Individual cristae within the same mitochondrion display different membrane potentials and are functionally independent

Abstract

The mitochondrial membrane potential (ΔΨ_m ) is the main driver of oxidative phosphorylation (OXPHOS). The inner mitochondrial membrane (IMM), consisting of cristae and inner boundary membranes (IBM), is considered to carry a uniform ΔΨ_m . However, sequestration of OXPHOS components in cristae membranes necessitates a re-examination of the equipotential representation of the IMM. We developed an approach to monitor ΔΨ_m at the resolution of individual cristae. We found that the IMM was divided into segments with distinct ΔΨ_m , corresponding to cristae and IBM. ΔΨ_m was higher at cristae compared to IBM. Treatment with oligomycin increased, whereas FCCP decreased, ΔΨ_m heterogeneity along the IMM. Impairment of cristae structure through deletion of MICOS-complex components or Opa1 diminished this intramitochondrial heterogeneity of ΔΨ_m . Lastly, we determined that different cristae within the individual mitochondrion can have disparate membrane potentials and that interventions causing acute depolarization may affect some cristae while sparing others. Altogether, our data support a new model in which cristae within the same mitochondrion behave as independent bioenergetic units, preventing the failure of specific cristae from spreading dysfunction to the rest.

Keywords: MICOS complex; Opa1; crista junction; cristae; membrane potential.

Figure 1. High‐resolution fluorescence imaging using Airyscan resolves the inner mitochondrial membrane (IMM) structure in live cells

High‐resolution imaging of mitochondria in live cells using the Airyscan module of Zeiss LSM880 confocal microscope.

1. A

   Images of IMM in living HeLa, L6, and H1975 cells, stained with 10‐N‐nonyl acridine orange (NAO). NAO preferentially binds phospholipids in the IMM, such as cardiolipin. Arrowheads indicate cristae in the IMM. Scale bar = 500 nm. N ≥ 3 independent experiments for each cell type.

2. B

   Mitochondrion cropped from HeLa cell shown in (A) (dashed line) and zoomed in to show cristae, inner boundary membrane (IBM), and matrix (arrows). Scale bar = 100 nm.

3. C

   H1975 cells transduced with matrix‐targeted DsRed and stained with NAO. Matrix‐targeted DsRed differentiates matrix from cristae stained with NAO. Arrowheads point to cristae. Scale bar = 500 nm. N = 1 independent experiment.

4. D

   The structure of the IMM in Mic13‐KO cells (HeLa), stained with NAO. The number of cristae is decreased in Mic13‐KO compared to control cells shown in panel (A), labeled with arrowheads. Scale bar = 500 nm. N = 3 independent experiments.

5. E, F

   Live‐cell imaging of the IMM in control and PTPMT1 KD H1975 cells, a model of cardiolipin deficiency. (E) A gallery of mitochondria from various control H1975 cells expressing scrambled shRNA and stained with NAO, showing cristae (arrowheads). Scale bars = 500 nm. N = 3 independent experiments. (F) A gallery of mitochondria from various H1975 cells expressing PTPMT1 shRNA and stained with NAO. Note the derangement of the ultrastructure (arrowheads). Scale bars = 500 nm. N = 3 independent experiments.

Figure EV1. Validation of KO and KD approaches of Cristae structure and CJ modulators

1. Representative image of Western blot, showing deletion of MIC13 in HeLa cells. N = 3 independent experiments.

2. Representative image of Western blot, showing deletion of PTPMT1 in H1975 cells. N = 3 independent experiments.

3. Representative image of Western blot, showing deletion of MIC60 and MIC10 in Hap1 cells. N = 3 independent experiments.

4. Representative image of Western blot, showing Opa1 KO and Opa1&Drp1 DKO in MEFs. N = 3 independent experiments.

Source data are available online for this figure.

Figure 2. The ΔΨ_m‐sensitive dye, TMRE, partitions to cristae stained with NAO

High‐resolution imaging of mitochondria in live cells using the Airyscan module of Zeiss LSM880 confocal microscope.

1. Mitochondria from HeLa cell, co‐stained with NAO and TMRE. Area from dashed white box, zoomed to right, shows the red and green intensities of TMRE and NAO colocalized (arrowheads). Scale bars = 500 nm. N = 3 independent experiments.

2. Mitochondria in L6 myoblast, co‐stained with NAO and TMRE. Area from dashed white box, zoomed to right, shows colocalizing NAO and TMRE at the cristae membrane (arrowheads). Scale bars = 500 nm. N = 3 independent experiments.

3. Mitochondria in HeLa cells, stained with NAO alone (top row) vs. NAO + TMRE (bottom row), and simultaneously excited with 488‐ and 561‐nm lasers. Note that, mitochondria stained with NAO alone do not emit noticeable fluorescence in the red channel; only after adding TMRE does strong signal appear in the red channel, showing negligible bleed‐through. Scale bars = 500 nm. N = 2 independent experiments.

Figure EV2. Membrane potential dyes, TMRE and Rho123, are concentrated at the cristae membranes in a heterogeneous pattern

1. Mitochondria in HeLa cells co‐stained with MTG and TMRE. Zoomed‐in region from dashed white box highlights colocalization of dyes, showing that signal intensities vary together across the long axis (green and red lines); arrowheads indicate cristae membranes. Scale bar = 500 nm. N = 3 independent experiments.

2. Mitochondria in HeLa cell stained with ΔΨ_m‐dependent dye, Rho123; zoomed‐in mitochondrion cropped from dashed line highlights more intense signal at cristae (arrowheads). Scale bar = 500 nm. N = 3 independent experiments.

Figure 3. Validation that TMRE partitioning to cristae is ΔΨ_m‐dependent

1. Time‐lapse Airyscan imaging of ΔΨ_m in living HeLa cell, co‐stained with MTG (ΔΨ_m‐insensitive after loading) and TMRE (ΔΨ_m sensitive). Arrowhead points to a flickering event where a mitochondrion depolarizes (˜ 9 s) and repolarizes (˜ 16 s), showing that heterogeneous signal from TMRE (but not MTG) disappears and reappears. Scale bar = 500 nm. N = 4 independent experiments.

2. Quantification of (A). Plot shows average TMRE fluorescence intensity (FI) of cristae (dark red line) vs. matrix (light red line) during the time series. The drop in TMRE FI during the depolarization phase of the flickering is the ΔΨ_m‐sensitive component of the TMRE signal. The remaining TMRE FI during depolarization can be considered as the ΔΨ_m‐insensitive portion of TMRE signal. Note that, the remaining TMRE FI after depolarization at the cristae and matrix is approximately identical, indicating that differences in TMRE FI between the cristae and matrix prior to depolarization are derived from differences in ΔΨ_m. N = 4 independent experiments.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM.

Figure 4. Cristae and IBM have different ΔΨ_m

1. A

   Live‐cell Airyscan image of mitochondrion in HeLa cell, showing different membrane potentials in different mitochondrial regions. Membrane potentials were calculated based on TMRE FI differences between compartments (FI_comp). Regions of interest from cristae and IBM used for ΔΨ_m calculations are labeled with arrows on the right‐hand side of the mitochondrion, using the cytosol as the reference compartment. Labeled with forks on the left‐hand side of the mitochondrion are the calculations of ΔΨ_m between individual cristae and their neighboring IBM. Nernst equation used to calculate different voltages. Scale bar = 500 nm.

2. B

   LUT of mitochondrion in HeLa cell shown in (A), color coding of TMRE FIs on scale of white (most intense) to blue (least intense): LUT scale shown in lower left‐hand corner. Arrowheads indicate cristae. Scale bar = 500 nm. Note that, most‐intense pixels (white) only associate with cristae.

3. C

   Quantification of ΔΨ_m (mV) at cristae and IBM relative to the average ΔΨ_m of the whole mitochondrion, calculated as in (A). Note: ΔΨ_m at cristae is significantly higher than ΔΨ_m at IBM, indicating electrochemical boundaries separate these two regions of the IMM. N = 3 independent experiments.

4. D–F

   Live‐cell Airyscan imaging of mitochondria in cultured primary mouse hepatocytes stained with TMRE. Arrowheads denote cristae. (D) Image showing TMRE‐labeled mitochondria. (E) LUT color coding of TMRE FI. (F) Quantification of ΔΨ_m differences using Nernst equation as shown in (A). Scale bar = 500 nm. N = 2 independent experiments.

5. G–I

   Live‐cell STED imaging of mitochondria in HeLa cell stained with TMRM. Arrowheads show cristae. Scale bar = 500 nm. (G) Image showing TMRM‐labeled mitochondrion. (H) LUT color coding of TMRM FI. (I) Quantification of ΔΨ_m differences using Nernst equation as shown in (A). Scale bar = 500 nm. N = 4 independent experiments. Note that, calculation of ΔΨ_m differences between the compartments based on images captured by STED and by Airyscan was very similar.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM.

Figure 5. ΔΨ_m differences between cristae and IBM are sensitive to F ₁ F ₀ ATP Synthase inhibition and uncoupling

Response of ΔΨ_m to the F ₁ F ₀ ATP Synthase inhibitor, oligomycin (10 μM), compared to the uncoupler, FCCP (10 μM), determined using live‐cell Airyscan imaging of HeLa cells co‐stained with MTG and TMRE.

1. Quantification of total mitochondrial TMRE FI. N = 3 independent experiments.

2. Quantification of cristae ΔΨ_m relative to IBM (ΔΨ_Cr‐IBM) in response to oligomycin compared to FCCP. F ₁ F ₀ ATP Synthase is exclusively localized to the cristae and significant increase in ΔΨ_Cr‐IBM in response to blocking of F ₁ F ₀ ATP Synthase with oligomycin indicates that differences in TMRE FI between cristae and IBM are driven by oxidative phosphorylation. N = 3 independent experiments.

3. Representative Airyscan images showing mitochondria from living HeLa cells stained with MTG and TMRE. Cells were first stained for 1 h and only then treated with oligomycin or FCCP. Images show hyperpolarization of cristae under oligomycin (middle row; arrowheads) and depolarization of cristae under FCCP (bottom row; arrowheads) compared to control (top row; arrowheads). Scale bar = 500 nm. N = 3 independent experiments. Note that, the ΔΨ_m‐independent fraction of TMRE staining is < 5% of the signal in control conditions.

4. Zoomed‐in region of FCCP‐treated mitochondria in (C). Increased contrast was used to visualize TMRE to compensate for TMRE loss induced by FCCP. These images show diminished heterogeneity of ΔΨ_Cr‐IBM values (arrowheads). Scale bar = 500 nm.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM.

Figure EV3. Effects of FCCP and oligomycin on mitochondrial morphology

1. Quantification of mitochondrial circularity, where mitochondria with circularity closer to 1 resemble a circular object and mitochondria with circularity closer to 0 resemble a branched object. Note: FCCP results in significant increase in mitochondrial circularity, consistent with numerous studies indicating FCCP induces fragmentation. N = 3 independent experiments.

2. Quantification of mitochondrial aspect ratio (a ratio of the long axis of the mitochondrion to the short axis), where mitochondria with an aspect ratio closer to 1 are rounder, whereas mitochondria higher and higher than 1 are increasingly elongated. Note: FCCP also leads to a significant decrease in aspect ratio, indicating loss of membrane potential results in fragmentation. N = 3 independent experiments.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM.

Figure 6. Crista junctions (CJs) regulate the difference in ΔΨ_m between cristae and IBM

The role of cristae structure as well as CJ formation and sealing on the generation of the difference in ΔΨ_m between cristae and IBM were studied by disrupting cristae using Mic10‐, Mic13‐, and Mic60‐deficient cells. Mic13 and Mic60 support cristae formation. Mic10 is essential for CJ formation, and, in its absence, cristae tend to remain as vesicles detached from IBM. Live‐cell Airyscan imaging of TMRE was used in all figure panels and models.

1. Representative images of mitochondria in control (top row) vs. Mic13‐KO (bottom row) HeLa cells. Note that, TMRE FI in Mic13‐KO mitochondria is distributed more evenly along the IMM, so that ΔΨ_Cr‐IBM is decreased (arrowheads). Scale bar = 500 nm. N = 3 independent experiments.

2. Quantification of (A), showing deletion of Mic13 in HeLa cells results in significant decrease in ΔΨ_Cr‐IBM. While Mic13‐KO cells show a substantial decrease in cristae number, the loss of cristae does not appear absolute, making it feasible to compare TMRE FI at cristae relative to IBM. N = 3 independent experiments.

3. Representative images of Hap1 control (top row) vs. Mic60 KO (middle row) and Mic10 KO (bottom row); Mic60‐KO mitochondria show very few cristae structures. TMRE staining along the IMM is relatively homogeneous, indicating decreased ΔΨ_Cr‐IBM. In general, Mic10 KO in Hap1 cells results in decreased TMRE signal intensity at cristae relative to IBM (arrowheads) as compared to control cells. Deletion of Mic10 induces detachment of cristae from IBM (arrows; see Fig 8). Except for the detached cristae vesicles, the TMRE staining is homogeneously distributed in Mic10‐KO cells, indicating decreased ΔΨ_Cr‐IBM. Scale bar = 500 nm. N = 3 independent experiments.

4. Quantification of (C), showing Mic60 KO in Hap1 cells results in significant decrease in ΔΨ_Cr‐IBM. While Mic60‐KO cells show a marked loss of normal cristae structures, depletion of cristae does not appear absolute, making it possible to compare TMRE FI at cristae relative to IBM. N = 3 independent experiments.

5. Quantification of (C), showing Mic10 KO in Hap1 cells results in a significant decrease in ΔΨ_Cr‐IBM. Note: quantification of ΔΨ_m at cristae relative to IBM was performed only on cristae structures that appeared to maintain attachment to IBM. N = 3 independent experiments.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM.

Figure 7. Opa1 regulates the difference in ΔΨ_m between cristae and IBM

Studies show that Opa1 interacts with MICOS complex, promoting closure of CJs. Thus, we tested whether Opa1 was required to maintain the difference in ΔΨ_m between cristae and IBM.

1. Live‐cell Airyscan images of MEF control (top row) vs. Opa1 KO (bottom row), co‐stained with NAO and TMRE. Arrowheads indicate decreased intensity of TMRE FI at cristae compared to IBM. Note the more even distribution of TMRE staining, indicating the cristae and IBM are relatively equipotential. Scale bar = 500 nm.

2. Quantification of (A), showing Opa1‐KO MEFs have significantly decreased ΔΨ_Cr‐IBM. N = 3 independent experiments.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM.

Figure 8. The largest differences in ΔΨ_{Cr‐ IBM} arise in cristae vesicles generated by deleting CJ regulators

1. Live‐cell Airyscan images of laser‐induced depolarization time series of Mic10 KO Hap1 cell stained with TMRE. Mitochondrion exposed to rapid, high 2‐photon laser power in region of white box, arrow (˜ 0.3 s); large, hyperpolarized vesicle (arrowhead) remains polarized after the rest of the mitochondrion loses ΔΨ_m, indicating electrochemical discontinuity between hyperpolarized vesicles and the rest of the mitochondrion, including the IBM. Scale bar = 500 nm. N = 2 independent experiments.

2. Live‐cell Airyscan image of time series showing hyperpolarized vesicles (arrowheads; region cropped from dashed white box) inside mitochondria lacking both Opa1 and Drp1. Scale bar = 500 nm. N = 2 independent experiments. Free movement of hyperpolarized vesicles within the matrix indicate their detachment from the IBM, representing an extreme case of cristae independency within one mitochondrion (see Movie EV1 for time series).

Data information: For quantification of hyperpolarized vesicles, see values in main text.

Figure 9. Laser‐induced depolarization results in non‐instantaneous loss of ΔΨ_m along the single mitochondrion

Laser‐induced mitochondrial membrane depolarization in L6 myoblasts shows gradual depolarization of elements along the IMM, visualized using live‐cell Airyscan microscopy.

1. A

   Representative images of laser‐induced depolarization time series from L6 myoblast, stained with ΔΨ_m‐dependent dye, Rho123. Images show LUT color‐coded for Rho123 FI. White and blue pixels represent most‐ and least‐intense ΔΨ_m, respectively. See legend of LUT colors on the right. Note that, at the first time point (˜ 0.15 s), white arrow points to box marking area of mitochondrion exposed to phototoxic pulse of 2‐photon laser (≤ 5 ms), inducing depolarization. Ensuing frames show wavelike depolarization away from site of perturbation (arrowheads). Scale bar = 500 nm. N = 3 independent experiments. (See Movie EV2 for time series.)

2. B

   Profile plot showing Rho123 pixel intensity as function of distance along mitochondrion in (A). Note that, zero value in the x‐axis corresponds to top area of the mitochondrion shown in (A), near site of 2‐photon laser exposure (box). Red line corresponds to frame immediately before depolarization (˜ 0.15 s); green, blue, purple, and dark purple lines correspond to ˜ 0.30, ˜ 0.60, ˜ 1.05, and ˜ 2.10 s, respectively, after initial depolarization, showing wavelike dissipation of ΔΨ_m over time.

3. C–F

   Laser‐induced mitochondrial membrane depolarization in living human skin fibroblasts co‐stained with TMRE and MTG and imaged with Airyscan, showing gradual depolarization of elements along the IMM. (C) Time‐lapse images of merged green (MTG) and red (TMRE) channels following laser‐induced mitochondrial membrane depolarization. Note that, at the first time point (0.3 s), white arrow points to box marking area of mitochondrion exposed to phototoxic pulse of 2‐photon laser (≤ 5 ms), inducing depolarization. Depolarization is marked by the dissipation of the ΔΨ_m‐dependent dye (TMRE) while the green (MTG) signal persists (arrowheads). Scale bar = 1 μm. N = 4 independent experiments. (D) TMRE channel from (C), showing gradual loss of TMRE along the IMM (arrowheads). Note that, although depolarization has largely completed at the area near the phototoxic stimulus already at 0.9 s, some specific regions at the very top maintain ΔΨ_m while others depolarize. (E) Quantification of ΔΨ_m using TMRE FI from (C, D). Measurement of TMRE pixel intensities immediately after laser‐induced depolarization at site distal (≥ 10 μm) vs. proximal (≤ 1 μm) to box. N = 4 independent experiments; see specific P values in panel. (F) Quantification of (C, D). Percentage of mitochondria that depolarize in a wavelike (i.e., gradual) vs. instantaneous manner after laser‐induced depolarization. Note: imaging at high temporal resolution (˜ 100–500 ms/frame) reveals wavelike depolarizations predominate, suggesting the ΔΨ_m is composed of multiple, disparate electrochemical domains along the IMM. The time scale of propagation of depolarization is slower than the propagation of electrical phenomena. N = 4 independent experiments.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM.

Figure 10. Cristae are unique electrochemical domains, thus constituting independent bioenergetic units

We calculated the ΔΨ_Cr‐IBM between neighboring cristae over time as a surrogate measure of cristae individuality and functional independence, using live‐cell Airyscan imaging.

1. A

   Stability of cristae ΔΨ_Cr‐IBM over time. Representative images from time series of HeLa cells co‐stained with MTG and TMRE. Top sequence shows merged green (MTG) and red (TMRE) channels. Bottom sequence shows color‐coded LUT of TMRE signal of the same sequence shown above. White arrowheads in the image point to specific cristae that were measured over time. Scale bar = 500 nm. N = 5 independent experiments.

2. B

   Quantification of (A), showing average ΔΨ_Cr‐IBM of Cristae “A–D” over time. Note that, Cristae (C) and (D) maintain more than double the ΔΨ_Cr‐IBM of Cristae (A) and (B). Error bars indicate STDEV.

3. C

   Quantification of (A), showing standard deviation in ΔΨ_Cr‐IBM (mV) of individual cristae over time compared to the standard deviation in ΔΨ_Cr‐IBM (mV) among different cristae along a single mitochondrion. N = 5 independent experiments. Note that, there is less variability in ΔΨ_Cr‐IBM of a single crista over time within the same mitochondrion compared to the variability of ΔΨ_Cr‐IBM among cristae over time, indicating cristae possess unique bioenergetic properties.

4. D–G

   Airyscan time series of TMRE flickering in HeLa cells co‐stained with MTG and TMRE, showing that depolarization is uncoordinated along the IMM with some cristae depolarizing while neighboring cristae remain polarized. (D) Representative live‐cell Airyscan image from time series showing merged green (MTG) and red (TMRE) channels. Area marked by dashed box is shown in the time series in (E). Scale bar = 500 nm. (E) Time series of cropped region from (D), showing tip of mitochondrion (arrowhead) maintaining ΔΨ_m while neighboring region of same mitochondrion depolarizes. The capacity of some cristae to maintain polarity, while adjacent cristae transiently depolarize, indicates cristae can exhibit a significant amount of electrochemical autonomy. Scale bar = 500 nm. N = 3 independent experiments. (See [Link], [Link] for time series). (F) Zoomed‐in LUT of mitochondrion from (E) (˜ 2 s), showing some cristae remaining polarized (red) while adjacent cristae depolarize (blue). (See Movie EV5 for time series). Scale bar = 100 nm. (G) Quantification of (D–F), showing that during a flickering event, some cristae maintain their membrane potentials, while other cristae depolarize, resulting in large differences in ΔΨ_Cr‐IBM along the individual mitochondrion. N = 3 independent experiments.

Data information: Data were analyzed with 2‐tailed Student's t‐tests, and P values < 0.05 were considered statistically significant. Specific P values are indicated in the figure. Error bars indicate SEM, except in panel (B), which show STDEV.

Figure 11. Model of cristae as individual bioenergetic units

1. An equipotential representation of the IMM. In this model, the IMM functions as a continuous electrical cable where cristae and IBM have similar ΔΨ_m and where all cristae respond simultaneously to any change in ΔΨ_m occurring in the mitochondrion where they reside. Transition from a hetero‐potential to an equipotential, via cleavage of Opa1, for example, could play a role in normal mitochondrial physiology when high conductance and low ATP synthesis are favored, such as thermogenesis or adaptation to excess nutrient.

2. Cartoon showing IMM as equipotential.

3. Model of mitochondrial hetero‐potential, showing cristae as individual bioenergetic units. Electrochemical autonomy of cristae protects the mitochondrion confronting a bioenergetic crisis, where dysfunction of one or more cristae would not inevitably cause the entire organelle to fail. In this model, cristae and IBM can have different membrane potentials. Closed CJs generate significant differences in charge between the cristae membranes and IBM, which could be related to the concentration of ETC subunits in the lateral side of the cristae and F ₁ F ₀ ATP Synthase on the cristae rims.

4. Cartoon showing IMM as hetero‐potential.