VBIT-4 induces defects in membrane bilayers, both in the presence and absence of VDAC1.

Upper panels: VDAC1 reconstituted into POPC:POPE:chol (60:32.5:7.5) liposomes was adsorbed onto mica for AFM imaging. Representative AFM images and corresponding height profiles before addition (A) and after addition of 1 µM (B) and 10 µM (C) VBIT-4. The purple rectangles in the height profiles indicate the 2.5 nm threshold: values above correspond to VDAC1 pores, whereas values below correspond to defects induced by VBIT-4. Asterisks (*) denote VBIT-4–induced pores. In (C), the left and middle panels show VDAC clusters, while the right panel presents an overview and a cross-section from a 3D zoom highlighting the depth of VBIT-4–induced pores in lipid regions devoid of VDAC. Images acquired at 10 µM VBIT-4 display reduced resolution, consistent with compound-induced alterations of membrane properties. D. Quantitative pore size distribution derived from pixel depth/height analysis of the overview images shown in A, B, and C. Histograms represent the distributions without VBIT-4 (green, panel A), in the presence of 1 µM (orange, panel B) or 10 µM (purple, panel C) VBIT-4. E. Monomer and micellar structure of VBIT-4 obtained by molecular dynamic simulations. On the monomer structure, Polar and hydrogen-bonding groups are highlighted in red, ionizable/potentially charged groups in blue, neutral aromatic regions in grey, and strongly hydrophobic (lipophilic) substituents in green. Overall, the molecule displays an amphipathic organization, with a charged/polar core (blue/red) flanked by lipophilic aromatic groups (grey/green). F. G. H. Models based on AFM topography from panels A, B (right), and C (left), respectively. VDAC1 is shown as a dark solid surface, the first lipid shell in yellow, and surrounding lipids in atom color representation. The red patches in H. show the position of the VBIT-4 pores, that cannot be modelized by this method. Lower panels: Lipid membranes (without VDAC) were adsorbed onto mica and imaged by AFM. I. Control membrane before VBIT-4 addition. After addition of 1 µM (J), 5 µM (K), 50 µM (L), and 100 µM (M) VBIT-4. Height profiles are overlaid on the images. The false-colour scale is 12 nm in all panels.

VBIT-4 incorporates into lipid membranes.

A. Laurdan GP analysis of POPC:POPE:chol (62.5:30:7.5) liposomes with VBIT-4. The plot shows a VBIT-4 concentration-dependent decrease in laurdan generalized polarization (n=5). The inset corresponds to the fluorescence spectra at VBIT-4 concentrations changing from 0 (light blue) to 100 μM (dark blue); the inset data can be found at a larger scale in Supplementary Figure 4. B. Liposome leakage test of encapsulated fluorescein sodium salt from POPC:POPE:chol (62.5:30:7.5) liposomes with increasing concentration of VBIT-4 and TX-100 (0.5 %) (n=3). The inset represents the kinetics of fluorescein leakage from the liposomes. C. Representative current traces obtained on the same PLM made from DOPC:DOPE:chol (60:32.5:7.5) before and after consequent additions of 20 and 30 μM of VBIT-4 to the cis compartment of the chamber at 80 mV of applied voltage. Large fast fluctuations of the membrane conductance at 30 μM of VBIT-4 preceded the membrane rupture shown by the upward red arrow. The inset shows the current-voltage (I/V) curves obtained in the experiments at different VBIT-4 concentrations. Grey lines are linear regressions indicating the nearly Ohmic behaviour of VBIT-4-induced conductances. Dashed grey lines show a zero current. The membrane-bathing solutions consisted of 150 mM KCl buffered with 5 mM HEPES at pH 7.4. The current was digitally filtered using a 500 Hz Bessel (8-poles) filter for presentation. D. Conductance of planar membrane increases with VBIT-4 concentration. PLM were made from DOPC/DOPE/chol as in (C) and from PLE as in Supplementary Figure 4A. Membrane conductance was calculated from the corresponding I/V curves. The dashed lines are an exponential fit to guide the eye. Error bars are ±SD from measured conductances at different voltages.

VBIT-4 destabilizes lipid membranes independently of VDAC1.

A. Representative single-channel current traces obtained with VDAC1 reconstituted in PLM formed from PLE before (control) and after the addition of 5 and 30 μM VBIT-4 to the cis compartment (n=8). Addition of VBIT-4 and the time of the recordings after additions are indicated by upward arrows. The jumps in the current correspond to the application of 10 mV voltage following the application of 0 mV. The record was obtained on the membrane with the same single channel of 4 nS conductance as in the control. The grey dash line indicates zero current (I=0). The dashed blue lines indicate the current through the open single channel. Addition of 30 μM VBIT-4 induced a monotonic increase in membrane conductance, leading to membrane rupture. Current records were digitally filtered using an averaging time of 0.2 ms. B. Characteristic bell-shaped plots of open probability as a function of the applied voltage obtained in a multichannel experiment with VDAC1 in a PLE membrane in control and after addition of 5 and 10 μM VBIT-4. In all panels, the membrane-bathing solutions consisted of 1 M KCl buffered with 5 mM HEPES at pH 7.4. C. Binding of VBIT-4 to fluorescently labeled VDAC1 into nanodiscs (blue) or to empty nanodiscs, with MSP1D1 labeled (yellow), monitored by microscale thermophoresis. The plot shows normalized fluorescence at 650 nm measured 2.5 s after IR-laser activation across increasing VBIT-4 concentrations. An average apparent 𝐾! of 70 µM with a confidence interval from 20-167 µM (n = 4) was obtained for VBIT-4 binding to empty nanodiscs and 260 µM with a confidence interval from 105-331 µM (n = 2) for VDAC1-containing nanodiscs. Supplementary Figure 5B shows the control with MSP1D1 protein only.

VBIT-4 partitions into lipid bilayers and disrupts membrane integrity through pore-like structures.

A. VBIT-4 chemical structure and Martini 3 model mapping. Atom-to-bead mappings are indicated by the coloured shapes. Assigned Martini 3 bead names are indicated for each bead as overlaid bold text, with the corresponding bead types in italic text. Protonable group is highlighted (⋆). B. VBIT-4 Martini 3 CG model. C. Potential of Mean Force (PMF) of VBIT-4 insertion into coarse-grained MOM membranes. Free energy profiles are shown for both the neutral (blue) and protonated (+1 charge, red) forms of VBIT-4. D. Snapshots of the insertion of neutral VBIT-4 at different distances from the bilayer center (ξ); lipid PO4 beads are represented in orange, and VBIT-4 in grey/red/blue. E. Predicted octanol–water partition coefficient (Log P) values for VBIT-4. VBIT-4 hydrophobicity was estimated using several computational predictors (grey) and the Martini 3 coarse-grained model developed in this study (blue). Higher Log P values indicate greater lipid solubility. The dashed black line shows the consensus value across predictors, while the orange dashed line marks the Log P of ibuprofen (∼3.9), a well-known hydrophobic drug used here for comparison. F. Representative snapshots showing the impact of increasing VBIT-4 concentrations on MOM membrane mimics after 10 μs of CG MD simulation. (Lipid phosphate beads are shown in orange; VBIT in grey, with chemically distinct beads highlighted in red and blue). Top and side snapshots are shown for a 50:50 mix of charged and neutral VBIT-4 at increasing VBIT-4:lipid ratios (0, 1:16, 1:8, 1:3, 1:1.5). Lipid biophysical properties of MOM mimic membranes were evaluated in the presence of increasing VBIT-4 concentrations, either in the charged (red) or neutral (blue) form, as well as a 50:50 mixture of both states (orange). We monitored several indicators of membrane integrity: G. Area per lipid, H. Bilayer thickness, I. POPC acyl-chain order (membrane packing and organization), J. Lipid tail protrusion (packing defects), K. Bilayer water crossing (membrane permeability), and L. Lipid flip-flop (bilayer asymmetry). Each system was simulated for 10 µs in triplicate, and error bars represent the standard deviation across replicates. M. PMF profiles for polar defect formation in coarse-grained MOM membrane systems, shown as a function of the pore formation reaction coordinate ξ, for membranes without VBIT-4 (black), with charged VBIT-4 (red), and with neutral VBIT-4 (blue). Curve uncertainty (under the single-digit kJ mol−1 range) is represented by shading in the same colours along the corresponding curve. N. Representative snapshots of the membrane systems in the presence of neutral VBIT along the reaction coordinate; lipid PO4 beads are represented in orange, cholesterol ROH in dark orange, water beads in blue (only those in the vicinity of the lipid headgroups or the established defect are shown), and VBIT-4 in grey.

VBIT-4 decreases cell viability, mitochondrial respiration, and mitochondrial membrane potential in HeLa cells.

A. The graph shows a decrease in viability (black and teal) and an increase in cytotoxicity (pink and purple) with increasing concentration of VBIT-4 in WT and VDAC1 KO HeLa cells. The data represent the average from 3 independent experiments, and the error bars represent the SD. The data was fit with nonlinear regression fit (GraphPad Prism 11.0.0). B. Bar graphs represent changes in the mitochondrial oxygen consumption rate (OCR) after 6 hours of treatment with 1.25 −10 µM VBIT-4 (red square symbols) compared to DMSO control (●) normalised to OCR before drug treatment. HeLa cells were treated with 0.5 µM oligomycin (▴), 2 µM (♦), and 4 µM FCCP (◊) as controls. Symbols represent 8 replicates, and error bars represent the standard deviation from the mean. Flow cytometry measurements of mitochondrial calcium using Rhod-2 (C) and mitochondrial membrane potential using TMRM (D) upon the addition of 5 µM (▪) and 10 µM (▴) VBIT-4 compared to vehicle control (●). Data from 6 independent experiments are represented in the box plots. The symbols represent data from independent experiments. The borders of the boxes define the 25th and 75th percentiles, with the median displayed as lines and error bars indicating the standard deviation from the mean. Significance was tested using one-way ANOVA followed by the Dunnett post hoc test (*p < 0.05, **p < 0.01, ***p<0.001, ****p<0.0001).

Putative model of VBIT-4 partitioning and pore formation into a VDAC1-containing membrane.

A. Without VBIT-4, VDAC1 forms clusters of different sizes and compactions with both direct and lipid-mediated contacts between VDAC1 β-barrels. B. At low concentrations (< 10 µM), VBIT-4 partitions into the membrane, destabilizing it, without measurably altering VDAC1 cluster compaction. C. At high concentrations (>10 µM), VBIT-4 forms water-soluble pores in the membrane, inducing membrane permeabilization, further affecting membrane integrity. In all three scenarios, VDAC1 monomers maintain their transport channel properties, conducting the water-soluble metabolites, such as ATP, ADP, and NADH, and small ions, including calcium.