1. Lists of target genes for reverse genetic analyses. The pathogen-induced genes that we have targeted for reverse genetics analysis are listed under the link “Genes of Interest”. The table labeled "For eds testing-proposal" is the list of genes from our original proposal for which we planned to test mutants for enhanced disease susceptibility phenotypes. The table labeled "For R-gene signaling-proposal" is the list of genes from our original proposal for which we planned to test mutants for defects in R-gene mediated signaling. These versions of the lists were created based on data from the 8,000 gene Arabidopsis Affymetrix array. As our project proceeded, we found that T-DNA insertion lines could not be obtained for some of the genes on these lists. Consequently, we will not be able to test them in our project. We have also added some target genes based on new data from the ATH1 Affymetrix array. To see which genes we are actively studying, use the lists of genes for which we have identified homozygous T-DNA insertion lines, described in the next section.

2. Lists of T-DNA insertion lines obtained. We have identified 195 homozygous T-DNA insertion lines that will be subjected to eds testing. They are shown in the table "T-DNA lines for eds testing". We will continue to add lines to this list, and update it periodically. We have identified 95 homozygous T-DNA lines with mutations in genes that are induced early during gene-for-gene resistance. These lines will be subjected to expression profiling using a custom mini-array to test for defects in signaling during R-gene mediated resistance. They are shown in the table "T-DNA lines for R-gene testing". Both of these lists can also be found under the link "Genes of interest".

3. Development of a miniarray. We have designed a small-scale microarray, which we call a “miniarray”.? Genes on the miniarray are selected for? representing diverse expression patterns during pathogen interactions.? The probes used for the miniarray are shown in the table “Miniarray design”.? We demonstrated high accuracy and precision of the measurements made by the miniarray.? See Sato et al. (2007) for details.

4. Reverse genetic screen with Pst DC3000 avrRpt2. We screened Arabidopsis T-DNA insertion lines for the genes listed in "For R-gene signaling-proposal" using the miniarray 6 hours after infection of Pst DC3000 avrRpt2. We identified 14 lines that showed significant changes in expression profiles, compared with the wild type.

5. NDR1 is involved in MAMP-triggered signaling. Using the miniarray, we demonstrated that the NDR1 gene is involved in the MAMP-triggered signaling, as well as the R gene-mediated signaling.? See Sato et al. (2007).? Indeed, Pst DC3000 hrcC mutant, which lacks the type III secretion system, grew more in ndr1 plant than wild type.? See Katagiri and Sato (2007).

6. Salicylic acid accumulation is triggered by MAMPs.? We demonstrated that the MAMP-triggered signaling leads to accumulation of salicylic acid (SA), and consequently, the SA-mediated signaling gets turned on.? Among the MAMP-inducible genes, there are SA-dependent and ?independent genes as well as partly dependent genes. See Tsuda et al. (2008).

7. A system snapshot of Arabidopsis responding to Psm ES4326. ?We used the Affymetrix ATH1 array to profile wild-type plants nine, 24, and 32 hours after infection by the moderately-virulent strain PsmES4326.? As the response was strongest at 24 hours, we chose this time point for profiling canonical defense signaling mutants: pad4, sid2, npr1, ein2, and coi1.? Of genes induced or repressed in wild-type plants, the change in expression of more than 80% was reduced in at least one signaling mutant, demonstrating major effects of these genes on both induction and repression of pathogen-responsive genes.? Most genes displayed one of a few major patterns of regulation, allowing construction of a simple network diagram.? The topology of this network was further explored with the miniarray using additional mutants.

8. Reverse genetics identification of mutants that cause enhanced susceptibility to Psm ES4326. ?Using profiling data from the ATH1 array, we selected genes that were induced by Psm ES4326 infection.? Plants homozygous for T-DNA insertions in approximately 220 induced genes were tested for enhanced disease susceptibility (eds) to Psm ES4326.? For eleven genes, we were able to identify at least two independent alleles with eds phenotypes, leading us to conclude that these genes play roles in resistance. ?We have studied three of these eleven genes in some detail.

9. Construction of a network graph describing responses to Pst DC3000 avrRpt2 infection.? We investigated the relationships among genes known to be involved in defense signaling.? For twenty-one Arabidopsis mutants, atnos1 (noa1), coi1, dde2, ein2, ein3, jar1, jin1, mpk3, mpk6, ndr1, nho1, nia2, npr1, pad4, pbs2 (rar1), pen2, pmr4, AtrbohD, rps2, sag101, and sid2 (ics1), triplicated miniarray expression profiles were obtained six hours after infection with Pst DC3000 avrRpt2.? Pst DC3000 avrRpt2 can activate multiple network sectors: the MAMP-inducible sector as it is a P. syringae strain, the resistance gene and SA sector as it is recognized by the R protein RPS2, and the JA-mediated sector as it produces the phytotoxin coronatine.? Therefore, using this strain as the input stimulus allows us to survey a very large part of the network activity.? The expression profiles were analyzed using RepEdLEGG (see section 12 below).? Our tentative model for the network graph for q-values <0.01 is shown here.

10. (Bioinformatics) Development of a method to reduce random noise from a large data set. A large data set could allow reduction of random noise using an appropriate statistical method.? However, a conventional method like principal component analysis (PCA) would lose small but significant signals together with noise. ?A new algorithm, RSPR-NR (Row-specific, Sorted PRincipal component-guided noise reduction), has been developed.? RSPR-NR retains small signals well while reducing random noise from the data.

11. (Bioinformatics) LEGG (Locally linear Embedding Graph Generator): a new algorithm for a non-linear dimensionality reduction method. We previously developed a method to convert non-linear dimensionality reduction information to a graphical representation (LCF (Local Context Finder); Katagiri and Glazebrook (2003)). We developed a distinct algorithm for this task with higher speed and accuracy.? We call the algorithm LEGG.? See van Poecke et al. (2007).

12. (Bioinformatics) Development of a multivariate analysis method that is sensitive to low levels of similarities. The concept of LEGG was extended to capture weak similarities among expression profiles.? The algorithm is called RepEdLEGG.? This procedure was used to infer the topology of the signaling network based on the expression profiles of Arabidopsis mutants (section 9 above).