AN INVESTIGATION OF MOLECULAR INDICATORS OF PLANT VIRUS INFECTION

Abstract

Aims. This thesis sought to advance what is known about how best to monitor the health of plants in order to determine what could be used as a molecular marker to detect deviation from health due to plant virus infection. The specific objectives were to (i) confirm the ability of five dissimilar viruses to infect Arabidopsis thaliana, (ii) to develop a method for the accurate quantification and analysis of small RNAs (sRNAs) from low molecular weight RNA (LMW-RNA) in response to the five viruses, (iii) to develop a real-time quantitative polymerase chain reaction (qPCR) method for the quantification of gene transcripts of interest in response to the five dissimilar viruses, and (iv) extend qPCR assays to a further biotic stress and two abiotic stresses in Arabidopsis in order to determine specificity of assay to virus infection. Results. It was established by PCR and qPCR, sequencing, and Immunostrip® assay that each of the five viruses were successfully inoculated into Arabidopsis and were absent from mock-inoculated tissue. LMW-RNA components were accurately quantified but not all viruses could be detected at every time point. Analyses of the ratio of sRNA to rRNA as a proportion of averaged mock-inoculation predicted a 94% correlation with known virus presence. SGS3 showed a statistically significant change in transcript accumulation compared to mock-inoculation in response to all five viruses as assessed by qPCR. A decision tree predictive model was devised from the sRNA/rRNA ratio and SGS3 transcript accumulation, resulting in > 94% positive predictive value. Conclusions. It is concluded that calculating a ratio of sRNA to rRNA accumulation as a proportion of averaged mock-inoculation can predict known virus infection to a high degree of certainty, if this response proves specific to virus infection. The decision tree predictive model developed from the sRNA/rRNA ratio and SGS3 transcript accumulation increases the likelihood of predicting virus infection to > 94%. Given further investigation and analysis, our ability to detect generic plant virus infection is likely to benefit from this host plant based method.