VDAC1-interacting molecules promote cell death in cancer organoids through mitochondrial-dependent metabolic interference

Abstract

The voltage-dependent anion-selective channel isoform 1 (VDAC1) is a pivotal component in cellular metabolism and apoptosis with a prominent role in many cancer types, offering a unique therapeutic intervention point. Through an in-silico-to-in-vitro approach we identified a set of VA molecules (VDAC Antagonists) that selectively bind to VDAC1 and display specificity toward cancer cells. Biochemical characterization showed that VA molecules can directly interact with VDAC1 with micromolar affinity by competing with the endogenous ligand NADH for a partially shared binding site. NADH displacement results in mitochondrial distress and reduced cell proliferation, especially when compared to non-cancerous cells. Experiments performed on organoids derived from intrahepatic cholangiocarcinoma patients demonstrated a dose-dependent reduction in cell viability upon treatment with VA molecules with lower impact on healthy cells than conventional treatments like gemcitabine. VA molecules are chemical entities representing promising candidates for further optimization and development as cancer therapy strategies through precise metabolic interventions.

Keywords: Functional aspects of cell biology; Molecular medicine; Small molecule.

Graphical abstract

Figure 1

VDAC1 architecture and small molecule identification (A) Cartoon representation of mouse VDAC1 crystal structure (PDB: 3EMN). In blue is identified the β-strand topology of the channel, and in cyan, its N-terminal α-helix. (B) Selected poses of NADH bound to the channel (PDB: 6TIR), with its nicotinamide anchoring point highlighted. (C) The large groove between VDAC1 N-terminal helix and the β-barrel used as a target for the AtomNet drug discovery campaign. NADH anchoring point is indicated. (D) Close-up view of the putative VA compounds binding pocket, showing the VA-D10 molecule docked in the pocket and assuming a V-shaped conformation. (E–I) Representative denaturation curves of mVDAC1 (0.9 mg/mL) in LDAO after treatment with different VA molecules at a final concentration of 1 mM. Black curves represent the denaturation profile of the protein pre-treated with DMF (0.5%), and the colored ones the protein pre-treated with the indicated VA molecules. Temperature denaturation inflection point (T_i) shifts toward higher temperatures in the presence of VA molecules VA-D11 (+4.7 ± 0.1); VA-D10 (+4.9 ± 0.1); VA-C1 (+4.6 ± 0.1); VA-C4 (+3.6 ± 0.2); VA-C6 (+4.6 ± 0.1), indicating protein stabilization/binding. Data are expressed as mean ± SEM (n = 3). (J) Three-rings architecture and overall shape shared by all VA molecules.

Figure 2

VA molecule binding measured by ¹⁹F NMR (A–C) ¹⁹F NMR spectra of indicated VA molecules in LDAO buffer (black lines) and in the presence of 5μM mVDAC1 (colored lines). VA-D10 (A) and VA-C4 (C) significantly move their chemical shifts upon VDAC1 binding. VA-C1 (B) shows a reduction and broadening of its peak upon VDAC1 interaction. (See also Figure S1).

Figure 3

Direct binding of VA molecules to VDAC1 revealed by fluorescence polarization competition assay (A) Binding of the NADH (at 5 μM) to increasing concentrations of mVDAC1 measured by fluorescence polarization (FP). 7 μM is the protein concentration used in all experiments. (B–F) Displacement curves of the NADH (5 μM) from mVDAC1 with increased concentrations of VA-molecules. All the five molecules are able to compete with NADH binding with an apparent K_D in the micromolar range (K_DVA-D11 = 15.9 ± 0.6, K_DVA-D10 = 18.5 ± 0.2, K_DVA-C1 = 18.7 ± 0.3, K_DVA-C4 = 19.4 ± 0.4, K_DVA-C6 = 18.7 ± 0.5). Data are expressed as mean ± SEM (n = 3).

Figure 4

Identification of VA molecules framing residues (A–E) ΔΔ inflection temperatures (ΔΔT_i) values were obtained by subtracting the ΔT_i of mVDAC1 mutants from the ΔT_i of wild-type protein. Positive or negative ΔΔT_i values show the effect of stabilizing and destabilizing mutations on inflection temperature after treatment with VA molecules, respectively. Data are expressed as mean ± SEM (n = 3). (F) Schematic representation of the V-shaped VA compounds and the interactions between VDAC1 residues and the three aromatic rings of the molecules. The thermal denaturation assays confirmed that a cation-π interaction between K236 and VA-L rings and a π−π stack between F18 and VA-R rings are essential for binding. The binding can be further improved by introducing additional π interactions in the neighboring residues.

Figure 5

VDAC1 mutants affect VA molecule binding (A–E) VA-induced NADH Fluorescence polarization displacement curves from mVDAC1 wild-type (dashed lines) and stabilizing (mVDAC1 N238K—solid line) or destabilizing (mVDAC1 K236A—dotted lines) mutants with increasing concentrations of VA-molecules. K236A and N238K shows decreased and increased affinity, respectively compared to the wild-type channel (K_DVA-D11_KA = 24.2 ± 0.6, K_DVA-D10_KA = 28.0 ± 0.7, K_DVA-C1_KA = 26.4 ± 0.4, K_DVA-C4_KA = 27 ± 1, K_DVA-C6_KA = 28.8 ± 0.2; K_DVA-D11_NK = 13.1 ± 0.4, K_DVA-D10_NK = 10.7 ± 0.1, K_DVA-C1_NK = 15.20 ± 0.04, K_DVA-C4_NK = 14.4 ± 0.2, K_DVA-C6 _NK = 14.9 ± 0.5). Experiments were performed using a final protein concentration of 7 μM. Data are expressed as mean ± SEM (n = 3). (See also Figure S2).

Figure 6

VA molecules affect the viability of cancer cells (A–F) Dose-response curves representing cell viability of NIH-3T3 and SKBR-3 treated with different concentrations of VA molecules (VA-D11, VA-D10, VA-C1, VA-C4, and VA- C6). Data are expressed as mean ± SEM (n = 3). (F) Bar plot reporting Half the effective dose (EC₅₀) retrieved after cell viability assay. SKBR-3 cell lines show greater sensitivity to treatment with the five molecules (VA-D11: 10 ± 2 μM; VA-D10 10 ± 1 μM; VA-C1 11 ± 2 μM; VA-C4 11 ± 3 μM; VA-C6 10 ± 2 μM) then NIH3T3cell lines (VA-D11: 118 ± 3 μM; VA-D10 160 ± 4 μM; VA-C1 146 ± 3 μM; VA-C4 137 ± 2 μM; VA-C6 127 ± 4 μM). (G) The effect of VA molecules (D10, C1, C4) on the viability of iCCA patient-derived and paired non-tumor (NT) 2D cells. Viability was measured by MTT assay after 72 h exposure using 2μM concentration (∗p < 0,05, ∗∗p < 0,01). Data are represented as mean ± SEM (n = 3). (See also Figure S3).

Figure 7

High-resolution respirometry shows cancer cell mitochondrial distress upon VA treatment (A) Quantitative analysis of the oxygen consumption in the analyzed states (N-Pathway and NS-Pathway) of SKBR-3 and NIH-3T3 attained after the exposure to 10 μM VA-molecules for 48h. Data are indicated as the mean ± SEM (n = 3) and expressed as the ratio of the treated cells to the untreated ones. Data were analyzed by unpaired t-test, with ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001. (B) Quantitative analysis of the net oxygen consumption in the NS-Pathway or ATP-linked flux of SKBR-3 and NIH-3T3 attained after the exposure to 10 μM VA-molecules for 48h. Data are indicated as the mean ± SEM (n = 3) and expressed as the ratio of the treated cells to the untreated ones. Data were analyzed by unpaired t-test, with ∗p < 0.05, ∗∗p < 0.01 and ∗∗∗p < 0.001. (C) Proposed mechanism for NADH imbalance through VDAC1. Panel C has been created with BioRender.com (See also Figure S6).

Figure 8

Cholangiocarcinoma 3D models are affected by VA treatment Tumor (iCCA) and non-tumor (NT) organoids were exposed to VA treatment for 72 h at the indicated concentration. DMSO-treated organoids were used as control. (A–C) Left panel: VA treatment (D10, C1, C4) reduce cell viability of iCCA organoids in a dose-dependent manner, while show a limited effect on non-tumor organoids. Data are represented as the percentage of control DMSO-treated organoids and are the mean of at least three independent experiments ±SEM. Right panel: representative bright-field images of tumor and non-tumor organoids exposed to VA treatment (D10, C1, C4) at the indicated concentration for 72h. Scale bar: 200 μm. (See also Figures S4 and S5).