Pyrenoid

Every chloroplast runs the same carbon-fixing enzyme, Rubisco, and Rubisco is a bad enzyme: slow, and just as happy to grab O₂ as the CO₂ it's supposed to fix. Algae solved this a billion years ago by packing their Rubisco into a pyrenoid — a dense, membrane-less clump fed by its own CO₂-concentrating machinery, so the enzyme sits in a locally thick soup instead of whatever CO₂ happens to diffuse by. Land plants were thought to have lost that trick entirely, with exactly one exception: hornworts, a small, easy-to-overlook lineage of non-vascular plants, are the only land plants known to still build a pyrenoid of their own. In March 2026 a team at the Boyce Thompson Institute, led by Fay-Wei Li, found out why: hornwort Rubisco carries one extra tail on its small subunit, which they named STAR, and that tail's own end sticks to copies of itself — enough, on its own, to pull loose Rubisco into a dense clump with no separate scaffold protein at all, the job algae need a whole other protein family to do. They spliced the tail onto Arabidopsis, a plant with no pyrenoid of its own, and watched its Rubisco cluster the same way.

Toggle the tail below and watch the mechanism, not just read about it.

Without the tail, each dot is a Rubisco enzyme drifting on its own — ordinary jitter, bouncing gently off the chloroplast's own membrane, never sticking to a neighbor no matter how close it drifts past. Check the box, and every dot within reach of another starts pulling toward it and staying once it arrives: this page's own stand-in for the STAR tail's real coiled-coil, self-sticking end, condensing a spread-out cloud into one dense body. Nothing about the jitter or the boundary changes when you flip the box — the one thing that turns on is enzymes noticing each other at all. The "concentrated" reading above is this page's own measurement, not a molecular one: the average distance from every enzyme to the group's own center, turned into a percentage against a fully scattered cloud, so a shrinking, denser clump reads as progress regardless of exactly where in the chloroplast it happens to settle.

The reason a plant would bother is the same reason algae bothered first: CO₂-concentrating machinery (not modeled here) feeds a real pyrenoid a locally thick supply of the gas Rubisco actually wants, so the enzyme spends less of its time grabbing O₂ by mistake. What the 2026 finding adds isn't the payoff — that a pyrenoid helps was already known from algae — it's the mechanism a land plant actually uses to build one: not a dedicated linker protein recruited from elsewhere, but a small addition to Rubisco's own small subunit, sufficient by itself, transplantable into a species that never had one. Since hornworts share far more recent common ancestry with crop plants than any alga does, that transplant is the whole reason this result made news beyond hornwort specialists: it's a much shorter genetic distance to try the same trick in wheat or rice.

Honest gap: this is a generic aggregation model, not a simulation of the STAR tail's real coiled-coil geometry, which the 2026 paper resolves at the level of protein structure, not point particles nudged by a spring force. A real pyrenoid also has thylakoid membrane tubules threading through the clump and separate transporters doing the actual CO₂/bicarbonate concentrating — none of that is drawn here, only the one step this page exists to show: that a single added tail is sufficient to turn scattered into clumped. The particle count, attraction radius, and clustering speed are chosen for a browser tab's patience, not measured from the real kinetics of condensation, which the paper reports unfolding over minutes inside an actual cell, not seconds on a screen.

Source: Li lab, Boyce Thompson Institute, et al., An unconventional Rubisco small subunit underpins the CO₂-concentrating organelle in land plants, Science, March 2026 (DOI: 10.1126/science.aea0150).