Designing better Rubisco for crops: Predicting effects of amino acid substitutions on Rubisco kinetics using machine learning

The rapid growth of human population commands to increase crop yields by 50 - 70 % by 2050 in order to feed the predicted 9 - 10 billion people. Extra food and biofuel production has to be achieved using the shrinking supply of arable land making it a key global challenge that requires ground-breaking innovations and ""thinking outside the box"". One of the innovative solutions for future prosperity of humankind is improved photosynthesis. All food production is based on photosynthesis either directly when growing crops or indirectly when plants used to feed livestock. However, despite being the most important biological process on the planet, photosynthesis is surprisingly inefficient with only 5 % of sun energy received by plants converted into biochemical energy of sugars.

One of the bottle necks of photosynthesis which curbs crop productivity, in particular under high light conditions, is the CO2 fixation catalysed by ribulose-1,5-bisphosphate carboxylase/oxygenase (aka Rubisco). Different species evolved Rubisco with different kinetics, and many crops don’t have the most efficient version of this enzyme. Recently we showed how crop performance under future climates might be improved if crop enzymes would be substituted by ones from other species (Sharwood et al., Nature Plants 2016). However, transplanting Rubisco between species is not a trivial task because of complex chaperone requirements as we demonstrated earlier (Whitney et al., PNAS 2015). Thus, improving crop Rubiscos by modifying existing crop enzymes with targeted amino acid substitutions could be a viable alternative.

However, time consuming nature of plant transformation and Rubisco kinetics assays curbs the progress in this field. The proposed project proposes to accelerate crop Rubisco engineering by using machine learning to find amino acid replacements that could act as catalytic switches in various crops and to predict values of Rubisco key kinetic characteristics in modified enzymes. This project will use existing data, in particular the wealth of Rubisco protein sequences and kinetic data, available for thousands and hundreds of species respectively. Machine learning will save time by providing customised solutions for particular crops obtained in silico. Testing a relatively small set of computationally obtained solutions will save resources compared to random mutagenesis, an approach used (unsuccessfully) to improve Rubisco until recently. Computationally designed enzymes will be recombinantly produced and their kinetics assessed using biochemical methods. In translational terms, the research outcome will provide essential knowledge for engineering better more productive crops.

The successful candidate will receive extensive training in computer science, bioinformatics as well as in molecular and synthetic biology and biochemistry as part of a collaborative multidisciplinary research group, and will have access to world-leading facilities in the School of Natural and Environmental Sciences at Newcastle University, and Department of Computer Science at the University of Liverpool.

For further information see the website: https://www.ncl.ac.uk/nes/

To apply
Please complete the online application form and attach a full CV and covering letter. Informal enquiries may be made to [Email Address Removed]