Modeling of gene and protein networks in the context of plant immunity

Thumbnail Image
Date
2021-08
Authors
Velásquez Zapata, Valeria
Major Professor
Advisor
Wise, Roger P
Dorman, Karin S
Nettleton, Daniel
Kelley, Dior
Whitham, Steven
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Authors
Research Projects
Organizational Units
Journal Issue
Is Version Of
Versions
Series
Department
Plant Pathology
Abstract
Cellular dynamics are regulated by coordinated biological networks that interact at multiple molecular levels. Functional characterization of such networks has shown that complex phenotypes are the product of profound changes in the interplay of gene and protein interactions. The aim of this work is to untangle the molecular properties of these networks, and thus, enable the construction of predictive models regarding the regulation of immunity in host-pathogen systems. Systems biology approaches were integrated to characterize the defense response of barley (Hordeum vulgare L.) to powdery mildew disease, caused by the ascomycete fungus, Blumeria graminis f. sp. hordei (Bgh). The barley MLA nucleotide-binding, leucine-rich-repeat (NLR) receptor was used as a model regulator from which the immune response is characterized. MLA represents a cross-species NLR model since its orthologs confer resistance to powdery mildew in wheat and transgenic Arabidopsis, wheat and rye Ug99 stem rust, as well as wheat stripe rust. High throughput Yeast-Two-Hybrid, Next-Generation-Interaction Screening (Y2H-NGIS) enabled identification and validation of protein-protein interactions (PPI) between MLA6 and barley targets, and these PPI were subsequently extended to additional MLA alleles. Novel interactors were positioned in a newly generated predicted barley interactome, to deduce in which cellular compartments and signaling processes the receptor was involved. Data integration between the predicted interactome, transcriptome and expression quantitative trait loci (eQTL) were performed to highlight nodes, edges, structures, and modules in the signaling network. Finally, transcriptional regulation models of the barley immune response were identified using a dynamic Gene Regulatory Network (GRN). Candidate transcription factor (TF) clusters and genes were prioritized for their role in the immune response, finding transcriptional associations with proteins in the MLA-signaling and resistant networks. To start, Y2H-SCORES, a statistical software to infer protein-protein interactions from Y2H-NGIS data, was proposed. NGIS combines yeast two-hybrid (Y2H) with deep sequencing to generate interactome networks in any organism, however, there were no standard metrics to rank high-confidence interacting proteins. Y2H-SCORES uses a Borda ensemble of three quantitative measures of Y2H-interactors, comprising: 1) an enrichment score associated with significant enrichment of prey interactors under selection, 2) a specificity score that measures the degree of interaction specificity of a prey among multi-bait comparisons, and 3) an in-frame score that ranks the selection of in-frame interactors coming from cDNA prey libraries. Using simulation, benchmarking with independent Y2H-NGIS datasets and an in-house experiment, we provided a quantitative assessment of the Y2H-SCORES performance to predict interacting partners under a wide range of experimental scenarios. These analyses helped to identify ideal experimental conditions that facilitate experimental validation including protocols such as prey library normalization, maintenance of larger culture volumes and replication of experimental treatments. Y2H-SCORES can be implemented in different yeast-based interaction screenings, with an equivalent or superior performance than existing methods. Y2H-SCORES software is available at GitHub repository https://github.com/Wiselab2/Y2H-SCORES/tree/master/Software. As a second objective, Y2H-SCORES was used to discover and validate novel interactions between barley MLA6 and fourteen new proteins. Integration between these experimentally validated interactors, additional defense-related datasets and a barley predicted interactome was performed to test hypotheses regarding MLA signaling and its link to plant immunity. Integration of interactome and eQTL data led to the generation of disease modules associated with two important moments in the Bgh infection cycle, penetration and haustorial development, revealing both core and unique responses. Using expression data from an Mla6-specified resistant progenitor and fast-neutron-derived susceptible mutants, a resistant interactome was built which presents unique properties associated with immunity, including higher protein essentiality and centrality, lower co-expression values and enrichment in Mla eQTL associations than the susceptible subnetwork. An MLA-associated subnetwork was also generated, which in combination with expression data, informed cellular localization of the immune receptor, and signaling response over time. Significant associations of MLA-signaling and defense related responses were found, comprising transcriptional regulation, MAP kinases, calcium signaling and intracellular trafficking. Results highlight master components of the signaling cascade, linking genomic, transcriptomic, and physical interactions during the MLA-based immune response. Lastly, the temporal regulation of immune signaling in response to powdery mildew was investigated. To achieve this, infection-time-course transcriptome data were evaluated from a wild-type resistant line carrying the Mla6 gene, and four fast-neutron-derived immune-signaling mutants. The time course spans from 0 to 48 hours after inoculation (HAI), covering key stages of Bgh development; appressorium formation, penetration of epidermal cells (16-20 HAI), and development of haustorial feeding structures (32-48 HAI). A GRN was generated using three different algorithms and ensembled based on the Biological Homogeneity Index (BHI) of the communities of the network. The resulting GRN covered 76% of the barley genome, including 1214 TFs, 20125 targets, and 54698 interactions. Hypergeometric tests were performed looking for enrichment of differentially expressed targets in each TF and TF-cluster, using comparisons between plants harboring the NLR-encoding gene, Mla6, and the corresponding mla6 mutant for the six timepoints. Additionally, other tests were performed looking for enrichment of eQTL associations (by timepoint and Mla-associated), NLRs, and targets in the resistant interactome. Ranks of the adjusted p-values of the tests were ensembled to identify the most important clusters and TFs in the GRN. Different families were found to regulate the transcriptional response at Bgh penetration including MADS box proteins, a B3 domain-containing protein, one homeobox protein and a NAC containing protein. Other families were associated with haustorial development, such as ethylene-responsive protein, zinc-finger TF, WRKY, MYB and scarecrow family protein. Remarkably, two MLA interactors were found as master drivers of defense gene transcription including a homeobox (HB) and a WRKY protein. These TFs are also predicted to regulate Mla transcription through different feedback loops, implying novel auto-regulatory mechanisms of the immune receptor.
Comments
Description
Keywords
Citation
Source
Copyright