Arabidopsis thaliana plants were infested i) with sucking insect herbivores (the generalist aphid Myzus persicae and the specialist aphid Brevicoryne brassicae), ii) with chewing insect herbivores (generalist caterpillars of Spodoptera exigua and specialist caterpillars of Pieris rapae) or iii) were treated by wounding. For each treatment, rosette leaves were harvested at two time points (6h and 24h) after removal of insects. For chewing herbivores and wounding both local, i.e. immediately damaged leaves, and systemic, i.e. undamaged leaves from the same plant, were collected. Control plants were uninfested, but otherwise equally treated and harvested in parallel. We tested the hypothesis that Arabidopsis can recognize and respond differentially to insect species at the transcriptional level using a genome wide microarray. Transcriptional reprogramming was characterized using co-expression analysis in damaged and undamaged leaves at two times in response to mechanical wounding and four insect species. In all, 2778 (10.6%) of annotated genes on the array were differentially expressed in at least one treatment. Responses differed mainly between aphid and caterpillar and sampling times. Responses to aphids and caterpillars shared only 10% of up-regulated and 8% of down-regulated genes. Responses to two caterpillars shared 21% and 12% of up- and down-regulated genes, whereas responses to the two aphids shared only 7% and 4% of up-regulated and down-regulated genes. Overlap in genes expressed between 6h and 24h was 3-15%, and depended on the insect species. Responses in attacked and unattacked leaves differed at 6h but converged by 24h. Genes responding to the insects are also responsive to many stressors and included primary metabolism. Aphids down-regulated amino acid catabolism; caterpillars stimulated production of amino acids involved in glucosinolate synthesis. Co-expression analysis revealed 17 response networks. Transcription factors were a major portion of differentially expressed genes throughout and responsive genes shared most of the known or postulated binding sites.
Overall design
For each treatment, four (or three) replicate arrays were hybridized for a total of 59 arrays. Total RNA from replicate control plants harvested at each time point and treatment was pooled in order to minimize variability between individuals. Labeled cDNA derived from the pooled control RNA was co-hybridized with labeled cDNA derived from RNA isolated from individual plants treated with the described herbivores. This results in four biological replicates for each treatment. In order to account for possible dye biases, within each treatment group two treatment samples were labeled using Cy3 (with controls labeled with Cy5) and the remaining two were labeled using Cy5 (with control samples labeled with Cy3). Hybridization orders were randomized in order to avoid biases due to the hybridization time. analyzed.
Less...
More...