Final Report Summary - ABSTRESS (Improving the resistance of legume crops to combined abiotic and biotic stress)
Executive Summary:
ABSTRESS has developed and implement protocols for the large scale production of Medicago truncatula (Mt) and pea subjected to combined drought and Fusarium oxysporum infection. Associated high throughput phenotyping data was captured from these plants as they were growing. This was the critical first step in ensuring that subsequent genetic and metabolite analyses were robust.
A successful protocol was developed and implemented for M. truncatula so that approximately 6000 plants have been generated under 4 different conditions being; control healthy plants (C), plants exposed to drought (D), Fusarium-infected plants (F) and plants both infected with Fusarium and exposed to drought (FD). Where applicable, plants were produced over a 15 day drought period following prior infection with F. oxysporum. Plant physiology was monitored during this period, with respiration rates and subsequent carbon assimilation measurements made to ensure that the relevant plants showed a typical stress response. The leaves and roots of these plants have been analysed using metabolomics and transcriptomics, and evaluated for other phenotypic properties such as root structure, nodulation and biomass. A full metabolomics dataset has been acquired using liquid chromatography, high-resolution mass spectrometry. This extensive dataset has been explored using multivariate analysis tools with patterns in the data indicating clear discrimination between stressed and unstressed plants. More detailed analysis has identified responses to both single and combined stressors and these observations relate to both anticipated stress responses and novel findings relating to specific perturbations in metabolic pathways.
Similarly, a large transcriptomics dataset has been collected using high throughput (next generation) sequencing. Bioinformatics models have been used to identify 36 hub genes in M. truncatula. Genetic maps for M. truncatula and pea have been used to identify 28 M. truncatula hub gene orthologs in pea. Some of these have been demonstrated to have a similar gene expression profile both in M. truncatula and in pea when exposed to combined stress. Tilling and eco-tilling for selected mutants of these 11 pea genes was conducted resulting in novel genetic material for testing in field trials, potentially offering genetic material for breeding resistance to combined stress. The identified hub genes have been shown to have homologues in a wide range of both legume and non-legume species providing the potential for the use of the data generated here to establish breeding programmes in, for example, tomato. ABSTRESS has therefore achieved its key aim of taking data from model plants into crops within approximately 4 years. However, to thoroughly test the success of the project and to provide SME partners with the value that they anticipated, field trials are required.
Project Context and Objectives:
Summary description of the project context and the main objectives
ABSTRESS applies combined, integrated systems biology and comparative genomics approaches to conduct a comprehensive study of the gene networks implicated in the interaction of drought stress and Fusarium infection in legumes using Medicago truncatula as a model for pea. The project demonstrates the advantages of applying advanced phenotyping methods for the generation of improved varieties of a commercial crop.
The Rationale of ABSTRESS
ABSTRESS was conceived based on three main drivers:
• Our belief that the combined application of comparative genomics and systems biology will deliver a rapid breeding solution to the poorly studied Fusarium - drought interaction in legumes.
• Our belief that a legume crop is essential for the future of European agriculture and our commitment to its improvement through a systems biology approach to plant breeding.
• Our desire to use ABSTRESS as an exemplar of how this technology can solve the Fusarium-Drought problem in non-legume crops as well as providing a generic methodology for resolving multi-component crop breeding challenges, such as the one described here.
Overarching objectives of ABSTRESS
The project had four major, tangible outputs to be delivered through the clear focus and unified approach of the project:
1. The identification of genes from Medicago and pea with a proven ability to alter the interaction between pea and Fusarium under drought conditions and thereby provide a major route to understanding the tripartite interaction of host, pathogen and drought stress.
2. The provision of alleles of genes identified in Output 1 which enhance the resistance of pea to Fusarium under drought conditions and do not negatively impact on crop performance in the absence of infection or drought. The pea genotypes that emerge from the project will be delivered to our SME plant breeding partners to be introgressed into elite breeding material using molecular breeding approaches to shorten the time taken to carry out this exercise.
3. The focus of this project has been the Fusarium – pea interaction under drought conditions, but this pathogen and abiotic stress are highly relevant to several other European crop species. Therefore, a third output was to identify the orthologs of the Medicago and pea genes involved in major crop genera of the Brassicaceae (e.g. oil seed rape), Poaceae (e.g. wheat), Solanaceae (e.g. potato), Compositeae (e.g. sunflower) and Chenopodiaceae (e.g. sugar beet), as well as other legume crops and demonstrate the validity of this approach in one non-legume crop species.
4. The development of molecular phenotyping tools using the latest Omics technologies to quantify gene expression (transcriptomics) and tolerance to combined stressors (metabolomics). Following the identification phase, simplified detection methodologies will be integrated within the workflows of the SME to enable high- throughput screening for advantageous phenotypes at the earliest possible stage in the breeding process. This will enable new varieties to be brought to market in a much shorter time than is currently feasible.
Context
The case for studying pea - an EU legume crop
The combining pea crop is well established as a valuable break crop in arable rotations in Europe. It offers the potential to reduce mineral nitrogen in the rotation due to bacterial nitrogen fixation. It also enables effective non-chemical control of problem grass weeds and provides a non-cereal break for soil borne pathogens such as take-all and pests such as nematodes. Peas have a zero requirement for soil nitrogen during the growing period and healthy crops produce a nitrogen residue immediately available to autumn planted cereals thereby reducing N requirement by between 25 and 55 kg/ha and reducing the diffuse pollution of more leachable forms of N in the early autumn. In the UK currently combining peas of high quality are mainly used as a food commodity, but in Europe, particularly France, they are used in animal feedstock as a high protein source to replace soya. With the increasing cost of imported soya, animal feed compounders would like to increase pea incorporation in feed and in some parts of Europe this demand is largely unfulfilled.
Biological nitrogen fixation is an extremely complex process which is very sensitive to environmental stresses (Zahran, 1999). Legumes are sensitive to abiotic stresses, most significantly water deficit and soil salinity. Drought is currently a major factor limiting crop productivity worldwide. In Mediterranean countries water deficit is encountered not only in arid and semiarid regions but also in areas where total precipitation is high but not evenly distributed during the growing season. In a context of increasing limitations to water use due to climate change and increased population, improving water use efficiency of crops is an important goal.
Rhizosphere microorganisms, particularly beneficial bacteria and fungi, can improve plant performance under stress environments and, consequently, enhance yield by providing plants with fixed nitrogen, phytohormones, iron and soluble phosphate, or by protecting the plant against soil-borne diseases mainly caused by pathogenic fungi. When consideration is given to the positive effects of plant growth promoting bacteria (PGPR) on plant growth under stress conditions, it is apparent that these strains can play an important role in developing strategies to control negative effects of biotic and abiotic stresses (Lugtemberg B. and Kamilova F., 2009).
However in what can be considered as an EU-wide market failure, investment in advanced technologies for legumes lags behind that of the cereals and other major crops such as potato and tomato. Therefore, for food security and economic reasons, it makes strategic sense for the EU to set about improving its own capacity to grow a major legume crop. Improved sustainability can be achieved by identifying and providing germplasm to the legume breeding community that can enable increased productivity with a reduced consumption of natural resources (e.g. reducing the need for pesticides, requiring mineral oil-based production, or scarce resources e.g. water). Advanced breeding techniques underpinned by modern post- genomic technologies promise to accelerate the production of novel germplasm for all crop species, but the means to do this for EU legumes lag behind crops such as the cereals.
The case for studying Fusarium and drought
Peas generally require temperate conditions and are suited to medium to light soil types. Drought during the flowering and pod filling period of spring varieties of combining peas can severely reduce yield. One of the climate changes predicted by forecasters is that summers will be drier in Europe, and one of the impacts of this climate change is likely to be an increase in drought affected spring crops.
An experiment carried out by partner PGRO demonstrated that drought during flowering and pod set reduced yield by around 55% under controlled conditions. However, within a range of 15 commercially available combining pea varieties, the effect of drought ranged from 60% reduction in some varieties to 30% in others. This indicates that there is some tolerance in existing varieties that could be exploited.
Fusarium wilts are widespread diseases caused by many forms of the soil-borne pathogen Fusarium oxysporum, affecting many agricultural crops, including most legumes, cucurbits, tomato, potato, pepper, strawberry, asparagus, cotton, banana etc. These soil borne pathogens can survive as thick-walled chlamydospores, which remain viable in the soil for many years. Control is problematic. It is very difficult to remove F. oxysporum when it becomes established, as it can grow saprophytically in the absence of a susceptible crop. The only effective response is soil sterilization, which is far too expensive for most farmers.
F. oxysporum f. sp. pisi (Fop) is an important and destructive pathogen of field pea, that has been reported in every country where pea is grown (Kraft and Pfleger, 2001). Successive cropping of peas can build up damaging populations of this pathogen. Crop rotation is the only means of maintaining safe levels of inoculum requiring the frequency of pea cropping in a field to be limited. In some regions, this is often no more than once in five years. Some control can be achieved with fungicides but the use of resistant cultivars of plants is the preferred approach (Lebeda et al., 2010).
Sources of resistance in peas are rather limited and difficult to estimate but single genes which have been identified and used in breeding, are rapidly overcome by new races of the pathogen (Infantino et al., 2006). The consortium has fully focussed on Fusarium because of its economic importance across a range of major EU crops. Knowledge on hub genes in M. truncatula and pea will be also of value to closely related legumes suffering from Fusarium wilt (i.e. common bean, cowpea, soybean, lentil, alfalfa, chickpea, etc) but also to non-legume crops (i.e. tomato).
Disease development is favoured by warm temperatures and dry weather, and is expected to be more severe and widespread with predicted climate change scenarios. The interactions between plants and pathogens involve a complex molecular dialogue. First, the pathogen is able to sense and recognize the plant host and to change its metabolism to provide favourable conditions for pathogenicity (Alfano and Collmer, 2004). Meanwhile, plants have evolved to identify structures associated with pathogens, leading to strengthening of existing defences and development of other powerful defence mechanisms. Successful plant infection by F. oxysporum is a complex phenomenon that requires a series of highly regulated processes (Di Pietro et al., 2001). The characteristic wilt symptoms appear as a result of severe water stress, mainly due to vessel clogging. Wilting is most likely caused by a combination of pathogen activities, such as the accumulation of fungal mycelium and/or toxin production and host defence responses, including production of gels, gums and xyloses and vessel crushing by proliferation of adjacent parenchyma cells (Beckman, 1987; Di Pietro et al., 2003). This mode of infection means that under warm and drought conditions the symptoms are more severe.
Response of plants to combined stresses
Whereas there is an extensive literature on the response of plants to single stresses under laboratory conditions, the study of plants’ responses to multiple stresses is still in its infancy (Mittler R & Blumwald E (2010)). The most information at the gene level comes from combined bioinformatics and experimental studies in which it is noted that cohorts of genes show responses to a diverse range of biotic and abiotic stresses suggesting, but not establishing, that such genes play roles in response to combined stress (Swindell WR et al. (2007)). Partner Essex identified the transcription factor HSFA1b as a major hub controlling resistance to biotrophic pathogens, increased tolerance to drought stress and improved seed yield under water-limiting conditions (Bechtold U et al. (2013)) and have also applied a Systems Biology approach to find hub genes implicated in controlling networks responsive to both abiotic and biotic stress. However, in all of these examples no attempt has been made at defining the response of the plants to multiple stresses and determining how it differs from application of the individual stresses. Where this has been done, the work reported has largely been descriptive, being confined to effects on symptoms , e.g. how changes in humidity, salinity or temperature affect resistance to pathogens (Yoshioka K et al (2001); Bechtold U et al (2005); AbuQamar S et al (2009)) or lists of genes from microarray experiments (Rizhsky L, et al (2004)).
ABSTRESS set out to conduct a rigorous temporal comparison between plants subjected to the two single and the combined stresses. Double stress-specific networks have emerged showing how the topology of these networks change compared with the situation under single stresses; identified hub genes underlying cellular mechanisms become targets for exploitation in our breeding programme.
The case for a Medicago truncatula-to-pea trait rapid exploitation pipeline.
Considerable effort has been expended on genome research on Medicago truncatula (hereafter called Medicago) in order to accelerate gene discovery for important agricultural traits. Medicago is highly susceptible to Fusarium infection combined with drought stress (Ramírez-Suero et al (2010)) and is therefore the model of choice for this project. Moreover, it has been demonstrated that Medicago and other legumes present several syntenic blocks, i.e. genomic regions with conserved sequences. Therefore, the identification of genes important in the Fusarium-drought interaction in Medicago can be used directly to identify the orthologous genes in pea.
ABSTRESS aimed at identifying hub genes involved in the control of resistance to combined biotic and abiotic stresses. Defined conditions were applied to Medicago first, and then the knowledge generated was transferred to pea in order to develop tools to accelerate breeding programmes.
Project Results:
Main results
ABSTRESS has combined integrated systems biology and comparative genomics approaches to conduct a comprehensive study of the gene networks implicated in the interaction of drought stress and Fusarium infection in legumes. The model legume Medicago truncatula has been used to rapidly identify characteristics for introgression into elite pea varieties.
The project is organised into 6 work packages (WPs). The first WP aimed to develop and implement protocols for the large scale provision of plant materials from Medicago truncatula and pea with associated high throughput phenotyping data. This work was critical to ensure that subsequent analysis was robust. Therefore much emphasis was placed on refining protocols for use in the high throughput phenotyping platform in Dijon (PPHD). A series of preliminary experiments (pre-pilots and pilots) with reduced numbers of plants were run with both Medicago truncatula and pea to establish optimal protocols for drought application and Fusarium infection. Multiple parameters had to be considered and tested, including pot size, levels of water, environmental humidity, length of treatment, Fusarium strains, infection protocols, etc. Various phenotypic and biological features were measured to monitor the response of the plants to the stress applied. After several iterations and much refinement of the methods, a successful protocol was developed and implemented for the large scale experiments, BOM (“big one” Medicago) and BOP (“big one” pea). Thousands of plants were generated under four different conditions: control healthy plants (C), drought-exposed plants (D), Fusarium infected plants (F) and plants both infected with Fusarium and exposed to drought (FD). Where applicable, plants were produced over a 15 day drought period following prior infection with Fusarium. Plant physiology was monitored during this period, with respiration rates and subsequent carbon assimilation measurements made to ensure that the relevant plants showed a typical stress response. Transcriptomics, epigenomics and metabolomics analyses were conducted to identify genes involved in stress response in the model plant Medicago. Comparative genomics was then employed to identify orthologue genes in pea, which were then investigated and mutants selected for introgression into elite pea cultivars in order to introduce characteristics associated with resistance to biotic and abiotic stress.
PHENOTYPIC ANALYSIS
Carbon and nitrogen acquisition, partitioning and allocation are tightly linked within plant parts as they both contribute to i) plant structures establishment (ie leaves, roots, nodules when plants interact with soil microbes, among them rhizobia and / or pathogens) and ii) their function for acquiring atmospheric and soil resources. Understanding how these two components (i.e. structure and function) interact when plants are faced with abiotic and biotic constraints remains a key issue towards selecting plants adapted to current agroecosystems. Both establishment and activity of the Rhizobium-legume symbiosis have been found to be extremely sensitive to drought stress. Several studies have shown that symbiotic nitrogen fixation is more sensitive to drought than other physiological parameters such as photosynthesis, transpiration, or nitrate assimilation.
Experiments were conducted to fully characterise plant status, focusing on shoots, roots and nodules. Plants were characterised dynamically throughout the experimental period, both destructively and non-destructively by image analysis. Non-destructive characterisation was undertaken with shoot and root imaging twice per week using the PPHD facilities at Dijon. Some plants for both the BOM and BOP were grown in parallel in rhizotubes in order to assess visually and non-destructively phenotypic changes across time. Images were acquired using the phenotyping cabin devoted to roots, which enabled the dynamically analysis of shoot, root and nodule growth on the same plants throughout the experiment.
Destructive measurements consisted of analysis of biomass partitioning and allocation, measurement of physiological traits (transpiration, stomatal conductance) and morphometric traits such as leaf area. A labelling experiment using 13C and 15N2 was further performed in a phytotron specially designed to measure precisely carbon and nitrogen entry and allocation to the various plant parts. Nodule gene expression studies were conducted by quantitative PCR. The nitrogen fixation genes nifA, nifH and fixJ of the symbiotic bacteria Sinorhizobium medicae and Rizhobium leguminosarum were analysed. Inhibition of N2-fixation in both Medicago and pea root nodules as a result of drought stress and Fusarium attack was analysed through the expression of structural genes of nitrogenase complex (represented by nifH expression) and on the positive activator of transcription nifA.
Elucidating plant strategy and nodulated root reaction to factorial abiotic and biotic stress: Biomass allocation to plant parts: Water stress mainly affected shoot compartments while carbon allocation to roots and its biomass were maintained. The situation observed was reversed with the addition of Fusarium infection, which might result from a strategy of the pathogen to “preserve” its host. While water stress negatively impacted nodule number in Medicago, mean nodule biomass was targeted in pea. Fusarium effect was principally detected on pea plants where nodule number was decreased while mean nodule weight increased. Using the precise labelling experiments, only the drought stress affected N and C flow within plants for both pea and Medicago.
C and N partitioning: In Medicago plants the stem mostly attracts carbon while pea leaves were the bigger carbon sink. Labelling experiment demonstrated that drought greatly decreased carbon incorporation of shoots, stems, and nodules for pea and Medicago while roots were less impacted, as observed for biomass. Fusarium infection did not impact carbon incorporation. N flow was greatly reduced for pea leaves, stems and nodules. Medicago seemed to be much less impacted for its entire compartment.
A differential strategy between pea and Medicago: In pea both leaf and nodule specific activity and biomass were severely decreased by drought stress and not by Fusarium infection. In contrast, in Medicago, only leaf specific activity was reduced by water stress while nodule specific activity was maintained.
Carbon and nitrogen uptake are tightly linked: a decrease of carbon acquisition directly impact nodule biomass (PEA) and or number (Medicago), which in turn reduces N flow through symbiotic nitrogen fixation. The relationship between C and N acquisition allows an estimate of the reactivity of plants facing a stress.
Characterising the expression levels of key bacterial genes: The application of singular or combined stresses lead to significant changes in the expression levels of bacterial nitrogen fixation genes over time. This suggests that potential hub genes to be found in Medicago plants have an effect on nif genes after the sixth day of treatment. Impact of single or double stress was similar for pea but down-regulation of N-fixation genes began earlier. The plants subjected to both stresses have a severe decay in the nitrogen fixation activity.
Towards Alternative Strategy to improve plant growth under abiotic and biotic stresses: Production of the auxin indole-3-acetic acid (IAA) is a common feature for plants with growth promoting rhizobacteria, having antagonistic activity against phytopathogenic microorganisms. The aim was to assess if the IAA overproduction in rhizobia had a positive effect on the response of plants to both pathogen attack and water stress.
Biotic stress resistance of IAA-overproducing rhizobia under free-living conditions: The bacterial strain S. medicae MD4, used to nodulate Medicago plant in Dijon, and its derivatives able to synthesize higher levels of IAA were tested for their antifungal activity against F. oxysporum sp. medicaginis through dual plate assay. The MD4 strains producing the highest IAA levels (MD4-65 and MD4-66) showed the highest antagonistic activity against Fusarium growth.
Effect of rhizobial IAA-overproduction on plant tolerance to drought stress and pathogen attack: Considering that the response mechanisms leading to salt- and drought tolerance are very similar, we investigated if IAA-overproducing rhizobia strains were able to improve the resistance of Medicago and pea to drought stress. For this study the modified strains producing the highest IAA levels have been selected. Preliminary results were obtained when Medicago and pea plants inoculated with the specific IAA-overproducing rhizobia were subjected to the double stress (drought stress and Fusarium attack). After stress treatment, visual signs of ion toxicity were present in the leaves of host plants. However, symptoms of senescence and zones of necrosis on the leaves of plants nodulated by the specific IAA-overproducing rhizobia were less evident as compared to the controls. The observed phenotype reflected changes in the expression levels of genes playing a key role in the mechanisms of drought response (the data are not shown here because they will be used for a patent application).
In conclusion, drought greatly decreased carbon incorporation in shoots, stems, and nodules while roots were less impacted. Only drought stress affected N and C flow within plants for both pea and Medicago; no clear trend could be distinguished in plants under pathogen attack. The labelling experiment also showed that N flow was greatly reduced by drought in pea while Medicago seemed to show much less of an impact. Water stress negatively impacted nodule number in Medicago while mean nodule biomass was targeted in pea. The Fusarium stress was principally detected in pea plants where nodule number was decreased while mean nodule weight increased. In pea both i) leaf specific activity and ii) nodule biomass and nodule specific activity were severely decreased by water stress, not by Fusarium. In Medicago, only leaf specific activity was reduced by water stress. This demonstrates differential plant-microorganism interactions between pea and Medicago. A tight carbon/nitrogen relationship allows estimating the degree of stress sensed by the plants and their efficiency to react when facing a stress. Drought stress led to significant down-regulation of genes involved in nitrogen fixation, an effect that was amplified by Fusarium attack. IAA production by rhizobacteria is probably one of the most important traits promoting plant growth under biotic stress conditions. The infection of three different legume plants with the specific IAA-overproducing rhizobia strains greatly improved the stress resistance of these plants.
MOLECULAR PHENOTYPING
METABOLOMICS
Pilot metabolomics experiments were undertaken in both Medicago truncatula and pea. Procedures were established and optimised for handling low sample amounts (< 10 mg), which was key for the subsequent full analysis, and for developing and validating analytical approaches using Liquid Chromatography – High Resolution Mass Spectrometry (LC-HRMS). Baseline metabolite profiles for both plants, leaves and roots, were established from a pooled reference material.
For the full BOM analysis, metabolites were studied in a non-targeted, holistic fashion for 288 Medicago samples (4 biological conditions x 3 biological replicates x 12 time points, shoots and nodulated roots analyzed separately). Data were assessed through a multi-faceted approach using Progenesis QI (NonLinear Dynamics), R (R Core team) and C# (Microsoft). Raw data files were uploaded into Progenesis QI for chromatographic alignment and feature detection. Outlier removal and QC (quality control) correction were applied using R before evaluation of the data using Pearson correlation and k-means clustering. Pearson correlation measures the correlation between two variables (metabolites) and is thus able to identify metabolites associated with particular patterns over time. Template patterns can be used to identify molecules whose abundances show correlation with time under the specified experimental conditions. Clustering methods can group together similar patterns in an unbiased manner. Interesting clusters of metabolites may then be investigated, for instance to identify common metabolic pathways or metabolite classes. k-means is a popular clustering method which starts with k randomly chosen cluster “centres”. Observations are assigned to the nearest centre and the centres are then moved to the average of all observations assigned to those centres. This iterative process is repeated until no further changes occur.
Metabolite identification confirmatory analyses were undertaken on an Orbitrap Velos Pro (Thermofisher Scientific) using the MS/MS and MSn capabilities of the instrument. Where a corresponding analytical standard was available, this was analysed concurrently to confirm compound identity. Where an analytical standard could not be obtained, compounds were “affirmatively” identified or dismissed using theoretical fragmentation from Mass Frontier software (HighChem Ltd) and / or library matching (if available) with the Metlin MS/MS database (Scripps Center for Metabolomics). Theoretical confirmation involved comparing accurate mass product ions (MS2 and MS3 ions) generated from a sample with a theoretical spectra produced by the software. Once confirmed, metabolites were tentatively assigned to pathways or a suggested biochemical impact using the open source databases KEGG (Kyoto Encyclopedia of Genes and Genomes, Kanehisa Laboratories, Japan.) and Mediccyc (http://mediccyc.noble.org/ The Samuel Roberts Noble Foundation, OK, USA). Thirty nine different metabolites were discovered to significantly change in abundance over time in leaves as a result of either one or both stresses. Twelve of them were identified with confidence, with the other 27 affirmatively identified as described above. Forty seven different metabolites were discovered to significantly change over time in root as a result of either one or both stresses. Fifteen of them were identified with confidence, with the other 32 affirmatively identified. Details of these metabolites are available on the Abstress website (Data for D2.1 D2.1 Metabolites identified).
TRANSCRIPTOMICS AND sRNAs
Long RNAs were extracted for 288 samples of Medicago truncatula (4 biological conditions x 3 biological replicates x 12 time points, shoots and nodulated roots), and analysed separately using GenXPro’s proprietary Massive Analysis of cDNA Ends (MACE). MACE libraries were generated for all 288 samples. The libraries were barcoded and sequenced at an average sequencing depth of 4.3 Million reads per library. Raw sequencing data were processed using GenXPro standard MACE workflow including cleaning of data from adaptor sequences, discarding of low-quality reads, and TrueQuant analysis to discard amplification artifacts. The MACE reads were mapped to the Medicago genome version 4.0 for transcriptome analysis. Unmapped reads were assembled into contigs which were consequently BLASTed against different publicly available data bases as e.g. SwissProt, Refseq, etc. providing functional assignment to the sequences.
Differential gene expression was evaluated using experiment design contrasts for pair-wise comparisons by means of tools including Deseq, EdgeR (R-framework), Cufflinks and also a similar library from the 'bioinfo' toolbox in the Matlab environment. All these tools use Negative Binomial Distribution in the statistical evaluation of pairwise comparisons to generate P-values for each tag in the dataset. Finally the obtained P-values were transformed to False Discovery Rates (FDR) using the Benjamini–Hochberg procedure.
Workflow used to process MACE transcriptomics data
The processing included quality control checking, taking account of biological variation, identifying differentially expressed genes (DEGs) across the time series and inputting the data from selected genes into VBSSM to generate 3 inferred gene regulatory network (GRN) models:
1. An inferred GRN in the leaves of Fusarium-infected/drought-stressed (FD) plants consisting of the top 100 DEGs in this dataset.
2. An inferred GRN in the roots of FD plants consisting of the top 100 DEGs in this dataset.
2. An inferred GRN consisting of all transcription factor (TF)-encoding DEGs expressed in the leaves of FD plants
The MDS tool was used for visual estimation of similarity between datasets. Sample replication is very important for establishing true events taking place in gene expression under different experimental conditions. Tissue samples from individual organisms tend to show high variance in distributed data among replicated samples. This occurs due to the natural variation of biological response between individuals. Additional factors associated with the quality of the RNA, sample preparation and library handling can also introduce significant artefacts that may lead to unwanted data dispersion. Therefore, we considered it prudent to examine data variability within sample groups prior to statistical analysis of the whole experiment. If individual replicates within one group exhibited significant dissimilarities and, in addition, were similar to those from a contrast group, they were removed. It was observed that the transcriptome data showed a dispersed distribution at almost all TPs (time points), which would be an issue for TPs represented by a single sample. Plots for all samples are available on the ABSTRESS website (Data for D2.4 D2.4 Datasets.zip).
Evaluation of differentially expressed genes
Tables showing the Differentially Expressed Genes (DGE, p-value threshold of <0,05) for drought vs control, Fusarium vs control and Fusarium-drought vs control conditions, for the kinetic analysis, both for shoot and root organs are available on the ABSTRESS website (Data for D2.1 file named “DGE genes”).The numbers of DEGS in each dataset are summarised in a Venn diagram is shown in Figure 1. The data reveal that a substantial portion of the DEGs in the FD datasets were unique to the double stress.
Small RNAs
To correlate the transcript regulations observed to small RNA populations, critical physiological time points of the kinetic were selected to identify differentially expressed sRNAs: day 1, 5 and 12, for single and combined stresses, shoots and nodulated roots, 3 biological replicates each. A protocol was set up to purify small RNA from Medicago leaves and roots. A bioinformatics workflow was established at IVS-CNRS to map RNA-seq data on the genome, and analyse the nature of small RNAs (miRNAs, siRNAs, nat-siRNAs and heterochromatin-derived siRNAs) as well as their organization on the genome (e.g. clusters, precursors). The corresponding 24 sRNA libraries were constructed and then sequenced. The data were processed with miR-Prefer and OmiRa software. The quality of the sequencing data was checked, the sRNAs were mapped to the Medicago truncatula genome (v4.0) and software to characterise the distribution between coding (ORFs) and promoter regions was developed. Analysis of miRNA and potential targets was also carried out using a combination of public degradome data and optimised plant software for target detection. The lists of differentially sRNA expressed (DSE) in the different stress conditions are uploaded on the ABSTRESS website (Data for D2.1 files named DSE-Root or DSE-Shoot). Overall, correct mapping frequencies of 24nt siRNAs (around 75%) and distribution of 8 to 10 M reads according to standard 21-24 nt ratios were obtained. Mapping of the sRNA revealed a general trend in all libraries: enrichment of 24nt siRNAs was obtained in 1kb promoter regions, and a significant fraction mapping also in introns and in un-annotated genomic regions. 54 promoters regions showed a differential expression of 24nt siRNAs depending on the stress condition, including 23 corresponding to genes with predicted functions. In leaf, 21/24 candidate promoters showed 24nt siRNA levels induced by drought (alone or combined) and in root, 24/31 candidates showed siRNA levels reduced by Fusarium (alone or combined). These differential patterns are listed in the file “D24-siRNA promoters” uploaded on the ABSTRESS website (Data for D2.1).
DNA METHYLATION
Genome-wide cytosine methylation analysis was conducted in DNA samples of Medicago leaf and root tissues exposed to combined stressors. These samples were from the same critical physiological time points used for sRNA profiling. For this purpose, MSDK libraries were obtained according to established protocols for high-throughput methylation-specific digital karyotyping improved by GenXPro’s proprietary SuperTag technology. In short, cytosine methylation was quantified at approx. 200,000 loci across the Medicago genome for each stress-situation using Illumina-sequencing. DNA was extracted from all root and leaf samples and was digested with the methylation-sensitive endonuclease HpaII. The flanking regions of the HpaII digestion-sites were sequenced using Illumina HiSeq2000 to assess differences in methylation at the level of the whole genome after quantification of the HpaII flanking sites. A total of 2.2 to 24 Million sequences were obtained per library, for a total of 24 libraries. The sequences obtained were subjected to bioinformatics processing which included quality assessment, de-multiplexing, mapping against the Medicago v.4 genome release, annotation, identification of “downstream” genes, and assessment of quantitative differential methylation. A robust correlation with genomic loci showing changes in methyl-cytosine content and the different level of expression of the transcripts (down regulation /activation) originating from the loci was carried out. A report on genome-wide quantitative methylation profiles containing the 1000 most likely differential methylated loci for the different stress/control comparisons can be found on the ABSTRESS website (Data for D2.1 file named “DME genes”). In addition, for the 54 promoter regions showing a differential expression of 24nt siRNAs depending on the stress condition, a correlation with differential methylation patterns was established (D24-siRNA promoters file), revealing several loci showing both changes in DNA methylation patterns and differential accumulations of 24nt siRNAs.
INFERRING GENE-INTERACTION NETWORKS FROM TIME SERIES EXPRESSION DATA
Hub genes are defined as the most highly connected genes in an inferred network and the working assumption is that such genes are more likely than others to be influential on whole plant processes, such as the response to combined Fusarium infection and drought stress being studied in ABSTRESS. Furthermore, by targeting hub genes for manipulation, the researcher can determine the accuracy of the model by seeing if proposed connections are disrupted when expression of a hub gene is perturbed. The data from such experiments can be used to develop later more refined draft models. Such re-iterative, unbiased approaches to model development underpin a Systems Biology approach in this context. However, the number of re-iterations to refine a gene regulatory network (GRN) is limited by the cost and time available. Nevertheless, the identification of hub genes by Bayesian Dynamic Modelling provides a mean of discovering novel genes that have a strong influence on both the GRN and the output phenotypes that the network influences.
We used VBSSM to infer gene regulation networks from the time series expression data. The VBSSM algorithm builds a probability model for each conditional state (time point) and scores the odds for fitting the expression data. One of the critical factors for robust evaluation of the models is the size of input population of genes. The VBSSM is effective in prediction of relatively small gene networks (max 100 genes). While this avoids over-fitting data to models, it does create a potential problem for choosing the ‘right’ population of gene candidates because the typical DEG analysis of RNA-SEQ data produces a much higher number of genes than VBSSM allows. We therefore modified the approach using the argument that the DEG analysis is most likely to give the best fit for a model assuming that potential hub genes are distributed among the top ranked genes. The ideal starting point for unbiased selection from our FD (Fusarium and drought combined treatments) DEG dataset would then be the top 100 genes.
To evaluate this assumption we generated an extra 20 networks for the same number of genes selected from randomly-permuted DEGs. The comparison of the Network Scores tests the hypothesis that hub genes are indeed amongst the top 100 DEGs and also challenges the stability of the inferred models for the same highly-connected genes. It should be noted that any inferred network can, in reality, reflect only a limited number of local connections. Therefore, the repeated identification of hub genes from random network modelling would reflect their involvement across a presumed wider gene regulatory network than could be computed and visualised as a single entity. It was these highly connected genes that appear as hubs in multiple models that were considered for further investigation. The full range of comparisons tested using VBSSM and the total number of selected genes provided as inputs over all modelling iterations are shown in Table 1.
Highly connected genes (HCGs) were classed as those which appeared in multiple models as hub genes, which in a single network were regarded as having the largest number of connections with neighbouring genes inferred from the DEG expression data. To evaluate HCGs from all networks, negative binomial statistics were used. A cumulative negative binomial function was used to obtain P values for distribution of connections for each gene in all networks and the genes with P < 0.05 were regarded as HCGs for the tested experimental condition. These are shown in Tables 2 and 3. The retrieved MACE transcripts corresponded to HCGs were verified using the BLASTX engine against the Medicago protein database.
Finally, we were able to temporally cluster by k-means clustering the FPM expression values for the HCGs from the leaf and root FD data into 16 and 12 groups respectively (Figures 2 and 3). This confirms that the modelling was picking out genes with different patterns of expression, making it likely that the changing expression of genes would impact on one another as would be the case if they were all part of an extended network. The inferred networks were returned in Simple Interaction Format (SIF) which can be visualised using, for example, Cytoscape freeware (http://www.cytoscape.org/). A series of interactions were inferred at different thresholds of probabilities (Z scores) and each .SIF file was marked accordingly. All files at all Z score thresholds that were set have been provided in folders for each VBSSM-generated GRN and are available on the ABSTRESS website project pages. The “VBSSM ready” data files with dummy replicates are available in their own folder on the ABSTRESS project website. A final selection of HCG hubs was made for further analysis (Tables 2 and 3, genes labelled in bold). In addition, analysis of the metabolite data showed that some of the metabolites unique to FD leaves or roots could be linked to the selected hub genes. This linkage is shown in Table 4.
GENETICS
Preliminary comparative genetic analysis of pea and Medicago was carried out in preparation for the orthologue identification exercise once the Medicago hub genes were defined. The pea genetic map was refined, alignment with Medicago genome verified and new reference markers were identified. Genotypes of some markers of F2 individuals from cross Mt (M. truncatula) A17 x Mt A20 displaying genetic linkage showed an unusual (cruciform) linkage pattern. Pollen viability test (~ 50 % lethality) suggested translocation. LG4 and LG8 were involved in the translocation. With the help of a cross between Mt 13U x Mt A20, markers were re-mapped and the breakpoint was delimited by genetic means. Mapping was performed with corresponding markers in diploid alfalfa (M. sativa) and pea. The results showed that M. truncatula reference accession A17 has an aberrant chromosomal configuration. The nucleotide sequence of the breakpoint was delimited at the breakpoint. In M. truncatula A17 genotype, the breakpoints separate the lower arm of LG4 from LG8 and vice versa. A17 stands alone because no other accession studied carries the same translocation in these chromosomal segments. The cross between M. truncatula A20 x 13U or any other crosses between M. truncatula accessions except A17 are suitable for mapping genes/markers in the translocation region. If M. truncatula A17 x A20 is needed for mapping the translocation point has to be taken into consideration.
Identification of the pea homologues of the M. truncatula hub genes
Based on their expression pattern in response to combined drought stress and Fusarium infection, 36 M. truncatula hub genes were selected as the most highly connected and differentially expressed. Twenty three hub genes were expressed predominantly in leaf and thirteen hub genes showed root specific transcription. In order to identify the pea homologs of the Medicago hub genes, two consecutive steps were applied; foremost sequences with high similarity to the Medicago genes were identified in the pea transcript and genomic databases. As a next step, the genomic locations of the identified pea sequences were determined and orthologous relationship was judged based on the genomic location and co-linearity between the Medicago and pea genomes. Sequences of pea genes were searched for at the Cool Season Food Legume Crop Database (https://www.coolseasonfoodlegume.org/) and at Genbank / EMBL databases with the provided 36 Medicago hub genes. In parallel, the Unigene set of pea was analysed to identify pea homolog sequences and obtain complete sequences of the pea genes. We could identify homolog sequences in the pea genome for 26 Medicago HUB genes with high sequence similarity and found sequences for 3 Medicago genes with lower similarity (Table 5). Unfortunately, no hit could be detected for 7 of the Medicago hub genes, most of which showed nodule-specific expression. This could be because the homologous genes in Medicago and pea either (1) do not share significant similarity although present in both genomes, (2) the genomic regions where they reside have not been sequenced in pea or (3) these genes have evolved since the divergence of the Medicago and pea lines.
Genetic mapping of pea hub gene candidates
As a next step, the genomic locations of the identified pea sequences were determined and orthologous relationship has been judged based on the genomic location and co-linearity between the Medicago and pea genomes. In order to increase the efficiency of the genetic mapping, an investigation aiming to resolve the anomaly of the genomic structure experienced in the reference genome of M. truncatula A17 genotype was conducted at the early phase of the project. Genetic mapping on the F2 population of the Medicago A17 x A20 accessions revealed a cruciform linkage between markers located on the lower arm of linkage group 4 and 8 (taking the diploid M. sativa genetic map as reference). We found that linear genetic linkage could be as observed between cross Mt A20 and Mt Jemalong plants. A former study suggested that the Mt A17 genome contains aberrant chromosomes which might cover a translocation between the affected chromosomes. In order to identify the translocation point in the Mt A17 genotype, published genomic sequence data was corrected and completed by de novo DNA sequencing. The translocation point was narrowed down by high resolution genetic mapping using generated markers from the improved genomic sequence using the Jemalong x A20 and A17 x A20 mapping populations. This analysis led to the discovery of a DNA segment where sister chromatid break and reunion occurred between the two chromosomes involved in the translocation event.
In order to prove the orthologous relationship between the Medicago HUB genes and their pea homologs, the genomic location of the pea homolog sequences were determined. Oligonucleotide primers were designed based on the pea homolog sequences and DNA fragments were amplified with PCR using the plants of JI281xJI399 RIL (Recombinant Inbred Lines) population. We determined the genomic location of the pea homologs with genetic mapping. Briefly, the genotypes of the RIL individuals were determined based on the detected polymorphism either in agarose gel or using SSCP analysis, genetic analyses were carried out and the co-linearity of genetic markers in the region was analysed.
In order to allow us to extend the genetic mapping of 8 pea homologs that did not show polymorphism between the parents JI281 and JI399 of the pea RIL population, we carried out an in silico genetic mapping project using a recently published WGS data of another pea genetic map which resolved map positions for 6 of the 8 genes. To do this, we constructed a composite genetic map containing pea core markers using the WGS data of the ‘Baccara x PI180693’ RIL mapping population. We analysed the sequences of the identified pea homologs of Medicago HUB genes for SNPs and based on the identified polymorphism we genotyped the selected plants of the RIL population. In summary using a combination of these two approaches, we identified the map location of pea homologs of 22 Medicago HUB genes which were positioned in co-linear genomic locations between Medicago and pea indicating their orthologous relationship (Table 6). We found 7 homolog sequences that were either mapped into non-syntenic genomic regions indicating that these genes are probably paralogs (homologs that developed as a result of gene-duplication events) or showed no polymorphism between the parental lines and therefore we were unable to map them.
Sequenced key genes for drought-Fusarium resistance Root and leaf hub genes responding to both drought and Fusarium infection were identified from transcriptomics and gene interaction network analyses. Sequences of the putative Pea orthologues were identified by using BLAST on the RNAseq-derived Pea Gene Expression Atlas. Confirmation of probable orthologues was established for many of the sequences by mapping and showing syntenic locations. The resulting set of sequenced genes, together with their predicted functions, is listed in Table 7.
EXPRESSION ANALYSIS OF HUB GENE ORTHOLOGUES IN PEA
In order to test the conservation of the Medicago inferred networks in pea, the expression of the pea orthologues of the Medicago hub genes was analysed. Peas were grown in symbiotic conditions and challenged with combined drought stress and Fusarium infection. Analysis was done by quantitative real time (q) PCR of cDNA from RNA prepared from leaf samples. There was not enough root material to test. The initiation of this task required the identification of the pea homolog sequences for transcribed sections of the Medicago hub genes. Twenty two (22) hub gene orthologues in pea were identified. Of these, 19 genes gave detectable and reproducible qPCR signals. Differential expression, with more than 2-fold change between the double stress treatments and controls was observed in 11 of these 19 genes. The panels in Figures 4A and 4B show the relative expression levels of the selected Medicago hub genes in pea grown in symbiotic conditions and challenged with combined drought stress and Fusarium infection. In our analysis, with 4 biological replicates, we used the confidence interval (CI) at 90% due to the marked variation within the replicates in some of the groups at certain time points. In most cases, the drought stress only and FD samples at 12 days post treatment showed similar expression profiles, suggesting that the main contribution to differential expression was the drought stress in the expression profile (Figures 4A/4B and Table 8). However, 6 of the genes analysed (PsCam006964, PsCam043610, PsCam002572, PsCam040229, PsCam039241 and PsCam058628) showed a synergistic response to the double stress, i.e. more or less than the sums of the fold increase in Fusarium or drought stress only.
Next, we compared the expression pattern of the 11 genes that showed a differential response to combined Fusarium infection and drought stress (FD) between the Medicago and pea over time. Heat map representations showed that both Medicago and pea adopt different expression states during the time course under double stress (FD) (Figure 5). In general, the expression of the hub genes peaked at 5 days after treatment (DAT) in Medicago, however, in pea the hub genes selected peaked at 12 DAT. This indicates that the combined stress effects were more rapid in Medicago than they were in pea. However, the general response of the pea hub gene orthologues was similar enough to indicate that the responses to the stresses in these two legume species were broadly qualitatively similar.
CHANGES IN ROOT AND LEAF METABOLOME OF PEA THAT CORRELATE WITH CHANGES IN MEDICAGO TRUNCATULA UNDER COMBINED DROUGHT AND FUSARIUM INFECTION
All metabolites found to significantly change in combined drought and Fusarium stressed Medicago plants in both leaf and roots are anticipated to be present in pea plants.
Table 9 shows the list of 22 pea orthologues of Medicago HUB genes, potential metabolic pathways affected and any associated metabolites found in the Medicago analyses that correlate with these pathways. Pathway information has been taken from Kegg (http://www.genome.jp/kegg/).
Pea material from the experiment named “Big One Pea” (BOP) was analysed in a non-targeted manner for metabolites by LC-HRMS following identical extraction and detection approaches as in the BOM.
Three of the 4 sample replicates for both pea leaf and root materials were analysed for days 2, 5, 8 and 12. All groups were analysed: control (C), drought only (D), Fusarium stress only (F) and combined stressors (FD).
The data of the BOP analysis corresponding to the metabolites discovered to significantly change over time for Medicago were selectively extracted and plotted in Excel and the overall pattern observed in order to ascertain if the metabolite was behaving in a similar manner in pea over time under dual stress. This pattern was either an increase or decrease in expression over time. For those metabolites that looked to agree in response, T-test statistics were employed on the final time point using the signal intensities of the control and dual stress groups.
From the 35 metabolites found to show a response to dual stress in Medicago leaves, 21 behaved in the same fashion in pea. Phenylalanine, arginine, lysine, proline, indoleacrylic acid and medicagenic acid glucuronide ester correlated in pea leaves up to day 8 time point only.
From the 41 metabolites found to show a response to dual stress in Medicago roots, 17 behaved in the same fashion in pea. Isopterofuran, proline, galactosyl-glycerol-3-phosphate, citrate, dehydroascorbate and dehydro-arabinono-1,4-lactone correlated in pea roots up to day 8 time point only.
Table 10 shows a list of metabolites that correlated in response to combined stressors between pea and Medicago. From this list, 11 metabolites were found to change uniquely in combined stress plants only.
TILLING and ECO-TILLING FOR PEA ORTHOLOGUE HUB GENES
Sequencing natural pea hub gene variants
In order to identify allelic variants for the HUB gene orthologues in different natural pea variants (eco-TILLING), primers were designed and synthesised for the 11 pea HUB gene orthologues that showed a response to combined drought and Fusarium infection in both Medicago and pea. Primer design was based on the sequences of the pea variety “Cameor”. To cover the coding sequences of these 11 genes, 21 primer pairs were designed and targeted amplicon sequencing (TAS) was carried out to determine allelic variants for the HUB gene orthologues in the selected pea genotypes. It was decided to access only the recently identified pea genotypes in groups 3.4 and 3.5 (Figure 6). This decision was based on a recent publication that showed that these groups represented the widest diversity of sequence variation in pea. The final list of JI accession number from these groups were as follows: Cameor (reference sequenced genome), JI157, JI186, JI189, JI252, JI281, JI399, JI707, JI1101, JI1105, JI1107, JI1109, JI1113, JI1116, JI1117, JI1118, J1124, JI1478, JI1543, JI2022, JI2360, JI2546, JI2603, JI2604, JI2606, JI2822, JI3003, JI3022, JI3150, JI3217, JI 3255, JI3256. The 31 accessions have been selected from a European Pisum germplasm collection to represent the maximum diversity.
Primers were designed to generate 24 amplicons in a total length of 12.5 kb to cover the CDSs of the 11 hub genes. The amplicons were prepared for Illumina next-generation sequencing using a Nextera XT kit on pooled amplicons from each accession, and sequenced on a MiSeq v2 platform across a single flowcell.
The sequencing generated all together ~35 M Illumina high quality reads (sequenced using paired-end technology) and about ~80% of them (~28 M reads) could be mapped to the 24 amplicon reference sequences of “Cameor” with the CLC genomic workbench 5.5.2 using its reference mapping function. Except for two amplicons, all the CDSs of the 11 HUB genes could be assembled for the 32 accessions although we found low coverage of the genes in a few cases.
The coding sequences of the HUB genes from the 31 pea accession were assembled based on the reference mapped contigs. DNA and amino acid sequence alignments were carried out with MEGA 7.0.21 and analysed with UPGMA hierarchical clustering method to compare the alleles of the 11 HUB genes. For two of the genes (PsCam002810 and PsCam006964) no polymorphism could be detected at amino acid level in the coding region.
Based on the detected polymorphisms at nucleotide and amino acid level, the alleles of the 31 accessions composed of two groups in the case of gene PsCam010311 and three groups in the case of PsCam034819. For other genes and proteins, the alleles were distributed into more than three clusters. In the case of the PsCam031330 gene early stop codons were detected in accessions 1543 and 3003, indicating that these alleles encode for truncated proteins. The positions of the amino acid substitutions and the number of each allele for each gene are summarised in Table 11.
The observations made in the limited time that the groups have had to analyse the data suggest no overall pattern. There was a range of amino acid substitutions, ranging from zero to ten. The one gene which stands out is PsCam031330, which codes for Heat Shock Transcription Factor (HSF). In accession JI1543 and JI3003 premature stop codons in the gene mean that a truncated form of HSF would be made, which was unlikely to be functional since this would terminate the protein within its essential DNA binding domain (DBD). The distance between the DBD and OD is less than 20 residues, which means PsCam031330 likely encodes for a class A HSF, its nearest match being HSFA1b from Arabidopsis. If so, this is consistent with a role for this gene in disease resistance and drought stress responses. The responses to both types of stress are controlled in part by HSFA1b in Arabidopsis.
In conclusion, there are pea accessions harbouring potential hub gene alleles which would be worthy of further investigation. The most notable of these is PsCam301330 coding for a HSF in which 2 alleles would be incapable of producing a functional protein. Further work is needed but post-ABSTRESS we will follow up this observation.
Primer design for TILLING mutants of pea hub gene orthologues
The primers for TILLING were designed and made for the 11 selected pea HUB gene orthologs chosen on the basis of a distinct response to combined drought and Fusarium infection. Primers were also designed for two newly identified hub genes (PsCam044570 and PsCam009624). All of these primer pairs were used to obtain pea mutants by TILLING the Pisum sativum var. Caméor population held at UMR Agroécologie. A summary of the mutants identified is shown in Table 12.
In addition, the previously identified drought resistance hub genes PsVPE (a putative vacuolar processing enzyme) and PsRP (a ribosomal protein) were also used for TILLING. Primers included (i) nested TILLING primers for identifying the mutant sequences, and (ii) dCAPS markers for following PsVPE in backrossing generations. Two of the mutants identified were multiplied in sufficient quantities to start field and lab testing (see below); the remaining ones are currently being multiplied.
PEA MATERIAL GENERATED FOR FIELD TESTING
Screening germplasm collections for accessions carrying combined resistance to both drought and Fusarium oxysporum
Collections of both pea and M. truncatula accessions were screened for resistance to F. oxysporum f. sp. pisi and f. sp. medicaginis, respectively, allowing the identification of resistant accessions. A detailed evaluation method coupling disease incidence, disease rating over time and its related area under the disease progression curve (AUDPC) was established and used to screen the collections.
Conditions for Fusarium infection were optimised to ensure reproducible and efficient F. oxysporum infection. We evaluated the responses to the disease of two contrasting M. truncatula genotypes, and the effect of several cultural conditions known to affect the disease incidence, such as plant age at infection time, growth substrate and the method of inoculation. Results indicated that the A17 accession harbours a moderate level of resistance to the disease. We also showed that the method of inoculation strongly affected development of fusarium wilt disease in this model species, whereas it was not significantly altered by plant age or the inorganic growth substrate tested. In addition, we describe a rapid change in leaf temperature after infection, which can be used as an indirect parameter to confirm fungal infection at a very early stage of the interaction. As selection of resistant accessions by assessing symptoms at timely intervals is highly time-consuming we tested the potential of an infra-red imaging system to speed up this process. For this, we monitored the changes in surface leaf temperature upon infection by F. oxysporum f. sp. pisi in several pea accessions with contrasting response to Fusarium wilt under a controlled environment. Using a portable infra-red imaging system we detected a significant temperature increase of at least 0.5 °C after 10 days post-inoculation in the susceptible accessions, while the resistant accession temperature remained at control level. The increase in leaf temperature at 10 days post-inoculation was positively correlated with the AUDPC calculated over a 30 days period. Thus, this approach allowed the early discrimination between resistant and susceptible accessions.
A set of 267 accessions of M. truncatula were screened against F. oxysporum f. sp. medicaginis, 26 of them were identified as resistant, 9 as susceptible, and all other accessions as partially resistant. The phenotype of 12 resistant accessions was confirmed in two independent experiments on a subset of 23 accessions. Quantification of F. oxysporum f. sp. medicaginis within plant tissue indicated that the resistance level of the accessions is correlated with the amount of Fusarium within its shoot. Inoculation with a different strain of the same fungus produced a similar response, indicating that the resistance phenotype was stable.
For pea, a Pisum spp. germplasm collection was screened against F. o. pisi race 2. A large variation in disease response, ranging from highly resistant to susceptible, was observed, indicating the quantitative expression of the resistance. The repetition of the inoculation experiments on a subset of 19 pea accessions, including two susceptible accessions, indicated that the scoring method was robust and reproducible and confirmed the resistant phenotypes of accessions P23, P42, P614, P615, P627, P632, P633, P638, P639, P656, P669, P650 and JI1760.
Twenty pea genotypes with different levels of resistance to F. oxysporum f. sp. pisi were tested for their response to F. oxysporum, drought and combined stresses using the conditions previously defined. A large variation across genotypes was detected in their response to both stresses. The results indicated that F. oxysporum resistance and drought tolerance are not genetically linked. Although most accessions susceptible to the disease were also susceptible to drought, a few of them, including P650, were tolerant to drought. Most accessions resistant to F. oxysporum were susceptible to drought but, interestingly, we identified one accession, JI1412 that combined Fusarium resistance and drought tolerance.
Generate segregating populations by crossing elite pea cultivars with sources of multiple resistances for field testing
According to the work programme, selected resistant/tolerant pea accessions and mutants would be crossed with elite pea cultivar to introgress the resistance/tolerance genes into agronomically interesting pea material. Since mutants were not ready in time, crosses were made between elite cultivars provided by the various SMEs involved in the project and resistant accessions. The resulting F2s and the F3s populations were studied at Cordoba by CSIC and/or Sumperk by Agritec. These populations (List provided in Table 13) were evaluated and further selections were made according to their breeding priorities. In addition, once the project-identified mutants became available, they were also crossed and segregating populations are being produced that will be further studied in coming years.
Multiplication for field testing
Project-generated mutants were multiplied at INRA-Dijon and provided to partners for field testing. This required several cycles of selfing, selection and multiplication. The mutants (C*(C*1556.1.1.4).1 C*(C*1244.1.4).3 and their wild types could be multiplied in sufficient amounts for sharing with SMEs for field testing and other partners for lab studies. Other mutants are still being multiplied.
In addition, a pea collection for ECO-TILLING studies provided by Aberystwyth University was multiplied at Córdoba during 2014/2015 to be later used for validations. This included accessions JI-45, JI-157, JI-186, JI-189, JI-196, JI-252, JI-281, JI-399, JI-707, JI-1101, JI-1105, JI-1107, JI-1109, JI-1113, JI-1116, JI-1117, JI-1118, JI-1124, JI-1478, JI-1543, JI-2022, JI-2314, JI-2360, JI-2546, JI-2603, JI-2604, JI-2606, JI-2822, JI-3003, JI-3022, JI-3150, JI-3151, JI-3217, JI-3255 and JI-3256 that were multiplied and seeds harvested and stored being available for use by partners outside the life span of the project.
Molecular and metabolic markers from HUB genes
As described earlier, 11 of the 22 pea hub gene orthologues were found to be differentially expressed synergistically under double stress conditions (Fusarium and drought) compared with their expression in healthy and single-stressed situations.
Also, the metabolomics analysis conducted in the project revealed a list of metabolites that show a similar response in both Medicago and pea plants subjected to dual stress and distinct to the responses observed under control or single stress conditions. The final list of these metabolic markers is shown in Table 14.
In addition, genetic markers were sought for nine drought resistance genes previously identified in Arabidopsis: AT4G32940.1 (GAMMA-VPE: gamma vacuolar processing enzyme), AT3G10985 (SAG20: senescence associated gene 20), AT1G68590.1 (Ribosomal protein PSRP-3/Ycf65), AT5G48220.1 (Aldolase-type TIM barrel family protein), AT2G31380.1 (STH: salt tolerance homologue), AT1G77180.1 (SKIP: chromatin protein family), AT3G24570.1 (Peroxisomal membrane 22 kDa (Mpv17/PMP22) family protein), AT3G43230.1 (RING/FYVE/PHD-type zinc finger family protein) and AT2G19830 (SNF7 family protein).These genes were mapped in our RIL pea population P665 x Messire. We searched for longer sequences in the pea transcriptome and then designed specific primers to amplify these genes and looked for polymorphisms in two pea accessions with contrasting responses to drought: P665 (tolerant to drought) and cv. Messire (moderately susceptible to drought). All nine genes were successfully amplified and sequenced and SNPs were identified for eight of them. Direct polymorphism was found for one of the genes (AT2G31380 homolog) and CAPs or dCAP markers were successfully developed for five of these genes (AT4G32940, AT2G19830, AT3G24570, AT5G48220 and AT1G77180 pea homologs). For AT3G10985 pea homolog gene no polymorphism was identified in the region amplified, while for AT3G43230 and AT1G68590 homologs dCAP were designed but they did not yield clear polymorphisms.
In conclusion, we suggest that a targeted analysis of all 11 metabolites and expression of the 11 pea orthologue genes may constitute a useful biomarker system for practical application in pea breeding programs in order to obtain elite cultivars showing resistance to Fusarium and drought combined stress. In addition, the molecular markers developed from the Arabidopsis resistance genes will enable the analysis of six additional pea hub genes for tolerance to drought. Polymorphisms for these genes were identified in two pea lines with contrasting responses to drought (P664 and Messire). These polymorphisms are currently being analysed in the RIL population P665 x Messire. QTLs (quantitative trait loci) for tolerance to drought have been previously identified in this population. After mapping these genes their potential location in these QTLs will be checked to assess whether they might be the genes conferring the drought tolerance associated with these chromosomal locations.
FIELD TEST EVALUATION AGAINST COMMERCIAL VARIETIES
Preparatory field studies under drought conditions started from first project year in Czech Republic (AGRITEC), Spain (AGROVEGETAL) and UK (PGRO) comparing the performance of elite pea materials. Protocols for evaluating the resistance to a combined stress (drought and Fusarium oxysporum) were designed for field conditions in 2012 to allow partners to get the facilities ready to test the mutant varieties once seeds were provided in sufficient quantity for field trials. Plots were established in December in Spain by AGROVEGETAL and in spring by the other partners, adjusting to their different agro-ecological conditions.
Prospective work was done by the three partners to identify a field plot heavily and uniformly infected with Fusarium oxysporum f.sp. pisi in which to run the field tests following years. For instance, exploratory work at AGRITEC showed that plots infested with Phoma, Rhizoctonia, Phytium, Aphanomyces and Fusarium solani, but less to F. oxysporum. Similarly, PGRO found that F. solani was dominant, but F. oxysporum was not present. Artificial inoculation with a local isolate was therefore needed for final testing.
A common set of 25 pea cultivars and a common protocol for field testing was agreed among partners AGROVEGETAL (Spain), AGRITEC (Czech Republic), and PGRO (UK). Each partner added 5 local cultivars at checks. These cultivars were field tested during 2014, 2015 and 2016 by AGRITEC and AGROVEGETAL, and during 2014 by PGRO. Agronomy and phenological observations were carried out during the growing season (date of sowing, germination, early flowering, end of flowering, lodging, maturity date, date of harvest, yield, TSW, moisture in harvested seeds. The occurrence of diseases, pests and weeds were evaluated by conventional methods. Results were shared among participants and are being considered in order to optimise the conditions in preparation for testing the project mutants when available.
In addition to field testing, as part of the baseline analysis with the 25 elite cultivars, their responses to infection with F. oxysporum f. sp. pisi races 1(FOX 1) and 2 (FOX 2) and F. solani (FS) were studied in laboratory conditions by AGRITEC. A polytunnel trial was also conducted on the provocation field, where pea has been repeatedly grown for forty years, what ensures the conditions for combined stress of resistance to drought and Fusarium. Assessments included emergence, beginning and end of flowering, maturity and agronomic characters such as state of vegetation after emergence, lodging during vegetation period and before harvest, occurrence of diseases and insects and yield. Fusarium infection was assessed by pulling off some plants and observing their roots. Based on the results achieved, Eso, Abarth, Cartouche, Andalusí, Toro and AGT 213.13 were selected for further field testing during 2017.
Agronomic evaluation of project mutants
Only by the end of the project, sufficient quantities of seeds for the mutant lines were made available. PGRO, AGRITEC and AGROVEGETAL received 570 seeds each of mutants (C*(C*1556.1.1.4).1 and C*(C*1244.1.4).3 and their WT lines coming from INRA. These were planted in three replicates by AGROVEGETAL in a field plot at Ecija, Sevilla in December 2016 together with control cultivars. This site was identified as infested by Fusarium oxysporum f.sp. pisi. and reserved for these trials.
AGRITEC will perfom these studies during spring 2017 on their own resources, including the following lines:
1. Abarth
2. Cartouche
3. ANDALUSÍ
4. ESO
5. C*(C*1556.1.1.4).1 line WT
6. C*(C*1556.1.1.4).1 line mutant
7. C*(C*1244.1.4).3 line WT
8. C*(C*1244.1.4).3 line mutant
In preparation for this, AGRITEC have secured a suitable poly tunnel space that can be used for the required voltage dual (i) biotic (F. oxysporum) and (ii) abiotic (drought). A Fusarium inoculum was grown in large quantities over a period of 6 weeks, after which it was evenly distributed and mechanically incorporated in the soil surface region corresponding to the poly tunnel. Sacrificial peas were planted in infected areas to provide host and pathogen levels increase. After two months, the soil was tested for the presence and level of F. oxysporum. Level was found to be very high and the surface is ready for the planned work.
Identification of potential hub gene orthologues in other legumes (chickpea, lentil, alfalfa) and non-legume (tomato) crops
Results from the Medicago and pea analysis led to the identification of 11 hub genes in pea. To identify the orthologues of these hub genes in chickpea, lentil, faba bean, alfalfa and in the non-legume species tomato, their respective genomic and transcriptomic databases have been mined (Table 15).
In chickpea, this approach identified the orthologue of each pea hub gene (Table 17). For lentil and alfalfa, one orthologue was identified for 10 out of the 11 pea hub genes identified. Search of the transcriptomic databases of faba bean identified 9 out of 11 pea hub genes illustrating the lower coverage of faba bean genome of these databases. By contrast, tomato orthologues were identified for only 6 out of 11 pea hub genes (Table 17) success of the identification being limited by the lower level of conservation and synteny between tomato and pea or M. truncatula.
To confirm that the sequence identified in each targeted legume and non-legume species corresponded to the true orthologue of pea, genetic and physical maps of M. truncatula, pea, chickpea, alfalfa and tomato were compared. In all cases, the identified orthologues was found in a conserved region with M. truncatula and pea genomes. The high level of conservation both in terms of location and order within the genome sequence and the overall homology of the sequences suggest that the respective orthologues in chickpea, lentil, faba bean, alfalfa and tomato might also be involved in the plant response to F. oxysporum and/or drought. As such they might be good candidate to improve resistance to Fusarium wilt and/or tolerance to drought in these species.
To further confirm the involvement of the hub gene orthologues of chickpea and lentil, one experiment was initiated at CSIC to monitor the expression of these genes in response to F. oxysporum and drought applied alone or in combination. Given the long delay incurred, this experiment could not be completed within the time frame of the project. However, it will be finished within the coming month. Currently, RNA samples are been generated from roots of two contrasting genotypes per species collected at two time-point after the application of the stress. These RNA samples will be used to assess the expression of the orthologues by Real-time PCR. In parallel, the water relative content of the soil and the fusarium wilt and/or drought symptoms have been evaluated to correlate gene expression level with stress development.
Evaluation of hub genes against other biotic stresses
Three lines of pea carrying independent mutations in the Gamma-VPE gene (drought resistance hub gene in Arabidopsis) have been tested against various biotic stresses including F. oxysporum, the causal agent of Fusarium wilt, Erysiphe pisi, the causal agent of the powdery mildew disease and Uromyces pisi responsible for pea rust. These are amongst the most damaging diseases of pea worldwide. Plants were artificially inoculated under controlled conditions and macroscopically evaluated after an incubation period. The three mutant lines were as susceptible as the control lines to all three pathogens. It is not therefore possible to confirm an involvement of the pea Gamma-VPE gene in the resistance against these biotic stresses.
Potential Impact:
Impact: Instilling resilience in important food crops
Projections of rising temperatures and changing precipitation patterns across Europe indicate that extended periods of drought will become more frequent, even in areas where drought is currently a rare occurrence. Climate change may also affect delicate ecosystem balances allowing new pests and diseases such as insects, fungi, bacteria and viruses to invade and proliferate in hitherto unpopulated areas.
ABSTRESS was a five-year project designed to revolutionise the development of new food crop varieties and so more rapidly increase resilience to stresses related to climate change. ABSTRESS applied combined, integrated systems biology and comparative genomics approaches to conduct a comprehensive study of the gene networks implicated in the interaction between drought stress and Fusarium infection in legumes and symbiotic organisms.
Legumes leave a low carbon footprint as they require no nitrogen fertilisers – they use symbiotic microorganisms to take nitrogen from the air and fix it in the soil close to their roots. Once a legume crop has been harvested, any other crop following will therefore have a reduced need for additional nitrogen fertiliser. A further factor is that industrial nitrogen fertiliser produces the potent green house gas nitrous oxide as it degrades in soil. This is in addition to the carbon dioxide produced from the fossil fuels used to manufacture nitrogen fertilisers. Growing legumes therefore has a positive environmental impact over many other crop plants. Legumes also provide high protein food and feed efficiently using natural nitrogen resources to make essential nutrients. However, crop legumes are susceptible to a number of stresses that may be amplified by climate change.
To improve the breeding of sustainable crop varieties that are better able to tolerate the diverse effects of climate change, ABSTRESS have selected a model plant for testing and analysis. Medicago truncatula, a small Mediterranean clover has a small and accessible genome, is very quick to propagate and has a short life cycle. As with many other legumes, it is also susceptible to a combination of biotic and abiotic stresses.
In this research, Medicago plants were subjected to drought conditions and infection by the soil fungus Fusarium oxysporum which infects roots and limits their development. The plants were monitored throughout the study using high throughput imaging technology supplemented with daily weighing and observations. Critically we were monitoring which genes are switched on when the plants are exposed to stress and establishing how these changes allow the plants to cope during the early stages of stress. Once patterns of response to drought and infection were established, changes were made to the genetic information of peas to create plants from which new varieties will be bred. Following the completion of ABSTRESS the new varieties of peas will be assessed against the current commercial varieties under stress conditions in field trials across Europe.
The project has rapidly delivered new material to plant breeders (who are SME partners) for further refinement and ultimately commercialisation of new crop varieties. We have demonstrated the advantages of applying advanced phenotyping methods to rapidly generate improved varieties of an important commercial crop,
ABSTRESS impact aligns with the European Strategic Research Agenda 2025. The European Strategic Research Agenda 2025 has been published by the European Technology Platform, “Plants for the Future”. This SRA provides a road map for guiding and integrating European plant research over the next 20 years. The key challenge will be to meet local needs for food in terms of both quantity and quality while conserving natural resources and biodiversity. The SRA identifies 5 specific challenges to address through research; healthy, safe and sufficient food and feed; plant derived chemicals and energy; sustainable agriculture; a vibrant and competitive research agenda; and consumer choice and governance. ABSTRESS very much focussed on achieving a step change in “sustainability in agriculture” by undertaking breeding research that sought to develop varieties having improved resistance to a combination of biotic and abiotic stresses. The project has contributed to the “sufficient food” and “vibrant research” objectives particularly through stakeholder activities. European political goals for producing 20% of transport fuels from bio-based processes by 2020, and generating 30% of raw materials for chemical manufacture from plants by 2030, seek to transfer demand for fossil fuel resource to a sustainable source. However equally important is to reduce chemical inputs in agriculture which a greater use of legume crops, resulting from this project, will facilitate. Achieving greater tolerance to biotic and abiotic stress for legumes, developed from this research project, will therefore have a substantial impact towards achieving these political goals.
The SRA recognises the risks posed by volatility of agricultural systems and vulnerability to uncontrollable climatic conditions. Equally, in spite of increased urbanisation, in order to achieve global food security, Europe must maintain and increase sustainable production. We have addressed this by applying systems biology to enhance plant breeding for improved varieties. The development of potentially more efficient new plant genotypes, particularly new varieties that use less water and have increased tolerance to biotic and other abiotic stresses, has been achieved by ABSTRESS and this will be confirmed with commercial trials outside of the project.
ABSTRESS results have supported the development of breeding and adaptation strategies to maintain and/or optimise yields under varying growth conditions and have provided:
• Focussed academic input into generation, identification and understanding new genetic materials.
• Input from SME’s who fully understand the commercial requirements for the development of a successful new crop variety.
• SME’s provided with new varieties to test in Northern, Eastern and South Western Europe so that a range of growing conditions may be experienced.
• The development of high throughput molecular phenotyping protocols, a key technology for tomorrow.
Current approaches to plant phenotyping lag behind those used in medical research where molecular phenotype is incorporated as a key assessment criterion for identifying therapeutic targets. Plant phenotyping is largely based on visual assessment of performance characteristics, leading to de facto evidence gathering. ABSTRESS has developed advance approaches to plant molecular phenotyping linking with high throughput phenotyping to modernise the current approach to plant breeding.
The benefits realised of a molecular approach to plant phenotyping include:
• Earlier assessment of breeding lines that both speed up the breeding cycle and enable more accurate endpoint predictions through computational modelling.
• Integration of evidence based predictive modelling tools within the plant breeding process, reducing spend on unviable genotypes.
• An established assessment platform for the development of enhanced varieties based on the key processes that enable tolerance to biotic and abiotic stressors.
• Improved integration of genomic data with that from other platforms (metabolomics, transcriptomics) within the plant breeding industry to underpin the need to future proof the food supply chain against food security threats.
• Development of functional genomics approaches to assess the dynamic molecular phenotype in response to combined stressors through the triggering of metabolic processes.
• Reduced time and costs associated with plant breeding through the broad implementation of automated high throughput molecular phenotyping platforms.
The major beneficiaries of this project are likely to be plant breeders (many of which are SMEs) who will commercialise the novel germlasm developed in this project . The project has further developed the capacity of ABSTRESS SME partners in:
• Epigenetic research
• Research to improve the rhizosphere.
• Opening the European market for microbial inoculants to become the segment leader.
• Growing new varieties resistant to combined stress.
• The opportunity to develop experience of the germplasm development process and how to influence it.
The project executed a clear plan to ensure that appropriate account was taken of other international research. The advisory board was chosen to focus heavily on integrating the current research with previous and on-going research undertaken in the 7th Framework and internationally. The board included Prof. Douglas Cook (USA) and Prof Luis Aguirrezabal (Argentina) to ensure truelly global integration of the project. ABSTRESS was actively engaged with the BIOTECSOJASUR consortium in South America so that we could take account of research being undertaken on soybean in Mercosur countries to provide a global dimension. We formed relationships with many other EU and nationally funded projects. Of particular note were training and promotional activities undertaken through Comnet, cross discipline fertilation through TiMet and direct interactions with the International Legume Society through which ABSTRESS held its final conference in collaboration with projects such as Legato, Eurolegume, Reforma and Medileg.
The legacy of ABSTRESS will focus on the inclusion of the genetic resource developed in this project into elite pea breeding lines and demonstrate an effective path for improving multi stress tolerance in crop plants generally. ABSTRESS has developed novel crop germplasm that has high potential for improved resistance to stress and has established a transferable set of tools for application to a wide variety of plant species. The outputs generated by ABSTRESS will facilitate rapid changes in strategy as weather patterns change. This will ensure the sustainability of agriculture in Europe, and the ultimate intention is that the techniques will be transferable for breeding new varieties of crop plants, such as potatoes and tomatoes. ABSTRESS also disseminated knowledge and best practice techniques to stakeholders. Indeed, many of the projects activities have been publicised widely and key dissemination activities are described below.
Main Dissemination Activities
ABSTRESS disseminated its research as widely as possible during the course of the project and will continue to do so afterwards having due regard to protection of IPR. Dissemination has encompassed many plant communities and was not restricted to stakeholders in the Legume sector. Some key highlights are described below:
1. International workshop in Argentina
From the 10th to the 14th of March 2016, an international workshop was held in Argentina that brought together participants from the ABSTRESS and BIOTECSOJASUR projects.
BIOTECSOJASUR is a collaborative project between several research organisations and two companies from the MERCOSUR involving Argentina, Brazil, Paraguay and Uruguay (the common market involving these four South American countries).
(Photograph 1.Legume plant in flower)
The project is part of BIOTECSUR, a biotechnology platform in Mercosur, and focuses on disease and drought stresses in soybean crops. The project not only brings together researchers from different Mercosur countries for the first time but also develops new phenotyping platforms, creates common germplasm banks and new genomic tools. During the two day international workshop held at Unidad Integrada Balcarce, excellent presentations on all aspects of the ABSTRESS and BIOTECSOJASUR projects were given with a particular focus on new technologies. The Guest Speaker, Prof. Douglas Cook from University of California Davis, who is a member of the ABSTRESS Advisory Board, gave an inspiring presentation on his work with chickpeas where the natural variation in wild progenitors is studied in order to increase climate resilience in chickpea crops.
(Photograph 2. International workshop was held in Argentina)
On Saturday, we visited a beautiful Estancia in Pergamino where we were not only offered a delicious lunch but also had the opportunity to discuss growing soybean crops with Mr. Gustavo Breuer Moreno, a farmer, whilst visiting one of his soya bean fields.
(Photograph 3 & Photograph 4 Discussion about growing soybean crops with Mr. Gustavo Breuer Moreno, a farmer, whilst visiting one of his soya bean fields – Argentina)
The next day was spent at Asociacion Cooperativas Argentinas harbour where Mr. Juan Carlos Piotto, the director of the export harbour, gave us a tour of the impressive site. He explained the day to day operations of the plant and showed us the grain unloading facilities, the transport to the ships, the vast storage silos and buildings for grains and fertilisers, respectively. His company alone produces 3000 tons of soybean oil per day. We are very grateful for the fantastic hospitality we received at both the farm and the harbour plant.
(Photograph 5 & Photograph 6. Asociacion Cooperativas Argentinas harbour – learning about the operations of the plant including the grain unloading facilities, the transport to the ships, the vast storage silos and buildings)
On the 14th March a stakeholder meeting was organised at the Bolsa de Comercio of Rosario and Ignacio Ibanẽz from the Ministry of Science, Technology and Innovation gave an introductory presentation about biotechnology in Argentina. This was followed by presentations of Dr Martin Crespi, Dr Atilio Castagnaro, Dr Luis Aguirrezabal and Dr Adrian Charlton who explained the work involved in ABSTRESS and BIOTECSOJASUR. A discussion with stakeholders led to an interesting and fruitful exchange. Overall, the workshop and stakeholder meeting were a great success enabling future collaborations between EU and Mercosur researchers, strengthening relationships within the projects and giving us the opportunity to visit Argentina, a potential market for ABSTRESS deliverables.
2. Second International Legume Society Conference, Troia, Portugal 11th – 14th October 2016
The Second International Legume Society Conference took place on the 11th to the 14th October, 2016 at the Tróia resort, (in the vicinity of Lisbon). It was attended by over 350 delegates from 53 countries, with an agenda comprising of over 120 oral presentations in plenary and parallel sessions, with approximately 190 additional poster presentations. The final Abstress project meeting was held as a satellite event during the conference on October 11th.
Dr Adrian Charlton (Fera) chaired the first plenary session of the conference on the morning of 12th October entitled Legumes Value Chain: Market Requirements and Economic Impact. This session featured presentations including the economic future for legumes, local supply chain limitations, novel uses of the protein fraction from legume crops and the future of legumes from an EU CAP perspective.
(Photograph 7. Dr Adrian Charlton – Chairing the first plenary session at the European conference in Troia Portugal. ABSTRESS Co-cordinator, WP5 & WP6 Leader)
Dr Richard Thompson (INRA) chaired the second plenary session of the conference on the morning of 12th October entitled Legumes and Environment. This session discussed topics relating to methods to validate environmental benefits of legumes and data on how legumes could aid in sustainable new (or modified) agricultural practices.
Dr Carmen Bianco (IBBR) chaired a parallel session on the afternoon of 12th October entitled Beneficial Legume-microbe Interactions A. This session covered topics relating to signalling in legume symbiosis, beneficial microbe populations in the soil environment and thoughts to improve the adaptation of legume-rhizobium symbiosis. Carmen also presented a 5 minute “flash presentation” on the evening of the 12th October on her poster “The auxin indole-3-acetic acid (IAA) is more than a plant hormone.”
Dr Christophe Salon (INRA) chaired a parallel session on the afternoon of the 13th October entitled Frontiers in Plant and Crop Physiology. This session presented topics on in vivo monitoring of legume root developments and changes in the soybean transcriptome induced by abiotic stress. Prof. Phil Mullineaux (University of Essex) gave the keynote lecture in this session, presenting on his work within the Abstress project. The presentation was entitled “The identification of novel genes controlling plant-environment interactions” and primarily discussed the transcriptomics and HUB gene discovery work undertake in Abstress.
(Photograph 8. Prof. Phil Mullineaux (University of Essex) gave the keynote lecture at the European Conference. ABSTRESS WPL WP3)
Dr Adrian Charlton (Fera) presented in the plenary session entitled Frontiers in Legume Breeding on the morning of the 14th October. Dr Charlton presented an overview of the Abstress project, specifically the methodologies / technology platforms used within the project to better understand resistance to combined abiotic and biotic stress. The presentation was entitled “Improving the resistance of legume crops to combined abiotic and biotic stress”.
Mike Dickinson (Fera) and Prof. Diego Rubiales (CSIC) presented in the final parallel session on the afternoon of the 14th October. The session was entitled Resistance to Biotic and Abiotic Stresses. Mike presented on the metabolomics methodology and results within the Abstress project, his talk was entitled “Exploring metabolic changes in legumes exposed to combined biotic and abiotic stress.” Professor Rubiales presented an overview of his recent work on disease resistance in pea, discussing his work in cloning orthologs of resistance genes in pea as part of Abstress.
(Photograph 9. Mike Dickinson (Fera) presented Exploring metabolic changes in legumes exposed to combined biotic and abiotic stress)
Martin Rusilowicz (Fera and The University of York) presented a poster entitled “Interactive exploration of the Medicago stress response through time-series analysis in the MetaboClust software package” which described the statistical analysis of the metabolomics data within Abstress.
Dr Peter Kalo (NARIC) presented a poster entitled “Identification of the translocation breakpoint between chromosome 4 and 8 in the genomes of Medicago truncatula A17 and A20” which described some of the genetic mapping work undertaken in Abstress.
(Photograph 10. Conference Delegates in Portugal)
Exploitation of results
Website: The ASBTRESS website can be found at www.abstress.eu. This contains a summary of the project status and is updated regularly. It is also a repository for all of the main data arising from ABSTRESS and as such is one of the main exploitation vehicles used by the project consortium. Interest in the project has clearly grown through its 5 year life span as shown by the webstats below:
(Figure 1. Webstats for life of project)
Stakeholder database: The front page of the ABSTRESS website contains a link for interested partners to join as formal ABSTRESS affiliates. This registration was used to create the ABSTRESS stakeholder database. Fifty-six organisations registered a direct interest in ABSTRESS including multinational coroperations and smaller plant breeders. ABSTRESS affiliates were reached across the globe with particular interest in India.
Stakeholder meetings: ABSTRESS was particular successful at reaching interested parties through local stakeholder meetings. With a target at the outset of the project of 10 stakeholder meetings, 19 were held during the course of the project
Oral Presentations: ABSTRESS was presented at international conferences by the consortium a total of 58 times against a target of 5 at the project outset. It is anticipated that many more presentations acknowledging ABSTRESS as a source of data and funding will be given as further conclusions are extracted from the large complex datasets we hold.
Newsletters: Three newsletters were produced against a target of 5. The ABSTRESS management board decided that there was insufficient high-impact content to produce 5 newsletters and that 3 would have a greater impact.
Peer reviewed papers: 9 papers have currently been published with several more in the pipeline. The initial target was 10 papers and this will be achieved comfortably albeit beyond the funded lifetime of the project.
Trade articles: A further 11 articles appeared in trade journals directly referencing ABSTRESS.
Press coverage: ABSTRESS was referenced in the press at least 11 times including an interview for the BBC with the project coordinator.
List of Websites:
www.abstress.eu
Project Co-ordinator: Dr Adrian Charlton, Fera Sciecne Limited
Email: abstess@fera.co.uk
ABSTRESS has developed and implement protocols for the large scale production of Medicago truncatula (Mt) and pea subjected to combined drought and Fusarium oxysporum infection. Associated high throughput phenotyping data was captured from these plants as they were growing. This was the critical first step in ensuring that subsequent genetic and metabolite analyses were robust.
A successful protocol was developed and implemented for M. truncatula so that approximately 6000 plants have been generated under 4 different conditions being; control healthy plants (C), plants exposed to drought (D), Fusarium-infected plants (F) and plants both infected with Fusarium and exposed to drought (FD). Where applicable, plants were produced over a 15 day drought period following prior infection with F. oxysporum. Plant physiology was monitored during this period, with respiration rates and subsequent carbon assimilation measurements made to ensure that the relevant plants showed a typical stress response. The leaves and roots of these plants have been analysed using metabolomics and transcriptomics, and evaluated for other phenotypic properties such as root structure, nodulation and biomass. A full metabolomics dataset has been acquired using liquid chromatography, high-resolution mass spectrometry. This extensive dataset has been explored using multivariate analysis tools with patterns in the data indicating clear discrimination between stressed and unstressed plants. More detailed analysis has identified responses to both single and combined stressors and these observations relate to both anticipated stress responses and novel findings relating to specific perturbations in metabolic pathways.
Similarly, a large transcriptomics dataset has been collected using high throughput (next generation) sequencing. Bioinformatics models have been used to identify 36 hub genes in M. truncatula. Genetic maps for M. truncatula and pea have been used to identify 28 M. truncatula hub gene orthologs in pea. Some of these have been demonstrated to have a similar gene expression profile both in M. truncatula and in pea when exposed to combined stress. Tilling and eco-tilling for selected mutants of these 11 pea genes was conducted resulting in novel genetic material for testing in field trials, potentially offering genetic material for breeding resistance to combined stress. The identified hub genes have been shown to have homologues in a wide range of both legume and non-legume species providing the potential for the use of the data generated here to establish breeding programmes in, for example, tomato. ABSTRESS has therefore achieved its key aim of taking data from model plants into crops within approximately 4 years. However, to thoroughly test the success of the project and to provide SME partners with the value that they anticipated, field trials are required.
Project Context and Objectives:
Summary description of the project context and the main objectives
ABSTRESS applies combined, integrated systems biology and comparative genomics approaches to conduct a comprehensive study of the gene networks implicated in the interaction of drought stress and Fusarium infection in legumes using Medicago truncatula as a model for pea. The project demonstrates the advantages of applying advanced phenotyping methods for the generation of improved varieties of a commercial crop.
The Rationale of ABSTRESS
ABSTRESS was conceived based on three main drivers:
• Our belief that the combined application of comparative genomics and systems biology will deliver a rapid breeding solution to the poorly studied Fusarium - drought interaction in legumes.
• Our belief that a legume crop is essential for the future of European agriculture and our commitment to its improvement through a systems biology approach to plant breeding.
• Our desire to use ABSTRESS as an exemplar of how this technology can solve the Fusarium-Drought problem in non-legume crops as well as providing a generic methodology for resolving multi-component crop breeding challenges, such as the one described here.
Overarching objectives of ABSTRESS
The project had four major, tangible outputs to be delivered through the clear focus and unified approach of the project:
1. The identification of genes from Medicago and pea with a proven ability to alter the interaction between pea and Fusarium under drought conditions and thereby provide a major route to understanding the tripartite interaction of host, pathogen and drought stress.
2. The provision of alleles of genes identified in Output 1 which enhance the resistance of pea to Fusarium under drought conditions and do not negatively impact on crop performance in the absence of infection or drought. The pea genotypes that emerge from the project will be delivered to our SME plant breeding partners to be introgressed into elite breeding material using molecular breeding approaches to shorten the time taken to carry out this exercise.
3. The focus of this project has been the Fusarium – pea interaction under drought conditions, but this pathogen and abiotic stress are highly relevant to several other European crop species. Therefore, a third output was to identify the orthologs of the Medicago and pea genes involved in major crop genera of the Brassicaceae (e.g. oil seed rape), Poaceae (e.g. wheat), Solanaceae (e.g. potato), Compositeae (e.g. sunflower) and Chenopodiaceae (e.g. sugar beet), as well as other legume crops and demonstrate the validity of this approach in one non-legume crop species.
4. The development of molecular phenotyping tools using the latest Omics technologies to quantify gene expression (transcriptomics) and tolerance to combined stressors (metabolomics). Following the identification phase, simplified detection methodologies will be integrated within the workflows of the SME to enable high- throughput screening for advantageous phenotypes at the earliest possible stage in the breeding process. This will enable new varieties to be brought to market in a much shorter time than is currently feasible.
Context
The case for studying pea - an EU legume crop
The combining pea crop is well established as a valuable break crop in arable rotations in Europe. It offers the potential to reduce mineral nitrogen in the rotation due to bacterial nitrogen fixation. It also enables effective non-chemical control of problem grass weeds and provides a non-cereal break for soil borne pathogens such as take-all and pests such as nematodes. Peas have a zero requirement for soil nitrogen during the growing period and healthy crops produce a nitrogen residue immediately available to autumn planted cereals thereby reducing N requirement by between 25 and 55 kg/ha and reducing the diffuse pollution of more leachable forms of N in the early autumn. In the UK currently combining peas of high quality are mainly used as a food commodity, but in Europe, particularly France, they are used in animal feedstock as a high protein source to replace soya. With the increasing cost of imported soya, animal feed compounders would like to increase pea incorporation in feed and in some parts of Europe this demand is largely unfulfilled.
Biological nitrogen fixation is an extremely complex process which is very sensitive to environmental stresses (Zahran, 1999). Legumes are sensitive to abiotic stresses, most significantly water deficit and soil salinity. Drought is currently a major factor limiting crop productivity worldwide. In Mediterranean countries water deficit is encountered not only in arid and semiarid regions but also in areas where total precipitation is high but not evenly distributed during the growing season. In a context of increasing limitations to water use due to climate change and increased population, improving water use efficiency of crops is an important goal.
Rhizosphere microorganisms, particularly beneficial bacteria and fungi, can improve plant performance under stress environments and, consequently, enhance yield by providing plants with fixed nitrogen, phytohormones, iron and soluble phosphate, or by protecting the plant against soil-borne diseases mainly caused by pathogenic fungi. When consideration is given to the positive effects of plant growth promoting bacteria (PGPR) on plant growth under stress conditions, it is apparent that these strains can play an important role in developing strategies to control negative effects of biotic and abiotic stresses (Lugtemberg B. and Kamilova F., 2009).
However in what can be considered as an EU-wide market failure, investment in advanced technologies for legumes lags behind that of the cereals and other major crops such as potato and tomato. Therefore, for food security and economic reasons, it makes strategic sense for the EU to set about improving its own capacity to grow a major legume crop. Improved sustainability can be achieved by identifying and providing germplasm to the legume breeding community that can enable increased productivity with a reduced consumption of natural resources (e.g. reducing the need for pesticides, requiring mineral oil-based production, or scarce resources e.g. water). Advanced breeding techniques underpinned by modern post- genomic technologies promise to accelerate the production of novel germplasm for all crop species, but the means to do this for EU legumes lag behind crops such as the cereals.
The case for studying Fusarium and drought
Peas generally require temperate conditions and are suited to medium to light soil types. Drought during the flowering and pod filling period of spring varieties of combining peas can severely reduce yield. One of the climate changes predicted by forecasters is that summers will be drier in Europe, and one of the impacts of this climate change is likely to be an increase in drought affected spring crops.
An experiment carried out by partner PGRO demonstrated that drought during flowering and pod set reduced yield by around 55% under controlled conditions. However, within a range of 15 commercially available combining pea varieties, the effect of drought ranged from 60% reduction in some varieties to 30% in others. This indicates that there is some tolerance in existing varieties that could be exploited.
Fusarium wilts are widespread diseases caused by many forms of the soil-borne pathogen Fusarium oxysporum, affecting many agricultural crops, including most legumes, cucurbits, tomato, potato, pepper, strawberry, asparagus, cotton, banana etc. These soil borne pathogens can survive as thick-walled chlamydospores, which remain viable in the soil for many years. Control is problematic. It is very difficult to remove F. oxysporum when it becomes established, as it can grow saprophytically in the absence of a susceptible crop. The only effective response is soil sterilization, which is far too expensive for most farmers.
F. oxysporum f. sp. pisi (Fop) is an important and destructive pathogen of field pea, that has been reported in every country where pea is grown (Kraft and Pfleger, 2001). Successive cropping of peas can build up damaging populations of this pathogen. Crop rotation is the only means of maintaining safe levels of inoculum requiring the frequency of pea cropping in a field to be limited. In some regions, this is often no more than once in five years. Some control can be achieved with fungicides but the use of resistant cultivars of plants is the preferred approach (Lebeda et al., 2010).
Sources of resistance in peas are rather limited and difficult to estimate but single genes which have been identified and used in breeding, are rapidly overcome by new races of the pathogen (Infantino et al., 2006). The consortium has fully focussed on Fusarium because of its economic importance across a range of major EU crops. Knowledge on hub genes in M. truncatula and pea will be also of value to closely related legumes suffering from Fusarium wilt (i.e. common bean, cowpea, soybean, lentil, alfalfa, chickpea, etc) but also to non-legume crops (i.e. tomato).
Disease development is favoured by warm temperatures and dry weather, and is expected to be more severe and widespread with predicted climate change scenarios. The interactions between plants and pathogens involve a complex molecular dialogue. First, the pathogen is able to sense and recognize the plant host and to change its metabolism to provide favourable conditions for pathogenicity (Alfano and Collmer, 2004). Meanwhile, plants have evolved to identify structures associated with pathogens, leading to strengthening of existing defences and development of other powerful defence mechanisms. Successful plant infection by F. oxysporum is a complex phenomenon that requires a series of highly regulated processes (Di Pietro et al., 2001). The characteristic wilt symptoms appear as a result of severe water stress, mainly due to vessel clogging. Wilting is most likely caused by a combination of pathogen activities, such as the accumulation of fungal mycelium and/or toxin production and host defence responses, including production of gels, gums and xyloses and vessel crushing by proliferation of adjacent parenchyma cells (Beckman, 1987; Di Pietro et al., 2003). This mode of infection means that under warm and drought conditions the symptoms are more severe.
Response of plants to combined stresses
Whereas there is an extensive literature on the response of plants to single stresses under laboratory conditions, the study of plants’ responses to multiple stresses is still in its infancy (Mittler R & Blumwald E (2010)). The most information at the gene level comes from combined bioinformatics and experimental studies in which it is noted that cohorts of genes show responses to a diverse range of biotic and abiotic stresses suggesting, but not establishing, that such genes play roles in response to combined stress (Swindell WR et al. (2007)). Partner Essex identified the transcription factor HSFA1b as a major hub controlling resistance to biotrophic pathogens, increased tolerance to drought stress and improved seed yield under water-limiting conditions (Bechtold U et al. (2013)) and have also applied a Systems Biology approach to find hub genes implicated in controlling networks responsive to both abiotic and biotic stress. However, in all of these examples no attempt has been made at defining the response of the plants to multiple stresses and determining how it differs from application of the individual stresses. Where this has been done, the work reported has largely been descriptive, being confined to effects on symptoms , e.g. how changes in humidity, salinity or temperature affect resistance to pathogens (Yoshioka K et al (2001); Bechtold U et al (2005); AbuQamar S et al (2009)) or lists of genes from microarray experiments (Rizhsky L, et al (2004)).
ABSTRESS set out to conduct a rigorous temporal comparison between plants subjected to the two single and the combined stresses. Double stress-specific networks have emerged showing how the topology of these networks change compared with the situation under single stresses; identified hub genes underlying cellular mechanisms become targets for exploitation in our breeding programme.
The case for a Medicago truncatula-to-pea trait rapid exploitation pipeline.
Considerable effort has been expended on genome research on Medicago truncatula (hereafter called Medicago) in order to accelerate gene discovery for important agricultural traits. Medicago is highly susceptible to Fusarium infection combined with drought stress (Ramírez-Suero et al (2010)) and is therefore the model of choice for this project. Moreover, it has been demonstrated that Medicago and other legumes present several syntenic blocks, i.e. genomic regions with conserved sequences. Therefore, the identification of genes important in the Fusarium-drought interaction in Medicago can be used directly to identify the orthologous genes in pea.
ABSTRESS aimed at identifying hub genes involved in the control of resistance to combined biotic and abiotic stresses. Defined conditions were applied to Medicago first, and then the knowledge generated was transferred to pea in order to develop tools to accelerate breeding programmes.
Project Results:
Main results
ABSTRESS has combined integrated systems biology and comparative genomics approaches to conduct a comprehensive study of the gene networks implicated in the interaction of drought stress and Fusarium infection in legumes. The model legume Medicago truncatula has been used to rapidly identify characteristics for introgression into elite pea varieties.
The project is organised into 6 work packages (WPs). The first WP aimed to develop and implement protocols for the large scale provision of plant materials from Medicago truncatula and pea with associated high throughput phenotyping data. This work was critical to ensure that subsequent analysis was robust. Therefore much emphasis was placed on refining protocols for use in the high throughput phenotyping platform in Dijon (PPHD). A series of preliminary experiments (pre-pilots and pilots) with reduced numbers of plants were run with both Medicago truncatula and pea to establish optimal protocols for drought application and Fusarium infection. Multiple parameters had to be considered and tested, including pot size, levels of water, environmental humidity, length of treatment, Fusarium strains, infection protocols, etc. Various phenotypic and biological features were measured to monitor the response of the plants to the stress applied. After several iterations and much refinement of the methods, a successful protocol was developed and implemented for the large scale experiments, BOM (“big one” Medicago) and BOP (“big one” pea). Thousands of plants were generated under four different conditions: control healthy plants (C), drought-exposed plants (D), Fusarium infected plants (F) and plants both infected with Fusarium and exposed to drought (FD). Where applicable, plants were produced over a 15 day drought period following prior infection with Fusarium. Plant physiology was monitored during this period, with respiration rates and subsequent carbon assimilation measurements made to ensure that the relevant plants showed a typical stress response. Transcriptomics, epigenomics and metabolomics analyses were conducted to identify genes involved in stress response in the model plant Medicago. Comparative genomics was then employed to identify orthologue genes in pea, which were then investigated and mutants selected for introgression into elite pea cultivars in order to introduce characteristics associated with resistance to biotic and abiotic stress.
PHENOTYPIC ANALYSIS
Carbon and nitrogen acquisition, partitioning and allocation are tightly linked within plant parts as they both contribute to i) plant structures establishment (ie leaves, roots, nodules when plants interact with soil microbes, among them rhizobia and / or pathogens) and ii) their function for acquiring atmospheric and soil resources. Understanding how these two components (i.e. structure and function) interact when plants are faced with abiotic and biotic constraints remains a key issue towards selecting plants adapted to current agroecosystems. Both establishment and activity of the Rhizobium-legume symbiosis have been found to be extremely sensitive to drought stress. Several studies have shown that symbiotic nitrogen fixation is more sensitive to drought than other physiological parameters such as photosynthesis, transpiration, or nitrate assimilation.
Experiments were conducted to fully characterise plant status, focusing on shoots, roots and nodules. Plants were characterised dynamically throughout the experimental period, both destructively and non-destructively by image analysis. Non-destructive characterisation was undertaken with shoot and root imaging twice per week using the PPHD facilities at Dijon. Some plants for both the BOM and BOP were grown in parallel in rhizotubes in order to assess visually and non-destructively phenotypic changes across time. Images were acquired using the phenotyping cabin devoted to roots, which enabled the dynamically analysis of shoot, root and nodule growth on the same plants throughout the experiment.
Destructive measurements consisted of analysis of biomass partitioning and allocation, measurement of physiological traits (transpiration, stomatal conductance) and morphometric traits such as leaf area. A labelling experiment using 13C and 15N2 was further performed in a phytotron specially designed to measure precisely carbon and nitrogen entry and allocation to the various plant parts. Nodule gene expression studies were conducted by quantitative PCR. The nitrogen fixation genes nifA, nifH and fixJ of the symbiotic bacteria Sinorhizobium medicae and Rizhobium leguminosarum were analysed. Inhibition of N2-fixation in both Medicago and pea root nodules as a result of drought stress and Fusarium attack was analysed through the expression of structural genes of nitrogenase complex (represented by nifH expression) and on the positive activator of transcription nifA.
Elucidating plant strategy and nodulated root reaction to factorial abiotic and biotic stress: Biomass allocation to plant parts: Water stress mainly affected shoot compartments while carbon allocation to roots and its biomass were maintained. The situation observed was reversed with the addition of Fusarium infection, which might result from a strategy of the pathogen to “preserve” its host. While water stress negatively impacted nodule number in Medicago, mean nodule biomass was targeted in pea. Fusarium effect was principally detected on pea plants where nodule number was decreased while mean nodule weight increased. Using the precise labelling experiments, only the drought stress affected N and C flow within plants for both pea and Medicago.
C and N partitioning: In Medicago plants the stem mostly attracts carbon while pea leaves were the bigger carbon sink. Labelling experiment demonstrated that drought greatly decreased carbon incorporation of shoots, stems, and nodules for pea and Medicago while roots were less impacted, as observed for biomass. Fusarium infection did not impact carbon incorporation. N flow was greatly reduced for pea leaves, stems and nodules. Medicago seemed to be much less impacted for its entire compartment.
A differential strategy between pea and Medicago: In pea both leaf and nodule specific activity and biomass were severely decreased by drought stress and not by Fusarium infection. In contrast, in Medicago, only leaf specific activity was reduced by water stress while nodule specific activity was maintained.
Carbon and nitrogen uptake are tightly linked: a decrease of carbon acquisition directly impact nodule biomass (PEA) and or number (Medicago), which in turn reduces N flow through symbiotic nitrogen fixation. The relationship between C and N acquisition allows an estimate of the reactivity of plants facing a stress.
Characterising the expression levels of key bacterial genes: The application of singular or combined stresses lead to significant changes in the expression levels of bacterial nitrogen fixation genes over time. This suggests that potential hub genes to be found in Medicago plants have an effect on nif genes after the sixth day of treatment. Impact of single or double stress was similar for pea but down-regulation of N-fixation genes began earlier. The plants subjected to both stresses have a severe decay in the nitrogen fixation activity.
Towards Alternative Strategy to improve plant growth under abiotic and biotic stresses: Production of the auxin indole-3-acetic acid (IAA) is a common feature for plants with growth promoting rhizobacteria, having antagonistic activity against phytopathogenic microorganisms. The aim was to assess if the IAA overproduction in rhizobia had a positive effect on the response of plants to both pathogen attack and water stress.
Biotic stress resistance of IAA-overproducing rhizobia under free-living conditions: The bacterial strain S. medicae MD4, used to nodulate Medicago plant in Dijon, and its derivatives able to synthesize higher levels of IAA were tested for their antifungal activity against F. oxysporum sp. medicaginis through dual plate assay. The MD4 strains producing the highest IAA levels (MD4-65 and MD4-66) showed the highest antagonistic activity against Fusarium growth.
Effect of rhizobial IAA-overproduction on plant tolerance to drought stress and pathogen attack: Considering that the response mechanisms leading to salt- and drought tolerance are very similar, we investigated if IAA-overproducing rhizobia strains were able to improve the resistance of Medicago and pea to drought stress. For this study the modified strains producing the highest IAA levels have been selected. Preliminary results were obtained when Medicago and pea plants inoculated with the specific IAA-overproducing rhizobia were subjected to the double stress (drought stress and Fusarium attack). After stress treatment, visual signs of ion toxicity were present in the leaves of host plants. However, symptoms of senescence and zones of necrosis on the leaves of plants nodulated by the specific IAA-overproducing rhizobia were less evident as compared to the controls. The observed phenotype reflected changes in the expression levels of genes playing a key role in the mechanisms of drought response (the data are not shown here because they will be used for a patent application).
In conclusion, drought greatly decreased carbon incorporation in shoots, stems, and nodules while roots were less impacted. Only drought stress affected N and C flow within plants for both pea and Medicago; no clear trend could be distinguished in plants under pathogen attack. The labelling experiment also showed that N flow was greatly reduced by drought in pea while Medicago seemed to show much less of an impact. Water stress negatively impacted nodule number in Medicago while mean nodule biomass was targeted in pea. The Fusarium stress was principally detected in pea plants where nodule number was decreased while mean nodule weight increased. In pea both i) leaf specific activity and ii) nodule biomass and nodule specific activity were severely decreased by water stress, not by Fusarium. In Medicago, only leaf specific activity was reduced by water stress. This demonstrates differential plant-microorganism interactions between pea and Medicago. A tight carbon/nitrogen relationship allows estimating the degree of stress sensed by the plants and their efficiency to react when facing a stress. Drought stress led to significant down-regulation of genes involved in nitrogen fixation, an effect that was amplified by Fusarium attack. IAA production by rhizobacteria is probably one of the most important traits promoting plant growth under biotic stress conditions. The infection of three different legume plants with the specific IAA-overproducing rhizobia strains greatly improved the stress resistance of these plants.
MOLECULAR PHENOTYPING
METABOLOMICS
Pilot metabolomics experiments were undertaken in both Medicago truncatula and pea. Procedures were established and optimised for handling low sample amounts (< 10 mg), which was key for the subsequent full analysis, and for developing and validating analytical approaches using Liquid Chromatography – High Resolution Mass Spectrometry (LC-HRMS). Baseline metabolite profiles for both plants, leaves and roots, were established from a pooled reference material.
For the full BOM analysis, metabolites were studied in a non-targeted, holistic fashion for 288 Medicago samples (4 biological conditions x 3 biological replicates x 12 time points, shoots and nodulated roots analyzed separately). Data were assessed through a multi-faceted approach using Progenesis QI (NonLinear Dynamics), R (R Core team) and C# (Microsoft). Raw data files were uploaded into Progenesis QI for chromatographic alignment and feature detection. Outlier removal and QC (quality control) correction were applied using R before evaluation of the data using Pearson correlation and k-means clustering. Pearson correlation measures the correlation between two variables (metabolites) and is thus able to identify metabolites associated with particular patterns over time. Template patterns can be used to identify molecules whose abundances show correlation with time under the specified experimental conditions. Clustering methods can group together similar patterns in an unbiased manner. Interesting clusters of metabolites may then be investigated, for instance to identify common metabolic pathways or metabolite classes. k-means is a popular clustering method which starts with k randomly chosen cluster “centres”. Observations are assigned to the nearest centre and the centres are then moved to the average of all observations assigned to those centres. This iterative process is repeated until no further changes occur.
Metabolite identification confirmatory analyses were undertaken on an Orbitrap Velos Pro (Thermofisher Scientific) using the MS/MS and MSn capabilities of the instrument. Where a corresponding analytical standard was available, this was analysed concurrently to confirm compound identity. Where an analytical standard could not be obtained, compounds were “affirmatively” identified or dismissed using theoretical fragmentation from Mass Frontier software (HighChem Ltd) and / or library matching (if available) with the Metlin MS/MS database (Scripps Center for Metabolomics). Theoretical confirmation involved comparing accurate mass product ions (MS2 and MS3 ions) generated from a sample with a theoretical spectra produced by the software. Once confirmed, metabolites were tentatively assigned to pathways or a suggested biochemical impact using the open source databases KEGG (Kyoto Encyclopedia of Genes and Genomes, Kanehisa Laboratories, Japan.) and Mediccyc (http://mediccyc.noble.org/ The Samuel Roberts Noble Foundation, OK, USA). Thirty nine different metabolites were discovered to significantly change in abundance over time in leaves as a result of either one or both stresses. Twelve of them were identified with confidence, with the other 27 affirmatively identified as described above. Forty seven different metabolites were discovered to significantly change over time in root as a result of either one or both stresses. Fifteen of them were identified with confidence, with the other 32 affirmatively identified. Details of these metabolites are available on the Abstress website (Data for D2.1 D2.1 Metabolites identified).
TRANSCRIPTOMICS AND sRNAs
Long RNAs were extracted for 288 samples of Medicago truncatula (4 biological conditions x 3 biological replicates x 12 time points, shoots and nodulated roots), and analysed separately using GenXPro’s proprietary Massive Analysis of cDNA Ends (MACE). MACE libraries were generated for all 288 samples. The libraries were barcoded and sequenced at an average sequencing depth of 4.3 Million reads per library. Raw sequencing data were processed using GenXPro standard MACE workflow including cleaning of data from adaptor sequences, discarding of low-quality reads, and TrueQuant analysis to discard amplification artifacts. The MACE reads were mapped to the Medicago genome version 4.0 for transcriptome analysis. Unmapped reads were assembled into contigs which were consequently BLASTed against different publicly available data bases as e.g. SwissProt, Refseq, etc. providing functional assignment to the sequences.
Differential gene expression was evaluated using experiment design contrasts for pair-wise comparisons by means of tools including Deseq, EdgeR (R-framework), Cufflinks and also a similar library from the 'bioinfo' toolbox in the Matlab environment. All these tools use Negative Binomial Distribution in the statistical evaluation of pairwise comparisons to generate P-values for each tag in the dataset. Finally the obtained P-values were transformed to False Discovery Rates (FDR) using the Benjamini–Hochberg procedure.
Workflow used to process MACE transcriptomics data
The processing included quality control checking, taking account of biological variation, identifying differentially expressed genes (DEGs) across the time series and inputting the data from selected genes into VBSSM to generate 3 inferred gene regulatory network (GRN) models:
1. An inferred GRN in the leaves of Fusarium-infected/drought-stressed (FD) plants consisting of the top 100 DEGs in this dataset.
2. An inferred GRN in the roots of FD plants consisting of the top 100 DEGs in this dataset.
2. An inferred GRN consisting of all transcription factor (TF)-encoding DEGs expressed in the leaves of FD plants
The MDS tool was used for visual estimation of similarity between datasets. Sample replication is very important for establishing true events taking place in gene expression under different experimental conditions. Tissue samples from individual organisms tend to show high variance in distributed data among replicated samples. This occurs due to the natural variation of biological response between individuals. Additional factors associated with the quality of the RNA, sample preparation and library handling can also introduce significant artefacts that may lead to unwanted data dispersion. Therefore, we considered it prudent to examine data variability within sample groups prior to statistical analysis of the whole experiment. If individual replicates within one group exhibited significant dissimilarities and, in addition, were similar to those from a contrast group, they were removed. It was observed that the transcriptome data showed a dispersed distribution at almost all TPs (time points), which would be an issue for TPs represented by a single sample. Plots for all samples are available on the ABSTRESS website (Data for D2.4 D2.4 Datasets.zip).
Evaluation of differentially expressed genes
Tables showing the Differentially Expressed Genes (DGE, p-value threshold of <0,05) for drought vs control, Fusarium vs control and Fusarium-drought vs control conditions, for the kinetic analysis, both for shoot and root organs are available on the ABSTRESS website (Data for D2.1 file named “DGE genes”).The numbers of DEGS in each dataset are summarised in a Venn diagram is shown in Figure 1. The data reveal that a substantial portion of the DEGs in the FD datasets were unique to the double stress.
Small RNAs
To correlate the transcript regulations observed to small RNA populations, critical physiological time points of the kinetic were selected to identify differentially expressed sRNAs: day 1, 5 and 12, for single and combined stresses, shoots and nodulated roots, 3 biological replicates each. A protocol was set up to purify small RNA from Medicago leaves and roots. A bioinformatics workflow was established at IVS-CNRS to map RNA-seq data on the genome, and analyse the nature of small RNAs (miRNAs, siRNAs, nat-siRNAs and heterochromatin-derived siRNAs) as well as their organization on the genome (e.g. clusters, precursors). The corresponding 24 sRNA libraries were constructed and then sequenced. The data were processed with miR-Prefer and OmiRa software. The quality of the sequencing data was checked, the sRNAs were mapped to the Medicago truncatula genome (v4.0) and software to characterise the distribution between coding (ORFs) and promoter regions was developed. Analysis of miRNA and potential targets was also carried out using a combination of public degradome data and optimised plant software for target detection. The lists of differentially sRNA expressed (DSE) in the different stress conditions are uploaded on the ABSTRESS website (Data for D2.1 files named DSE-Root or DSE-Shoot). Overall, correct mapping frequencies of 24nt siRNAs (around 75%) and distribution of 8 to 10 M reads according to standard 21-24 nt ratios were obtained. Mapping of the sRNA revealed a general trend in all libraries: enrichment of 24nt siRNAs was obtained in 1kb promoter regions, and a significant fraction mapping also in introns and in un-annotated genomic regions. 54 promoters regions showed a differential expression of 24nt siRNAs depending on the stress condition, including 23 corresponding to genes with predicted functions. In leaf, 21/24 candidate promoters showed 24nt siRNA levels induced by drought (alone or combined) and in root, 24/31 candidates showed siRNA levels reduced by Fusarium (alone or combined). These differential patterns are listed in the file “D24-siRNA promoters” uploaded on the ABSTRESS website (Data for D2.1).
DNA METHYLATION
Genome-wide cytosine methylation analysis was conducted in DNA samples of Medicago leaf and root tissues exposed to combined stressors. These samples were from the same critical physiological time points used for sRNA profiling. For this purpose, MSDK libraries were obtained according to established protocols for high-throughput methylation-specific digital karyotyping improved by GenXPro’s proprietary SuperTag technology. In short, cytosine methylation was quantified at approx. 200,000 loci across the Medicago genome for each stress-situation using Illumina-sequencing. DNA was extracted from all root and leaf samples and was digested with the methylation-sensitive endonuclease HpaII. The flanking regions of the HpaII digestion-sites were sequenced using Illumina HiSeq2000 to assess differences in methylation at the level of the whole genome after quantification of the HpaII flanking sites. A total of 2.2 to 24 Million sequences were obtained per library, for a total of 24 libraries. The sequences obtained were subjected to bioinformatics processing which included quality assessment, de-multiplexing, mapping against the Medicago v.4 genome release, annotation, identification of “downstream” genes, and assessment of quantitative differential methylation. A robust correlation with genomic loci showing changes in methyl-cytosine content and the different level of expression of the transcripts (down regulation /activation) originating from the loci was carried out. A report on genome-wide quantitative methylation profiles containing the 1000 most likely differential methylated loci for the different stress/control comparisons can be found on the ABSTRESS website (Data for D2.1 file named “DME genes”). In addition, for the 54 promoter regions showing a differential expression of 24nt siRNAs depending on the stress condition, a correlation with differential methylation patterns was established (D24-siRNA promoters file), revealing several loci showing both changes in DNA methylation patterns and differential accumulations of 24nt siRNAs.
INFERRING GENE-INTERACTION NETWORKS FROM TIME SERIES EXPRESSION DATA
Hub genes are defined as the most highly connected genes in an inferred network and the working assumption is that such genes are more likely than others to be influential on whole plant processes, such as the response to combined Fusarium infection and drought stress being studied in ABSTRESS. Furthermore, by targeting hub genes for manipulation, the researcher can determine the accuracy of the model by seeing if proposed connections are disrupted when expression of a hub gene is perturbed. The data from such experiments can be used to develop later more refined draft models. Such re-iterative, unbiased approaches to model development underpin a Systems Biology approach in this context. However, the number of re-iterations to refine a gene regulatory network (GRN) is limited by the cost and time available. Nevertheless, the identification of hub genes by Bayesian Dynamic Modelling provides a mean of discovering novel genes that have a strong influence on both the GRN and the output phenotypes that the network influences.
We used VBSSM to infer gene regulation networks from the time series expression data. The VBSSM algorithm builds a probability model for each conditional state (time point) and scores the odds for fitting the expression data. One of the critical factors for robust evaluation of the models is the size of input population of genes. The VBSSM is effective in prediction of relatively small gene networks (max 100 genes). While this avoids over-fitting data to models, it does create a potential problem for choosing the ‘right’ population of gene candidates because the typical DEG analysis of RNA-SEQ data produces a much higher number of genes than VBSSM allows. We therefore modified the approach using the argument that the DEG analysis is most likely to give the best fit for a model assuming that potential hub genes are distributed among the top ranked genes. The ideal starting point for unbiased selection from our FD (Fusarium and drought combined treatments) DEG dataset would then be the top 100 genes.
To evaluate this assumption we generated an extra 20 networks for the same number of genes selected from randomly-permuted DEGs. The comparison of the Network Scores tests the hypothesis that hub genes are indeed amongst the top 100 DEGs and also challenges the stability of the inferred models for the same highly-connected genes. It should be noted that any inferred network can, in reality, reflect only a limited number of local connections. Therefore, the repeated identification of hub genes from random network modelling would reflect their involvement across a presumed wider gene regulatory network than could be computed and visualised as a single entity. It was these highly connected genes that appear as hubs in multiple models that were considered for further investigation. The full range of comparisons tested using VBSSM and the total number of selected genes provided as inputs over all modelling iterations are shown in Table 1.
Highly connected genes (HCGs) were classed as those which appeared in multiple models as hub genes, which in a single network were regarded as having the largest number of connections with neighbouring genes inferred from the DEG expression data. To evaluate HCGs from all networks, negative binomial statistics were used. A cumulative negative binomial function was used to obtain P values for distribution of connections for each gene in all networks and the genes with P < 0.05 were regarded as HCGs for the tested experimental condition. These are shown in Tables 2 and 3. The retrieved MACE transcripts corresponded to HCGs were verified using the BLASTX engine against the Medicago protein database.
Finally, we were able to temporally cluster by k-means clustering the FPM expression values for the HCGs from the leaf and root FD data into 16 and 12 groups respectively (Figures 2 and 3). This confirms that the modelling was picking out genes with different patterns of expression, making it likely that the changing expression of genes would impact on one another as would be the case if they were all part of an extended network. The inferred networks were returned in Simple Interaction Format (SIF) which can be visualised using, for example, Cytoscape freeware (http://www.cytoscape.org/). A series of interactions were inferred at different thresholds of probabilities (Z scores) and each .SIF file was marked accordingly. All files at all Z score thresholds that were set have been provided in folders for each VBSSM-generated GRN and are available on the ABSTRESS website project pages. The “VBSSM ready” data files with dummy replicates are available in their own folder on the ABSTRESS project website. A final selection of HCG hubs was made for further analysis (Tables 2 and 3, genes labelled in bold). In addition, analysis of the metabolite data showed that some of the metabolites unique to FD leaves or roots could be linked to the selected hub genes. This linkage is shown in Table 4.
GENETICS
Preliminary comparative genetic analysis of pea and Medicago was carried out in preparation for the orthologue identification exercise once the Medicago hub genes were defined. The pea genetic map was refined, alignment with Medicago genome verified and new reference markers were identified. Genotypes of some markers of F2 individuals from cross Mt (M. truncatula) A17 x Mt A20 displaying genetic linkage showed an unusual (cruciform) linkage pattern. Pollen viability test (~ 50 % lethality) suggested translocation. LG4 and LG8 were involved in the translocation. With the help of a cross between Mt 13U x Mt A20, markers were re-mapped and the breakpoint was delimited by genetic means. Mapping was performed with corresponding markers in diploid alfalfa (M. sativa) and pea. The results showed that M. truncatula reference accession A17 has an aberrant chromosomal configuration. The nucleotide sequence of the breakpoint was delimited at the breakpoint. In M. truncatula A17 genotype, the breakpoints separate the lower arm of LG4 from LG8 and vice versa. A17 stands alone because no other accession studied carries the same translocation in these chromosomal segments. The cross between M. truncatula A20 x 13U or any other crosses between M. truncatula accessions except A17 are suitable for mapping genes/markers in the translocation region. If M. truncatula A17 x A20 is needed for mapping the translocation point has to be taken into consideration.
Identification of the pea homologues of the M. truncatula hub genes
Based on their expression pattern in response to combined drought stress and Fusarium infection, 36 M. truncatula hub genes were selected as the most highly connected and differentially expressed. Twenty three hub genes were expressed predominantly in leaf and thirteen hub genes showed root specific transcription. In order to identify the pea homologs of the Medicago hub genes, two consecutive steps were applied; foremost sequences with high similarity to the Medicago genes were identified in the pea transcript and genomic databases. As a next step, the genomic locations of the identified pea sequences were determined and orthologous relationship was judged based on the genomic location and co-linearity between the Medicago and pea genomes. Sequences of pea genes were searched for at the Cool Season Food Legume Crop Database (https://www.coolseasonfoodlegume.org/) and at Genbank / EMBL databases with the provided 36 Medicago hub genes. In parallel, the Unigene set of pea was analysed to identify pea homolog sequences and obtain complete sequences of the pea genes. We could identify homolog sequences in the pea genome for 26 Medicago HUB genes with high sequence similarity and found sequences for 3 Medicago genes with lower similarity (Table 5). Unfortunately, no hit could be detected for 7 of the Medicago hub genes, most of which showed nodule-specific expression. This could be because the homologous genes in Medicago and pea either (1) do not share significant similarity although present in both genomes, (2) the genomic regions where they reside have not been sequenced in pea or (3) these genes have evolved since the divergence of the Medicago and pea lines.
Genetic mapping of pea hub gene candidates
As a next step, the genomic locations of the identified pea sequences were determined and orthologous relationship has been judged based on the genomic location and co-linearity between the Medicago and pea genomes. In order to increase the efficiency of the genetic mapping, an investigation aiming to resolve the anomaly of the genomic structure experienced in the reference genome of M. truncatula A17 genotype was conducted at the early phase of the project. Genetic mapping on the F2 population of the Medicago A17 x A20 accessions revealed a cruciform linkage between markers located on the lower arm of linkage group 4 and 8 (taking the diploid M. sativa genetic map as reference). We found that linear genetic linkage could be as observed between cross Mt A20 and Mt Jemalong plants. A former study suggested that the Mt A17 genome contains aberrant chromosomes which might cover a translocation between the affected chromosomes. In order to identify the translocation point in the Mt A17 genotype, published genomic sequence data was corrected and completed by de novo DNA sequencing. The translocation point was narrowed down by high resolution genetic mapping using generated markers from the improved genomic sequence using the Jemalong x A20 and A17 x A20 mapping populations. This analysis led to the discovery of a DNA segment where sister chromatid break and reunion occurred between the two chromosomes involved in the translocation event.
In order to prove the orthologous relationship between the Medicago HUB genes and their pea homologs, the genomic location of the pea homolog sequences were determined. Oligonucleotide primers were designed based on the pea homolog sequences and DNA fragments were amplified with PCR using the plants of JI281xJI399 RIL (Recombinant Inbred Lines) population. We determined the genomic location of the pea homologs with genetic mapping. Briefly, the genotypes of the RIL individuals were determined based on the detected polymorphism either in agarose gel or using SSCP analysis, genetic analyses were carried out and the co-linearity of genetic markers in the region was analysed.
In order to allow us to extend the genetic mapping of 8 pea homologs that did not show polymorphism between the parents JI281 and JI399 of the pea RIL population, we carried out an in silico genetic mapping project using a recently published WGS data of another pea genetic map which resolved map positions for 6 of the 8 genes. To do this, we constructed a composite genetic map containing pea core markers using the WGS data of the ‘Baccara x PI180693’ RIL mapping population. We analysed the sequences of the identified pea homologs of Medicago HUB genes for SNPs and based on the identified polymorphism we genotyped the selected plants of the RIL population. In summary using a combination of these two approaches, we identified the map location of pea homologs of 22 Medicago HUB genes which were positioned in co-linear genomic locations between Medicago and pea indicating their orthologous relationship (Table 6). We found 7 homolog sequences that were either mapped into non-syntenic genomic regions indicating that these genes are probably paralogs (homologs that developed as a result of gene-duplication events) or showed no polymorphism between the parental lines and therefore we were unable to map them.
Sequenced key genes for drought-Fusarium resistance Root and leaf hub genes responding to both drought and Fusarium infection were identified from transcriptomics and gene interaction network analyses. Sequences of the putative Pea orthologues were identified by using BLAST on the RNAseq-derived Pea Gene Expression Atlas. Confirmation of probable orthologues was established for many of the sequences by mapping and showing syntenic locations. The resulting set of sequenced genes, together with their predicted functions, is listed in Table 7.
EXPRESSION ANALYSIS OF HUB GENE ORTHOLOGUES IN PEA
In order to test the conservation of the Medicago inferred networks in pea, the expression of the pea orthologues of the Medicago hub genes was analysed. Peas were grown in symbiotic conditions and challenged with combined drought stress and Fusarium infection. Analysis was done by quantitative real time (q) PCR of cDNA from RNA prepared from leaf samples. There was not enough root material to test. The initiation of this task required the identification of the pea homolog sequences for transcribed sections of the Medicago hub genes. Twenty two (22) hub gene orthologues in pea were identified. Of these, 19 genes gave detectable and reproducible qPCR signals. Differential expression, with more than 2-fold change between the double stress treatments and controls was observed in 11 of these 19 genes. The panels in Figures 4A and 4B show the relative expression levels of the selected Medicago hub genes in pea grown in symbiotic conditions and challenged with combined drought stress and Fusarium infection. In our analysis, with 4 biological replicates, we used the confidence interval (CI) at 90% due to the marked variation within the replicates in some of the groups at certain time points. In most cases, the drought stress only and FD samples at 12 days post treatment showed similar expression profiles, suggesting that the main contribution to differential expression was the drought stress in the expression profile (Figures 4A/4B and Table 8). However, 6 of the genes analysed (PsCam006964, PsCam043610, PsCam002572, PsCam040229, PsCam039241 and PsCam058628) showed a synergistic response to the double stress, i.e. more or less than the sums of the fold increase in Fusarium or drought stress only.
Next, we compared the expression pattern of the 11 genes that showed a differential response to combined Fusarium infection and drought stress (FD) between the Medicago and pea over time. Heat map representations showed that both Medicago and pea adopt different expression states during the time course under double stress (FD) (Figure 5). In general, the expression of the hub genes peaked at 5 days after treatment (DAT) in Medicago, however, in pea the hub genes selected peaked at 12 DAT. This indicates that the combined stress effects were more rapid in Medicago than they were in pea. However, the general response of the pea hub gene orthologues was similar enough to indicate that the responses to the stresses in these two legume species were broadly qualitatively similar.
CHANGES IN ROOT AND LEAF METABOLOME OF PEA THAT CORRELATE WITH CHANGES IN MEDICAGO TRUNCATULA UNDER COMBINED DROUGHT AND FUSARIUM INFECTION
All metabolites found to significantly change in combined drought and Fusarium stressed Medicago plants in both leaf and roots are anticipated to be present in pea plants.
Table 9 shows the list of 22 pea orthologues of Medicago HUB genes, potential metabolic pathways affected and any associated metabolites found in the Medicago analyses that correlate with these pathways. Pathway information has been taken from Kegg (http://www.genome.jp/kegg/).
Pea material from the experiment named “Big One Pea” (BOP) was analysed in a non-targeted manner for metabolites by LC-HRMS following identical extraction and detection approaches as in the BOM.
Three of the 4 sample replicates for both pea leaf and root materials were analysed for days 2, 5, 8 and 12. All groups were analysed: control (C), drought only (D), Fusarium stress only (F) and combined stressors (FD).
The data of the BOP analysis corresponding to the metabolites discovered to significantly change over time for Medicago were selectively extracted and plotted in Excel and the overall pattern observed in order to ascertain if the metabolite was behaving in a similar manner in pea over time under dual stress. This pattern was either an increase or decrease in expression over time. For those metabolites that looked to agree in response, T-test statistics were employed on the final time point using the signal intensities of the control and dual stress groups.
From the 35 metabolites found to show a response to dual stress in Medicago leaves, 21 behaved in the same fashion in pea. Phenylalanine, arginine, lysine, proline, indoleacrylic acid and medicagenic acid glucuronide ester correlated in pea leaves up to day 8 time point only.
From the 41 metabolites found to show a response to dual stress in Medicago roots, 17 behaved in the same fashion in pea. Isopterofuran, proline, galactosyl-glycerol-3-phosphate, citrate, dehydroascorbate and dehydro-arabinono-1,4-lactone correlated in pea roots up to day 8 time point only.
Table 10 shows a list of metabolites that correlated in response to combined stressors between pea and Medicago. From this list, 11 metabolites were found to change uniquely in combined stress plants only.
TILLING and ECO-TILLING FOR PEA ORTHOLOGUE HUB GENES
Sequencing natural pea hub gene variants
In order to identify allelic variants for the HUB gene orthologues in different natural pea variants (eco-TILLING), primers were designed and synthesised for the 11 pea HUB gene orthologues that showed a response to combined drought and Fusarium infection in both Medicago and pea. Primer design was based on the sequences of the pea variety “Cameor”. To cover the coding sequences of these 11 genes, 21 primer pairs were designed and targeted amplicon sequencing (TAS) was carried out to determine allelic variants for the HUB gene orthologues in the selected pea genotypes. It was decided to access only the recently identified pea genotypes in groups 3.4 and 3.5 (Figure 6). This decision was based on a recent publication that showed that these groups represented the widest diversity of sequence variation in pea. The final list of JI accession number from these groups were as follows: Cameor (reference sequenced genome), JI157, JI186, JI189, JI252, JI281, JI399, JI707, JI1101, JI1105, JI1107, JI1109, JI1113, JI1116, JI1117, JI1118, J1124, JI1478, JI1543, JI2022, JI2360, JI2546, JI2603, JI2604, JI2606, JI2822, JI3003, JI3022, JI3150, JI3217, JI 3255, JI3256. The 31 accessions have been selected from a European Pisum germplasm collection to represent the maximum diversity.
Primers were designed to generate 24 amplicons in a total length of 12.5 kb to cover the CDSs of the 11 hub genes. The amplicons were prepared for Illumina next-generation sequencing using a Nextera XT kit on pooled amplicons from each accession, and sequenced on a MiSeq v2 platform across a single flowcell.
The sequencing generated all together ~35 M Illumina high quality reads (sequenced using paired-end technology) and about ~80% of them (~28 M reads) could be mapped to the 24 amplicon reference sequences of “Cameor” with the CLC genomic workbench 5.5.2 using its reference mapping function. Except for two amplicons, all the CDSs of the 11 HUB genes could be assembled for the 32 accessions although we found low coverage of the genes in a few cases.
The coding sequences of the HUB genes from the 31 pea accession were assembled based on the reference mapped contigs. DNA and amino acid sequence alignments were carried out with MEGA 7.0.21 and analysed with UPGMA hierarchical clustering method to compare the alleles of the 11 HUB genes. For two of the genes (PsCam002810 and PsCam006964) no polymorphism could be detected at amino acid level in the coding region.
Based on the detected polymorphisms at nucleotide and amino acid level, the alleles of the 31 accessions composed of two groups in the case of gene PsCam010311 and three groups in the case of PsCam034819. For other genes and proteins, the alleles were distributed into more than three clusters. In the case of the PsCam031330 gene early stop codons were detected in accessions 1543 and 3003, indicating that these alleles encode for truncated proteins. The positions of the amino acid substitutions and the number of each allele for each gene are summarised in Table 11.
The observations made in the limited time that the groups have had to analyse the data suggest no overall pattern. There was a range of amino acid substitutions, ranging from zero to ten. The one gene which stands out is PsCam031330, which codes for Heat Shock Transcription Factor (HSF). In accession JI1543 and JI3003 premature stop codons in the gene mean that a truncated form of HSF would be made, which was unlikely to be functional since this would terminate the protein within its essential DNA binding domain (DBD). The distance between the DBD and OD is less than 20 residues, which means PsCam031330 likely encodes for a class A HSF, its nearest match being HSFA1b from Arabidopsis. If so, this is consistent with a role for this gene in disease resistance and drought stress responses. The responses to both types of stress are controlled in part by HSFA1b in Arabidopsis.
In conclusion, there are pea accessions harbouring potential hub gene alleles which would be worthy of further investigation. The most notable of these is PsCam301330 coding for a HSF in which 2 alleles would be incapable of producing a functional protein. Further work is needed but post-ABSTRESS we will follow up this observation.
Primer design for TILLING mutants of pea hub gene orthologues
The primers for TILLING were designed and made for the 11 selected pea HUB gene orthologs chosen on the basis of a distinct response to combined drought and Fusarium infection. Primers were also designed for two newly identified hub genes (PsCam044570 and PsCam009624). All of these primer pairs were used to obtain pea mutants by TILLING the Pisum sativum var. Caméor population held at UMR Agroécologie. A summary of the mutants identified is shown in Table 12.
In addition, the previously identified drought resistance hub genes PsVPE (a putative vacuolar processing enzyme) and PsRP (a ribosomal protein) were also used for TILLING. Primers included (i) nested TILLING primers for identifying the mutant sequences, and (ii) dCAPS markers for following PsVPE in backrossing generations. Two of the mutants identified were multiplied in sufficient quantities to start field and lab testing (see below); the remaining ones are currently being multiplied.
PEA MATERIAL GENERATED FOR FIELD TESTING
Screening germplasm collections for accessions carrying combined resistance to both drought and Fusarium oxysporum
Collections of both pea and M. truncatula accessions were screened for resistance to F. oxysporum f. sp. pisi and f. sp. medicaginis, respectively, allowing the identification of resistant accessions. A detailed evaluation method coupling disease incidence, disease rating over time and its related area under the disease progression curve (AUDPC) was established and used to screen the collections.
Conditions for Fusarium infection were optimised to ensure reproducible and efficient F. oxysporum infection. We evaluated the responses to the disease of two contrasting M. truncatula genotypes, and the effect of several cultural conditions known to affect the disease incidence, such as plant age at infection time, growth substrate and the method of inoculation. Results indicated that the A17 accession harbours a moderate level of resistance to the disease. We also showed that the method of inoculation strongly affected development of fusarium wilt disease in this model species, whereas it was not significantly altered by plant age or the inorganic growth substrate tested. In addition, we describe a rapid change in leaf temperature after infection, which can be used as an indirect parameter to confirm fungal infection at a very early stage of the interaction. As selection of resistant accessions by assessing symptoms at timely intervals is highly time-consuming we tested the potential of an infra-red imaging system to speed up this process. For this, we monitored the changes in surface leaf temperature upon infection by F. oxysporum f. sp. pisi in several pea accessions with contrasting response to Fusarium wilt under a controlled environment. Using a portable infra-red imaging system we detected a significant temperature increase of at least 0.5 °C after 10 days post-inoculation in the susceptible accessions, while the resistant accession temperature remained at control level. The increase in leaf temperature at 10 days post-inoculation was positively correlated with the AUDPC calculated over a 30 days period. Thus, this approach allowed the early discrimination between resistant and susceptible accessions.
A set of 267 accessions of M. truncatula were screened against F. oxysporum f. sp. medicaginis, 26 of them were identified as resistant, 9 as susceptible, and all other accessions as partially resistant. The phenotype of 12 resistant accessions was confirmed in two independent experiments on a subset of 23 accessions. Quantification of F. oxysporum f. sp. medicaginis within plant tissue indicated that the resistance level of the accessions is correlated with the amount of Fusarium within its shoot. Inoculation with a different strain of the same fungus produced a similar response, indicating that the resistance phenotype was stable.
For pea, a Pisum spp. germplasm collection was screened against F. o. pisi race 2. A large variation in disease response, ranging from highly resistant to susceptible, was observed, indicating the quantitative expression of the resistance. The repetition of the inoculation experiments on a subset of 19 pea accessions, including two susceptible accessions, indicated that the scoring method was robust and reproducible and confirmed the resistant phenotypes of accessions P23, P42, P614, P615, P627, P632, P633, P638, P639, P656, P669, P650 and JI1760.
Twenty pea genotypes with different levels of resistance to F. oxysporum f. sp. pisi were tested for their response to F. oxysporum, drought and combined stresses using the conditions previously defined. A large variation across genotypes was detected in their response to both stresses. The results indicated that F. oxysporum resistance and drought tolerance are not genetically linked. Although most accessions susceptible to the disease were also susceptible to drought, a few of them, including P650, were tolerant to drought. Most accessions resistant to F. oxysporum were susceptible to drought but, interestingly, we identified one accession, JI1412 that combined Fusarium resistance and drought tolerance.
Generate segregating populations by crossing elite pea cultivars with sources of multiple resistances for field testing
According to the work programme, selected resistant/tolerant pea accessions and mutants would be crossed with elite pea cultivar to introgress the resistance/tolerance genes into agronomically interesting pea material. Since mutants were not ready in time, crosses were made between elite cultivars provided by the various SMEs involved in the project and resistant accessions. The resulting F2s and the F3s populations were studied at Cordoba by CSIC and/or Sumperk by Agritec. These populations (List provided in Table 13) were evaluated and further selections were made according to their breeding priorities. In addition, once the project-identified mutants became available, they were also crossed and segregating populations are being produced that will be further studied in coming years.
Multiplication for field testing
Project-generated mutants were multiplied at INRA-Dijon and provided to partners for field testing. This required several cycles of selfing, selection and multiplication. The mutants (C*(C*1556.1.1.4).1 C*(C*1244.1.4).3 and their wild types could be multiplied in sufficient amounts for sharing with SMEs for field testing and other partners for lab studies. Other mutants are still being multiplied.
In addition, a pea collection for ECO-TILLING studies provided by Aberystwyth University was multiplied at Córdoba during 2014/2015 to be later used for validations. This included accessions JI-45, JI-157, JI-186, JI-189, JI-196, JI-252, JI-281, JI-399, JI-707, JI-1101, JI-1105, JI-1107, JI-1109, JI-1113, JI-1116, JI-1117, JI-1118, JI-1124, JI-1478, JI-1543, JI-2022, JI-2314, JI-2360, JI-2546, JI-2603, JI-2604, JI-2606, JI-2822, JI-3003, JI-3022, JI-3150, JI-3151, JI-3217, JI-3255 and JI-3256 that were multiplied and seeds harvested and stored being available for use by partners outside the life span of the project.
Molecular and metabolic markers from HUB genes
As described earlier, 11 of the 22 pea hub gene orthologues were found to be differentially expressed synergistically under double stress conditions (Fusarium and drought) compared with their expression in healthy and single-stressed situations.
Also, the metabolomics analysis conducted in the project revealed a list of metabolites that show a similar response in both Medicago and pea plants subjected to dual stress and distinct to the responses observed under control or single stress conditions. The final list of these metabolic markers is shown in Table 14.
In addition, genetic markers were sought for nine drought resistance genes previously identified in Arabidopsis: AT4G32940.1 (GAMMA-VPE: gamma vacuolar processing enzyme), AT3G10985 (SAG20: senescence associated gene 20), AT1G68590.1 (Ribosomal protein PSRP-3/Ycf65), AT5G48220.1 (Aldolase-type TIM barrel family protein), AT2G31380.1 (STH: salt tolerance homologue), AT1G77180.1 (SKIP: chromatin protein family), AT3G24570.1 (Peroxisomal membrane 22 kDa (Mpv17/PMP22) family protein), AT3G43230.1 (RING/FYVE/PHD-type zinc finger family protein) and AT2G19830 (SNF7 family protein).These genes were mapped in our RIL pea population P665 x Messire. We searched for longer sequences in the pea transcriptome and then designed specific primers to amplify these genes and looked for polymorphisms in two pea accessions with contrasting responses to drought: P665 (tolerant to drought) and cv. Messire (moderately susceptible to drought). All nine genes were successfully amplified and sequenced and SNPs were identified for eight of them. Direct polymorphism was found for one of the genes (AT2G31380 homolog) and CAPs or dCAP markers were successfully developed for five of these genes (AT4G32940, AT2G19830, AT3G24570, AT5G48220 and AT1G77180 pea homologs). For AT3G10985 pea homolog gene no polymorphism was identified in the region amplified, while for AT3G43230 and AT1G68590 homologs dCAP were designed but they did not yield clear polymorphisms.
In conclusion, we suggest that a targeted analysis of all 11 metabolites and expression of the 11 pea orthologue genes may constitute a useful biomarker system for practical application in pea breeding programs in order to obtain elite cultivars showing resistance to Fusarium and drought combined stress. In addition, the molecular markers developed from the Arabidopsis resistance genes will enable the analysis of six additional pea hub genes for tolerance to drought. Polymorphisms for these genes were identified in two pea lines with contrasting responses to drought (P664 and Messire). These polymorphisms are currently being analysed in the RIL population P665 x Messire. QTLs (quantitative trait loci) for tolerance to drought have been previously identified in this population. After mapping these genes their potential location in these QTLs will be checked to assess whether they might be the genes conferring the drought tolerance associated with these chromosomal locations.
FIELD TEST EVALUATION AGAINST COMMERCIAL VARIETIES
Preparatory field studies under drought conditions started from first project year in Czech Republic (AGRITEC), Spain (AGROVEGETAL) and UK (PGRO) comparing the performance of elite pea materials. Protocols for evaluating the resistance to a combined stress (drought and Fusarium oxysporum) were designed for field conditions in 2012 to allow partners to get the facilities ready to test the mutant varieties once seeds were provided in sufficient quantity for field trials. Plots were established in December in Spain by AGROVEGETAL and in spring by the other partners, adjusting to their different agro-ecological conditions.
Prospective work was done by the three partners to identify a field plot heavily and uniformly infected with Fusarium oxysporum f.sp. pisi in which to run the field tests following years. For instance, exploratory work at AGRITEC showed that plots infested with Phoma, Rhizoctonia, Phytium, Aphanomyces and Fusarium solani, but less to F. oxysporum. Similarly, PGRO found that F. solani was dominant, but F. oxysporum was not present. Artificial inoculation with a local isolate was therefore needed for final testing.
A common set of 25 pea cultivars and a common protocol for field testing was agreed among partners AGROVEGETAL (Spain), AGRITEC (Czech Republic), and PGRO (UK). Each partner added 5 local cultivars at checks. These cultivars were field tested during 2014, 2015 and 2016 by AGRITEC and AGROVEGETAL, and during 2014 by PGRO. Agronomy and phenological observations were carried out during the growing season (date of sowing, germination, early flowering, end of flowering, lodging, maturity date, date of harvest, yield, TSW, moisture in harvested seeds. The occurrence of diseases, pests and weeds were evaluated by conventional methods. Results were shared among participants and are being considered in order to optimise the conditions in preparation for testing the project mutants when available.
In addition to field testing, as part of the baseline analysis with the 25 elite cultivars, their responses to infection with F. oxysporum f. sp. pisi races 1(FOX 1) and 2 (FOX 2) and F. solani (FS) were studied in laboratory conditions by AGRITEC. A polytunnel trial was also conducted on the provocation field, where pea has been repeatedly grown for forty years, what ensures the conditions for combined stress of resistance to drought and Fusarium. Assessments included emergence, beginning and end of flowering, maturity and agronomic characters such as state of vegetation after emergence, lodging during vegetation period and before harvest, occurrence of diseases and insects and yield. Fusarium infection was assessed by pulling off some plants and observing their roots. Based on the results achieved, Eso, Abarth, Cartouche, Andalusí, Toro and AGT 213.13 were selected for further field testing during 2017.
Agronomic evaluation of project mutants
Only by the end of the project, sufficient quantities of seeds for the mutant lines were made available. PGRO, AGRITEC and AGROVEGETAL received 570 seeds each of mutants (C*(C*1556.1.1.4).1 and C*(C*1244.1.4).3 and their WT lines coming from INRA. These were planted in three replicates by AGROVEGETAL in a field plot at Ecija, Sevilla in December 2016 together with control cultivars. This site was identified as infested by Fusarium oxysporum f.sp. pisi. and reserved for these trials.
AGRITEC will perfom these studies during spring 2017 on their own resources, including the following lines:
1. Abarth
2. Cartouche
3. ANDALUSÍ
4. ESO
5. C*(C*1556.1.1.4).1 line WT
6. C*(C*1556.1.1.4).1 line mutant
7. C*(C*1244.1.4).3 line WT
8. C*(C*1244.1.4).3 line mutant
In preparation for this, AGRITEC have secured a suitable poly tunnel space that can be used for the required voltage dual (i) biotic (F. oxysporum) and (ii) abiotic (drought). A Fusarium inoculum was grown in large quantities over a period of 6 weeks, after which it was evenly distributed and mechanically incorporated in the soil surface region corresponding to the poly tunnel. Sacrificial peas were planted in infected areas to provide host and pathogen levels increase. After two months, the soil was tested for the presence and level of F. oxysporum. Level was found to be very high and the surface is ready for the planned work.
Identification of potential hub gene orthologues in other legumes (chickpea, lentil, alfalfa) and non-legume (tomato) crops
Results from the Medicago and pea analysis led to the identification of 11 hub genes in pea. To identify the orthologues of these hub genes in chickpea, lentil, faba bean, alfalfa and in the non-legume species tomato, their respective genomic and transcriptomic databases have been mined (Table 15).
In chickpea, this approach identified the orthologue of each pea hub gene (Table 17). For lentil and alfalfa, one orthologue was identified for 10 out of the 11 pea hub genes identified. Search of the transcriptomic databases of faba bean identified 9 out of 11 pea hub genes illustrating the lower coverage of faba bean genome of these databases. By contrast, tomato orthologues were identified for only 6 out of 11 pea hub genes (Table 17) success of the identification being limited by the lower level of conservation and synteny between tomato and pea or M. truncatula.
To confirm that the sequence identified in each targeted legume and non-legume species corresponded to the true orthologue of pea, genetic and physical maps of M. truncatula, pea, chickpea, alfalfa and tomato were compared. In all cases, the identified orthologues was found in a conserved region with M. truncatula and pea genomes. The high level of conservation both in terms of location and order within the genome sequence and the overall homology of the sequences suggest that the respective orthologues in chickpea, lentil, faba bean, alfalfa and tomato might also be involved in the plant response to F. oxysporum and/or drought. As such they might be good candidate to improve resistance to Fusarium wilt and/or tolerance to drought in these species.
To further confirm the involvement of the hub gene orthologues of chickpea and lentil, one experiment was initiated at CSIC to monitor the expression of these genes in response to F. oxysporum and drought applied alone or in combination. Given the long delay incurred, this experiment could not be completed within the time frame of the project. However, it will be finished within the coming month. Currently, RNA samples are been generated from roots of two contrasting genotypes per species collected at two time-point after the application of the stress. These RNA samples will be used to assess the expression of the orthologues by Real-time PCR. In parallel, the water relative content of the soil and the fusarium wilt and/or drought symptoms have been evaluated to correlate gene expression level with stress development.
Evaluation of hub genes against other biotic stresses
Three lines of pea carrying independent mutations in the Gamma-VPE gene (drought resistance hub gene in Arabidopsis) have been tested against various biotic stresses including F. oxysporum, the causal agent of Fusarium wilt, Erysiphe pisi, the causal agent of the powdery mildew disease and Uromyces pisi responsible for pea rust. These are amongst the most damaging diseases of pea worldwide. Plants were artificially inoculated under controlled conditions and macroscopically evaluated after an incubation period. The three mutant lines were as susceptible as the control lines to all three pathogens. It is not therefore possible to confirm an involvement of the pea Gamma-VPE gene in the resistance against these biotic stresses.
Potential Impact:
Impact: Instilling resilience in important food crops
Projections of rising temperatures and changing precipitation patterns across Europe indicate that extended periods of drought will become more frequent, even in areas where drought is currently a rare occurrence. Climate change may also affect delicate ecosystem balances allowing new pests and diseases such as insects, fungi, bacteria and viruses to invade and proliferate in hitherto unpopulated areas.
ABSTRESS was a five-year project designed to revolutionise the development of new food crop varieties and so more rapidly increase resilience to stresses related to climate change. ABSTRESS applied combined, integrated systems biology and comparative genomics approaches to conduct a comprehensive study of the gene networks implicated in the interaction between drought stress and Fusarium infection in legumes and symbiotic organisms.
Legumes leave a low carbon footprint as they require no nitrogen fertilisers – they use symbiotic microorganisms to take nitrogen from the air and fix it in the soil close to their roots. Once a legume crop has been harvested, any other crop following will therefore have a reduced need for additional nitrogen fertiliser. A further factor is that industrial nitrogen fertiliser produces the potent green house gas nitrous oxide as it degrades in soil. This is in addition to the carbon dioxide produced from the fossil fuels used to manufacture nitrogen fertilisers. Growing legumes therefore has a positive environmental impact over many other crop plants. Legumes also provide high protein food and feed efficiently using natural nitrogen resources to make essential nutrients. However, crop legumes are susceptible to a number of stresses that may be amplified by climate change.
To improve the breeding of sustainable crop varieties that are better able to tolerate the diverse effects of climate change, ABSTRESS have selected a model plant for testing and analysis. Medicago truncatula, a small Mediterranean clover has a small and accessible genome, is very quick to propagate and has a short life cycle. As with many other legumes, it is also susceptible to a combination of biotic and abiotic stresses.
In this research, Medicago plants were subjected to drought conditions and infection by the soil fungus Fusarium oxysporum which infects roots and limits their development. The plants were monitored throughout the study using high throughput imaging technology supplemented with daily weighing and observations. Critically we were monitoring which genes are switched on when the plants are exposed to stress and establishing how these changes allow the plants to cope during the early stages of stress. Once patterns of response to drought and infection were established, changes were made to the genetic information of peas to create plants from which new varieties will be bred. Following the completion of ABSTRESS the new varieties of peas will be assessed against the current commercial varieties under stress conditions in field trials across Europe.
The project has rapidly delivered new material to plant breeders (who are SME partners) for further refinement and ultimately commercialisation of new crop varieties. We have demonstrated the advantages of applying advanced phenotyping methods to rapidly generate improved varieties of an important commercial crop,
ABSTRESS impact aligns with the European Strategic Research Agenda 2025. The European Strategic Research Agenda 2025 has been published by the European Technology Platform, “Plants for the Future”. This SRA provides a road map for guiding and integrating European plant research over the next 20 years. The key challenge will be to meet local needs for food in terms of both quantity and quality while conserving natural resources and biodiversity. The SRA identifies 5 specific challenges to address through research; healthy, safe and sufficient food and feed; plant derived chemicals and energy; sustainable agriculture; a vibrant and competitive research agenda; and consumer choice and governance. ABSTRESS very much focussed on achieving a step change in “sustainability in agriculture” by undertaking breeding research that sought to develop varieties having improved resistance to a combination of biotic and abiotic stresses. The project has contributed to the “sufficient food” and “vibrant research” objectives particularly through stakeholder activities. European political goals for producing 20% of transport fuels from bio-based processes by 2020, and generating 30% of raw materials for chemical manufacture from plants by 2030, seek to transfer demand for fossil fuel resource to a sustainable source. However equally important is to reduce chemical inputs in agriculture which a greater use of legume crops, resulting from this project, will facilitate. Achieving greater tolerance to biotic and abiotic stress for legumes, developed from this research project, will therefore have a substantial impact towards achieving these political goals.
The SRA recognises the risks posed by volatility of agricultural systems and vulnerability to uncontrollable climatic conditions. Equally, in spite of increased urbanisation, in order to achieve global food security, Europe must maintain and increase sustainable production. We have addressed this by applying systems biology to enhance plant breeding for improved varieties. The development of potentially more efficient new plant genotypes, particularly new varieties that use less water and have increased tolerance to biotic and other abiotic stresses, has been achieved by ABSTRESS and this will be confirmed with commercial trials outside of the project.
ABSTRESS results have supported the development of breeding and adaptation strategies to maintain and/or optimise yields under varying growth conditions and have provided:
• Focussed academic input into generation, identification and understanding new genetic materials.
• Input from SME’s who fully understand the commercial requirements for the development of a successful new crop variety.
• SME’s provided with new varieties to test in Northern, Eastern and South Western Europe so that a range of growing conditions may be experienced.
• The development of high throughput molecular phenotyping protocols, a key technology for tomorrow.
Current approaches to plant phenotyping lag behind those used in medical research where molecular phenotype is incorporated as a key assessment criterion for identifying therapeutic targets. Plant phenotyping is largely based on visual assessment of performance characteristics, leading to de facto evidence gathering. ABSTRESS has developed advance approaches to plant molecular phenotyping linking with high throughput phenotyping to modernise the current approach to plant breeding.
The benefits realised of a molecular approach to plant phenotyping include:
• Earlier assessment of breeding lines that both speed up the breeding cycle and enable more accurate endpoint predictions through computational modelling.
• Integration of evidence based predictive modelling tools within the plant breeding process, reducing spend on unviable genotypes.
• An established assessment platform for the development of enhanced varieties based on the key processes that enable tolerance to biotic and abiotic stressors.
• Improved integration of genomic data with that from other platforms (metabolomics, transcriptomics) within the plant breeding industry to underpin the need to future proof the food supply chain against food security threats.
• Development of functional genomics approaches to assess the dynamic molecular phenotype in response to combined stressors through the triggering of metabolic processes.
• Reduced time and costs associated with plant breeding through the broad implementation of automated high throughput molecular phenotyping platforms.
The major beneficiaries of this project are likely to be plant breeders (many of which are SMEs) who will commercialise the novel germlasm developed in this project . The project has further developed the capacity of ABSTRESS SME partners in:
• Epigenetic research
• Research to improve the rhizosphere.
• Opening the European market for microbial inoculants to become the segment leader.
• Growing new varieties resistant to combined stress.
• The opportunity to develop experience of the germplasm development process and how to influence it.
The project executed a clear plan to ensure that appropriate account was taken of other international research. The advisory board was chosen to focus heavily on integrating the current research with previous and on-going research undertaken in the 7th Framework and internationally. The board included Prof. Douglas Cook (USA) and Prof Luis Aguirrezabal (Argentina) to ensure truelly global integration of the project. ABSTRESS was actively engaged with the BIOTECSOJASUR consortium in South America so that we could take account of research being undertaken on soybean in Mercosur countries to provide a global dimension. We formed relationships with many other EU and nationally funded projects. Of particular note were training and promotional activities undertaken through Comnet, cross discipline fertilation through TiMet and direct interactions with the International Legume Society through which ABSTRESS held its final conference in collaboration with projects such as Legato, Eurolegume, Reforma and Medileg.
The legacy of ABSTRESS will focus on the inclusion of the genetic resource developed in this project into elite pea breeding lines and demonstrate an effective path for improving multi stress tolerance in crop plants generally. ABSTRESS has developed novel crop germplasm that has high potential for improved resistance to stress and has established a transferable set of tools for application to a wide variety of plant species. The outputs generated by ABSTRESS will facilitate rapid changes in strategy as weather patterns change. This will ensure the sustainability of agriculture in Europe, and the ultimate intention is that the techniques will be transferable for breeding new varieties of crop plants, such as potatoes and tomatoes. ABSTRESS also disseminated knowledge and best practice techniques to stakeholders. Indeed, many of the projects activities have been publicised widely and key dissemination activities are described below.
Main Dissemination Activities
ABSTRESS disseminated its research as widely as possible during the course of the project and will continue to do so afterwards having due regard to protection of IPR. Dissemination has encompassed many plant communities and was not restricted to stakeholders in the Legume sector. Some key highlights are described below:
1. International workshop in Argentina
From the 10th to the 14th of March 2016, an international workshop was held in Argentina that brought together participants from the ABSTRESS and BIOTECSOJASUR projects.
BIOTECSOJASUR is a collaborative project between several research organisations and two companies from the MERCOSUR involving Argentina, Brazil, Paraguay and Uruguay (the common market involving these four South American countries).
(Photograph 1.Legume plant in flower)
The project is part of BIOTECSUR, a biotechnology platform in Mercosur, and focuses on disease and drought stresses in soybean crops. The project not only brings together researchers from different Mercosur countries for the first time but also develops new phenotyping platforms, creates common germplasm banks and new genomic tools. During the two day international workshop held at Unidad Integrada Balcarce, excellent presentations on all aspects of the ABSTRESS and BIOTECSOJASUR projects were given with a particular focus on new technologies. The Guest Speaker, Prof. Douglas Cook from University of California Davis, who is a member of the ABSTRESS Advisory Board, gave an inspiring presentation on his work with chickpeas where the natural variation in wild progenitors is studied in order to increase climate resilience in chickpea crops.
(Photograph 2. International workshop was held in Argentina)
On Saturday, we visited a beautiful Estancia in Pergamino where we were not only offered a delicious lunch but also had the opportunity to discuss growing soybean crops with Mr. Gustavo Breuer Moreno, a farmer, whilst visiting one of his soya bean fields.
(Photograph 3 & Photograph 4 Discussion about growing soybean crops with Mr. Gustavo Breuer Moreno, a farmer, whilst visiting one of his soya bean fields – Argentina)
The next day was spent at Asociacion Cooperativas Argentinas harbour where Mr. Juan Carlos Piotto, the director of the export harbour, gave us a tour of the impressive site. He explained the day to day operations of the plant and showed us the grain unloading facilities, the transport to the ships, the vast storage silos and buildings for grains and fertilisers, respectively. His company alone produces 3000 tons of soybean oil per day. We are very grateful for the fantastic hospitality we received at both the farm and the harbour plant.
(Photograph 5 & Photograph 6. Asociacion Cooperativas Argentinas harbour – learning about the operations of the plant including the grain unloading facilities, the transport to the ships, the vast storage silos and buildings)
On the 14th March a stakeholder meeting was organised at the Bolsa de Comercio of Rosario and Ignacio Ibanẽz from the Ministry of Science, Technology and Innovation gave an introductory presentation about biotechnology in Argentina. This was followed by presentations of Dr Martin Crespi, Dr Atilio Castagnaro, Dr Luis Aguirrezabal and Dr Adrian Charlton who explained the work involved in ABSTRESS and BIOTECSOJASUR. A discussion with stakeholders led to an interesting and fruitful exchange. Overall, the workshop and stakeholder meeting were a great success enabling future collaborations between EU and Mercosur researchers, strengthening relationships within the projects and giving us the opportunity to visit Argentina, a potential market for ABSTRESS deliverables.
2. Second International Legume Society Conference, Troia, Portugal 11th – 14th October 2016
The Second International Legume Society Conference took place on the 11th to the 14th October, 2016 at the Tróia resort, (in the vicinity of Lisbon). It was attended by over 350 delegates from 53 countries, with an agenda comprising of over 120 oral presentations in plenary and parallel sessions, with approximately 190 additional poster presentations. The final Abstress project meeting was held as a satellite event during the conference on October 11th.
Dr Adrian Charlton (Fera) chaired the first plenary session of the conference on the morning of 12th October entitled Legumes Value Chain: Market Requirements and Economic Impact. This session featured presentations including the economic future for legumes, local supply chain limitations, novel uses of the protein fraction from legume crops and the future of legumes from an EU CAP perspective.
(Photograph 7. Dr Adrian Charlton – Chairing the first plenary session at the European conference in Troia Portugal. ABSTRESS Co-cordinator, WP5 & WP6 Leader)
Dr Richard Thompson (INRA) chaired the second plenary session of the conference on the morning of 12th October entitled Legumes and Environment. This session discussed topics relating to methods to validate environmental benefits of legumes and data on how legumes could aid in sustainable new (or modified) agricultural practices.
Dr Carmen Bianco (IBBR) chaired a parallel session on the afternoon of 12th October entitled Beneficial Legume-microbe Interactions A. This session covered topics relating to signalling in legume symbiosis, beneficial microbe populations in the soil environment and thoughts to improve the adaptation of legume-rhizobium symbiosis. Carmen also presented a 5 minute “flash presentation” on the evening of the 12th October on her poster “The auxin indole-3-acetic acid (IAA) is more than a plant hormone.”
Dr Christophe Salon (INRA) chaired a parallel session on the afternoon of the 13th October entitled Frontiers in Plant and Crop Physiology. This session presented topics on in vivo monitoring of legume root developments and changes in the soybean transcriptome induced by abiotic stress. Prof. Phil Mullineaux (University of Essex) gave the keynote lecture in this session, presenting on his work within the Abstress project. The presentation was entitled “The identification of novel genes controlling plant-environment interactions” and primarily discussed the transcriptomics and HUB gene discovery work undertake in Abstress.
(Photograph 8. Prof. Phil Mullineaux (University of Essex) gave the keynote lecture at the European Conference. ABSTRESS WPL WP3)
Dr Adrian Charlton (Fera) presented in the plenary session entitled Frontiers in Legume Breeding on the morning of the 14th October. Dr Charlton presented an overview of the Abstress project, specifically the methodologies / technology platforms used within the project to better understand resistance to combined abiotic and biotic stress. The presentation was entitled “Improving the resistance of legume crops to combined abiotic and biotic stress”.
Mike Dickinson (Fera) and Prof. Diego Rubiales (CSIC) presented in the final parallel session on the afternoon of the 14th October. The session was entitled Resistance to Biotic and Abiotic Stresses. Mike presented on the metabolomics methodology and results within the Abstress project, his talk was entitled “Exploring metabolic changes in legumes exposed to combined biotic and abiotic stress.” Professor Rubiales presented an overview of his recent work on disease resistance in pea, discussing his work in cloning orthologs of resistance genes in pea as part of Abstress.
(Photograph 9. Mike Dickinson (Fera) presented Exploring metabolic changes in legumes exposed to combined biotic and abiotic stress)
Martin Rusilowicz (Fera and The University of York) presented a poster entitled “Interactive exploration of the Medicago stress response through time-series analysis in the MetaboClust software package” which described the statistical analysis of the metabolomics data within Abstress.
Dr Peter Kalo (NARIC) presented a poster entitled “Identification of the translocation breakpoint between chromosome 4 and 8 in the genomes of Medicago truncatula A17 and A20” which described some of the genetic mapping work undertaken in Abstress.
(Photograph 10. Conference Delegates in Portugal)
Exploitation of results
Website: The ASBTRESS website can be found at www.abstress.eu. This contains a summary of the project status and is updated regularly. It is also a repository for all of the main data arising from ABSTRESS and as such is one of the main exploitation vehicles used by the project consortium. Interest in the project has clearly grown through its 5 year life span as shown by the webstats below:
(Figure 1. Webstats for life of project)
Stakeholder database: The front page of the ABSTRESS website contains a link for interested partners to join as formal ABSTRESS affiliates. This registration was used to create the ABSTRESS stakeholder database. Fifty-six organisations registered a direct interest in ABSTRESS including multinational coroperations and smaller plant breeders. ABSTRESS affiliates were reached across the globe with particular interest in India.
Stakeholder meetings: ABSTRESS was particular successful at reaching interested parties through local stakeholder meetings. With a target at the outset of the project of 10 stakeholder meetings, 19 were held during the course of the project
Oral Presentations: ABSTRESS was presented at international conferences by the consortium a total of 58 times against a target of 5 at the project outset. It is anticipated that many more presentations acknowledging ABSTRESS as a source of data and funding will be given as further conclusions are extracted from the large complex datasets we hold.
Newsletters: Three newsletters were produced against a target of 5. The ABSTRESS management board decided that there was insufficient high-impact content to produce 5 newsletters and that 3 would have a greater impact.
Peer reviewed papers: 9 papers have currently been published with several more in the pipeline. The initial target was 10 papers and this will be achieved comfortably albeit beyond the funded lifetime of the project.
Trade articles: A further 11 articles appeared in trade journals directly referencing ABSTRESS.
Press coverage: ABSTRESS was referenced in the press at least 11 times including an interview for the BBC with the project coordinator.
List of Websites:
www.abstress.eu
Project Co-ordinator: Dr Adrian Charlton, Fera Sciecne Limited
Email: abstess@fera.co.uk