Transcript levels of several individual genes of these BINs were similarly affected by both treatments (Fig. 6b; Resource S4). In both data sets, transcripts for genes associated with ethylene-related processes showed increased levels (BINs 17.5 and 17.5.5, Resource S2) and auxin-related gene groups (BINs 17.2 and 17.2.2, Resource S2) were changed also, but transcript levels generally decreased. With MFA only, salicylic acid-related categories (BINs 17.8, 17.8.1; Resource S2) were affected with overall up-regulation.Cluster Analysis of Leaf Transcriptomes
Figure 5. Correlation of gene expression changes from cytochrome pathway inhibition (AA) and TCA cycle inhibition (MFA). Graph of fold change (log2) from AA treatment (abscissa) versus fold change (log2) from MFA treatment (ordinate) for 215 genes that showed transcript level changes with q#0.05 for both treatments. 10.6) as was BIN 31.4 for vesicle transport, which showed overall induction (Resource S2). For both transcriptome data sets, metabolism-related gene categories were identified as responsive. With AA treatment, gene groups encoding enzymes for amino acid synthesis showed overall transcript level increases (BIN 13 and nested BINs, including those for aromatic amino acid and tryptophan synthesis, Fig. 6a and Resources S2, S3, S4), while for MFA treatment the gene groups for amino acid degradation showed overall transcript level decreases (BIN 13 and nested BINs, Fig. 6a and Resources S2, S3, S4). Another group of categories affected oppositely between the two inhibitor treatments were related to protein degradation (BIN 29.5 and nested BINs). These showed overall downregulation with AA treatment and overall up-regulation with MFA (Resource S2). The MFA treatment data set had additional metabolic gene categories showing overall transcript decreases, including starch synthesis (BIN 2 and nested BINs), minor carbohydrate metabolism (BINs 3 and 3.5), lipid metabolism (BIN 11 and nested BINs), C-1 metabolism (BIN 25), nucleotide metabolism (BIN 23), photorespiration (BIN 1.2), and tetrapyrrole synthesis (BIN 19; Resource S2).
For both AA and MFA treatments, the broad functional category “biotic stress” (BIN 20.1, Fig. 6a and Resources S2, S3, S4) was highly statistically significant, with overall induction. Some categories for genes encoding enzymes to ameliorate effects of oxidative stress were affected in both AA and MFA data sets, showing overall induction: glutaredoxins (BIN 21.4) and glutathione S-transferases (BIN 26.9; Fig. 6a and Resources S2, S3, S4). A gene category associated with ascorbate and glutathione metabolism (BIN 21.2) showed overall up-regulation with AA treatment, but not MFA treatment (Fig. 6a and Resources S2, S3, S4). Most other statistically significant stress-related functional categories were associated with MFA treatment, showing overall repression, including those for dismutases, catalases and thioredoxins (BINs 21.1 and 21.6, Fig. 6a and Resources S2, S3, S4) and a group of genes that respond to abiotic stresses such as drought and salt (BIN 20.2.3, Resource S2). Notably, the general functional category “abiotic stress” (BIN 20.2) was not statistically significant for either of the inhibitor treatments (adjusted p = 0.97 and 0.46 for AA and MFA treatment, respectively; Resource S4). Functional categories for some signaling-related genes and processes were affected in the same way by both inhibitors.
In order to determine whether the genes whose transcript levels were affected by AA or MFA treatment are affected similarly by biotic, abiotic, or oxidative stresses imposed on aerial tissue, two cluster analyses were performed. The genes with transcript abundance changing in response to either AA or MFA were used as separate query sets and compared to the responses of these same genes (termed here “transcript subset”) in 47 other experiments in which a stress was applied (Resource S5), including stresses representative of the biotic, abiotic, and oxidative categories and the other inhibitor experiment; the MFA experiment was included in the cluster analysis using the AA-affected transcript subset and vice versa. Two trees (Fig. 7) were derived from correlation coefficients (Resource S6) for the cluster nodes. The corresponding heat maps for the trees are shown in Resources S7 to S15. A photomorphogenesis experiment, low red/far red light, served as an out-group (circled in Fig. 7a and b). The profile of the transcript subset from this light treatment was not positively correlated with the gene transcript abundance changes in response to AA or MFA. The transcript expression pattern resulting from AA treatment, when used as the query set, clustered at node 13 (correlation coefficient 0.76) with transcript subsets from ozone treatment, and two of the three UV-B treatments. The next-closest node, 15 (correlation coefficient 0.73), joins with fungal (Botrytis cinerea) and oomycete (Phytophthora infestans) pathogens and the P. infestans elicitor, NPP1. Subsequent nodes join with the bacterial pathogens Pseudomonas syringae avrRpm1 (node 18, correlation coefficient 0.67) and 6 h P. syringae phaseolicola, and the bacterial elicitors flagellin [45], lipopolysaccharide [46], and harpin, which disrupts mitochondria [47] (node 19, correlation coefficient 0.66; Fig. 7a). The AA transcriptome overlaps the MFA transcriptome by less than 16% (Fig. 4) which probably accounts for their low correlation in this tree (node 28, correlation coefficient 0.49; Fig. 7a). When genes whose expression was affected by MFA treatment were used as the query, essentially the same clustering with respect to ozone, UV-B, pathogens and pathogen elicitors relative to AA was observed (see Fig. 7b for nodes and correlation coefficients) with two exceptions. For one, in this tree, MFA closely correlated with AA, at node 10 (correlation coefficient 0.79), likely due to the MFA transcriptome having a large percent of genes also affected by AA treatment (Figs. 4 & 5). For the other, NPP1 treatment clustered with P. syringae phaseolicola, close to the bacterial elicitors (Fig. 7b). Thus, in both trees, there was an enrichment of pathogen and pathogen-related treatments in the nodes closest to AA and MFA, and, except for ozone and UV-B, abiotic stresses occurred at more weakly correlated nodes. However, four pathogen-treatment experiments did not cluster close to AA, MFA, or the other pathogen and pathogen-related treatments.
