Multifactorial and Species-Specific Feedback Regulation of the RNA Surveillance Pathway Nonsense-Mediated Decay in Plants.

Nonsense-mediated decay (NMD) is an RNA surveillance mechanism that detects aberrant transcript features and triggers degradation of erroneous as well as physiological RNAs. Originally considered to be constitutive, NMD is now recognized to be tightly controlled in response to inherent signals and diverse stresses. To gain a better understanding of NMD regulation and its functional implications, we systematically examined feedback control of the central NMD components in two dicot and one monocot species. On the basis of the analysis of transcript features, turnover rates and steady-state levels, up-frameshift (UPF) 1, UPF3 and suppressor of morphological defects on genitalia (SMG) 7, but not UPF2, are under feedback control in both dicots. In the monocot investigated in this study, only SMG7 was slightly induced upon NMD inhibition. The detection of the endogenous NMD factor proteins in Arabidopsis thaliana substantiated a negative correlation between NMD activity and SMG7 amounts. Furthermore, evidence was provided that SMG7 is required for the dephosphorylation of UPF1. Our comprehensive and comparative study of NMD feedback control in plants reveals complex and species-specific attenuation of this RNA surveillance pathway, with critical implications for the numerous functions of NMD in physiology and stress responses.

Metabolomics of tomato xylem sap during bacterial wilt reveals Ralstonia solanacearum produces abundant putrescine, a metabolite that accelerates wilt disease.

Ralstonia solanacearum thrives in plant xylem vessels and causes bacterial wilt disease despite the low nutrient content of xylem sap. We found that R. solanacearum manipulates its host to increase nutrients in tomato xylem sap, enabling it to grow better in sap from infected plants than in sap from healthy plants. Untargeted GC/MS metabolomics identified 22 metabolites enriched in R. solanacearum-infected sap. Eight of these could serve as sole carbon or nitrogen sources for R. solanacearum. Putrescine, a polyamine that is not a sole carbon or nitrogen source for R. solanacearum, was enriched 76-fold to 37 µM in R. solanacearum-infected sap. R. solanacearum synthesized putrescine via a SpeC ornithine decarboxylase. A ΔspeC mutant required ≥ 15 µM exogenous putrescine to grow and could not grow alone in xylem even when plants were treated with putrescine. However, co-inoculation with wildtype rescued ΔspeC growth, indicating R. solanacearum produced and exported putrescine to xylem sap. Intriguingly, treating plants with putrescine before inoculation accelerated wilt symptom development and R. solanacearum growth and systemic spread. Xylem putrescine concentration was unchanged in putrescine-treated plants, so the exogenous putrescine likely accelerated disease indirectly by affecting host physiology. These results indicate that putrescine is a pathogen-produced virulence metabolite.

A comparison of classical and machine learning-based phenotype prediction methods on simulated data and three plant species.

Genomic selection is an integral tool for breeders to accurately select plants directly from genotype data leading to faster and more resource-efficient breeding programs. Several prediction methods have been established in the last few years. These range from classical linear mixed models to complex non-linear machine learning approaches, such as Support Vector Regression, and modern deep learning-based architectures. Many of these methods have been extensively evaluated on different crop species with varying outcomes. In this work, our aim is to systematically compare 12 different phenotype prediction models, including basic genomic selection methods to more advanced deep learning-based techniques. More importantly, we assess the performance of these models on simulated phenotype data as well as on real-world data from Arabidopsis thaliana and two breeding datasets from soy and corn. The synthetic phenotypic data allow us to analyze all prediction models and especially the selected markers under controlled and predefined settings. We show that Bayes B and linear regression models with sparsity constraints perform best under different simulation settings with respect to explained variance. Further, we can confirm results from other studies that there is no superiority of more complex neural network-based architectures for phenotype prediction compared to well-established methods. However, on real-world data, for which several prediction models yield comparable results with slight advantages for Elastic Net, this picture is less clear, suggesting that there is a lot of room for future research.