Early Pep-13-induced immune responses are SERK3A/B-dependent in potato

Potato plants treated with the pathogen-associated molecular pattern Pep-13 mount salicylic acid- and jasmonic acid-dependent defense responses, leading to enhanced resistance against Phytophthora infestans, the causal agent of late blight disease. Recognition of Pep-13 is assumed to occur by binding to a yet unknown plasma membrane-localized receptor kinase. The potato genes annotated to encode the co-receptor BAK1, StSERK3A and StSERK3B, are activated in response to Pep-13 treatment. Transgenic RNAi-potato plants with reduced expression of both SERK3A and SERK3B were generated. In response to Pep-13 treatment, the formation of reactive oxygen species and MAP kinase activation, observed in wild type plants, is highly reduced in StSERK3A/B-RNAi plants, suggesting that StSERK3A/B are required for perception of Pep-13 in potato. In contrast, defense gene expression is induced by Pep-13 in both control and StSERK3A/B-depleted plants. Altered morphology of StSERK3A/B-RNAi plants correlates with major shifts in metabolism, as determined by untargeted metabolite profiling. Enhanced levels of hydroxycinnamic acid amides, typical phytoalexins of potato, in StSERK3A/B-RNAi plants are accompanied by significantly decreased levels of flavonoids and steroidal glycoalkaloids. Thus, altered metabolism in StSERK3A/B-RNAi plants correlates with the ability of StSERK3A/B-depleted plants to mount defense, despite highly decreased early immune responses.

enhanced resistance to P. infestans infection 13 . In parsley, biochemical analyses revealed that Pep-13 is recognized by a plasma membrane-bound receptor 10 . The specificity of eliciting defense responses by variants of Pep-13 is similar in parsley and potato, suggesting a similar mechanism of perception 10,12,13 .
The potato homologue of BAK1 was identified as a Pep-13-activated gene in microarray experiments. Transgenic plants with reduced expression of StSERK3A/B displayed altered morphology that was reminiscent of a brassinosteroid-deficiency phenotype and which correlated with differential accumulation of phenolics, flavonoids and sterols in untreated plants. Importantly, StSERK3A/B-RNAi plants were unable to activate early defense responses in response to Pep-13 treatment. Despite this, defense gene expression was induced by Pep-13 in StSERK3A/B-depleted plants.

Results
StSERK3B transcript levels are increased by Pep-13 treatment. StSERK3B was identified in microarray analyses 14,15 as a gene activated in response to treatment by Pep-13 in wild type, as well as in transgenic plants impaired in jasmonic acid biosynthesis (StAOC-RNAi and StOPR3-RNA) or perception (StCOI1-RNAi; Fig. 1A). The originally identified EST (MICRO.11825.C1) corresponds to PGSC0003DMT400032797, annotated to encode the receptor kinase SERK3B (Sotub01g042020; http://solanaceae.plantbiology.msu.edu/). The 60mer located on the potato chips 15 , corresponds to the 3′ untranslated region of StSERK3B, but not StSERK3A (Sotub10g013940). In subsequent qRT-PCR analyses, primers were used, which are predicted to amplify both StSERK3A and StSERK3B transcripts. These analyses revealed significantly enhanced StSERK3A/B transcript levels in Pep-13-infiltrated potato leaves four hours after treatment, which declined after 24 hours (Fig. 1B).
The protein coding regions of StSERK3A and StSERK3B are located on 11 exons on chromosome 10 and 1, respectively ( Supplementary Fig. S1A,B). The full length proteins StSERK3A (KJ625629, 615 amino acids) and StSERK3B (XP_006351807, 617 amino acids) display 79% sequence identity to AtBAK1 (At4G33430). Protein domain prediction programs describe a similar structure of StSERK3A and B to tomato SERK3B 7 , with a signal peptide, a leucine zipper region, four LRR domains, a proline-rich domain preceding a transmembrane domain and a C-terminal kinase domain ( Supplementary Fig. S1C,D). StSERK3A and B share 89% and 87% sequence identity at the amino acid ( Supplementary Fig. S1E) and nucleotide level, respectively.

Defense responses in StSERK3A/B-RNAi plants.
To assess the function of StSERK3A/B for Pep-13-induced defense repsonses, RNA interference constructs were generated targeting the 3′ end of the gene ( Supplementary Fig. S1B). Due to the high sequence similarity of StSERK3B to StSERK3A, the RNAi fragment is predicted to affect the expression of both genes. Transgenic potato plants expressing the RNAi construct were generated by Agrobacterium-mediated leaf disk transformation. qRT-PCR was performed with RNA from Pep-13-treated leaf disks of four independent transformants using primers that amplify both StSERK3A and StSERK3B transcripts. Significantly reduced levels of StSERK3A/B transcripts were detected in all plant lines ( Fig. 2A). To differentiate between StSERK3A and StSERK3B expression, gene-specific primers were used. These experiments revealed that, in wild type plants, StSERK3A is activated twofold in response to Pep-13, but generally expressed at lower levels than StSERK3B, whose transcripts increase threefold (Fig. 2B,C). Importantly, both genes were affected by the RNAi construct, since transcript levels after Pep-13 treatment were significantly lower in the RNAi compared to control plants. Despite this decrease, Pep-13-induced StSERK3A/B transcript levels in Data are derived from three independent experiments (n = 6). Statistical analysis of Pep-13-induced expression versus W2A treatment was performed using Mann-Whitney two-tailed U test (W2A versus Pep-13-treatment); **p < 0.01, ***p < 0.001. www.nature.com/scientificreports www.nature.com/scientificreports/ the RNAi plants were higher than those induced by W2A treatment (Fig. 2B,C), suggesting that residual levels of StSERK3A/B were sufficient to induce a weak Pep-13-specific response.
Since the RNAi fragment also showed similarity to StSERK1 (Sotub04g027320), StSERK1 transcript levels were determined in wild type, empty vector and StSERK3A/B-RNAi plants using gene-specific primers. StSERK1 transcripts did not accumulate in response to Pep-13 infiltration, nor did they show differences between control and StSERK3A/B-RNAi plants (Fig. 2D), suggesting that the RNAi fragment specifically reduced the levels of StSERK3A/B transcripts.
The formation of reactive oxygen species (ROS), the oxidative burst, is a hallmark of early defense responses. In a luminol-based assay, Pep-13 elicited the oxidative burst in wild type and empty vector plants, but not in StSERK3A/B-RNAi plants (Fig. 3A, Supplementary Fig. S2A). Application of the nearly inactive analog W2A did not lead to a strong ROS production (Fig. 3B, Supplementary Fig. S2B). The peptide elicitor flg22, whose activity is BAK1-dependent in Arabidopsis 16,17 , elicited a strong ROS burst in control, but not in StSERK3A/B-RNAi plants (Fig. 3C, Supplementary Fig. S2C), suggesting a requirement of StSERK3A/B for both PAMPs, Pep-13 and flg22. In contrast, the oligosaccharide chitin, a fungal PAMP, induced ROS formation in a StSERK3A/B-independent manner in all plants tested (Fig. 3D, Supplementary Fig. S2D).
The activation of defense-related MAP kinases was monitored by Western blot using an antibody specific for phosphorylated MAPK-pTEpY motifs. Pep-13, but not W2A, induced the activation of a MAP kinase of about 48 kD in wild type and empty vector plants (Fig. 3E, Supplementary Fig. S2E). Importantly, MAPK activation was highly reduced in all StSERK3A/B-RNAi lines tested, indicating that StSERK3A/B are required also for this early defense response.
Despite the inability to mount an oxidative burst and to activate MAP kinases, enhanced levels of transcripts of selected defense genes were detected in Pep-13-treated leaf disks. While Pep-13-induced expression of StSERK3A/B was highly reduced in StSERK3A/B-RNAi plants (    (Fig. 5A). These included dwarfism in tissue culture, darker green leaves with a crinkled surface and leaf curling, resulting in a reduced expansion of the leaves. A delay in senescence was accompanied by reduced numbers and weight of tubers compared to control plants, leading to decreased overall tuber yield (Fig. 5B). The striking phenotype of the StSERK3A/B-RNAi lines is reminiscent of a brassinosteroid-deficiency phenotype observed in other plants 18,19 .
To further characterize differences between wild type and   Table S1). Despite this discrepancy, all three experiments were evaluated for the data shown in Figs. 6 and 7.
With more than 2000 metabolite features detected, we observed changes in branches of the phenylpropanoid pathway. Hydroxycinnamic acid amides, typical defense compounds of potato, were present at enhanced levels in StSERK3A/B-RNAi lines (Supplementary Table S1, Fig. 6). The fold changes varied from only minor increases up to factors of more than 5 fold, as visualized with bar charts for N-feruloyltyramine, p-coumaroylagmatine and caffeoylputrescine ( Fig. 7A-C). The biogenic amines, putrescine and agmatine, were not detected in our experiments, whereas the levels of tyramine, a precursor of N-feruloyltyramine, were significantly enhanced.
In contrast to the enhanced levels of specific hydroxycinnamic acid amides, those of a number of coumarin and flavonoid compounds were significantly reduced (Fig. 6, Supplementary Table S1). The highest reduction was observed for esculin, a glycoside of the coumarin esculetin, with 10 fold lower levels in StSERK3A/B-RNAi plants. Similarly, flavonoids such as kaempferol and quercetin derivates displayed significantly reduced abundance (Figs. 6, 7D, Supplementary Table S1). Another class of compounds with reduced abundance in StSERK3A/B-RNAi lines was identified as chlorogenic acid derivatives. Four peaks with identical MS/MS were detected ( Supplementary Fig. S3) with reduced abundance in the RNAi lines, suggesting that these chlorogenic acid-like compounds are derived from the same pathway. Finally, the steroidal glycoalkaloids solanine and chaconine, identified by MS-MS and analytical standards, were both significantly lower in StSERK3A/B-RNAi plants (Fig. 7E,F; Supplementary Table S1). In summary, elevated levels of hydroxycinnamic acid amides correlated with a concomitant reduction in the levels of coumarin and flavonoid compounds, suggesting that the common precursor of these pathways, coumaroyl-CoA, is preferentially converted by the HCAA branch of the phenylpropanoid pathway in StSERK3A/B-RNAi lines.

Discussion
Reduced ROS formation and loss of MAP kinase activation in StSERK3A/B-RNAi lines suggests a requirement of StSERK3A/B for perception of Pep-13 in potato. ROS formation is a hallmark of early defense responses to pathogen and PAMP treatment 20 . In Arabidopsis, perception of PAMPs by a receptor complex comprising the PRR and BAK1 has been shown to activate the cytoplasmic RLK BOTRYTIS INDUCED KINASE 1 (BIK1), which www.nature.com/scientificreports www.nature.com/scientificreports/ subsequently phosphorylates and activates the ROS-forming enzyme RBOHD 21,22 . In Arabidopsis bak1 mutants, the oxidative burst and MAP kinase activation in response to treatment with the PAMPs flg22 or elf18 are significantly reduced 16,17,23,24 , highlighting the importance of AtBAK1 for PAMP responsiveness. However, the degree of reduction varies in different bak1 mutants 17 . Moreover, in response to bacterial infection, Arabidopsis bak1-4 mutants still show a reduced oxidative burst, suggesting redundancy 25 . Indeed, a double mutant defective in BAK1 (bak1-5) and the gene encoding the LRR-RLK SERK4/BKK1 shows even higher reduction in ROS formation and MAPK activation than bak1-5 alone 23 . Searches for SERK4 homologous sequences from potato revealed highest sequence homology to PGSC0003DMP400047882 (Sotub04g027320), which is annotated as SERK1 in the potato genome database (http://solanaceae.plantbiology.msu.edu). Since transcript levels from this gene are not affected by the RNAi construct (Fig. 2D), we conclude that StSERK3A and B are required for Pep-13-induced ROS formation and MAP kinase activation in potato.
Despite the inability of StSERK3A/B-RNAi plants to accumulate ROS in response to Pep-13 (Fig. 3A,B) and to activate MAP kinases (Fig. 3E), they show defense gene activation upon treatment with Pep-13 (Fig. 4), which is similar to or even higher than that in wild type plants. Thus, the early responses that occur within minutes, i.e. ROS formation and MAPK activation, are clearly different from the later responses that are detectable after hours, i. e. defense gene activation. This is in contrast to reports from other plants in which reduced BAK1 expression also affects late responses, such as PAMP-induced cell death or growth inhibition. For example, the cell death response to the Phytophthora elicitin INF1 was reduced in Nicotiana benthamiana plants that were transiently silenced for NbBAK1 expression 8 . Also, Arabidopsis bak1 mutants displayed reduced growth inhibition in response to flg22 17 . On the other hand, in accordance with our data, potato plants silenced for BAK1 with a StSERK3A-specific RNAi construct showed Pep-13-inducible expression of three defense genes 26 , which led the authors to conclude that Pep-13 induces immunity in a SERK3/BAK1-independent manner. Our data do not support this conclusion, since Pep-13 neither induces ROS formation, nor activates MAPK in StSERK3A/B-depleted plants. Thus, our data show that perception of Pep-13 is dependent on StSERK3A/B.
The activation of defense responses in a BAK1-depleted background has been reported before 27 . In Arabidopsis bak1 mutant plants, defense gene activation and cell death is elicited by treatment with endogenous plant peptide signals, such as Pep2 28 , which act as damage-associated molecular patterns (DAMPs). Apparently, Arabidopsis can sense the absence of BAK1 and responds with the activation of immune responses 29 . Thus, similar to elicitation by Pep2 in Arabidopsis bak1 mutants, Pep-13 treatment of StSERK3A/B-depleted potato plants results in the activation of immune signaling.
The morphological alterations that were observed in all StSERK3A/B-RNAi plants might be a consequence of enhanced activation of immune responses, i.e. autoimmunity. In general, autoimmunity is accompanied by reduced growth, enhanced levels of salicylic acid, constitutive expression of defense genes as well as spontaneous lesion formation 30 . In potato, such a phenotype was observed in plants with reduced expression of StSYR1, a syntaxin required for the formation of callose-containing papillae 31 . In contrast, the StSERK3A/B-RNAi plants described here did not display spontaneous lesions (Fig. 5), nor constitutive defense gene expression (Fig. 4). Rather, the phenotype of StSERK3A/B-RNAi plants is more reminiscent of brassinosteroid-deficiency, with darker  www.nature.com/scientificreports www.nature.com/scientificreports/ Along with the striking phenotype of the StSERK3A/B-RNAi plants, major alterations in the metabolite pattern of untreated transgenic plants compared to control plants were observed. The central precursor for the different branches of the phenylpropanoid pathway, 4-coumaroyl-CoA, was differentially channeled into the formation of HCAAs, while flavonoid levels were reduced (Fig. 6). As typical phytoalexins of Solanaceous plants, levels of HCAAs were at least threefold higher in the transgenic lines. Caffeoylputrescine, a compound which is found in a number of Colorado Potato Beetle-resistant wild species of Solanum 35 , is more than 25 times more abundant in StSERK3A/B-RNAi lines than in control lines. These observations correlate with reports that Arabidopsis brassinosteroid-deficient mutants contain higher amounts of aliphatic and indolic glucosinolates, typical defense compounds of Arabidopsis 36 . Moreover, exogenous application of brassinosteroids to Arabidopsis seedlings reduced the levels of glucosinolates, suggesting that brassinosteroids negatively affect defense compounds 36 . Our observation that StSERK3A/B-RNAi plants contain higher levels of defense metabolites is also in accordance with the analysis of Arabidopsis serk1-3serk3-2 roots, which had higher levels of aliphatic glucosinolates as well as 4-methoxy-3-indol-3-ylmethyl glucosinolate 37 . The latter compound is a substrate of the atypical myrosinase PEN2 38,39 , which is required for penetration resistance against nonhost pathogens, such as P. infestans 40 .
In Leaf disk and infiltration assay. Leaf disks were cut out from 4-week-old potato plants with a biopsy puncher (4 mm diameter) and placed with the abaxial side onto the surface of 250 µl of water in a 96-well plate. The plate was incubated overnight at 22 °C in the dark. Water from the wells was removed and 100 µl sterilized fresh water per well was added. The plate was incubated for 30 min in the phytochamber (20 °C). Elicitation was performed by adding 5 nM Pep-13 or the nearly inactive analog W2A 11 . For whole plant assays, PAMP treatment was performed by infiltrating a 100 µM elicitor solution into the abaxial side of leaves of 3-week-old potato plants growing in a phytochamber. RNA expression analyses. RNA was isolated from potato leaves or leaf disks as described 12 . DNase digestion (RNase-free DNase Set, Qiagen) and cDNA synthesis using Maxima H Minus First Strand cDNA Synthesis Kit (Thermo Fisher Scientific) were performed according to the manufacturer's instructions. For quantitative PCR, Maxima Probe qPCR MasterMix (Thermo Fischer Scientific) was used and the samples were run on an Mx3005P qPCR system (Agilent).
ROS assay. ROS analyses were performed as described 44 with the following modifications: Each well contained 200 µl of water supplied with 5 µM luminol L-012 (Wako Chemicals), 2 µg horseradish peroxidase (Fluka) and 5 nM Pep-13 or W2A peptide. Immunoblot analysis. Protein extraction was performed as described 45 .
Liquid chromatography-mass spectrometry measurements. Leaf disks from 4-week-old potato plants were cut out and methanolic extracts were prepared as described 46 .
To reveal a comprehensive MS/MS library for structural annotations of compounds, the autoMS/MS method of the Bruker Otof control software was optimized. Ions were selected for MS/MS according to their intensity (highest first) and an intensity threshold of at least 500 counts, isolation width 0.5 Da, active exclusion after 2 spectra, reconsideration of excluded ions after 5 spectra. Preferred charge was 1+ or 2+, single charged ions were fragmented with 15 eV collision energy, double charge ions with 25 eV. For selected metabolites, the MS/MS collision energy was modified for optimized fragmentation.
Metabolite profiling was performed in MetaboScape 3.1 (Bruker Daltonics). Data files were assigned to sample groups of StSERK3A/B-RNAi lines "ABSF" and controls "empty vector and WT" with distinction of biological sampling in year 2016, 2017, 2018. The following settings were applied: Peak picking from 0.6 min to 10.5 min with intensity threshold: 1000 counts; minimum peak length: 9 spectra; re-extract feature if detected in 21 of 139 analyses; consider feature if found in 40 of 139 analyses or in 75% of samples in one sample group; mass calibration from 12.05 min to 12.2 min on lithium formate. [M + H] + was set as primary ion, [M + Na] + annotated as an adduct if the EIC correlation was above R = 0.9. Peak area was selected as an indicator for feature abundance. AutoMSMS data were mapped on the intensity matrix with the following settings: m/z tolerance 100 ppm, retention time (RT) tolerance 6 seconds.
Quantification of alpha-solanine (m/z 868.505, RT 5.8) and alpha-chaconine (m/z 852.510, RT 6.0) was performed on extracted ion chromatograms in QuantAnalysis 4.4 (Bruker), because peaks were not merged and quantified correctly by automated processing due to their width and peak shape.
Identification and annotation of the compounds was based on comparison of m/z and retention times to analytical reference standards, comparison of MS/MS patterns to analytical standards, published former annotations of hydroxycinnamic acid amides 47 and interpretation of tandem mass spectra in combination with retention time systematics (Supplementary Table S1, Supplementary Figs. S3-S6).
Metabolic pathways were created in PathVisio 3.0.0 48 and the log2 fold change of StSERK3A/B-RNAi vs. controls (wildtype, empty vector) was mapped via color code using MetaboScape (Bruker). For pathway mapping, all data of the experiments 2016, 2017, 2018 were combined; single experiment fold changes and p-values (Student's t-test) were calculated in Excel and presented as Supplementary Table S1. Data in Fig. 7 was processed using GraphPad Prism 7.04 (www.graphpad.com).