This post is a reflection on and summary of my recent work with collaborators: Flamholz AI, Prywes N, Moran U, Davidi D, Bar-On YM, Oltrogge LM, Alves R, Savage D, Milo R. 2019. Revisiting Trade-offs between Rubisco Kinetic Parameters. Biochemistry 58:3365–3376. doi:10.1021/acs.biochem.9b00237

All plants, algae and cyanobacteria rely on the Calvin-Benson-Bassham (CBB) cycle for growth. Rubisco is the central enzyme of the CBB cycle and the most abundant enzyme on the planet: it does the tricky bit where CO2 gets “fixed” onto a soluble sugar. While it is often said that rubisco is “slow,” it is actually an average enzyme in terms of “turnover number” (maximum rate per active site, kcat). The true rate of rubisco carboxylation is much slower than its kcat, however, because rubisco can react non-specifically with O2. I certainly think it’s surprising that rubisco is not faster or more specific given how important and abundant it is.

RuBisCO illustration by David Goodsell

What’s the deal with rubisco? Why isn’t it faster? Why isn’t it more CO2-specific? People often argue that Rubisco can’t be both fast and specific at the same time. Correlations between rubisco kinetic parameters are taken to support this argument since they show that faster Rubiscos (higher kcat) are less CO2-specific. But these arguments were based on a very small dataset (about 20 rubiscos). We collected a much larger dataset of about 300 rubiscos and found that the correlations are now substantially weaker.

Our main takeaway from this dataset is that rubisco kinetics display surprisingly little variation. Turnover numbers for most enzymes vary over 5-10 fold but less than twofold for rubisco. Rubisco CO2-specificity varies even less - about 30%. Narrow variation suggests tight chemical constraints on rubisco evolution. We offer a coarse-grained model of rubisco catalysis that could explain the constraints imposed on the Rubisco evolution. I think this work should cause us to re-evaluate prospects for Rubisco engineering, since a tightly-constrained enzyme is probably hard to engineer. We put a fair bit of effort into suggesting experiments and engineering strategies that could test our model, so I do hope that someone will try to prove us wrong.

This work started as a journal club of Tcherkez et al. PNAS 2006 and Savir et al. PNAS 2010 that Dave Savage suggested when we visited Ron Milo’s lab in January 2018. Noam Prywes and I scrounged together data for the JC and Uri Moran did a lot of the later heavy lifting, combing the literature for as much Rubisco data as we could find.