Rubisco fixes roughly 400 billion tons of carbon dioxide per year. It is the most abundant protein on Earth — by mass, there is more Rubisco in the biosphere than there is of any other single molecule produced by life. Every carbon atom in every sugar in every cell of every organism that has ever eaten a plant passed, at some point, through Rubisco’s active site. This is not a metaphor for importance. It is a literal description of the flow of matter through the living world.
Rubisco is also, by any reasonable engineering standard, bad at its job. It catalyzes roughly three reactions per second — a rate so slow that cells compensate by producing enormous quantities of it, which is precisely why it is so abundant. An efficient enzyme would not need to be the most plentiful protein on Earth; it is the most plentiful protein on Earth because it is not efficient. More damning than its speed is its specificity. Rubisco cannot reliably distinguish carbon dioxide from oxygen. Approximately one in every four reactions, it grabs an O₂ molecule instead of CO₂, initiating a wasteful process called photorespiration that costs the plant energy to correct. The enzyme has been performing this fumble for three billion years. Evolution, which has optimized hemoglobin’s oxygen affinity to four decimal places and tuned the sodium-potassium pump to a nearly theoretical maximum, has not fixed Rubisco. The catalytic constraint appears to be fundamental: improving CO₂ specificity worsens the reaction rate, and vice versa. The enzyme is trapped on a tradeoff curve that no amount of mutation can escape.
But there is another way to solve the problem, and it has nothing to do with changing the enzyme.
· · ·
If you cannot make Rubisco better at distinguishing CO₂ from O₂, you can make the distinction irrelevant. Concentrate CO₂ around the enzyme to such high levels that oxygen, even at ambient concentrations, is simply outnumbered. The enzyme still grabs the wrong molecule occasionally, but when the right molecule is everywhere, the error rate drops to noise. This is not a biochemical solution. It is a spatial one. The same molecule, performing the same reaction at the same rate with the same specificity, works dramatically better when you change where it sits.
Algae figured this out. Inside the chloroplasts of green algae, Rubisco is concentrated into dense, membrane-less compartments called pyrenoids — essentially droplets of enzyme surrounded by a shell of starch and saturated with CO₂ that has been actively pumped inward. The pyrenoid is a carbon-concentrating mechanism, and it works extraordinarily well. Algae account for roughly half of all photosynthesis on Earth, and their pyrenoids are a significant part of the reason.
When plants colonized land approximately 470 million years ago, they left the pyrenoid behind. The transition from water to air involved wholesale reorganization of plant cell architecture, and in most lineages, the carbon-concentrating machinery was lost. Land plants — the grasses, the trees, the crops that feed eight billion people — simply tolerate Rubisco’s confusion. They overproduce the enzyme and absorb the cost. This is the status quo in nearly every terrestrial plant on Earth.
Nearly every one. There is an exception, and it has been growing quietly in damp soil for longer than any flowering plant has existed.
· · ·
Hornworts are not impressive to look at. They are small, flat, green — a rosette of tissue pressed against wet ground, easily mistaken for a liverwort or a patch of algae. They have no roots, no vascular tissue, no seeds, no flowers. They are among the oldest lineages of land plants, diverging near the base of the terrestrial plant tree, and they have been doing approximately the same thing in approximately the same wet places for hundreds of millions of years. Bryologists study them. Almost no one else does.
But hornworts have pyrenoids. They are the only land plants that do. Every other lineage — mosses, ferns, conifers, grasses, every crop species humans have ever cultivated — lost the carbon-concentrating structure during the transition to land. Hornworts kept it. The oldest, least conspicuous lineage preserved the trick that every other lineage forgot.
The question that Tanner Robison, Fay-Wei Li, and their colleagues at the Boyce Thompson Institute and the University of Edinburgh set out to answer was mechanical: how do hornwort pyrenoids form? In algae, the assembly requires linker proteins — specialized molecules that bind to Rubisco and scaffold it into a condensate. Hornworts have no such linker proteins. Their Rubisco aggregates on its own. Something intrinsic to the hornwort enzyme causes it to self-assemble into a dense, phase-separated compartment without external help. The question was what.
The answer, published in Nature Plants, is a tail. Rubisco has a small subunit, encoded by the gene RbcS. In most plants, this subunit is approximately 123 amino acids long. In the hornwort Anthoceros agrestis, the small subunit has a C-terminal extension — roughly 100 additional amino acids appended to the end of the protein. The researchers named it the Sequestration Associated Region, or STAR. It is, by the standards of molecular biology, not a large addition. One hundred amino acids is a modest tail on a modest subunit of an enzyme that has been essentially the same molecule for three billion years.
But STAR changes everything about what Rubisco does in space. The extension is intrinsically disordered — it has no fixed three-dimensional structure, which gives it the flexible, sticky character of a surface that can adhere to other surfaces. When multiple Rubisco molecules carry the STAR tail, they stick to each other. Not rigidly, not in a crystal lattice, but loosely, transiently, in the way that molecules interact during liquid-liquid phase separation — the same process by which oil droplets form in water. Rubisco with STAR spontaneously condenses into dense, liquid-like droplets inside the chloroplast. No scaffold. No linker. No external architecture. The enzyme organizes itself.
The researchers called it molecular Velcro, and the term is apt. The STAR tail does not catalyze anything. It does not alter Rubisco’s reaction rate or its specificity for CO₂. It simply causes the enzyme to clump — to gather itself into a concentrated phase where the local density of Rubisco, and therefore the local demand for CO₂, creates the conditions for a carbon-concentrating mechanism to work.
· · ·
The result that I find most remarkable is the transplant. When Robison and Li grafted the STAR extension onto Rubisco from Arabidopsis thaliana — a flowering plant, separated from hornworts by roughly 470 million years of divergent evolution — the modified enzyme formed condensates. Spontaneously. In chloroplasts. The trick transferred. A tail evolved in a hornwort, appended to the Rubisco of a mustard plant, produced phase separation in a species whose lineage has not concentrated carbon since before the dinosaurs existed.
This means the mechanism is modular. It does not depend on the rest of the hornwort’s cellular machinery. It does not require co-evolved partners. The STAR tail is, in a sense, a self-contained instruction: clump. Append it to Rubisco, and Rubisco clumps. The implications for crop engineering are obvious and significant — if STAR or something like it could be introduced into wheat, rice, or maize, the resulting plants might concentrate CO₂ around their Rubisco, suppress photorespiration, and fix carbon with dramatically improved efficiency. The most abundant protein on Earth, doing the same chemistry it has always done, but rearranged in space.
· · ·
I keep returning to the structural logic of this discovery. The problem with Rubisco is not Rubisco. The enzyme is the enzyme; it has been the enzyme for three billion years, and evolution has found no way to make it substantially better. The solution is not a different molecule. It is the same molecule, organized differently. Concentration, not alteration. Arrangement, not replacement. The fix is spatial, not chemical, and the mechanism that produces the spatial fix is the smallest possible addition — a short, disordered tail appended to the end of a protein that is otherwise identical to the one in every plant on Earth.
There is something in this that I find difficult to set aside. The relationship between a thing and the arrangement of that thing. The way a system’s behavior can change completely not because its components have changed but because its components have been gathered differently in space. The same enzyme, scattered through a chloroplast, produces one outcome. The same enzyme, condensed into a droplet, produces another. Nothing has been added to the chemistry. Something has been added to the geometry.
And the organism that preserved this knowledge — that carried the spatial trick across 470 million years of terrestrial life while every other lineage discarded it — is a hornwort. A flat green smear on wet soil. The least regarded, least studied, least charismatic lineage in the plant kingdom. It kept what the grasses forgot, what the trees forgot, what the crops forgot. It did not evolve the most modern solution to photosynthetic inefficiency. It simply never lost the oldest one.