The interplay between BAX and BAK tunes apoptotic pore growth to control mitochondrial-DNA-mediated inflammation

Abstract

BAX and BAK are key apoptosis regulators that mediate the decisive step of mitochondrial outer membrane permeabilization. However, the mechanism by which they assemble the apoptotic pore remains obscure. Here, we report that BAX and BAK present distinct oligomerization properties, with BAK organizing into smaller structures with faster kinetics than BAX. BAK recruits and accelerates BAX assembly into oligomers that continue to grow during apoptosis. As a result, BAX and BAK regulate each other as they co-assemble into the same apoptotic pores, which we visualize. The relative availability of BAX and BAK molecules thereby determines the growth rate of the apoptotic pore and the relative kinetics by which mitochondrial contents, most notably mtDNA, are released. This feature of BAX and BAK results in distinct activation kinetics of the cGAS/STING pathway with implications for mtDNA-mediated paracrine inflammatory signaling.

Keywords: AFM; BAK; BAX; BCL-2; inflammatory cell death; membrane pore; mitochondria; pore-forming protein; single-molecule imaging; super-resolution microscopy.

Graphical abstract

Figure 1

BAK assembly into distinct structures correlates with mitochondrial outer membrane permeabilization (A) Scheme of the assay to monitor Smac release (magenta) from mitochondria to the cytosol due to mEGFP-BAK (green) foci formation. (B) Representative confocal images of a BAX/BAK DKO HCT116 cell transfected with mEGFP-BAK (green) and Smac-(1-60)-mCherry (magenta) before (0 min) and 16 and 36 min after apoptosis induction with 1-μM STS. Scale bars, 10 μM. (C) Maximum fluorescent intensity of mEGFP-BAK (green) and mean fluorescent intensity of cytosolic Smac-(1-60)-mCherry (magenta) for individual cells as in (B) (n = 4). Time point 0 min corresponds to the normalized time point of Smac-mCherry release. The release of Smac-mCherry in the cytosol correlates with an increase in the mEGFP-BAK intensity, which is indicative of foci formation. Bold lines represent the mean of individual dataset (light lines). (D) Percentage of TMRE positive BAX/BAK DKO HCT116 cells transfected with mEGFP-BAK after treatment with 1 μM STS (n = 2). Error bars represent the SD. (E and F) Representative SMLM images of mEGFP-BAK in BAX/BAK DKO HCT116 cells in healthy (E, mitochondria profile is defined by the dotted lines) and apoptotic conditions (F). Right panels are zoomed regions (indicated by white boxes in the main image). Scale bars: 1 μM in the main images and 500 nm in zoomed regions.

Figure 2

Super-resolution microscopy reveals nanoscopic BAK structures in apoptotic cells (A) Gallery of BAK structures in apoptotic BAX/BAK DKO HCT116 cells transfected with mEGFP-BAK and labeled with an anti-GFP Alexa Fluor 647 nanobody. Scale bars, 100 nm. (B) Relative distribution of the imaged BAK structure types over the total number of structures in all measured cells (n = 12). (C and D) Quantification of the line and arc length and of the ring perimeter (C) and of the arc and ring radius (D) for all analyzed cells (n = 12). (E) Quantification of the diameter of dots (blue) and outer diameter of rings (light blue) for all analyzed cells (n = 12). (F) Exemplary images representing dots with less intensity in the center. Scale bars, 100 nm. (G) Relative distribution of all collected line (minimum 84), arc (minimum 112), and ring (minimum 24) structures (cells n = 12) at different time points after first foci appearance. See also Figures S1 and S2.

Figure 3

BAK assemblies are associated with stable membrane pores smaller than BAX (A and B) Representative image of an EPC:CL (80:20 mol %) membrane without (A) and with (B) BAK pores (indicated by the blue arrows) imaged by atomic force microscopy (AFM). Scale bars, 200 nm. (C) 3D representative image of a BAK pore (left) and its height profile (right, corresponding to the blue line in the 2D image inset). Scale bars, 20 nm. (D) Representative BAK structures imaged by AFM. Picture size 100 nm. The full-color height range of the topographs from low (brown-orange) to high (yellow-white) is 2 nm. (E) Percentage distribution of each BAK structure types over the total number of analyzed structures (lines = 14, arcs = 31, and rings = 14). All data (A–E) were obtained from at least four independent experiments. (F) Representative images of giant unilamellar vesicles (GUVs, black and red membrane in the merged channel) of EPC:CL (80:20 mol %) lipid composition without (top) or with (bottom) incubation with cBid-activated BAK. After 1-h incubation, Cytc488 (green) and APC (blue) were added and images were taken 30 min later. White and yellow arrow heads indicate Cytc488 and Cytc488-APC permeabilized GUVs, respectively. Scale bars, 10 μm. (G) Percentage of GUVs internalizing the fluorescent probes at different concentrations of BAK and 20 nM of cBid. Data (F and G) were obtained from at least three independent experiments and a minimum of 50 vesicles were analyzed per condition. Error bars represent the SD. ^∗p < 0.05; ^∗∗p < 0.01 (one-way ANOVA with Dunnett's multiple comparison test). (H) Schematic representation of the protocol used for sample preparation of stoichiometry experiments in model membranes (BAX and BAK depicted in yellow, see STAR Methods). (I) Representative TIRF image of an SLB containing BAK oligomers (bright spots). Scale bars, 10 μm. (J) Upper panel: representative fluorescence intensity distribution of BAK-488 and BAX-488 oligomers (minimum 1,500 particles per condition) obtained from SLB samples with 10 nM of protein. The obtained brightness distribution was plotted as a probability density function (Pdf, black) or, alternatively, as a histogram, and fitted with a linear combination of Gaussians to estimate the percentage of occurrence of particles containing n-mer-labeled molecules (see color code in the graph). The three panels in the background indicate each a 50,000 counts range and are for a visual comparison between BAX and BAK intensity distribution graphs. Lower panel: percentage of occurrence of BAK-488 and BAX-488 different oligomeric species calculated as the average value from two different experiments. Data provided are corrected for partial labeling (see STAR Methods). The error bars correspond to the SD from the different experiments. See also Figures S3 and S4.

Figure 4

BAX and BAK foci assembly proceeds via a different mechanism of oligomerization (A) Representative confocal images of a BAX/BAK DKO U2OS cell transfected with mEGFP-BAX (green), Tomm20-BFP (blue) and Smac-1-60-mCherry (magenta) and treated with ABT-737, S63845 and qVD-Oph for apoptosis induction. Time points are after Smac-1-60-mCherry release from mitochondria. Scale bars, 10 μm. Inset: apoptotic foci of mEGFP-BAX growing over time on mitochondria. Scale bars, 3 μm. (B) Representative photon-counting confocal images of BAX/BAK DKO U2OS cells transfected with mEGFP-BAK, mEGFP-BAX, or mEGFP-BAX(S184V) and treated for apoptosis induction. Scale bars, 10 μm. (C) Average mEGFP expression level per mitochondrial area in individual cells (represented by individual dots in the plot) transfected with mEGFP-BAK, mEGFP-BAX, or mEGFP-BAX(S184V) detected by single-cell fluorescent intensity analysis of mEGFP signal. The error bars correspond to the SD from the different experiments. (D–F) Representative distribution of foci intensity, in stoichiometric units, at different time points in individual apoptotic cells overexpressing mEGFP-BAX (D), mEGFP-BAK (E), or mEGFP-BAX(S184V) (F). (G) Average foci intensity obtained from the fitting of foci distributions as in (D–F) at each time point for individual apoptotic cells expressing mEGFP-BAX (yellow, n = 9), mEGFP-BAK (blue, n = 9) or mEGFP-BAX(S184V) (purple, n = 4). Lines in the graph correspond to the average values from all measured cells and colored areas correspond to data variability from single cells (mean ± SD). (H and I) Number of foci with molecularity higher than 200 (H) or 400 (I) normalized to the total number of foci in the cell at different time points after Smac release. Average total n° foci over analyzed cells is ranging from 67 (at 5 min) to 517 (60 min) for mEGFP-BAX (n = 9 cells) and from 250 (at 5 min) to 481 (60 min) for mEGFP-BAK (n = 9 cells). The error bars correspond to the SD from the different experiments. (J) Number of foci per mitochondrial area over time for apoptotic cells as in (G). See also Figures S5 and S6.

Figure 5

Reciprocal contribution of BAK and BAX to their assembly kinetics and molecularity (A) Representative photon-counting confocal images of BAX/BAK KO U2OS cells transfected with mEGFP-BAX or mEGFP-BAK, respectively, after apoptosis induction by ABT-737, S63845, and qVD-OPh. Time points are after Smac-mCherry release. Scale bars, 10 μm. (B) Comparison of the expression levels of mEGFP-BAK and mEGFP-BAX in single or double BAK/BAX KO cells detected by western blot (upper) or by single-cell (individual dots in the plot) fluorescent intensity analysis of mEGFP signal (bottom). WT, wild type.The error bars correspond to the SD from the different experiments. (C and D) Representative foci intensity distribution, in stoichiometric units, at different time points in an apoptotic BAX (C) and BAK (D) KO cell expressing mEGFP-BAX or mEGFP-BAK, respectively. (E) Average foci intensity obtained from the fitting of foci distributions as in (C and D) at each time point for individual apoptotic cells expressing mEGFP-BAX (orange, n = 5) or mEGFP-BAK (green, n = 9). (F and G) Comparison between the average foci intensity of mEGFP-BAK in BAX/BAK DKO (blue, n = 9) and BAK KO (green, n = 9) cells over time (F) and of mEGFP-BAX in BAX/BAK DKO (yellow, n = 9) and BAX KO (orange, n = 5) cells over time (G). ^∗p < 0.1 (unpaired t test with Welch’s correction). (H) Number of mEGFP-BAX (orange, n = 5) and mEGFP-BAK (green, n = 9) foci per mitochondrial area in single KO cells over time (total number of events per cell as in C and D). For (E–H), lines in the graph correspond to the average values from all measured cells and colored areas correspond to data variability from single cells (mean ± SD). (I) Average size of the different structural parameters for mEGFP-BAK assemblies in BAX/BAK DKO (n = 7) and BAK KO (n = 6) cells measured by SMLM. Error corresponds to SD. (J) Distribution of the different BAK structure types found in BAX/BAK DKO (n = 7) and BAK KO (n = 6) cells (minimum number of analyzed structure type = 86). (K–M) Comparison of the quantifications of the ring radius (K), arc radius (L), and arc length (M). Number of analyzed structures in single and double KO cells, respectively: 86 and 109 (K), 310 and 363 (L), and 310 and 363 (M).

Figure 6

Recruitment of BAX by BAK and BAX-BAK co-localization (A) Live-cell STED super-resolution microscopy in BAX/BAK DKO U2OS cell transfected with Snap-BAX (red), Halo-BAK (green) and 4xmt-mTurquoise (magenta, confocal) to stain mitochondria after apoptosis induction. Zoomed images (right) correspond to crops of the overview image (left) as indicated by numbers. Scale bars, 5 μm and 500 nm for cropped images. (B) Gallery of BAX and BAK line, arc, and ring structures in apoptotic mitochondria by STED microscopy. Scale bars, 500 nm. Images in (A) and (B) are representative of three independent experiments. (C) (Left) Scheme of proximity-dependent labeling with APEX2 (see STAR Methods). (Central and right) Immunoblots of streptavidin immunoprecipitation of APEX2-BAX (central) or APEX2-BAK (right) from apoptotic HeLa cells expressing FLAG-APEX2-BAX or FLAG-APEX2-BAK. Immunoblots are representative of two independent experiments. Only regions of the gel with bands of interest are shown for clarity. (D) (Left) Principle of dimerization-dependent fluorescent protein (ddFP). RA only becomes fluorescence when it is in complex with GB. (Right) Representative confocal microscopy image of a BAX/BAK DKO U2OS cell transfected with GB-BAK, RA-BAX, and Mito-BFP (magenta) after apoptosis induction. Zoomed images correspond to cropped regions as indicated. The interaction between BAX and BAK (green dots) takes place in mitochondria upon apoptosis (negative control for collisions is not shown, it was tested in independent experiments). Scale bars, 10 μm; crop, 5 μm. Images are representative of two independent experiments. (E) Schematic representation of sample preparation for SMALPs pull-down assay (see STAR Methods). (F) Representative healthy (control) or apoptotic cells transfected with mEGFP-BAK (green) used in the assay. (G) Immunoblots of SMALPs immunoprecipitation of GFP-BAK and pull-down of endogenous BAX. Immunoblots show total input (T), flow through (FT), wash (W), and elution (E) fractions of the immunoprecipitation and are representative of three independent experiments. IB, immunoblot; IP, immunoprecipitation. Only regions of the gel with bands of interest are shown for clarity. (H) Representative confocal microscopy image of a BAX/BAK DKO U2OS cell transfected with mCherry-BAK (magenta), mEGFP-BAX (green), and Tomm20-BFP (blue) at indicated time points after apoptosis induction showing BAK foci appearing before BAX foci. Scale bars, 10 μm. Images are representative of three independent experiments. (I) Representative confocal microscopy image of optogenetic activation of BAK and recruitment of BAX to apoptotic mitochondria (MitoTrackerTM Deep Red FM, cyan). Scale bars, 20 μm. Images are representative of 4 independent experiments. (J) Representative images of BAK (green) activation followed by BAX (red) recruitment on a GUV (black hole). Alexa 555 (magenta) is used as a soluble dye to detect GUV permeabilization. Scale bars, 10 μm. Images are representative of 3 independent experiments. (K) Quantification of the binding of BAK-Atto 488 (green) and BAX-Atto 655 (red) to GUVs by radial profile corrected fluorescence intensity (cFU) measurements. Thin lines indicate individual measurements and thick lines correspond to the average of n = 7 individual GUVs from n = 3 independent experiments. See also Figure S7.

Figure 7

BAK pores enable faster mtDNA release and downstream inflammatory responses than BAX pores (A) Scheme of the experimental setup to investigate the kinetics of mtDNA release (green) after Smac(1-60)-mCherry (magenta) release by airy scan super-resolution microscopy. (B) Representative airy scan microscopy images of WT, BAK KO, and BAX KO U2OS cells transfected with Smac-(1-60)-mCherry, Tfam-GFP (green) and immunostained for Tomm20 (magenta) without induction of apoptosis (0 min) and 15 and 30 min after Smac(1-60)-mCherry release. Cropped panels on the right represent zoomed region of the white box in the main image. Scale bars, 5 μm and 1 μm for the main and zoomed images, respectively. (C) Percentage of cells with mtDNA release at different time points after Smac release. Numbers of analyzed cells (n) is on top of the bar graphs. (D) Representative photon-counting confocal images of BAX/BAK DKO U2OS cells transfected with mEGFP-BAX or mEGFP-BAX(T182I), after apoptosis induction. Time points are after Smac-mCherry release. Scale bars, 10 μm. (E) Average foci intensity at each time point for individual apoptotic cells expressing mEGFP-BAX (yellow, n = 2) or mEGFP-BAX(T182I) (orange, n = 3). Lines in the graph correspond to the average values from all measured cells and colored areas to data variability (mean ± SD). (F) Average mEGFP expression level detected by single-cell fluorescent intensity analysis per mitochondrial area in the individual cells (represented by individual dots in the plot) analyzed in (E). The error bars correspond to the SD from the different experiments. (G) Representative airy scan microscopy images of cells transfected with Smac-(1-60)-mCherry, HaloTag-BAX or HaloTag-BAX(T182I), Tfam-GFP (green) and immunostained for Tomm20 (magenta) at indicated time points after Smac release. Cropped panels on the right are zoomed region of the white box in the main image. Scale bars, 5 and 2 μm for the main and zoomed images, respectively. (H) Percentage of cells with mtDNA release at different time points after Smac release. Numbers of analyzed cells (n) is presented on top of the bar graphs. (I and J) STING degradation (I) and TBK1 phosphorylation (J) in WT, BAX KO, and BAK KO U2OS cells at 0 (untreated), 1, 2, and 3 h after apoptosis induction. Only regions of the gel with bands of interest are shown for clarity. (K) Changes of CD62L and CD44 expression in spleenocytes (CD45⁺, top row) and CD4⁺ T cells (bottom row) with SVEC cells either untreated or pre-treated with ABT-737, S63845 and QVD for 3 h. Two-way ANOVA and Sidak multiple comparison test was used for the statistical analyses. n = 8 spleen donors. ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. See also Figure S8.