Mitochondria-lysosome contacts regulate mitochondrial fission via RAB7 GTP hydrolysis

Abstract

Both mitochondria and lysosomes are essential for maintaining cellular homeostasis, and dysfunction of both organelles has been observed in multiple diseases. Mitochondria are highly dynamic and undergo fission and fusion to maintain a functional mitochondrial network, which drives cellular metabolism. Lysosomes similarly undergo constant dynamic regulation by the RAB7 GTPase, which cycles from an active GTP-bound state into an inactive GDP-bound state upon GTP hydrolysis. Here we have identified the formation and regulation of mitochondria-lysosome membrane contact sites using electron microscopy, structured illumination microscopy and high spatial and temporal resolution confocal live cell imaging. Mitochondria-lysosome contacts formed dynamically in healthy untreated cells and were distinct from damaged mitochondria that were targeted into lysosomes for degradation. Contact formation was promoted by active GTP-bound lysosomal RAB7, and contact untethering was mediated by recruitment of the RAB7 GTPase-activating protein TBC1D15 to mitochondria by FIS1 to drive RAB7 GTP hydrolysis and thereby release contacts. Functionally, lysosomal contacts mark sites of mitochondrial fission, allowing regulation of mitochondrial networks by lysosomes, whereas conversely, mitochondrial contacts regulate lysosomal RAB7 hydrolysis via TBC1D15. Mitochondria-lysosome contacts thus allow bidirectional regulation of mitochondrial and lysosomal dynamics, and may explain the dysfunction observed in both organelles in various human diseases.

Extended Data Figure 1. Correlative light electron microscopy and 3D structured illumination microscopy of mitochondria-lysosome contacts

a–c, Representative electron microscopy images of mitochondria (M) and lysosome (L) contacts (yellow arrows; <30 nm) in untreated HeLa cells (insets shown on right) (n = 55 examples from 20 cells). d,e, Representative correlative light electron microscopy and confocal images of HeLa cells (from n = 14 images from 6 cells) incubated with LysoTracker Red to label lysosomes/late endosomes (red arrows) which (d) contain electron-dense lumen with irregular content and/or multilamellar membrane sheets (see insets on right), while early endosomes lacking electron-dense lumen are LysoTracker-negative (blue arrows), and (e) form a stable membrane contact site with mitochondria (yellow arrows; see inset on right), while simultaneously forming contact sites with the ER (purple arrows). f, Representative structured illumination microscopy (N-SIM) images of mitochondria-lysosome contacts (yellow arrows) in fixed HeLa cells stained for endogenous Lamp1 (lysosomes) or TOM20 (mitochondria) and imaged in Z-stacks showing contacts extending >200nm in the Z-plane (n = 210 examples from 26 cells). Scale bars, 200nm, a–d; 100nm, a–d (insets on right), e (left, middle); 50 nm, e (right); 500nm, f.

Extended Data Figure 2. Characterizing mitochondria-lysosome contacts in living cells

a–d, Representative images of mitochondria-lysosome contacts (>10 sec) in living HeLa cells expressing Lamp1-mGFP (lysosomes) and mApple-TOM20 (mitochondria) (n = 23 cells). (a) Examples of small Lamp1 vesicles (vesicle diameter <0.5 μm) contacting mitochondria. (b) Examples of larger Lamp1 vesicles (vesicle diameter >1 μm) contacting mitochondria. (c) Examples of a single Lamp1 vesicles contacting multiple mitochondria. (d) Examples of multiple Lamp1 vesicles contacting a single mitochondria. e, Representative images of contacts (yellow arrows) in fixed HeLa cells stained for endogenous Lamp1 (green) and TOM20 (red) (n = 341 examples from 25 cells). f,g, Representative images of living HeLa cells (n = 23 cells) expressing Lamp1-mGFP (lysosomes) and mApple-TOM20 (outer mitochondrial membrane) with corresponding linescan showing (f) a mitochondria-lysosome contact at close proximity, distinct from (g) lysosomal engulfment of mitochondrial TOM20. Scale bars, 0.5 μm, a–g.

Extended Data Figure 3. Structured illumination microscopy and FRET imaging of mitochondria-lysosome contacts in living cells

a–c, Representative structured illumination microscopy (N-SIM) images of mitochondria-lysosome contacts (yellow arrows) in living HeLa cells (n = 43 examples from 10 cells) expressing Lamp1-mGFP (lysosomes) and mApple-TOM20 (mitochondria) and quantitation of duration of mitochondria-lysosome contacts from N-SIM time lapse images. d, Model of newly generated FRET pairs targeted to the outer mitochondrial membrane (TOM20-Venus) and the lysosomal membrane (Lamp1-mTurquoise2). e, Representative time lapse images of a living HeLa cell (n = 200 cells) expressing FRET pairs (TOM20-Venus, Lamp1-mTurquoise2) and Q67L Rab7a-mRuby3 demonstrating preferentially increased SE-FRET signal over 60 seconds at the interface between mitochondria and lysosomes (white arrows). f, Quantification of normalized SE-FRET intensity per cell in conditions expressing wild-type Rab7a or Q67L Rab7a (n = 200 cells per condition) showing ~2-fold increase from Q67L Rab7a. Data are means ± SEM. (***P<0.0001, unpaired two-tailed t test (f)). Scale bars, 4 μm, a; 1 μm, b,e.

Extended Data Figure 4. Mitochondria-lysosome contacts are distinct from MDVs and mitophagy

a–c, Representative images of living HeLa cells expressing Lamp1-RFP (lysosomal membrane), mito-BFP (mitochondrial matrix) and SMAC-EGFP (mitochondrial intermembrane space) and corresponding linescans showing mitochondrial intermembrane space and matrix proteins do not undergo bulk transfer into lysosomes at contacts (yellow arrows) (n = 57 events from 12 cells). d, e, Representative images in a living HeLa cell expressing mApple-TOM20 (mitochondrial outer membrane), mito-BFP (mitochondrial matrix) and Lamp1-mGFP (lysosomes) and linescan (corresponding to top panel in d) showing mitochondria that form contacts with lysosomes (yellow arrows) are positive for mitochondrial matrix protein mito-BFP (not TOM20-positive MDVs) (n = 104 events from 23 cells). f, Representative linescan in a living HeLa cell expressing mEmerald-TOM20 (mitochondrial outer membrane), DsRed2-Mito (mitochondrial matrix) and mBFP2-Lys (lysosomes) showing mitochondria that form contacts with lysosomes (yellow arrows) are positive for mitochondrial matrix protein DsRed2-mito (not TOM20-positive MDVs) (n = 94 events from 16 cells). g–i, Representative images in a living HeLa cell expressing mApple-TOM20 (outer mitochondrial membrane), Lamp1-mGFP (lysosomal membrane) and fluid-phase marker dextran blue pulse-chased into the lysosomal lumen, and corresponding linescans showing lysosomal luminal content (blue) do not undergo bulk transfer into mitochondria at contacts (yellow arrows) (n = 66 events from 18 cells). j, Representative images in a living HeLa cell expressing Lamp1-RFP (lysosomes), mito-BFP (mitochondrial matrix) and EGFP-LC3 (autophagosome) showing mitochondria that form contacts with lysosomes (yellow arrows) are not engulfed by autophagosomes (not undergoing mitophagy) (n = 142 events from 17 cells). k, Autophagosome biogenesis proteins (ULK1-GFP, mCherry-Atg5, mEmerald-Atg12, GFP-DFCP1 and EGFP-LC3) do not mark sites of mitochondria-lysosome contacts in living cells (number of events analyzed n = 14 cells (ULK1), n = 17 cells (Atg5, Atg12, LC3) or n = 13 cells (DFCP1), top; quantification, bottom). Mitochondria (M) and lysosomes (L) are indicated in linescans. Data are means ± SEM. Scale bars, 0.5 μm, a; 1 μm, d,g,j.

Extended Data Figure 5. Fis1 recruits TBC1D15 to mitochondria

a–e, Representative images and quantitation of HA-TBC1D15 immunofluorescent localization to mitochondria (stained with endogenous TOM20) in fixed HeLa cells showing that mitochondrial localization is not disrupted by TBC1D15 GAP mutants (D397A or R400K) but by mutating its Fis1 binding site (Δ231-240) (n = 293 cells, WT; n = 228 cells, D397A; n = 181 cells, R400K; n = 379 cells, Δ231-240) (Δ231-240 versus WT (*P = 0.0178), D397A (*P = 0.0131), and R400K (*P = 0.0112), ANOVA with Tukey’s post-hoc test). f, Quantification showing that YFP-TBC1D15 localization to mitochondria is dramatically decreased by the Flag-Fis1 (LA) mutant (unable to bind TBC1D15) as compared to wild-type Flag-Fis1 (n = 290 cells, Fis1; n = 281 cells, Fis1 (LA)) (***P<0.0001, unpaired two-tailed t test). g, Examples of HA-TBC1D15 GAP mutants (D397A and R400K) or Fis1-binding mutant (Δ231-240) inducing enlarged lysosomes (white arrows) (Lamp1-mGFP) not observed in HA-TBC1D15 wild-type-expressing cells (n = 293 cells, WT; n = 228 cells, D397A; n = 181 cells, R400K; n = 379 cells, Δ231-240). Data are means ± SEM. Scale bars, 10 μm, a–d,g; 1 μm, a–d (insets).

Extended Data Figure 6. Recruitment of TBC1D15 by Fis1 to mitochondria promotes mitochondria-lysosome contact untethering

a,b, Representative time-lapse images of stable mitochondria-lysosome contacts (yellow arrows) for >120 sec before untethering (white arrow) in living HeLa cells expressing mApple-TOM20 (mitochondria), Lamp1-mGFP (lysosome) and Rab7-GAP mutant TBC1D15 D397A (n = 38 events from 10 cells). c, TBC domain mutants TBC1D15 D397A and R400K lacking GAP activity do not alter the percentage of lysosomes in contacts (n = 12 cells per condition), as compared to wild-type TBC1D15 (N.S. not significant). d,e, TBC1D15−/− HCT116 cells have increased duration (d, n = 18 events from 6 cells; WT; n = 16 events from 7 cells, TBC1D15−/−) but no change in the number of mitochondria–lysosome contacts (e, n = 15 cells, WT; n = 14 cells, TBC1D15−/−) compared to wild-type HCT116 cells (*P < 0.0491, N.S. not significant). f, Expression of the Flag–FIS1(LA) mutant (unable to bind TBC1D15) increases the percentage of lysosomes in mitochondria–lysosome contacts compared to wild-type FIS1 in living HeLa cells (n = (n = 18 cells, FIS1; n = 16 cells, FIS1(LA); *P < 0.0117). g,h Fis1 −/− HCT116 cells have increased duration (n = 15 events from 5 cells; WT; n = 14 events from 6 cells, Fis1 −/−) and number of mitochondria-lysosome contacts compared to wild-type HCT116 cells (n = 15 cells, WT; n = 13 cells, Fis1 −/−) (*P < 0.0442, ***P<0.0001). i,j, HA-TBC1D15 (n = 293 cells) and Flag-Fis1 (n = 272 cells) localization to mitochondria in fixed HeLa cells does not only occur at mitochondria-lysosome contacts. Data are means ± SEM. ANOVA with Tukey’s post-hoc test (c), unpaired two-tailed t test (d–h)). Scale bars, 0.5 μm, a; 1 μm, b, i–j (insets); 10 μm, i,j.

Extended Data Figure 7. Mitochondrial fission sites are marked by mitochondria-lysosome contacts in multiple cell types

a,b, Representative time-lapse images of lysosomes contacting mitochondria at site of mitochondrial fission (yellow arrow; top panel) prior to mitochondrial fission (white arrows; middle panels) in living HeLa cells expressing mGFP-Lamp1 (lysosomes) and mApple-TOM20 (mitochondria) with corresponding linescans showing lysosomes at the site of fission (yellow arrow; linescan) after mitochondrial division into two daughter mitochondria (grey arrows; linescan) (n = 62 events from 23 cells). c, EM image of mitochondria (M) in contact (<30 nm) with a lysosome (L; yellow arrows) at site of mitochondrial constriction in untreated HeLa cells (from n = 20 cells imaged). d–g, Lysosomes (yellow arrows; mGFP-Lamp1) mark sites of mitochondrial fission (white arrows; mApple-TOM20) at similar rates in living H4 neuroglioma, HEK293 and HCT116 cells as in HeLa cells by time-lapse confocal imaging (n = 49 events from 10 cells, HeLa; n = 36 events from 13 cells, H4; n = 18 events from 9 cells, HEK293; n = 9 events from 6 cells, HCT116). Data are means ± SEM. (N.S. not significant, ANOVA with Tukey’s post-hoc test). Scale bars, 1 μm, a,b, e–g (inset); 200 nm, c; 2.5 um, e–g.

Extended Data Figure 8. Mitochondria-lysosome contacts mark sites of mitochondrial fission upon induction of mitochondrial fragmentation

a–d, Lysosomes (yellow arrows; mGFP-Lamp1) mark sites of mitochondrial fission (white arrows; mApple-TOM20) at similar rates in untreated living HeLa cells as when treated with 0–20 min of Actinomycin D (a), STS (b) or CCCP (c) (n ≥ 49 events from ≥ 10 cells, Control; n = 29 events from 14 cells, Actinomycin D; n = 36 events from 10 cells, STS; n = 49 events from 14 cells, CCCP). Data are means ± SEM. (N.S. not significant, ANOVA with Tukey’s post-hoc test). Scale bars, 5 μm, a–c;.1 um, a–c (inset).

Extended Data Figure 9. Mitochondrial fission sites marked by lysosomes are positive for DRP1 and endoplasmic reticulum tubules

a, Representative time-lapse image of lysosome contacting mitochondria at site of mitochondrial division (yellow arrow) prior to fission event (white arrows) in living HeLa cell expressing mEmerald-TOM20 (mitochondria), mBFP2-Lys (lysosomes) and mCherry-Drp1 showing Drp1 oligomerization at site of mitochondrial division (n = 41 events from 11 cells). b,c, Representative time-lapse image (inset in c) of lysosome contacting mitochondria at site of mitochondrial division (yellow arrow) prior to fission event (white arrows) in living HeLa cell expressing mEmerald-TOM20 (mitochondria), mBFP2-lys (lysosomes) and mCherry-ER (ER) showing ER tubule at site of mitochondrial division (n =84 events from 19 cells). Scale bars, 1 μm, a,c; 5 μm, b.

Extended Data Figure 10. Regulation of mitochondrial network dynamics by Rab7 GTP hydrolysis

a, Examples of mitochondria not undergoing fission for >120 sec in living HeLa cells expressing mApple-TOM20 (mitochondria) and Rab7Q67L-GFP (n = 13 cells). b,c, Examples of mitochondria undergoing fission (white arrows) after 36 sec in living HeLa cells expressing mApple-TOM20 (mitochondria) and wild-type Rab7-GAP TBC1D15 (n = 13 cells). d,e, Examples of mitochondria not undergoing fission for >240 sec in living HeLa cells expressing mApple-TOM20 (mitochondria) and Rab7-GAP mutants TBC1D15 D397A (d) or TBC1D15 R400K (e) (n = 13 cells per condition). f–i, The percentage of mitochondrial fission sites marked by lysosomes (mGFP-Lamp1) or ER (mCherry-ER) is not disrupted by the Rab7Q67L GTP-hydrolysis deficient mutant (n = 12 events from 15 cells) or by TBC1D15 GAP mutants (D397A or R400K) (n = 22 events from 10 cells, WT; n = 17 events from 19 cells, D397A; n = 27 events from 22 cells, R400K). j–l, Examples of Rab7Q67L and HA-TBC1D15 GAP mutants (D397A and R400K) inducing elongated mitochondria (yellow arrows; >10μm length) compared to control cells, and quantification of Rab7Q67L (*P = 0.0321) and HA-TBC1D15 GAP mutants (D397A and R400K) (*P = 0.0297, **P = 0.0051) leading to decreased percentages of cells with normal mitochondrial networks (no elongated mitochondria > 10μm length or hyperfused/tethered networks) (n = 47 cells, Rab7; n = 72 cells, Rab7Q67L; n = 88 cells, TBC1D15 WT; n = 168 cells, TBC1D15 D397A; n = 132 cells, TBC1D15 R400K). Data are means ± SEM. (N.S. not significant, ANOVA with Tukey’s post-hoc test (h,i,l), unpaired two-tailed t test (f,g,k)). Scale bars, 0.5 μm, a, 1 μm, b–e, 10 μm, j.

Figure 1. Mitochondria and lysosomes form stable membrane contact sites

a,b, Representative electron microscopy image of mitochondria (M) and lysosome (L) contact (yellow arrow) in untreated HeLa cells and quantification of distance between contact membranes and length of contact (n=55 examples from 20 cells). c, Representative structured illumination microscopy (N-SIM) images of mitochondria-lysosome contacts (yellow arrows) in fixed HeLa cells stained for endogenous Lamp1 (lysosome) and TOM20 (mitochondria) and imaged in Z-stacks showing contacts extending >200nm in the Z-plane (3D N-SIM; left; n=210 examples from 26 cells) and in living HeLa cells expressing Lamp1-mGFP and mApple-TOM20 (Live N-SIM; right; n=43 examples from 10 cells). d–h, Representative time-lapse confocal images of stable mitochondria-lysosome contacts (yellow arrows) in living HeLa cells expressing Lamp1-mGFP (lysosomes) and mApple-TOM20 (mitochondria) (n=67 examples from 23 cells). White arrows in h mark lysosomes before or after contact tethering to mitochondria i,j, Quantitation of duration of mitochondria-lysosome contacts and percent of lysosomes contacting mitochondria (for >10 sec) from confocal time lapse images (n=45 examples from 10 cells). k, Quantification of percent of mitochondria (TOM20) or lysosomes (Lamp1) positive for mitochondrial intermembrane space protein (SMAC-EGFP; n=57 examples from 12 cells), mitochondrial matrix protein (mito-BFP; n=104 events from 23 cells) or lysosomal lumen marker (pulse-chased dextran; n=66 events from 18 cells) at mitochondria-lysosome contacts in living HeLa cells. Data are means ± SEM. (***P<0.0001, unpaired two-tailed t test). Scale bars, 200 nm, a; 500nm, c (3D N-SIM); 500 nm, c (Live N-SIM; left, right); 100 nm, c (Live N-SIM; middle); 1 μm, d; 0.5 μm, e–h.

Figure 2. Rab7 GTP hydrolysis promotes mitochondria-lysosome contact untethering

a–c, Representative time-lapse images of lysosome in cytosol (white arrow; top panel) approaching mitochondria to form a stable contact (yellow arrows; black line) before leaving mitochondria (white arrow; bottom panel) in living HeLa cells expressing mApple-TOM20 (mitochondria) and lysosomal markers Lamp1-mGFP (a), Rab7-GFP (b) or constitutively active GTP-bound Rab7 Q67L - GFP mutant unable to undergo GTP hydrolysis (c) (n=45 events from 9 cells per condition). d, Representative time-lapse images of mitochondria-lysosome contacts (yellow arrows) for >150 seconds in Rab7Q67L-GFP cells (n=45 events from 9 cells per condition). e–g, Rab7 Q67L mutant leads to increased percentage of lysosomes in contacts (n=12 cells per condition), and increased minimum duration of mitochondria-lysosome contacts (n=45 events from 9 cells per condition). Data are means ± SEM. (***P<0.0001, unpaired two-tailed t test). Scale bars, 1 μm, a–c; 0.5 μm, d.

Figure 3. Mitochondrial recruitment of TBC1D15, a RAB7 GAP, by FIS1 drives RAB7 GTP hydrolysis to promote mitochondria–lysosome contact untethering

a–e, Representative time-lapse images of stable mitochondria-lysosome contacts (yellow arrows) for >300 sec in living HeLa cells expressing mApple-TOM20 (mitochondria), Lamp1-mGFP (lysosome) and TBC domain mutants TBC1D15 D397A or R400K lacking GAP activity, which increase the minimum duration of mitochondria-lysosome contacts compared to wild-type TBC1D15 (n=34 events from 12 cells, wild-type; n=38 events from 10 cells, D397A; n=36 events from 11 cells, R400K) (*P = 0.0404, ***P = 0.0002, ANOVA with Tukey’s post-hoc test (d)). f,g, Fis1 (LA) mutant (unable to recruit TBC1D15 to mitochondria) leads to increased minimum duration of mitochondria-lysosome contacts compared to wild-type Fis1 (n=45 events from 9 cells per condition) (*P = 0.049, unpaired two-tailed t test). Data are means ± SEM. Scale bars, 1 μm, a, b; 0.5 μm, c, 5 μm, f; 1 μm, f (insets).

Figure 4. Mitochondria–lysosome contacts mark sites of mitochondrial fission regulated by RAB7 GTP hydrolysis

a,b, Representative time-lapse images of lysosomes contacting mitochondria at the site of mitochondrial division (yellow arrow) prior to fission (white arrows) in living HeLa cells expressing Lamp1-mGFP (lysosomes) and mApple-TOM20 (mitochondria) (n=62 events from 23 cells). c, Linescan corresponding to Fig. 4a showing a lysosome contacting mitochondria pre-fission (yellow arrow; top panel) and remaining in contact post-fission (yellow arrow; middle panel) after the mitochondria has divided into two daughter mitochondria (grey arrows; middle panel). d, Percentage of mitochondrial division events marked by Lamp1 vesicles in living HeLa cells expressing Lamp1-mGFP (lysosomes) and mApple-TOM20 (mitochondria) (n=54 events from 18 cells, ***P < 0.0001, Fisher’s exact test). Significantly more events were marked by Lamp1 vesicles (81.5%) than expected by random chance (12.6%), or by early endosomes (GFP-EEA1) (n=45 events from 17 cells), or peroxisomes (mEmerald-peroxisomes) (n=49 events from 17 cells; ***P<0.0001). e–g, Rab7Q67L GTP-hydrolysis deficient mutant (n=10 cells, Rab7; n=13 cells, Rab7Q67L; ***P = 0.0008), TBC1D15 GAP mutants (D397A or R400K) (n=13 cells per condition; *P = 0.451, ***P = 0.001) or Fis1 (LA) mutant (unable to bind TBC1D15) (n=19 cells, Fis1 WT; n=18 cells, Fis1 (LA); **P = 0.0027) lead to decreased rates of mitochondrial fission events (n ≥ 10 cells per condition). Data are means ± SEM. (ANOVA with Tukey’s post-hoc test (d,f), unpaired two-tailed t test (e,g)). Scale bars, 0.5 μm, a, b.