Module 0: Origin & Endosymbiosis

In 1967 Lynn Sagan (later Margulis) published “On the Origin of Mitosing Cells” in the Journal of Theoretical Biology after more than a dozen journal rejections. She proposed that three organelles in the eukaryotic cell—mitochondria, plastids, and (controversially) undulipodia (flagella)—originated as free-living prokaryotes that were engulfed by a host cell and never digested. This is the Serial Endosymbiotic Theory (SET).

The SET was an old idea in new clothes. Andreas Schimper (1883), Konstantin Mereschkowski (1905), and Ivan Wallin (1927) had all proposed bacterial ancestry for plastids or mitochondria; Wallin even claimed to have cultured mitochondria in vitro (he had not). By the mid-20th century the hypothesis had been dismissed as fringe. Margulis revived it, added rigor, and backed it with the newly available electron-microscopy record—double membranes, bacterial-sized ribosomes inside the organelles, a circular DNA molecule with a bacterial G+C content, antibiotic sensitivities that matched bacteria rather than cytoplasm.

SET was fully vindicated by molecular phylogenetics in the 1970s and 1980s. The rise of 16S ribosomal-RNA sequencing in Carl Woese’s laboratory placed mitochondrial rRNA squarely inside the α-proteobacterial clade, with closest affinities to modern Rickettsiales and SAR11-like marine aerobes. The plastid branch landed equally clearly inside the cyanobacteria.

Michael Gray’s review in Science (1999) consolidated the 16S rRNA, cytochrome oxidase, and ATP synthase phylogenies. The mitochondrial ancestor branched inside the α-proteobacteria—a single origin, an unambiguous topology, though the precise sister taxon (Rickettsiales? SAR11?) depended on the gene chosen and the outgroup tree. Modern genome-scale phylogenomics (Wang & Wu 2015, Martíjn et al. 2018) still converges on a deep-branching α-proteobacterium, possibly related to the free-living marine SAR11 clade of Pelagibacter.

The phylogenetic signal is preserved in the 13 mitochondrial-encoded proteins of the electron transport chain (cytochrome b, three subunits of Complex IV, seven subunits of Complex I, two subunits of ATP synthase). These are the genes that refused to migrate to the nucleus, for reasons that remain debated (the CoRR hypothesis—co-location for redox regulation—of John Allen 2015 is the leading model). Their bacterial origin is transparent in sequence.

A crucial point from Gray is that mitochondria share a single common ancestor. All extant eukaryotes—animals, plants, fungi, protists—descend from a lineage that already contained mitochondria. The few apparently amitochondriate protists (diplomonads, parabasalids, microsporidia, pelobionts, entamoebids) are not primitive; they are secondarily derived, harboring reduced organelles called hydrogenosomes or mitosomes.

Margulis left the host cell ambiguous. The question of “who swallowed the bacterium?” was finally resolved in the 2010s by metagenomic sequencing of deep-sea sediment from Loki’s Castle (Spang et al. 2015, Nature; Zaremba-Niedzwiedzka et al. 2017, Nature). The Asgard archaeal superphylum—Lokiarchaeota, Thorarchaeota, Odinarchaeota, Heimdallarchaeota, Helarchaeota—encodes hundreds of eukaryotic signature proteins (ESPs): actin homologs (profilin, gelsolin-like, Arp2/3), tubulin relatives, Ras-superfamily GTPases, ubiquitin modifier cascades, ESCRT machinery.

Roger, Muñoz-Gómez & Kamikawa (2017, Curr. Biol.) synthesized the state of the art into a coherent eukaryogenesis model: the host of the proto-mitochondrion was an Asgard archaeon, most likely a close relative of Lokiarchaeum; the nucleus, the endomembrane system, and the cytoskeleton all emerged during the integration with the bacterial symbiont, not before. In 2020 Imachi et al. reported the first cultivated Asgard archaeon, Candidatus Prometheoarchaeum syntrophicum—a small coccoid cell with long membrane protrusions that interact with partner bacteria.

A consequence: the eukaryotic cell is a chimera. The cytoplasm, ribosomes, replication and transcription machinery trace to archaea; the mitochondria and the phospholipids of the cellular membranes (a bacterial-type ester-linked bilayer replaced the archaeal ether-linked system) trace to bacteria. This transition is possibly the single most unusual evolutionary event in 3.8 billion years of life on Earth.

The mitochondrial endosymbiosis happened once, at the stem of all eukaryotes. Despite tens of thousands of independent origins of eukaryotic lineages spanning 1.5+ billion years, no other bacterial endosymbiosis has produced a new branch of the tree of life with comparable complexity. Why?

Nasir, Forterre & Caetano-Anollés (2020) and Martin, Tielens & Müller (2020) argue that this singularity reflects a genuinely rare evolutionary transition. The prerequisites are stringent:

• A host and a symbiont metabolically interlocked (syntrophy): H₂ exchange, or a redox partnership
• An engulfment event (phagocytosis or cell-cell fusion) without digestion
• Survival of the symbiont through many host divisions: vertical transmission
• Gene transfer to the host nucleus—hundreds of independent transfer events
• Targeting machinery to return the transferred products to the organelle

Each step has non-trivial probability; their product is small. The fact that plastids arose independently several times (primary endosymbiosis in Archaeplastida; secondary/tertiary in chromalveolates, chlorarachniophytes, euglenoids) suggests that after the invention of eukaryotic cell biology—phagocytosis, endomembrane trafficking, a nucleus to receive transferred genes—endosymbiosis becomes much easier. The first event, starting from a prokaryote, seems to require a singular coincidence.

Nick Lane and William Martin’s 2010 Nature paper, “The energetics of genome complexity,” asks why prokaryotes have stayed bacteria for 3+ billion years while eukaryotes evolved nuclei, meiosis, sex, mitosis, multicellularity, brains. Their answer: bioenergetic constraint.

A bacterium runs its electron-transport chain on its plasma membrane. Its power output scales with surface area \(S \sim r^2\) while the genome it must maintain scales with cell volume \(V \sim r^3\). The available ATP per unit of DNAtherefore falls as \(S/V \sim 1/r\): a giant bacterium is energy-starved per gene.

A eukaryote with \(N_{\text{mito}}\) mitochondria, each of which amplifies its inner membrane surface by a factor \(A_{\text{crist}} \sim 25\text{-}30\)through cristae, hosts a total bioenergetic membrane that scales linearly with cell volume. The power per gene therefore becomes size-independent—and the cell can afford orders of magnitude more genes.

The empirical evidence is compelling. Truly giant bacteria do exist—Thiomargarita namibiensis (750 μm), Epulopiscium fishelsoni (700 μm)—but they pay by carrying tens of thousands of copies of their genome to keep up with the volume-scaling metabolic demand. They have not radiated. Bacteria cannot use their membrane estate to grow complex because that membrane must also run the ETC.

Mitochondria decouple information storage (the nucleus) from energy generation (many bioenergetic membranes in each cell). This is Lane’s central claim: eukaryotic complexity is a thermodynamic phenomenon, not a merely genetic one.

When the α-proteobacterial ancestor entered the archaeal host it carried ~3000–4000 genes. The modern human mitochondrial genome encodes only 37. Where did the others go?

The answer is endosymbiotic gene transfer (EGT): over evolutionary time, bacterial DNA leaked into the host nucleus, was integrated there, and evolved a targeting peptide that returned the protein to the organelle. Timmis, Ayliffe, Huang & Martin (2004, Nat. Rev. Genet.) estimate that over 90% of the proteobacterial ancestor’s gene complement moved to the nucleus. The nuclear genome today contains roughly 1000–1500 genes of mitochondrial origin.

Genes that moved to the nucleus needed a re-entry ticket. An N-terminal mitochondrial targeting sequence (MTS)—an amphipathic α-helix of ~20–50 residues, rich in positive and hydrophobic residues—was added, typically by recombination with a pre-existing targeted gene. The MTS is recognized by the TOM complex (translocase of the outer membrane) at the OMM and then passed to the TIM23 complexat the inner membrane, after which it is cleaved by mitochondrial processing peptidase (MPP).

In plastids, an analogous machinery (TOC/TIC) handles import. Some eukaryotic lineages with secondary plastids (derived by engulfing an already-plastid-bearing alga) use a SELMA translocon (“symbiont-specific ERAD-like machinery”) to cross the extra membrane layers; the SELMA motif is a feature of four-membrane plastids in diatoms and related organisms. No SELMA-equivalent exists in mitochondria; the two-membrane topology is handled by TOM/TIM alone.

EGT is still occurring. NUMTs (nuclear-mitochondrial DNA segments) are fragments of mtDNA that have recently jumped to the nucleus; the human genome contains hundreds. The transfer is one-way: once in the nucleus, a gene competes under a completely different selective regime.

The modern mitochondrion is not just a bacterium in a double membrane; its inner membrane is folded into invaginations called cristae. Cristae amplify inner-membrane surface area by ~25–30× and concentrate the ETC into locally curved compartments with a high pH at the crista junction.

Cristae topology varies with tissue and organism: lamellar (flattened) in most vertebrate tissues, tubular in steroidogenic cells (adrenal cortex, Leydig cells, corpus luteum) to favor hydroxylase chemistry, and paracrystalline in insect flight muscle and brown adipose tissue where the ETC is packed almost as a 2D crystal for maximum ATP throughput.

Cristae shape is set by the MICOS complex(Mitofilin/MIC60, MIC19, MIC10, MIC26/27) at crista junctions (Rabl et al. 2009; Pfanner et al. 2014) and by OPA1 oligomerization on the matrix face. A reduction of OPA1 unravels cristae, exposes cytochrome c, and initiates the intrinsic apoptotic cascade (Scorrano et al. 2002, Dev. Cell).

The diversity of cristae geometry across eukaryotes suggests the MICOS/OPA1 system has been remodeled many times to match each lineage’s metabolic style.

Not every eukaryote runs aerobic OxPhos. Parasitic and anaerobic lineages have reduced their mitochondria drastically, and for decades these organelles were mistaken for primitive (pre-mitochondrial) states. Molecular markers—heat-shock proteins, iron-sulfur cluster biosynthesis—eventually revealed them as degenerate mitochondria.

• Hydrogenosomes (trichomonads, anaerobic ciliates, chytrid fungi): produce ATP via substrate-level phosphorylation with pyruvate: ferredoxin oxidoreductase and an Fe-hydrogenase, exhaling H₂. No ETC, usually no DNA. Müller et al. (2012, Microbiol. Mol. Biol. Rev.) reviews.
• Mitosomes (Giardia lamblia, Entamoeba histolytica, microsporidia): minimal organelles (~200–400 nm) whose sole retained function is iron-sulfur cluster biosynthesis. No ATP production, no cristae, often no genome. Their sole evolutionary raison d’être is that Fe-S assembly must occur in an enclosed compartment for chemical reasons (Tovar et al. 2003,Nature).
• Anaerobic mitochondria (Nyctotherus, intestinal parasites): retain partial ETC using alternative terminal acceptors (fumarate, nitrate).

That mitosomes exist and are universally derived from full mitochondria is, perhaps, the strongest single argument that the endosymbiotic event occurred once at the stem of eukaryotes: every extant amitochondriate lineage carries a clear molecular fingerprint of the lost organelle.

Rickettsia prowazekii, the agent of epidemic typhus, cannot grow outside a eukaryotic host. It is an obligate intracellular parasite with a genome of only ~1.1 Mb (Andersson et al. 1998). Its metabolism is strikingly reduced: it imports ATP directly from the host cytoplasm via a ADP/ATP translocase that is structurally and mechanistically identical to the one mitochondria use to export ATP to the host—just run backwards.

Rickettsial biology is, in effect, a controlled experiment in endosymbiosis: the early stages of the mitochondrial integration may have resembled a parasitic intrusion that became mutualistic only when gene transfer stabilized the system. The closely related Wolbachia (a symbiont of arthropods and nematodes) sits in the ambiguous zone between parasite and mutualist—sometimes required for host reproduction, sometimes a manipulator.

Human OxPhos complexes are chimeras: Complex I has 7 mtDNA-encoded and ~38 nuclear-encoded subunits; Complex IV has 3 mtDNA + 11 nuclear; ATP synthase has 2 mtDNA + ~15 nuclear. Every complex must be assembled from parts encoded in two genomes with radically different evolutionary dynamics: mtDNA mutates 10–100× faster than nDNA and is inherited uniparentally.

This imposes mitonuclear coadaptation. A mtDNA variant that changes a mitochondrial Complex I subunit must be matched by a compensating nuclear variant. Mismatched pairs—introduced by crossing distant populations— cause hybrid breakdown in many species (Dowling, Friberg & Lindell 2008,Trends Ecol. Evol.; Sloan et al. 2018, Curr. Biol.).

Mitonuclear incompatibility is one reason mitochondrial replacement therapy (“three-parent IVF”) is medically regulated: replacing the mitochondria of a fertilized egg can introduce a mismatch with the nuclear background. We return to this in Module 8.