Specificity of resistance and tolerance to cucumber vein yellowing virus in melon accessions and evidence for resistance breaking associated with a single mutation in VPg

18 Cucumber vein yellowing virus (CVYV) is an emerging virus on cucurbits in the Mediterranean 19 Basin, against which few resistance sources are available, particularly in melon. The melon 20 accession PI 164323 displays complete resistance to isolate CVYV-Esp, and accession HSD 2458 21 presents a tolerance, i.e. very mild symptoms in spite of virus accumulation in inoculated plants. 22 The resistance is controlled by a dominant allele Cvy-1 1 , while the tolerance is controlled by a 23 recessive allele cvy-2 , independent from Cvy-1 1 . Before introducing the resistance or tolerance in 24 commercial cultivars through a long breeding process, it is important to estimate their specificity 25 and durability. Upon inoculation with eight molecularly diverse CVYV isolates, the resistance was 26 found to be isolate-specific since many CVYV isolates induced necrosis on PI 164323, whereas the 27 tolerance presented a broader range. A resistance-breaking isolate inducing severe mosaic on PI 28 164323 was obtained. This isolate differed from the parental strain by a single amino acid change


Cucumber vein yellowing virus (CVYV) is an emerging virus on cucurbits in the Mediterranean
Basin.CVYV belongs to the genus Ipomovirus in the family Potyviridae.Its genetic organization is similar to that of viruses in the genus Potyvirus, except that it encodes no helper component protein (HC-Pro) but a duplicated P1 protein that replaces the HC-Pro functionally as a silencing suppressor (Valli et al. 2006).CVYV is transmitted by the whitefly Bemisia tabaci in a semipersistent manner (Lecoq et al. 2000;Desbiez et al. 2019).First described in Israel in 1960, CVYV is now present throughout the Middle East and the western Mediterranean Basin (Velasco et al. 2016).Surveys in Sudan have shown that the virus has also been present for decades in Sub-Saharan Africa (Desbiez et al. 2019).CVYV induces symptoms of vein clearing and yellowing of the leaves, as well as discoloration and sometimes cracking on fruits that make them unmarketable (Navas-Castillo et al. 2014).As for many other viruses, control of CVYV relies on the control of whitefly populations and the use of genetic resistance when available (Janssen et al. 2003;Gomez et al. 2009).The use of resistant cultivars is a sustainable and environmentally-friendly way of controlling viruses but presents several limitations.First, resistance factors must be available in the germplasm of the species of interest or wild relatives and introduced in high-yielding cultivars through a time-and labor-consuming breeding process.Besides, upon large-scale deployment of resistances in the field, their efficiency can be threatened because they happen to be too specific, i.e. efficient only against a subset of the existing viral isolates, or broken through the evolution of the pathogen (Garcia-Arenal and McDonald 2003;Kobayashi et al. 2014).Therefore, it is important to estimate the potential durability of a resistance factor before using it in breeding programs.
Most accessions displayed either susceptibility (mosaic and vein clearing symptoms) or very high susceptibility (severe mosaic, yellowing and stunting), a few accessions showed systemic necrosis followed by rapid plant death, one accession was tolerant (very mild mottle despite virus accumulation in the plant), and one accession was resistant (no symptoms, no virus detection) (Pitrat et al. 2012) (Figure 1).
In this work, we studied the specificity of the resistance and tolerance phenotypes in melon cultivars against several CVYV isolates.We showed that the tolerance was efficient against a large number of isolates, whereas the resistance was very specific and easily overcome by the virus through an E to K point mutation at amino acid position 90 in the viral protein genome-linked (VPg).We also explored the possible impact of this mutation on VPg conformation.

Melon accessions
Melon accessions representative of the five phenotypes observed after infection with CVYV-Esp (Pitrat et al. 2012) were used in the study: i) Védrantais, a Cantaloupe type from France, is susceptible to CVYV-Esp, showing symptoms of vein clearing and mosaic; ii) Ouzbeque 2 from Uzbekistan is highly susceptible with symptoms of severe mosaic, stunting and yellowing; iii) accession HSD 93-20-A from Sudan exhibits necrotic symptoms with rapid death upon infection; C. Desbiez Phytopathology iv) HSD 2458 from Sudan is tolerant to CVYV-Esp, i.e. the virus can be detected in inoculated plants but the plants display only a very mild mottle ; and v) accession PI 164323 from India is resistant to CVYV-Esp : no virus is detected and no symptoms are visible in inoculated plants.The resistance was first observed after mechanical inoculation of the virus but it was shown to be also effective after whitefly transmission (data not shown).Five to 10% of the mechanically infected PI 164323 plants developed a top necrosis after inoculation (Figure 1).F1, F2 and back-cross (BC) progenies were obtained between the above mentioned accessions in order to study the inheritance of the different phenotypes (Pitrat et al. 2012).Three loci appeared to be involved.At a first locus tentatively named Cucumber vein yellowing resistance (symbol Cvy-1), three alleles have been identified: Cvy-1 + for susceptibility (present in Védrantais), Cvy-1 1 controlling resistance in PI 164323 and Cvy-1 2 controlling necrosis in HSD 93-20-A.At an independent second locus, the recessive allele cvy-2 present in HSD 2458 controls the tolerance.
And at a third locus, the allele Cvy-3, present in Ouzbeque 2, controls the highly susceptible type of symptoms (Pitrat et al. 2012).

CVYV isolates and mechanical inoculation of melon accessions
Isolates CVYV-Esp, SU96-84, SU12-10, TN05-56, OM04-06, ISR, LIB and JOR are as described in Desbiez et al. (2019).They originate from Spain, Sudan, Tunisia, Oman, Israel, Lebanon, and Jordan, respectively, and are representative of the known molecular variability of CVYV (Desbiez C. Desbiez Phytopathology et al. 2019).All isolates were stored as dried leaf material on calcium chloride or as frozen leaf tissue in liquid nitrogen.For inoculations, dry or frozen plant tissues were first ground in 0.03 M Na2HPO4 + 0.2% diethyldithiocarbamate (DIECA).Activated charcoal and carborundum (75 mg/ml each) were added before rub-inoculating the two cotyledons of young plantlets of cucumber cf.
Beit Alpha.Symptomatic cucumber leaves were used for mechanical inoculation of each melon accession with the same protocol, using 5 or 10 young plantlets of each accession.A second inoculation was performed one week later on the first leaf of the same melon plantlets to increase infection efficiency.Inoculated plants were kept in a quarantine greenhouse (C3 biosecurity level).The presence of the virus at the plant apex, six weeks after inoculation, was assessed by DAS-ELISA with a CVYV-specific antiserum (Desbiez et al. 2019) or by RT-PCR as described (Lecoq et al. 2007).The experiment was repeated three times.

Full-length sequencing of resistance breaking isolate CVYV-PI1
After mechanical inoculation of CVYV-Esp on melon PI 164323, symptoms of top necrosis were observed in 5 to 10% of the plants (Figure 1).In one instance, an axillary shoot developed that showed severe mosaic symptoms.Back inoculations were performed on PI 164323 from mosaicshowing plant material and the inoculated plants displayed mosaic symptoms.After five successive passages on PI 164323, the mosaic-inducing variant CVYV-PI1 was stored as dried material on calcium chloride and used for full-length sequencing.The complete sequence of CVYV-PI1 was determined through Sanger sequencing as described by Desbiez et al. (2019).Fulllength sequences of CVYV-PI1 and CVYV-Esp were aligned with ClustalW included in MEGA6 (Tamura et al. 2013) and compared manually.

Cloning and inoculation of plants with an infectious cDNA clone of CVYV
Total RNA was extracted from a young plant of cucumber cv.Beit Alpha inoculated with CVYV-Esp using TRI-reagent (Molecular Research Center, Cincinnati, OH).The precise sequence of the 5' extremity from CVYV-Esp was reassessed by 5'-RACE (Ferriol et al. 2018).The genome of CVYV-Esp was amplified in five overlapping fragments of 1.2 to 1.8 kb through a two-step RT-PCR using a high-fidelity Pfu DNA polymerase for PCR and specific primers (Supplementary table S1) as described (Desbiez et al. 2012).Two introns were added by fusion PCR to fragments 2 and 3, in the P3 and CI coding regions, respectively.The five final fragments were purified from agarose gel, mixed with a linearized vector derived from pAGUS1 and used for transformation of competent Saccharomyces cerevisiae YHP 501 cells as described (Desbiez et al. 2012), before plating on CAU selective medium.The yeast colonies were then grown in CAU liquid medium, yeast DNA was extracted as described (Desbiez et al. 2012) and used to transform electrocompetent Escherichia coli DH5 bacteria.Bacterial plasmid DNA was extracted by alkaline lysis and a screening was performed by restriction enzyme analysis.Since the yields of bacterial DNA were very low, 0.5 µl of plasmid DNA were submitted to rolling circle amplification (RCA) using the TempliPhi kit (GE Healthcare) according to the manufacturer's manual.One µl of RCA product was then used for restriction analysis.Inoculation was performed by biolistics using a GeneGun according to Gal-On et al. (1995), on the cotyledons of eight cucumber Beit Alpha plantlets.Briefly, 30 µl of plasmid DNA, or 5 µl RCA product + 25 µl distilled water, were mixed with 30 µl of tungsten M20 at 50 mg/ml in 50% glycerol and 30 µl of Ca(NO 3 ) 2 1.25M, pH 8.After 5 minutes at room temperature, a short centrifugation was performed and 60 µl of the supernatant were removed.Three µl of the remaining mix were used for the bombardment of Since the expression level of CVYV-Esp infectious clone in E. coli was very low, clone 3.1 was submitted to RCA with the TempliPhi kit according to the manufacturer's manual before digestion with restriction enzyme NdeI that cuts at positions 5210 and 8140 in the intron-C.Desbiez Phytopathology containing CVYV clone.The 12.7-kbp resulting vector DNA as well as the two PCR fragments were purified from gel and used for transformation of competent S. cerevisiae YHP 501 cells as described (Desbiez et al. 2012).After growth on CAU medium, yeast DNA was extracted and used to transform electrocompetent E. coli DH5 bacteria.Bacterial plasmid DNA was extracted by alkaline lysis and a screening was performed by restriction enzyme analysis.The presence of the expected E90K mutation in the VPg was checked by amplification with primers CVYV-VPg-F (5'-GCAGGAGTCATCACGATAAC-3') and CVYV-6250-R (5'-ARCAATTGCGGTTTCAGCGAC-3') followed by Sanger sequencing of the targeted area.Positive clones (0.5 µl DNA extracted by alkaline lysis) were amplified by RCA.One µl of RCA product was used for inoculation by biolistics on the cotyledons of cucumber Beit Alpha plantlets.

Infectivity of CVYV-VPg-C1
One infectious clone containing the VPg mutation, CVYV-VPg-C1, was inoculated by GeneGun to five plantlets of cucumber cv.Beit Alpha.Leaf tissue of symptomatic plants was used for mechanical inoculation of 5 or 10 plantlets of cucumber Beit Alpha and melons cv.Védrantais, PI 164323, Ouzbeque 2, HSD 93-20-A and HSD 2458.The plants were kept in a quarantine greenhouse until symptom development.Six weeks after inoculation, all asymptomatic plants were tested by DAS-ELISA for the presence of CVYV.The experiment was repeated 3 times.
Four methods were used to predict the intrinsic disorder of CVYV VPg and the impact of the mutation on this disorder: (1) Disopred (http://bioinf.cs.ucl.ac.uk/disopred), ( 2

Full-length sequencing of CVYV-Esp and the resistance-breaking variant CVYV-PI1
An evolved isolate CVYV-PI1, developing severe mosaic symptoms on PI 164323, was obtained after mechanical inoculation of CVYV-Esp.The full-length sequences of CVYV-PI1 and CVYV-Esp were obtained (Desbiez et al. 2019).The accurate sequence of the 5' extremity of CVYV-Esp was obtained by using 5'-RACE, indicating that the CVYV sequence available before was truncated and the first 17 nt were missing.The sequence of CVYV-Esp sequence presented an A-rich 5' extremity as expected for a potyvirid, and the first 15 nt were identical to those of the SqVYV reference isolate (accession NC_010521).Compared to CVYV-Esp, three mutations were found in CVYV-PI1,

Infectious cDNA clone of CVYV-Esp
After amplification of the CVYV-Esp genome and yeast transformation, clones with the expected size and restriction profiles were obtained but the level of expression of plasmid DNA in E. coli was very low, and RCA amplification was performed to enhance the plasmid concentration.After GeneGun inoculation on cucumber cv.Beit Alpha, 100% of inoculated plants with clone 3.1 showed symptoms of vein clearing two weeks after inoculation with the RCAamplified plasmid, whether or not it was digested with SnaBI, showing that direct GeneGun inoculation of very long RCA products was as efficient as inoculation with the 15 kbp linearized plasmid.In contrast, no infection was observed with inoculation of the plasmid without RCA amplification.The full-length sequence of clone 3.1 was determined, and it differed from the sequence obtained ten years earlier for the original field isolate by 26 mutations including 7 nonsynonymous located in the P1 (3 mutations), NIb (2 mutations), CI and CP (1 mutation each).
There was no non-synonymous mutation in the VPg.

Diversity and structure of the VPg coding region of CVYV and other ipomoviruses
The amino acid VPg sequences of 11 CVYV isolates and the evolved variant CVYV-PI1 were aligned and compared.There was no extensive sequence polymorphism in the VPg coding region for CVYV (183 aa for all isolates) (Figure 2a).Most CVYV isolates had an E at amino acid position 90 in the VPg, except CVYV-ISR and CVYV-ISM (sharing identical VPg sequences) that had a D. Among the 11 CVYV sequences excluding CVYV-PI1, 14 non-synonymous mutations were present in the VPg coding region, nine of which were observed in only one isolate.Ten mutations were in the central part of the VPg between aa 90 and 137 out of 183.CocMoV, closely related molecularly to CVYV, displayed ~75% amino acid identity in the VPg with different CVYV isolates.It had a Q at the position corresponding to CVYV VPg90, immediately followed by an E, and a K at position VPg93 instead of the E present in all CVYV isolates (Figure 2a).Melon accession PI 164323 was susceptible to the Su12-25 isolate of CocMoV in mechanical inoculation (H.Lecoq, unpublished).
The other ipomoviruses displayed between 30% and 50% aa conservation in the VPg with CVYV, Page 14 of 25 C. Desbiez Phytopathology without conservation at or around the E5962 position (Figure 2a).
Based on PONDR prediction, the central part of CVYV VPg where the resistance-breaking mutation was located appeared as a highly ordered region, and the E 90 K mutation, despite the change from a negatively charged glutamic acid to the positively charged lysine, did not have a strong effect on its disorder profile (Figure 2b).The VPg disorder profile of CVYV was very different from that of the potyviruses PVY, ZYMV, PRSV and TuMV (Supplementary figure S2), where the central region of the VPg was confirmed as an intrinsically disordered domain (Supplementary figure S2).
The disorder profiles of the ipomoviruses CVYV, SqVYV, CBSV, and UCBSV were very variable, the central domain being either ordered or disordered, with contrasted situations even for the closely related CBSV and UCBSV (Supplementary figure S2), suggesting that the VPg interaction patterns between potyviruses and ipomoviruses, and also between different ipomoviruses are not conserved.

Discussion
In this work, we found that resistance to CVYV present in melon accession PI 164323 and conferred by the allele Cvy-1 1 was isolate-specific since only some isolates failed to infect, whereas others induced a rapid and lethal top necrosis.Besides, the resistance to CVYV-Esp could be overcome through a point mutation in the virus genome, resulting in an E to K amino acid change (E90K) in the viral VPg.We validated the effect of that mutation by reverse genetics using a newly-developed infectious cDNA clone of CVYV.Previous attempts to obtain infectious clones of CVYV were unsuccessful, probably due to the fact that the sequences available so far lacked 17 nts at their 5' extremity, including the first four "As" required for potyvirid infectivity (Simon-C.Desbiez Phytopathology Buela et al. 2000).Despite the presence of two introns in the P3 and CI coding regions of the cDNA, the yield of CVYV-containing plasmids were very low in E. coli, probably in relation to the toxicity of P1a or P1b proteins in bacteria.No infection was obtained after direct biolistic inoculation of the plasmid, indicating that the threshold concentration of plasmid DNA for infectivity was not reached.For potyviruses, less than 0.1 ng plasmid DNA can be sufficient for high infectivity after bombardment (Gal-On et al. 1995), but it seems more DNA may be required for CVYV.Indeed, CVYV is generally less efficiently transmitted than cucurbit-infecting potyviruses in mechanical inoculation (H.Lecoq, unpublished).The use of RCA prior to inoculation improved the infectivity of the plasmid.Resistance-breaking mutants, both the natural CVYV-PI1 and the cloned CVYV-VPg-C1, induced severe mosaic symptoms on PI 164323.Such severe symptoms could be expected since PI 164323, besides the broken Cvy-1 1 resistance allele, possesses the Cvy-3 "high susceptibility" allele also present in Ouzbeque 2 (Pitrat et al. 2012).
The E to K mutation at aa position 90 in the VPg of CVYV-VPg-C1 was sufficient to induce resistance breaking in PI 164323.Indeed, contrary to other pathogens, most situations of resistance breaking in plant viruses are related to point mutations rather than indels or loss of expression (Sacristan and Garcia-Arenal 2008).This is probably due to the constraints of virus genomes where all proteins are required for infectivity and are frequently multifunctional (Revers and Garcia 2015).
The E 90 K mutation in the VPg resulted in charge change in the protein, which probably affects its interaction with so far undetermined plant or virus factors.The area where the mutation took place appears as highly ordered, and the mutation does not seem to modify deeply the level of disorder of the protein (Figure 2b).Surprisingly, although the resistance related to Cvy-1 1 was efficient only against some CVYV Page 16 of 25 C. Desbiez Phytopathology isolates that display an E or a D at position 90 in the VPg, all the other isolates tested had an E at position 90 and presented few mutations in the direct vicinity of this site.CVYV-JOR had a V to I mutation at position 91, and OM04-06 had an E to G mutation at position 101, but no common mutation was detected.Top necrosis was frequently observed upon infection with different CVYV isolates, which might be considered as a systemic hypersensitive response (Abebe et al. 2021).As described for cucumber mosaic virus, a small decrease in the level of resistance induction could change a hypersensitive response (not visible in PI 164323 but that may take place at the cellular level) to systemic necrosis that may account for the low and variable frequency of top necrosis in PI 164323 upon inoculation with CVYV-Esp.Determinants for the top necrosis response may thus be different from those involved in the bona fide resistance breaking observed for CVYV-PI1.PI 164323 was also observed to display occasionally top necrosis even in the absence of virus infection, although the mechanisms involved are not known (M.Pitrat, unpublished).
Mutations in the VPg have been associated with resistance-breaking virus isolates in several plantvirus interactions (Truniger and Aranda 2009), notably among potyviruses where direct interaction between the VPg and the plant translation initiation factors eIF4E and/or eIF(iso)4E is required for infectivity (Le Gall et al. 2011;de Oliveira et al. 2019).In these cases, the resistance is recessive (Truniger and Aranda 2009 ;Revers and Garcia 2015), and mutations in the highly disordered central region of the VPg affect the direct interaction with eIF4E (Ala- Poikela et al. 2019 ;Charon et al. 2018).Melon lines silenced for eIF4E showed resistance to several potyviruses and also to CVYV (Rodriguez-Hernandez et al. 2012), suggesting that the resistance mechanisms could be related in both cases to a lack of functional interaction between virus VPg and plant eIF4E.Cucumbers with the eIF4E gene disrupted by CRISPR/Cas9 technology were also immune C. Desbiez Phytopathology to CVYV, the resistance being recessive as for potyviruses (Chandrasekaran et al. 2016).Contrary to most potyvirus resistance, resistance to CVYV in melon PI 164323 appeared to be controlled by a dominant allele Cvy-1 1 (Pitrat et al. 2012), which does not fit with a hypothesis of the resistance allele being a plant component, e.g.eIF4E, required for infectivity.There was no amino acid difference in the eIF4E between Védrantais and PI 164323 (C. Dogimont, pers. comm.).PI 164323 presented a nonsynonymous mutation in the eIF(iso)4E protein but the same mutation was also present in accession 90625 (C.Dogimont, pers.comm.) that shows no resistance to CVYV-Esp (Pitrat et al. 2012).Thus, there was no obvious correlation between eIF4E or eIF(iso)4E coding sequences and resistance to CVYV.Besides eIF4E, the VPg of potyviruses interacts with several host and virus proteins (Jiang and Laliberté 2011).Potyvirus VPg contributes directly to inhibiting plant antiviral silencing response (Rajamäki et al. 2014 ;Cheng and Wang 2017), and it also interacts with the viral silencing suppressor HC-Pro (Jiang and Laliberté 2011).CVYV VPg may play a role in silencing suppression as well, either by itself or cooperatively through interaction with other viral products including the silencing suppressor P1b that functionally replaces the potyviral HC-Pro (Valli et al. 2006).These properties might be affected in the resistance-breaking mutant.Alternatively, CVYV VPg may be recognized as an avirulence factor by a host resistance gene in PI 164323, inducing an effector-triggered immune response leading to complete resistance or to top necrosis depending on the timing or efficiency of the reaction.More studies will be needed to understand the nature of the Cvy 1 gene and all the details of the interaction between CVYV and melon.
Since the resistance conferred by Cvy 1 1 is isolate-specific, it should not be used in the field in areas where necrosis-inducing isolates are present.A single mutation, corresponding to an A Page 18 of 25 C. Desbiez Phytopathology to G transition in the virus genome, was sufficient to induce resistance breaking in PI 164323.This constitutes the least favorable situation for resistance durability (Fabre et al. 2009).
Resistance breaking might have a fitness cost, as observed for other plant-virus interactions (Desbiez et al. 2003 ;Janzac et al. 2010), which could contribute to preserving resistance durability through optimized management of the resistance deployment (Rimbaud et al. 2021).
However, this may not be sufficient to ensure satisfactory durability of the resistance in the field, and thus a durable control of CVYV, without being associated with other control measures.

Figure 2a :
Figure 2a: Alignment of VPg amino acid sequences of twelve CVYV isolates and eight other

Figure 2b :
Figure 2b : Overlay of predictions of disorder for the VPg of CVYV-Esp (in red) and CVYV-PI1 (in Figure 1

Table 1
Phytopathologycases no symptoms were visible and no virus was detected by DAS-ELISA but occasionally systemic necrosis was observed.Accession HSD 2458 [Cvy-1 + cvy-2 Cvy-3 + ] carrying the cvy-2 allele conferring tolerance to CVYV-Esp was tolerant and displayed only very mild mottle symptoms upon inoculation with the different isolates, even though the Jordanian and Oman isolates were more severe than the others.The melon accession PI 164323 [Cvy-1 1 cvy-2 + Cvy-3] carrying the resistance allele Cvy-1 1 displayed complete resistance with no visible symptoms and no virus detection by DAS-ELISA to isolate CVYV-SU12-10.Upon inoculation with CVYV-Esp and CVYV-Isr, most inoculated plants (>90%, n>30) displayed no symptoms and no virus accumulation at the systemic level (Table 2 allele responsible for lethal necrosis upon inoculation with CVYV-Esp, all isolates except SU96-84 induced 100% top necrosis or systemic necrosis.Upon inoculation with SU96-84, in most C. Desbiez