Candidate Susceptibility Genes for Powdery and Downy Mildew in Watermelon and Squash

Powdery mildew (PM) caused by Podosphaera xanthii and downy mildew (DM) caused by Pseudoperonospora cubensis are two of the most economically important diseases for watermelon (Citrullus lanatus) and squash (Cucurbita pepo, C. maxima and C. moschata). Traditional breeding for resistance to PM and DM is resource intensive, often requiring decades’ long phenotyping and selection processes. As an alternative, durable and broadspectrum resistance to PM and DM can be obtained through loss-of-function of susceptibility genes in elite breeding material. Susceptibility genes for PM [Mildew-Locus-O (MLO) and Powdery Mildew Resistance (PMR)] and DM [Downy Mildew Resistance (DMR)] have been functionally proven in model plant species. Previous studies have reported candidate MLO genes for C. lanatus and C. pepo, but none for C. maxima and C. moschata. On the contrary, no PMR or DMR candidate genes have been identified for C. lanatus or any of the Cucurbita species. The current study used bioinformatics approaches based on sequence similarity, phylogenetic relationships and presence of conserved domains to predict candidate MLO genes in C. maxima and C. moschata and PMR and DMR genes in C. lanatus, C. pepo, C. maxima and C. moschata. Four MLO homologs in C. maxima and five in C. moschata clustered within Clade V, a clade containing all MLO susceptibility genes in dicots, and had highly conserved transmembrane domains and C-terminal PM interaction motif. Sixty-three candidate PMR genes were identified among the four species, 16 of which had close similarity to functionally proven PMR homologs in model species. Similarly, 37 candidate DMR genes were identified 12 among which clustered with functionally proven DMR homologs in model species. Functional analysis of the genes identified in the current study will reveal their role in pathogenesis and assess their potential for manipulation through gene editing methods to generate novel resistant plant genotypes.


Introduction
Watermelon (Citrullus lanatus) and squash (Cucurbita pepo, C. maxima and C. moschata) are economically important cucurbits in the U.S. with a combined annual value of approximately 0.75 billion dollars [1]. Powdery mildew (PM) caused by Podosphaera xanthii and downy mildew (DM) caused by Pseudoperonospora cubensis are economically important diseases for the two vegetable crops [2] and can cause significant yield and quality losses to growers if not properly managed. Strict preventative fungicide spray regimens are often required to keep plants healthy and free from disease [3]. Although effective, persistent use of fungicides poses toxicity hazards to humans and non-target organisms, as well as a risk for fungicide-resistance [4][5][6]. Genetic resistance is the most preferred management option for the two diseases. However, traditional breeding for PM and DM resistance is resource intensive, often requiring decades' long phenotyping and selection processes.
Knockout of susceptibility genes, also known as S-genes, may be a rapid and effective method for conferring resistance in cultivated plant species. Inactivation of S-genes through gene-knockout, geneknockdown or virus induced gene-silencing has been proven a viable option for generating resistance genotypes against important diseases such as rice blast in rice (Oryza sativa) [7], citrus canker in citrus (Citrus paradisi) [8] and powdery mildew in grapevine (Vitis vinifera) [9], tomato and pepper [10]. Currently, two classes of PM S-genes are known, Mildew-Locus-O (MLO) genes and Powdery Mildew Resistance (PMR), both of which are loss-of-function genes [11][12]. Since the first discovery of MLO genes in barley in 1942, many MLOlike genes have been described and annotated in several model species including Arabidopsis thaliana and Solanum lycopersicum [11][12].
Phylogenetic analysis of MLO proteins reveal as many as 6 distinct clades, denoted clades I through VI, and Clade V harbors all the proven MLO susceptibility genes for PM in dicots [11]. MLO genes are characterized by 7 transmembrane helices, similar to metazoan and fungal G-protein coupled receptors [13]. All the MLO susceptibility genes in clade V share highly conserved domains within transmembrane portions of the protein and harbor a four amino acid long motif at the C-terminus that has been associated with PM interaction [14]. Functional clade V MLO proteins negatively regulate defense pathways at the site of PM inoculation, thus allowing the pathogen to induce pathogenesis [11]. Therefore, a loss of function in MLO genes consequently induce resistance to PM [10,15]. PMR genes on the other hand are far less studied and most are putative, but some have been shown to play a role in PM susceptibility in Arabidopsis (AtPMR4, AtPMR5, and AtPMR6) and tomato (SlPMR4) [16][17][18][19]. PMR genes are involved in cell wall biology where they mediate structure formation and pectin accumulation [18]. In Arabidopsis, loss-of-function in AtPMR4, which encodes callose synthase, resulted in PM resistant genotypes [16].
In cucurbits, MLO candidate genes have been identified in cucumber (Cucumis sativus [12], watermelon (C. lanatus), melon (Cucumis melo) and squash (C. pepo) [22]. However, no MLO candidate genes have been identified for C. maxima or C. moschata. Similarly, information on candidate genes for PMR and DMR in C. lanatus, C. pepo, C. maxima and C. moschata is currently lacking.
Identification of MLO, PMR and DMR genes across the Cucurbitaceae family is an indispensable prerequisite for fundamental studies into the functional role of candidate genes in pathogenesis, and subsequently, identification of potential targets for genetic manipulation to generate novel resistant plant genotypes. Therefore, the objective of the current study was to use bioinformatics approaches based on sequence similarity, phylogenetic relationships and presence of conserved domains to predict candidate MLO genes in C. maxima and C. moschata and PMR and DMR genes in C. lanatus, C. pepo, C. maxima and C. moschata.
Each dicot MLO protein was used in a blast search against the C. maxima and C. moschata genomes, while PMR and DMR proteins were searched against C. lanatus, C. pepo, C. maxima and C. moschata genomes (http://cucurbitgenomics.org/). For C. lanatus, C. maxima and C. moschata, a blastp search was used and the top five results were saved after removal of duplicates. However, since blastp function is not available for C. pepo, a tblastn search was performed against the unigene database (v1.0). The chromosomal distribution of MLO, PMR and DMR protein homologs across the C. lanatus, C. pepo, C. maxima and C. moschata genomes was visualized using Mapchart [23].

Sequence alignment, phylogenetic analysis and clade annotation
All MLO (monocots and dicots), PMR and DMR protein sequences from model species were aligned with protein sequences extracted from C. lanatus, C. pepo, C. maxima and C. moschata genomes using the ClustalW alignment program in MEGA6 software [24]. The neighbor-joining clustering method was used to generate bootstrap consensus phylogenetic trees using 100 replicates. For MLO genes, clades were annotated as described by Devoto et al. [25].

Conserved domain analysis of MLO proteins
All MLO-like sequences from C. maxima and C. moschata that clustered in clade V were aligned to functionally proven MLO proteins in dicots to confirm presence of conserved domains in the transmembrane protein (TM1 -TM7), and the PM interaction Cterminus (D/E-F-S/T-F) using ClustalO alignment with boxshade [14,26]. To build a MLO consensus sequence, an amino acid was regarded as conserved if at least 7 out of 8 of the functionally proven MLO proteins harbored it, or an amino acid with similar chemical properties as determined by the Rasmol color scheme [12,27]. The number of amino acids deviating from the conserved sequence in each of the candidate proteins as well as in all the clade V proteins was counted to determine degree of similarity.

Homolog Designation in genome database
Cucurbita maxima  All the ten homologs clustered in Clade V, a clade containing all MLO susceptibility genes in dicots (Figure 1). A similar study by Lovieno et al. [22] identified 14 MLO homologs in C. lanatus and 18 in C. pepo, three of which clustered in Clade V in the former and latter, respectively. In cucumber, Schouten et al. [12] identified 14 MLO-like proteins, 3 of which placed within clade V. Examination of the transmembrane portions of the MLO proteins as well as the PM interaction motif at the C-terminus (D/E-F-S/T-F) revealed a total of 119 conserved amino acids in the consensus sequence. Alignment of the consensus sequence to the functionally proven MLO proteins in clade V revealed high similarity, ranging from 94.8% to 100% (Table 2 and Figure 2). High level of similarity among clade V proteins has been noted in other plant species including cucumber [12], apricots (Prunus

Chromosomal distribution of MLO, PMR and DMR homologs
The candidate genes identified in this study were distributed across the C. lanatus, C. maxima and C. moschata genomes ( Figure 5A). In watermelon, all the 11 chromosomes had at least one candidate susceptibility gene and clustering of MLO [previously identified by Lovieno et al. [22] and PMR genes was noted on chromosome 3 in an interval (0.5 Mb to 5.9 Mb) known for high nucleotide divergence and enrichment of disease response genes [28] (Figure 5A). Other gene clusters for PMR and DMR homologs were identified on Chromosome 2. In C. maxima and C. moschata, candidate susceptibility genes were found in all chromosomes except chromosomes 9,10,11 and 12. For the two species, most candidate homologs were located at similar positions on respective chromosomes ( Figure 5B-C), thus suggesting high levels of synteny between C. maxima and C. moschata.

Significance of MLO, PMR and DMR candidate genes
The current study identified multiple MLO, PMR and DMR candidate gene in C. lanatus, C. pepo, C. maxima and C. moschata which should be investigated to elucidate their role in PM and DM pathogenesis. Approaches such as gene knock out, gene knock down or gene expression analysis will further narrow down the candidate gene list to just a few that can be manipulated through gene-editing methods to create novel PM or DM resistant genotypes. Figure 4B: Downy Mildew Resistance phylogenetic analysis. Phylogenetic tree for DMR6 proteins in Arabidopsis (AtDMR1 and AtDMR6) and AtPMR6), tomato (SlDMR1) and cucumber (CsaDMR1, CsaDMR6-1 and CsaDMR6-2) and DMR-like proteins in Citrullus lanatus, Cucurbita pepo, Cucurbita maxima and Cucurbita moschata. Underlined proteins represent those closest in previously identified by Lovieno et al. [22]. Red, purple and green font represent MLO, PMR and DMR homologs, respectively. Underlined proteins represent those closest in similarity to functionally proven homologs in other species. Chromosomes (Chr) without any candidate genes are not included.