Understanding the gene regulatory processes involved in the absorption, translocation, and metabolism of iron is essential for producing plants with increased nutrient capacity and tolerance to nutrient-poor soils.  These efforts would aid in the identification of meaningful ways to stretch resources, increasing crop yields to feed the projected 9 billion people in 2050 and address global iron deficiency induced anemia, which is the most prevalent global nutritional disorder.  Prior to the project,, root-specific transcriptional datasets that measure whole-genome responses to low iron in the model species Arabidopsis thaliana across time, in specific root developmental regions, and across cell-types had been generated. Although efforts had been made to mine these data, an unbiased assessment of the potential regulators involved in the iron deprivation response had not been conducted.  This was, in part, due to the lack of computational and mathematical tools needed to generate dynamic models of the gene regulatory response to iron deprivation stress at the whole root and at the cell-specific level. In this work, we developed broadly applicable, yet novel computational and analytical paradigms for understanding the dynamic transcriptional cascade associated with the iron deficiency response in A. thaliana at the whole-root level and at the cell-specific level (epidermis cells).  

Research Findings: We developed the Clustering and Differential Alignment Algorithm (CDAA), which identified 7 putative regulators previously unlinked to the iron homeostasis response using whole-root time-course transcriptome profiles. The CDAA predicted 32 potential regulatory connections of which 53% were experimentally validated (Figure 1).  We formulated an integrative modelling approach to generate a dynamic ordinary differential equation model describing the expression of these genes and their regulatory interactions under iron deficient conditions (Figure 2). The trained model was able to capture and account for a significant difference in mRNA decay rates under iron sufficient and iron deficient conditions (Figure 3), approximate the expression behavior of currently unknown gene regulators, unveil potential synergistic effects between the modulating transcription factors and predict the effect of double knockdown mutants (Figure 4). The presented modelling approach illustrates a framework for experimental design, data analysis and information aggregation in an effort to gain a deeper understanding of various aspects of a biological process of interest.  

While these findings shed new light on responses within the whole root, we also developed tools to gain insight into responses within specific root cell types.. We analyzed transcriptional profiles of A. thaliana root epidermal cells that were stressed by iron deficiency for 36 hours, taking samples at a 6 hour increments. We used Dirichlet Process - Gaussian Process (DPGP) clustering approach and gene ontology (GO) analysis, to identify groups of genes enriched in iron response genes (Figure 5). Finally, we applied random sampling scheme and utilized existing DNA affinity purification sequencing (DAP-Seq) data to infer the Fe responsive-epidermal GRN. 

Our analysis of the transcriptional data also revealed that STOP2, a TF with no known role in iron deficiency response, was differentially expressed five out of the six time points in epidermal cells (Figure 6).  Our network and molecular analysis also indicated that STOP2 also regulates the expression of AHA genes important for acidifying the rhizosphere in response to iron deprivation. 

Broader Impacts: As part of our broader impact and outreach efforts, we developed a electronic-based tool that would allow K-12 and undergraduate students to understand metabolic pathways and how transgenics can influence the flow of pathway biochemical reactions.  We developed the Enzyme metabolic circuit (EzMC), which is a physical circuit model of the Lignin biosynthesis pathway. This circuit uses current flow as an analog to the mass flow of metabolites through the network. Transgenics are simulated by adjusting potentiometers strategically placed through the network, which emulates the increase or decrease in available enzymes.  Students can then use voltmeters to read the voltage at various points in the network, which is equivalent to the metabolite concentrations. The lignin EzMc tool was showcased by North Carolina State University communication department (https://news.ncsu.edu/2017/03/hands-on-ge-model-2017/). The story about our tool reached not only undergraduates, graduates, faculty and staff at our university, but it was also picked up by external news outlets (Techxplore:  https://techxplore.com/news/2017-03-hands-on-students-genetic.html, NSF Science360: https://news.science360.gov/archives/20170323). The next step for this tool is to perform limited usage in classrooms at the k-12 level. We plan to seek private funding to construct several of these devices so that we can effectively engage these students at the secondary level.

Last Modified: 10/30/2019
Modified by: Cranos M Williams