A hornwort's modified enzyme subunit could teach crop plants to concentrate carbon dioxide
Science, March 2026, AAAS
Photosynthesis feeds the world, but its central enzyme has a well-known flaw. Rubisco (ribulose-1,5-bisphosphate carboxylase/oxygenase) is responsible for capturing carbon dioxide from the air, yet it also reacts with oxygen, triggering energy-wasting photorespiration. Many algae solved this problem long ago by concentrating CO2 around Rubisco inside specialized microcompartments called pyrenoids. Crop plants never evolved this adaptation.
A new study reveals that hornworts, the only land plants known to possess pyrenoid-like structures, achieved their version of Rubisco clustering through a mechanism entirely different from what algae use. And that mechanism appears to be transferable to other plants.
An extension embedded in the enzyme itself
In algae, Rubisco clustering is typically driven by specialized linker proteins that evolved independently across lineages. These linkers bind Rubisco externally, pulling enzyme molecules together into dense aggregates. But the algal linker proteins are species-specific and incompatible with plant Rubisco, which has stalled efforts to engineer pyrenoids into crops.
Tanner Robison and colleagues report a different solution found in the hornwort Anthoceros agrestis. Instead of relying on a separate linker protein, this species carries a distinctive variant of the Rubisco small subunit (RbcS) with an approximately 100-amino-acid extension at its C-terminal, which the researchers named the Sequestration Associated Region (STAR).
The STAR domain embeds the condensation mechanism directly into the enzyme. Rather than needing an external linker to pull Rubisco molecules together, the enzyme carries its own molecular adhesive.
Successful transfer to Arabidopsis
The critical question was whether STAR could function outside its native hornwort. The researchers tested this by attaching the STAR domain to native Arabidopsis RbcS. The result: Rubisco condensates formed within the chloroplasts of the modified plants.
This is significant because previous algal linker proteins could not achieve this in land plants. The STAR approach bypasses the compatibility barrier by modifying the enzyme itself rather than introducing foreign linker machinery.
Biochemical and structural analyses revealed the molecular interactions driving condensate assembly. The STAR domain causes Rubisco molecules to stick together in a self-assembling process, forming dense compartments analogous to the pyrenoids found in algae and hornworts.
A step toward engineering more efficient crops
The potential agricultural implications are substantial. If Rubisco can be concentrated into compartments within crop plant cells, and if those compartments can be paired with a CO2 delivery system, photosynthetic efficiency could increase meaningfully. Even modest improvements in photosynthetic efficiency translate to significant gains in crop yield.
But the clustering alone is insufficient. A functional carbon-concentrating mechanism requires not just a compartment to house Rubisco but also molecular machinery to actively pump CO2 into that compartment. The STAR discovery addresses the first challenge. The second remains an active area of research.
Limitations and next steps
The study demonstrates Rubisco condensate formation in Arabidopsis, a model plant used in laboratory research, not in actual crop species. Whether the same approach works in wheat, rice, or maize has not been tested. Additionally, condensate formation was confirmed structurally but the functional impact on photosynthetic efficiency in the modified Arabidopsis plants has not yet been fully characterized.
Engineering a complete carbon-concentrating mechanism in crop plants will require integrating multiple components. STAR provides one critical piece, a modular, transferable tool for Rubisco clustering, but a functioning system demands more.
In a related Perspective article, Moritz Meyer and Howard Griffiths discussed the study's implications for the field.