A high-throughput screening approach to discover potential colorectal cancer chemotherapeutics: Repurposing drugs to disrupt 14-3-3 protein-BAD interactions

doi:10.21203/rs.3.rs-5242408/v1

Download PDF

Article

A high-throughput screening approach to discover potential colorectal cancer chemotherapeutics: Repurposing drugs to disrupt 14-3-3 protein-BAD interactions

https://doi.org/10.21203/rs.3.rs-5242408/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Selectively inducing apoptosis in cancer cells is an effective therapeutic strategy, but the reality of success of existing chemotherapeutics is compromised by emergent tumor cell resistance and systemic off-target effects. Therefore, the discovery of new classes of pro-apoptotic compounds with minimal systemic side-effects remains an urgent need. 14-3-3 proteins are molecular scaffolds that serve as important regulators of cell survival. Our previous study demonstrated that 14-3-3ζ can sequester BAD, a pro-apoptotic member of the BCL-2 protein family, in the cytoplasm to inhibit the induction of apoptosis. Despite being a critical mechanism of cell survival, it is unclear whether disrupting 14-3-3 protein:BAD interactions could be harnessed as a chemotherapeutic approach. Herein, we established a BRET-based, high-throughput drug screening approach (Z’-score = 0.52) capable of identifying molecules that can disrupt 14-3-3ζ:BAD interactions. An FDA-approved drug library containing 1971 compounds was used for screening, and the capacity of identified hits to induce cell death was examined in NIH-3T3 fibroblasts and colorectal cancer cell lines, HT-29 and Caco-2. Our in vitro results suggest that terfenadine, penfluridol, and lomitapide could be potentially repurposed for treating colorectal cancer. An in silico structural analysis, validated by grounding in the experimental data, provides insight into specific molecular interactions and highlights proposed binding modes that can be further modified to refine the affinity and selectivity of identified hits. This multi-modal screening method demonstrates the feasibility of identifying pro-apoptotic agents that can be applied towards conditions where aberrant cell growth or function are key determinants of disease pathogenesis.

Biological sciences/Cell biology

Health sciences/Diseases/Cancer

High-throughput screening

BRET

apoptosis

BCL-2

BAD

14-3-3

colorectal cancer

Apoptosis, or programmed cell death, is a highly regulated process of cell suicide. During apoptosis, cells break down into apoptotic bodies and are eventually engulfed by phagocytes like macrophages and neutrophils^[1]. With limited leakage of a cell’s content into the extracellular environment, apoptosis can minimize the damage to surrounding cells. The identification of molecules that safely and selectively induce apoptosis holds significant potential in treating a variety of conditions, such as cancer, infectious diseases, and autoimmune disorders^[2–4].

Colorectal cancer (CRC) ranks as the second most deadly cancer and accounted for nearly 10% of cancer-induced mortality in 2020^[5]. CRC carcinogenesis originates from either the colon or the rectum, and most malignant adenomas develop from benign polyps^[6]. Current therapies for CRC involve chemotherapy, radiation therapy, and surgery; however, resistance to existing chemotherapeutics often leads to poor clinical outcomes^[7]. Accordingly, it remains important to discover new mechanisms and novel compounds that can selectively induce apoptosis in CRC cells.

An example of a chemotherapeutic that exploits the intrinsic pathway of apoptosis is Venetoclax (ABT-199), which is a specific inhibitor of the anti-apoptotic protein, BCL-2^[8]. The effectiveness of ABT-199 in treating chronic lymphocytic leukemia (CLL) and acute myeloid leukemia (AML) highlights the potential of targeting the actions of members of the BCL-2 protein family to treat cancers. We previously demonstrated that 14-3-3ζ, a member of the 14-3-3 scaffold protein family, plays an essential role in maintaining the survival of MIN6 insulinoma cells through its inhibitory actions on pro-apoptotic BCL-2 proteins, such as BAD and BAX^[9]. In murine cells, 14-3-3 proteins sequester BAD in the cytoplasm by interacting with the phosphorylated Ser112 and Ser136 residues on BAD^[10]. However, the induction of cell death or prolonged cell stress results in these serine residues becoming dephosphorylated, leading to the dissociation of 14-3-3 protein:BAD complexes. This allows BAD to translocate to the outer mitochondrial membrane, where it activates or inhibits pro-apoptotic or anti-apoptotic BCL-2 proteins to initiate apoptosis^[11–13].

14-3-3 proteins have been found to play important roles in cancer cell survival^[14]. Alterations in 14-3-3 protein expression, especially the 14-3-3ζ isoform, have been observed in a variety of cancers, such as those of colon, breast, lung, and pancreas^[15], and overexpression of 14-3-3ζ may mediate tumor resistance to chemotherapy due to its anti-apoptotic functions^[16]. Depletion of 14-3-3ζ has been found to induce the apoptosis of CRC cells in vitro and in vivo^[17], suggesting that identifying or developing novel disruptors of 14-3-3 protein:BAD protein-protein interactions (PPIs) might represent a promising approach towards the treatment of CRC and other cancers.

Drug development involving de novo synthesis and validation of novel chemical entitites is an incredibly expensive and time-consuming endeavor that typically yields a low success rate^[18–20]. From preclinical studies to clinical trials, and ultimately to approval by the US Food and Drug Administration (FDA), the estimated average cost per drug was over $1.5 billion between 2009 to 2018, with a development time of up to over 20 years^{[18, 19]}. Much of this cost is driven by failure, as only 10% of drugs that enter phase I clinical trials are approved^[20]. Given the challenges of new drug development, repurposing already approved compounds for new indications is an attractive strategy, to save both time and costs^[21–23]. To date, the most comprehensive compound screen aimed at identifying disruptors of 14-3-3 protein:BAD PPIs was performed with a time-resolved fluorescence resonance energy transfer (TR-FRET)-based approach, and while 16 hits were discovered from over 52,100 examined compounds, an important caveat was that this assay was based on a cell-free system involving recombinant proteins^[24].Thus, it was not possible to discern if identified hits would act via receptor-mediated pathways or be transported into a cell to directly disrupt 14-3-3 protein:BAD PPIs. The essential follow-up assays to evaluate these activities were not conducted.

Herein, we have developed an innovative BRET (bioluminescence resonance energy transfer)-based biosensor to detect 14-3-3 protein:BAD PPIs in intact, living cells^[25]. Using this sensor and an FDA-approved drug library containing 1971 compounds, we first identified 101 hits through a high throughput screening (HTS) approach in NIH-3T3 fibroblasts. We also examined the possible binding modes of the top compounds using molecular docking simulations and compared them to known inhibitors, in order to better understand the molecular contexts of our constructs. We next evaluated the capacity of these hits to induce cell death, and 41 compounds emerged as potential candidates. Based on their original indications and routes of administration, we selected 13 of these drugs for further assessment of their effectiveness in inducing apoptotic cell death in the well-characterized HT-29 & Caco-2 CRC cell lines. Our screening workflow has identified terfenadine, a withdrawn antihistamine, penfluridol,a 1st generation antipsychotic, and lomitapide (a non-statin cholesterol control medication, as potent candidate molecules that can potentially be repurposed as chemotherapies for inducing CRC cell death with activity mediated by 14-3-3:BAD PPIs^[26–30].

BRET sensor construction

The original plasmids containing 14-3-3ζ, BAD, and BAD mutants were kind gifts from Dr. Herman Spaink and Dr. Aviva M Tolkovsky, respectively^{[31, 32]}. To conjugate mTurquoise, a cyan fluorescent protein (CFP) to the C- or N-termini of 14-3-3ζ, 14-3-3ζ was subcloned into pmTurquoise2-N1 (Addgene, Massachusetts, USA; plasmid # 54843) using restriction enzymes EcoRI (NEB, Massachusetts, USA; # R0101S) and AgeI (NEB; # R0552S), and also into pmTurquoise2-C1 (Addgene; plasmid # 60560) using EcoRI and KpnI-HF (NEB; # R0145S), respectively. pmTurquoise2-N1 and pmTurquoise2-C1 were gifts from Michael Davidson and Dorus Gadella^[33]. For the conjugation of Renilla luciferase-8 (Rluc8) to the C-terminus of 14-3-3ζ, Rluc8 was subcloned from pcDNA-Rluc8 (a kind gift from Dr. Jace Jones-Tabah and Dr. Terry Hébert, McGill University) and used to replace the mTurquoise of constructed 14-3-3ζ-mTurquoise using AgeI and NotI-HF (New England Biolabs, NEB; # R3189S). To conjugate Rluc8 to the N-termini of 14-3-3ζ, Rluc8 was subcloned to the original 14-3-3ζ-containing plasmid using EcoRI and KpnI-HF^[31]. Specific primers were used to generate truncated forms of BAD, which were then subcloned into pmCitrine-C1 (Addgene; plasmid #54587) and pmCitrine-N1 (Addgene; plasmid #54594) using EcoRI and BamHI (NEB; #R0136S). This process attached mCitrine, a yellow fluorescent protein (YFP), to the N- and C-termini of BAD, respectively. pmCitrine-C1 and pmCitrine-N1 were gifts from Robert Campbell, Michael Davidson, Oliver Griesbeck, and Roger Tsien^[34]. To construct bi-directional plasmids, Rluc8-conjugated 14-3-3ζ and BAD variants conjugated to mCitrine were subcloned to the multiple cloning site 2 (MCS-2) of the pBI-CMV1 vector (Takara, Shiga, Japan; # 631630), using EcoRI and XbaI (NEB; # R0145S) and to its MCS-1 using MluI-HF (NEB; # R3198S) and SalI-HF (NEB; # R3138S), respectively. Primers were designed with SnapGene Viewer (7.1.0) and synthesized by Integrated DNA Technologies (IDT, California, USA). Phusion® High-Fidelity DNA Polymerase (NEB; # M0530S) was used for PCR amplification. QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany; # 28704) was used to recover DNA products from agarose gels. T4 DNA ligase (NEB; # M0202S) was used to insert genes into vectors. Primer sequences are listed in Table 1.

Table 1

Primers and restriction enzymes for the construction of BRET sensor
Gene	Primer
5'-14-3-3	ATGGATAAAAATGAGC
3'-14-3-3	ATTTTCCCCTCCTTCTCCTG
5'-BAD	ATGGGAACCCCAAAGCAG
3'-BAD*	TGGATCCTGGGAGGGGGTG
3'-BAD-136F	CGCTGCCCAGAGATTGGG
5'-112-136F	ATGGAGACTCGGAGTCGC
5'-Rluc8	ATGGCTTCCAAGGTGTACGAC
3'-Rluc8	CTGCTCGTTCTTCAGCACGC
5'mCitrine	ATGGTGAGCAAGGGCGAG
3'-mCitrine	CTTGTACAGCTCGTCC
5'-mTurquoise	ATGGTGAGCAAGGGCG
3'-mTurquoise	CTTGTACAGCTCGTCCATGCC

Cell Culture

NIH-3T3 cells were kindly provided by Dr. Marc Prentki (CRCHUM, Montreal, Canada) and maintained in 25 mM glucose DMEM (Gibco, Massachusetts, USA; # 11995065) supplemented with 10%FBS (Gibco; # 12483020) and 1% penicillin-streptomycin (Gibco; # 15140122). HT-29 and Caco-2 cells were kind gifts from Dr. Petronela Ancuta (CRCHUM, Montreal, Canada) and maintained in Advanced MEM (Gibco; # 12492013) supplemented with 20% FBS and 1% penicillin-streptomycin or McCoy 5A media (Gibco; # 16600082) supplemented with 10%FBS and 1% penicillin-streptomycin, respectively. All cells were cultured in a humidified incubator at 37°C with 5% CO₂ and passaged upon reaching 70%-80% confluency. All studies were repeated with cells at different passages to ensure reproducibility.

Plasmid transfection

Plasmid DNA was amplified in Subcloning Efficiency DH5α competent cells (Invitrogen, Massachusetts, USA; # 18265-017) and extracted using QIAprep Spin Miniprep Kit (Qiagen; # 27104) or PureLink™ HiPure Plasmid Maxiprep Kit (Invitrogen; # K210007), following the manufacturer's protocol. The purity and concentration of each plasmid were assessed with a NanoDrop™ Lite Spectrophotometer (Invitrogen). Two days before transfection, cells were plated in 12-well plates with a density of 30,000 cells/well or proportionally adjusted based on the surface area of different well sizes. Plasmids were transfected with Lipofectamine™ 3000 Transfection Reagent (Invitrogen; # L3000015), according to manufacturer's instructions.

Confocal imaging and colocalization analysis

Confocal images were acquired with a Leica TCS-SP5 inverted microscope using an HCX PL APO CS 100x/1.4 oil objective with the Las-AF software. Excitation was performed using the 458nm and 514nm laser lines of an Argon laser for CFP and YFP, respectively, and a 633nm HeNe laser for TOPRO-3. Detection bandwidth were set to 468-500nm for CFP using a HyD detector under the Standard mode, 524-623nm for YFP using a PMT, and 643-748nm for TOPRO-3 using a HyD detector under the Standard mode. A sequential acquisition consisting of CFP and YFP in sequence 1, and TOPRO-3 in sequence 2 was performed. A line average 4 was applied for each sequence. Images were acquired at zoom 2.5 with a 400Hz scan speed and final images are 12bits, 1024x1024pixels (axial pixel size of 60nm).

To visualize BAD translocation, 35mm glass bottom dishes (ibidi, Gräfelfing, Germany; # 81158) were coated with 0.1 mg/mL Poly-d-lysine hydrobromide for 2 hours at 37°C. After coating, dishes were washed twice with sterilized water. Subsequently, cells transfected with BAD-mCitrine were seeded at a density of 20,000 cells per dish and incubated for 24 hours to allow for cell attachment. For live-cell imaging (Leica TCS SP5, Leica Microsystems, Wetzlar, Germany), cells were then incubated for 1 hour with the mitochondrial membrane potential dye tetramethylrhodamine ethyl ester (TMRE) (Invitrogen; # T668) at a final concentration of 100nM, according to the manufacturer's instructions. Super-resolution images were acquired on a Zeiss (Oberkochen, Germany) LSM-900 Airyscan inverted confocal microscope equipped with an environmental incubation system, laser lines including 488 and 561 nm, and an oil immersion 60×/1.40 Plan achromat objective. The signal from the YFP fluorophore (ex. 488 nm, det. 525–542 nm) and the TMRE (ex. 561 nm, det. 600–700 nm) were collected in super-resolution mode. The cell viability dye, DRAQ (BioLegend, California, USA; # 424101), was added to the medium at a 1:1000 dilution from the commercial stock and incubated at room temperature for 30 minutes prior to live-cell imaging. To quantify the colocalization of BAD-mCitrine signal and the mitochondria labelled with the TMRE, JACoP plugin in ImageJ was employed^[35]. A correlation analysis was performed based on Pearson’s coefficient between pixel-grey values for the two channels.

Protocol for the high-throughput screening (HTS) of drugs

The Z’-score, which evaluates the robustness of an HTS assay readout, was calculated according to standard protocols^[36]. Fo the HTS, 384-well plates were coated with 100ug/mL poly-l-ornithine (Sigma, Missouri, United States; # P3655) overnight at 37°C, followed by rinsing with sterilized water. NIH-3T3 cells were transfected with the BRET sensor two days prior to experimentation, followed by replating onto the pre-coated 384-well plates at a density of 20 000 cells/well. After 24 hours incubation at 37 ºC, growth media was replaced by Krebs buffer (pH 7.4). A robotic liquid handling system (Eppendorf, Hamburg, Germany; epMotion 5075) was used for appropriately diluting and adding the drug library (APExBIO, Texas, USA; # L1021, 2021 edition) to corresponding wells. Cells were incubated (37 ºC) for 3 hours with each drug at final concentrations of 200 µM, 20 µM, 2 µM, and 200nM, followed by the addition of coelenterazine-h (NanoLight Technology, Arizona, USA; # 301), the substrate of Rluc8, to each well (final concentration of 5 µM). After incubation for 10 minutes, emission from Rluc8 at 460 nm and mCitrine at 535 nm were measured (Perkin-Elmer, Massachusetts, USA; 1420 Multilabel counter). BRET ratios were calculated by the 535/460 nm ratio of emission readings^[37], and BRET reduction was quantified by normalizing the BRET ratio of drug-treated cells BRET_drug to the BRET ratio of solvent-treated control cells BRET_control, using the formula: (BRET_control – BRET_drug) / BRET_control.

Cell death and apoptosis assays

Cells were plated at a density of 20,000 cells/well into 96-well plates (PerkinElmer; # 6055302) and incubated overnight at 37 ºC before the addition of the drugs. One hour prior to imaging, Hoechst 33342 (Invitrogen, # H3570) and propidium iodide (Invitrogen; # P3566) were added to the wells to achieve final concentrations of 50 ng/ml and 0.5 µg/ml, respectively. To measure apoptosis, AlexaFluor 488-conjugated Annexin V (ThermoFisher, Massachusetts, USA; # A13201) was also added at a dilution of 1:400 from the commercial stock. A high-content imaging system (PerkinElmer; Operetta CLS), was used to quantify cell death and apoptosis by calculating the number of propidium iodide-positive cells relative to Hoechst-positive cells, and the number of Annexin V-positive cells relative to Hoechst-positive cells, respectively. To assess caspase-dependent apoptosis, cells were pre-treated for 30 minutes with the caspase inhibitor Z-VAD-FMK (BD biosciences, New Jersey, USA; #550377; 20uM) or the negative control Z-FA-FMK (BD biosciences; #550411; 20uM) prior to the incubation of drugs.

Computational Methodology

Structure Preparation

Structures of 14-3-3ζ (PDB Code 2C1J, chosen as it had no missing components or loops), mCitrine (PDB Code 3DQO) and Rluc8 (PDB Code 7OMO) were obtained from the protein data bank while full length BAD was calculated using AlphaFold under standard parameters (Deepmind, London UK)^[38–40]. BAD is a poorly structured protein and a full-length crystal structure is unavailable. Solvent and if present, ligands, were removed from available structures that were prepared using the protein preparation module in Schrödinger (New York, NY). The protonation states of proteins were set, assuming a pH of 7.4, and the structures were then minimized. As the confidence of the structure generated by Alphafold was low, full length BAD was subjected to a 500ns Gaussian accelerated molecular dynamics (GaMD) simulation to ensure an equilibrated structure. The BAD 112-136F fragment spanned M104-A144 and was similarly obtained using Alphafold and prepared using Schrodinger. As the fragment is smaller and more flexible, three 500ns GaMD simulations were performed to ensure proper sampling of conformations. GaMD simulations have been shown to significantly enhance sampling and has been used to accurately fold peptides^[41]. To generate the 14-3-3ζ-RLuc and BAD-mCit constructs, the missing N terminal sequence of 14-3-3ζ was built using Schrödinger and connected to Rluc8 using the Protein Linker Design module. Similarly full length BAD and truncated BAD were attached to mCitrine.

Gaussian Accelerated Molecular Dynamics

Molecular dynamic (MD) simulations were performed using Amber20^[42]. Structures were prepared using Leap with the ff19SB forcefield for the proteins and the OPC water model^[43]. The protein was surrounded in a 12Å octahedral periodic solvent box, and charges were neutralized using either Na⁺ or Cl⁻ ions as appropriate. Na⁺ and Cl⁻ were then added to achieve a 0.15M ionic strength concentration for the water box. Three minimization steps, each with 10,000 steps conjugate gradient & 10,000 steps steepest descent with decreasing restraints on the protein, of 100, 3 and 0 kcal·mol^− 1·Å⁻² l were performed. For the remaining steps, SHAKE was used to constrain hydrogen bonds, along with a nonbonded cutoff of 12Å, with a 2 fs time step^[44]. Pressure was regulated using the Berendsen barostat, and the Langevin thermostat with a collision frequency of 2 ps-1 was used for temperature. The system was then heated to 300K over 50ps with a harmonic restraint of 1 kcal·mol^− 1·Å⁻² on the protein. This was followed by two density equilibration steps, 2 fs time steps, to allow for adjustment of periodic box conditions. The first for 50 ps with a restraint of 2 cal·mol^− 1·Å⁻² on the protein and the second for 200 ps with no restraint. For the GaMD simulation, dual boost on both dihedral and total potential energy was used with the threshold energy set to the lower bound. For equilibration, 200,000 steps of conventional MD with a 2 fs time step were performed to equilibrate the system, followed by 1,000,000 steps of initial conventional MD simulations to collect potential energies. 800,000 steps of preparation biasing MD steps were performed followed by 50 ns of GaMD equilibration. This was followed by 450 ns of GaMD simulations. First and second potential boosts of 3 kcal·mol^− 1 were utilized for all GaMD simulations.

Clustering

Trajectories were clustered using cpptraj and the kmeans clustering algorithm based on the backbone atoms of the protein. For full length BAD, the clusters were similar with minor differences in the loop containing residues Ser112 and Ser136 and the orientation of the residues^[45]. The lowest energy top pose was chosen for use in further studies as Ser112 and Ser136 both point away from the protein core and are solvent exposed, while in the other clusters the side chains are rotated inwards towards BAD and cannot form interactions with 14-3-3ζ. For truncated BAD, due to the flexible nature of the protein all of the top 5 poses were used for further study.

Docking Protocol

Structures of small molecule inhibitors were prepared using the ligprep module in Schrödinger at a pH of 7.4. If multiple protonation states were possible, all were used in subsequent docking. Induced fit docking was performed targeting the groove of 14-3-3ζ using the standard protocol in Schrödinger with residues within 5Å of ligand poses being refined with XP precision for Glide redocking.

Protein-Protein Docking

Phosphorylated full length BAD or truncated BAD was docked to the 14-3-3ζ dimer using the Piper protein-protein docking module in Schrödinger. No constraints were used for full length BAD, while constraints on the BAD fragment were used to ensure that the BAD fragment docked to 14-3-3ζ and not the conjugated mCitrine.

Molecular Dynamics Simulations of 14-3-3ζ-BAD Complexes

To observe the distances and interactions between mCit and Rluc8, MD simulations were performed. Due to the large size of the systems, explicit solvation could not be used due to runtime constraints. Instead, implicit solvation was used. Ff14SB was used for the protein with the OBC-2 solvent model with a salt concentration of 0.15M, no nonbonded cutoff and a 2fs time step. Similar minimization and heating steps we’re perfomed as above and similar parameters used for equilibration however periodic boundary conditions were not used.

Statistical analysis

R programming (4.3.2) was used for generating scatter density plots. Harmony® high-content analysis software (4.9) was used for processing data collected by the Operetta CLS system. Data were analyzed by GraphPad Prism (version 9.5.0), and the statistical significance of the results was examined by one-way ANOVA with a Dunnett post hoc test or two-way ANOVA with a Tukey post hoc test. Statistical significance in the figures is shown as follows: *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Data are presented as mean ± SEM.

Inhibition of 14-3-3 proteins promotes BAD translocation

The canonical mechanism by which 14-3-3 proteins regulate cell survival is through the sequestration of pro-apoptotic BCL-2 proteins, such as BAD, in the cytoplasm^[46]. To visualize how 14-3-3 protein inhibition impacts BAD localization and subsequent translocation to mitochondria, BAD-mCitrine was transiently expressed in NIH-3T3 cells, followed by incubation with TMRE (tetramethylrhodamine, ethyl ester) to label mitochondria. R18 and FTY720, which are established 14-3-3 protein inhibitors that act through distinct mechanisms of action^{[47, 48]}, were used to inhibit 14-3-3 proteins. BAD-mCitrine-expressing cells or control mCitrine cells were exposed to R18 (10 µM) and FTY720 (2 µM) for 24 hours, and localization of mCitrine was visualized by confocal microscopy. A BAD mutant containing S112A and S136A double mutations (BAD-AA), which prevent 14-3-3 protein:BAD PPIs, was used as a positive control. In the absence of 14-3-3 protein inhibitors, BAD-mCitrine was distributed diffusely throughout the cytoplasm, and inhibition of 14-3-3 proteins with FTY720 and R18 promoted BAD translocation to TMRE-labelled mitochondria. This redistribution was also observed in cells expressing the BAD-AA-mCitrine mutant (Fig. 1A). An increased degree of colocolization between the signal of mCitrine and TMRE was found in BAD-AA-expressing cells, and BAD-expressing cells treated with R18 and FTY-720, as determined by Pearson's correlation coefficient (R) (Fig. 1B). These observations indicate the essential role of 14-3-3 proteins in the cytoplasmic sequestration of BAD.

Development of a BRET-based living-cell sensor to detect interactions between 14-3-3ζ and BAD

To measure 14-3-3 protein:BAD PPIs in living cells, we focused our efforts on developing a BRET-based reporter whereby 14-3-3ζ and BAD would be conjugated to Rluc8 and mCitrine, respectively (Fig. 2A). As BRET is highly dependent on the close proximity of the donor (Rluc8) and acceptor (mCitrine), we examined all possible pairs. This involved generating six BAD-mCitrine constructs: the label affixed to either the N- or C-termini of either full-length murine BAD, or of two truncated versions of BAD (Fig. 2B, Supplemental Fig. 1). These two truncated BAD fragments were BAD-136F, which comprises the N-terminus of BAD to residue A144, and BAD-112-136F which spanned from M104 to A144. We also ligated Rluc8 to both the N- and C-termini of 14-3-3ζ (Fig. 2C, D).

To confirm the interactions between 14-3-3ζ-Rluc8 and BAD-112-136F-mC, we introduced single serine-to-alanine mutations at either S112 (BAD-112-136F-112A-mC), S136 (BAD112-136F-136A-mC), or a double S112/136AA mutation (BAD-112-136F-AA-mC), as these mutations are known to prevent binding of 14-3-3 proteins to BAD. We found that BAD-112-136F-136A-mC and BAD-112-136F-AA-mC could not elicit a detectable BRET signal, which indicated an inability for 14-3-3ζ-Rluc8 to interact with either BAD variant (Fig. 2E).

Using confocal microscopy, the subcellular localizations of 14-3-3ζ and BAD-112-136F were determined. Conjugation of (mT)urquoise to 14-3-3ζ revealed that 14-3-3ζ is primarily restricted to the cytoplasm (Fig. 2F). We next compared the subcellular localization of BAD-mC to BAD-112-136F-mC and found that the fragment was similarly restricted to the cytoplasm. In contrast, BAD-112-136F-AA-mC was distributed throughout the cell, including the nucleus (Fig. 2F).

With our observations that co-transfection of 14-3-3ζ-Rluc8 and BAD-112-136F-mC plasmids resulted in a detectable BRET signal, we constructed a bi-directional BRET sensor plasmid whereby the donor and the acceptor were expressed at near stoichiometric ratios due to equal promoter activities (Fig. 3A). Co-expression of BAD-112-136F-mCitrine and 14-3-3ζ-Rluc8 in NIH-3T3 cells resulted in an average of 34.56% increase in BRET compared to the co-expression of 112-136F-AA-mCitrine and 14-3-3ζ-Rluc8 (Fig. 3B). To test the performance of our bi-directional BRET sensor, we evaluated the capacity of our sensor to detect 14-3-3 protein:BAD interactions with two well-recognized 14-3-3 inhibitors, FTY720 and I,2–5^[49]. Treatment of sensor-expressing cells with FTY720 and I,2–5 significantly reduced BRET in a dose-dependent manner (Fig. 3C, 3D). To optimize conditions for the following HTS, we next determined the optimal incubation time and found that 3 hours was required to reduce the magnitude of BRET to the same degree as BAD-112-136-AA, which indicated a maximal effect (Figs. 2E,3E).

In silico modeling of BAD and 14-3-3ζ interactions confirms the performace of the BRET sensor

To better understand the molecular interactions between BAD or BAD-112-136F and 14-3-3ζ, in silico approaches were used. Examination of the crystal structures of 14-3-3ζ shows that the N-terminus sits at the interface of the 14-3-3ζ dimer and is located on the “rear” of the protein, opposite the binding groove (Fig. 4A). In contrast, the C-terminus is adjacent to the binding groove, potentially much closer to any BAD ligand, and this would increase the probability of interactions with mCitrine (Fig. 4B). Models of 14-3-3ζ-Rluc8 and BAD-mC and BAD-112-136F-mC were built and Gaussian accelerated MD simulations were performed to sample protein conformations. Clustering was performed to obtain representative structures. All the top clusters obtained for full length BAD were similar and had two conserved alpha helices, but the domain between residues 112–136 is highly flexible. Ser112 and Ser136 can rotate and were inaccessible in some poses, so the topmost (lowest energy) cluster was used for further analysis as its residues were solvent exposed allowing for interactions while in some of the other clusters they were blocked (Fig. 4C,D,E). As a shorter peptide, and focused on the unstructured region, the BAD-fragment was significantly more flexible and may not have any actual defined structure, so the top five poses were used in the studies to sample a wide variety of possibilities (Supplemental Fig. 2A,B). Protein-protein docking was performed between 14-3-3ζ and full length BAD and BAD-112-136F (Fig. 4F). Full length BAD can bind with a phosphorylated serine residue in each of the aliphatic grooves of the 14-3-3ζ dimer, interacting with R56, and is adjacent to K49 and R127, which is consistent with previous reports that BAD can bind both subunits simultaneously (Fig. 4G,H)^[12].

Molecular dynamic situations were then performed to compare the structures of 14-3-3ζ-Rluc8 when bound to BAD-mC or BAD-112-136F-mC. With 14-3-3ζ-Rluc8 and BAD-mC or mC-BAD, the C-termini containing mCitrine extend out to the side, away from Rluc8 on 14-3-3ζ (Supplemental Fig. 3). In contrast, docking of BAD-112-136F-mC had mCitrine positioned much closer to Rluc8 (Supplemental Fig. 2C). The top cluster of BAD-112-136F-mC sits significantly closer to Rluc8 and binds 14-3-3ζ via S136, consistent with experimental results. Molecular dynamic simulations were also performed on full length BAD-mCitrine bound to 14-3-3ζ-Rluc8 and the average distance between Rluc8 and mCitrine over the course of the simulations was ~ 80Å based on center of mass of Rluc8 and mCit and potentially explains the low BRET signal (Supplemental Fig. 4C). The pairing of 14-3-3ζ-Rluc8 with BAD-112-136F-(mC)itrine generated the most robust BRET signal compared to other combinations (Fig. 2C,D, G,H). In contrast, the pairing of 14-3-3ζ-Rluc8 with BAD-112-136F-mC generated the most robust BRET signal compared to other combinations (Fig. 2D,E).

Screening for FDA-approved pro-apoptotic drugs

The successful implementation of our BRET-based sensor in living cells permitted the screening of previously-approved drugs (PADs) that could disrupt 14-3-3 protein:BAD interactions (Fig. 5A). The re-purposing or re-positioning of these drugs could lead to the successful identification of new functions in the induction of cell death. The primary screen was conducted in a 384-well plate format, and 1971 compounds were tested (Fig. 5B). We found the median BRET reduction caused by compounds capable of reducing BRET was approximately 25%, and there were 416, 162, 31, and 16 PADs that reduced the BRET signal by 25% at concentrations of 200 µM, 20 µM, and 2 µM, respectively (Fig. 5B). We next adapted our screen into 96-well plates to re-screen PADs that were effective at 20 µM, a concentration commonly used in HTS assays^[50]. Of these, 101 PADs showed consistent results and were further assessed for their capacity to induce cell death in NIH-3T3 cells via Hoechst/propidium iodide incorporation assays (Fig. 5C). Scatter density plots were used to visualize the relationship between BRET reduction and the induction of cell death at 24 and 48 hours (Fig. 5D, 4E), and 41 PADs were found to reduce BRET by more than 34%, which is consistant with the reduction in BRET caused by 112-136F-AA-mCitrine (Fig. 3B), and induce cell death greater than 30% (Zone A; Fig. 5D,E).

14-3-3ζ inhibits BAD-induced cell death in CRC cells

To explore the potential of identified hits to treat diseases by triggering apoptosis, CRC was selected as our disease model. We first determined if disruption of 14-3-3 protein:BAD PPIs could induce cell death in CRC cells. Colorectal cell lines, Caco-2 and HT-29, were transfected with either BAD-mCitrine or BAD-AA-mCitrine, followed by incubation with FTY-720 or R18. BAD translocation to mitochondria following 14-3-3 protein inhibition was similar to what was observed in NIH-3T3 cells (Figs. 1; 6A-D). Additionally, cell death was assessed following the over-expression of BAD or BAD-AA. After 72 hours post-transfection, a significantly higher degree of cell death was observed in cells co-expressing 14-3-3ζ and BAD-AA, compared to those only expressing 14-3-3ζ (control) or co-expressing 14-3-3ζ and BAD (Fig. 6E,F). These observations imply that when 14-3-3ζ is unable to sequester BAD due to S112/136A mutations or due to the presence of 14-3-3 inhibitors in CRC cells, BAD translocates to the mitochondria to induce cell death.

Examination of hits in CRC cells

Among the 41 identified PAD hits from our primary screen in NIH-3T3 fibroblasts, some were immediately found to be unsuitable for systemic administration if re-purposed as chemotherapy. For example, crystal violet is a synthetic dye used for cell staining^[51]; whereas some PADs are topical treatments, such as cetylpyridinium chloride, benzethonium chloride, and thonzonium bromide^[52–54]. These are all problematic compounds, and meet the criteria as pan-assay interference compounds (PAINS), and likely should not have been included in the library, as their activity is due to non-specifc effects^[55]. In addition, various other PADs are also already used as chemotherapeutics, such as entrectinib, ceritinib, and ponatinib, so the cytotoxicity observed would be expected from their other mechanism of action^[56]. After excluding these two classes of PADs, 25 were left to be assessed on Caco-2 and HT-29 CRC cells. Of these, 15 PADs caused more than 30% of cell death at 24 hours or 48 hours in both cell lines (Fig. 7A), and a further filtering to remove pro-drugs, salts, and low potency narrowed our list to 13 PAD hits (Fig. 7A). Dose-response studies were then conducted with these 13 agents (Fig. 7B-6N). Lomibuvir, terfenadine, penfluridol, and lomitapide were found to be the most effective, as they significantly induced cell death at concentrations as low as 5 µM (Fig. 7G, I, L, N). Although lomibuvir consistently induced cell death at different concentrations, its efficacy in the magnitude of cell death attained was inferior to other candidates (Fig. 7N), and it was excluded from further study.

Table 2

Cell death observed in respective cell line and docking score obtained from induced fit docking
Compound	%Cell death		Docking score (kcal/mol)
Compound	Caco-2	HT-29	Docking score (kcal/mol)
Azelnidipine	13.7	91.6	-7.14
Bardoxolone methyl	19.2	67.5	-7.43
Cinacalcet	49.4	70.3	-4.75
Clomiphene	34.7	52.0	-5.21
Doramectin	32.3	51.6	-4.95
Dronedarone	86.9	79.0	-6.48
Efavirenz	20.5	22.0	-3.43
Embelin	1.7	51.8	-4.86
FTY720	67.9	69.1	-6.06
Ivermectin	5.2	78.9	-4.98
Lomitapide	81.7	72.9	-7.48
Moxidectin	8.4	78.4	-3.82
Nebivolol	45.0	67.3	-6.66
Oxethazaine	13.9	53.1	-4.55
Penfluridol	90.0	92.2	-8.31
Pimozide	59.9	64.6	-6.5
Saikosaponin A	1.2	87.6	-6.56
Simeprevir	5.5	70.9	-4.57
Tamoxifen	36.5	63.3	-5.07
Terfenadine	79.7	89.6	-5.91
Vortioxetine	52.1	52.1	-5.31
VX-222 (VCH-222, Lomibuvir)	36.3	78.5	-5.9

Observed activity of identified compounds correlates with their predicted interactions with 14-3-3

We also performed induced fit docking to explore the possible binding modes of the compounds (Table 2). BV02 (IC₅₀ = 5.2µM ), I,2–5(IC₅₀ = 2.6µm) and FTY-720 were used as reference compounds, as they are known 14-3-3 inhibitors^{[49, 57]}. While no crystal structures of these complexes exist, the binding mode of BV02 obtained in our study matches the binding mode reported previously in similar studies (Supplemental Fig. 5A)^{[57, 58]}. BV02, I,2–5 and FTY-720 had docking scores of 6.69, 7.21, and 6.06 kcal/mol respectively. BV02 and I,2–5 are both negatively charged and binding is largely mediated by the positively charged residues in the 14-3-3 binding groove (Supplemental Fig. 5A,C). BV02 forms salt bridges with both R56 and R127 and additional hydrogen bonds with K49, K120 and N173. Similarly I,2–5 forms salt bridges with R56, R127 and K49 through the phosphate and additional hydrogen bonds with K120 and N173. FTY720, however, is positively charged and instead forms hydrogen bonds through the OH groups with S45, K120, Y125 and Y128 and a π-cation interaction with K49(Supplemental Fig. 5C). The interaction of FTY-720 was surprising, as it has been established that FTY-720 potently promotes the dissociation of 14-3-3 protein dimers^[48].

Despite the preference for negatively charged ligands, Penfluoridol, Lomitapide and Terfenadine have good docking scores (-8.31, -7.48 and − 5.91), though this is largely driven through aromatic H-bonds and π-cation interactions rather than salt-bridge formation (Supplemental Fig. 5D-F). Penfluridol forms a hydrogen bond with N173 as well as an aromatic H bond, and an aromatic H bond with E131. It could also potentially form several π-cation interactions with adjacent R58, R60 and R127 however this was not observed due to the limited flexibility in IFD. Terfenadine can form several H-bonds with R56, N173, K120 and D120 as well as a π-cation with R60, aromatic H bonds with S45 and D124 and a π-π stacking interaction with P117. From the compounds examined, Penflridol, and Lomitapide had the best docking scores which correlated with observed experimental results where they induced the most cell death (Table 2; Fig. 7G,I). Badoxolone and azelnipidine also had good docking scores, while terfenadine was lower at -5.91 (Table 2). Generally, with the exception of Azelnipidine, Bardoxolone, and Saikosaporin A, compounds that had similar or better docking scores than FTY-720 strongly induced cell death in Caco-2 cells after 48 hours and docking scores correlate well with observed cell death (Fig. 7O). Some compounds, cinacalcet − 4.75, Clomiphene, -5.21, doramectin − 4.95, had weaker docking scores and were found to induce moderate levels of cell death. Overall, the docking score appears to be a good predictor of whether compounds were capable of inducing cell death in Caco-2 cells.

Verifying the capacity of terfenadine, penfluridol, and lomitapide to induce apoptosis in cancer cells

Cell death can emerge from a variety of different pathways: including apoptosis, necrosis, and autophagy^[1]. To ensure that these hit PADs induced apoptotic cell death via the disruption of 14-3-3 protein:BAD PPIs, we further assessed the relationship between apoptosis via Hoechst/propidium iodide/Annexin V incorporation and the mitochondrial translocation of BAD upon drug exposure. Given that caspase activation is a hallmark of apoptosis, a pan-caspase inhibitor, Z-VAD-FMK, was used to attenuate hit PAD-induced apoptosis^[59]. According to time courses of PI and Annexin V incorporation (Figs. 8A, B, F, G, K, L), differences in the kinetics of cell death or apoptosis were seen with the PAD hits in HT-29 and Caco-2 cells, such that terfenadine and penfluridol were able to induce cell death and apoptosis more rapidly than lomitapide. To assess caspase-dependent apoptosis, HT-29 and Caco-2 cells were treated for 24 hours with terfenadine (10 µM) (Fig. 8C,D) and penfluridol (10 µM) (Fig. 8H,I), or for 48 hours for lomitapide (10 µM) (Fig. 8M,N), along with either Z-VAD-FMK or its control, Z-FA-FMK^[60]. Significantly diminished hit PAD-induced propidium iodide and/or annexin-V incorporation was observed in cells pre-treated with Z-VAD-FMK compared to those pre-treated with either DMSO or a control inhibitor Z-FA-FMK, implying caspase activation following drug administration. Additionally, confocal imaging showed that mitochondrial translocation of BAD occurred after the addition of the drugs (Fig. 8E,J,O; Supplemental Fig. 6), demonstrating that disruption of 14-3-3:BAD PPIs by the identified hits induces apoptosis.

Cancer arises from uncontrolled cell proliferation—one of the primary mechanisms of chemotherapies is to induce cell death in proliferating cells. The overexpression of 14-3-3ζ and its related isoforms in the context of cancer has long been associated with poor clinical outcomes due to increased cell survival in the face of chemotherapeutic treatment^{[15, 61]}. Although inhibiting 14-3-3ζ triggers apoptosis in cancer cells, there are currently no approved therapeutics that target 14-3-3ζ:BAD interactions^[17]. The primary aim of this study was to explore the possibility of identifying anti-CRC compounds by their capacity to interrupt 14-3-3ζ:BAD interactions, but for this to be achieved, a suitable, mechanistically-specific assay is needed and was not available. We created this required tool by innnovating a biosensor capable of detecting 14-3-3ζ:BAD PPIs. Importantly, our BRET-based biosensor was capable of detecting 14-3-3ζ:BAD PPIs in living cells, which provides physiological relevance.

In this study, a short fragment of murine BAD was used to construct the BRET sensor in place of full-length BAD. Previous research has demonstrated that BAD overexpression leads to apoptosis in various cell types^{[62, 63]}, and an advantage of using the BAD-112-136F is its inability to interact with its client BCL-2 proteins due to the lack of BH-3 domain to induce apoptosis^[64]. Furthermore, it was not possible to detect BRET when the Rluc8 acceptor and mCitrine donor were fused to 14-3-3ζ and full-length BAD, respectively. Since energy transfer between Rluc8 and mCitrine requires a distance shorter than 10nm, we assumed that the inability to detect BRET was due to the distance between the fusion sites and interaction sites. In reported crystal structures, the N-terminus of 14-3-3ζ is positioned on the rear of the protein on the opposite side of the aliphatic groove and sits on the interface of the 14-3-3ζ^[38]. This could potentially interfere with dimer formation and function and due to its location, Rluc8 could be blocked from interacting with the mCitrine attached to bound ligands. Additionally our modelling suggests that when full length BAD-mCitrine bound to 14-3-3ζ, mCitrine is positioned far away from from Rluc8 at the end of a flexible tether. This increased distance, and the low occupancy of any state within the BRET distance to Rluc8, predicts that there would be no meaningful BRET signal. TR-FRET, a technique where a fluorescent protein is used as an energy donor instead of luciferase, has been previously used to screen for disruptors of 14-3-3:BAD PPIs, but a limitation was the fusion of the FRET acceptor to the serine residue that mediates interactions^[24]. Thus, we chose not to pursue this option as it would directly interfere with the needed binding mode and generate false positives. Instead, different truncated forms of BAD were generated to identify the optimal fusion strategy for measuring BRET efficiency between 14-3-3ζ-Rluc and BAD-truncate-mCitrine. In our modelling of BAD-112-136F-mC, we saw that the fragment could bind 14-3-3ζ with mCitrine positioned much closer to Rluc8 than in full length BAD. Although the peptide is more flexible and allows the fluorophore to move through a greater range, almost all the lowest energy conformations keep the Rluc8 and mCitrine within the necessary BRET distance, leading to a strong signal upon binding. Given that the interactions between 14-3-3ζ and BAD occur between the C-terminus of 14-3-3 and S112 or/and S136 of BAD, it was not surprising that the combination of 14-3-3ζ-Rluc8 and 112-136F-mCitrine represented the optimal combination in the constructing the BRET sensor^{[12, 65]}.

Since it was uncertain if the BAD-112-136F could represent the full-length BAD in its interactions with 14-3-3ζ, further evaluations were conducted by introducing mutations at Ser-Ala mutations at S112 and/or S136. Unlike the S112A mutation, S136A significantly disrupted the association between 14-3-3 and BAD-112-136F, as indicated by reduced BRET. This aligns with a prior study suggesting that S136, rather than S112, primarily mediates 14-3-3:BAD interactions^[66]. This hypothesis is completely in line with our computational modelling as S136 on BAD’ engages in important H-bonds with R56, while S112 appears to merely form a weaker electrostatic interaction with R127 and K49 of 14-3-3ζ. We see no meaningful predicted difference in these key binding motifs between the full length BAD and the truncated versions from the in silico calculations. Additionally, in contrast to 112-136F-AA, 112-136F is specifically sequestered in the cytoplasm. This indicates that 14-3-3ζ interacts with this truncated form similarly to how it would interact with full-length BAD, but only if the serine residues crucial for 14-3-3:BAD interactions remain intact.

To assess the capacity of this sensor to discover disruptors of 14-3-3ζ:BAD PPIs, we introduced two well-known 14-3-3 inhibitors, FTY720 and I-2,5, and both compounds signficantly reduced BRET. It is worth mentioning that we did not assess if there were any differences in affinity between 14-3-3ζ:BAD and 14-3-3ζ:112-136F. Nevertheless, drugs identified to disrupt 14-3-3ζ:112-136F PPI should be effective, as this smaller fragment likely accesses the binding groove of 14-3-3ζ more readily than the full-length BAD^[67]. Additionally, it is worth mentioning that the high homology among different 14-3-3 isoforms permits them to share client proteins and form homo- or hetero-dimers, suggesting that the identified PADs are highly likely to disrupt PPIs not only between BAD and 14-3-3ζ but also between BAD and other 14-3-3 isoforms^[65]. An important caveat of our reporter system is that we cannot experimentally distinguish if PADs directly block or disrupt the amphipathic groove of 14-3-3ζ where PPIs occur or if PADs promote the dephosphorylation of Ser112 and Ser136F on the BAD fragment^{[10, 62, 63]}. However, the in silico calculations strongly suggest that the hits, for the most part, do simply work through direct competitive target engagement; although additional studies, both experimental and in silico, combined with a structure-activity relationship campaign and/or confirmatory experimental structural biological data, are required to examine the mechanisms of action of each identified PAD hit.

The efficiency of utilizing HTS to develop novel anti-cancer compounds has been underscored by the discovery of sorafenib, palbociclib, and ABT-199^[68–71]. However, a recognized drawback of this drug discovery approach is the increased risk of false positives and false negatives due to the lack of replication and the use of miniaturized reaction systems^{[72, 73]}. To increase the chances of identifying potential compounds, we first ensured the robustness of our sensor by achieving a Z-factor greater than 0.5^[36]. Second, to minimize false negatives, we tested each compound twice at four different concentrations in our primary screens but only recorded the highest BRET reduction for each concentration. After evaluating the capacity of identified compounds to induce cell death in NIH-3T3 fibroblasts, a group of drugs that decreased BRET by more than 34% and triggered more than 30% of cell death emerged as potential hits capable of killing target cells by disrupting 14-3-3ζ:BAD PPIs. Interestingly, this BRET reduction aligns with that caused by 112-136F-AA-mCitirine, suggesting that the ability of a compound to completely dissociate the 14-3-3ζ:112-136F complex is indicative of its potential to induce cell death. Another group of drugs that were capable of reducing BRET reduction by more than 34%, but without notable efficacy in inducing cell death in NIH-3T3 cells, arose from our screens. A possible explanation for this is that our screening is based on a cell-based assay^[50], and the complex intracellular environment makes it challenging to determine whether the dissociation of 14-3-3ζ:BAD resulted from a direct inhibitory action on 14-3-3ζ or BAD, or an indirect effect on the upstream signaling pathways that promote 14-3-3ζ:BAD interactions, or possibly a general mechanism of interference of the assay through non-specifc absorption to Rluc8. Therefore, other than disrupting 14-3-3:BAD interactions, these compounds may have additional effects, such as up-regulating anti-apoptotic BCL-2 proteins, which promote cell survival^[11].

Although the altered expression of 14-3-3 in CRC has been reported in several studies, the role of 14-3-3ζ:BAD in the survival of CRC cells remains unclear^{[17, 74, 75]}. Our research provides the first evidence that disruption of 14-3-3ζ:BAD PPIs can promote CRC cell death. We used two representative CRC cell lines, Caco-2 and HT-29, to validate our assay^[76–78]. Terfenadine, penfluridol, and lomitapide were advanced as the most promising hits after conducting dose-response studies with the 13 most potent compounds.

To support our experimental work, we modelled the possible binding modes of hits to 14-3-3ζ and compared them with the most likely binding modes of known inhibitors BV-02 and I,2–5. All compounds are capable of fitting the aliphatic groove but had varying docking scores, and many were predicted to only have very moderate affinity. None of the compounds were predicted to have nM affinity based on the docking scores. Compounds BV-02 and I-2,5 were designed take advantage of the phosphate binding region which contains numerous Arg and Lys residues and have their binding largely driven by the formation of hydrogen bonds and salt bridges. However, for many screened compounds, aromatic H-bonds and π-cation interactions played a significant role, especially in top compounds Penfluridol and Lomitapide. Interestingly the top two compounds experimentally, penfluridol and lomitapide, had the best docking scores. Docking scores also align well with observed Caco-2 cell death at the 48 hours mark and better docking scores correlated with higher levels of cell death. Exceptions were Azelnipidine, Bardoxolone methyl, and Saikosaponin A; however, as these did induce cell death in HT29 cells but are not predicted to have strong binding, this may be due to completely separate cytotoxic mechanisms that are independent on 14-3-3ζ and BAD^[79–82]. Surprisingly, some compounds also had similar or better docking scores than known inhibitors, which suggested that they indeed bound to 14-3-3ζ. We also note that this computational model does not account for any variability in cell permeability, localization, cell-driven degradation, or more importantly off-target effects. Corrections for these features likely would improve linearity, and it must be remembered that all compounds in the assay are existing drugs with established biological activity through target engagement with other proteins. Despite these caveats, the model still shows predictive power in separating effective from less effective compounds that are all active over a tight range. This suggests that both the model is likely reasonable and that the observed cell toxicity is at least partially ascribable to this mechanism, rather than due to the known other activity of these compounds.

Additionally, we also collected data on the efficacy of all 101 compounds that were identified from the primary screen in inducing cell death in these two CRC cell types at 20uM (Supplemental Table 1). These data would be invaluable for future research exploring the different capacities of drugs to target CRC cells, and provide the essential information needed to initiate a rational drug design campaign starting from any of these hits.

To confirm that lead PADs induce apoptotic cell death by disrupting 14-3-3ζ:BAD PPIs, we further tested whether inhibiting caspase activation could mitigate lead PAD-induced cell death and if these lead PADs could promote the mitochondrial translocation of BAD. In most cases, Z-VAD-FMK treatment prevented increases in PI-positive and Annexin-V-positive cells; however, in lomitapide-treated Caco-2 cells, Z-VAD-FMK had no effect on propidium iodide incorporation, despite preventing annexin V incorporation (Fig. 8M). This is likely due to the kinetics of lomitapide in Caco-2 cells whereby the early stages of apoptosis, marked by the binding of annexin V to phosphatidylserine, is being observed, without a loss of membrane integrity that is needed for propidium iodide entry into the cell^[83]. As Z-VAD-FMK cannot inhibit necroptosis, a process where cells shift to necrosis when they cannot complete apoptosis^{[84, 85]}, other forms of cell death may also be occuring. Nevertheless, confocal imaging showed that all three lead PADs trigger the mitochondria translocation of BAD, confirming apoptotic cell death.

With the variability of 14-3-3 protein expression across individuals, a personalized medicine approach could be undertaken to explore the potential of lead PADs to treat CRC by careful evaluation of tissues from people living with CRC^[75]. A significant advantage of our screening strategy is our focus on repurposing drugs from PADs. Therefore, all of our identified hits have been tested for their safety in prior phase 1 clinical trials. Interestingly, penfluridol and lomitapide have been previously suggested to have potential for treating CRC, but their mechanisms were not fully defined—their designed mechanisms of action are not related to driving cell death^[86–88]. Our study not only further validates their therapeutic value but also provides insight into the mechanism by which these drugs may ameliorate CRC. Nevertheless, additional in-depth pre-clinical studies in animal models are required. We recognize that PADs may also have effects in other cell types, and improving cell type specificity is clearly warranted^[8]. A potential approach could also be to adopt a localized use of PADs to treat colorectal tumors, whereby intratumural drug administration might also enhance the specificity of these compounds^[89].

With the use of a novel BRET-based sensor to monitor 14-3-3ζ:BAD interactions in living cells, we successfully identified terfenadine, penfluridol, and lomitapide as having the abilities to disrupt 14-3-3ζ:BAD interactions and induce apoptosis of CRC cells. Although further research is critical to validate the ability of these compounds to ameliorate CRC in animal models and in humans, these hits represent potential chemical backbones that can be modified and translated into new chemical entities for the treatment of CRC. In addition, our screening approach has shown significant potential for the discovery of novel therapeutics for the treatment of other apoptosis-related diseases.

Disclosure statement: SH, DM, and LD have nothing to disclose. GEL is a consultant for Diogenix and Ambagon Therapeutics and has received research funding from Inversago Pharma, a Novo Nordisk Company. GAR has received grant funding from, and is a consultant for, Sun Pharmaceuticals Inc. JFT is CEO of Binary Star Research Services, which has no interests in the subject of this work, holds no relevant IP, and neither received, nor provided, any funding associated with this work. BSRS had no input into the methods used, or the conclusions of this work.

CREDIT Statement

Conceptualization: GEL, JFT; Funding acquisition: GEL, JFT, GAR; Investigation — Biology, SH, LDS; Investigation — in silico analysis, DM;; Writing original draft, SH, DM, GEL; Writing–review and editing, All authors. GEL is the guarantor of this work.

Funding

GEL was supported by CIHR Project (PJT-186121) and NSERC Discovery (RGPIN-2017-05209) grants, as well as funding from the Centre d’expertise en diabète du CHUM. GEL holds the Canada Research Chair in Adipocyte Development. G.A.R. was supported by a Wellcome Trust Investigator award (212625/Z/18/Z); UKRI-Medical Research Council (MRC) Programme grant (MR/R022259/1), an NIH-NIDDK project grants (R01DK135268; 1R01DK139630-01A1 PI), a CIHR-JDRF Team grant (CIHR-IRSC TDP-186358 and JDRF 4-SRA-2023-1182-S-N), CRCHUM start-up funds, and an Innovation Canada John R. Evans Leader Award (CFI 42649). JFT acknowledges support from NSERC Discovery (RGPIN-2024-04113) for salary support for DM. SH was supported by doctoral awards from the Fonds de recherche due Québec- Santé (FRQS) and the NSERC-CREATE-supported Canadian Islet Research Training Network in partnership with the JDRF. LDS was supported by a CIHR Postdoctoral Fellowship (#489982 CIHR-IRSC:0745000255).

Acknowledgments

The authors would like to thank Drs. Jace Jones-Tabah and Terry Hébert (McGill University) for providing the protocol and original Rluc8 vectors for the generation of the BRET-based sensor. We would also like to thank Dr. Aurélie Cleret-Buhot of the Cell Imaging core facility of the CRCHUM for performing the confocal microscopy acquisitions, Dr. Alexis Vivoli (CRCHUM) for data visualization assistance, George Vornicu (University of Montreal) for conducting preliminary cell death assays, and Dr. Petronela Ancuta (CRCHUM) for providing the Caco-2 and HT-29 cell lines. The authors also thank the Cellular physiology core facility (CRCHUM) for their help with the Operetta CLS and the Cellular imaging core facility (CRCHUM) for their help with confocal imaging. DM and JFT wish to recognize that this work was made possible by the facilities of the Compute Ontario (https://www.computeontario.ca) and the Digital Research Alliance of Canada (www.alliancecan.ca).

Data availability

Computational co-ordinates, including the docking poses, prepared proteins and peptides, and representative frames from the MD simulations are available from the Borealis Dataverse, a repository jointly operated by the Canadian Universities and Research Institutes, at https://doi.org/10.5683/SP3/IT6KHN.

D'Arcy MS. Cell death: a review of the major forms of apoptosis, necrosis and autophagy. Cell Biol Int 2019; 43(6):582–592.
Labbe K, Saleh M. Cell death in the host response to infection. Cell Death Differ 2008; 15(9):1339–1349.
Pfeffer CM, Singh ATK. Apoptosis: A Target for Anticancer Therapy. Int J Mol Sci 2018; 19(2).
Mayer CT, Nieke JP, Gazumyan A, Cipolla M, Wang Q, Oliveira TY, et al. An apoptosis-dependent checkpoint for autoimmunity in memory B and plasma cells. Proc Natl Acad Sci U S A 2020; 117(40):24957–24963.
Xi Y, Xu P. Global colorectal cancer burden in 2020 and projections to 2040. Transl Oncol 2021; 14(10):101174.
Simon K. Colorectal cancer development and advances in screening. Clin Interv Aging 2016; 11:967–976.
Hammond WA, Swaika A, Mody K. Pharmacologic resistance in colorectal cancer: a review. Ther Adv Med Oncol 2016; 8(1):57–84.
Souers AJ, Leverson JD, Boghaert ER, Ackler SL, Catron ND, Chen J, et al. ABT-199, a potent and selective BCL-2 inhibitor, achieves antitumor activity while sparing platelets. Nature Medicine 2013; 19(2):202–208.
Lim GE, Piske M, Johnson JD. 14-3-3 proteins are essential signalling hubs for beta cell survival. Diabetologia 2013; 56(4):825–837.
Datta SR, Katsov A, Hu L, Petros A, Fesik SW, Yaffe MB, et al. 14-3-3 proteins and survival kinases cooperate to inactivate BAD by BH3 domain phosphorylation. Mol Cell 2000; 6(1):41–51.
Kale J, Osterlund EJ, Andrews DW. BCL-2 family proteins: changing partners in the dance towards death. Cell Death Differ 2018; 25(1):65–80.
Zha J, Harada H, Yang E, Jockel J, Korsmeyer SJ. Serine phosphorylation of death agonist BAD in response to survival factor results in binding to 14-3-3 not BCL-X(L). Cell 1996; 87(4):619–628.
Yang E, Zha J, Jockel J, Boise LH, Thompson CB, Korsmeyer SJ. Bad, a heterodimeric partner for Bcl-XL and Bcl-2, displaces Bax and promotes cell death. Cell 1995; 80(2):285–291.
Hermeking H. The 14-3-3 cancer connection. Nat Rev Cancer 2003; 3(12):931–943.
Neal CL, Yu D. 14-3-3ζ as a prognostic marker and therapeutic target for cancer. Expert Opinion on Therapeutic Targets 2010; 14(12):1343–1354.
Maxwell SA, Li Z, Jaya D, Ballard S, Ferrell J, Fu H. 14-3-3zeta mediates resistance of diffuse large B cell lymphoma to an anthracycline-based chemotherapeutic regimen. J Biol Chem 2009; 284(33):22379–22389.
Ding J, Zhu YT, Yang L, Tang J, Wang YY, Chen Y, et al. 14-3-3zeta is involved in the anticancer effect of metformin in colorectal carcinoma. Carcinogenesis 2018; 39(3):493–502.
Wouters OJ, McKee M, Luyten J. Estimated Research and Development Investment Needed to Bring a New Medicine to Market, 2009–2018. JAMA 2020; 323(9):844–853.
Brown DG, Wobst HJ, Kapoor A, Kenna LA, Southall N. Clinical development times for innovative drugs. Nat Rev Drug Discov 2022; 21(11):793–794.
Takebe T, Imai R, Ono S. The Current Status of Drug Discovery and Development as Originated in United States Academia: The Influence of Industrial and Academic Collaboration on Drug Discovery and Development. Clin Transl Sci 2018; 11(6):597–606.
Begley CG, Ashton M, Baell J, Bettess M, Brown MP, Carter B, et al. Drug repurposing: Misconceptions, challenges, and opportunities for academic researchers. Sci Transl Med 2021; 13(612):eabd5524.
Pushpakom S, Iorio F, Eyers PA, Escott KJ, Hopper S, Wells A, et al. Drug repurposing: progress, challenges and recommendations. Nat Rev Drug Discov 2019; 18(1):41–58.
De Rosa MC, Purohit R, Garcia-Sosa AT. Drug repurposing: a nexus of innovation, science, and potential. Sci Rep 2023; 13(1):17887.
Du Y, Fu RW, Lou B, Zhao J, Qui M, Khuri FR, et al. A time-resolved fluorescence resonance energy transfer assay for high-throughput screening of 14-3-3 protein-protein interaction inhibitors. Assay and drug development technologies 2013; 11(6):367–381.
Boute N, Jockers R, Issad T. The use of resonance energy transfer in high-throughput screening: BRET versus FRET. Trends Pharmacol Sci 2002; 23(8):351–354.
McTavish D, Goa KL, Ferrill M. Terfenadine. Drugs 1990; 39(4):552–574.
Roy M, Dumaine R, Brown AM. HERG, a primary human ventricular target of the nonsedating antihistamine terfenadine. Circulation 1996; 94(4):817–823.
Janssen PA, Niemegeers CJ, Schellekens KH, Lenaerts FM, Verbruggen FJ, Van Nueten JM, et al. The pharmacology of penfluridol (R 16341) a new potent and orally long-acting neuroleptic drug. Eur J Pharmacol 1970; 11(2):139–154.
Wu HT, Wu BX, Fang ZX, Wu Z, Hou YY, Deng Y, et al. Lomitapide repurposing for treatment of malignancies: A promising direction. Heliyon 2024; 10(12):e32998.
Stefanutti C. Lomitapide-a Microsomal Triglyceride Transfer Protein Inhibitor for Homozygous Familial Hypercholesterolemia. Curr Atheroscler Rep 2020; 22(8):38.
van Hemert MJ, Niemantsverdriet M, Schmidt T, Backendorf C, Spaink HP. Isoform-specific differences in rapid nucleocytoplasmic shuttling cause distinct subcellular distributions of 14-3-3 sigma and 14-3-3 zeta. J Cell Sci 2004; 117(Pt 8):1411–1420.
Virdee K, Parone PA, Tolkovsky AM. Phosphorylation of the pro-apoptotic protein BAD on serine 155, a novel site, contributes to cell survival. Curr Biol 2000; 10(23):R883.
Goedhart J, von Stetten D, Noirclerc-Savoye M, Lelimousin M, Joosen L, Hink MA, et al. Structure-guided evolution of cyan fluorescent proteins towards a quantum yield of 93%. Nat Commun 2012; 3:751.
Griesbeck O, Baird GS, Campbell RE, Zacharias DA, Tsien RY. Reducing the environmental sensitivity of yellow fluorescent protein. Mechanism and applications. J Biol Chem 2001; 276(31):29188–29194.
Bolte S, Cordelieres FP. A guided tour into subcellular colocalization analysis in light microscopy. J Microsc 2006; 224(Pt 3):213–232.
Zhang JH, Chung TD, Oldenburg KR. A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays. J Biomol Screen 1999; 4(2):67–73.
Nicole E. Brown JBB, John R. Hepler. Protein-Protein Interactions. 2015.
Macdonald N, Welburn JP, Noble ME, Nguyen A, Yaffe MB, Clynes D, et al. Molecular basis for the recognition of phosphorylated and phosphoacetylated histone h3 by 14-3-3. Mol Cell 2005; 20(2):199–211.
Barstow B, Ando N, Kim CU, Gruner SM. Alteration of citrine structure by hydrostatic pressure explains the accompanying spectral shift. Proc Natl Acad Sci U S A 2008; 105(36):13362–13366.
Schenkmayerova A, Toul M, Pluskal D, Baatallah R, Gagnot G, Pinto GP, et al. Catalytic mechanism for Renilla-type luciferases. Nature Catalysis 2023; 6(1):23–38.
Miao Y, Feher VA, McCammon JA. Gaussian Accelerated Molecular Dynamics: Unconstrained Enhanced Sampling and Free Energy Calculation. J Chem Theory Comput 2015; 11(8):3584–3595.
D.A. Case KB, I.Y. Ben-Shalom, S.R. Brozell, D.S. Cerutti, T.E. Cheatham, III, V.W.D. Cruzeiro, T.A. Darden, R.E. Duke, G. Giambasu, M.K. Gilson, H. Gohlke, A.W. Goetz,R Harris, S. Izadi, S.A. Izmailov, K. Kasavajhala, A. Kovalenko, R. Krasny, T. Kurtzman, T.S. Lee, S. LeGrand, P. Li, C. Lin, J. Liu, T. Luchko, R. Luo, V. Man, K.M. Merz, Y. Miao, O. Mikhailovskii, G. Monard, H. Nguyen, A. Onufriev, F. Pan, S. Pantano, R. Qi, D.R. Roe, A. Roitberg, C. Sagui, S. Schott-Verdugo, J. Shen, C.L. Simmerling, N.R. Skrynnikov, J. Smith, J. Swails, R.C. Walker, J. Wang, L. Wilson, R.M. Wolf, X. Wu, Y. Xiong, Y. Xue, D.M. York and P.A. Kollman. AMBER 2020. In. San Francisco: University of California; 2020.
Izadi S, Anandakrishnan R, Onufriev AV. Building Water Models: A Different Approach. J Phys Chem Lett 2014; 5(21):3863–3871.
Ryckaert J-P, Ciccotti G, Berendsen HJC. Numerical integration of the cartesian equations of motion of a system with constraints: molecular dynamics of n-alkanes. Journal of Computational Physics 1977; 23(3):327–341.
Roe DR, Cheatham TE, 3rd. PTRAJ and CPPTRAJ: Software for Processing and Analysis of Molecular Dynamics Trajectory Data. J Chem Theory Comput 2013; 9(7):3084–3095.
Pennington KL, Chan TY, Torres MP, Andersen JL. The dynamic and stress-adaptive signaling hub of 14-3-3: emerging mechanisms of regulation and context-dependent protein-protein interactions. Oncogene 2018; 37(42):5587–5604.
Wang B, Yang H, Liu Y-C, Jelinek T, Zhang L, Ruoslahti E, et al. Isolation of High-Affinity Peptide Antagonists of 14-3-3 Proteins by Phage Display. Biochemistry 1999; 38(38):12499–12504.
Woodcock JM, Ma Y, Coolen C, Pham D, Jones C, Lopez AF, et al. Sphingosine and FTY720 directly bind pro-survival 14-3-3 proteins to regulate their function. Cell Signal 2010; 22(9):1291–1299.
Wu H, Ge J, Yao SQ. Microarray-assisted high-throughput identification of a cell-permeable small-molecule binder of 14-3-3 proteins. Angewandte Chemie 2010; 49(37):6528–6532.
He S, Lim GE. The Application of High-Throughput Approaches in Identifying Novel Therapeutic Targets and Agents to Treat Diabetes. Adv Biol (Weinh) 2023; 7(2):e2200151.
Feoktistova M, Geserick P, Leverkus M. Crystal Violet Assay for Determining Viability of Cultured Cells. Cold Spring Harb Protoc 2016; 2016(4):pdb prot087379.
Brookes Z, McGrath C, McCullough M. Antimicrobial Mouthwashes: An Overview of Mechanisms-What Do We Still Need to Know? Int Dent J 2023; 73 Suppl 2(Suppl 2):S64-s68.
Maillard JY. Impact of benzalkonium chloride, benzethonium chloride and chloroxylenol on bacterial antimicrobial resistance. J Appl Microbiol 2022; 133(6):3322–3346.
Simon-Soro A, Kim D, Li Y, Liu Y, Ito T, Sims KR, Jr., et al. Impact of the repurposed drug thonzonium bromide on host oral-gut microbiomes. NPJ Biofilms Microbiomes 2021; 7(1):7.
Baell JB, Nissink JWM. Seven Year Itch: Pan-Assay Interference Compounds (PAINS) in 2017-Utility and Limitations. ACS Chem Biol 2018; 13(1):36–44.
Roskoski R, Jr. Properties of FDA-approved small molecule protein kinase inhibitors: A 2021 update. Pharmacol Res 2021; 165:105463.
Iralde-Lorente L, Cau Y, Clementi L, Franci L, Tassone G, Valensin D, et al. Chemically stable inhibitors of 14-3-3 protein-protein interactions derived from BV02. J Enzyme Inhib Med Chem 2019; 34(1):657–664.
Corradi V, Mancini M, Manetti F, Petta S, Santucci MA, Botta M. Identification of the first non-peptidic small molecule inhibitor of the c-Abl/14-3-3 protein-protein interactions able to drive sensitive and Imatinib-resistant leukemia cells to apoptosis. Bioorg Med Chem Lett 2010; 20(20):6133–6137.
Van Noorden CJ. The history of Z-VAD-FMK, a tool for understanding the significance of caspase inhibition. Acta Histochem 2001; 103(3):241–251.
Andersson Y, Juell S, Fodstad O. Downregulation of the antiapoptotic MCL-1 protein and apoptosis in MA-11 breast cancer cells induced by an anti-epidermal growth factor receptor-Pseudomonas exotoxin a immunotoxin. Int J Cancer 2004; 112(3):475–483.
Yu J, Chen L, Chen Y, Hasan MK, Ghia EM, Zhang L, et al. Wnt5a induces ROR1 to associate with 14-3-3zeta for enhanced chemotaxis and proliferation of chronic lymphocytic leukemia cells. Leukemia 2017; 31(12):2608–2614.
Jiang L, Luo M, Liu D, Chen B, Zhang W, Mai L, et al. BAD overexpression inhibits cell growth and induces apoptosis via mitochondrial-dependent pathway in non-small cell lung cancer. Cancer Cell Int 2013; 13(1):53.
Cao X, Wang X, Lu L, Li X, Di R, He X, et al. Expression and Functional Analysis of the BCL2-Associated Agonist of Cell Death (BAD) Gene in the Sheep Ovary During the Reproductive Cycle. Front Endocrinol (Lausanne) 2018; 9:512.
Adachi M, Imai K. The proapoptotic BH3-only protein BAD transduces cell death signals independently of its interaction with Bcl-2. Cell Death Differ 2002; 9(11):1240–1247.
Obsilova V, Obsil T. Structural insights into the functional roles of 14-3-3 proteins. Frontiers in Molecular Biosciences 2022; 9.
Masters SC, Yang H, Datta SR, Greenberg ME, Fu H. 14-3-3 inhibits Bad-induced cell death through interaction with serine-136. Mol Pharmacol 2001; 60(6):1325–1331.
Yang X, Lee WH, Sobott F, Papagrigoriou E, Robinson CV, Grossmann JG, et al. Structural basis for protein-protein interactions in the 14-3-3 protein family. Proc Natl Acad Sci U S A 2006; 103(46):17237–17242.
Wilhelm S, Carter C, Lynch M, Lowinger T, Dumas J, Smith RA, et al. Discovery and development of sorafenib: a multikinase inhibitor for treating cancer. Nat Rev Drug Discov 2006; 5(10):835–844.
Toogood PL, Harvey PJ, Repine JT, Sheehan DJ, VanderWel SN, Zhou H, et al. Discovery of a potent and selective inhibitor of cyclin-dependent kinase 4/6. J Med Chem 2005; 48(7):2388–2406.
Petros AM, Dinges J, Augeri DJ, Baumeister SA, Betebenner DA, Bures MG, et al. Discovery of a potent inhibitor of the antiapoptotic protein Bcl-xL from NMR and parallel synthesis. J Med Chem 2006; 49(2):656–663.
Mureddu LG, Vuister GW. Fragment-Based Drug Discovery by NMR. Where Are the Successes and Where can It Be Improved? Front Mol Biosci 2022; 9:834453.
Macarron R, Banks MN, Bojanic D, Burns DJ, Cirovic DA, Garyantes T, et al. Impact of high-throughput screening in biomedical research. Nat Rev Drug Discov 2011; 10(3):188–195.
Malo N, Hanley JA, Cerquozzi S, Pelletier J, Nadon R. Statistical practice in high-throughput screening data analysis. Nat Biotechnol 2006; 24(2):167–175.
Shao Z, Cai Y, Xu L, Yao X, Shi J, Zhang F, et al. Loss of the 14-3-3σ is essential for LASP1-mediated colorectal cancer progression via activating PI3K/AKT signaling pathway. Scientific Reports 2016; 6(1):25631.
Young GM, Radhakrishnan VM, Centuori SM, Gomes CJ, Martinez JD. Comparative analysis of 14-3-3 isoform expression and epigenetic alterations in colorectal cancer. BMC Cancer 2015; 15:826.
Hidalgo IJ, Raub TJ, Borchardt RT. Characterization of the human colon carcinoma cell line (Caco-2) as a model system for intestinal epithelial permeability. Gastroenterology 1989; 96(3):736–749.
Hekmati M, Polak-Charcon S, Ben-Shaul Y. A morphological study of a human adenocarcinoma cell line (HT29) differentiating in culture. Similarities to intestinal embryonic development. Cell Differ Dev 1990; 31(3):207–218.
Gheytanchi E, Naseri M, Karimi-Busheri F, Atyabi F, Mirsharif ES, Bozorgmehr M, et al. Morphological and molecular characteristics of spheroid formation in HT-29 and Caco-2 colorectal cancer cell lines. Cancer Cell Int 2021; 21(1):204.
Zhu HL, Tomoda T, Aishima M, Ito Y, Teramoto N. The actions of azelnidipine, a dihydropyridine-derivative Ca antagonist, on voltage-dependent Ba2 + currents in guinea-pig vascular smooth muscle. Br J Pharmacol 2006; 149(6):786–796.
de Zeeuw D, Akizawa T, Audhya P, Bakris GL, Chin M, Christ-Schmidt H, et al. Bardoxolone methyl in type 2 diabetes and stage 4 chronic kidney disease. N Engl J Med 2013; 369(26):2492–2503.
Yates MS, Tauchi M, Katsuoka F, Flanders KC, Liby KT, Honda T, et al. Pharmacodynamic characterization of chemopreventive triterpenoids as exceptionally potent inducers of Nrf2-regulated genes. Mol Cancer Ther 2007; 6(1):154–162.
Kang SJ, Lee YJ, Kang SG, Cho S, Yoon W, Lim JH, et al. Caspase-4 is essential for saikosaponin a-induced apoptosis acting upstream of caspase-2 and gamma-H2AX in colon cancer cells. Oncotarget 2017; 8(59):100433–100448.
Nagata S, Suzuki J, Segawa K, Fujii T. Exposure of phosphatidylserine on the cell surface. Cell Death Differ 2016; 23(6):952–961.
Guo R, Lin B, Pan JF, Liong EC, Xu AM, Youdim M, et al. Inhibition of caspase-9 aggravates acute liver injury through suppression of cytoprotective autophagy. Scientific Reports 2016; 6(1):32447.
Wu YT, Tan HL, Huang Q, Sun XJ, Zhu X, Shen HM. zVAD-induced necroptosis in L929 cells depends on autocrine production of TNFalpha mediated by the PKC-MAPKs-AP-1 pathway. Cell Death Differ 2011; 18(1):26–37.
Varalda M, Antona A, Bettio V, Roy K, Vachamaram A, Yellenki V, et al. Psychotropic Drugs Show Anticancer Activity by Disrupting Mitochondrial and Lysosomal Function. Front Oncol 2020; 10:562196.
Zuo Q, Liao L, Yao Z-T, Liu Y-P, Wang D-K, Li S-J, et al. Targeting PP2A with lomitapide suppresses colorectal tumorigenesis through the activation of AMPK/Beclin1-mediated autophagy. Cancer Letters 2021; 521:281–293.
Lee B, Park SJ, Lee S, Lee J, Lee E, Yoo ES, et al. Lomitapide, a cholesterol-lowering drug, is an anticancer agent that induces autophagic cell death via inhibiting mTOR. Cell Death Dis 2022; 13(7):603.
Hua S. Advances in Oral Drug Delivery for Regional Targeting in the Gastrointestinal Tract - Influence of Physiological, Pathophysiological and Pharmaceutical Factors. Front Pharmacol 2020; 11:524.

There is no duality of interest

Download PDF

Version 1

posted

You are reading this latest preprint version

A high-throughput screening approach to discover potential colorectal cancer chemotherapeutics: Repurposing drugs to disrupt 14-3-3 protein-BAD interactions

Status:

Version 1

Abstract

Figures

Introduction

Materials and Methods

BRET sensor construction

Cell Culture

Plasmid transfection

Confocal imaging and colocalization analysis

Protocol for the high-throughput screening (HTS) of drugs

Cell death and apoptosis assays

Computational Methodology

Structure Preparation

Gaussian Accelerated Molecular Dynamics

Clustering

Docking Protocol

Protein-Protein Docking

Molecular Dynamics Simulations of 14-3-3ζ-BAD Complexes

Statistical analysis

Results

Inhibition of 14-3-3 proteins promotes BAD translocation

Development of a BRET-based living-cell sensor to detect interactions between 14-3-3ζ and BAD

In silico modeling of BAD and 14-3-3ζ interactions confirms the performace of the BRET sensor

Screening for FDA-approved pro-apoptotic drugs

14-3-3ζ inhibits BAD-induced cell death in CRC cells

Examination of hits in CRC cells

Observed activity of identified compounds correlates with their predicted interactions with 14-3-3

Verifying the capacity of terfenadine, penfluridol, and lomitapide to induce apoptosis in cancer cells

Discussion

Declarations

CREDIT Statement

Funding

Acknowledgments

Data availability

References

Additional Declarations

Supplementary Files

Status:

Version 1