henhaobio
Sign in

Proceedings of the National Academy of SciencesCell BiologyPeer-reviewed journal article

In situ cryo-ET visualization of mitochondrial depolarization and mitophagic engulfment

Kevin M. RoseiD, Eric HerrmanniD, Eve V. KakudjiiD, Javier LizarrondoiD, A. Yasemin Celebi, Florian WilflingiD, Samantha C. LewisiD, James H. HurleyiD

PDFDOIPreprint
In-depth full-text digest available · Jump to digest ↓

Abstract

Defective mitochondrial quality control in response to loss of mitochondrial membrane polarization is implicated in Parkinsons disease by mutations in PINK1 and PRKN. Application of in situ cryo-electron tomography (cryo-ET) made it possible to visualize the consequences of mitochondrial depolarization at higher resolution than heretofore attainable. Parkin-expressing U2OS cells were treated with the depolarizing agents oligomycin and antimycin A (OA), subjected to cryo-FIB milling, and mitochondrial structure was characterized by in situ cryo-ET. Phagophores were visualized in association with mitochondrial fragments. Bridge-like lipid transporter (BLTP) densities potentially corresponding to ATG2A were seen connected to mitophagic phagophores. Mitochondria in OA-treated cells were fragmented and devoid of matrix calcium phosphate crystals. The intermembrane gap of cristae was narrowed and the intermembrane volume reduced, and some fragments were devoid of cristae. A subpopulation of ATP synthases re-localized from cristae to the inner boundary membrane (IBM) apposed to the outer membrane (OMM). The structure of the dome-shaped prohibitin complex, a dodecamer of PHB1-PHB2 dimers, was determined in situ by sub-tomogram averaging in untreated and treated cells and found to exist in open and closed conformations, with the closed conformation is enriched by OA treatment. These findings provide a set of native snapshots of the manifold nano-structural consequences of mitochondrial depolarization and provide a baseline for future in situ dissection of Parkin-dependent mitophagy.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (25 Mar 2025) · CC BY 4.0

Cryo-ET of Parkin-expressing U2OS cells shows oligomycin/antimycin depolarization moves some ATP synthase to the inner boundary membrane, abolishes matrix calcium granules, and favors closed prohibitin domes.

Depolarization triggers PINK1-Parkin mitophagy (implicated in Parkinson's disease) but also perturbs mitochondria broadly, so in situ structural study of mitophagy first needs a baseline of depolarization effects. This work provides that baseline in cells at nanometer scale, reporting ATP synthase redistribution and a C1 prohibitin dome model that, unlike a prior C11-symmetrized model, fills the density with inter-subunit packing typical of its protein family.

Key results

  1. ShownSub-tomogram averaging resolved the native prohibitin complex as a C1 dome of twelve PHB1-PHB2 heterodimers in closed and open conformations. Depolarization with oligomycin plus antimycin A (OA) shifted the balance toward the closed state.

    ~2500 manually picked particles; both classes ~20 Å; 12 density peaks under C1; AlphaFold3 screen of 7 to 16 heterodimer pairs favored 11-12 copies; closed dome ~20 nm wide, 10 nm tall vs open 8 nm with ~25% of wall missing; text states open fraction fell from 75% to 56%, but legend counts (untreated 930 open/677 closed vs OA 297 open/511 closed; Fisher's exact test) do not match those percentages. · Structural characterization of the Prohibitin…; Cryo-electron tomography data processing and…; Figure legends

  2. ShownAfter depolarization, part of the F0F1 ATP synthase population moved from cristae arrays to the flat inner boundary membrane (IBM). Live-cell Airyscan imaging in U2OS cells without Parkin overexpression confirmed this shift.

    Template matching seeded by 503 manual picks: untreated 623 cristae vs 17 IBM complexes; OA 413 cristae vs 257 IBM (Fisher's exact test), nearly 40% relocalized per authors. Airyscan of ATP5F1B-tGFP vs mCherry-TOMM20 showed significant IBM enrichment during OA (n=3-4 cells per replicate, three replicates, paired t test). · Mis-localization of ATP synthases to the IBM…; Discussion; Live Airyscan microscopy of mitochondrial…; Figure legends

  3. ShownOA-treated mitochondria lost every electron-dense matrix granule (presumed calcium phosphate). They also had fewer and thinner cristae, which points to a smaller intracristal space.

    Segmented cristae per volume fell from 12 to 6 (n=37 untreated vs n=81 OA tomograms); cristae surface-area-to-volume ratio rose from 3.2 to 3.6 nm (n=25 cristae per condition, unpaired t test); Dragonfly segmentation found ~32 granule clusters per untreated tomogram vs zero after OA. · Depolarization alters the matrix architecture of…; Figure legends

  4. SuggestedPhagophores at several stages, including multilamellar envelopes, engaged depolarized mitochondrial fragments. Rods bridging a membrane sheet (possibly an omegasome) and a phagophore were tentatively assigned to ATG2A, the main autophagy bridge-like lipid transporter, from their ~20 nm length and VPS13C-like layout.

    14 of n=85 OA fragments (~17%) were targeted; earliest phagophore indentation ~1 nm, a cup opening of ~250 nm and a ~500 nm mitophagosome; cup rims within 10 nm of the outer membrane; bridging rods ~20 nm long; stacked phagophore layers spaced less than 10 nm apart. · Phagophores target and envelop mitochondrial…; Discussion; Figure legends

  5. ShownA Parkin-expressing U2OS reporter line showed OA-induced Parkin recruitment and lysosome-dependent mitophagy flux. It also supported fluorescence-guided cryo-focused ion beam (FIB) milling, yielding a large in situ tomography dataset of untreated and OA-fragmented mitochondria.

    Parkin recruitment required combined oligomycin plus antimycin A; ~28% of Su9-HALO processed after 3 h of 3 μM OA, reversed by BafA1; mitochondrial section diameter 1261 nm (n=66) untreated vs 732 nm (n=93) OA; 92 lamellae and 157 tomograms (47 untreated, 110 OA), 141 analyzed. · Validation of Parkin-mitophagy CLEM reporter…; Figure legends

Methods, data and reproducibility

Parkin-expressing U2OS reporter cells were treated for 3 h with oligomycin plus antimycin A, vitrified, milled by fluorescence-guided cryo-focused ion beam and imaged by cryo-electron tomography. Membranes and granules were segmented, ATP synthase was template-matched, and prohibitin was sub-tomogram averaged and modeled with AlphaFold3/ISOLDE; Airyscan imaging served as validation.

  • Stable U2OS osteosarcoma lines expressing mCherry-Parkin with BFP-mito or Su9-HALO, used for Parkin recruitment and mitophagy flux assays · human · n = 3 fields/condition/replicate (recruitment); quadruplicates (flux)
  • Cryo-FIB lamellae (~200 nm) and tilt-series tomograms of untreated vs OA-treated Parkin-expressing U2OS cells · human (U2OS) · n = 92 lamellae (27/65); 157 tomograms (47/110); 141 analyzed
  • ATP synthase particles used for sub-tomogram averaging and PyTOM template matching on membrane-masked tomograms · human (U2OS) · n = 503 manual picks; 675 template-match picks · EMDB-45272 (reference map); PDB 8H9T (docked model)
  • Prohibitin complex particles from all tomograms, 3D classified into open and closed conformations · human (U2OS) · n = ~2500 manually picked particles
  • Live-cell Airyscan imaging of U2OS cells without Parkin overexpression, transiently expressing mCherry-TOMM20 and ATP5F1B-turboGFP during OA treatment · human · n = 3-4 cells per replicate, three independent replicates
  • Published HFLC/K (bacterial prohibitin-family complex) structures in open and closed states, used for structural comparison · bacteria (species not stated) · PDB 9CZ2 (open); PDB 7VHP (closed)
  • Prior in situ human prohibitin model built with C11 symmetry, and a putative prohibitin density map from Chlamydomonas, used for comparison · human (8RRH); Chlamydomonas (EMDB-50212) · PDB 8RRH; EMDB-50212
Methods & tools
Cryo-FIB milling (Aquilos 2 with integrated fluorescence, Auto-TEM), Scipion (processing framework), MotionCor3, CTFfind 5, AreTomo2 (tilt-series alignment and reconstruction), Membrain (membrane segmentation), Dragonfly (calcium cluster segmentation), Relion 5 (sub-tomogram averaging, 3D classification), PyTOM (template matching), AlphaFold3, ISOLDE, Arivis Vision4D (Airyscan segmentation)
Reagents & environment
Relion 5, CTFfind 5, MotionCor3, AreTomo2, GraphPad Prism 10 (10.1.0), Arivis Vision4D 4.2.1, ZEN Blue 3.7; untreated data on Krios G4 (Falcon 4i, 10 eV slit), OA data on Krios G2/G3 (K3, 25 eV slit).
Reproducibility gaps
Maps, models and raw tomograms are described as being deposited to PDB, EMDB and EMPIAR, but no accession IDs are given; no code repository is cited; step-by-step protocols are linked on protocols.io.
Limitations (6)
  • Too few phagophore events were captured to establish whether mitophagy preferentially targets the most severely distorted mitochondrial fragments; the authors say more events must be analyzed. · stated by the authors
  • It is unresolved whether ATP synthase relocation causes or follows cristae remodeling. Oligomycin binding to the F0 sector might drive relocation directly, and the authors say the mechanism may need sub-nanometer resolution or conformation-specific template matching. · stated by the authors
  • The open prohibitin state has lower density quality and resolution than the closed state, and the authors judge that it probably averages an ensemble of related open conformations rather than one defined structure. · stated by the authors
  • Tomography covered only a 3 h depolarization snapshot, and PINK1-Parkin-TOMM20 complexes were not localized in situ. The authors note that finer time sampling is needed to link these structures to mitophagic uptake. · stated by the authors
  • Untreated tomograms came from a Krios G4 with a Falcon 4i detector and 10 eV slit, while OA tomograms came from G2/G3 instruments with a K3 detector and 25 eV slit. This could bias template matching, granule detection and class assignment between conditions. · noted in this summary
  • The ATP synthase and prohibitin comparisons apply Fisher's exact tests to pooled particle counts. This treats particles as independent despite clustering within mitochondria, tomograms and cells, which may overstate significance. · noted in this summary

Read this if you study PINK1-Parkin mitophagy, cristae and inner-membrane organization, or prohibitin-like dome complexes and want quantitative in situ cryo-ET snapshots of mitochondrial depolarization.

Citations
25
FWCI
17.33
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISLin '20Georgiou '26Kraus '22Maruszczak '22Antico '21Ravindran '20
68% semantic matchCell Biology
C.-C. Lin, J. Yan, M. D. Kapur+9 more·bioRxiv·Feb 28, 2020