Two Organelles, One Bargain: Mitochondria, Chloroplasts, and the Genome Each One Kept

Two arrivals, not one

The mitochondrion descends from a free-living bacterium. So does the chloroplast. Neither statement is speculative any longer: both organelles carry their own DNA, replicate on their own schedule, are bounded by a double membrane, build proteins on ribosomes of the bacterial rather than the cytosolic type, and sit — by gene sequence — inside recognisable bacterial families. The mitochondrion's relatives are alphaproteobacteria. The chloroplast's are cyanobacteria, the lineage that invented oxygenic photosynthesis and remade the atmosphere this volume has already spent chapters inside.

The argument was made most forcefully by Lynn Margulis, publishing in 1967 as Lynn Sagan under the title “On the Origin of Mitosing Cells,” reviving and hardening proposals that Mereschkowsky and others had floated decades earlier and that the field had let lapse. The evidence that settled it was molecular and came later. What is worth taking from her and what is worth refusing are both specific: the endosymbiotic account of the organelles is taken, and taken completely, while the wider claim she pressed afterwards — that symbiosis rather than accumulated mutation is the principal engine of evolutionary novelty — is refused as overreach. One correct large idea does not license the next one.

The two arrivals are not symmetrical, and the asymmetry matters. On present evidence the mitochondrion appears to have been acquired once: every eukaryote examined either has one or has a reduced derivative of one — a hydrogenosome, a mitosome — and the single reported exception, the oxymonad *Monocercomonoides*, described by Karnkowska and colleagues in 2016 as lacking any mitochondrion-related organelle, is read as a secondary loss rather than a lineage that never had one. Plastids are the opposite case. They have been taken up repeatedly, including by cells that ate other cells which already carried one, so that some algal plastids sit behind three or four membranes and retain, in one lineage, the vestigial nucleus of the eaten alga. Once, against many times. The chapter records that difference before drawing the pairing tight.

What each one kept

Human mitochondrial DNA is a circle of 16,569 base pairs carrying thirty-seven genes: thirteen proteins, twenty-two transfer RNAs and two ribosomal RNAs. That is the whole local library. It is transcribed inside the organelle, into the organelle's own messenger RNA, which is then translated on the organelle's own ribosomes — and translated under a genetic code that is not quite the standard one, since in mammalian mitochondria the codon that reads as a stop in the nucleus reads as tryptophan here, and two that read as arginine outside read as stops inside. A dialect, kept behind a membrane.

The chloroplast keeps a larger library. A typical land-plant plastid genome runs to something like a hundred and twenty to a hundred and sixty thousand base pairs and a hundred-odd to a hundred and thirty genes — the core photosystem subunits, the large subunit of the carbon-fixing enzyme, a set of transfer RNAs, the plastid's own ribosomal RNAs and a bacterial-type RNA polymerase to transcribe them. Both organelles therefore do, at small scale, the full sequence the nucleus does at large: transcribe, process, translate. And both organelle ribosomes remain sensitive to antibiotics that act on bacterial ribosomes and not on cytosolic ones — chloramphenicol among them — which is descent showing up as a pharmacological fact rather than as a phylogenetic argument.

Set that against the size of the operation each organelle runs. A human mitochondrion depends on something in the order of eleven to fifteen hundred distinct proteins. It encodes thirteen. Everything else is transcribed in the nucleus, translated in the cytosol, and imported — through the translocase of the outer membrane and its inner-membrane partners, a dedicated machinery of recognition, threading and folding that exists because the genes moved out and the proteins still have to get in. The plastid's arithmetic is the same shape: a few hundred genes retained at most, some three thousand proteins required. The overwhelming majority of each former bacterium's genetic estate now sits in the host's nucleus. Some of it has left entirely, and some sits in the nucleus as recognisable transferred fragments still accumulating today.

The bargain

So the interesting question is not what was lost. Loss is the easy half: a gene inside an organelle that has been copied into the nucleus is redundant, and redundancy erodes. The interesting question is what was not lost. Why did thirteen genes stay behind a membrane when fourteen hundred moved out, and why — across lineages that diverged a billion years and more ago — is it so nearly the same thirteen? The retained set is not a random residue. In both organelles it is dominated by core subunits of the machinery that moves electrons and pumps protons, together with the local apparatus needed to express them.

The chapter reads that pattern as a bargain, and states plainly that the reading is its own. Almost everything about these organelles can be governed from a distance — built in the cytosol on nuclear instructions and delivered. What appears to resist government at a distance is the machinery that must be adjusted in response to its own instantaneous electrical and redox state: the parts of the apparatus whose correct expression level depends on what the membrane is doing right now. Those stayed where the reading is taken. This is the substance of John Allen's colocation-for-redox-regulation hypothesis, proposed in 1993 and developed since — genes are retained in the organelle so that their expression can be controlled directly by the redox state of the electron-transport chain they belong to, without the delay and indirection of a signal sent to the nucleus and a protein sent back.

It is a hypothesis, and it has a serious rival that requires no governance argument at all. The retained proteins are, almost without exception, extremely hydrophobic multi-pass membrane proteins. A protein that hydrophobic may simply be un-importable — mistargeted to the wrong membrane, or aggregated in transit — in which case its gene could not move even if moving were otherwise advantageous. That account explains the same retained set with a constraint rather than a strategy, and it is the more parsimonious of the two. The chapter does not adjudicate between them; it names both, and notes which discriminating experiments would separate them.

One word needs a fence around it. “Bargain” names a division of function that can be described without any party intending anything. No negotiation occurred. What occurred was that lineages in which particular genes left the organelle did worse, or could not be constructed at all, and lineages in which those genes stayed persisted. The volume's ban on ordination applies to its own metaphors: a retained division of labour is a survival of what carried, not a deal anybody struck.

The same machinery, run in opposite directions

Here the pairing earns its place in this part of the volume. Both organelles run chemiosmosis — Peter Mitchell's 1961 proposal, resisted for a decade and then decisive: protons are pumped across an inner membrane, the membrane holds the resulting store, and a rotary enzyme lets the protons back through and turns their return into ATP. Chapter 68 measured that store as a field and Chapter 72 placed the rotary synthase as the meter in the middle. Both organelles have it. Both use it. They point it in opposite directions.

The chloroplast is an accumulator. Light strips electrons from water at photosystem II, oxygen leaves as a by-product, the electrons are passed along and re-energised at photosystem I, and the energy is banked twice over: as a proton gradient that drives ATP synthesis, and as reducing power carried by NADPH. Both are then spent in the stroma on the reaction the whole apparatus exists to perform — the carbon-fixing enzyme, rubisco, pulling carbon dioxide out of the air and committing it to sugar. Capture, store, build.

The mitochondrion is a furnace with a meter on it. Carbon compounds arrive already built, are oxidised through the citric-acid cycle, and the electrons released are handed to the respiratory chain, which spends them pumping protons out across the inner membrane and finally passes them to oxygen, making water. The gradient is then spent through the synthase to pay for everything the cell does. Break down, release, spend. Thus the two halves of a cycle this volume has circled before: what the chloroplast assembles from light, the mitochondrion disassembles at living temperature, and the oxygen and water pass between them in opposite directions.

And even in the shared mechanism the two are not interchangeable, which is the point most worth keeping. The store is held differently. The mitochondrion holds the great majority of its proton-motive force as voltage — roughly a hundred and fifty to two hundred millivolts across four nanometres, with only a modest difference in acidity, because it pumps into a large buffered space. The chloroplast's thylakoid pumps into a small enclosed lumen that acidifies markedly, holding its store mostly as a difference in acidity of around three pH units with comparatively little voltage across the membrane. Same total currency, different denominations. A cell that inherited the mitochondrial arrangement cannot be described by the chloroplast's numbers, and any argument that treats the two gradients as one quantity has already lost the specific case.

Things in common, not the same thing

The volume's rule holds here as strictly as anywhere. Mitochondria and chloroplasts have things in common; they are not the same thing, and they are not the same thing twice. Their ancestors belong to different bacterial groups. Their acquisitions were separate events, separated by a long stretch of time. Their retained genes overlap in character but not in identity. Their internal architectures differ — cristae folded inward against stacked thylakoid discs — and their gradients, as just shown, are held in different coin. What they hold in common is a principle of energy handling inherited from a shared bacterial past, and a shared fate as former organisms that kept part of their own government.

Two of those commonalities have different sources, and conflating them would be the error. The chemiosmotic mechanism is genuinely inherited: both organelles have it because their bacterial ancestors had it, which is homology in the strict sense. The pattern of gene retention, by contrast, looks like the same pressure acting twice on two independent histories — convergent, not inherited — and that is the weaker and more interesting of the two facts, because a pressure that produces the same outcome twice is a pressure one can look for elsewhere.

The carrying test, run on the pairing, gives an honest split. Does the organelle's local genome carry its function? For the retained core subunits the answer is yes and it has been tested in the hardest way available: human mitochondrial diseases are, in large part, what it looks like when a handful of those thirty-seven genes fail. Remove or corrupt the local library and the apparatus does not stand, even with fourteen hundred nuclear-encoded parts intact and delivered. Does the governance reading carry? Not yet — because the measurement that would separate redox colocation from an import constraint has not been made decisive, and until it is, the bargain is a reading of a pattern rather than a mechanism established.

One last thing is worth saying without ornament. A reader's mitochondria are inherited from her mother and from no one else; the chloroplasts in the plant on the windowsill are, in most species, inherited the same one-sided way. Two lineages of former bacteria have been carried forward inside larger cells for something on the order of a billion and a half years, transcribing their own messages behind their own membranes the whole time, and nothing in the arrangement was designed. It was kept because it carried.