Global expression changes in saul1-1

to the immune system an incorrect or missing N-glycosylation of the PRRs EFR (EF-TU RECEPTOR) and FLS2 (FLAGELLIN-SENSITIVE 2) has been shown to lead to non-functional receptors and retention in the endoplasmic reticulum ²⁷². Therefore, a down-regulation of N-glycosylation would be unexpected, since this posttranslational modification is vital for correct protein folding. In contrast, it is known that S-glycosylation is an important step in secondary metabolite synthesis ²⁷³. Although those pathways, which are down-regulated in saul1-1, are quite diverse, most of them have one characteristic in common: they are highly connected to the interplay between JA and SA ^274–276. For instance, the down-regulated TF MYB28 and MYB29, which promote glucosinolate biosynthesis, are known to be positively regulated by JA and suppressed by SA ^276,277. Additionally, repression of these TFs is known to lead to decreased growth ²⁷⁸. This is especially interesting, because a characteristic of the saul1-1 and other autoimmune-associated phenotypes is growth arrest ^123,134. Genes regulated by MYB28 or MYB29 comprise other TFs such as WRKY18 and WRKY40, which in turn regulate members of the CYP (Cytochrome P450) family ²⁷⁶. These, in particular CYP79F2 and CYP83A1, were found to be down-regulated in saul1-1 as well.

Interestingly, it has been reported that a mutant of CYP83A1 leads to a higher resistance against powdery mildew. This was due to an accumulation of camalexin, one of the major phytoalexins ⁴³. Phytoalexins are known to be expressed upon infection with necrotrophic pathogens and to be fungistatic ²⁷⁹. In context of ETI, they have not been excessively studied, but are thought to be expressed to prevent a secondary infection ²⁷¹. In addition, the regulations in cyp83a1 are linked to SA accumulation, because the increased resistance was found to be EDS1-, PAD4- and NDR1 (NON RACE-SPECIFIC DISEASE RESISTANCE 1)-dependent ⁴³. Other genes involved in S-glycosylation, like FMOGS-OX1 (Flavin Monooxygenase GS-OX1) and FMOGS-OX3, are known to be JA- and SA-responsive as well ²⁷⁵. Therefore, those diverse pathways connected to S-glycosylation, could be part of downstream regulations reacting to elevated SA levels in saul1-1 ¹³⁵.

In summary, the gene expression changes 6 h and especially 3 h after the temperature shift in saul1-1 appeared to be regulated in a quite complex manner involving an interplay of the hormones SA, JA, ET and ABA. These regulations seem to initiate all downstream regulations, since they were dominantly detected in cluster III and this cluster contained genes that were differentially expressed from 3 h onwards. This assumption is supported by the observation that after 6 h (Cluster I) fewer genes were differentially expressed for the first time. Consequently, the saul1-1 autoimmunity was defined by a large number of genes, which initiates after 3 h pathways that result in the saul1-1 phenotype. Therefore, clustering all DEGs in saul1-1 for the different time points was extremely helpful to get first insights into genes with similar behavior and into activation of pathways. However, genes that were down- and up-regulated were separated during this analysis. This may result in an insufficient amount of enough genes being detected to significantly enrich specific GO terms. In conclusion, some signaling pathways involved in saul1-1 may not be detected in GO analysis because they contain up- and down-regulated genes.

Figure 23: GO analysis of early regulated genes in saul1-1. Hierarchical clustering of GO terms in saul1-1.

All GO term nodes are represented with the respective percentage of DEGs of a certain time point. Larger groups are defined by their color and the most significant enriched GO term. GO terms were determined as enriched, if the initial cluster consisted of at least 1 % of the respective DEGs and the final term had an adjusted p value ≤ 0.01. These were defined as connected, if their corresponding kappa score was higher or equal to 0.3. In addition, functional groups were clustered together if they shared 30 % or more of their genes. After all selection criteria were applied, 2183 (56.03 %) genes were associated to the pathways.

The clustering analysis identified the enrichment of four different groups of global GO terms. The two groups containing most GO terms were defense response to bacterium, incompatible reaction

and cellular response to salicylic acid stimulus (Figure 23, orange and yellow cluster). These pathways are highly connected to the ETI response. An incompatible reaction generally reflects a defense response against an infection with an avirulent pathogen, resulting in HR ²⁸⁰. Accordingly, an incompatible defense reaction here reflects the SOC3-dependent autoimmune phenotype of saul1-1 (orange cluster).

The saul1-1 phenotype has been reported to depend on SA ¹³⁵. In line with this SA dependency the largest GO term cluster was determined to be cellular response to salicylic acid stimulus (yellow cluster). An additional smaller array, involving three different metabolic pathways, toxin metabolic process, protein phosphorylation and camalexin biosynthetic process, respectively, was detected as well (blue clusters).

Preceding cluster analysis already pointed towards an elevated camalexin synthesis in saul1-1, however, only down-regulated genes were detected in this analysis (see chapter 4.1.6). This global GO network in saul1-1 autoimmunity identified additional up-regulated genes in this pathway.

Among others, CYP79B2 and CYP79B3 were detected to be up-regulated after 3 h. Both are involved in the conversion of tryptophan to indole-3-acetaldoxime, a precursor of camalexin ²⁸¹. Acting further downstream PAD3 gene expression was up-regulated after 6 h. PAD3 is a cytochrome P450 enzyme that catalyzes the final step during camalexin synthesis ²⁸¹. In conclusion, camalexin synthesis could be induced in saul1-1 as early as 3 h. Camalexin, as well as other phytoalexins, are hypothesized to prevent a secondary infection with necrotrophic pathogens ²⁸². This hypothesis was supported by up-regulation of camalexin synthesis genes as shown in other ETI-inducing experiments ²⁷¹. An involvement of camalexin in the onset of the HR could take place as well, because higher concentrations were found to be cytotoxic and could therefore be the cause of subsequent HR ⁵⁶. This is consistent with previous reports that HR may not be a finely tuned defense pathway, but a consequence of downstream signaling during the immune response ⁸⁵. Thus, the defense against a secondary necrotrophic infection during ETI may be highly important, and first regulatory mechanisms can be detected a long time before the onset of HR.

This is underlined by findings showing that cell wall thickening, which could function as a physical barrier to necrotrophic pathogens, was detectable 3 d earlier in saul1-1 than the lesioning of all aboveground organs. Accordingly, this analysis does not only confirm that saul1-1 exhibits an ETI-related phenotype, but gives insights into the early regulations and their complexity.

Figure 24: Excerpts of the GO analysis. Excerpts of GO terms, which contained several DEGs that were regulated after 1 h/2 h and 3 h. These are (A) response to oxygen-containing compound as part of the response to oxygen-containing compound cluster (red), which involved mostly DEGs being regulated after 1 h/2 h, (B) defense response part of the defense response to bacterium as part of the incompatible reaction cluster (bright orange), which involved mostly DEGs being regulated after 1 h/2 h, (C) response to hormone being part of the cellular response to salicylic acid stimulus cluster (yellow), which involved mostly DEGs being regulated after 1 h/2 h and (D) response to acid chemical being part of the cellular response to salicylic

This regulatory map was used thereafter to detect pathways containing DEGs, which were regulated early. These DEGs may function to trigger subsequent regulations (Figure 24). The detected GO terms, which contained at least three DEGs being regulated after 1 h/2 h, were response to oxygen-containing compound, defense response, response to hormone, response to acid chemical and organonitrogen compound metabolic process. The early regulated genes were THI2.1 (THIONIN2.1), ERF2 and AT3G28580. The comparison of these GO terms revealed that nearly all regulations were previously characterized as responses to signaling molecules, which are known to be essential immune regulators ²⁸³. This was consistent with the analysis of Cluster III (Figure 22) pointing towards the involvement of the interplay between JA and SA in early regulations in saul1-1 (see chapter 4.1.5). Increasing SA levels have been detected not before 12 h after the initiation of the saul1-1 phenotype ¹³⁵. However, it has been proposed that such regulatory networks may depend on low levels of these hormones only ²⁵². The regulation of more than 50 SA and JA responsive genes after 3 h may suggest that a putative rise in SA levels was already present but below the detection limit in previous experiments ¹³⁵. Therefore, an interplay of JA, SA, ET and ABA appears to be necessary to successfully initiate ETI.

When comparing these GO terms, an additional similarity was detected. Nearly all GO terms contain the same early regulated genes (Figure 24). This could hint towards their importance in the regulation of the ETI phenotype, because they have been associated with multiple regulatory pathways.

The THI2.1 gene encodes a thionin, which is expressed upon pathogen infection and has been used as a marker gene for increasing JA levels ^252,284. In addition, THI2.1 is involved in the interplay between JA and SA at low concentrations. It was shown that THI2.1 is detectable at low SA and JA concentrations and is down-regulated in response to higher amounts of SA ²⁵². This was also demonstrated in this RNA-seq experiment, since THI2.1 was not differentially expressed after 3 h and 6 h, but only after 1 h/2 h. Accordingly, THI2.1 could be important for the early regulations in saul1-1. Nevertheless, it was not possible to confirm the regulation of THI2.1 using qPCR (Figure 16). Thus, those results have to be interpreted carefully.

In contrast, the differential expression of ERF2 was confirmed by qPCR (Figure 16). Interestingly, the ERF2 gene was previously connected to defense response and has been reported to be ET-dependent ²⁸⁵. In addition, the clustering analysis has already shown that other members of the ERF family were also up-regulated in saul1-1 (Figure 22, Cluster III), emphasizing that this class of TFs may well be important for saul1-1 autoimmunity. The putative function of ERF2 and other TFs will be discussed below (see chapter 4.1.8).

The AT3G28580 gene encodes a putative AAA-ATPase. Whereas the protein function has not been shown yet, the AT3G28580 promoter has been shown to specifically respond to singlet oxygen and has been used in promoter-reporter gene constructs to detect singlet oxygen levels ^235,236. Singlet oxygen and other ROS are rapid signaling molecules, which are of high importance during the defense response ³². In addition, ROS contributes to HR in case of an incompatible defense response ⁸⁵. Hence, the up-regulation of AT3G28580 may point towards the involvement of ROS in the early regulation of the ETI observed in saul1-1. In silico analysis support this hypothesis, because AT3G28580 was shown to be up-regulated in plants overexpressing NLRs and in infection experiments ^186,187. In addition, further ROS responsive genes were detected after 3 h (89 genes) and after 6 h (320 genes), highlighting the importance of pathways, which are regulated by ROS.

The GO term analysis showed that already after 1 h/2 h different hormone-responsive pathways appeared to be active. This suggested that the initiation point of the ETI response could be a finely tuned interplay of small amounts of different plant hormones. These changes of hormonal levels could be very low and would therefore not be detectable with most techniques. Another problem with preceding infection experiments would be that they cannot be synchronized. This would result in the detection of averaged hormone levels. Thus smaller changes in hormone levels may be missed ²⁸⁶. Consequently, saul1-1 appears to be an ideal ETI model that for the first time suggests the presence of finely tuned hormone regulations resulting in the initiation of ETI.