Gene ontology (GO) analysis based on Redundancy Analysis
(RDA)
We identified the GO categories contributing to the differences in RDA
analyses by averaging the RDA loadings of the genes in each of the
categories, and tested whether the values significantly differ from
random scores obtained by permutation. Genes contributing to the
separation between inoculation and control were enriched for ABA and
cytokine signaling, primary metabolism and chloroplast activity. ABA
induces resistance to powdery mildew in barley (Wiese, Kranz, &
Schubert, 2004), and repression of ABA biosynthesis as well as genes
regulated by ABA, such as cold/dehydration/salinity responsive genes,
are associated with mildew resistance in nonhost plants in general
(Jensen et al., 2008). Cytokinin suppresses programmed cell death and
plays a role in the synthesis and maintenance of chlorophyll (Walters &
McRoberts, 2006); both processes were enriched among the differentially
expressed genes. Additionally, cytokinin levels regulate cell division
together with auxin. Interestingly, in Arabidopsis, Golovinomyces
orontii inoculation induced cell cycle related genes and
endoreduplication, possibly due to increased metabolic demands of the
pathogen (Chandran, Inada, Hather, Kleindt, & Wildermuth, 2010). On the
other hand, Choi, Choi, Lee, Ryu, and Hwang (2011) have shown that plant
based cytokinins systematically induce plant resistance against
pathogens by cytokinin and salicylic acid signaling.
Genes associated with the differences between the phenotypes showed GO
enrichments for kinase activity, carbohydrate metabolism, plant cell
wall organization, photosystem II and response to cold GO categories,
whereas the genes contributing to the differences between genotypes were
enriched for tryptophan metabolism, plant cell wall and chloroplast
(Figure 4a and Figure 2d/e; Supplementary Figure 3). In Arabidopsis
(Chandran et al., 2010), the expression of cold/drought responsive genes
were decreased together with ABA biosynthesis after inoculation withG. orontii . Together with the observed induction of ABA during
inoculation, this suggests that the phenotypes may differ in how
strongly ABA activates its targets such as cold responsive genes.
The responses to infection start to play a role in
genotype-by-experiment effect where the genes are mainly enriched for
immune and defense responses, illustrating that the genes in these
processes highly differ by the transcription levels between genotypes.
Overall, the enriched GOs show a clear activation of defense responses
in general, and defense responses to fungi in particular. The GO
category with highest positive average of RDA loadings (and therefore,
high contribution to separation) is aldose 1−epimerase activity
(GO:0004034) which, based on Sheshukova et al. (2017), is activated
because of the mechanical damage inflicted by the pathogen and results
in methanol emission and priming of the non-infected leaves. The
presence of hydrogen peroxide metabolic process and salicylic acid
mediated signaling pathway are both well-established pathogen-induced
defense mechanisms (Hua, 2009; Kuniak & Urbanek, 2000; Niu & Liao,
2016; Sheshukova et al., 2017), further demonstrating the activation of
the defense processes due to the pathogen infection. The GO category
with most negative average RDA loadings is RNA splicing, via
endonucleolytic cleavage and ligation (GO:0000394). An emerging field in
plant molecular biology, it is becoming increasingly clear that plants
use alternative RNA splicing extensively as a means to respond to their
environment and defend against pathogens (Shang, Cao, & Ma, 2017;
Staiger, Korneli, Lummer, & Navarro, 2013). Within the
signaling-specific GOs (Figure 4b) the genotype by experiment effect
showed the activation of jasmonic acid (JA) and abscisic acid signaling
(as expected, (Yang et al., 2019)), but again, in a genotype-specific
manner. Further inspection of putative orthologs of marker genes for
different hormonal signaling pathways showed activation of auxin
biosynthesis and signaling, as well as differences in the activation of
JA signaling and NLR signaling through EDS1 ortholog (Supplementary
Table 4). In R1, specific induction of the SA biosynthesis and systemic
acquired resistance was also detected.
The most significant contributor to phenotype-by-experiment is
photosystem II activity. The GO category with highest average RDA
loadings for phenotype by experiment effect is oligopeptide
transmembrane transporter activity (GO:0035673). The perception and
transduction of fungal oligopeptides will trigger multiple defense
responses (Hahlbrock et al., 1995; Nürnberger et al., 1994). Multitude
of photosynthetic processes were also enriched; their role in defense
was discussed above. The categories with most negative average loadings
were response to fungus, and cytokinin biosynthetic process
(GO:0009691).