Genomic structure and transcript analysis of the Rapid Alkalinization Factor (RALF) gene family during host-pathogen crosstalk in Fragaria vesca and Fragaria x ananassa strawberry

Rapid Alkalinization Factors (RALFs) are cysteine-rich peptides ubiquitous within plant kingdom. They play multiple roles as hormonal signals in diverse processes, including root elongation, cell growth, pollen tube development, and fertilization. Their involvement in host-pathogen crosstalk as negative regulators of immunity in Arabidopsis has also been recognized. In addition, peptides homologous to RALF are secreted by different fungal pathogens as effectors during early stages of infection. Previous studies have identified nine RALF genes in the diploid strawberry (Fragaria vesca) genome. This work describes the genomic organization of the RALF gene families in commercial octoploid strawberry (Fragaria × ananassa) and the re-annotated genome of F. vesca, and then compares findings with orthologs in Arabidopsis thaliana. We reveal the presence of 15 RALF genes in F. vesca genotype Hawaii 4 and 50 in Fragaria x ananassa cv. Camarosa, showing a non-homogenous localization of genes among the different Fragaria x ananassa subgenomes. Expression analysis of Fragaria x ananassa RALF genes upon infection with Colletotrichum acutatum or Botrytis cinerea showed that FanRALF3-1 was the only fruit RALF gene upregulated after fungal infection. In silico analysis was used to identify distinct pathogen inducible elements upstream of the FanRALF3-1 gene. Agroinfiltration of strawberry fruit with deletion constructs of the FanRALF3-1 promoter identified a 5’ region required for FanRALF3-1 expression in fruit, but failed to identify a region responsible for fungal induced expression.


Introduction
In plants, several small secreted peptides (SSPs) function as hormonal signalling molecules that respond to internal and external stimuli [1]. SSPs are known to be involved in different processes, ranging from organ growth to biotic and abiotic responses [2,3].
Rapid alkalinization factors (RALFs) are cysteine-rich SSPs originally identified for their ability to rapidly alkalinize tobacco cell culture [4]. They are ubiquitous in the plant kingdom with 37 members identified in Arabidopsis thaliana genome alone [5,6]. RALF genes are translated as pre-pro-proteins and are activated in the apoplast through proteolytic cleavage, generating~50 amino acid long active peptides. Aside from the signal sequence necessary for extracellular extrusion, canonical RALF peptides contain distinctive amino acid motifs, such as the RRILA motif for S1P protease recognition [7] and the YISY motif, important for the activation of the signaling cascade [8,9]. In addition, four conserved cysteines form two disulfide bonds that stabilise the mature RALF peptide. Based on these features, RALFs have been grouped into four major clades [6]; clades I, II and III contain typical RALF peptides, whereas clade IV includes the most divergent RALF proteins lacking, both the RRILA and YISY conserved motifs and in some cases containing only three of the four cysteines.
RALF peptides bind to the Catharanthus roseus Receptor Like Kinases 1-like family proteins (CrRLK1L) known to be involved in cell expansion and reproduction throughout the plant kingdom [10]. The large CrRLK1L receptor family which includes FERONIA (FER) proteins, previously were reported to interact with Arabidopsis RALF1 and RALF23 [11,12], BUDDHA'S PAPER SEAL 1 and 2 (BUPS1/2), ANXUR1, 2 (ANX1/2) proteins interact through their ectodomain and bind to RALF4 and 19 in the pollen tube [13] and THESEUS1 (THE1) the RALF34 receptor in the root [14]. Binding RALF peptides to CrRLK1L receptors also involves other interacting partners such as Lorelei-like-Glycosylphosphatidylinositol-Anchored proteins (LLG1,2,3) [9,15] and Leucine-Rich Repeat Extensins (LRX). The latter has been reported to bind RALF4/19 in the pollen tube [16,17] and to interact with FER, as part of cell wall sensing system responsible for vacuolar expansion and cellular elongation [18]. Binding of RALFs to their receptors leads to a number of different intracellular signaling events involving different molecular components, mostly still unidentified. However, it is known that in A. thaliana binding of RALF1 to FER receptor results in the phosphorylation of plasma membrane H(+)-adenosine triphosphatase 2, inhibition of proton transport and subsequent apoplastic alkalinization [11].
RALF peptides regulate a variety of different functions such as cell expansion [11], root growth, root hair differentiation [19,20,21] stress response [22], pollen tube elongation and fertilization [13,16,22,23]. In A. thaliana, RALF peptides can also act as negative regulators of the plant immune response following bacterial infection [12]. For example, RALF23 binding to the FER receptor inhibits physical interaction between the immune receptor kinases EF-TU RECEPTOR (EFR) and FLAGELLIN-SENSING 2 (FLS2) with their co-receptor BRASSINOS-TEROID INSENSITIVE 1-ASSOCIATED KINASE 1 (BAK1), inhibiting immune signalling. Interestingly, biologically-active RALF homologs have also been identified in fungal plant pathogens, possibly following horizontal gene transfer, suggesting a role for fungal RALF genes as virulence factors [24]. In fact, a fungal RALF homolog is required for host alkalinisation and infection by Fusarium oxysporum [25]. Because alkalinisation is important for activation of virulence factors and successfully infection of plant tissues for many pathogenic fungi, secreted RALF peptides may promote host alkalinisation at early stage infection when hyphal biomass is not sufficient to secrete a large quantities of ammonia [24] Strawberry is an important crop but is susceptible to fungal and bacterial pathogens that can have a significant effect on marketable yield [26]. Post-harvest molds are particularly difficult to manage as infection initiates during flowering, and can become problematic after a long asymptomatic quiescent period on ripe fruits [27]. Severe strawberry fruit post-harvest molds are caused by Colletotrichum acutatum, causal agent of anthracnose disease, and Botrytis cinerea, causal agent of grey mold [28].
F. × ananassa fruits infected with C. acutatum show increase in the accumulation of a RALF transcripts in the ripe susceptible fruit at an early stage of infection [29]. Overexpression of the F. × ananassa ortholog of A. thaliana RALF33, through transient agroinfiltration, increases strawberry fruit susceptibility to anthracnose, leading to increased fungal growth on fruits and the induction of the host immune response [30,31]. Upregulation of RALF transcripts during plant infection has also been observed in mature red tomato fruits (Solanum lycopersicum) upon infection with Colletotrichum gleosporioides [32] and in rice (Oryza sativa) upon Magnaporthe oryzae infection [31]. These findings suggest a role for RALF gene expression as a susceptibility factor in fungal infection. Furthermore Dobón et al. [33], studying the expression pattern of four Arabidopsis transcription factors mutants (at1g66810, pap2, bhlh99, zpf2) with increased susceptibility to B. cinerea and Plectosphaerella cucumerina, observed a coincidental increase of RALF23, RALF24, RALF32 and RALF33 transcripts.
The woodland strawberry RALF gene family members have been previously characterized based on the F. vesca genome annotation v1 [6], and nine FveRALF genes were identified and grouped in the four RALF clades. Analysis of an updated and annotated version of the Fragaria vesca genome (v4.0.a2) [34] and of the Fragaria x ananassa cv. Camarosa genome sequence (v1.0.a1) [35,36], allowed exploration of RALF gene family composition and genomic organization. Genome localization, phylogenetic and transcript analyses were conducted, based on available genomic and transcriptomic sources, with an aim to gain insights into the functional roles of specific RALF genes. Furthermore, the induction of RALF gene expression upon infection of F.× ananassa fruits with C. acutatum and B. cinerea was studied. In silico analysis of the FanRALF3-1 promoter was conducted in order to identify putative pathogen responsive motifs. We then tested in vivo if progressively truncated FanRALF3-1 promoter fragments could induce reporter genes expression in agroinfiltarted strawberry fruits infected with C. acutatum.

Materials and methods
Identification of RALF family genes, phylogenetic analysis, and chromosome assignment F. vesca RALF genes identified in v1.1 genome and reported by Campbell and Turner [6] were detected in the Genome Database for Rosaceae (GDR; v4.0.a2) through keyword gene search 'RALF' and through a BLASTn search using the previously annotated RALF sequences against the F. vesca v4.0.a1 chromosome database. The v4.0.a2 [34] and the Blastn gene search outputs were then used to update the previous data on v1.1 genome. Nucleotide sequences of the 15 F. vesca RALF members (both previous and new found ones), were used as query for BLASTx on the F. × ananassa cv. Camarosa Genome v1.0.a1 proteome [37] to find octoploid RALF orthologs. Sequences of F. × ananassa RALF peptides were aligned with MUSCLE [38] and the phylogeny was inferred using the Maximum Likelihood method and JTT matrix-based model [39]. The tree with the highest log-likelihood was chosen. The Initial tree was obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model, and then selecting the topology with a superior log-likelihood value. Evolutionary analyses were conducted in MEGA X [40]. To define the classification of the new members in different clades, all the RALF protein sequences available from the Campbell and Turner annotation F. vesca genes were aligned with the newly identified RALFs and a phylogenetic tree was constructed as mentioned above. Fragaria x ananassa genes annotation and position were retrieved from GDR (F. × ananassa cv. Camarosa genome v1.0.a1), and progenitor lineage were inferred according to [36]. Chromomap package in R [41] was used to create a FanRALF gene chromosome map.

Expression profile of RALF family genes in F. vesca
Transcript accumulation of RALF family genes from different F. vesca tissues was depicted using heatmap3 package in R [42] from Transcripts Per Kilobase Million (TPM) values calculated by Li et al. [34,43,44,45,46].

Infection of Fragaria x ananassa fruits
F. × ananassa cv. Alba plants were grown in the greenhouse at 25˚C and 16 h of light. White (21 d after anthesis) and red fruits (28 d after anthesis) were harvested and infected. Each treatment contained at least three biological replicates. Colletotrichum acutatum (Isolate Maya-3, from CRIOF-UniBo fungi collection) and Botrytis cinerea strain B05.10 were grown on PDA plates for 15 d. Detached fruits were dipped for 30 s in a 10 6 conidia per mL suspension or in water as negative control and incubated in plastic bags for 24 h at approximately 21˚C (room temperature) as reported in Guidarelli et al. (2011) [29].

RNA extraction and qRT-PCR analysis
The surface of experimental fruits was excised with a scalpel and immediately frozen in liquid nitrogen. RNA was extracted according to Gambino et al. [47], run on an 2% agarose gel and quantified with NanoDrop™ 3300 for integrity and quality control, respectively. The cDNA was prepared from 1 μg of RNA using Promega ImProm-II™ Reverse Transcription system. Quantitative RT-PCR analysis was performed using ThermoFisher MAXIMA SYBR GREEN/ ROX QPCR 2x supermix. The relative accumulation of RALF transcripts was calculated using standard curve method and Elongation Factor 1 gene as reference (XM_004307362.2) [48]. Primers for RALF gene expression analysis were designed on F. × ananassa, F. vesca subgenome sequences and specificity was verified by observing a single peak in the dissociation curve for each primer pair. All primers used for gene expression analyses are listed in S2 Table. Statistical analyses RALF and eGFP reporter transcripts were quantified as the average of three independent biological replicates, each formed by a group of at least three treated fruits. For eGFP quantification, all mean normalized expression values were expressed relatively to the negative empty pKGWFS7 vector infiltrated fruits. Student t-test was used to assess statistical significance between samples and controls.

In silico prediction of regulatory elements in FanRALF3-1 and FanRALF6-1 promoters
Analysis of regulatory elements required for pathogen-induced transcripts was performed by examining pathogen-upregulated and downregulated transcripts found in datasets of red strawberry (F. × ananassa) fruits 24 h post infection with C. acutatum [29] and B. cinerea [49]. Promoters of 87 upregulated and 36 downregulated genes were analysed in response to C. acutatum infection, whereas 97 upregulated and 16 predicted promoters were analysed B. cinerea infected fruits. For each gene, 1500 bp upstream the ATG start codon were analyzed from the F. vesca genome v4.0.a1 assembly. MotifLab software v1.08 [50] was used for in silico analysis using PLACE database for Motif Scanning [51], and AlignACE algorithm for Motif Discovery [52] and motif similarity analysis algorithm for motifs comparison. Briefly, Motif Scanning was performed using the Simple Scanner program with default parameters. A regulatory element was considered statistically significant to expression by performing the same Motif scanning program on randomly generated DNA sequences starting from input predicted promoter sequences using a third order background model. The frequency measured for each cis-acting element on random DNA was used as background occurrence for statistical significant evaluation using a binomial test with p-value threshold of 0.05. Motif Discovery was performed using AlignACE method with default parameters and motif significance were calculated as mentioned above for Motif scanning method.

Fragaria × ananassa RALF3 promoter characterization
Preliminary assessment of putative promoter allelic variants were conducted on F. × ananassa cv. Alba and cv. Florida Elyana. Genomic DNA was extracted from leaves using Wizard 1 Genomic Purification kit (Promega). The Plant Tissue Protocol (Manufacturer protocol 3.E.) was modified adding two consecutive chloroform:isoamyl alcohol (24:1) purification steps after Protein Precipitation solution was added and before 2-propanol precipitation. The Fan-RALF3-1 putative promoter was amplified using primers For 5'-TGCATCTGTTACATCAT CCCTTG-3' and Rev 5'-GTAGTCGACTCTCCCATCTTG-3', cloned into pGEM 1 -T easy vector (Promega). Five clones for each variety were sequenced and aligned with the F. × ananassa cv. Camarosa genomic sequence available from GDR, using Clustal Omega.

Cloning of progressively truncated promoters and agrobacterium-mediated transient transformation
The upstream sequence of FanRALF3-1 was PCR-amplified starting from F. × ananassa cv. Elyana genomic DNA, using primer 5'-GGGGACCACTTTGTACAAGAAAGCTGGGTNCTGA AAGGACAAAAC ATTTTCT-3' as the reverse primer for all promoter fragments, 5'-GG GGACAAGTTTGTACAAAAAAGC AGGCTNNTGCATCTGTTACATCATCCCTTG-3' as the forward primer for the complete promoter fragment (T6), 5'-GGGGACAAGTTTGTACAAAA AAGCAGGCTNNTGCTTAAGTGGCTCTCAAAG-3' as the forward primer for the 400 bp fragment (T4) and 5'-GGGGACAAGTTTGTACAAAAAAGCAGGCTNNCCGCTAAGTGGTTCAA TTCA-3' as the forward primer for the 200 bp fragment (T2). Truncated FveRALF3-1 promoter constructs and a double p35S promoter used as a positive control were cloned into pDONR222 using the Gateway BP reaction and consequently cloned into pKGWFS7 (S1 Fig) vector by LR Reaction. The resulting plasmids were then introduced into chemically competent Agrobacterium tumefaciens strain EHA105 by heat shock transformation. Briefly, liquid nitrogen-frozen cells were thawed, and then incubated for 5 min at 37˚C with 1 μg of plasmid DNA. The cells were then incubated at 30˚C for 2 h in LB medium with shaking and plated. Positive colonies were then grown in selective media (Rifampicin 100 μg/mL and spectinomycin 50 μg/mL) until the culture reached an OD 600 of 0.8. Cells were collected by centrifugation and resuspended in fresh MS medium (Murashige and Skoog Basal Medium 4.4 g/L plus 20 g/ L sucrose) and grown to OD 600 = 2.4. At the end, acetosyringone (4'-Hydroxy-3',5'-dimethoxyacetophenone) was added to the culture to final concentration of 200 μM. At least three white attached fruits for each condition were injected with about 3-5 mL of the Agrobacterium solution using a needle and syringe until culture filled strawberry fruit tissues. Five days after agroinfiltration, fruit were harvested and infected with C. acutatum conidial suspension or mock-inoculated with water, as was described above. After 24 h the fruits were dissected and one half was used for RNA extraction and eGFP transcript analysis, and the other half was used for histochemical assay of GUS activity.

3D modelling of FanRALFs interaction with MRLK and LLG2 proteins
Homology models of FanRALF3, in complex with FERONIA and LLG2 were generated using Modeller [53]

Identification of RALF gene family members in fragaria vesca
RALF proteins belonging to 51 plant species have been previously classified in four clades based on sequence conservation [6]. Typical distinctive amino acid sequence motifs, such as the RRILA proteolytic cleavage site [7] and the YISY receptor binding site, are present in RALF peptides of clade I to III, and missing in clade IV, which contains more divergent members of the family. Nine RALF genes have been previously identified in the Fragaria vesca v1 genome.
In order to classify the members of RALF gene family in Fragaria vesca, the recent genome annotation (v4.0.a2) [34] was searched using 'RALF' as keyword gene name, revealing the presence of 13 RALF genes. These F. vesca RALF (FveRALF) [54] genes are named with progressive numbers according to their chromosome position, from Chr1 to 6 (Table 1), with one RALF gene in Chr1, two in Chr2, -3, and -5, and three in Chr4, and -6. No RALF genes are found in Chr7. Out of the nine RALF genes previously reported by Campbell and Turner [6], eight genes are confirmed both for identity and chromosome position. These are the Table 1). The gene00145, previously annotated as gene encoding for a protein with the typical RALF motifs RRILA and YISY [6], was discarded since in the new v4.0.a2 annotation its sequence corresponds to gene FvH4_6g07633 encodes for a shorter protein lacking most of the conserved motifs.
In order to find new, previously unannotated RALF genes, a BLASTn analysis was performed on the genome of the F. vesca (v4.0.a1 assembly) [55], using the 13 known FveRALF genes as query sequences. The BLAST search identified a putative new FveRALF gene located on chromosome 2 (Fvb2:15886205..15886408) and homolog to FveRALF10 and coding for a protein lacking the RRILA cleavage site and the YISY active site, and having only two out of four cysteines at conserved positions, but presenting the RALF conserved domain in the C-terminal part of the protein sequence (S2 Fig). This hypothetical new RALF gene was named FveRALF14 (Table 1). Furthermore the BLASTn analysis revealed the presence of another predicted new FveRALF gene, with two RALF conserved domains both homolog to Arabidopsis AtRALF19 and alternatively translated in frame 1 or frame 2. This gene was annotated also in the previous V4.0.a1 version of F. vesca genome (FvH4_3g15010) and here named FveRALF15 ( Table 1).

Identification, evolution and chromosome organization of RALF genes in F. x ananassa
F. vesca RALF gene sequences were used as query against a database of predicted proteins (v1.0.a1 Proteins source [35]) in F. × ananassa cv. Camarosa (Fxa). Fifty RALF members were identified in the Fxa octoploid strawberry (S1 Table). Paralogous genes (orthologs to particular V4.0.a2 annotation according to Li et al., [35]; Former annotation corresponds to v1.1 (geneN˚) or v4.0.a1 (FvH4_N˚) [51]. Asterisks ( � ) indicate those genes that were previously identified in v1 F. vesca genome [6] and confirmed in the v4.0.a2 annotation version; clade classification according to Campell and Turner [6]. https://doi.org/10.1371/journal.pone.0226448.t001 FveRALF genes) were identified in the Fxa subgenomes from the alignment and phylogenetic analysis ( S4 Fig and Fig 1A). Progenitors lineage was inferred from chromosome localization ( Fig 1B) according to the prediction by Edger et al. [35]. Fourteen genes out of these 50 are localized in the F. vesca subgenome, 15 members in the F. nipponica subgenome, 13 in the F. iinumae subgenome and only eight in the F. viridis subgenome (Fig 1C). It is important to note that these assignments were based on the best subgenome match in the 'Camarosa' genotype. This is one accession and octoploid strawberry subgenomes structure may not be the same among all genotypes at this ploidy.
Fxa RALF genes (FanRALF) were named based on corresponding FveRALF orthologs and the subgenome localization, with progressive numbers from 1 to 4 to indicate F. vesca, F. iinumae, F. nipponica and F. viridis progenitors, respectively (accordingly to Edger et al. [35]), and progressive letters to nominate genes orthologous to the same FveRALF genes and localized on the same chromosome (e.g. FanRALF8-2a is one of the two Fxa genes orthologous to FveR-ALF8 mapping on F. iinumae subgenome).
Gene homology and chromosome localization analysis showed that only four out of 13 FveRALF genes, namely FveRALF1, FveRALF4, FveRALF7, FveRALF9, have orthologs in all the four subgenomes. For all the other cases the FveRALF orthologs are not represented in all the  Table. Progenitor lineage were inferred from gene chromosome location (b), and pink was used for F. vesca subgenome, green for F. nipponica, light blue for F. viridis and orange for F. iinumae. The color legend also reports the code number used to name different FanRALF genes according to subgenome lineage, as is reported in S1 Table.  different Fxa subgenomes, probably due to gene loss events occurred during evolution or polyploidy adjustment (Fig 1A). In particular, the F. viridis derived subgenome has the lowest number of RALF gene members and is lacking genes orthologous to FveRALF2, FveRALF5, FveRALF6, FveRALF8, FveRALF10, FveRALF11 and FveRALF14. Similarly to F. vesca, no RALF genes are localized in Chr7 of the different progenitors. It is likely that in the Fxa genome some of the RALF genes are the result of duplication events. For example Fan-RALF11-1a and FanRALF11-1b are both orthologous to FveRALF11 and are positioned close together on Chr6 (F. vesca subgenomes Fig 1B). Another atypical gene organization is found for FanRALF15 and FanRALF4-1 genes, located on Chr3 (F. vesca subgenome), which are annotated as single genes but contain two tandem RALF conserved domains, suggesting that a duplication event occurred during genome evolution. In addition, the FanRALF7-4 gene is predicted to encode a 325 aa protein containing a conserved RALF domain within the first 104 amino acids and a domain homologous to chloroplastic NADPH-dependent aldehyde-reductase like protein, from aa 124 to 325. There are many potential explanations for how this novel domain architecture could occur, and it will be interesting to test for functional relevance of this particular variant.
The F. viridis derived subgenome contains the fewest detected putative RALF genes. This was also the case for R gene family in Fxa [35]. However, in contrast to the R gene family, there is not a clear dominance of F.vesca progenitor in the FanRALF gene family composition, since genes are similarly distributed in the F. iinumae, F. nipponica and F.vesca subgenomes ( Fig 1C). As was speculated by Edger et al. [35] the lack of RALF genes in F. viridis subgenome could be related to the higher TE content of this subgenome which can cause both higher mutation rates and gene loss. Fxa RALF gene classification in specific clades agrees with the structure in diploid woodland strawberry, with 21 genes in clade IV, 17 genes in clade III, 12 in clade I-II (S1 Table).

Transcriptome dataset analysis of RALF genes in F.vesca
To provide insights into the RALF gene members functions in strawberry, RNA-seq datasets that have been mapped onto the new genome annotation v4.0.a2 by Li et al. [34] were analyzed in different tissues and developmental stages. RALF members were grouped based on similar expression profile and hierarchical clustering resulted in four major RALF expression groups (Fig 2): i) RALF genes specifically expressed in mature male gamete (FveRALF4, FveRALF5,  FveRALF10, FveRALF11); ii) a gene expressed only in two anther developmental stages (FveR-ALF2) iii) FveRALF3 and FveRALF12 genes mainly expressed in roots, and in roots after two days after infection with Phytophthora cactorum iv) and genes mainly expressed in different fruit developmental stages (FveRALF1, FveRALF6, FveRALF7, FveRALF8, FveRALF9 and  FveRALF13).
Contrary to what has been observed in Arabidopsis, where clade IV RALF genes were highly expressed in flower tissues [6], the woodland strawberry FveRALF transcripts included in each of the four expression groups belong to different clades, suggesting that members of the same clade are have different roles in different contexts. The highly specific expression of four FveR-ALF genes in male gamete and late stage of anther development (FveRALF4, FveRALF5, FveR-ALF10, FveRALF1) shown by the heatmap (Fig 2), suggests a role for these RALF genes in the ovule-pollen, cell-cell communication during the sequence of events precisely regulated during fertilization. A recent study reports that in Arabidopsis AtRALF34 gene, expressed in the ovule, competes with RALF4 and RALF19, expressed in the pollen tube, for binding to BUPs and ANXs receptors [13]. The interaction between AtRALF34, expressed by the female gamete, and receptor complex formed by BUPS1/2 and ANX1/2, present on pollen tube membrane leads to pollen tube rupture and sperm release [13]. The competitive binding of AtRALF4 and AtRALF19 to this receptor complex suggest that they have a redundant function in regulating pollen tube growth and integrity [16]. It is possible that these RALF genes functional redundancy is conserved in woodland strawberry.
Among the FveRALF transcripts expressed in flower and fruit organs, FveRALF1 and FveR-ALF7 are the most abundant at the early stage of development in shoot apical meristem (SAM), floral meristem (FM) and receptacle meristem (RM), with FveRALF1 also being the family member most highly expressed in the leaves. As for fruit, FveRALF8 is the gene most highly expressed in mature fruits, whereas FveRALF6, FveRALF7, FveRALF9 are detected during fruit growth and in the mature organ at 15 d post anthesis (15 DPA) both in Yellow Yonder and Red Rugen genotypes (the two F. vesca genotypes used for RNA seq) and FveRALF3 and FveRALF13 transcripts are most abundant in the mature fruits at 20 DPA. FveRALF1, FveR-ALF6, FveRALF7 and FveRALF13 expression decreases during fruit development both in the inner and the outer tissues of fruit while FveRALF9 expression gradually increases with fruit development.
In a recent work, Jia et al. [56] analyzed the expression of the woodland strawberry (F. vesca) Malectin Receptor Like Kinases (MRLK) also known as the Catharanthus roseus RLKlike proteins (RLK1Ls). F. vesca MRLKs are encoded by more than 60 genes, and more than 50% of these are expressed during fruit development. The majority of fruit FvMRLK genes are expressed at high level only at the early stage of fruit ripening, and decrease at ripe stages. Transiently silencing and overexpression of MRLK47 in strawberry fruit, severely affects ripening regulation [50]. Moreover, a recent report describes how MRKL47 changes the sensitivity of ripening-related genes to ABA, a key hormone for strawberry fruit ripening [57]. Consistently, both RALF and ABA were found to be FERONIA-mediated cross-talk signals in stressresponse and cell growth in Arabidopsis [58]. FERONIA receptor is known to be important for cell-wall integrity and in Ca 2+ signaling [59], both known to be important during fruit growth and ripening [60]. The similar expression profile of FveMRLKs to the one we observe here for FveRALF1, FveRALF6, FveRALF7 and FveRALF13 supports the idea that the RALF-MRLK signalling plays an important role for also for strawberry fruit development. Future studies will test the interaction between RALF peptides and FveMRLK in vivo.

Expression profile of RALF genes in Fragaria x ananassa fruit and induction upon pathogen infection
RALF peptides are known to play a role in plant-pathogen interaction [24] since they were found to negatively regulate plant immunity response in Arabidopsis [12]. They were also found to be secreted by fungal pathogen as crucial virulence factors [25,61]. Furthermore, it was reported that genes homologous to AtRALF33 were upregulated both in tomato (Solanum lycopersicon) and commercial strawberry (F. × ananassa) susceptible ripe fruits interacting with C. gloeosporioides and C. acutatum, respectively [29,31,32].
FanRALF3 transcripts show a significant increase in abundance upon infection with either pathogen in the susceptible ripe stage, whereas FanRALF9 expression decreases in white fruits upon C. acutatum but not upon B. cinerea infections (Fig 3). The expression of FanRALF8 and FanRALF13 were not affected by infection, neither in white nor in red fruits (Fig 3). Out of the other FanRALF genes analyzed, only FanRALF6 shows a clear downregulation in infected fruits at both ripening stage and in response to both pathogens, while FanRALF1 and Fan-RALF7 gene expression is significantly decreased only in red ripe stage with B. cinerea and in white stage with C. acutatum, respectively.
The expression profiles of fruit RALF genes on Fxa strawberry fruits infected with two post-harvest pathogens are consistent with our previous results [30]. This report showed that transient overexpression of a FanRALF33-like gene (here now named FanRALF3) affected disease susceptibility. Notably, the FanRALF3 and FanRALF13 genes encode mature peptides differing only by two amino acids (S2 Fig) but accumulate differently to pathogen infection, suggesting that the genes possess subfunctionalized promoters that allow responses to different stimuli, or that the two amino acid difference leads to contrasting roles for the two different peptides.
Furthermore, N-terminal sequence alignment of FanRALF1-1, FanRALF3-1, FanRALF7-1, FanRALF8-1a and FanRALF13 peptides with AtRALF23 ( Fig 4A) shows that residues directly involved in AtRALF23 binding with AtLLG2 are conserved, suggesting that FanRALF peptides in strawberry fruits may be the strawberry orthologs in the LLG-MRLK heterotypic complex. Consistently, the homology models of FanRALF3-1 peptide interaction with FanMRLK47-  the MRLK mostly expressed in Fxa fruits [56] and with the Fxa LLG2 homolog (maker-Fvb3-4-snap-gene-34.65), show that the structural components necessary to bind MRLK47 and LLG2 are conserved in FanRALF3 (Fig 4B and 4C), suggesting a similar binding mechanism and complex formation. In Arabidopsis, AtRALF23 binding to FERONIA and LLG proteins leads to negative immunity response regulation, and this hypothesis can be tested by analysis of FanRALF3-1 binding to MRLK47 in fruits.
Finally, the expression of FanRALF1, FanRALF6 and FanRALF9 decreases during infection, thus it is possible that these FanRALF gene members may play different roles than RALF3 homologs during plant-pathogen interaction and immune response.

FanRALF3 promoter analysis in Fragaria x ananassa subgenomes and varieties
As shown above and reported in previous studies, RALF gene expression is triggered by different biotic and abiotic stimuli, however the signaling events regulating its expression are not yet known. Identification of the promoter elements necessary for gene induction by fungal pathogens could provide a basis to identify the transcription factors and other mechanisms involved in immunity signaling and ultimately provide the necessary knowledge to develop synthetic pathogen-responsive promoters to fight infections. Among FanRALF family genes, FanRALF3 shows increases in transcripts in response to pathogen, and its overexpression in strawberry fruits is related to susceptibility [30]. For this reason FanRALF3 was chosen for promoter truncation analysis.
To study allelic variability, the FanRALF3-1 putative promoter sequence similarity was assessed also in genomes of F. × ananassa varieties with different susceptibility to fungal pathogens, the cv. Florida Elyana from Florida (U.S.A.) which is resistant to anthracnose disease [62], and cv. Alba, an italian variety which is highly susceptible to C. acutatum infection (https://plantgest.imagelinenetwork.com/it/varieta/frutticole/fragola/alba/59). The promoters were amplified with specific primers and five clones for each variety were sequenced and aligned with that of the v1.0.a1 genome sequence cv. Camarosa (S5 Fig). Only a single nucleotide polymorphism was detected between cvs. Alba and Florida Elyana. This suggests that the function associated with the FanRALF3-1 5' upstream sequence in octoploid strawberry may be based on small differences in sequence, or that factors upstream of the promoter are the basis of the mechanism of resistance.

Prediction of FveRALF3 promoter pathogen-responsive regulatory elements
The FanRALF3-1 (from F.vesca subgenome) putative promoter sequence was chosen for pathogen-responsive regulatory element analysis because of the reported F. vesca subgenome dominance in octoploid genome [35]. Since Fa-and FveRALF3 upstream putative regulatory sequences share 99% level of identity and the available Fxa pathogen-responsive transcriptome data have all been mapped onto F. vesca genome, the analysis of the FanRALF3-1 promoter regulatory elements responsive to pathogen infection were carried out on F. vesca genome FveRALF3 promoter.
The 656 bp genomic sequence upstream of FveRALF3 ATG start codon was analysed using truncation analysis. FveRALF3 putative promoter sequence was compared with known transcription factor binding sites of genes known to be regulated in Fxa strawberry fruit upon C. acutatum and B. cinerea infections [29,49] in the PLACE database [51] (Motif Scanning analysis). These latter sequences and FveRALF3 putative promoter were then scored for motif frequency (Motif Discovery analysis) (Fig 5).
Motif scanning analysis of cis-acting elements enriched for fungal-induced transcripts revealed the presence of an element initially identified as initiator of PsaDb gene promoter (INRNTPSADB, PLACE ID: S000395). The putative promoter lacks a TATA-box, which is common to FveRALF3 and the majority of the 5' upstream sequence (78 out of 87 (90%)) of the C. acutatum responsive genes promoter sequences and 92 out of 97 (94%) of B. cinerea genes (Fig 5). Other TATA-like elements such as TATABOX2 (PLACE ID: S000109) and TATAPVTRNALEU (S000340), which have the role of recognition and initiator of transcription complex [63,64] were also found in several putative promoter sequences (59 (67%) from C. acutatum upregulated genes and in 41 (42%) of B. cinerea induced genes). Among the group of genes upregulated by C. acutatum, 33 (37%) and 47 (57%) predicted promoters possess MYBPLANT (S000167) [65] and MYBZM (S000179) [66] regulatory elements, which are binding sites for MYB-type transcription factors. Additional analysis on the putative promoters of Arabidopsis RALF33 (AT4G15800] and its homologous in tomato (S. lycopersicum Solyc09g074890.1), revealed the presence of the same binding site (data not shown). MYB proteins are a large family of transcription factors involved in regulation of many processes in plants, such as phenylpropanoid metabolism [67], ABA and JA signaling [68], responses to abiotic and biotic stress [69], cell death and output from the circadian clock. MYB proteins generally interact with basic helix-loop helix (bHLH) family member and are regulated by cytosolic WD40 repeat proteins through formation of MYB/bHLH/WD40 dynamic complexes, which regulate various gene expression pathways [70]. Interestingly, MYB46 is involved in enhancing B. cinerea resistance through down-regulation of cell wall associated genes (CESA) during early stage infection [71]. Arabidopsis T-DNA insertion mutants of genes regulated by MYB46, such as the a zinc-finger containing protein gene zfp2, the Basic Helix-Loop-Helix TF bhlh99, the AUX/IAA-type transcriptional repressor pap2 and the At1g66810 gene coding for a Zing Finger Transcription Factor, showed enhanced susceptibility to the necrotrophic fungal pathogens B. cinerea and P. cucumerina [33]. The transcript analysis of these four mutants revealed a coincident upregulation of RALF23, RALF24, RALF32 and RALF33 [33] showing that MYB46 and RALF genes are part of a co-expression network, and possibly functionally related.
A Motif Discovery analysis was performed using AlignACE algorithm [52] on the putative promoter sequences of both FveRALF3 and the identified C. acutatum and B. cinerea strawberry upregulated genes, and significantly overrepresented motifs were assessed. The motifs identified using MotifLab software are named 'AlignACE' followed by progressive numbers. For C. acutatum genes group the AlignACE00085 element (consensus GxTxxxTGTGAxTTG) was found in 20 (23%) predicted promoters, and is partially overlapping with MYBZM elements at the position between bases 57 and 62 of FveRALF3 putative promoter. The AlignACE00098 (xxTGGxCTTGG) element was found in the 14 (16%) C. acutatum upregulated genes and align to the elements AlignACE00021 (TTGGxxTTGG) found in B. cinerea upregulated group (19% of sequences analyzed). It is tempting to speculate that the position of this element might be a regulatory component important for FveRALF3 fungal-induced expression. Furthermore in B. cinerea gene group the AlignACE00040 (AAxxTTGAxxGxxxAA), AlignACE00048 (TxGAAxxGGTC) and AlignACE00034 (AxTGAAxTGA) motifs, located between bases 262 and 317 in FveRALF3 putative promoter, were found significantly overrepresented.
Besides upregulation of FanRALF3, both pathogen infections led to the downregulation of FanRALF6 transcript accumulation (Fig 3), suggesting a role also for this gene during fruit defence response. In order to explore why these two genes showed opposite behaviour and to highlight the presence of similar cis-regulatory sequences in FanRALF3 and FanRALF6 promoters, the predicted promoter sequence of FanRALF6 was also scored for the precence of annotated cis-acting elements and motif discovery by analysing sets of downregulated genes in datasets from C. acutatum and B. cinerea infected fruit tissues (Fig 6A).
Similar to the FanRALF3-1 predicted promoter, the FanRALF6-1 upstream sequence (F. vesca progenitor) has very high identity (99%) with F. vesca (data not shown), thus the regulatory element analysis was performed considering 1500 bp from ATG of FveRALF6 gene. Contrary to the high numbers of genes found upregulated upon C. acutatum and B. cinerea infections, examined in the FveRALF3 promoter analysis, the number of genes downregulated similarly to FveRALF6 is limited to 36 for C. acutatum and 16 for B. cinerea. The motif scanning analysis revealed that 100% of downregulated genes during C. acutatum early infection (Fig 6B) contain the INRNTPSADB element (S000395, [63]), found also in upregulated genes analysis (Fig 5), and known to particiapte in transcription initiation. Analysis also uncovered the "REalpha" element (REALPHALGLHCB, S000362), a feature initially identified in Lhcb21 gene promoter of duckweed (Lemna gibba) [72,73], that is involved in phytochrome regulation. This element is present in 30 out of 36 (83%) sequences analyzed with the consensus sequence AACCAA, and shares high similarity with sequences recognized by MYB family TFs. The motif has been identified in promoters of stress responsive genes, such as MYB1AT (S000408, aAAC-CAA) and MYBATRD22 (S000175, CTAACCA) (S3 Table). MYB binding motifs (MYBPLANT and MYBZM) were also found in FanRALF3 and other genes upregulated upon fungal pathogen infection (Fig 5B). These preliminary analyses suggest a role of these conserved sequences in pathogen responsive gene regulation. On the other hand, the motif scanning analysis of B. cinerea downregulated genes did not show any significant result (Fig 6C).
In addition, the motif discovery analysis revealed the occurrence of conserved motifs among C. acutatum downregulated genes ( Fig 6B) and in particular AlignACE00001 (AAAAAxAAAx) motif was found in 28 (77%), AlignACE00020 (TTTggATGAA) in 12 (33%), AlignACE00045 (TGxagAxtGAAa) in 16 (44%), AlignACE0004 (GAGAGAaAxxa) in 18 (50%) and AlignACE00026 (ctCGCGGxga) in 9 (25%) out of 36 sequences analyzed. The AlignACE00001 consensus motif (AAAAAxAAAx) is a A-reach sequence that has been reported to correlate with scaffold attachment regions [73] and in chromatin loop domain organization and transcription [74]. On the other hand, the AlignACE00004 motif (GAGAGAaAxxa) shows the typical sequences recognized by GAGA-binding proteins involved in chromatin silencing modification [75]. Motif discovery analysis performed on B. cinerea induced-downregulated genes revealed the enrichment of AlignACE00015 motif (gxTTTxATGaxg) which was found in 9 out of 16 (56%) sequences analyzed. . Codes and names of regulatory elements are reported, together with elements percentage abundance among sequences analyzed (Sequences), total count of elements found (total), element background frequency calculated as number of elements found in a third order random generated sequences (Exp), ranking score values calculated by MotifLab software using a binomial test with p-value threshold of 0.05 (Score). Consensus sequence of each element is shown along the position of the element in the putative FanRALF6 promoter (Position bp) and the strand in which it is found. https://doi.org/10.1371/journal.pone.0226448.g006

FveRALF3 promoter-reporter assay
In order to explore the functional relevance of the FanRALF3 putative promoter elements idenfied, three progressive truncated fragments of the FanRALF3-1 upstream sequence were cloned into pKGWFS7 vector and fused to two tandem reporter genes eGFP and β-glucoronidase (GUS) (S1 Fig). Deletions were 200 bp (T4) and 400 bp deletion (T2) (Fig 7A), Agrobacterium-mediated transient transformation of white Fxa fruits was performed through injection of bacteria transformed with each of the three constructs. Fruits were infected with C. acutatum and analyzed for both reporter genes activity at 24 hpi, through quantification of eGFP transcript in qRT-PCR and histochemical GUS assay for β-glucuronidase activity. GUS reporter activity, visualized as blue color of fruits, showed great variability among infected and mock-infected fruits (Fig 7B). Consistent with this, no significant difference was shown in the eGFP transcript level quantified in C. acutatum infected versus control fruit (Fig 7C). This is possibly due to the fruit response to Agrobacterium itself, independently from the fungal pathogen. Indeed Agrobacterium can be perceived as a pathogen by the fruit and stimulate similar responses as C. acutatum, including those leading to FanRALF3 expression. The variability affecting strawberry Agrobacterium-mediated transformation depending on technical and environmental conditions has recently been described [76]. It was shown that the expression level of a reporter gene is normally distributed in a population of 30 treated fruits, with huge variation among different fruits. Other important factors affecting agroinfiltration methodology are the quantity of bacteria injected for each fruits, the stage of fruit ripening and the temperature and incubation time after transient transformation. Perhaps for all of these reasons it was not possible to infer the identity of FanRALF3-1 promoter elements inducible by fungal pathogens such as C. acutatum in these experiments. However, the truncated promoter tests were informative. The GUS activity and eGFP expression were measurable and comparable from the T4 and T6 sequences, while reporter activity from the T2 element was almost undetectable. These findings suggest that T4, comprising 400 bp sequence upstream FanRALF3-1 ATG, contains the necessary promoter sequence elements necessary to drive reporter gene expression in strawberry fruits and that the 200 bp region is not sufficient. The full-length upstream region which includes the 3'-UTR of the adjacent gene, shows no more activity that the 400 bp sequence, suggesting this region lacks additional elements to affect transcript accumulation over the T4 construct. According to Motif Scanning analysis, the T4 promoter fragment includes at least two regulatory elements known to be recognized by transcriptional activation complex, (TATA-boxes and Initiator of activation in TATA-less promoter) (Figs 5 and 7A).

Conclusions
Rapid Alkalinization Factors are small signal peptides with multiple roles in plant growth, fertilization and disease response. RALF genes are upregulated in different plant hosts upon pathogens attack and sequences similar to RALF genes are also expressed by many fungal pathogens as virulence factors, suggesting a role as susceptibility factors during plant pathogen interaction. The present work aimed to characterize the RALF gene family in F. vesca woodland and F. x ananassa octoploid strawberries according to tissue specific expression and similarity to Arabidopsis RALFs. The RALF gene family members distribution among F × ananassa subgenome is consistent with octoploid genome evolution, which is characterized by different transposable element activity in the different subgenomes and consequently differential gene distribution. A putative involvement of a MYB transcription factor as regulator of Fan-RALF3-1 infection-inducibility is speculated based on in silico promoter prediction and MYB motif recognition. This element is present in the 400 bp upstream the start codon of Fan-RALF3-1 gene sequence. Because FanRALF3-1 contains the same conserved N-terminal sequence as AtRALF23 it is speculated that FanRALF3-1 interaction with receptor FERONIA (MRLK47) and the coreceptor FanLLG2 may follow the Arabidopsis complex interaction structure. Future efforts will seek to identify specific pathogen-responsive promoter elements and their role in strawberry disease resistance.