---
id: "2026-10-09-ml-assisted-directed-evolution-plant-rubisco"
url: "https://postcutoff.com/e/2026-10-09-ml-assisted-directed-evolution-plant-rubisco/"
as_of: "2026-10-12T07:22:00+02:00"
date: "2026-10-09"
date_precision: day
category: science
importance: 2
confidence: high
status: [Result confirmed]
verification: Peer-reviewed
sources: 1
editor: Adam Bicz
human_review: null
version: null
---

As of: 2026-10-12 07:22 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code). Human editor: Adam Bicz.
Canonical page: https://postcutoff.com/e/2026-10-09-ml-assisted-directed-evolution-plant-rubisco/

# Machine-learning-guided evolution makes a plant Rubisco 29% faster (MIT, Stanford, ANU

Full title: Machine-learning-guided evolution makes a plant Rubisco 29% faster (MIT, Stanford, ANU; Science Advances)

A paper in Science Advances (vol. 12, issue 41, published 9 Oct 2026) used the structure-informed protein model ESM-IF1 to choose which positions of tobacco Rubisco's large subunit to mutate, then selected variants in Rubisco-dependent E. coli. The best single change, T391I, raised the carboxylation rate by 29% and aerobic carboxylation efficiency by 43%. Several improved variants carry amino acids not seen in plant evolution. Rubisco is the enzyme that fixes CO₂ in photosynthesis and has long resisted engineering.

## Key facts
- Paper: 'Machine learning–assisted directed evolution of plant Rubisco', Science Advances 12(41), doi 10.1126/sciadv.aeh3246; authors Julie L. McDonald, Jiacheng Lin, Yunlong Zhao, Brian L. Hie, Rosemary Birch, Mary Gehring, Bryan D. Bryson, Spencer M. Whitney, Matthew D. Shoulders, Robert H. Wilson (MIT; Stanford; Australian National University)
- Method (abstract): 'We applied the structure-informed machine learning (ML) model ESM-IF1 to identify plausible amino acid sites in the large subunit of Nicotiana tabacum Rubisco to target for directed evolution', followed by selection in Rubisco-dependent Escherichia coli
- Result (abstract): 'the T391I substitution improved carboxylation rate by 29% and aerobic carboxylation efficiency by 43%'
- Abstract: several improved variants 'carried amino acid changes not found in the evolutionary lineage of plants, despite being assembly competent in plant chloroplasts'
- Limits: the abstract reports enzyme kinetics, not crop yield; it speaks of 'potential for enhancing crop productivity'. Written from the abstract (Crossref); the full text was not read

## What happened

Rubisco catalyses the step of photosynthesis that captures CO₂. It is slow and inefficient and, in the words of the abstract, "has proven challenging to engineer".
The MIT-led team used ESM-IF1, an inverse-folding protein model, to pick positions in the tobacco enzyme
where changes were plausible, built libraries at those positions and let bacteria that depend on Rubisco select the better
variants. Some winning mutations do not occur in any plant lineage.

## Why it matters

"A better Rubisco" is on the list of grand challenges for biology (see the related entry). This is a peer-reviewed, lab-validated
step on a plant enzyme, using a small specialised model as a guide for experiments. It is an enzyme-level gain, not yet a
demonstrated gain in crop yield.

## Your AI and this story

- GPT-6 Astra (training cutoff April 2026): 162 days after its cutoff
- Claude Opus 5.5 (training cutoff June 2026): 101 days after its cutoff
- Gemini 3.8 Flash (training cutoff March 2026): 192 days after its cutoff
- Grok 4.7 (training cutoff May 2026): 131 days after its cutoff

## Sources

1. [Science Advances: Machine learning–assisted directed evolution of plant Rubisco](https://www.science.org/doi/10.1126/sciadv.aeh3246) (science.org, paper)

## Changes

- 2026-10-12 (filed): Created from the Science Advances table of contents and the Crossref abstract

## Related

- 2026-10-05: [Google DeepMind's AlphaProtein Novo designs new-to-nature enzymes from scratch (piperidine synthesis, DEHP degradation), with Frances Arnold's lab](https://postcutoff.com/e/2026-10-05-deepmind-alphaprotein-novo-enzyme-design/index.md)
- 2026-09-19: [FutureHouse and Edison Scientific publish twelve 'Millennium Problems for Biology', pitched as the 'last reasonable eval' for AI in biology](https://postcutoff.com/e/2026-09-19-millennium-problems-for-biology/index.md)