The Endosymbiotic Theory

The endosymbiotic theory proposes that two of the eukaryotic cell's defining organelles — the mitochondrion and the chloroplast — originated as free-living prokaryotes that took up permanent residence inside a host cell. "Endosymbiosis" literally means "living together, inside": one organism (the symbiont) lives within the body, and in this case within the cytoplasm, of another (the host). Unlike a transient parasitic or commensal relationship, this symbiosis became obligate and heritable — the engulfed cell's descendants have divided alongside the host cell for well over a billion years, coevolving into an organelle that can no longer live independently and that the host cell can no longer live without. Two separate endosymbiotic events are now supported by strong evidence. The first and older event gave rise to the mitochondrion, acquired when a host cell — probably an archaeon or an archaeon-related lineage — engulfed an alphaproteobacterium (a group related to modern Rickettsia and Agrobacterium). This event is thought to have occurred at or near the very origin of the eukaryotic lineage itself, meaning essentially all eukaryotes either possess mitochondria or possess reduced remnant organelles (mitosomes or hydrogenosomes) descended from them. The second, later event produced the chloroplast (plastid), acquired when a eukaryotic cell — already possessing mitochondria — engulfed a photosynthetic cyanobacterium. This "primary" plastid endosymbiosis gave rise to the lineage that includes plants, green algae, red algae, and glaucophytes. Later still, several eukaryotic lineages acquired chloroplasts secondarily, by engulfing a already-photosynthetic eukaryote rather than a cyanobacterium directly — explaining why some algae (such as euglenids and dinoflagellates) have plastids surrounded by three or four membranes rather than two. Crucially, this is not merely an analogy to symbiosis — it is a literal claim about ancestry, testable using the tools of molecular phylogenetics, genomics, and biochemistry, and it makes specific, falsifiable predictions: organelle genomes should be circular, organelle ribosomes should resemble bacterial ribosomes, and organelle genes should form phylogenetic trees that nest inside specific bacterial groups rather than clustering with the eukaryotic nuclear genome. Every one of these predictions has been confirmed.

Step 1: Engulfment. A host cell, capable of phagocytosis-like membrane deformation, engulfs a prokaryotic cell rather than digesting it. This produces a bacterium wrapped in a host-derived membrane (the outer membrane of the future organelle) surrounding the bacterium's own original plasma membrane (the future inner membrane) — explaining the double membrane seen in both mitochondria and chloroplasts today. This detail is mechanistically important: it is the single most direct structural fossil of the engulfment event, distinguishing organelles built this way from single-membrane organelles like the peroxisome or lysosome, which bud from the endomembrane system rather than being swallowed whole.

Step 2: Establishment of metabolic interdependence. Instead of being degraded, the symbiont begins supplying the host with a useful product — ATP from oxidative phosphorylation in the case of the proto-mitochondrion, or sugars from photosynthesis in the case of the proto-chloroplast. The host, in turn, supplies the symbiont with protection, needed metabolites, and a stable environment. Natural selection favors host cells that keep the symbiont alive and functioning, and favors symbiont lineages that stop wasting energy on functions the host now provides (such as cell wall synthesis or independent motility), locking in mutual dependence.

Step 3: Genome reduction and endosymbiotic gene transfer (EGT). Over evolutionary time, DNA is repeatedly transferred from the organelle genome to the host nuclear genome. This happens through fragments of organelle DNA escaping (often during organelle lysis or via vesicle-mediated transfer) and becoming incorporated into the nucleus by non-homologous recombination. Once a gene's function is available from a nuclear-encoded, cytoplasmically-synthesized protein that can be re-imported into the organelle, the original organelle copy becomes redundant and is free to accumulate mutations and eventually be lost. This is why: ``` Free-living cyanobacterium genome: ~2,000–3,000 genes Modern chloroplast genome: ~100–250 genes Free-living alphaproteobacterium: ~1,000–2,000 genes Modern human mitochondrial genome: 37 genes (13 proteins, 22 tRNAs, 2 rRNAs) ```

Step 4: Protein import machinery evolves. Because most organelle proteins are now encoded in the nucleus, made on cytosolic ribosomes, and must be sent back into the organelle, the cell evolves dedicated import channels — the TOM/TIM complexes (translocase of the outer/inner mitochondrial membrane) for mitochondria and the TOC/TIC complexes (translocase of the outer/inner chloroplast membrane) for chloroplasts. These import systems recognize N-terminal targeting sequences on precursor proteins and thread them through the double membrane, often with the assistance of chaperones such as Hsp70 on both sides of the membrane.

Step 5: Genome and ribosome retention (the "why keep any genes at all" problem). Neither organelle lost its genome entirely. The leading hypothesis, called CoRR (co-location of gene and gene product for redox regulation), proposes that genes for the core proteins of the electron transport chain and photosynthetic reaction centers must remain locally encoded so that their expression can respond rapidly to local redox conditions inside the organelle — a fast, direct feedback loop that would be lost if the gene were moved to the distant nucleus. This is why the retained organelle genes are overwhelmingly those encoding subunits of the respiratory chain complexes (mitochondria) or photosystems (chloroplasts), plus the rRNAs and tRNAs needed to translate them.

Step 6: Division synchronization. Mitochondria and chloroplasts cannot be made de novo — they arise only by growth and fission of pre-existing organelles, using a bacterial-derived division machine. In many eukaryotes this involves a dynamin-related GTPase on the cytosolic face (Drp1 for mitochondria) constricting the organelle, sometimes assisted by an internal ring built from FtsZ, a bacterial cell-division protein that chloroplasts and some mitochondria still retain — direct molecular continuity with bacterial binary fission.

The theory is unusually well corroborated because it makes many independent, falsifiable predictions, all of which have been tested and confirmed.

1. Membrane topology. Both mitochondria and chloroplasts are bounded by two membranes, unlike organelles of the endomembrane system (ER, Golgi, lysosomes, peroxisomes), which are bounded by one. Freeze-fracture electron microscopy and lipid analysis show the outer membrane resembles the host's plasma/endomembrane lipids and contains porins similar to those of Gram-negative bacterial outer membranes, while the inner membrane's lipid composition (notably cardiolipin in mitochondria, and galactolipids in chloroplasts) closely matches bacterial and cyanobacterial membranes respectively.

2. Organelle genomes are circular and bacterial in organization. Unlike the linear, histone-wrapped chromosomes of the eukaryotic nucleus, mitochondrial DNA (mtDNA) and chloroplast DNA (cpDNA) are typically circular (with some documented exceptions and complex forms), lack histones, and are organized in operon-like gene clusters — a hallmark of bacterial genome architecture. Human mtDNA is a 16,569 base-pair circle; chloroplast genomes range roughly from 120,000 to 220,000 base pairs.

3. Organelle ribosomes are bacterial-sized, not eukaryotic. Cytosolic eukaryotic ribosomes are 80S (large 60S + small 40S subunits). Mitochondrial and chloroplast ribosomes are 70S (large 50S + small 30S), matching bacterial ribosomes almost exactly in sedimentation coefficient, RNA sequence, and sensitivity to antibiotics. This is direct, mechanistic evidence, not just structural resemblance: drugs like streptomycin and chloramphenicol, which block bacterial 70S ribosomes, also block protein synthesis inside mitochondria and chloroplasts, while leaving cytosolic 80S ribosome function untouched. Conversely, cycloheximide, which blocks eukaryotic 80S ribosomes, does not stop organelle protein synthesis. This differential drug sensitivity was among the earliest biochemical proofs (1960s–70s) that organelles possess an independent, bacterial-type translation system.

4. Molecular phylogenetics places organelle genes squarely inside bacterial clades. When ribosomal RNA and protein-coding gene sequences from mitochondria are used to build phylogenetic trees, mitochondrial genes consistently nest within the alphaproteobacteria, close to modern Rickettsia-like intracellular bacteria — not as a sister lineage to the eukaryotic nuclear genome. Likewise, chloroplast genes nest within the cyanobacteria. This is the strongest and most quantitative line of evidence, because tree topology is a directly testable, statistically evaluable claim (bootstrap support values for the placement of mitochondria within alphaproteobacteria, and chloroplasts within cyanobacteria, are typically very high, often >95%, across independent gene sets).

5. Independent replication of division machinery. Mitochondria and chloroplasts divide by a fission process resembling bacterial binary fission, using FtsZ-ring constriction (retained from the bacterial ancestor in most lineages) combined with a host-derived dynamin-family GTPase acting from the cytosolic side. They are never assembled de novo from cytoplasmic components; a cell without mitochondria cannot spontaneously generate new ones, a fact confirmed by classic cell-fractionation and radioactive-labeling experiments showing that new organelle proteins and lipids are added only to pre-existing organelles.

6. Direct laboratory demonstration in amoebae. In a series of experiments beginning in the 1960s, Kwang Jeon infected a laboratory strain of Amoeba proteus with a bacterial pathogen; initially the infection was highly lethal, killing most of the population. Over about five years of continuous culture, the surviving amoebae and their bacterial infection evolved into a stable, obligate relationship: the amoebae now required the bacteria to survive (removing them killed the host), and the bacteria could no longer live free outside the amoeba. This experiment did not create a new organelle, but it demonstrated, in real time and in a controlled laboratory setting, exactly the kind of pathogen-to-permanent-symbiont transition endosymbiotic theory proposes for the deep past — turning a hypothesis about ancient, unobservable events into a mechanism verified to actually happen.

7. Modern genomic confirmation. Whole-genome sequencing has found that a substantial fraction of the nuclear genome in plants and other eukaryotes consists of genes clearly derived from the mitochondrial or chloroplast ancestor (endosymbiotic gene transfer), still bearing the phylogenetic signature of their bacterial origin even though they are now transcribed from the nucleus and their protein products re-imported into the organelle. In Arabidopsis thaliana, an estimated 18% of nuclear genes appear to derive from the cyanobacterial ancestor of the chloroplast.

Example 1 — reading a phylogenetic tree to test the theory. Suppose a researcher sequences the small-subunit ribosomal RNA (16S-like) gene from the mitochondrion of a fungus, a plant, and a human, and builds a maximum-likelihood tree together with 16S rRNA sequences from ten free-living bacterial genera spanning alphaproteobacteria, gammaproteobacteria, and cyanobacteria, plus 18S rRNA from the corresponding nuclear genomes. Prediction of endosymbiotic theory: the three mitochondrial sequences should cluster together, and that cluster should nest inside the alphaproteobacterial branch (closer to Rickettsia and Agrobacterium than to E. coli, a gammaproteobacterium), with strong bootstrap support. The nuclear 18S sequences should form a completely separate eukaryotic clade, unrelated in branching position to any bacterial group. Result actually obtained in real studies: this is exactly the topology found. Mitochondrial genes fall within alphaproteobacteria with bootstrap values typically 95–100%, while nuclear rRNA genes form the deeply diverging eukaryotic branch. A rival hypothesis — that mitochondria arose by gradual internal specialization of an ancestral eukaryotic membrane system, with no bacterial ancestor — predicts mitochondrial genes should cluster with the nuclear eukaryotic genes, or at least not nest specifically within one bacterial phylum. That prediction is falsified by the observed tree. The nesting inside a specific bacterial clade, rather than mere "resemblance to bacteria in general," is the discriminating test, and it is why phylogenetics — not just morphology — was needed to settle the debate definitively in the 1970s–1980s.

Example 2 — a quantitative reasoning problem on genome reduction and gene transfer. A free-living alphaproteobacterium ancestor is estimated to have carried about 1,500 protein-coding genes. The modern human mitochondrial genome encodes only 13 proteins. Roughly 1,000 to 1,500 nuclear genes in the human genome encode proteins that function inside mitochondria (the "mitochondrial proteome"), most of them recognizably homologous to bacterial genes and imported back into the organelle after cytosolic synthesis. Question: does this mean the ancestral bacterium's genes were mostly lost, or mostly transferred? Reasoning: if genes were simply lost with no transfer, the modern mitochondrial proteome should be dramatically smaller than the ancestral bacterial gene set, because there would be no compensating nuclear source for those functions — and mitochondria would have to survive with a drastically reduced protein complement. But biochemically, mitochondria still perform essentially all of the core metabolic functions the ancestral bacterium had (TCA cycle, electron transport, membrane lipid synthesis, its own transcription/translation machinery), which requires on the order of hundreds to over a thousand proteins. Since only 13 are still made from mitochondrial DNA, the remaining ~1,000+ required proteins must be supplied from somewhere — and comparative genomics shows they are supplied by the nucleus, many still phylogenetically traceable to alphaproteobacterial origin. This tells us the dominant process was transfer, not simple loss: genes moved to the nucleus and their protein products evolved mitochondrial targeting sequences so they could be re-imported, rather than the functions themselves disappearing. This calculation — comparing the size of the retained organelle genome to the size of the organelle's actual functional proteome — is exactly the kind of quantitative argument that distinguishes "loss" from "relocation" and is a key piece of evidence for ongoing endosymbiotic gene transfer.

The idea that organelles might be domesticated bacteria was proposed piecemeal as early as the 1880s–1920s by botanists such as Andreas Schimper and Konstantin Mereschkowski, who noted that chloroplasts divide independently and resemble cyanobacteria, and by Ivan Wallin, who suggested mitochondria were bacterial symbionts. But the idea was largely dismissed for decades as unsupported speculation, lacking any molecular mechanism or rigorous evidence — it was not taken seriously by mainstream cell biology. It was Lynn Margulis who transformed this scattered, disreputable idea into a rigorously argued, evidence-based theory. In her landmark 1967 paper, published (under her earlier name, Lynn Sagan) as "On the Origin of Mitosis in Eukaryotic Cells" in the Journal of Theoretical Biology, she synthesized cytological, biochemical, and paleontological evidence into a detailed, falsifiable model proposing that mitochondria, plastids, and (more speculatively and less successfully) undulipodia (cilia/flagella) all arose from separate bacterial symbioses. The paper was reportedly rejected by roughly fifteen journals before publication, and her broader ideas met heavy skepticism from a scientific establishment that viewed organelle origins as a matter of gradual internal cellular specialization, not bacterial ancestry. Margulis persisted, publishing the fuller argument in her 1970 book Origin of Eukaryotic Cells, continuing to refine the theory as molecular sequencing data began to arrive in the late 1970s and 1980s and provided the decisive phylogenetic confirmation her original cytological arguments alone could not.

Mitochondrial disease genetics. Because mitochondria retain their own small genome, mutations in mtDNA cause a distinct class of maternally inherited diseases (e.g., Leber's hereditary optic neuropathy, MELAS syndrome) that follow non-Mendelian inheritance patterns, a direct clinical consequence of the organelle's bacterial ancestry and cytoplasmic (not nuclear) genome. Understanding this ancestry is essential for genetic counseling, since mtDNA is inherited almost exclusively through the maternal lineage in most animals.

Mitochondrial replacement therapy ("three-parent babies"). Clinics now transfer a prospective mother's nuclear DNA into a donor egg with healthy mitochondria, preventing transmission of severe mtDNA disease — a technique only conceivable because mitochondrial and nuclear genomes are separable, heritable units, exactly as endosymbiotic theory predicts.

Antibiotic side effects and drug design. Because mitochondrial ribosomes remain bacterial-like (70S), certain antibiotics that target bacterial ribosomes (e.g., aminoglycosides, chloramphenicol) can also impair mitochondrial protein synthesis in patients, contributing to known side effects such as antibiotic-induced hearing loss in individuals with susceptible mtDNA variants — a direct clinical echo of the same 70S-ribosome evidence that helped establish the theory.

Forensic and population genetics. Mitochondrial DNA's maternal, non-recombining inheritance makes it a standard tool for tracing maternal lineages, human migration history ("mitochondrial Eve" studies), and species identification in conservation forensics.

Agricultural and bioenergy engineering. Efforts to improve crop photosynthetic efficiency, or to engineer algae for biofuel production, directly manipulate chloroplast genomes and their import machinery — chloroplast transformation is now a standard biotechnology platform precisely because the organelle retains an independently transformable, bacterial-style genome.

Evolutionary cell biology and the study of the cytoskeleton. The mitochondrion's dependence on host-derived cytoskeletal transport (mitochondria are moved along microtubules by motor proteins, and their division depends on cytosolic dynamin-family constriction) foreshadows how deeply organelle function is now intertwined with the host cytoskeletal system — a partnership layered on top of, but never replacing, the organelle's ancient bacterial identity.

Ask the tutor: How did the host cell in the first endosymbiotic event — the one that produced mitochondria — avoid simply digesting the engulfed bacterium, and what do we actually know about what that host cell was?