Global pressure on agriculture is set to increase due to a number of reasons, including population growth and climate change. Climate change induced increases in both abiotic and biotic crop stressors (heat, drought, flooding, as well as pathogen and herbivore spread) have already resulted in regional crop failures and yield loss across the globe. Prevention and mitigation can be accomplished by improving the resilience of crops, through targeted breeding programmes, increased plant protection practices, and utilising the genomic potential of crop wild relatives. To effectively apply crop improvement measures, a holistic and deep understanding of the plant response to stress interactions is required. This stress response triggers a complex signalling network, involving a multitude of transcriptional regulators, massive gene activity changes and phytohormone synthesis and signalling. Ideally, the response results in a variety of defensive processes and compounds, allowing the plant to overcome or tolerate the stress.
Basic research in model species has led to improved molecular understanding of signalling and regulation mechanisms in plants. Many resources are available, including molecular interaction data (STRING, MINT, IntAct), metabolic models (KEGG, AraCyc), and large knowledge graphs (CKN, KnetMiner). However, these resources are all either missing detailed mechanistic information, transcription factors and other gene regulatory components, or the context of plant stress. At the National Institute of Biology, information from a number of resources and publications has been compiled into a single large plant stress signalling (PSS) network. PSS is a bottom-up manually curated and highly detailed resource of plant stress related molecular interactions, including stress perception and defensive processes. The majority of information in PSS originated from the model species Arabidopsis thaliana.
A similar level of insight is missing in most crop species. This can be improved by inferring the function of crop genes from better understood and annotated model species. Application of “model-to-crop” translations have many hurdles, but show promise in leading to the development of improved crops, including brassicas, legumes, maize, and others. Such inference is most often done by identifying orthologous genes through primary sequence similarity. Unfortunately, due to diversification in evolution and gene duplication events, phylogenetically orthologous genes between model and crop species do not always perform exactly the same function, especially in response to stress. For actionable translations, we need to instead identify “functional orthologs” that may have only moderate sequence similarity, but shared molecular function.
Orthologs are commonly identified through sequence similarity approaches, which alone are not enough to predict functional orthology. The first objective of this proposal is to generate a high quality, multi-factor approach to translate genes involved in plant stress signalling from model to crop species. The approach will include elements such as functional annotations, protein domains, gene regulatory regions, protein structure, as well as interactions with upstream regulators or downstream interactors to identify true functional orthologs. This will require the utilisation of existing datasets, including PLAZA, and the use of algorithms and tools, such as AlphaFold, FoldSeek, collinearity detention, GO and Interpro annotations, and topological alignment of networks produced from high throughput multi-omic data. A novel, multi-graph approach will integrate the various levels of information. The resultant orthology translations will both improve our current knowledge in stress signalling by integrating knowledge from multiple species to fill knowledge gaps, and translate current available knowledge to crops to better understand the stress response. A visualisation tool will be developed to enable the exploitation of this resource by plant researchers. A comparative analysis of the distribution of genes within the plant stress signalling network across the multiple species will determine the core signalling clusters that are shared between the species and unique potential within subsets of species. Analysis will also allow putative annotation of genes with unknown function and inference of pathways across species.
Even with quality model-to-crop orthology translations, we can expect differences between species in the dynamic responses of genes to stress. Therefore, the second objective of this proposal is to quantify and investigate the dynamical responses of the translated stress networks in an important set of example plants: the model species Arabidopsis, a well studied crop tomato, and an important, but less studied, crop potato. This will be done using Boolean networks and semi-quantitative modelling, integrated with time series transcriptomic data. Multiple models will be parameterised and used to compare response dynamics in a number of stress related conditions. This will also enable quantification of the translations by ensuring the models fit observed data. Exploiting the fitted models, perturbations will be done under multiple stress conditions. This will enable the identification of important components of plant stress resilience, since they will have larger impacts on the plant growth and defence outputs of the model. These identified componentes are hypothesised targets for crop resilience improvements. This task will provide understanding of how responses differ between species under multiple stress conditions, and of special interest, the differences in responses between weeds and crops. Experimental validation of the predicted target will provide tools for further use in improvement of crop resilience to environmental stress.
The approaches taken in this proposal are novel in their integrative and extensible concepts, utilising a variety of prior knowledge resources to generate new hypotheses. The methodologies will enable a breakthrough in understanding of how plants perform in diverse environments, through development of new multifaceted functional orthology detection, modelling approaches and novel target prediction, providing opportunity for faster breeding of more tolerant plants. Additionally, each of the outcomes of the project will have value for all researchers in plant stress science and are also designed to be continually useful community resources. The research proposed in this project will thus be of great importance for expanding knowledge in comparative genomics, plant stress responses and crop stress resilience improvements.
