://BioRender.com/h55b699.

The Bcl-2 family is composed of two subfamilies: the antiapoptotic factors (Bcl-2, Bcl-xL, Bcl-w, Bfl-1, and Mcl-1) and the pro-apoptotic factors, of which there are activators (Bid, Bim, and weaker activators Puma and Noxa), sensitizers (Bad, Bik, Bmf, Hrk, Noxa, and Puma), and effectors (Bak, Bax, and Bok).³¹ The BH3-only pro-apoptotic proteins are of particular interest as gene therapy agents, due to their abilities to target and localize to the mitochondria as well as inhibit antiapoptotic factors. Of these proteins, the transport and regulation of Bad is well characterized, with phosphorylation of Ser-112 and Ser-136 residues resulting in binding and inactivation by cytoplasmic 14–3–3 scaffold protein.^36–41 Toward the objective of designing a constitutively mitochondrial variant of Bad, we previously performed compound S112A/S136A mutations on Bad to produce Bad*, an unphosphorylated variant capable of targeting and binding the mitochondria outer membrane (MOM) through C-terminal lipid binding motifs (47 aa), where it can thereby exert its apoptotic potential.⁴²

Iterating upon our previous fusion construct p53-MTS,^{19,28,35,40,42} we re-engineered p53 as a fusion with Bad* (p53-Bad*) to further enhance apoptotic activity while also retaining mitochondrial targeting. Indeed, fusion of p53 with Bad* demonstrably improves mitochondrial localization and enhances apoptosis in both HCC and ovarian cancers relative to wild-type p53.^19,42 However, we have yet to demonstrate the mechanism for this enhanced activity. p53-Bad* is designed toward the improvement of two functions: (i) mitochondrial targeting and localization and (ii) binding and neutralization of antiapoptotic factors. With regard to the first function, we have already demonstrated superior mitochondrial localization,^19,42 although it remains unclear whether mitochondrial targeting alone is sufficient to explain the potent apoptotic activity of p53-Bad* or if this activity is indeed influenced by improvements to the second function. Fusion to Bad* is hypothesized to enhance neutralization of antiapoptotic factors as a potent inhibitor of antiapoptotic Bcl-2, Bcl-xL, and Bcl-w,³⁶ where Bad* is expected to act in concert with p53, which is itself a known inhibitor of antiapoptotic Bcl-2, Bcl-xL, and Bcl-w in addition to Mcl-1.^36,41,43 Mitochondrial p53 is also believed to liberate tBID from sequestration^44–49 and directly activate Bak/Bax^50,51 (Figure 1A–C), although p53-Bad* is not designed explicitly toward the improvement of these latter functions. All three functions are believed to result in consequent homo-oligomerization of Bak/Bax (Figure 1D), the subsequent release of cytochrome C and assemblage of APAF1 (Figure 1E), and cell death (Figure 1F).^52–60 In this way, inhibition of Bak/Bax oligomerization functions as a checkpoint against apoptosis, and p53-Bad* should be rendered ineffectual via any of its proposed mitochondrial mechanisms in the event that liberated, monomeric Bak/Bax is unable to oligomerize.

Mitochondrial p53 is proposed to bind and inhibit Mcl-1, Bcl-2, and Bcl-xL through its transcription activation (TAD) and DNA-binding (DBD) domains, while Bad/Bad* is also believed to bind and inhibit Bcl-2 and Bcl-xL through its BH3 domain.^{33,50,61–69} If the greatly enhanced apoptotic activity of p53-Bad* relative to p53-WT is indeed a consequence of enhanced binding interactions with antiapoptotic factors,^19,42 we hypothesize that, by introducing point mutations into p53-Bad* targeted against these interactions, a correlative reduction in apoptosis and protein binding should be observed. To identify critical residues for targeted mutagenesis, known structural and function profiles of p53, its common mutants in cancer, and Bad in complex with Mcl-1/Bcl-2/Bcl-xL were reviewed.^61–72

With respect to p53, the Mcl-1 and Bcl-2/Bcl-xL binding profiles are distinct; while both Mcl-1 and the Bcl proteins appear to be bound by the p53 TAD and DBD domains simultaneously, Mcl-1 exhibits far greater binding affinity for the TAD while Bcl-2/Bcl-xL exhibits far greater binding affinity for the DBD,⁶⁸ requiring mutagenesis of both domains. Several naturally occurring mutants have been identified within the p53 DBD, which abolish both transcriptionally dependent and independent functions simultaneously, including hot-spot DNA contact mutant R273H (Figure 2), which has been shown to disrupt binding between p53 and Bcl-2/Bcl-xL.^33,70–72 While the binding interface between p53 and Mcl-1 has not been as intimately studied, NMR titrations of the TAD/Mcl-1 complex reflect high perturbations at residues 16–28 and 45–56 of TAD1 and TAD2 respectively,⁶⁸ indicating these residues as the dominant contributors toward the binding interaction.⁶⁸ Residues L22/W23 and W53/F54 specifically exhibit average to above average perturbation, and their compound hot-spot mutants L22Q/W23S and W53Q/F54S (Figure 2) have been shown to render p53 ineffective in vitro and in vivo, producing transcriptional profiles comparable to that of p53-null cells.^73–75 It then follows that these TAD compound mutants may have a behavior similar to that of other p53 hot-spot mutations, acting against both its transcriptional and mitochondrial functions. With respect to Bad/Bad*, the BH3-binding groove of Bcl-2/Bcl-xL and the BH3 domain of Bad/Bad* form the dimer interface. It has been shown that introduction of a single L115A (Figure 2) mutation into Bad BH3 disrupts this interface and completely abrogates interactions with Bcl-2/Bcl-xL.^62,69,76

Figure 2.

Domains and binding mutations of p53-Bad*. The p53 transcription activation domain (TAD) and DNA-binding domain (DBD) are primarily responsible for binding to antiapoptotic proteins Mcl-1 and Bcl-2/Bcl-xL, respectively. The Bad* BH3 domain is responsible for binding to Bcl-2 and Bcl-xL. Compound mutations L22Q/W23S and W53Q/F54S in the p53 TAD are intended to disrupt binding to Mcl-1, while R273H and L115A point mutations in the p53 DBD and Bad* BH3 domains are intended to disrupt binding to Bcl-2 and Bcl-xL. Elongated p53 structure adapted from Demir, Ieong, and Amaro (2017) with permission.⁷⁷

Herein, we investigate the nature of our p53-Bad* construct and its activity at the mitochondria in three HCC cell lines with varying p53 statuses. We probe functional activity in the upstream intrinsic apoptotic pathway by exploring protein–protein interactions between p53-Bad* and target antiapoptotic proteins Mcl-1, Bcl-2, and Bcl-xL using binding interaction assays. As Bak/Bax oligomerization acts as the checkpoint for the intrinsic pathway, committing the cell to apoptosis, we also explore functional activity in the downstream Bak/Bax oligomerization event by evaluating cell survival and proliferation when challenged with p53-Bad* and rescued by a potent Bak/Bax oligomerization inhibitor. We further validate the correlation between antiapoptotic binding interactions and apoptosis by generating a series of p53-Bad* mutants designed to disrupt binding interactions between p53 and Bad* with Mcl-1, Bcl-2, and Bcl-xL, both as individual and combination mutants.

EXPERIMENTAL SECTION

Cell Culture, Cell Lines, and Transient Transfection.

All cell lines were purchased from ATCC and grown as monolayers at 37 °C with 5% CO₂. Epithelial HCC cell lines include HepG2/C3A (p53-WT; abbreviated as C3A henceforth), Hep3B2.1–7 (p53-null; abbreviated as Hep3B), and PLC/PRF/5 (p53-R249S, dominant-negative; abbreviated as PP5). Epithelial liver cell line THLE-2 was included as a nontumorigenic control. HCC cells were maintained in DMEM media supplemented with 1% l-glutamine (Thermo Scientific; #25030081), 1% penicillin–streptomycin (Thermo Scientific; #15070063), and 10% FBS (R&D Systems; #S11150). THLE-2 cells were maintained in precoated flasks with BEGM media using the BEBM Bullet Kit (Lonza/Clonetics Corporation; #CC-3170), without the gentamycin/amphotericin and epinephrine and supplemented with an additional 5 ng/μL EGF, 70 ng/mL phosphoethanolamine, and 10% FBS as indicated by ATCC. For precoated THLE-2 flasks, a mixture of 0.01 mg/mL fibronectin (MilliporeSigma; #F1141), 0.03 mg/mL bovine collagen I (MilliporeSigma; #C4243), and 0.01 mg/mL bovine serum albumin (MilliporeSigma #10735078001) dissolved in BEBM media was applied to the flask floors until covered, followed by incubation at 37 °C for 1 h. The coating mixture was then aspirated, and the flasks were dried at room temperature before storage at 4 °C.

For the CheckMate Mammalian Two-Hybrid System (Promega; # E2440), TMRE (Invitrogen; # T669), and 7-AAD (Invitrogen; # A1310) assays, cells were seeded between 3 × 10⁵ and 4 × 10⁵ cells/well into 6-well plates (Genesee Scientific; #25–105). After 24 h, each well was transfected with 1 pmol (500 μM) of DNA using jetPRIME reagent (Polyplus; #101000001) per the manufacturer’s protocol. For CCK8 cell viability assays, cells were sown at a density of 1 × 10⁴ cells/well into 96-well plates (Corning; #3300). After 24 h, each well was transfected with 0.05 pmol (500 μM)of DNA using jetPRIME reagent. All experiments were performed between cell passages 5–15.

p53-Bad* Mutants: Nomenclature and Descriptions.

Five p53-Bad* mutants designed to disrupt various binding interactions between p53-Bad* and Mcl-1, Bcl-2, and Bcl-xL were produced by site-directed mutagenesis. Mutations were introduced into the p53 and Bad* regions of p53-Bad* individually and in combination, targeting binding to either Mcl-1 or Bcl-2/Bcl-xL or Mcl-1, Bcl-2, and Bcl-xL collectively. Exact mutation(s), short-hand identities, and design descriptions are provided in Table 1.

Table 1.

p53-Bad* Binding Mutants^a

Identity	Mutation	Description
[All]p53-[Bcl]Bad*	[L22Q, W23S, W53Q, F54S; R273H]p53-[L115A]Bad*	L22Q/W23S and W53Q/F54S are mutations in p53 that disrupt binding between p53 and Mcl-1. R273H mutation in p53 disrupts binding between p53 and Bcl-2/Bcl-xL, and L115A mutation in Bad* disrupts binding between Bad* and Bcl-2/Bcl-xL.
[Bcl]p53-Bad*	[R273H]p53-Bad*	R273H mutation in p53 disrupts binding between p53 and Bcl-2/Bcl-xL.
[Bcl]p53-[Bcl]Bad*	[R273H]p53-[L115A]Bad*	R273H mutation in p53 disrupts binding between p53 and Bcl-2/Bcl-xL, and L115A mutation in Bad* disrupts binding between Bad* and Bcl-2/Bcl-xL.
[Mcl]p53-Bad*	[L22Q, W23S, W53Q, F54S]p53-Bad*	L22Q/W23S and W53Q/F54S mutations in p53 disrupt binding between p53 and Mcl-1.
p53-[Bcl]Bad*	p53-[L115A]Bad*	L115A mutation in Bad* disrupts binding between Bad* and Bcl-2/Bcl-xL.

^aTo indicate which protein in the fusion construct was mutated, we inserted brackets before the protein name. Designated within the brackets is a short-hand identifier corresponding to the protein interaction partner targeted by a mutant, where [Bcl] represents Bcl-2/Bcl-xL, [Mcl] represents Mcl-1, and [All] represents Mcl-1, Bcl-2, and Bcl-xL collectively.

Plasmid Constructs.

Human p53-WT, Bad*, and p53-Bad*^19,42 were cloned into vector pACT while human Bcl-2, Bcl-xL, and Mcl-1 cDNA (Sino Biological) were cloned into vector pBIND using SalI and EcoRV restriction sites (New England Biolabs). Cloned constructs and their primer pairs are listed in Table 2.

Table 2.

Mammalian Two-Hybrid System Constructs and Primer Pairs^a

Construct	Sequence
p53-WT	5’-CTAGTCGACTTGACATGGAGGAGCCGCAGTC-3’ 5’-CTAGATATCAACGCTTAGTCTGAGTCAGGCCC-3’
Bad*	5’-CGAGTCGACTTGACATGTTCCAGATCCCAGAG-3’ 5’-CTAGATATCAACGCTCACTGGGAGGGGGCG-3’
p53-Bad*	5’-CTAGTCGACTTGACATGGAGGAGCCGCAGTC-3’ 5’-CTAGATATCAACGCTCACTGGGAGGGGGCG-3’
Mcl-1	5’-CTAGTCGACTTGACATGTTTGGCCTCAAAAG-3’ 5’-GCCGATATCAACGCCTATCTTATTAGATATGCC-3’
Bcl-2	5’-CTAGTCGACTTGACATGGCGCACGCTGGGAGAAC-3’ 5’-CGGGATATCAACGCTCACTTGTGGCCCAGATAG-3’
Bcl-xL	5’-CTAGTCGACTTGACATGTCTCAGAGCAACCGGG-3’ 5’-GCCGATATCAACGCTTATTTCCGACTGAAGAG-3’

^aFor each construct, forward primers are listed first and reverse primers are listed second.

All p53-Bad* mutants were produced using pCMV-EGFP-p53-Bad* and pACT-p53-Bad* constructs, as described above. Mutants [R273H]p53-Bad* ([Bcl]p53-Bad*) and p53-[L115A]Bad* (p53-[Bcl]Bad*) were produced using the QuikChange II Site-Directed Mutagenesis Kit (Agilent # 200524) and forward/reverse primer pairs, while mutants [L22Q, W23S, W53Q, F54S]p53-Bad* ([Mcl]p53-Bad*), [R273H]p53-[L115A]Bad* ([Bcl]p53-[Bcl]Bad*), and [L22Q, W23S, W53Q, F54S, and R273H]p53-[L115A]Bad* ([All]p53-[Bcl]Bad*) were produced using the QuikChange Multi Site-Directed Mutagenesis Kit (Agilent #200515) and sets of forward primers. All mutant constructs and their primer pairs/sets are provided in Table 3.

Table 3.

Mutagenesis Constructs and Primer Pairs/Sets^a

Construct	Sequence
[Bcl]p53-Bad*	5’-GCTTTGAGGTGCATGTTTGTGCCTGTCC-3’ 5’-GGACAGGCACAAACATGCACCTCAAAGC-3’
p53-[Bcl]Bad*	5’-GCTATGGCCGCGAGGCCCGGAGGATGAGTGG-3’ 5’-CCACTCATCCTCCGGGCCTCGCGGCCATAGC-3’
[Mcl]p53-Bad*	5’-GGAAACATTTTCAGACCAATCGAAACTACTTCC-3’ 5’-GGACGATATTGAACAACAGAGCACTGAAGACCCAGG-3’
[Bcl]p53-[Bcl]Bad*	5’-GCTTTGAGGTGCATGTTTGTGCCTGTCC-3’ 5’-GCTATGGCCGCGAGGCCCGGAGGATGAGTGG-3’
[All]p53-[Bcl]Bad*	5’-GGAAACATTTTCAGACCAATCGAAACTACTTCC-3’ 5’-GGACGATATTGAACAACAGAGCACTGAAGACCCAGG-3’ 5’-GCTTTGAGGTGCATGTTTGTGCCTGTCC-3’ 5’-GCTATGGCCGCGAGGCCCGGAGGATGAGTGG-3’

^aFor [Bcl]p53-Bad* and p53-[Bcl]Bad*, forward primers are listed first and reverse primers are listed second. For [Mcl]p53-Bad*, [Bcl]p53-[Bcl]Bad*, and [All]p53-[Bcl]Bad*, all listed primers are forward primers used a set for multi-site mutagenesis.

Quantitative Real-Time PCR (qRT-PCR).

Cells were plated in complete media at a concentration of 3 × 10⁵ cells/well in six well plates and incubated for approximately 24 h. Cells were harvested, and mRNA was extracted using the Zymo Research Quick-RNA MiniPrep Kit (Genesee Scientific #11–327) according to the manufacturer’s instructions; mRNA samples were all diluted to 30 ng/μL. cDNA was then made using the SuperScript III First-Strand Synthesis System (Thermo Fisher Scientific #18080051). Quantitative real-time PCR was run using a QuantStudio 3 from Thermo Fisher Scientific using TaqMan Universal PCR Master Mix (Thermo Fisher Scientific #4304437) and TaqMan probes from Thermo Fisher Scientific targeting Bak (Hs00832876_g1), Bax (Hs00180269_m1), Bcl-2 (Hs0104832_g1), Bcl-xL (Hs00236329_m1), and Mcl-1 (Hs01050896_m1) with control 18S (Hs99999901_s1). Data was analyzed using a standard curve in a 96-well plate for each experiment. A Hep3B sample was used to generate the standard curve for Bak, Bax, Bcl-xL, and Mcl-1 experiments, and a C3A sample was used to generate the standard curve for Bcl-2 experiments in order to use the cell line with the highest expression of each construct to generate the standard curve. Each cell line was harvested in three separate experiments with at least three replicates per experiment. Each replicate was run in duplicate during qRT-PCR, and the mean Ct value normalized to the 18S expression of the same sample was reported.¹⁹ One outlier was removed from the Mcl-1 data set using Grubb’s test (alpha = 0.01) and replaced with a new replicate.

Mammalian Two-Hybrid System.

Using the CheckMate Mammalian Two-Hybrid System (Promega; #E2440), anti-apoptotic gene constructs (Mcl-1, Bcl-2, and Bcl-xL) and p53 constructs (p53-WT, p53-Bad*, [Mcl-1]p53-Bad*, [Bcl]p53-Bad*, p53-[Bcl]Bad*, [Bcl]p53-[Bcl]Bad*, [All]p53-[Bcl]Bad*) were subcloned into both pACT and pBIND vectors as a precaution toward potential vector directionality and vector-dependent interaction.⁷⁸ pACT (which includes the VP16 activation domain), pBIND (which includes the Gal4 DNA-binding domain and also encodes the Renilla reniformis luciferase gene for normalization), and pG5luc (which encodes the firefly luciferase reporter gene downstream of five Gal4 binding sites and a TATA box) plasmids were cotransfected at a 1:1:1 molar ratio of 0.5 pmol each with a fourth plasmid (pCMV-EGFP) acting as an empty vector to account for mass difference. As provided by the vendor, pACT-MyoD and pBIND-Id plasmids were used as positive control, while empty pACT and pBIND plasmids were used as negative control. Approximately 24 h post-transfection, samples were analyzed for firefly and Renilla luminescence using the Dual-Glo Luciferase Assay System (Promega; #E2940).

Luciferase Reporter Gene Assay.

The Dual-Glo Luciferase Assay System (Promega; #E290) was used to determine the activities of firefly luciferase and Renilla reniformis per the manufacturer’s protocol by measuring the luminescence signal with an Infinite M1000 microplate reader (Tecan, Männedorf) 24 h post-transfection. The Dual Glo luciferase reagent was added to each well and incubated for 15 min at room temperature, and the firefly luminescence signal was measured to determine luciferase activity. Next, Stop-Glo reagent was added to each well and incubated for 15 min at room temperature, followed by measuring Renilla luminescence. The Dual-Glo luciferase assay was performed in triplicate with independent transfections of each construct (n = 3). The relative response ratios (RRRs) were then calculated using the formula: RRR = (x_i − x_min)/(x_max − x_min), where x_i is the normalized luciferase value for each construct, x_min is the lowest value observed across all samples, and x_max is the highest.

CCK-8 Cell Viability Assay.

The Cell Counting Kit-8 (CCK-8) Assay (Dojindo; #CK04) utilizes a water-soluble tetrazolium salt that is reduced by dehydrogenases in living cells to yield a yellow formazan dye that can be measured colorimetrically (proportional to the number of living cells). Changes in cell viability over time were assessed using this kit according to the manufacturer’s instructions at 12, 16, 20, 24, 36, and 48 h time points post-transfection (time points based on our observations of the start and end of cell death using our constructs). Approximately 10 h post-transfection, cells were treated with 10 μM Bak/Bax inhibitor MSN-125 (MedChemExpress; #HY-12007). At scheduled time points above, 10 μL of CCK-8 solution was added to wells and allowed to incubate at 37 °C for 1 h before measuring absorbance at 450 nm (Agilent BioTek Cytation 3 Imaging Microplate Reader). Nontransfected, untreated cells were used as a positive control, while cells treated with 10% DMSO were used as a negative control.

TMRE Assay.

Tetramethylrhodamine ethyl ester (TMRE) is an orange-red cationic dye that accumulates in active mitochondria and is used to determine the integrity of the mitochondria by assessing the mitochondrial membrane potential. As performed previously,^19,42 cells were harvested 24 h post-transfection, resuspended in 400 μL of PBS, and stained with 50 nM of TMRE reagent (Invitrogen, #T669). The stained cells were incubated at 37 °C for 25 min and then analyzed by flow cytometry (FACS-Canto-II; Flow Core Facility, University of Utah) using FlowJo. Gating was performed to select transfected, EGFP-positive cells, and singlet cells. EGFP was excited at 488 nm with an emission filter of 530/30 nm, while TMRE was excited at 561 nm with an emission filter of 585/15 nm. A reduction in the intensity of the TMRE fluorescence was used as an indicator of depolarization of the mitochondria, corresponding to an increase in mitochondrial outer membrane permeabilization (MOMP). Experiments were performed in triplicate with three biological replicates per experiment (n = 9). The lowest IMM (inner mitochondrial membrane) depolarization value (EGFP control) was set as 0%, and the highest value (p53-Bad*) was set as 100%.

7-AAD Assay.

As performed previously,^19,42 cells were harvested 24 h post-transfection and were resuspended in 400 μL of PBS containing 3 μg/mL 7-aminoactinomycin D (7-AAD; Invitrogen, #A1310). Cells were incubated on ice and analyzed by flow cytometry (FACS Canto-II) using FlowJo. Gating was performed to select transfected, EGFP-positive cells, and singlet cells. EGFP was excited at 488 nm with an emission at 535 nm, while 7-AAD was excited at 561 nm with an emission at 780 nm. As 7-AAD is a membrane impermeable dye and is generally excluded from viable cells, increases in signal intensity are used as indicators of late-stage apoptosis. Experiments were performed in triplicate with three biological replicates per experiment (n = 9). The lowest IMM (inner mitochondrial membrane) depolarization value (EGFP control) was set as 0% and the highest value (p53-Bad*), as 100%.

Statistical Analysis.

Unless otherwise indicated, assays were performed in biological triplicates with at least three technical replicates per experiment. Statistical analyses and multiple comparison corrections were performed by two-way ANOVA with Tukey’s posthoc test for mammalian two-hybrid assays; one-way ANOVA with Tukey’s posthoc test for qRT-PCR experiments and TMRE and 7-AAD apoptosis assays; and two-way ANOVA with Bonferroni’s posthoc test for CCK8 cell viability assays. For all, ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, and ns = not significantly different. Normality assumptions were validated by quantile (QQ) plots for contingent ANOVA analyses (Supplemental Figures S1–S3).

RESULTS

Baseline Gene Expression of Bak, Bax, Bcl-2, Bcl-xL, and Mcl-1 in HCC Cell Lines.

In order to determine baseline gene expression levels of the pro- and antiapoptotic factors involved in this study (Bak, Bax, Bcl-2, Bcl-xL, and Mcl-1), quantitative real-time PCR was performed in all three liver cancer cell lines (Hep3B, PP5, and C3A) as well as in an immortalized normal liver epithelial cell line, THLE2, to serve as a control. Gene expression of pro-apoptotic factors Bak and Bax in the cancer cell lines was not significantly different from THLE2 cells (Figure 3A,B). For the antiapoptotic factors (Bcl-2, Bcl-xL, and Mcl-1), gene expression in the cancer cells was upregulated or not significantly different from expression in THLE2 cells (Figure 3C–E). C3A cells showed significantly elevated Bcl-2 expression levels compared with THLE2 (p < 0.0001), PP5 (p < 0.0001), and Hep3B (p < 0.0001) cells (Figure 3C). Bcl-xL was significantly upregulated in C3A and Hep3B cells compared to THLE2 (p < 0.01 and p < 0.001, respectively) and PP5 (p < 0.05 and p < 0.01, respectively) cells (Figure 3D). Hep3B cells displayed significantly increased expression levels of Mcl-1 compared to THLE2 (p < 0.05), PP5 (p < 0.05), and C3A (p < 0.01) cell lines (Figure 3E) (statistical comparison not shown). Overall, the gene expression of Bcl-2, Bcl-xL, Mcl-1, Bak, and Bax varies between the different liver cancer cell lines (as similarly reported in the literature using Western blots^79–83). Those expression levels that do vary significantly from the baseline THLE2 levels display upregulation for antiapoptotic factors, with C3A cells displaying increased gene expression of Bcl-2 and Bcl-xL and Hep3B cells showing increased expression of Bcl-xL and Mcl-1. PP5 cells showed no significant difference in gene expression of any of the tested factors compared to THLE2 control cells.

Figure 3.

Gene expression of pro- and anti-apoptotic factors in C3A, Hep3B, PP5, and THLE2 cell lines. Gene expression of Bak (A), Bax (B), Bcl-2 (C), Bcl-xL (D), and Mcl-1 (E) measured via quantitative real-time PCR in C3A, Hep3B, and PP5 liver cancer cell lines and THLE2 (immortalized hepatocytes) cell lines. Statistical analysis was performed using one-way ANOVA with Tukey’s posthoc test (****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, ns = not significant). n ≥ 9 for each group.

p53-Bad* Exhibits Superior Binding to Mcl-1, Bcl-2, and Bcl-xL Relative to p53 Wild-Type.

To ascertain if our p53-Bad* construct exhibits superior binding and neutralization of antiapoptotic factors relative to p53-WT, both were evaluated against antiapoptotic Bcl-xL, Bcl-2, and Mcl-1 using the mammalian two-hybrid system (Figure 4). Experiments were performed in three HCC cell lines with varying p53 statuses: p53-null Hep3B (Figure 4A), p53-WT C3A (Figure 4B), and dominant negative PP5 (Figure 4C). The dominant-negative inhibition of wt p53 has notably been proven to be an obstacle in the translation of p53-WT gene therapy in the past.^22,84,85

Figure 4.

Protein–protein interactions via the mammalian two-hybrid system. p53-WT and p53-Bad* were evaluated for binding interactions with antiapoptotic proteins Bcl-xL, Bcl-2, and Mcl-1 in (A) p53 null Hep3B, (B) wild-type p53 C3A, and (C) dominant-negative p53 PP5 HCC cells 24 h post-transfection. Statistical analysis was performed using two-way ANOVA with Tukey’s posthoc test following validation of approximate normality by the QQ plot (****p < 0.0001, **p < 0.01, *p < 0.05, ns = not significant). Error is shown as mean ± standard deviation from triplicates (n = 9).

In all cell lines, binding interactions between Bcl-2/Bcl-xL and p53-Bad* (Figure 4, blue dotted bars) are greatly enhanced relative to binding interactions between Bcl-2/Bcl-xL and p53-WT (Figure 4, orange lined bars) (p < 0.0001 for all comparisons). Trends are consistent regardless of cellular p53 status, with the greatest relative binding observed in Bcl-xL (Figure 4A–C, first blue bar) followed by Bcl-2 (Figure 4A–C, second blue bar). In contrast, binding enhancements between p53-Bad* and Mcl-1 (Figure 4A–C, third blue bar) relative to p53-WT (Figure 4A–C, third orange bar) are lower and more variable, ranging from nonsignificant in Hep3B to significant in PP5 (p < 0.05) and more significant in C3A (p < 0.01). These results suggest that, while fusion of Bad* to p53 enhances binding to Bcl-xL and Bcl-2, it does not consistently enhance binding to Mcl-1, nor enhance binding to the same extent. Confirmation of normality for two-way ANOVA analysis is shown in Supplemental Figure S1.

Bak/Bax Inhibition Protects Cells From p53-Bad* Induced Cytotoxicity.

To validate that p53-Bad* induces apoptosis by committing cells to death through the intrinsic apoptotic pathway, we attempted to rescue cells treated with p53-Bad* using a potent Bak/Bax oligomerization inhibitor, MSN-125.⁸⁶ This inhibitor is unique in its ability to inhibit both Bak and Bax by altering the BH3-groove dimer interface to prevent MOMP, an important consideration as Bak and Bax are functionally interchangeable in most cell lines.^54,55 Cells were treated with transfected p53-Bad* alone, MSN-125 alone, and a combination (combo) of p53-Bad* and MSN-125, measuring cell survival and proliferation over time (Figure 5). A 10% DMSO treatment group was included as a positive control, while an untreated group was included as a negative control. Due to the high potency and rapid toxicity of p53-Bad*,^19,42 it is imperative that cells be incubated with the inhibitor in advance to p53-Bad* activity onset (10 h post-transfection). However, early incubation with the inhibitor also results in early cytotoxicity, requiring analysis over a large window of time to resolve differences between the treatment groups. While MSN-125, both as a stand-alone treatment and in combination (combo) with p53-Bad*, exhibits mild to moderate cytotoxicity at early time points relative to untreated groups, cells do recover to varying degrees. Combo treated Hep3B cells recover by 24 h and continue to proliferate until 48 h (Figure 5A, orange circles), a stark contrast to combo treated PP5 cells which do not fully recover until 48 h (Figure 5E, orange circles) and combo treated C3A combo cells plateau through 48 h (Figure 5C, orange circles).

Figure 5.

CCK8 cell viability and proliferation. Optical density (450 nm) was measured at 12, 16, 20, 24, 36, and 48 h time points post-transfection. (A, C, E) Cells can be rescued from p53-Bad* cell killing at later time points by treatment with MSN-125. (A, B) MSN-125 rescues p53 null Hep3B cells from 20 h onward, (C, D) p53-WT C3A cells from 24 h onward, and (E, F) dominant-negative PP5 cells from 36 h onward. Statistical analysis was performed using two-way ANOVA with Bonferroni’s posthoc test following validation of approximate normality by the QQ plot (****p < 0.0001, ***p < 0.001, ns = not significant). Error is shown as mean ± standard deviation (n = 6).

In p53 null Hep3B (Figure 5A,B), p53 WT C3A (Figure 5C,D), and dominant-negative PP5 (Figure 5E,F) cells, viability in the MSN-125 (red squares), p53-Bad* (blue diamonds), and combo (orange circles) treatment groups is either reduced or nonsignificant relative to the negative control (magenta triangles) at early time points. However, a cytoprotective effect is observed in the combo treatment group (orange circles) relative to the p53-Bad* treatment group (blue diamonds) from 20 h onward in Hep3B (Figure 5A,B), 36 h onward in C3A (Figure 5C,D), and 48 h onward in PP5 (Figure 5E,F) cells. In all scenarios, cells treated with MSN-125 alone (red squares) exhibit higher proliferation than combination (orange circles) treatment (Figure 5A,C,E), but combination treatment (orange lined bar) also exhibits higher proliferation than treatment with p53-Bad* alone (blue dotted bar) at 48 h (Figure 5B,D,F). Maximal protection/proliferation is observed at the 48-h time point for all cell lines. Confirmation of normality for the two-way ANOVA analysis is shown in Supplemental Figure S2.

Mutations in p53-Bad* Disrupt Binding to Mcl-1, Bcl-2, and Bcl-xL.

To validate that the causality of enhanced apoptosis of p53-Bad* is indeed due to enhanced binding interactions with antiapoptotic factors, we generated a series of p53-Bad* mutants designed to impair interactions with Mcl-1, Bcl-2, and/or Bcl-xL including mutations against: Bcl-2/Bcl-xL binding in p53 ([Bcl]p53-Bad*), Bcl-2/Bcl-xL binding in Bad* (p53-[Bcl]Bad*), Mcl-1 binding in p53 ([Mcl]p53-Bad*), combined Bcl-2/Bcl-xL binding in p53 and Bad* ([Bcl]p53-[Bcl]Bad*), and a total combination of all mutants ([All]p53-[Bcl]Bad*). As performed for p53-WT and p53-Bad* previously,^19,42 these five mutants were subcloned with and evaluated against antiapoptotic Bcl-xL, Bcl-2, and Mcl-1 in p53 null Hep3B (Figure 6A), p53 WT C3A (Figure 6B), and dominant-negative PP5 (Figure 6C) cell lines using the mammalian two-hybrid system.

Figure 6.

Mutant binding interactions measured by the mammalian two-hybrid binding assay. Constructs were evaluated for binding interaction with antiapoptotic proteins Bcl-xL, Bcl-2, and Mcl-1 in (A) p53 null Hep3B, (B) wild-type p53 C3A, and (C) dominant-negative p53 PP5 HCC cells 24 h post-transfection. Statistical analysis was performed using two-way ANOVA with Tukey’s posthoc test following validation of approximate normality by the QQ plot (****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, ns = not significant). Error is shown as mean ± standard deviation from triplicates (n = 9). All indicated statistical comparisons were performed against p53-Bad* within the same sample sets.

Across all cell lines, the same general trends are observed (Figure 6A–C): (i) mutations in Bad* (purple gridded bars) appear more impactful and produce more significant reductions in protein–protein interaction than mutations in p53 (blue dotted bars and orange diagonally lined bars), (ii) the combination of p53 and Bad* mutants produce the greatest total reductions in protein–protein interaction (yellow checkered bars and gray bricked bars), and (iii) mutations targeted against Mcl-1 (blue dotted bars) most significantly reduce interactions with Mcl-1 as compared to Bcl-xL and Bcl-2, while mutations targeted against Bcl-xL and Bcl-2 (orange diagonally lined bars and purple gridded bars) most significantly reduce interactions with Bcl-xL and Bcl-2 as compared to Mcl-1, as expected. Notably, mutations targeted against Bcl-xL and Bcl-2 do occasionally reduce interactions between p53 and Mcl-1 as well (Figure 6B, Mcl-1 group), as observed for mutants [Bcl]p53-Bad* (orange diagonally lined bar) and [Bcl]p53-[Bcl]Bad* (yellow checkered bar) in C3A, while the p53 Mcl-1 mutant [Mcl]p53-Bad* (blue dotted bar) also reduced the interaction between p53 and Bcl-xL on one occasion (Figure 6C, Bcl-xL group). While these nuances are inconsistent between cell lines, they are not unexpected. As p53 interacts with each of these targets through both the TAD and DBD domains, it seems likely that a mutation targeting binding to Mcl-1 in the TAD might also impact binding to Bcl-2/Bcl-xL. Confirmation of normality for two-way ANOVA analysis is shown in Supplemental Figure S3.

Mutations in p53-Bad* Reduce Early and Late-Stage Apoptosis.

TMRE.

To determine if binding disruption between p53-Bad* and its antiapoptotic targets affects the apoptotic ability of p53-Bad*, cells were treated with p53-Bad* binding mutants (Table 1) designed to specifically disrupt binding between: p53 and Bcl-2/Bcl-xL ([Bcl]p53-Bad*), p53 and Mcl-1 ([Mcl]p53-Bad*), Bad* and Bcl-2/Bcl-xL (p53-[Bcl]Bad*), p53/Bad* with Bcl-2/Bcl-xL ([Bcl]p53-[Bcl]Bad*), and p53/Bad* with Bcl-2/Bcl-xL/Mcl-1 collectively ([All]p53-[Bcl]Bad*). Following treatment with p53-Bad* and its mutants in p53 null Hep3B (Figure 7A), p53 WT C3A (Figure 7B), and dominant-negative PP5 (Figure 7C) cell lines, samples were stained and analyzed using tetramethylrhodamine ethyl ester perchlorate (TMRE), a cationic dye that accumulates and fluoresces in healthy, polarized mitochondria and provides a measure of early stage apoptosis.

Figure 7.

TMRE early stage apoptosis. Assays were conducted 24 h post-transfection with p53-Bad* and five binding mutants in HCC cell lines (A) p53 null Hep3B, (B) wild-type p53 C3A, and (C) dominant-negative p53 PP5. Statistical analysis was performed using one-way ANOVA with Tukey’s posthoc test (****p < 0.0001, ***p < 0.001, *p < 0.05, ns = not significant). Error is shown as mean ± standard deviation from triplicates (n = 9).

In all cell lines, total binding disruption by [All]p53-[Bcl]Bad* (Figure 7A–C, yellow checkered bar) showed the greatest reduction in apoptosis relative to p53-Bad* (Figure 7A–C, gray bricked bar). In contrast, disruption between p53 and Mcl-1 by [Mcl]p53-Bad* reduced apoptosis the least of all mutations (Figure 7A–C, blue dotted bar). Disruption of binding to Bcl-2 and Bcl-xL in Bad* by p53-[Bcl]Bad* (Figure 7A–C, orange diagonally lined bar) resulted in a far more significant reduction in apoptosis compared to disruption of the same interaction in p53 by [Bcl]p53-Bad*(Figure 7A–C, blue dotted bar). Despite the various mutations that disrupt binding in p53-Bad*, all mutants exhibited a higher percentage of apoptosis compared to p53-wt (Figure 7A–C, cyan horizontally lined bar) across all three cell lines. TMRE gating parameters for control and experimental samples are provided in Supplemental Figures S4 and S5, respectively.

7-AAD.

To further evaluate the impact of binding disruptions between p53-Bad* and its antiapoptotic targets toward apoptotic activity, p53-Bad* binding mutants (Table 1) were again used to treat p53 null Hep3B (Figure 8A), p53 WT C3A (Figure 8B), and dominant-negative PP5 (Figure 8C) cell lines prior to analysis using 7-aminoactinomycin D (7-AAD), a fluorescent intercalator that exhibits a spectral shift upon binding to DNA and provides a measure of late-stage apoptosis.

Figure 8.

7-AAD late-stage apoptosis. Assays were conducted 24 h post-transfection with p53-Bad* and five binding mutants in (A) p53 null Hep3B, (B) wild-type p53 C3A, and (C) dominant-negative p53 PP5 cell lines. Statistical analysis was performed using one-way ANOVA with Tukey’s posthoc test (****p < 0.0001, ***p < 0.001, *p < 0.05, ns = not significant). Error is shown as mean ± standard deviation from triplicates (n = 9).

In all cell lines, combined Bcl-2/Bcl-xL disruption by [Bcl]p53-[Bcl]Bad* (Figure 8A–C, magenta gridded bar) and total binding disruption by [All]p53-[Bcl]Bad* (Figure 8A–C, yellow checkered bar) showed the greatest reductions in apoptosis relative to p53-Bad* (Figure 8A–C, gray bricked bar, p < 0.0001). Binding disruption between Bad* and Bcl-2/Bcl-xL by p53-[Bcl]Bad* (Figure 8A–C, orange diagonally lined bar) provides the greatest reduction in apoptosis for any single mutant with varying degrees of significance between cell lines. Binding disruption between p53 and Mcl-1 by [Mcl]p53-Bad* consistently exhibits the least reduction in apoptosis relative to p53-Bad*, with significant reduction only observed in PP5 but not Hep3B or C3A (Figure 8C, red triple-dotted bar, p < 0.05). 7-AAD gating parameters for control and experimental samples are provided in Supplemental Figures S6 and S7, respectively.

The apoptotic impact of p53-Bad* was also directly compared between cell lines for both early- and late-stage apoptosis assays, normalizing results against p53 WT to obtain relative apoptosis for p53-Bad* between cell lines (Supplemental Figure S8). Early apoptosis induced by p53-Bad* is similar between cell lines, with no significant differences observed (Supplemental Figure S8A). However, variations are observed in late-stage apoptosis, where PP5 appears to be the least sensitive to treatment by p53-Bad* (Supplemental Figure S8B).

DISCUSSION

Apoptosis acts as a natural bulwark against the development of cancer, and thus, emergent resistance against apoptosis has become a hallmark of human cancers.^87,88 The central mechanism to apoptosis, by either intrinsic or extrinsic pathways, is the activation of caspases to collapse the cell. Caspase activation is strictly regulated in both pathways, occurring at the mitochondria for the intrinsic pathway—whereby caspase recruitment, activation, and commitment to cell death occur following MOMP. The Bcl-2 protein family members serve as wardens to the mitochondria, regulating MOMP through a delicate homeostasis between pro- and antiapoptotic factors.^89,90 Misregulation and imbalanced expression between pro- and antiapoptotic factors disrupts this equilibrium—conferring apoptotic resistance to the cell, as evidenced in many forms of cancer.^91–104 With respect to HCC specifically, upregulation of Bcl-xL and Mcl-1 have each been observed in approximately one-half of patient samples, and high Bcl-xL expression specifically is correlated with a 31.4 month decrease in patient median survival time.^19,92,98 Moreover, elevated expression of Bcl-xL, Bcl-2, and Mcl-1 in HCC has been associated with tumor growth.^105–111

The functional role of these antiapoptotic Bcl-2 family members in cancer has resulted in development of BH3 mimetics and Bcl-2 inhibitors, including ABT-199/venetoclax, a potent Bcl-2 inhibitor and the only FDA-approved BH3 mimetic thus far.¹¹² Multiple Mcl-1 inhibitors, including S63845,¹¹³ AZD5991,¹¹⁴ and AMG176/tapotoclax,¹¹⁵ have proceeded to clinical trials, in addition to Bcl-2/Bcl-xL/Bcl-w inhibitor ABT-263/navitoclax.¹¹⁶ However, while GX15/obatoclax was designed as a pan-Bcl-2 inhibitor capable of targeting Bcl-2, Bcl-xL, and Mcl-1, this drug did not advance to FDA approval, succumbing to low affinity and clinical activity.¹¹⁷ Moreover, no such pan-Bcl-2 inhibitor has been approved with indication for HCC. This unrealized potential for targeting and pan-inhibition of antiapoptotic proteins at the mitochondria, coupled to the withstanding need for safer and more tolerable treatment strategies, has thus inspired our continual iteration of a mitochondrially targeted p53 fusion construct, resulting in our most recent iteration, p53-Bad*.^19,42

We previously described the design, improved mitochondrial targeting, and enhanced apoptosis of p53-Bad* in HCC and ovarian cancer.^19,42 Here, we investigate and validate the nature of our p53-Bad* construct and its activity at the mitochondria, demonstrating (i) increased binding of anti-apoptotic proteins Mcl-1, Bcl-2, and Bcl-xL relative to p53-WT, (ii) dependence upon mitochondrial Bak/Bax oligomerization to induce cell death, supporting action via the intrinsic pathway, and (iii) a relationship between functional binding capability and apoptotic activity in vitro.

While nuclear p53 is known to exert its transcriptional activity as a tetramer, the oligomeric state of mitochondrial p53 and the effect of oligomerization upon its mitochondrial function remains unclear.^30,118–121 There is evidence that deletion of the tetramerization domain does not impact apoptotic efficiency, and moreover, it has also been shown that tumor-derived dominant-negative p53 mutants that are known to impact transcriptional activity do not impact mitochondrial activity, suggesting a monomeric state for mitochondrial p53.^118,119 However, cross-linking studies have yielded conflicting results—with p53 occupying dimeric and tetrameric states in one and a near-total monomeric state in another.^118,120 As such, the possibility for oligomer-induced inactivation of p53-Bad* with cytosolic p53 prior to or during transit to the mitochondria cannot be precluded in p53-expressing cells.^27,28 This consideration is pertinent not only when evaluating intrinsic apoptotic/cytotoxic activity but also when evaluating binding in the two-hybrid assay, which necessarily requires transport of monomeric p53-Bad* through the cytoplasm to the nucleus for evaluation. As dominant-negative or wild-type inhibition effects on p53-Bad* during transit would thus expectedly be similar for nuclear and mitochondrial destinations, all studies (binding, survival, apoptosis) were performed using three cell lines with variable p53 status to allow for comparison of performance and potential inhibitory effects within individual cell lines between experiments.

The expression status of BH3 genes is another important consideration in cell line selection, as p53-Bad* could be less effective in an environment in which antiapoptotic genes or Bak/Bax are downregulated. High upregulation of Bcl-xL (58%), Mcl-1 (51%), and Bax (63%) are commonly observed in primary HCC cells, while high upregulation of Bcl-2 is less commonly observed (16%).^92,98,104 Quantitative reverse transcription PCR (qRT-PCR) analysis of our secondary HCC cell lines reveals that Bak and Bax expression does not differ significantly between HCC and liver cell lines (Figure 3A–B), while Mcl-1, Bcl-2, and Bcl-xL genes are consistently either upregulated or similar to nontumorigenic THLE-2 (Figure 3C–E). Moreover, Bcl-2 appears exceptionally upregulated in C3A (p < 0.0001), while Bcl-xL is considerably upregulated in C3A and Hep3B. As fusion of Bad* to p53 is hypothesized to enhance binding interactions with Bcl-2 and Bcl-xL specifically, notable upregulation in C3A and Hep3B suggests these cell lines might be particularly sensitive to treatment with p53-Bad*. It should be noted however that mRNA expression is not an exact correlate with protein expression, and these results serve primarily to qualify our cell lines for use rather than to support mechanistic inferences.

p53-Bad* consistently exhibits a greater degree of functional binding to Bcl-xL and Bcl-2 targets relative to p53-WT, regardless of the status of p53 in the cell line (Figure 4). In contrast, interaction with Mcl-1 is not consistently enhanced, nor is it enhanced to the same degree of significance as with Bcl-xL and Bcl-2. This finding is unsurprising as Mcl-1 is not a known target of Bad/Bad* and, in fact, has been shown to not bind to Bad in solution binding studies;¹²² therefore, p53-Bad* fusion was not expected to enhance Mcl-1 binding meaningfully over p53-WT. In all cell lines, binding between p53-Bad* and Bcl-xL is most improved relative to p53-WT, with 8.5-, 7.8-, and 3.5-fold enhancements in Hep3B, PP5, and C3A, respectively (Figure 4). Binding between p53-Bad* and Bcl-2 exhibits lower enhancements of 2.5-, 4.1-, and 3.3-fold relative to those of p53-WT (Figure 4). These findings demonstrate that p53-Bad* is indeed characterized by superior functional binding of mitochondrial antiapoptotic proteins, specifically highlighting the extraordinary contributions of Bad*. It is inferred that binding enhancements are primarily a consequence of added interactions between the Bad* BH3 domain and Bcl-2/Bcl-xL. However, unexpected enhancement with Mcl-1, as observed in C3A and PP5 (Figure 4B,C), suggests that this is not a complete explanation. As Mcl-1 is not a known target of Bad/Bad* and has been repeatedly shown to not interact,^122–124 it seems possible that C-terminal fusion of Bad* to p53 may either stabilize the p53-Mcl-1 complex or alter the Bad* BH3 fold such that it supports binding to the Mcl-1 BH3-binding groove.

An apparent limitation of the two-hybrid system is its inability to constitute a non-nuclear interactive milieu. While p53-Bad* demonstrably interacts strongly with Mcl-1, Bcl-2, and Bcl-xL as nuclear, soluble proteins, this is not a perfect reflection of the mitochondrial environment where they are membrane-bound. However, if p53-Bad* indeed acts as a pro-apoptotic agent at the mitochondria via binding and neutralization of functional, membrane-bound antiapoptotic proteins, then this activity would be blocked by inhibition of Bak/Bax oligomerization, which acts as the checkpoint event for mitochondrially programmed apoptosis. When challenged by transfection with p53-Bad*, cells begin to reflect reductions in cell viability between 12 and 24 h post-transfection, with cell viability minima observed at 48 h for all cell lines (Figure 5). However, treatment with potent Bak/Bax oligomerization inhibitor MSN-125 rescues cells from p53-Bad*-induced cytotoxicity with effects apparent as early as 20 h post-transfection for Hep3B (Figure 5A,B), at 36 h post-transfection for C3A (Figure 5C,D), and as late as 48 h post-transfection for PP5 (Figure 5E,F) cells. Despite early toxicity, treatment with MSN-125 following p53-Bad* transfection indeed improves survival overall, fully rescuing cells from p53-Bad*-induced cytotoxicity. Moreover, an upward trend in viability, indicative of continual proliferation, is apparent in both Hep3B and PP5 cell lines at 48 h. These findings confirm that the pro-apoptotic activity of p53-Bad* is dependent upon downstream Bak/Bax oligomerization and subsequent MOMP, implicating its activity with regulatory Bcl-2 family members and substantiating findings from the two-hybrid assay. To better characterize the binding activity between p53-Bad* and membrane inserted antiapoptotic proteins, thereby mitigating risk of false positive results, future studies that more accurately simulate the native mitochondrial outer membrane are required. Using mitochondrial fractionation to harvest intact HCC mitochondria, recombinant p53-Bad* could be evaluated for a comprehensive profile of binding partners and interactions from the outer mitochondrial membrane, potentially elucidating unexpected interactions with and beyond Bak, Bax, Mcl-1, Bcl-2, and Bcl-xL. Alternatively, the two-hybrid assay could be recycled with transmembrane deleted Mcl-1, Bcl-2, and Bcl-xL interaction partners, as the transmembrane domains of the native proteins would ordinarily be inaccessible by p53-Bad* due to insertion within the outer mitochondrial membrane, therefore providing a more accurate representation of the interaction expected within the native mitochondrial environment.

p53-Bad* is hypothesized to increase apoptosis versus p53-WT due to enhanced binding interactions with Bcl-2 and Bcl-xL conferred by the Bad* BH3 domain. However, p53-Bad* is also expected to retain the intrinsic ability of p53-WT to directly activate Bak/Bax and liberate tBID from the complex with Bcl-xL, with all mechanisms of action converging at downstream Bak/Bax oligomerization (Figure 1). While the Bak/Bax oligomerization blockade does implicate p53-Bad* as a pro-apoptotic agent at the mitochondria, it does not necessarily establish that this activity occurs primarily due to enhanced binding interactions with antiapoptotic proteins, the premise upon which it was designed. Mutant p53-Bad* variants intended to disrupt binding interactions with Mcl-1, Bcl-2, and Bcl-xL (Table 1) demonstrate consistent reductions in protein–protein interactions in all cell lines (Figure 6A–C). In agreement with established binding profiles for p53 and Bad* and regardless of cell line p53 status, as shown in Figure 6, mutations introduced into p53 TAD1/2 ([Mcl]p53-Bad*; blue bar) significantly reduce binding interactions with Mcl-1, while p53 DBD ([Bcl]p53-Bad*; orange bar) and Bad* BH3 (p53-[Bcl]Bad*; purple bar) mutations significantly reduce binding interactions with Bcl-2 and Bcl-xL. In Figure 6, combinations of p53 DBD, Bad* BH3 ([Bcl]p53-[Bcl]Bad*; yellow), and p53 TAD1/2 ([All]p53-[Bcl]Bad*; gray) mutations most significantly reduce binding interactions with all target proteins. Of particular interest, the Bad* BH3 L115A mutation has the most pronounced effect against binding of all single point-mutations, highlighting again the extraordinary and disproportionate contribution provided by Bad* toward the enhanced binding interactions observed in the p53-Bad* fusion (Figure 4). It is important to note that, while combination mutants [Bcl]p53-[Bcl]Bad* and [All]p53-[Bcl]Bad* do generally reflect the greatest reductions in binding, these constructs retain approximately 6–41% of relative functional binding depending upon the target protein and cell line, indicating that the incorporated mutants do not totally abrogate interactions between p53-Bad* and its targets. The p53-BH3 interface is complex and involves both the p53 TAD and DBD domains for each Mcl-1, Bcl-2, and Bcl-xL. Therefore, the individual interactions between p53 and its BH3 antiapoptotic targets cannot be adequately described by a single residue or set of residues within one domain, nor apparently can these interactions be disrupted by single point-mutations targeting one domain. Our p53-Bad* mutants target critical residues only within the domain most dominantly involved in a particular interaction—the p53 DBD for Bcl-2 and Bcl-xL and the p53 TAD1/2 for Mcl-1. It is thus unsurprising that, even upon combination, our p53-Bad* mutants exhibit some retention of activity with their Bcl-2 family targets.

The same p53-Bad* binding mutants demonstrate consistent reductions in both early (Figure 7) and late-stage (Figure 8) apoptosis in both C3A and Hep3B cell lines, although PP5 appears to be significantly less sensitive to p53-Bad*, with fewer cells escalating to late-stage terminal apoptosis despite a similar early apoptosis profile to Hep3B and C3A at the same time point (Supplementary Figure S8). Mutant [Mcl]p53-Bad* does not generally reduce early or late-stage apoptosis despite significantly reducing binding interactions between p53 and Mcl-1 (Figure 6), suggesting that inhibition of Mcl-1 alone may not be sufficient to induce apoptosis in HCC, which is supported by previous precedent.¹²⁵ The effect of mutant [Bcl]p53-Bad* against apoptosis is variable, only slightly reducing early stage apoptosis in Hep3B (Figure 7A) and PP5 (Figure 7C) and not at all in C3A (Figure 7B), with a similar but more pronounced trend reflected in late-stage apoptosis (Figure 8A–C). In good agreement with binding assays (Figure 6), Bad* mutant p53-[Bcl]Bad* consistently exhibits the single greatest reduction in both early and late stage apoptosis of all single mutants—further emphasizing its pronounced ability to mutually reduce apoptosis and binding interactions between Bad* and Bcl-2/Bcl-xL. Additionally, compound mutants [Bcl]p53-[Bcl]Bad* and [All]p53-[Bcl]Bad* consistently exhibit the greatest and most significant total reductions in both early and late-stage apoptosis regardless of endogenous p53 status.

The remarkably close agreement between trends observed in both binding and apoptosis strongly suggests a direct relationship between the enhanced binding interactions and superior apoptotic activity of p53-Bad* relative to p53-WT. As in binding assays, combination mutants [Bcl]p53-[Bcl]Bad* and [All]p53-[Bcl]Bad* do not result in a total reduction of apoptosis, with each retaining more apoptotic activity than even p53-WT. As discussed previously, it is likely that our p53-Bad* mutants do not completely abrogate interactions with target Mcl-1, Bcl-2, and Bcl-xL. Moreover, the antiapoptotic BH3 subfamily includes members unstudied here, including Bcl-w and Bfl-I, whose interactions with p53 and Bad* may be conserved, in addition to interactions with Bak/Bax. Retention of these interactions provides one explanation for the conserved, albeit limited, apoptotic capability of the p53-Bad* mutants.

Although the same trend is observed across cell lines in both early- and late-stage apoptosis assays, PP5 is demonstrably less sensitive to p53-Bad* than Hep3B and C3A, with fewer cells escalating to late-stage apoptosis at the same time point (Supplementary Figure S8) and is consequently impacted by its binding mutants to a lesser degree (Figures 7C and 8C). In contrast to C3A and Hep3B, PP5 does not exhibit meaningful upregulation of Mcl-1, Bcl-2, or Bcl-xL with respect to nontumorigenic THLE-2 cells (Figure 3C–E), and thus, a possible explanation for this phenomenon is that upregulation of Mcl-1, Bcl-2, and Bcl-xL in C3A and Hep3B uniquely sensitizes these cell lines to p53-Bad* as opposed to PP5 (although p53-Bad* still kills these cells).

CONCLUSIONS

Here, the functional activity and apoptotic mechanism of action of our p53-Bad* construct was investigated. Fusion of pro-apoptotic Bad* to p53 enhances interactions with antiapoptotic Bcl-2 family members Mcl-1, Bcl-2, and Bcl-xL relative to p53-WT. Binding and neutralization of antiapoptotic factors by p53-Bad* subsequently leads to downstream Bak/Bax oligomerization. Inhibition of this Bak/Bax oligomerization rescues HCC cells from p53-Bad* induced cytotoxicity and death, confirming that p53-Bad* acts as a pro-apoptotic agent at the mitochondria to induce MOMP and cell death. Mutant p53-Bad* variants designed to disrupt binding interactions with antiapoptotic targets also reduce early and late-stage apoptosis in HCC cells, establishing a direct relationship between the functional binding interactions and improved apoptotic activity of p53-Bad*. Correlation between Mcl-1, Bcl-2, and Bcl-xL gene expression and construct performance suggest that cell lines in which antiapoptotic Bcl-2 family genes are upregulated might exhibit greater sensitivity to treatment with p53-Bad*. These findings represent encouraging and exciting evidence for the use of p53-Bad* as a pan-Bcl-2 inhibitor and synergistic pro-apoptotic agent in cancers with upregulation of multiple antiapoptotic and prognostic markers.

Supplementary Material

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.molpharmaceut.5c00773.

Normal quantile-quantile plots of mammalian two-hybrid assay; normal quantile-quantile plots of CCK8; normal quantile-quantile plots of mutant mammalian two-hybrid assay; flow cytometry gating for TMRE assay control samples; flow cytometry gating for TMRE assay experimental samples; flow cytometry gating for 7-AAD assay control samples; flow cytometry gating for 7-AAD assay experimental sample; relative apoptosis of p53-Bad* (PDF)

ABBREVIATIONS

[All]p53-[Bcl]Bad*

[L22Q, W23S, W53Q, F54S; R273H]p53-[L115A]Bad* where L22Q/W23S and W53Q/F54S are targeted against binding between p53 and Mcl-1, R273H is targeted against binding between p53 and Bcl-2/Bcl-xL, and L115A is targeted against binding between Bad* and Bcl-2/Bcl-xL

APAF1

Apoptotic protease activating factor 1

Bad

Bcl-2 antagonist of cell death protein

Bad*

Bad* = [S112A, S136A]Bad where compound S112A/S136A mutation prevents phosphorylative inactivation

Bak

Bcl-2 homologous antagonist/killer protein

Bax

Bcl-2 associated X-protein

Bcl-2

B-cell lymphoma 2 protein

Bcl-xL

B-cell lymphoma-extra large protein

[Bcl]p53-Bad*

[Bcl]p53-Bad* = [R273H]p53-Bad* where R273H mutation in p53 disrupts binding between p53 and Bcl-2/Bcl-xL

[Bcl]p53-[Bcl]Bad*

[Bcl]p53-[Bcl]Bad* = [R273H]p53-[L115A]Bad* where R273H mutation in p53 disrupts binding between p53 and Bcl-2/Bcl-xL and L115A mutation in Bad* disrupts binding between Bad* and Bcl-2/Bcl-xL

BH3

Bcl-2 homology 3

CCK8

Cell Counting Kit 8

C3A

HepG2/C3A

GFP

Green fluorescent protein

HCC

Hepatocellular carcinoma

Hep3B

Hep3B.1-7

[Mcl]p53-Bad*

[Mcl]p53-Bad* = [L22Q, W23S, W53Q, F54S]p53-Bad* where L22Q/W23S and W53Q/F54S mutations in p53 disrupt binding between p53 and Mcl-1

Mcl-1

Myeloid cell leukemia 1 protein

MOMP

Mitochondrial outer membrane permeability

PP5

PLC/PRF/5

p53-Bad*

p53-Bad[S112A, S136A] where S112A/S136A mutation in Bad prevents phosphorylative deactivation

p53-[Bcl]Bad*

p53-[Bcl]Bad* = p53-[L115A]Bad* where L115A mutation in Bad* disrupts binding between Bad* and Bcl-2/Bcl-xL

TMRE

Tetramethylrhodamine ethyl ester

WT

Wild-type

7-AAD

7-aminoactinomycin D