Module 1: Ultrastructure & Membranes

Mitochondria had been visible in light microscopy since Altmann (1890) described “bioblasts”—granules of uniform size inside cells—but their internal structure was inaccessible. In 1952 George Palade, working at the Rockefeller Institute, published an electron-microscopy study in the Anatomical Record that first resolved the double-membrane architecture and the internal projections he named cristae mitochondriales (“crests of the mitochondria,” Latin).

Palade’s images showed the familiar picture still reproduced in every textbook: a smooth outer membrane, a highly folded inner membrane whose projections pushed into the interior, a clear matrix space, and a thin rind between the two bilayers (the intermembrane space). Palade shared the 1974 Nobel Prize with Claude and de Duve for founding cell biology with this work.

The outer mitochondrial membrane (OMM) is a ~7 nm phospholipid bilayer of roughly “ordinary” composition— phosphatidylcholine (PC) 54%, phosphatidylethanolamine (PE) 29%, phosphatidylinositol (PI) ~4%, cholesterol <5%—but it is pierced by the voltage-dependent anion channel (VDAC / porin) at very high density (~50,000 copies per mitochondrion, Colombini 2004). VDAC gates passage of molecules smaller than ~5 kDa; ions, metabolites, ATP, ADP, P_i, NAD⁺, and pyruvate move essentially freely.

The inner mitochondrial membrane (IMM) is radically different. It is thinner (~5 nm), extraordinarily protein-rich (75% protein by mass—the densest protein-to-lipid ratio of any biological membrane), highly negatively charged on the matrix face, and impermeable to ions. Its tightness is what makes it possible to maintain the ~180 mV protonmotive force that drives ATP synthesis. Specific transporters (ANT, UCP, phosphate carrier, dicarboxylate carrier, etc.) let selected metabolites cross.

The OMM and IMM enclose two aqueous compartments:

• Intermembrane space (IMS): ~10 nm thick peripherally, 10–30 nm inside cristae. Contains cytochrome c, adenylate kinase, creatine kinase, and the apoptosis executioners Smac/DIABLO. The IMS pH is close to the cytosolic value (~7.0–7.2) because the OMM is porous.
• Matrix: contains TCA-cycle enzymes, β-oxidation enzymes, PDH complex, amino-acid catabolism enzymes, mtDNA nucleoids, mitoribosomes, tRNAs, and the protein-import machinery. The matrix is alkaline (~pH 7.9) and negatively charged.

Cristae are not simple infoldings but distinct IMM subcompartments with their own topology. Electron tomography (Perkins et al. 1997; Frey & Mannella 2000; Mannella 2006) revealed that cristae open to the IMS through narrow 15–25 nm cristae junctions rather than broad continuous folds. The cristae compartment is semi-enclosed: it concentrates cytochrome c and maintains a locally higher H⁺ gradient.

Cristae shape is dynamic: OPA1 oligomer stability, cardiolipin remodeling by taffazin, and MICOS assembly respond to metabolic state. During apoptosis cristae unfold and cytochrome c is released; during starvation, mitochondrial elongation and cristae compaction occur (Gomes, Di Benedetto & Scorrano 2011).

The MICOS complex (mitochondrial contact site and cristae-organizing system) is a ~1 MDa IMM assembly that stabilizes cristae junctions (Pfanner et al. 2014, J. Cell Biol.). First identified as “Mitofilin” or “Fcj1” (formation of crista junctions, Rabl et al. 2009, J. Cell Biol.), the full complex now includes:

• MIC60 / Mitofilin: central scaffold; yeast Fcj1; “boomerang” shape bridging IMM leaflets
• MIC19 / CHCHD3: Mic60 partner, also at contact sites
• MIC10 / MINOS1: small transmembrane, induces membrane bending via tilted helix dimerization (Barbot et al. 2015)
• MIC26 / APOO and MIC27 / APOOL: paralogs that bind cardiolipin
• MIC13 / QIL1: recently discovered, links MIC10 and MIC60 subcomplexes

MIC60 and MIC10 form two partially independent subcomplexes. MIC60 also bridges the IMM to the OMM at contact sites, where it interacts with components of the OMM protein-translocase TOM and with SAM (sorting and assembly machinery for β-barrel proteins).

Loss of MIC60 or MIC10 produces mitochondria with disorganized cristae, onion-like membrane whorls, and severely compromised OxPhos. MICOS mutations cause human disease: MIC13 mutations cause a hepatic-encephalopathic phenotype (Guarani et al. 2016); MIC26 variants cause Parkinsonism-dementia syndromes.

OPA1 (optic atrophy 1) is a dynamin-family GTPase anchored in the IMM. It plays two linked roles: it fuses the IMM during mitochondrial fusion (covered in Module 6) and it stabilizes cristae junctions by oligomerizing at the matrix face. Frezza et al. (2006, Cell) showed that OPA1 depletion triggers cristae widening, cytochrome c redistribution, and apoptosis.

OPA1 exists in two forms, produced by proteolytic cleavage:

• L-OPA1 (long): IMM-anchored; assembles into fusion-competent oligomers
• S-OPA1 (short): soluble in the IMS; cannot drive fusion by itself

Two proteases balance L-/S- OPA1: YME1L(constitutive) and OMA1 (stress-activated). Loss of \(\Delta\Psi\), heat shock, or ATP depletion activates OMA1, which converts L-OPA1 to S-OPA1 and inhibits fusion. This is how mitochondria “sense” damage: unhealthy mitochondria fall out of the fusion cycle and are segregated for mitophagy.

Heterozygous OPA1 mutations cause autosomal dominant optic atrophy (DOA), the most common inherited optic neuropathy, with a selective vulnerability of retinal ganglion cells. This tissue-specific sensitivity reflects the high OxPhos demand of unmyelinated RGC axons in the optic nerve.

Cardiolipin (CL) is a tetra-acyl phospholipid with two phosphate head-groups connected by a glycerol bridge: essentially two PG molecules linked tail-to-tail. It is the only phospholipid with four fatty-acyl tails; it carries two negative charges at physiological pH. CL is found almost exclusively in the IMM (~10–20% of IMM lipids by mass; <2% of OMM) and in the plasma membranes of bacteria, reflecting its endosymbiotic origin.

Cardiolipin has three functions:

• Curvature stabilization: the conical shape of the CL molecule (small head, wide tails) produces negative spontaneous curvature that is ideal for the inner leaflet of curved cristae (Horvath & Daum 2013).
• Supercomplex assembly: CL is an obligate glue for respiratory supercomplexes. Complex I–III₂–IV (“respirasome”) assembly requires CL binding sites on each subunit (Pfeiffer et al. 2003; Schägger & Pfeiffer 2000).
• Cytochrome c anchoring: CL binds cytochrome c via electrostatic interactions, tethering it to the IMM and modulating its apoptotic release.

CL is synthesized in the matrix from phosphatidic acid via CDP-DAG, and its acyl composition is remodeled post-synthesis by tafazzin (TAZ), an IMM acyltransferase. Loss-of-function mutations in TAZ cause Barth syndrome: X-linked cardiomyopathy, skeletal myopathy, neutropenia, with abnormally high monolyso-cardiolipin (MLCL) and low mature tetralinoleoyl-CL. Barth syndrome is the paradigmatic lipid mitochondrial disease.

Mitochondria do not float as isolated organelles. Vance (1990) identified a biochemical fraction of the endoplasmic reticulum tightly associated with mitochondria—the MAMs—that engages in privileged exchange of Ca²⁺ and phospholipids. The gap between the ER membrane and the OMM at MAMs is 10–30 nm, bridged by protein tethers.

The IP3R–Grp75–VDAC complex (Hayashi et al. 2009; Szabadkai et al. 2006) couples ER Ca²⁺ release to mitochondrial uptake. IP₃ receptors on the ER cluster at MAMs; the chaperone Grp75 bridges to VDAC on the OMM; Ca²⁺ then crosses the IMM via the mitochondrial calcium uniporter (MCU). Local [Ca²⁺] at MAMs can reach 20 μM, far above cytosolic 100 nM, which is why MCU (K_d ~ 5 μM) is loaded efficiently only at MAMs.

Ca²⁺ uptake stimulates three matrix dehydrogenases (pyruvate DH, isocitrate DH, α-ketoglutarate DH), linking ER signaling to OxPhos activation. Excessive loading opens the mitochondrial permeability transition pore (MPTP) and triggers cell death (Rizzuto et al. 2012, Nat. Rev. Mol. Cell Biol.).

MAMs are now recognized as a hub for neurodegeneration: presenilins (Alzheimer’s), Parkin, PINK1, DJ-1, SOD1, and VAPB (ALS) all localize to or regulate MAMs. Disordered MAM-mediated Ca²⁺ transfer is a shared pathogenic feature.

Roughly 99% of mitochondrial proteins are nuclear-encoded and imported post-translationally from the cytosol. They cross up to four compartments (cytosol → OMM → IMS → IMM → matrix) through dedicated translocases (Neupert & Herrmann 2007; Wiedemann & Pfanner 2017).

• TOM complex (translocase of the outer membrane): Tom40 forms the β-barrel channel; Tom20, Tom22, Tom70 are receptors. Tom20 recognizes N-terminal amphipathic MTS; Tom70 binds hydrophobic internal targeting signals (for carrier-family proteins).
• TIM23 complex: presequence translocase of the IMM. Delivers MTS-bearing pre-proteins into the matrix. Powered by\(\Delta\Psi\) (electrophoresis) and by the PAM motor.
• TIM22 complex: carrier translocase, inserts polytopic IMM proteins (ADP/ATP translocator, phosphate carrier, UCPs) into the lipid bilayer. Independent of MTS; uses internal targeting signals.
• PAM (presequence-associated motor): mtHsp70, Pam16, Pam18, Mge1. ATP-driven ratchet that pulls the unfolded pre-protein into the matrix.
• SAM complex: sorting and assembly machinery for OMM β-barrels (VDAC, Tom40 itself, Sam50).
• MIA40 / Erv1 pathway: disulfide-relay import into the IMS for small cysteine-rich proteins.

The mitochondrial targeting sequence (MTS) is the N-terminal 20–50 residue extension of imported proteins. It forms an amphipathic α-helix with a positive face (several Arg/Lys residues) and a hydrophobic face; it is recognized by Tom20 through an induced-fit interaction and cleaved by matrix processing peptidase (MPP) upon arrival.

Hsp70/Hsp90 chaperones keep newly translated pre-proteins unfolded in the cytosol so they can thread through Tom40 (a ~20 Å channel incompatible with a folded domain). On arrival in the matrix, mtHsp70 and mtHsp60 (GroEL-type chaperonin) refold the protein. The import machinery is thus tightly coupled to the cellular chaperone network.

mtDNA does not float naked in the matrix. It is packaged into ~500 discrete nucleoids per mitochondrion, each ~100 nm in diameter and containing 1–2 copies of the 16 569 bp genome (Kukat et al. 2011,PNAS). The major protein component is TFAM(mitochondrial transcription factor A, Ngo et al. 2011), an HMG-box DNA-bending protein analogous to bacterial HU but evolved from nuclear HMG proteins.

Nucleoids are tethered to the IMM via interactions with POLG (polymerase γ) and the mitochondrial single-stranded DNA-binding protein (mtSSB). They associate preferentially with cristae junctions where the MICOS complex provides anchoring sites.

Each nucleoid is an autonomous unit for replication and transcription. Heteroplasmy— the presence of different mtDNA variants in the same cell—is partitioned at the nucleoid level: mutant and wild-type genomes can coexist in neighboring nucleoids and segregate unequally during mitophagy or mitochondrial division (Kukat & Larsson 2013). We develop heteroplasmy drift in Module 2.

Mitochondrial size spans orders of magnitude. Individual mitochondria range from 0.3 μm short spheroids to >10 μm elongated rods; in some cell types (cardiomyocytes, yeast buds, some neurons) they form a continuous reticulum extending many cell diameters (“one mitochondrion per cell” topology).

Number is equally variable:

• Erythrocyte: 0 (mitochondria are eliminated during terminal differentiation)
• Hepatocyte: ~1000–2000 per cell
• Cardiac myocyte: ~5000–8000 per cell; ~35% of cell volume
• Oocyte: ~100 000–400 000 (the reservoir for the embryo)
• Skeletal muscle (slow oxidative fiber): ~5000 / fiber segment, subsarcolemmal + intermyofibrillar populations

Mitochondrial network topology is set by the balance between fusion (OPA1, MFN1, MFN2) and fission (Drp1), covered in detail in Module 6. Neurons have especially dynamic networks because mitochondria must be actively trafficked along axons (kinesin-1 / KIF5B, Miro, Milton) to reach distal synapses.