Reprogramming of Strawberry (Fragaria vesca) Root Transcriptome in Response to Phytophthora cactorum

Crown rot (Phytophthora cactorum) causes significant economic losses in strawberry production. The best control strategy would be to use resistant cultivars, but polygenically inherited resistance makes the breeding of the garden strawberry (Fragaria × ananassa) challenging. The diploid wild strawberry Fragaria vesca Hawaii 4 genotype was shown previously to have resistance against crown rot. To explore the resistance mechanisms, we inoculated the roots of Hawaii 4 with P. cactorum in a novel in vitro hydroponic system to minimize interference caused by other microbes. Major reprogramming of the root transcriptome occurred, involving 30% of the genes. The surveillance system of the plant shifted from the development mode to the defense mode. Furthermore, the immune responses as well as many genes involved in the biosynthesis of the defense hormones jasmonic acid, ethylene and salicylic acid were up-regulated. Several major allergen-like genes encoding PR-10 proteins were highly expressed in the inoculated plants, suggesting that they also have a crucial role in the defense responses against P. cactorum. Additionally, flavonoids and terpenoids may be of vital importance, as several genes involved in their biosynthesis were up-regulated. The cell wall biosynthesis and developmental processes were down-regulated, possibly as a result of the down-regulation of the key genes involved in the biosynthesis of growth-promoting hormones brassinosteroids and auxin. Of particular interest was the expression of potential resistance genes in the recently identified P. cactorum resistance locus RPc-1. These new findings help to target the breeding efforts aiming at more resistant strawberry cultivars.


Introduction
Strawberry crown rot caused by the hemibiotrophic oomycete Phytophthora cactorum hampers strawberry production in many parts of the world. It causes stunting and wilting of the plants and spoils the fruits by causing leather rot [1]. The wilting symptoms first appear in the young leaves and as the disease progresses the whole plant may collapse. The most characteristic symptom is brown necrosis in the vascular tissue of the crown. As P. cactorum is a soil-borne pathogen and forms sexual oospores that may survive many years in the soil, the removal of the disease from the infested fields is difficult. The cultivation practices promoting soil drainage may help in managing the disease. In addition, fosetyl-Al, phosphite-based products or elicitors could provide some protection against crown rot [2]. However, the most effective method to control the disease is the use of resistant cultivars.
Strawberry displays resistance against most P. cactorum strains, and only specialized P. cactorum isolates have shown potential to cause crown rot [3,4]. On the other hand, many of the widely used strawberry cultivars are susceptible to specialized P. cactorum crown rot isolates, and only a few cultivars display resistance quite consistently [5,6]. Shaw et al. have suggested that resistance against crown rot pathotype is a polygenic trait [7,8]. Besides the genetic background, the physiological status of the plant plays a role: cold-stored and/or wounded plants are particularly prone to the disease, and young plants tend to be more susceptible than the older ones [9]. Eikemo et al. tested 60 diploid Fragaria sp. accessions for their resistance/susceptibility to P. cactorum [10]. The accessions varied from resistant to highly susceptible, but a majority was categorized as resistant or moderately resistant. Recently, Davik et al. analyzed the F 2 population of a cross between crown rot resistant and susceptible F. vesca genotypes (Bukammen and Haugastøl 3, respectively) [11]. They were able to identify a single major gene locus (named RPc-1, Resistance to Phytophthora cactorum 1), which explained 74.4% of the phenotypic variance. The 3.3 Mb QTL region contained 801 predicted genes, of which 69 may play a role in disease resistance. However, a major effort is still needed to identify those genes actually conferring crown rot resistance.
The defense mechanisms against strawberry pathogens have been extensively reviewed [12]. As in other plants, a two-layered inducible defense system probably functions also in strawberry. Immunity is induced, if the plant recognizes a highly conserved pathogen-or microbeassociated molecular pattern (PAMP or MAMP) with the help of a membrane-localized pattern-recognition-receptor (PRR), or if it perceives a pathogen effector with the help of a resistance (R) protein [13]. The recognition triggers a complex signaling network, leading to the reprogramming of gene expression and to the activation of defense responses, including the synthesis of antimicrobial secondary metabolites and the expression of pathogenesis-related (PR) genes. The magnitude of the defense varies depending on the pathogen, the effector-triggered immunity (ETI) being a stronger and faster response than the PAMP-triggered immunity (PTI). Plant hormones play important roles in the modulation of defense responses. In general, the salicylic acid (SA)-dependent signaling pathway is effective against biotrophic and hemibiotrophic pathogens and is required for the activation of hypersensitive cell death response, whereas the jasmonate (JA)-and ethylene (ET)-mediated defense responses act against necrotrophs [14]. Crosstalk between these two signaling pathways is often antagonistic, and defense responses can be modulated also by other plant hormones [15].
Until now, no large-scale transcriptome analyses on strawberry defense responses against P. cactorum have been reported. To gain a better understanding of the defense mechanisms, we inoculated F. vesca plantlets with P. cactorum zoospores, and sequenced the root transcriptome. The diploid woodland strawberry, F. vesca, is the wild relative of the octoploid garden strawberry. The accession Hawaii 4 (H4) was chosen as it is quite resistant against crown rot [10], and is the reference accession used in the genomic sequencing project [16]. In order to reduce biological variation between the replicates, micropropagated plant clones were used, and a novel hydroponic in vitro culture system was exploited to eliminate interference caused by other microbes (S1 Fig). The samples were collected two days post-inoculation, representing the early infection stage according to Chen et al. [17]. The data provide comprehensive insight into the transcriptional reprogramming that occurs in the roots of F. vesca upon inoculation.
To induce the root formation, the micropropagated plants were transplanted to MS agar without hormones. After 4 weeks, when the roots were sufficiently developed, the plants were transferred to aerated hydroponic culture in modified RITA1 containers (VITROPIC, Saint-Mathieu-de-Tréviers, France) (S1 Fig). Two plants of each clone (H4.4, H4.5, 'Senga Sengana') were placed in each container (six containers in total), and the roots were protected from light by wrapping aluminum foil around the lower part of the container.

Oomycete material
The P. cactorum isolate Pc407 originates from infected rhododendrons and is highly virulent to garden strawberry [19]. The P. cactorum isolate was maintained on potato dextrose agar (PDA) at the room temperature. To induce sporangia and zoospore production, the isolate was cultured in pea broth and soil extract water. Pea broth was prepared as described by Zentmyer and Chen [20]. Frozen green peas were mixed in a blender in deionized water (100 g/250 ml). After centrifugation, the supernatant was filtered through Whatman1 qualitative filter paper (GE Healthcare Bio-Sciences, Pittsburgh, PA, USA), deionized water was added to 500 ml, and the preparation was autoclaved (121°C, 15 min). The soil extract water was prepared according to Hamm & Hansen [21]. The soil (Puutarhamulta, Kekkilä Oy, Vantaa, Finland) was mixed in deionized water (1:1) and kept overnight at room temperature. After centrifugation and filtering through qualitative filter paper, the soil extract water was autoclaved.
The P. cactorum was grown in the pea broth (10 ml) in Petri dishes for three days at room temperature. To induce sporangia development, the cultures were washed three times with sterilized deionized water, and 10 ml of soil extract water was added. The plates were incubated overnight at room temperature under fluorescent light. The soil extract water was removed and the plates were kept under fluorescent light for another 24 h, after which plenty of sporangia were visible. Sterile deionized water (25 ml, 4°C) was added, and the cultures were chilled for 30 min at 4°C and then returned to room temperature. After 2 h, most of the zoospores were liberated, and the suspensions were decanted to sterile Erlenmeyer flasks. The zoospores were counted (Bürker chamber), and the concentration was adjusted to 5000 zoospores ml -1 by adding sterile deionized water.

Inoculation with P. cactorum zoospores and sample collection
For the inoculation with P. cactorum zoospores, the scaffolds, each holding six plants (S1 Fig), were transferred to RITA1-containers (modified for hydroponic growth) containing 390 ml of zoospore suspension or sterile deionized water (negative control). The roots were immersed in the liquid for 30 min, and then returned to half-strength Hoagland solution. All treatments were made in triplicate.
Samples for RNA extraction were collected 2 days after inoculation, as the hyphae were visible on the root surfaces of the inoculated plants (S6 Fig). Three biological replicates were collected, all from different containers. Roots, crown and leaves were excised with scalpel, weighed separately and snap frozen in liquid nitrogen. Samples were stored at-80°C.

RNA extraction
Roots of H4.4 plants were ground to fine powder in liquid nitrogen using a mortar and pestle, each individual separately.

Data analysis
The raw reads were first trimmed using Trimmomatic (version 0.32) [24]. The adapter sequences were removed, and low quality bases (phred quality score below 3) were deleted from both ends of the reads. The reads were also scanned with 4-base sliding window and cut if the average quality per base dropped below 15. Reads shorter than 36 bases were discarded. The trimmed reads were mapped against F. vesca nuclear (version v1.0) and chloroplast genome using STAR RNA-seq aligner (version 2.4.0) [16,25]. Maximum of ten mismatches were allowed, and the minimum and maximum intron lengths were set to 20 and 6000, respectively. Reads that were mapped to coding sequences (CDS) of annotated genes were counted using featureCounts with default settings [26]. Chloroplast genome annotation was released on 12.05.2011 and genome annotation was last updated on 04.03.2015. The unmapped paired-end reads were mapped against expressed sequence tag (EST) sequences of Phytophthora species (330 482 sequences downloaded from NCBI EST database) using Bowtie (version 0.12.7) [27].
Differential expression analysis was made using EdgeR [28]. Only the genes with a minimum expression level of one count per million (cpm) in at least three replicates were used. The genes were considered differentially expressed, if the false discovery rate (FDR) was < 0.05 and log 2 fold change < -1 or > 1.
Gene Ontology (GO) terms were searched for the longest protein sequence of each gene using PANNZER with default settings [29]. GO term enrichment analysis was made using BiNGO plugin in Cytoscape [30]. REVIGO was used to remove redundant terms [31]. Protein domains for receptor-like kinases were searched with Interproscan in Geneious 8.1.5 (http:// www.geneious.com) [32]. Separate GO term enrichment analysis was also made for the up-and down-regulated RLK gene sets, using the whole RLK gene set as a reference.

Results and Discussion
To understand the molecular events resulting from the interaction of P. cactorum with strawberry roots, we first needed to establish a suitable plant-pathogen pair. Of the P. cactorum isolate selection in our position, the isolate 407, originating from infected rhododendrons, was known to be highly virulent to garden strawberry [19]. It killed or nearly killed 80% of the inoculated garden strawberry cv. Jonsok plants in pathogenicity tests. When we tested the isolate on another strawberry cv. Senga Sengana known to be one of the most resistant cultivars against crown rot [5,6] We concluded that the P. cactorum isolate 407 together with F. vesca genotype Hawaii 4.4 served our purpose, as the isolate was not pathogenic to Hawaii but was pathogenic at least to two garden strawberry cultivars. The responses that would be seen were expected to indicate how F. vesca protects itself against P. cactorum and crown rot disease.

Overview of RNA sequencing results
Illumina sequencing generated 47.36-54.28 million paired-end raw reads (101 bp) per sample (Table 1). Of the read pairs, 94.3% passed the quality filtering, and 61.77-76.18% of those were uniquely mapped to coding DNA sequences (CDS) of the nuclear or chloroplast genome of F. vesca. In total, 104.79 and 97.24 million reads were uniquely assigned to CDS in water control and inoculated plants, respectively. Of the filtered reads, 4.50%, 8.23% and 1.67% aligned to Phytophthora EST sequences in the inoculated samples 1-3, respectively. Genes expressed at a low level were filtered out, leaving 16 793 genes for further analysis (S1 Table). Of these genes, 4576 and 4243 were down-and up-regulated (false discovery rate (FDR) < 0.05) in the inoculated plants compared to water controls, respectively, if no foldchange cut-off values were applied. However, to improve the reliability of the results, more stringent analysis was carried out with cut-off values of log 2 fold-change < -1 or > 1, resulting in 2993 significantly down-regulated and 2371 up-regulated genes (FDR < 0.05). These more stringent cut-off values were used below. The differences between the groups were mainly quantitative, since only 15 and 10 genes were expressed exclusively in the controls and inoculated plants, respectively. GO terms were assigned to 13 144 genes (78% of the genes) using PANNZER annotation program (S1 Table). GO term enrichment analysis of the expressed genes revealed 656 and 475 biological processes, 252 and 84 molecular functions and 37 and 85 cellular components significantly enriched (adjusted p-value < 0.05) in the up-and down-regulated gene sets, respectively (S2 Table). The most significantly (adjusted p-value < 1E-15) enriched biological processes are shown in S4 Fig. As one might expect, "response to biotic stimulus" was the most significantly enriched biological process among the gene set up-regulated in the inoculated plants. This confirms that the experimental set-up was successful. Other up-regulated biological processes included many GO terms associated with metabolic/biosynthetic processes, signaling, transport, responses to various stimuli, defense responses and regulation of immune processes (S2 Table). Salicylic acid, jasmonic acid and ethylene biosynthetic/metabolic processes were up-regulated as well.
On the other hand, "cell wall organization or biogenesis" was the most significantly downregulated biological process. Several GO terms related to polysaccharide metabolism, morphogenesis, development and growth were also down-regulated (S2 Table). Of the hormonal processes, auxin transport and brassinosteroid metabolic/biosynthetic processes were the most significantly down-regulated ones. Down-regulation of growth and development is a well-known phenomenon in pathogen-challenged plants [33]. When the resources are limited, the plant needs to prioritize the defense over development and growth in order to survive. However, constant, unnecessary activation of the defense responses at the expense of development is harmful for the fitness of the plant. In this study, growth repression caused by P. cactorum inoculation was apparently transient, as there were no significant differences in the biomass between the inoculated and uninoculated plants in the long-term experiments (S2 and S3 Figs). This minuscule effect on growth may be associated to the tight regulation of defense responses. In the next chapters, a closer look is taken to the genes related to the most significantly up-and down-regulated biological processes and the genes that could play important roles in plant immunity.
Genes expressed in P. cactorum resistance locus RPc-1 Of particular interest was the expression of 69 defense related genes, located in the QTL region associated with P. cactorum resistance in F. vesca [11]. Of these genes, 47 were expressed highly enough (at least one cpm in three replicates) to be included in the differential expression analysis ( Table 2). Seven of these genes were significantly down-regulated, while 19 were up-regulated upon P. cactorum inoculation. Many of the up-regulated genes, such as cyclic nucleotidegated ion channels, calcium-binding protein and kinases are probably involved in signaling events, whereas WRKY transcription factors are important transcriptional regulators of plant defense responses. Phenylalanine ammonia-lyase (PAL) catalyzes the first committed step in the phenylpropanoid pathway, leading to the biosynthesis of various polyphenolic compounds, and is arguably one of the central players in plant defense. However, the most interesting resistance gene candidates are receptor-like protein, receptor kinases and nucleotide-binding siteleucine-rich repeat (NBS-LRR) resistance proteins possibly functioning in pathogen perception. In total, four NBS-LRR protein-encoding genes (101306457, 101297569, 101300750, 101304699) were expressed in the RPc-1 locus, two of them being significantly down-regulated in the inoculated plants. The most highly expressed NBS-LRR gene was 101297569, which had much higher expression level compared to the other NBS-LRR genes, and can thus be considered as one of the resistance gene candidates in the RPc-1 locus. Receptor-like kinases (RLK) are discussed in more detail in the next chapter. Approximately 66% of these RLKs were differentially expressed upon P. cactorum attack. Among the up-regulated RLK gene set, several enriched GO terms were related to stress and defense responses, the most significant term being "response to salicylic acid stimulus". In contrast, most of the enriched GO terms in the down-regulated RLK set were related to the regulation of development and growth, indicating that the surveillance system of the plant was shifted from the development mode to the defense mode. Interestingly, two L-type-lectin-RLKs (101310048, 101309756), two G-type-lectin-RLKs (101305393, 101305094) and one receptor-like protein (RLP) (101290881), significantly upregulated in the inoculated plants in our study, are located in the QTL region associated with P. cactorum resistance in F. vesca [11] (Table 2). Especially, two L-type-lectin-RLKs (101310048, 101309756) were highly expressed and markedly up-regulated in the inoculated plants ( Table 2); they were also the most highly expressed L-type-lectin-RLKs in the inoculated roots of F. vesca. In addition, one gene (101292033) located in the RPc-1 locus, with typical L-typelectin receptor domain but lacking kinase domain, was constitutively highly expressed in the roots ( Table 2). As several studies have demonstrated the role of L-type-lectin-RLKs in Phytophthora resistance [34][35][36], these genes can be considered as the strongest resistance gene candidates in the RPc-1 locus. Some G-type-lectin-RLKs are involved in innate immunity as well, although they are better known for their role in the control of self-incompatibility in Brassicaceae [37]. For example, NgRLK1 is the interactor of Phytophthora capsici elicitin, capsicein, that is able to trigger a hypersensitive response [38]. Up-regulated RLP in the RPc-1 locus contains a leucine-rich repeat (LRR)-domain in the extracellular part but lacks a cytoplasmic kinase domain. As this gene was highly expressed in the inoculated plants and LRR-RLPs have been implicated in disease resistance in several plants [39], this gene is also a good resistance gene candidate in F. vesca. To unravel the potential role of these RLKs in the crown rot resistance, mechanistic studies should be carried out.
A closer look at the RLKs revealed a new group of proteins, which often contained wallassociated receptor kinase galacturonan-binding domain (IPR025287) in their extracellular part, although being otherwise more similar to pathogenesis-related 5 (PR5)-like receptor kinase-related (PTHR24420:SF561) subfamily in the PANTHER database. Over 80% of the members of this RLK group were up-regulated in F. vesca roots after P. cactorum inoculation (Table 3), suggesting an important role in defense responses. Wall-associated kinases (WAK)/ WAK-like (WAKL) may serve as sensors monitoring cell wall integrity, since they are directly cross-linked to cell wall pectins, but are also able to bind shorter pectin fragments released from the cell wall in pathogen attacks and activate defense responses [40]. We also identified an additional RLK group that contained LRR-domain followed by malectin-like carbohydratebinding domain (IPR024788). Some of the members of this group showed similarity to Arabidopsis LysM RLK1-interacting kinase 1, which regulates negatively the innate immunity triggered by chitin and flg22, and positively JA and ET signaling pathways [41]. Of the other RLK groups, the members of which were most often up-regulated, cysteine-rich receptor kinases (CRKs) may perceive extracellular ROS signals and activate intracellular signaling in response to abiotic or biotic stress [42].
Three RLK groups, in which the members were more often down-regulated than up-regulated, were LRR-RLKs, Catharanthus roseus RLK1-like kinases (CrRLK1Ls) and MLD-LRR-RLKs (Table 3). CrRLK1Ls regulate cell elongation and developmental processes in plants [43], and many of the down-regulated LRR-RLKs seem to control plant development as well. These data are in line with the results of GO term enrichment analysis indicating the down-regulation of developmental processes upon P. cactorum inoculation (S4 Fig). Examples of MLD-LRR-RLKs include symbioses receptor-like kinase (SYMRK) required for symbiotic plant-pathogen-interactions [44,45], and impaired oomycete susceptibility 1 (IOS1) that contributes to downy mildew disease development in Arabidopsis [46]. As some of the MLD-LRR-RLKs may function as susceptibility factors, it may be beneficial to down-regulate their expression upon pathogen attack.

Phenylalanine ammonia-lyase genes and positive regulators of SA signaling
The SA-related biological processes were up-regulated in F. vesca upon P. cactorum inoculation. In SA-rich plants such as poplar, rice and potato, SA synthesis mainly takes place through phenylalanine ammonia-lyase (PAL)-dependent phenylpropanoid pathway [47]. As strawberry leaves have high basal levels of SA [48], it seems probable that phenylpropanoid pathway is the main SA biosynthesis pathway also in Fragaria spp. Of the two PAL-encoding genes expressed in our transcriptome, one (101315259) is located in the P. cactorum resistance locus in F. vesca genome [11].
In Arabidopsis, SA is mostly synthesized through isochorismate synthase 1 (ICS1) [49], which has not been identified in F. vesca. However, some other gene homologs known to be involved in SA signaling in Arabidopsis were up-regulated in inoculated F. vesca roots. These included enhanced disease susceptibility 1 (EDS1) and phytoalexin-deficient 4 (PAD4) that function upstream of SA synthesis and potentiate SA signaling in a positive feedback loop [50]. EDS1 is able to form heterodimers with PAD4 and also with senescence-associated gene101 (SAG101) that has been implicated in plant immunity [51]. In our study, three SAG101-genes were up-regulated in the inoculated plants. In addition, two AGD2-like defense response protein 1 (ALD1) genes and two flavin-dependent monooxygenase 1 (FMO1) genes were upregulated. FMO1 is required for systemic accumulation of SA in systemic acquired resistance (SAR) along with ALD1, which is responsible for the synthesis of pipecolic acid, a long distance signal for SA [52]. The results suggest that systemic responses are activated in the inoculated F. vesca roots.

JA and ET biosynthesis and metabolism
The biosynthetic and metabolic processes of JA and ET were significantly enriched among the gene set up-regulated in the P. cactorum inoculated F. vesca roots. Two 1-aminocyclopropane-1-carboxylate (ACC) synthase-like (ACS) genes and at least two ACC oxidase-like (ACO) genes were up-regulated in the ET biosynthetic pathway. Also three ethylene response factor 1 (ERF1) genes were up-regulated. ERF1 is known to activate transcription of several PR genes in response to ET and JA and is an important transcription factor in plant defense [53]. Also the negative regulators of ET signaling were up-regulated upon P. cactorum attack: reversion-toethylene sensitivity 1 (RTE1) and two ET receptors (ethylene receptor 2 (ETR2) and ethylene response sensor 1 (ERS1)). These receptors serve as negative regulators of ET signaling, ensuring that the responses are down-regulated as soon as the ET level is diminished [54]. As the prolonged up-regulation of immune responses can be detrimental, this feedback regulation system is probably crucial for the plant.
Most genes in key steps of JA biosynthesis were up-regulated upon P. cactorum attack: three LOX-like genes, two allene oxide synthase-like (AOS) genes, two allene oxide cyclase-like (AOC) genes, and three 12-oxophytodienoate reductase 3-like genes (OPR3). Activation of negative regulation was also observed. Four negative regulators of JA-responsive genes, jasmonate ZIM-domain (JAZ) genes, were significantly up-regulated, whereas the Coronatine-insensitive protein 1 (COI1)-encoding transcripts were reduced (log 2 FC -0.955, FDR 4.12E-05). This could cause desensitization to JA-Ile, since COI1 functions as JA-Ile receptor, and forms SCF COI1 E3 ubiquitin ligase complexes with other proteins to recruit JAZ repressors for ubiquitination and degradation [55]. Similar responses have been observed in potato cultivar Ackersegen that displays quantitative resistance against Phytophthora infestans: concentrated P. infestans culture filtrate caused up-regulation of genes implicated in SA-related defense responses and JA biosynthesis, but also up-regulation of JAZ1 and down-regulation of COI1 [56]. This suggests that the seemingly contradictory responses could represent a highly functional strategy against Phytophthora pathogens.

WRKY transcription factors potentially involved in SA/JA crosstalk
In total, 21 genes encoding WRKY transcription factors were significantly up-regulated in response to P. cactorum inoculation. Four of these (101296502, 101297653, 101297362, 101311683) are located in the QTL region associated with P. cactorum resistance in F. vesca [11]. In Arabidopsis, several of these up-regulated WRKY transcription factors, including WRKY70, promote SA-related defense responses and suppress JA responses [57]. On the other hand, two genes showing similarity to Arabidopsis WRKY33, and two probable WRKY40 genes that function as negative regulators of SA/EDS1-signaling pathway and positive regulators of JA signaling were also up-regulated [58,59]. The results suggest that hormone crosstalk in F. vesca is highly complex and the defense responses are under strict control.

PR-10 proteins and flavonoid biosynthesis
In total, 18 genes with similarity to Mal d 1, Pru ar 1 or Pru av 1, major allergens of the Rosaceae family, were significantly up-regulated by P. cactorum. Four of these were among the 10 most highly expressed genes in the P. cactorum inoculated F. vesca. All of these proteins belong to the PR-10 pathogenesis-related protein family induced by various stresses [60]. Previous studies have shown that levels of PR-10 transcripts/proteins are elevated in Fragaria species challenged with pathogens such as Colletotrichum acutatum, Fusarium oxysporum f. sp. fragariae, Botrytis cinerea and Podosphaera aphanis [61][62][63][64][65]. The importance of these proteins in disease resistance is supported by the fact that their expression patterns differ between resistant and susceptible plants: some PR-10 proteins are expressed exclusively in the resistant genotype, or their up-regulation is faster or stronger compared to the susceptible genotype.
The PR-10 proteins can bind biologically important ligands in their hydrophobic cavity, including fatty acids, cytokinins, sterols and flavonoids [66]. For example, the strawberry Fra a allergens bind metabolic intermediates of flavonoid biosynthesis and thereby control their biosynthesis [67,68]. Thus, it is possible that the up-regulation of the PR-10 proteins in P. cactorum inoculated F. vesca is related to the enhanced synthesis of secondary metabolites. In accordance with this, the data suggest that the biosynthesis of phenylpropanoids and flavonoids were up-regulated (Fig 1). In particular, the biosynthesis of flavan-3-ols, such as catechin, was probably increased, since leucoanthocyanidin reductase-like gene was up-regulated, but flavanol synthase and anthocyanidin synthase were significantly and nearly significantly (log 2 FC: -0.92, FDR: 0.008) down-regulated, respectively. Previous studies have shown that catechin is one of the most abundant flavonoids in strawberry fruits and leaves [69][70][71]. Catechin seems to be involved in resistance against Alternaria alternata and Botrytis cinerea in the leaves and Strawberry Transcriptome in P. cactorum Attack fruits of strawberry, respectively [72,73]. We thus suggest that catechin or other flavan-3-ol derived compounds play a role in strawberry defense responses against P. cactorum.

Terpenoid biosynthesis
Several genes involved in terpenoid biosynthesis according to KEGG (Kyoto encyclopedia of genes and genomes) PATHWAY database [74] were up-regulated in P. cactorum inoculated F. vesca (Fig 2). This includes eight genes that catalyze the steps from acetyl-CoA to isopentenyl pyrophosphate (IPP) in the mevalonate (MVA) pathway. In contrast, none of the genes involved in the methylerythritol phosphate (MEP) pathway were up-regulated. The MVA pathway supplies precursors for the biosynthesis of sesquiterpenoids and triterpenoids and functions in the cytosol/endoplasmic reticulum/peroxisomes, whereas the MEP pathway is located in the plastids and gives rise to monoterpenoids, diterpenoids and tetraterpenoids [75]. Even though the MEP pathway was not up-regulated, there could be some changes in the monoterpenoid biosynthesis, as an (-)-alpha-pinene synthase-like gene was strongly up-regulated.
Genes involved in sesquiterpenoid and triterpenoid synthesis were strongly up-regulated. Indeed, the most highly expressed gene in the inoculated plants encodes (-)-germacrene D synthase-like (GDS) protein included in sesquiterpenoid biosynthesis (S1 Table). Three other GDS genes were also significantly up-regulated. Transcript levels of squalene synthase, five squalene epoxidases and two beta-amyrin synthase-like genes were significantly higher in the inoculated roots, suggesting that triterpenoid biosynthesis is increased upon P. cactorum inoculation. Alpha/beta-amyrin synthase gene, along with 3-hydroxy-3-methylglutaryl coenzyme-A synthase gene, were induced also in the fruits of garden strawberry 24 h after Colletotrichum acutatum inoculation [62]. Moreover, Hirai et al. have identified three antifungal triterpenoids from the inoculated and wounded fruits of strawberry cultivar that displays resistance against Colletotrichum fragariae [76]. These results suggest that terpenoid biosynthesis is part of the general defense responses in strawberry.
Triterpenoids can be further processed to triterpenoid saponins that consist of triterpenoid backbone with one or more covalently linked sugar moieties [77]. Prior to glycosylation, triterpenoid backbones are often oxygenated by cytochrome P450 enzymes. In our study, ten of the fifteen beta-amyrin 28-oxidase-like genes expressed in F. vesca were up-regulated, suggesting that triterpenoid saponin biosynthesis is induced in the roots in response to P. cactorum attack. In a non-targeted metabolite analysis, triterpenoid saponins and glycosylated sesquiterpenoid derivatives have been identified in the strawberry flower [78]. They are enriched in the sepals that protect floral buds from biotic and abiotic stresses, supporting an important role in defense. The molecular mechanisms of antimicrobial and antiherbivorous activities of saponins are not completely understood, but many of them are able to perturb the membranes [79].

Brassinosteroid biosynthesis
In contrast to terpenoid biosynthesis, some of the genes catalyzing the biosynthesis of steroids and brassinosteroids (BR) were down-regulated in F. vesca upon P. cactorum inoculation. This includes two genes encoding cycloartenol synthases (Fig 2), a gene encoding homolog of Arabidopsis DWF1 (DWARF1) and a gene encoding CYP85A2/BR6OX2 (BRASSINOSTER OID-6-OXIDASE 2). Cycloartenol synthases catalyze the first step in the pathway leading to campesterol and BR, i.e. the conversion of (S)-2,3-oxidosqualene to cycloartenol. DWF1 catalyzes the conversion of 24-methylenecholesterol to campesterol and CYP85A2/BR6OX2 catalyzes the last, rate-limiting step [80] in BR biosynthesis. We suggest that upon P. cactorum inoculation squalene is directed to triterpenoid biosynthesis rather than to BRs or other steroids. Furthermore, a homologue of Arabidopsis BR receptor, BRI1 (BRASSINOSTEROID INSENSITIVE1), was significantly down-regulated, presumably causing desensitization to BRs. Strawberry Transcriptome in P. cactorum Attack Down-regulation of BR biosynthesis and reduction of BR responsiveness is in agreement with the regulatory role of BRs in growth-immunity tradeoffs [81]. Even though BRs can enhance pathogen resistance in some cases, several reports indicate negative effects on immunity, in particular on the activation of PAMP-triggered immunity [82,83]. BR treatment also blocks benzothiadiazole (BTH)-induced resistance in rice [84]. It makes sense that BR synthesis would be diminished and BR responsiveness reduced in resistant F. vesca when inoculated with P. cactorum, as BR signaling could otherwise override defense signaling. In addition, it would be beneficial for F. vesca to channel isoprenoid metabolites to terpenoid biosynthesis rather than to BR synthesis.

Auxin biosynthesis and transport
The transcriptome results suggest a decrease in auxin biosynthesis in the inoculated plants. All four YUCCA genes of F. vesca (YUC1, YUC3, YUC5, YUC7) [85] expressed in the roots of the F. vesca H4.4 genotype, were significantly down-regulated. The YUC flavin monoxygenase-like proteins catalyze the rate-limiting step in the indole-3-pyruvic acid (IPA) pathway, which has been confirmed as the main auxin biosynthesis route in Arabidopsis [86,87]. According to GO term enrichment analysis, auxin transport was apparently down-regulated as well, and the expression levels of at least seven homologues of genes implicated in auxin transport in Arabidopsis were reduced in the inoculated F. vesca roots.
Auxin is an important regulator of development, and its impact on immunity can be negative or positive [88]. In general, auxin signaling seems to improve resistance against necrotrophs but enhance susceptibility against biotrophic pathogens, apparently because of the antagonistic relationship between SA and auxin. It has been reported that a mutation in WALLS ARE THIN1 (WAT1) gene, encoding a vacuolar auxin transporter, reduces auxin level and increases SA level in Arabidopsis roots, which confers resistance to several vascular pathogens [89,90]. As WAT1 gene was down-regulated in the inoculated F. vesca, it could play a role also in the defense responses against P. cactorum. Auxin may enhance the susceptibility also by other mechanisms, e.g. by modulating cell wall properties [91,92]. On the other hand, it has been shown that suppression of auxin response renders Arabidopsis more susceptible to Phytophthora cinnamomi [93]. Because of these contradictory findings it is difficult to conclude whether the changes in auxin-related processes promote or impede immunity in F. vesca.

Cell wall biosynthesis
Several GO terms related to cell wall biosynthesis, development and growth were enriched in the gene set down-regulated upon P. cactorum inoculation. Of the transcription factors regulating cell wall biosynthesis, two MYB46, VASCULAR-RELATED NAC-DOMAIN 4 (VND4) and SECONDARY WALL-ASSOCIATED NAC DOMAIN 2 (SND2) genes were down-regulated in the P. cactorum inoculated plants. In Arabidopsis, MYB46 acts as a master switch activating genes involved in the secondary cell wall synthesis [94], VNDs (VASCULAR-RELATED NAC-DOMAIN) are central activators of the secondary wall synthesis in the vessels [95], and SND2 regulates genes involved in cellulose and hemicellulose biosynthesis and lignin polymerization in fibers [96].
Of the cell wall biosynthetic genes, six out of eight cellulose synthase A catalytic subunit-like genes were down-regulated in the inoculated F. vesca (Table 4). Another group of genes systematically down-regulated was Fasciclin-like arabinogalactan proteins (FLAs) ( Table 4), which are suggested to have a role in cellulose deposition in the secondary cell wall [97]. Of the genes encoding Trichome birefringence-like (TBL) proteins, over 50% were down-regulated ( Table 4). The functions of TBLs are not well known, but some of them appear to contribute to the acetylation of cell wall polymers [98]. The observed down-regulation of several cell wall synthesis genes may promote the activation of defense responses, as mutant plants with defects in the cell wall synthesis sometimes display changes in hormone homeostasis and enhanced resistance to pathogens [99]. The only gene group involved in the cell wall synthesis that showed strong up-regulation in the inoculated plants, was the lignin-forming anionic peroxidase-like genes ( Table 4), suggesting that the plant attempts to compensate reduced cellulose synthesis by enhancing lignin deposition. This is in agreement with the finding that reduced cellulose synthesis leads to ectopic lignin synthesis in Arabidopsis [100].

Conclusion
This RNA-seq analysis provides comprehensive insight into the molecular survival mechanisms triggered in F. vesca roots upon P. cactorum attack, and emphasizes the balance between development and defense. Accordingly, cell wall synthesis is down-regulated, RLK-based surveillance system and biosynthesis pathways of hormones are reprogrammed and expression of genes involved in the biosynthesis of potential defense compounds is enhanced. Interesting changes were also seen in the recently identified P. cactorum resistance locus RPc-1. Particularly, we want to highlight NBS-LRR resistance gene (101297569), which was the most highly expressed NBS-LRR gene in this locus, and the L-type-lectin-RLKs that were highly expressed and strongly up-regulated in the inoculated roots. Because L-type-lectin-RLKs are known to play important roles in Phytophthora-plant interactions, these RLKs are among the most interesting resistance gene candidates in F. vesca and their function should be further characterized.
In addition, metabolomic studies should be carried out to confirm the changes observed in the biosynthesis pathways.  Fasciclin-like protein 13 0 9

Supporting Information
Laccase 33 9 15 Lignin-forming anionic peroxidase (TIF) S1 Table. Differential expression analysis. Expression levels (cpm) of F. vesca genes in the controls and in the inoculated roots, log 2 fold changes, false discovery rates (FDR) and GO terms. Expression data of all genes are presented on the first spreadsheet of the excel table and the discussed gene groups are presented on separate spreadsheets. (XLSX) S2 Table. GO term enrichment analysis. Enriched GO terms (biological processes, molecular functions and cellular components) in the up-regulated and down-regulated gene set. (XLSX)