Aphid- borne viruses infecting cultivated watermelon and squash in Spain: Characterization of a variant of cucurbit aphid-borne yellows virus (CABYV)

Aphid- borne viruses are responsible for major cucurbit diseases and hamper the sustainability of crop production. Systematic monitoring can reveal the occurrence and distribution of these viruses, in addition to unadvertised viruses, facilitating the control of diseases. For three consecutive (2018– 2020) seasons, the presence of aphid- borne viruses was monitored from a total of 292 samples of watermelon and squash plants that showed yellowing symptoms in three major cucurbit- producing areas (Castilla La- full- length isolates LP63 isolate, amino acid potential higher RNA plant CABYV variant severe yellowing causing cucurbit


| INTRODUC TI ON
Cucurbits are among the most important horticultural vegetables in the Mediterranean basin. However, the sustainability of their production is threatened by viral diseases that include at least 28 different viruses (Lecoq & Desbiez, 2012). Among them, the prevalence of aphid-borne viruses appears to be favoured not only because of the lack of effective countermeasures against plant viruses, but also by the increase of these crops in organic cultivation systems that may alter the vector population size, encouraging the incidence of these viral diseases. Thus, there is a need to perform systematic monitoring of the occurrence of viral diseases, as well as of their genetic structure and evolutionary epidemiology, which may facilitate the early detection of emerging diseases and control of viral diseases.
In particular, cucurbit aphid-borne yellows virus (CABYV, family Luteoviridae, genus Polerovirus) is one of the most prevalent viruses Mnari-Hattab et al., 2009). CABYV was initially described in France in 1992 (Lecoq et al., 1992), and subsequently it has been reported in cucurbit crops from Algeria, Greece, Italy, Iran, Lebanon, Spain, Tunisia, and Turkey (Juárez et al., 2004;Mnari-Hattab et al., 2009;Tomassoli & Meneghini, 2007). Most recently, it has also been reported in Germany (Menzel et al., 2020), Slovenia (Mehle et al., 2019), and Poland (Minicka et al., 2020). CABYV is especially spreading through the northern hemisphere where it is known to cause the most harmful epidemic viral diseases in agriculture (Lecoq et al., 1992). CABYV is transmitted primarily by two aphids, Aphis gossypii and Myzus persicae, and the transmission is considered to be circulative, persistent, and nonpropagative (Dogimont et al., 1996). In Spain, CABYV was first identified in 2004 in the Murcia region (southeastern Spain; Juárez et al., 2004). From that point on, it has spread to cucurbit crops, becoming highly prevalent in melon and zucchini crops in southeastern Spain Kassem et al., 2007).
Additionally, potyviruses, such as watermelon mosaic virus (WMV), zucchini yellow mosaic virus (ZYMV), and papaya ringspot virus (PRSV), as well as cucumber mosaic virus (CMV, genus Cucumovirus), among others, have been reported to affect cucurbit crops (Bertin et al., 2020;Desbiez et al., 2020;Juárez et al., 2013), with a relatively high importance in Spain (De Moya- Ruiz et al., 2021;Juárez et al., 2013). Furthermore, the number of multiple viral infections is comparatively high in cucurbit plant species in Spain, and CABYV and WMV are often found in mixed virus combinations Kassem et al., 2007Kassem et al., , 2013. The sequencing of different CABYV isolates has identified four major groups: Asian or Chinese (C), Mediterranean (N), and Taiwanese (TW), with the phylogenetic analysis further distinguishing four clusters, including recombinant (R) groups (Knierim et al., 2010). Thus, CABYV populations are considered to be genetically structured. Further analysis of the open reading frames (ORF3; coat protein, CP) has shown that CABYV has a high nucleotide substitution rate (0.01 substitutions/ site/year) (Pagán & Holmes, 2010). This could indicate a relatively high mutation rate as the source of genetic diversity. Additionally, it has been reported that the genetic diversity of Polerovirus populations can be prone to recombination amongst themselves or viruses belonging to other families (Knierim et al., 2010). These factors provide genetic plasticity that may allow for the rapid evolution and adaptation of viruses to new agricultural conditions and may therefore be contributing to the creation of new CABYV variants that could challenge the prevailing control strategies.
Heterogeneous host genotype populations (i.e., different host plants and varieties of cucurbits along with particular agroecological practices) may alter the epidemiological patterns of viral diseases (De Moya- Ruiz et al., 2021;Juárez et al., 2019;Valverde et al., 2020). Hence, the systematic monitoring of the occurrence of viral diseases and their causative agents, combined with ecological and quantitative epidemiological approaches, is essential to understand and mitigate their consequences on crops and natural ecosystems (Jeger, 2020;McLeish et al., 2020). In addition, the genetic and biological characterization of the emerging viruses increases our understanding of the ecological epidemiology of the diseases and could facilitate their early detection and prompt action for their control.
Thus, the aim of this study was to increase our understanding of the occurrence and distribution of aphid-borne viruses that cause yellowing diseases in watermelon and squash crops. We monitored the occurrence of cucurbit aphid-borne viruses in these cultivated plants for three consecutive seasons in three major crop-producing areas. Additionally, we observed and further characterized an emerging CABYV variant that causes bright yellowing disease in watermelon plants. The construction of two CABYV infectious cDNA clones allowed us to assess the symptom expression and viral RNA accumulation in five cucurbit plant species in order to understand the impact of this novel CABYV variant on the cultivated cucurbits.

| Cucurbit virus detection
The total RNA from the plant samples was extracted using TRI reagent (Sigma-Aldrich), and 1.5 µl was used for virus detection either by dot-blot hybridization or reverse transcription (RT)-PCR. Because no differences were found between both techniques for the viral detection, we provide data from molecular hybridization. RNA from each sample was placed twice on five positively charged nylon membranes, and the RNA was fixed with a UV light crosslinker. A dot-blot molecular hybridization was carried out using specific RNA probes for CABYV, CMV, PRSV, WMV, and ZYMV detection (Kassem et al., 2007). The membranes were incubated overnight at 68 °C with the specific digoxygenin (DIG)-labelled probes. After the hybridization, a series of washes were carried out, followed by an incubation with the anti-DIG antibody conjugated to alkaline phosphatase (Roche Diagnostics) and the chemiluminescent substrate CDP-Star (GE Healthcare) (Gómez-Aix et al., 2019). The membrane analyses were performed using a chemiluminescent detector Amersham Imager 600 (GE Healthcare Bio-Sciences). Furthermore, the watermelon samples with exacerbated yellowing symptoms were analysed by RT-PCR using generic polerovirus primers, in addition to specific primers (

| Full-length CABYV genome amplification and construction of CABYV infectious clones
One CABYV-positive sample from a watermelon with symptoms (2019) was randomly chosen to construct a full-length CABYV infectious clone. Concurrently, another CABYV isolate from a melon sample collected in 2014 was also cloned and sequenced for further comparison. Both full-length genome sequences were aligned with other CABYV isolates from the NCBI GenBank in order to examine the phylogenetic relationship among CABYV isolates.

| Sequencing and phylogenetic relationships of both CABYV isolates
Full-length genomes of both CABYV clones (LP63 and MEC12.1) were sequenced by using a set of eight internal primers (Table S2) and the Sanger method (STAB VIDA, Caparica, Portugal). Each primer was able to provide a sequence of up to 800 bp, matching with the contiguous genome sequence. All the contigs were assem-

| Agroinoculation of cucurbit plant species
Agrobacterium tumefaciens C58C1 was transformed with the purified plasmids that contained either LP63 or MEC12.1 isolates. The cultures were incubated overnight at 28 °C, centrifuged, and resuspended in the same volume of MES buffer (Gómez et al., 2009). The inoculations were carried out in the leaves of the cotyledons 2 weeks after potting. The plants were grown in a greenhouse (16 hr photoperiod and 24 °C in a day/night cycle). Note that virus-like symptoms suggestive of CABYV infection were observed during the experiment in older leaves, and in watermelon plants, LP63 symptoms were only observed after 31 dpai. Thus, to ensure Koch's postulates with this LP63 isolate, another group of watermelon plants was inoculated and maintained for a longer term, and the yellowing severity symptoms were confirmed at 50 dpai.

| Viral load quantification
All leaves above the cotyledons were harvested in groups of three replicated plants for each group of plant species and virus infection at 6, 12, 21, and 31 dpai. All material was ground in a mortar using liquid nitrogen. Total RNA was extracted using TRI reagent, purified by phenol-chloroform extraction, and treated with DNase I (Sigma-Aldrich). Thereafter, CABYV viral load was estimated by measuring the viral RNA accumulation via real-time quantitative RT-PCR (RT-qPCR) with an AB7500 System (Applied Biosystems) using the One-step NZYSpeedy RT-qPCR Green kit, ROX plus (NZYTech).
Two specific primers, CE-2879 Fw 5′-GAGAGCCCAGCATTCAGC-3′ and CE-2880 Rv 5′-TGCAGTGGGGGTCCAA-3′, were designed to target the CP region (3,805-3,941 nucleotides), and their specificity was monitored with melting curve analysis. The reaction mix was prepared according to the manufacturer's instructions (NZYTech), using 2 µl of RNA (70 ng/µl). Both non-template controls and mockinoculated plants were included to ensure product-specific amplifications and the absence of primer-dimers. The CP gene fragment (c.598 nucleotides) from MEC12.1 isolate was cloned into pGEM T-Easy vector by using the CE-9 and CE-10 primers (Table S1) in order to synthesize viral RNA from a plasmid. After linearizing the plasmid with SphI and RNA transcription using SP6 RNA polymerase, the RNA transcripts were quantified twice with a Qubit 3.0 fluorometer following the manufacturer's instructions (Thermo Fisher Scientific). The initial RNA concentration was used in serial dilutions (10-fold) to generate external standard curves for RT-qPCR from the CP gene. From each sample, the RNA concentration (ng of viral RNA per 100 ng of total RNA) was estimated by plotting the cycle threshold (C t ) values from each biological assay (n = 3, at each time point) with three experimental replicates for each biological replicate.

| Statistical analysis
The analysis of the viral load for each plant species was performed with a general linear model (GLM). Values from each plant were independent among treatments, data were transformed with a logarithmic function to meet the assumption of normality and homoscedasticity of variance. We thus fitted the viral isolate type, the five cucurbit plant species, and the time of infection (dpai) as threefactor fixed effects, including replicates as random effects (REML).
For the analysis of each CABYV clone effect, virus accumulation from each plant species was considered separately and analysed using one-way repeated-measures analysis of variance (ANOVA). All analyses were performed with JMP v. 9 software (SAS institute). Plot graphs of the viral RNA accumulation for each isolate and plant species were drawn using R v. 4 software (R Core Team).  ORF0 had the highest level of similarity, followed by ORF1, ORF3, and ORF4 that shared a 97% similarity. However, ORF2 had 96%, and ORF3a and ORF5 95% nucleotide similarity. Accordingly, the highest amino acid similarity between both isolates was found in ORF0 (99%), followed by ORF3 and ORF5 (98%), and then by ORF3a and ORF1 (97%), ORF2 (96%), and ORF4 with the lowest amino acid similarity (93%) recorded, suggesting the lack of any frameshift mutation.

| Field symptoms and genetic characterization of a novel CABYV variant
We then explicitly carried out an analysis of the number of mutations for each ORF that were exclusive to the MEC12.1 and LP63 genomes, compared to two other full-length genomes sourced from GenBank (JF939812 and JF939814) that also belonged to the Mediterranean subpopulation (Table 1). We found 87 positions with single nucleotide polymorphisms (SNPs) in the LP63 genome, with 58 and 29 synonymous and nonsynonymous SNP mutations, respectively. In the MEC12.1 genome, we found 79 SNPs, with 55 synonymous and 24 nonsynonymous SNP mutations. This indicated a moderate genetic variability in both isolates, although interestingly, ORF2 (replication-associated protein) from LP63 had a higher number of nonsynonymous mutations as compared to the MEC12.1 isolate (17 versus 5), as well as to the rest of the ORFs (Table 1).
Furthermore, the detection of evidence pointing to recombination was attempted, using the full-length genomes of both CABYV-LP63 and -MEC12.1 isolates, including 28 sequences used for phylogenetic analysis. This recombination analysis suggested that CABYV-LP63 may have originated from two breakpoints in the P0-P1 region (40-1,068 nucleotides), and regions P1, 2, and 3a (1,069-3,464 nucleotides). In the first potential recombination event, all the methods used significantly identified (p < 0.01) the sq/2004/1.9 isolate as the major parent and the MEC12.1 isolate as the minor parent. In the second recombination event, the methods identified the MEC12.1 isolate as the major parent and an unknown minor parent (Figure 3b).

| Comparison of symptom expression and viral accumulation of the CABYV isolates
To examine whether genetic differences of the CABYV-LP63 isolate were attributed to biological differences in cucurbit plants, a in the three major producing areas in Spain. This result is in agreement with previous studies conducted in southeastern Spain Kassem et al., 2007Kassem et al., , 2013 that confirmed CABYV as the most prevalent aphid-borne virus in these crops. CABYV seems to be widespread throughout the Mediterranean basin and Europe (Desbiez et al., 2020;Mehle et al., 2019;Menzel et al., 2020;Minicka et al., 2020;Mnari-Hattab et al., 2009). In fact, CABYV appears to be the most common virus affecting cucurbit crops across different climatic regions, temperate, Mediterranean, and subtropical (Lecoq et al., 1992). This high occurrence and prevalence of CABYV could be linked to either cultivated or wild plant species that may provide the source of inoculum for spread, in addition to the aphid-associated transmission, because CABYV is transmitted in a persistent and circulative manner by at least two aphid species, A. gossypii and M. persicae Lecoq et al., 1992). Furthermore, it has been reported that the aphid feeding behaviour can be influenced  (Desbiez & Lecoq, 2008), and this could be favouring the occurrence of WMV in those yellowing plants. On the other hand, although the impact of aphid-vector performance on the ecology and evolution of CABYV and WMV in mixed infections is still unexplored, it is worth mentioning that mixed-infection melon plants, infected with WMV and cucurbit yellow stunting disorder crinivirus (CYSDV), may prompt ecological advantages that allow for the coexistence of both viruses in the field (Domingo-Calap et al., 2020). Thus, it is likely that this combination of CABYV and WMV may be influencing physiological and chemical plant changes, with an effect on the vector behaviour that supports the occurrence and distribution of these aphid-borne virus diseases in cucurbit crops (Carmo-Sousa et al., 2016;Mauck, 2016;Schoeny et al., 2020).
Multiple infections are more common in nature than what would be expected to occur by chance (Alcaide et al., 2020;Syller, 2012). We observed that the frequency of the combination of CABYV + WMV varied between 8% and 32% of the infected plants, depending on the species and cultivation area.
This combination of CABYV + WMV infections has also previously been described in the Mediterranean basin (Desbiez et al., 2020;Juárez et al., 2013;Kassem et al., 2007). The epidemiological consequences of mixed virus infections are unclear and seem to be con- have been defined to be agronomically important (Costa et al., 2020;Knierim et al., 2010), and this demands further research on the genetic variability of CABYV populations in cucurbit crops. This LP63 accumulation was in agreement with the symptom expression that was more severe with the higher viral RNA accumulation.
Because the level of LP63 accumulation in cucurbit plants was higher than MEC12.1, it is likely that this novel variant has a competitive advantage in cucurbit crops, with new outcomes on the CABYV spread and distribution, especially in areas of watermelon production. In this sense, the occurrence of CABYV and genetic diversity of the viral populations needs to be further investigated.
In conclusion, despite the extent of this new yellowing disease in watermelon and other cucurbit crops being unknown, our findings suggest that this novel CABYV variant may be currently threatening cucurbit crops. It is thus fundamental to consider that plant viral epidemics are often initiated and spread through different host plants and varieties, along with different agroecological practices that may alter the viral disease epidemiology (Moya- Ruiz et al., 2021;Jeger, 2020;Juárez et al., 2019;Valverde et al., 2020). This exchange of viral diseases between overlapping crops at spatial/temporal scales could play an important role, as it may hinder the early detection of the disease. Therefore, there is a need to establish a comprehensive detection programme to monitor the occurrence, distribution, and genetic diversity of CABYV populations in different cultivated and wild plant species to facilitate the design and implementation of control measures against this disease.

ACK N OWLED G EM ENTS
We thank all technicians from each cucurbit-producing area for as- funds. We also thank M. C. Montesinos for technical assistance, and Mario Fon (mariogfon@gmail.com) for English editing assistance.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in DIGITAL-CSIC at https://digit al.csic.es, reference number http:// dx.doi.org/10.20350/ digit alCSI C/13757.