Photo: Jesper Cairo Westergaard
Microbiome Assisted Triticum Resilience In X-dimensions (The MATRIX)
MATRIX is a part of the Collaborative Crop Resilience Program (CCRP), together with the projects INTERACT and InRoot, all funded by the Novo Nordic Foundation. In the MATRIX, new microbiome-assisted approaches combined with deep-learning and modelling will be used to quantitatively and predictably improve crop resilience management strategies.
We will zoom in on the taxonomic diversity and functional potential of the wheat flag leaf microbiome. Which key microbes are involved in plant protection against biotic and abiotic stress? Combined with machine learning, this allows us to predict microbiome-related changes and their effect on crop resilience and productivity under climate change scenarios.
The following four Research Questions (RQ1-RQ4) will be addressed in the project:
RQ1: What is the microbiological and chemical diversity of the wheat phyllosphere (here encompassing the surface as well as the interior, the endosphere of the wheat flag leaf)?
RQ2: How can we best integrate multiple data types to develop a predictive model of microbiome-mediated wheat resilience to biotic and abiotic stress?
RQ3: What are the relationships between the leaf microbiota, phyllosphere chemistry, plant genetics, environmental factors, and stress resilience of wheat?
RQ4: How do we optimize use of the predictive model to engineer the wheat phyllosphere microbiome to improve resilience and productivity?
MATRIX will address six scientific Work Packages (WP1-WP6). The experimental Tasks within each WP will contribute to answer the Research Questions
WP1: Field logistics, environmental and phenotype monitoring
This WP is responsible for the design of field experiments at UCPH and NCSU in order to investigate how the flag leaf microbiome of wheat is influenced by cropping factors, microclimate and wheat genotypes.
WP Lead: Svend Christensen (UCPH); Co-lead: Gina Brown-Guedira (NCSU)
- Task 1.1 Establishment of field study infrastructure and field microbiome gradients.
- Task 1.2 Optimize logistics for targeted sampling.
- Task 1.3 Optimize logistics for basal environmental measurements.
- Task 1.4 Design field intervention experiments.
WP2: Microbiomics & hub taxa
This WP is responsible for sequencing the wheat flag leaf phyllosphere metagenome and metatranscriptome from samples obtained from the different fields in the US and DK (see WP1). The sequencing data will together with the environmental metadata create the backbone of the initial data input into the MATRIX predictive model.
WP Lead: Jos Raaijmakers (NIOO-KNAW); Co-lead: Lars Hestbjerg Hansen (UCPH)
- Task 2.1 Optimize protocols and pipelines.
- Task 2.2 Sequence the wheat phyllosphere metagenome and metatranscriptome.
- Task 2.3 Isolate, cultivate and identify phyllosphere microbes.
- Task 2.4 Isolate bacteriophages with potential to regulate hub taxa and pathogens.
- Task 2.5 Investigate microbe-microbe interactions and screening isolates for chemical signals.
This WP is responsible for providing the chemical profiling/characterization of the wheat flag leaf phyllosphere microbiome and in particular of hub microbial taxa that affect crop resilience. The targeted and untargeted analysis (metabolomics) performed in this WP covers finding carbohydrates, amino acids, organic acids as well as volatile organic compounds (VOCs) associated with the wheat phyllosphere microbiome.
WP Lead: Jan H. Christensen (UCPH); Co-lead: Jos Raaijmakers (NIOO-KNAW)
- Task 3.1 Build a chemical database for targeted screening and suspect screening analysis of phyllosphere metabolites.
- Task 3.2 Establish a pipeline for collection, handling and long-term storage of VOCs, semi-VOCs and non-VOCs in the phyllosphere.
- Task 3.3 Targeted and untargeted metabolomics of organic compounds and elements in field and greenhouse samples.
- Task 3.4 Metabolite profiling of isolated strains from flag leaves.
- Task 3.5 Tentative and full identification of key metabolites.
WP4: Effect of plant-microbiome interactions on resilience
This WP is responsible for empirically testing, under controlled settings in greenhouse experiments, how the wheat phyllosphere microbiome and its members affect wheat growth, physiology, and yield under stress. Putative mechanisms underlying microbiome effects, as well as interactions within microbial consortia, will be identified by examining changes in gene expression and secondary metabolites. The WP serves as validation of the results obtained from field microbiomes (WP1-3), informatics analysis (WP5) and modelling (WP6).
WP Lead: Christine Hawkes (NCSU); Co-lead: Ross Sozzani (NCSU)
- Task 4.1 Optimize methods and select putative drivers of foliar microbiome effects on plant hosts.
- Task 4.2 Identify minimal microbiomes needed to achieve resilient phenotypes.
- Task 4.3 Identify putative mechanisms for how addition of hub taxa affects wheat phenotype resilience.
- Task 4.4 Examine how consortia of hub taxa affect wheat resilience.
- Task 4.5 Targeted microbiota manipulations by bacteriophage applications.
- Task 4.6 Verify genetic mechanisms underlying microbiome effects on wheat resilience.
- Task 4.7 Translate to real-world agriculture.
WP5: Informatics and data integration
This WP is responsible for establishing a common computational platform for handling major computational tasks related to storing, sharing and analysing large-scale data generated from the other WPs, such as metabolomics, environmental and microbiome data.
WP Lead: Simon Rasmussen (UCPH); Co-lead: Lars Hestbjerg Hansen (UCPH)
- Task 5.1 Establish informatics infrastructure and accessible databases for chemical, environmental and microbial data, which can be used by all WPs.
- Task 5.2 Preparation of microbiomics, metabolomics and environmental data for unsupervised learning and input to the deep learning model.
- Task 5.3 Identify data manifolds for reconstruction of microbial genomes and metabolomic signatures.
- Task 5.4 Network analysis and deep learning approaches for identification of microbial consortia, genes, and metabolites important for plant resilience.
- Task 5.5 Determining the relationship between leaf, root, and seed microbiomes.
- Task 5.6 Continuous update, validation and dissemination of data and results to the plant-microbiome experimental interactions (WP4) and the deep learning model (WP6).
WP6: Building a deep-learning model
This WP is responsible for the integration of all data and results produced in WP1-WP5, which will be based on deep neural networks. These networks will, by using representation learning, serve to integrate the environmental, microbial, metabolomics and network data in a single deep learning model that can be used to predict yield and crop resilience of the wheat system.
WP Lead: Mads Nielsen (UCPH); Co-lead: Simon Rasmussen (UCPH)
- Task 6.1 Determine the prerequisites for building an integrative deep learning model and identify existing mechanistic models of plant growth.
- Task 6.2 Transform microbiome, chemical, environmental and imaging data to appropriate formats for feeding into the deep learning models.
- Task 6.3 Identify resilience trajectories through representation learning of the diverse omics datasets.
- Task 6.4 Create supervised deep learning models for prediction of phenotype membership.
- Task 6.5 Develop time resolved deep learning models for prediction of resilience trajectories and for identifying possible interventions in a given field.
- Task 6.6 Validating, refine and optimize the models after field trial data acquisition and test model transferability.
The following partners are involved in the project
- University of Copenhagen, Denmark
- North Carolina State University, United States
- Netherlands Institute of Ecology, Netherlands
- Technical University of Denmark, Denmark
Christine Hawkes, Professor, North Carolina State University
Ross Sozzani, Associate Professor, North Carolina State University
Gina Brown-Guedira, USDA Professor, North Carolina State University
Jos Raaijmakers, Professor, Netherlands Institute of Ecology (NIOO-KNAW)
Anders Bjorholm Dahl, Professor MSO, Technical University of Denmark
Rasmus Kjøller, Associate Professor, University of Copenhagen, Department of Biology
Mads Nielsen, Professor, University of Copenhagen, Department of Computer Science
Simon Rasmussen, Associate Professor, University of Copenhagen, Novo Nordisk Foundation Center for Protein Research
This project is funded by the Novo Nordisk Foundation (Grant number: NNF19SA0059348).
Project start: 1 Oct 2019
Duration: 6 years