A molecular marker set combining a retrotransposon insertion and SSR polymorphisms is useful for assessing diversity in Vitis

Molecular markers, based on DNA polymorphisms, are useful tools for identifying individuals, establishing phylogenetic relationships, managing collections of genetic material or assisting breeding. In the present study, we developed a marker set to differentiate Vitis species, grapevine varieties or clones belonging to the same variety. This novel marker set combines, in four PCR amplifications, the presence/absence of a remarkable retrotransposon, Tvv1-Δ3460, inserted at its single locus and the SSR polymorphism present within its two LTRs. By studying a collection of Vitaceaeaccessions, we showed the prevalence of two allelic forms of Tvv1-Δ3460 - one of which was partially truncated - in Vitis species. Out of the twenty-five studied Vitis species, the insertion of a Tvv1-Δ3460 element was detected in twenty, including Vitis vinifera. The homozygous vs heterozygous state of the element insertion was determined by amplifying the empty site. Additionally, each Tvv1-Δ3460 LTRs included a microsatellite sequence useful for designing markers based on LTR length. The LTR-SSR markers distinguished most of the fifty-two cultivars and revealed polymorphism within five of the seven varieties studied.


INTRODUCTION
Molecular markers are useful tools for identifying individuals, establishing phylogenetic relationships, managing collections of genetic material or assisting breeding. They are based on various types of DNA polymorphisms found in genomes (Gupta et al., 1999;Schulman, 2006). Microsatellite markers, whose polymorphism takes advantage of the variable number of simple sequence repeats (SSR) at a given locus, is undoubtedly the most extended molecular marker system for grapevine. SSR markers are highly transferable, co-dominant, and very useful for identifying grapevine cultivars (Merdinoglu et al., 2005; and for studying Vitis phylogeny (Di Gaspero et al., 2000). However, most available SSR markers fail to distinguish clones that derive from repeated vegetative propagation cycles from a unique single individual. Nevertheless, a standard set of five SSR markers (VMC3a9, VMC5g7, VVS2, VVMD30 and VVMD32) which can reveal clonal polymorphism has been proposed (Pelsy et al., 2010). Somatic variations giving rise to clone diversity within grapevine varieties have also been investigated by genomic approaches; 15 distinguishable Chardonnay clones have been identified by 1620 SNPs and InDels, which can be exploited to define markers for clone-specific genotyping (Roach et al., 2018).
Mobile elements are abundant, rapidly evolving and widespread in the genomes of plants. They actively contribute to molecular polymorphism and, therefore, form the basis of other molecular marker systems. They take advantage of the activity or the structural variations of transposable elements.
Sequence-specific amplification polymorphisms (SSAP) markers, which reveal the pattern of insertion of elements belonging to the same family (Waugh et al., 1997), have been developed to analyse genetic diversity and relatedness in the genus Vitis (Moisy et al., 2008b). Retrotransposon polymorphism fingerprinting (RUP), which amplifies the highly variable untranslated leader (UTL) region of the Tvv1, has revealed a unique pattern in each of 94 Vitaceae accessions and is conserved between clones (Pelsy, 2007). Finally, inter retrotransposon amplified polymorphism (IRAP) and retrotransposonmicrosatellite amplified polymorphism (REMAP) have been shown to discriminate the white table grape cultivar Italia from its coloured variants (Rubi, Benitaka, Brasil and Black Star) derived from clonal propagation of somatic mutations (Strioto et al., 2019).
In the grapevine genome, several retrotransposons have been found inside genes, such as Vine1 in Adhr (Verriès et al., 2000), or close to genes, such as Gret1 in the promotor of VVMybA1 (Kobayashi et al., 2004). Others have been characterised by a computerized sequence similarity search and FIGURE 1. Schematic representation of the three alleles of the locus. grouped into families sharing an amino acid identity of ≥ 90 % (Moisy et al., 2008b), among which the Tvv1 family, which is a Ty1-copia like LTR-retrotransposon, has been extensively studied (Moisy et al., 2008a). Tvv1 full-length copies of around 5 kb in size share an internal region with a single highly conserved open reading frame and, upstream, an untranslated leader (UTL) region which is highly variable in size. This internal region is flanked by LTRs between 149 and 157 bp long (Pelsy and Merdinoglu, 2002). The Tvv1 family also comprises unique copies of remarkable Tvv1 elements that have suffered large deletions and are fixed at single loci. Among them, Tvv1-Δ3460, which has undergone a major 3,460 bp long deletion in the coding sequence compared to the full-length copies, is located on chromosome 8 of the grapevine genome. In Pinot noir cv, Tvv1-Δ3460 is 2,074 pb long, but it slightly differs in size in the genomes of other varieties hosting this element, mainly because of the number of TA motifs in a microsatellite stretch located in both LTR sequences (Moisy et al., 2008a).
Molecular markers based on the insertion of retrotransposons can be used for identifying grapevine species and cultivars (D'Onofrio et al, 2010), but they generally fail to reveal clonal polymorphism (Pelsy, 2007). Nevertheless, transposable elements are responsible for the main proportion of somatic mutations affecting four Pinot Noir clones (Carrier et al., 2012) and have been found to cause diversification of 15 Zinfandel clones (Vondras et al., 2019). In the present study, we developed a marker set based on the combination of the presence/absence of the remarkable retrotransposon Tvv1-Δ3460 at its insertion site and the SSR polymorphism within its LTRs. By studying a collection of Vitaceae accessions, we showed the relevance of Tvv1-Δ3460-based markers to distinguish in a simple way most of the Vitis accessions, and to reveal clonal polymorphism within different varieties.

Plant material and DNA extraction
The plant material consisted in Vitaceae accessions divided into three groups. The Vitaceae group comprised Ampelopsis (5 accessions), Parthenocissus (1 accession), and Vitis genera (Muscadinia rotundifolia (4 accessions), Asian species (6 accessions), North American species (16 accessions)) and inter-specific hybrids (2 accessions). The V. vinifera group contained cultivated grapevine varieties (52 accessions) and wild vines (6 accessions). Due to the dioicity of wild vines, a male and a female were chosen for each origin of sampling: Ste Croix en Plaine (53) and Mandeure (C25) in France, and Martigny (50) in Switzerland. All accessions are kept in the ampelographic collection of INRAE-Colmar (France). In addition, the segregation of the LTR-SSR alleles was assessed in a progeny of Riesling x Gewurztraminer comprising 11 randomly chosen individuals. The third group comprised seven wine grape varieties represented by a random collection of certified clones conserved in the French national repository (ENTAV, Le Grau du Roi, France), and by accessions recovered from field selections and kept in different germplasm repositories at INRAE, the French National Research Institute for Agriculture, Food and Environment : Cabernet franc (17 clones) , Cabernet-Sauvignon (37 clones), Chenin blanc (19 clones), Grolleau (48 clones), Pinot noir (23 clones), Riesling (27 clones), and Savagnin (45 clones including 22 Savagnin blanc, 19 Gewurztraminer, 4 Savagnin rose).
Total DNA was purified from young expanded leaves from individual plants using Dneasy TM Plant Mini-Kit (Qiagen, Hilden, Germany) as described by the supplier.

PCR conditions and fragment analysis
According to Moisy et al. (2008a), the internal PCR products A-750 and B-1500 characterised the Tvv1-Δ3460 copy (Figure1). Due to the presence of full-length copies of Tvv1 elsewhere in the genome, both amplifications also produce 4,3Kb and 4Kb long fragments respectively (Moisy et al., 2008a). When present at its insertion site, Tvv1-Δ3460 was revealed by a long-range PCR product 2500 bp long, F-2500, using Pvv78x primers designed in the host flanking regions of the insertion. These primers also amplify the empty site, E-450, as well as a potential fulllength Tvv1 element that can be inserted at the site (Moisy et al., 2008a). To take advantage of the polymorphism of the microsatellite stretch located in both LTR sequences of Tvv1-Δ3460, 5' and 3' LTR-SSR markers primers were designed using Primer3 software (Rozen & Skaletsky, 1998) and synthesised by MWG Biotech AG (Ebersberg, Germany). One primer of each pair was HEX and 6-FAM fluorophore-labelled (PE Applied Biosystems, Warrington, UK) respectively to amplify the 5' and 3' LTR-SSR markers in multiplex (Table 1). All primer locations are given in Figure 1 and all new sequences in Table 1. PCR amplifications were carried out according to Hocquigny et al. (2004). The programme consisted of the following steps: 5 min at 94 °C, followed by 30 cycles of 30 s at 92 °C, 30 s at 52 °C, 30 s at 72 °C, and a final extension step of 7 min at 72 °C. PCR fragments were resolved on an automated 310C ABI PRISM DNA sequencer (PE Applied Biosystems, Foster City, CA), and sized with an ROX labeled-internal standard (50-654 bp) (PE Applied Biosystems, Foster City, CA). Microsatellite alleles were scored using GenScan (version 3.1) and Genotyper (version 2.5.2) software (PE Applied Biosystems, Foster City, CA). All polymorphisms were confirmed by at least two analysis.
1. Tvv1-Δ3460 insertion; 2: 5'ΔTvv1-Δ3460 insertion and 3: the empty site Arrows boxes represent LTRs including the SSR stretch, the black box the UTL and the triple line ORF regions. The 5 bp-sequence duplicated to flank the retrotransposon is represented by a triangle. Dotted lines represent the 5' deletion of 5'ΔTvv1-Δ3460 and of its flanking region. Black arrows represent primers used in the study, whose numbers are shown in Table 1. Primers 1 and 2 amplify 5'LTR-SSR and primers 3 and 4 amplify 3'LTR-SSR. Pvv78x primers, located in the host region flanking the insertion, amplify either Tvv1-Δ3460 (F-2500) or the empty site (E-450). Primer positions are given according to the LTRs. Primers that generate A-750 an B-1500 fragments are given in Moisy et al. (2008a).

RESULTS
The presence/absence of Tvv1-Δ3460 at its expected insertion site was revealed by three PCR amplifications. The A-750 and B-1500 fragments, which overlap the specific internal region of Tvv1-Δ3460, characterised this element (Moisy et al., 2008a). The Pvv78x primers designed in the host flanking regions of the insertion amplified a 2500 bp-long fragment, F-2500, when Tvv1-Δ3460 was present. Conversely, this pair of primers produced a 450 bp-long fragment, E-450, when the site was empty. In some accessions, A-750 was amplified but not B-1500 or E-450. This combination of PCR fragments revealed a new element whose 5'LTR was truncated, which was named 5'ΔTvv1-Δ3460. No amplification of this new element was possible using the Pvv78x primers leading to the conclusion that its 5'host flanking region must have been deleted as well. Finally, the insertion of a full-length Tvv1 copy, with an average size of 5 kb, was investigated at the given site by long-range PCR using Pvv78x primers. Among the accessions in the study, none revealed the expected 5,5 Kb-long fragment.
In addition, as can be seen in Table 1, two markers, 5' and 3' LTR-SSR, were developed to characterise Tvv1-Δ3460 LTRs. 5'LTR-SSR was amplified using primer 5_3460_fw, annealing upstream of Tvv1-Δ3460, and paired with primer 5_3460_rv designed in the UTL sequence of the element. Similarly, 3'LTR-SSR was amplified using primer 3_3460_fw, designed in the RNAse sequence of Tvv1-Δ3460, and paired with primer 3_3460_rv located downstream of the element.

Tvv1-Δ3460 in Vitaceae species
A collection of 34 Vitaceae species (Table 2) was investigated for the presence/absence of Tvv1-Δ3460. Ampelopsis species did not amplify A-750, B-1500 or E-450, which is most likely due to a lack of homology of the primers designed from the sequence of V. vinifera with the target sequence of Ampelopsis. The Parthenocissus, the 4 Muscadinia and 5 of 24 Vitis accessions amplified E-450 and not A-750 or B-1500, leading to the conclusion that Tvv1-Δ3460 is absent in these accessions, while the Pvv78x primers are effective at amplifying the empty site.  the empty site of insertion of Tvv1-Δ3460, but did not amplify any of the 6 Asian Vitis.
The polymorphism of the LTRs of the inserted elements were investigated using 5' and 3' LTR-SSR markers (Table 1). In the collection of the 19 Vitaceae species hosting at least one copy of Tvv1-Δ3460, a total of 18 alleles of 240 to 290 bp in size and 16 alleles of 322 to 371 bp in size were scored for 5'and 3'LTR-SSR markers respectively.

Vitis aestivalis
North America --+ ---  The two LTR-SSR markers defined a specific genotype for all of these Vitis species, except for V. riparia Millardet and V. riparia Muller, which shared the same genotype. The LTR-SSR genotype of V. rupestris du Lot is consistent with the association of two copies of 5'ΔTvv1-Δ3460, whose 3'LTR were different in size.

Tvv1-Δ3460 in Vitis vinifera
The presence/absence of Tvv1-Δ3460 was also evaluated in a collection of V. vinifera accessions comprising 52 cultivated varieties and 6 wild vines. On the one hand, all the accessions amplified A-750, while none amplified E-450, indicating the prevalence of Tvv1-Δ3460 insertion in the V. vinifera species. On the other hand, 55 of the 58 accessions amplified B-1500. The three varieties Mourvèdre, Roussanne and Semillon that did not amplify B-1500 presumably hosted only 5'ΔTvv1-Δ3460 (Table 3).
In the V. vinifera collection, the amplification of 5' and 3' LTR-SSR markers each provided 13 alleles of 239 to 297 pb and 324 to 353 bp in size respectively. These two markers made it possible to characterise 38 genotypes which differed from those previously characterised in the collection of 34 Vitaceae species. Twenty-nine genotypes were displayed by one variety only, 5 genotypes by 2 varieties, 1 genotype by 3 varieties and 4 genotypes by 4 varieties. Varieties sharing the same genotype were known to be related such as the progeny of Pinot noir × Gouais blanc cross. Out of the 58 Vitis vinifera, 22 showed two alleles at each locus, indicating the presence of two full copies of Tvv1-Δ3460 with polymorphic LTRs. Thirty-one accessions, among them Pinot noir and Gewurztraminer, displayed one allele for 5'LTR-SSR and 2 alleles for 3'LTR-SSR. These varieties associated either two copies of Tvv1-Δ3460 with 5'LTRs of the same length

LTR-SSR polymorphism within seven French wine grape variety collections
The capacity of the two Tvv1-Δ3460 LTR-SSR markers to reveal intra-varietal polymorphism was evaluated in seven clone collections of wine grape varieties: Cabernet franc, Cabernet-Sauvignon, Chenin blanc, Grolleau, Pinot noir, Riesling, Savagnin. These collections comprised These collections of clones were chosen for comparison since they had previously been evaluated with 12 SSR markers for the collections of Grolleau, Cabernet franc, Chenin blanc, and with 30 markers for those of Cabernet-Sauvignon, Pinot noir, Riesling and Savagnin (Pelsy et al., 2010).
In this study, the reference genotype was defined as that the genotype shared by the majority of the accessions of a collection (Table 4). All varieties amplified one allele at locus 5'LTR-SSR and six amplified two alleles at locus 3'LTR-SSR, except Riesling. Two variants were detected at locus 5'LTR-SSR: one Savagnin that displayed a 275 bp-long allele, instead of the 277 bp-long reference allele, and one Cabernet franc that showed the heterozygous genotype [277:285], instead of the reference genotype, which was most probably homozygous [277:277]. With the appearance of 10 new alleles, the 3'LTR-SSR locus was far more susceptible to polymorphism than 5'LTR-SSR (2 new alleles). One to 5 variant genotypes were observed in 5 of the varieties, mainly as a result of the addition of a new allele to the reference genotype, leading to triple-allele genotypes characterising periclinal chimeras.
Thus, clonal polymorphism was revealed in 5 of the 7 clone collections. Four variants were characterised out of the 17 clones of Cabernet franc, 3 out of the 19 clones of Chenin, 2 out of the 23 clones of Pinot noir, 3 out of the 27 clones of Riesling and 10 out of the 45 clones of Savagnin, most of the latter being Savagnin blanc. One clone of Cabernet franc was polymorphic at both loci. Altogether, it was possible to unambiguously distinguish 11 clones out of the 216 studied by a unique genotype, while 2 to 5 variants shared the same genotype.
Conversely, no variants were detected among the 37 and 48 clones of Cabernet-Sauvignon and Grolleau respectively.

DISCUSSION
Having effective and easy-to-use markers capable of quickly identifying Vitis species, grapevine varieties and clones belonging to the same variety is still a challenge today. In this study, we describe a novel set of molecular markers based on a combination of the variation in the presence of the remarkable retrotransposon Tvv1-Δ3460 at its insertion locus and the SSR polymorphism of its LTRs to assess grapevine diversity. We used it to explore species, varietal and intra-varietal diversity. In 4 PCR amplifications (B-1500, E-450 and 5' and 3' LTR-SSR in multiplex), it was possible to characterise all the species and varieties. In addition, the 3'LTR-SSR marker was shown to be highly relevant in revealing clonal polymorphism.

Presence vs absence of Tvv1-Δ3460 is informative
The Tvv1-Δ3460 locus showed different alleles in the Vitaceae species revealed by the long-range PCR using Pvv78x primers. Ampelopsis accessions did not amplify Tvv1-Δ3460 or its empty site, which is likely because the primers designed from a V. vinifera sequence were not homologous enough for their target sites in Ampelopsis, due to the phylogenetic distance between both genera. The Parthenocissus, the four Muscadinia and five of the American Vitis or hybrid accessions (V. aestivalis, V. candidans, V. cinerea, V. cordifolia Couderc and V. linsecumii) only amplified the empty site of the locus (Figure 2). This result agrees with previous studies which showed the divergence between the genera Ampelopsis and Vitis, but brought Parthenocissus quinquefolia as close as Muscadinia to Vitis (Pelsy, 2007).
Four American Vitis or hybrid accessions (V. berlandieri Colombard, V. berlandieri Planchon, V. doaniana, V. monticola Large Bell) amplified both the full and the empty site of Tvv1-Δ3460 ( Figure 2). All the remaining accessions, including the Asian accessions and the wild or cultivated V. vinifera, displayed a Tvv1-Δ3460 element, but never the empty site.
These results indicate that the Tvv1-Δ3460 insertion is specific to the Vitis species. Either the allele with the full site or the one with the empty site of Tvv1-Δ3460 were present homozygous or heterozygous in the American Vitis or hybrids accessions. Conversely, all the Asian accessions and V. vinifera varieties hosted a Tvv1-Δ3460 insertion at the locus. This insertion was most probably dispersed through natural intermixing, due to the close proximity of Asian Vitis with the European species. However, the empty site which is remained in half of American Vitis confirms the disjunction between the Old and New World (Zecca et al., 2012).
The formation of deleted elements may occur either during the retrotransposition process prior to integration, or by illegitimate recombination within an integrated full-length element. Since no full copy of Tvv1 was amplified via long-range PCR amplification using the Pvv78x pair, the large internal deletion characteristic of Tvv1-Δ3460 most likely occurred during the retrotransposition process before integration.
Finally, 5'ΔTvv1-Δ3460, a new element that was subject to a deletion of the 5' region of Tvv1-Δ3460 and of its 5'host region, was identified in V. rupestris du Lot and in many V. vinifera varieties. Nevertheless, the 4 PCR does not allow us to conclude that the V. rupestris du Lot and V. vinifera varieties share the very same 5'ΔTvv1-Δ3460 element, or that independent deletions lead to different 5'ΔTvv1-Δ3460 elements.
In some varieties, two copies of Tvv1-Δ3460 or of 5'ΔTvv1-Δ3460 can be combined, such as in Sylvaner or Semillon respectively, but with 3'LTRs of different lengths. Other varieties, such as Pinot noir, Riesling and Gewurztraminer, combine Tvv1-Δ3460 and 5'ΔTvv1-Δ3460. The elements of Pinot noir and Gewurztraminer have 3'LTRs of different lengths, while those of Riesling have 3'LTR of the same length ( Figure 2).

5' and 3' LTR-SSR: highly informative markers which can be used to identify Vitis accessions
In the Vitis species which hosted at least one copy of Tvv1-Δ3460, the 5' and 3' LTR-SSR markers are well-conserved, defining a specific genotype for all of these Vitis species. Moreover, when considering the 58 V. vinifera accessions (cultivated grapevine varieties and wild vines), the same group studied with 14 SSR markers showed that the number of alleles detected per locus ranged from 1 for VMC1e11a to 14 for VVS2, with an average of 8.6 (Pelsy, 2007). 5' and 3' LTR-SSR each amplified 13 alleles; they are therefore among the most informative markers for identifying V. vinifera accessions.

LTR-SSR polymorphism within seven French wine grape variety collections
Not all microsatellite markers can reveal clonal polymorphism within clone collections, and therefore a standard set of five microsatellite markers (VMC3a9,VMC5g7, VVS2, VVMD30, and VVMD32) has been proposed (Pelsy et al., 2010).
It was possible to develop 5' and 3' LTR-SSR markers due to the systematic presence of an insertion -either of Tvv1-Δ3460 or 5'ΔTvv1-Δ3460 -at the locus of different grape varieties, and to a TA microsatellite stretch in the LTR sequences of these Tvv1 elements. Using these markers, polymorphism was assessed in the clone collections of seven varieties each comprising 17 to 48 clones. Variant genotypes were revealed by 3'LTR-SSR in five of the seven clone collections. Thus, this marker is more effective than all the previously identified SSR markers: indeed, VMC3a9, VMC5g7 and VVS2 revealed variants in four of these collections of clones, while VVMD30 and VVMD32 in three and two collections respectively.
The 5 'LTR-SSR marker, meanwhile, often only amplified a fragment in the studied varieties, as was the case for Pinot noir, Riesling and Gewurztraminer, which are all hemizygous due to the combination of Tvv1-Δ3460 and 5'ΔTvv1-Δ3460, and 5'ΔTvv1-Δ3460 resulting in a null allele with 5' LTR-SSR. Therefore, the 5' LTR-SSR marker is less effective at revealing clonal polymorphism than the 3' LTR-SSR marker.
In a previous study, the same clones were studied with SSRs to identify a standard set of five markers (VMC3a9, VMC5g7, VVS2, VVMD30 and VVMD32) which revealed clonal polymorphism within different varieties (Pelsy et al., 2010). The addition of the LTR-SSR markers improved the identification of the clones in 4 of the 7 varieties by distinguishing new clones. Therefore, while the standard set markers made it possible to assign a unique genotype to 1 to 5 clones in the collections, the additional LTR-SSR markers allowed 2 to 4 new clones with a unique genotype to be identified, thus increasing the number of distinguishable clones; for example, for Savagnins, the number of clones displaying a unique genotype increased from 5 to 9 (Table 5).
Genomic analysis provides access to a large number of SNPs and InDels that can be exploited to define markers capable of characterising interclonal diversity in different grapevine varieties. Nevertheless, these approaches are limited by sequencing technology or the lack of a reference genome for the studied varieties (Roach et al., 2018). Furthermore, SSR markers, which are highly transferable between different varieties and Vitis species, are simple tools which can be used to identify variant clones in a wide range of varieties. Thus, to increase the capacity for revealing clonal polymorphism with SSR markers, we recommend that 3 ′ LTR-SSR be added to the standard set of previously characterised VMC3a9, VMC5g7, VVS2, VVMD30 and VVMD32 (Pelsy et al., 2010) CONCLUSION Deleted retroelements are unique, very stable and can be considered as Mendelian loci. They can be identified within different families of retrotransposon by PCR amplification with primers derived from the LTRs or the conserved PBS or PPT sequences. Using available grapevine genome sequences, their flanking sequences can then be identified to design primers that will yield easy-to-use co-dominant markers. Their specific dispersion in the Vitaceae make these markers valuable tools for studying their diversity.
Butterlin, Roxanne Hardy and Angélique Ardiller for the laboratory experiments.