(May 14, 2018) — A team of biologists and computer scientists has adopted a time-based machine-learning approach to deduce the temporal logic of nitrogen signaling in plants from genome-wide expression data. The work potentially offers new ways to monitor and enhance crop growth using less nitrogen fertilizer, which would benefit human nutrition and the environment.
The research, which appears in the journal Proceedings of the National Academy of Sciences (PNAS), centers on gene regulatory networks (GRNs) that identify which transcription factors serve to regulate genes needed to respond to nitrogen, which is a nutrient vital to plant development and human nutrition.
“By constructing these regulatory networks based on dynamic gene responses to nitrogen treatment, we can see, in time-lapse detail, the genetic process necessary for the intake of nitrogen and its conversion into amino acids used in the synthesis of all N-containing compounds including DNA, proteins and chlorophyll” explains Gloria Coruzzi, a professor in New York University’s Department of Biology and Center for Genomics and Systems Biology and the paper’s senior author. “Armed with these new insights, we can now look ahead for ways to bolster the efficiency of food production and enrich sustainable agriculture measures on lower nitrogen input, which would benefit the environment.”
The study also included researchers from Purdue University, the University of Illinois at Urbana Champaign, Cold Spring Harbor Laboratory, and the French National Institute for Agricultural Research.
The research exploited time—the fourth and largely unexplored dimension of GRNs—with the aim of better elucidating the transcription factors (TFs) relevant to genetic responses to nitrogen. Specifically, understanding how transcription factors function at different points in time can allow scientists to target the early responders and to make predictions on the temporal operation of the entire gene regulatory network.
This time-based GRN now provides a wealth of regulatory knowledge to inform testable hypotheses on how 155 transcription factors exert regulatory control of nitrogen response and its effect on core plant life processes, such as circadian rhythm, photosynthesis, and RNA metabolism, among other phenomena affecting plant growth, development and yield.
The research was supported by grants from the National Institutes of Health (NIH) (R01-GM032877), a National Science Foundation Plant Genome Grant (IOS-1339362), an NIH National Research Service Award (GM095273), and an NIH National Institute of General Medical Sciences Fellowship (1F32GM116347).
West Lafayette, IN (May 14, 2018) – When a plant is introduced to an abundant patch of water or fertilizer, a cascade of genetic and molecular actions lead to beneficial physiological responses such as root development and increased biomass. Identifying the genes involved in the early molecular responses has been difficult, but a Purdue University scientist has led an effort that can identify the genetic mechanisms and predict targets for improving crop plants.
Kranthi Varala, an assistant professor in the Department of Horticulture and Landscape Architecture, is first author of a report published May 14 in the “Proceedings of the National Academy of Sciences” that describes a method for identifying the genes activated when plants are exposed to external stimuli. The process, called temporal transcriptional logic for plant response, streamlines searches that can often involve tens of thousands of genes that work together in complex ways. He collaborated with colleagues from the University of Illinois and New York University.
“Putting together experimental and computational methods, we developed a general methodology that can be applied to almost any area of interest, whether it’s plant stress, pathogen interactions or nutrition aspects,” Varala said. “Even in animal systems, any time an organism has to respond to something external, you have to have a methodology to find what is controlling that response.”
Arabidopsis plants were grown for in the absence of nitrogen, a critical plant nutrient. After two weeks, some plants were given nitrogen while others went without. Gene expression was measured 11 times, from the moment of nitrogen application through two hours.
Gene activity changes in the nitrogen-fed plants were compared against the gene activity in plants that didn’t receive the nitrogen. Those genes that increased or decreased activity, when compared to the controls were considered likely to be involved with a plant’s nitrogen use.
In particular, the scientists were interested in the transcription factors, genes that essentially act as controls for other genes. When a transcription factor is activated, it turns up or down the activity of its target genes.
Using a machine-learning algorithm, they predicted which genes work together to influence nitrogen use. The results identified 155 transcription factors and 608 other genes - out of 28,000 total - that are closely linked to nitrogen use in plants and predicted how changes to the transcription factors might influence this process.
Tests on plants with some of those transcription factors knocked out showed which were actually associated with the predicted genes and controlled plant growth, root development and other physiological responses. Overall, the prediction algorithm was correct one in three times.
“That might not seem high, but when you think about 28,000 genes and the number of combinations there might be, getting one-third of these correct is highly significant,” Varala said.
The small set of transcription factors identified in this research are now targets for improving a plant’s ability to take nitrogen from the soil and use it as efficiently as possible, reducing the amount of fertilizers that might need to be used in future crops.
“If we can apply less fertilizer and get the same crop yield, that’s both an ecological and economic success,” Varala said. “To do that, you have to know how the plant responds to nitrogen and how it uses that nitrogen”.
Going forward, the method can be used to narrow the scope of genes associated with responses to heat, drought and other plant stresses.
The National Institutes of Health and the National Science Foundation supported this research.