Functional conservation of the grapevine candidate gene INNER NO OUTER for ovule development and seed formation

Seedlessness represents a highly appreciated trait in table grapes. Based on an interesting case of seedless fruit production described in the crop species Annona squamosa, we focused on the Vitis vinifera INNER NO OUTER (INO) gene as a candidate. This gene encodes a transcription factor belonging to the YABBY family involved in the determination of abaxial identity in several organs. In Arabidopsis thaliana, this gene was shown to be essential for the formation and asymmetric growth of the ovule outer integument and its mutation leads to a phenotypic defect of ovules and failure in seed formation. In this study, we identified in silico the V. vinifera orthologue and investigated its phylogenetic relationship to INO genes from other species and its expression in different organs in seeded and seedless varieties. Applying cross-species complementation, we have tested its functionality in the Arabidopsis ino-1 mutant. We show that the V. vinifera INO successfully rescues the ovule outer integument growth and seeds set and also partially complements the outer integument asymmetric growth in the Arabidopsis mutant, differently from orthologues from other species. These data demonstrate that VviINO retains similar activity and protein targets in grapevine as in Arabidopsis. Potential implications for grapevine breeding are discussed.


Introduction
Grapevine (Vitis vinifera L.) is one of the most cultivated and appreciated fruit crop trees in many regions of the world. It is cultivated mainly for wine-making, but also for fresh consumption and raisins. Breeding programs are focused either on pathogen resistance or on qualitative traits appreciated by consumers, among which seedlessness for table grape 1 . Two different types of seedless fruits have been observed in grape, caused either by parthenocarpy or stenospermocarpy 1,2 . In parthenocarpy, fruit development occurs in the absence of ovule fertilization, leading to a complete lack of seeds 3 . In stenospermocarpy, ovule fertilization takes place but seed development fails because of the embryo and/or endosperm degeneration. Stenospermocarpic cultivars are not strictly seedless but they contain seminal rudiments or seed traces of different sizes 2 . Genetic studies have mapped grape QTLs for seeds related trait to several genomic regions (see refs. 4,5 for a summary). The Sultanine (or Thomson seedless) cultivar contributes to the SEED DEVELOPMENT INHIBITOR (SDI) locus, the major source of seedlessness exploited for breeding purposes and in commercial grapevine cultivars [6][7][8][9][10] . A recent study has elucidated the molecular basis of this trait demonstrating this is associated with a missense mutation in the MADS-Box gene VviAGL11 controlling seed coat development and lignification 11 . Besides this major gene, genetic studies indicate that alternative less exploited genes contribute to the seedless phenotype in grapevine.
An interesting case of seedless fruit production was described in the spontaneous mutant Thai seedless (Ts) of the crop species Annona squamosa (sugar apple), belonging to an early divergent angiosperm clade 12 . This mutant produces fully seedless normal size fruits following pollination. The authors demonstrated that failure in seed formation is due to a defect in the ovules which lack the outer of the two normal integuments, phenocopying the Arabidopsis inner no outer (ino-1) mutant 13 . They isolated the Annona INO orthologue gene and showed that the mutant was indeed associated with the deletion of the INO locus, revealing the molecular basis of seedlessness in the Ts mutant and providing an interesting candidate with the potential of introducing seedlessness in further crop species.
The INO gene encodes a putative transcription factor belonging to the YABBY gene family, involved in the determination of abaxial identity in a variety of plant organs [14][15][16] . INO and its role in ovule development were mainly characterized in the model plant species Arabidopsis. Phenotypic analyses in the ino-1 mutant have shown, besides the lack of the outer of the two integuments which normally cover the nucellus in plant ovules, also the absence of the typical hoodlike structure (amphitropous) characteristic of wild-type ovules. Indeed, the amphitropous configuration is due to the asymmetric growth of the outer integument, and it is therefore lost in the mutant plants that show the micropyle not adjacent, but in a line, with the funiculus. Moreover, ino-1 plants exhibit a strongly reduced number of viable seeds and, differently from the Ts Annona crop mutant, reduced siliques expansion [17][18][19] . Further molecular characterizations have shown that the ino-1 mutant phenotype in Arabidopsis was due to a G-to-A transition close to a splice acceptor site of the gene, which leads to a frameshift mutation affecting the coded protein. Investigation on the spatial distribution of INO transcript accumulation showed that INO, prior to the visible emergence of the integuments, is expressed specifically in the epidermal cells on the abaxial half of ovule primordium, the region corresponding to the site of outer integument initiation. Therefore, INO was suggested to be involved in the polar determination and to work as a primary determinant of abaxial identity in the ovule part from which the outer integument originates through strict control of its expression pattern 13 . Further studies showed that INO expression is regulated by a positive autoregulatory loop and that this loop is attenuated by the SUPERMAN(SUP) repressor in the adaxial ovule side. Indeed, in the sup mutant, the outer integument grows also on the adaxial side of the ovule primordium, resulting in ovules with a nearly radially symmetrical tubular shape 17 . This molecular mechanism directly controls the polar development of ovule outer integument 13,20 .
Beside studies in the model plant Arabidopsis, INO orthologues were characterized in a number of different taxa and species. The highly specific expression pattern described in Arabidopsis was compared to the expression pattern in early-diverging bitegmic angiosperms such as Nymphaea alba and Cabomba caroliniana, from the Nymphaeales order, or Amborella trichopoda, from the Amborellales [21][22][23][24] . These studies showed that INO expression pattern in these early-diverging lineages exactly parallels that observed in Arabidopsis indicating that exclusive expression of INO in the abaxial epidermis of the outer integument is primitive and has been conserved from early stages of angiosperm evolution. Further comparison of INO orthologues expression pattern in unitegumic species confirmed the specific expression in the abaxial outermost cell layer also of the single ovule integument, indicating widely conserved INO function across all angiosperms 12,[25][26][27] .
Despite the described conservation in the expression pattern, studies addressing INO functional conservation, including comparative characterization of protein targets and activities are still few. While virus-induced knockdown of expression of the INO orthologue in Nicotiana benthamiana, a representative of the unitegumic Solanales in the asterids clade, inhibited growth of the outermost cell layer of the unique ovule integument, leading to a decrease in both integument extension and ovule curvature, the tomato SlINO CDS failed to complement the Arabidopsis ino-1 mutant phenotype, indicative of divergences in protein targets and activity 27 . Therefore, despite the claim of a widely conserved role for INO also to unitegumic species and to near the base of the angiosperms, cross-species complementation did not support that so far.
In this study, starting from the recent characterization of the YABBY gene family in grapevine 28 we have identified the grapevine VviINO gene and investigated its phylogenetic relationship to INO orthologues from other species and its expression in different plant organs and seeded and seedless cultivars. Moreover, we conclusively demonstrated that the grapevine orthologue VviINO retains similar protein targets and activity in the grape as in Arabidopsis, since it could fully restore the outer integument growth in Arabidopsis ino-1 mutant. Some specificities in the asymmetric growth were also observed and discussed. These data provide relevant information on a new candidate with potential implications for table grape breeding.

Results
In silico identification and annotation of the grapevine VviINO gene The genomic organization of the YABBY gene family members in V. vinifera, including the gene encoding VviINO, was recently described by Zhang et al. 28 . To validate and refine the VviINO annotation, structural information for the whole gene family was updated to the newest genome assembly 12X.v2 (Supplementary Table  S1). Minor differences in genomic positions, length of genes, related CDS structures and predicted protein sequences were noted according to the different available annotations. However, careful inspection of multialignments showed that predicted proteins, according to two latest and most used annotations V2 or VCost.v3 29,30 were congruent for all family members, with the only exception of VIT_201s0011g00140 which is different in the N-terminal protein portion and VIT_206s0009g00880 missing 2aa in the VCost.v3 protein version compared to V2 annotation ( Supplementary Fig. S1). In silico analysis of predicted protein sequences confirmed the presence of nuclear localization signals, in line with the expected function, as well as the presence of the conserved YABBY domain located from 2nd to 162th aa, including both hallmarks motifs of the family, i.e., the Zinc finger toward N-terminal and the helix-loop-helix YABBY domain toward C-terminal protein portion [14][15][16] (Supplementary  Table S2).
Direct orthologous relationships between V. vinifera and Arabidopsis YABBYs were established based on CDS analysis. Among all grape YABBYs coding sequences, VIT_201s0127g00330 located on chromosome 1 formed a clade with A. thaliana AT1G23420 encoding for INO, suggesting this gene as the corresponding grapevine INO orthologue (Fig. 1). Distance matrix estimated from both DNA coding sequence and protein sequence alignments confirmed this closer relationship (Supplementary Table  S3), consistent with previous analysis and the genomic location in syntenic blocks between grape and Arabidopsis genomes 28 . Following guidelines established by the Super-Nomenclature Committee for Grape Gene Annotation (sNCGGa) 31 , we have revisited the whole grapevine YABBY family nomenclature, and propose to rename VIT_211s0016g05590 as VviYAB5 according to its highest sequence similarity to Arabidopsis YAB5. VIT_206s0009g00880 and VIT_208s0032g01110, showing both highest sequence similarity to Arabidopsis YAB2, were renamed as VviYAB2a and VviYAB2b as a paralogous set of genes located on different chromosomes. Another putative paralogous set, VIT_202s0154g00070, and VIT_215s0048g00550, showed equal sequence similarity to the Arabidopsis YAB1 and YAB3 genes which are recognized as belonging to a subclade in the family 22 . Since no one2one orthologous relationship could be established, we renamed these as VviYAB6 and VviYAB7 respectively, using numbers higher than the highest already used for both Vitis and Arabidopsis according to rules defined by sNCGGa 31 . VIT_201s0011g00140 was finally renamed as VviCRC due to the high similarity with the characterized Arabidopsis CRC.

Phylogenetic relationship among INO proteins from different species
Evolutionary relationships for grape INO with species in which this gene was previously characterized were enquired.
INO protein function has been first deeply characterized in the model plant A. thaliana, where its role in integument development during ovule formation was widely studied 24 . Sequence comparison revealed that grape INO shares 79% similarity and 69% identity with the Arabidopsis INO. However, while the Arabidopsis INO encodes a 231 aa protein 13 VviINO encodes a shorter protein (176 aa). Sequence alignment revealed nearly complete conservation inside both the zinc finger domains as well as in the HLH domain, each presenting only five unconserved amino acids. High conservation was found also in border regions of both conserved domains and in the protein portion among the two domains, while terminal protein portions were more divergent, with a long C-terminal unshared sequence only present in the Arabidopsis INO ( Supplementary Fig. S2).
Beside Arabidopsis, studies on INO have been extended to further species also with divergent ovule morphology, in the attempt to clarify the evolutionary steps behind ovule development across angiosperms and especially the reduction of integuments number [21][22][23]25,26 . Comparison of expression domains of INO orthologues suggested so far that the role of INO for outer integument growth was established early in angiosperms lineage and is widely conserved, despite some divergences in protein function 12,27 . To explore the relationship of the grape INO  Table S4). INO proteins from monocots and basal Angiosperms including Annona species clustered independently, in line with their highest phylogenetic distance. The grapevine INO protein clustered closer to the Arabidopsis protein and INO orthologues from other Rosidae, while INO from Asteridae, including tomato, grouped as a separate clade, independently of the number of integuments found in their ovules (Fig. 2).

Cloning of VviINO.1 and VviINO.2 cDNAs and expression characterization
To further characterize VviINO, its coding sequence was isolated from flowers of the grapevine cv. Italia harvested at 10% anthesis. Surprisingly, two cDNA were cloned (VviINO.1 and VviINO.2), 531 and 612 bp long, respectively. Sequence alignment to the grapevine reference genomic sequence showed that the shortest cDNA corresponded to the expected CDS according to the V2 prediction, including six exons and five introns with no polymorphisms and encoding a 176 aa protein. The second cDNA clone corresponded to an incompletely spliced mRNA entirely retaining the intron IV (81 bp). In silico translation of this produced a shorter protein (131 aa), due to the presence of an in-frame stop codon. Comparative 3D modeling showed that the 176 aa protein predicted from fully processed mRNA consisted of five small alpha-helices followed by longer unstructured loops, typically observed in regulatory proteins 32 . The removal of the last 45 residues, due to intron retention in the incompletely spliced form, determines the loss of two and a half helices (blue portion, Supplementary Fig. S3), which are predicted to be deeply involved in the stabilization of the interactions with DNA.
We tested the expression of each of the two mRNA in different plant organs by applying specific assays. Both VviINO.1 and VviINO.2 were specifically expressed in flowers and young fruits (Fig. 3a, b), as expected for INO gene and differently to other YABBY family members which were expressed also in other vegetative organs; the only exception was the VviCRC, that similarly as VviINO mRNAs also showed either low or no expression in vegetative tissues ( Supplementary Fig. S4). The highest expression was found at the onset of flowering, while expression decreased at full bloom and in berries at the pea-size stage. However, the incompletely spliced mRNA showed a lower expression compared to the fully processed form (Fig. 3a, c). The accumulation of both VviINO.1 and VviINO.2 mRNAs was also compared in flowers at pre-bloom and bloom stages in seeded and seedless cultivars. Samples were collected from three seedless cultivars Big Perlon, Fiammetta, and Crimson seedless, and two additional seeded cultivars Baresana and Vittoria (Fig. 3b, d). No relevant differences in expression levels associated with seedless and seeded cultivars were found, but the incompletely spliced mRNA always consistently showed a lower accumulation compared to the fully processed form. Recent experiments in Solanaceous species demonstrated that, despite the conservation in expression pattern and role, the SlINO gene was not able to complement the Arabidopsis ino-1 mutant 27 . Therefore, we wanted to  Supplementary Table S1. The phylogenetic tree was built using the Neighbor-Joining method and the p-distance in MEGA7. All positions with <85% site coverage were eliminated. The optimal tree is shown, and bootstrap support is shown next to the branches test whether VviINO, differently from SlINO, can complement the Arabidopsis mutant.
The Arabidopsis ino-1 mutant is vegetatively normal, but, as described, lacks the outer integument development on both sides of the ovule primordium and asymmetric growth during ovule development 13,18,33 . At anthesis, the nucellus is covered only by the inner integument, and bending of the chalaza does not progress and the micropyle lay in a line with the funiculus. This mutant was chosen for our cross-species complementation analysis. Both cloned VviINO.1 and VviINO.2 CDS were transferred to a segregating Arabidopsis population derived from the mutant, under the control of the AthINO promoter, previously described 13,20,34 . After selection of homozygous T3 plants in ino-1 background, vegetative and ovule phenotypes were observed. No differences in the vegetative growth were observed for any of the six and nine lines selected carrying VviINO.1 and VviINO.2, respectively (Supplementary Fig. S5), even though some variability was observed in the size of silique in mature plants (Supplementary Fig. S6). Careful observation of ovule phenotypes both by stereo and optical microscope showed that both VviINO.1 and VviINO.2 coding regions were able to restore outer integument growth. However, differently to complementation with AthINO transgene, which leads to the high frequency of full complementation 13,35 , transgenic plants complemented with grape INO CDS exhibited different ovule morphologies consistent inside lines. We have classified lines as "wild type-like" when normal ovule development was fully rescued or alternatively as "sup-like", "weakino-like", or "ino-like" when some atypical growth of outer integument or no growth at all was observed, similarly to 35,36 . To further support our scoring, some lines with representative morphologies were also analyzed by CRYO-SEM ( Supplementary Fig. S7).
Observation of transgene effects on ino-1 ovule morphologies are summarized in Table 1, and representative photos are shown in Fig. 4. Ovules of transgenic lines expressing VviINO.1 exhibited either wild-type growth ( Fig. 4e-g) of the outer integument or a "sup-like" phenotype ( Fig. 4c, d, h) with outer integument growing symmetrically on both sides of the ovule  1 (a, b) and VviINO.2 (c, d) transcripts was quantified using an MGB Taqman-specific assay. Expression was analyzed in different organs at different developmental stages (pre-bloom, bloom, berry pea-size, young leaves, medium leaves, old leaves, tendrils, and stems) in the cultivar Italia (a and b) or only in flowers at the pre-bloom and bloom stage in additional cultivars with seeds (Baresana, Vittoria) and in seedless cultivars (Fiammetta, Crimson seedless, and Big Perlon) (b, d). Means are calculated from a technical triplicate of three biological replicates. Error bars show standard errors primordium. This demonstrates that VviINO.1 can restore the compromised outer integument growth during ovule development and, to some extent, also an asymmetric growth. Surprisingly, all transgenic lines expressing VviINO.2 rescued the outer integument growth too. In addition to two lines fully recovered to normal ovule morphology (Fig. 4i, p), six lines showed a partial "suplike" phenotype. However, symmetric growth in these lines was much weaker compared to the sup mutant (Fig.  4j, k, m-o). Finally, one transgenic line showed a typical "weak-ino" phenotype (Fig. 4l).
Ino-1 mutant is strongly affected in female fertility and homozygous plants produce approximately one to three seeds 18 . Thus, the seed set allows additional evaluation of the complementation. In our hands, wild-type plants produced more than 50 seeds on average per silique. Transgenic lines presented a variable number of seeds per silique, but for all the seed set was significantly different compared to the ino-1 mutant. Only two lines carrying VviINO.1 showed a seed set comparable to that of wildtype plants. Interestingly, both lines were previously classified as "wild type-like" according to their ovule morphology. Seed set in all other lines was significantly different to the wild type, even though also significantly different to ino-1, with an average seed number per silique proportional to the rescue of the wild-type ovule morphology scored by microscope (Table 1).
Finally, we have investigated the molecular basis of the different levels of complementation. The expression of both transgenes was quantified by using a specific assay (Fig. 5).
As expected, no expression was found in the wild-type Arabidopsis, and lines transformed with VviINO.1 did not show any expression of VviINO.2. On the contrary, lines transformed with VviINO.2 produced both mRNAs, confirming that the VviINO.2 is likely an mRNA intermediate that can be successfully further processed in Arabidopsis to encode the complete functional protein, thus explaining the observed rescue in ovule morphology in these lines. Unfortunately, since we only conducted a relative quantification, the expression of VviINO.1 cannot be compared either to its endogenous expression level in grape nor to the expression of AthINO. Relative expression of both transgenes was variable in the different lines, and an inverted relationship between transgene expression and complementation level was observed, which was especially evident in lines carrying the VviINO.1 transgene.
Altogether these results demonstrate that VviINO can restore outer integument growth in A. thaliana ino-1 Ovule phenotypes were assessed by flower observation from independent plants with a dissecting stereomicroscope and optical microscope for each transgenic line. The average seed set was measured counting the average number of seeds per silique in four siliques and three independent plants. The seed set was also assessed in wild-type and ino-1 plants for comparison. (1) A statistical t test was conducted to assess the significance of observed differences in seeds mutant as well as a partial asymmetric growth, therefore its function is conserved across the two species.

Discussion
In this work, we have first of all identified the grapevine orthologue of the AthINO transcription factor, starting from the recent characterization of the grapevine YABBY gene family 28 . We confirmed that, according to all recent grapevine genome annotations, VviINO is encoded by a unique gene located on top of chromosome 1 comprising six exons and five introns. Differing from what expected from current annotations, cDNA cloning resulted in the . Asymmetric growth of the outer integument is always visible in these lines. Some symmetric growth is also visible leading to a "sup-like" phenotype (arrows in c, d, and h). i-p show representative phenotypes scored in VviINO.2 transgenic lines (#2, #23, #28, #29, #30, #31, #32, and #33, respectively). Asymmetric growth of the outer integument was visible in all lines. Reduced outer integument growth leading to the only partial covering of the inner integument as in the weak-ino mutant (allele ino-4) can be also observed (l, double arrow shows the uncovered inner integument). Symmetric growth of variable importance was often appreciated (arrows in j, k, m, n, o). Single arrow: outer integument growth from the adaxial side. Double arrow: inner integument uncovered by outer integument. Ovule phenotypes were consistent inside genotypes and representative pictures were chosen for each genotype. All pictures were taken with the same magnification 20x. Bar, in a, 100 µm identification of two mRNAs, the second one retaining the intron four. Alternative splicing due to intron retention could have functional implications, especially when translation results in a mis-functional protein due to a frameshift mutation 37,38 . However, we demonstrated that the VviINO.2 mRNA can be further spliced, even in a heterologous system, thus suggesting its presence is more likely related to an incomplete processing event or to an mRNA storing mechanism than working through translation in a mis-functional protein, as already reported in the literature for other plant genes 39 . We confirmed a specific expression in flowers and fruit (pea size) while no expression was found in other organs similarly as reported by other authors 28 , in agreement with the specific role of INO in ovule development.
In the past, detailed characterization of INO expression patterns in several species suggested wide conservation of INO function down to the early divergent Angiosperm clades [21][22][23][24] . Despite that, more recent functional studies provided controversial results. While VIGs silencing of the NbINO orthologue in tobacco inhibited the growth of the outer cell layer of the integument, leading to a decrease in both integument extension and ovule curvature, the tomato SlINO coding region was not able to complement the Arabidopsis ino-1 mutant 27 . With the final aim to deepen our knowledge of grapevine reproductive biology and especially ovule development to identify new targets for breeding purposes, we have enquired VviINO protein function conservation. By comparing protein sequences from 30 INO orthologues, we found that VviINO grouped in a clade with AthINO while both were more distantly related to SlINO. This supported the hypothesis that VviINO could complement the Arabidopsis ino-1 mutant phenotype, unlike SlINO. Results of cross-species complementation demonstrate that the VviINO can indeed successfully functionally complement the Arabidopsis ino-1 mutant phenotype when expressed from the AthINO promoter. We analyzed the ovules of 6 independent T3 transgenic lines expressing the VviINO.1 cloned CDS and 9 independent T3 transgenic lines expressing the VviINO.2 alternative CDS which, as previously mentioned, can be further spliced in Arabidopsis to encode a functional VviINO protein identical to that encoded by the VviINO.1 transcript. Careful microscopic observations showed that in all lines the outer integument growth and some ovule curvature was rescued. Seed set evaluation also confirmed a significantly different behavior in the transgenic lines compared to the background ino-1 line. However, differently from Arabidopsis lines complemented with the AthINO, we observed a high number of transgenic lines displaying a "sup-like" phenotype 35 , with also outer integument growth from the adaxial side of the ovule primordium associated with a still partially reduced seed set. This was reminiscent of behaviors previously observed in domain swap experiments. Replacement of the C-terminus of AthINO with AthCRC resulted in a significant proportion of transgenic lines that contained ovules with a "sup-like" phenotype, supporting the involvement of this region in the repressive action of SUP 36 . Our results strongly resembled these results, suggesting that, although VviINO can effectively complement the outer integument growthpromotive effects in the ino-1 mutant, it was less responsive than the endogenous AthINO to the SUP inhibition. Interestingly, alignment of VviINO and AthINO protein sequences highlighted low conservation of the C-terminal portion in the grapevine protein (Supplementary Fig. S2), which could explain the less effective SUP suppression and the consequent "sup-like" phenotype. Furthermore, we speculated that the consequent abaxalization of the adaxial domain would be likely more pronounced in lines with higher transgene expression and that this could likely explain the observed variability in phenotypes. Accordingly, we found a correlation between compromised ovule asymmetric growth and the expression levels of VviINO.1 and VviINO.2 transcripts, which further supports our data interpretation (Table 1 and Fig.  5).
These data improve our understanding of grapevine ovule development, with potential implications also for table grape breeding. The major source of seedlessness currently exploited in cultivated grapevine has been recently characterized, being due to an amino acid substitution in the VviAGL11 gene controlling seed coat VviINO.2 (white bars) transcripts in Arabidopsis flowers were quantified using the Taqman MGB-specific assay. Several pre-bloom flowers were collected for each transgenic line. Flowers of wild-type Arabidopsis plants were included as a control. Relative expression to actin was plotted in the graph. Transgenic lines showing fully complemented phenotype ("wild type-like") either only by microscope ovule observation or also by seed set evaluation are highlighted with dashed and solid arrows, respectively. Transgenic lines with no associated arrow were complemented but showed "sup-like" phenotype and lower seed set development and lignification, downstream of ovule development 11 . Accordingly, no difference in the VviINO.1 and VviINO.2 transcripts expression was found in seeded and seedless grape varieties (Fig. 3). However, alternative sources with potential implications for breeding purposes could also exist. The description of seedless normal size fruit production in Ts mutant of A. squamosa lacking the orthologous INO gene 12 , beside the demonstrated functional conservation in ovule development of VviINO (this work), supports the VviINO gene as a candidate for grapevine seedlessness. However, despite findings in Annona crop, in Arabidopsis ino-1, as previously mentioned, "fruit" development appears as compromised 18,19 . The different effect of ino defects in ovule development and on seeds and fruit production in Annona and Arabidopsis species has been the subject of further characterizations 12,40 . In these studies, the Arabidopsis ino-1 pollen tubes grow through the transmitting tract but were never observed inside the micropyle, and the majority of ovules fail to form embryo sacs. In contrast, in the A. squamosa Ts mutant, pollen tube growth was more normal often targeting the micropyle, and most of the ovules contained fully developed, but sometimes degenerating, embryo sacs. Authors have speculated that the outer integument role in pollen tube guidance and embryo sac development could have been a recent acquisition in Arabidopsis. They suggested that the absence of an essential role of the outer integument in pollen tube guidance in Annona could be related to its endostomal type of micropyle, with the outer integument not fully covering the inner integument and participating in the micropyle 12,41 . Furthermore, concerning embryo sac development, a higher sensitivity to changes in integument development could be due to a much thinner fraction of tissue around the female germline in the tenuinucellate Arabidopsis compared to the crassinucellate Annona. Interestingly, grapevine also shows an endostomal micropyle and several cell layers surrounding the embryo sac 12,42 . Only comparative studies in grapevine plants knocked-out for the now confirmed grapevine functional VviINO will conclusively allow to define implications of the defect in outer integument development for pollen tube guidance and embryo sac development and thus on seed and fruit production, eventually validating the utility of VviINO for table grape breeding. Interestingly, early genetic studies on grape seedlessness reported that seed coat hardness and endosperm/embryo development were behaving as separate sub-traits, confirming the existence of alternative contributions 43 . Moreover, more recently QTLs studies have revealed the contribution of a region located on top of chromosome 1 including VviINO gene to the total seeds fresh weight (TSFW) per berry trait 44 , further supporting VviINO as a candidate for grapevine seedlessness. An association to seedlessness in this genomic part of Chr1 was also confirmed by resequencing of seedless and seeded varieties, even though no associated SNP located in this gene were found in the studied panel 10 .
In conclusion, sequence comparison and the rescue of the outer integument growth in all Arabidopsis ino-1 lines expressing the VviINO protein from the AthINO promoter demonstrate that VviINO is the AthINO orthologue and that it plays the same function in promoting outer integument growth during ovule development. The high number of transgenic lines displaying a "sup-like" phenotype found in our cross-species complementation suggests a reduced sensitivity of VviINO compared to AthINO to the Arabidopsis SUP-mediated repression of expression in the adaxial side of the ovule primordia. Therefore, the mechanism involved in the tight control of INO spatial expression for proper ovule asymmetric growth could have partially diverged in the grapevine. Now that the functional involvement of VviINO in outer integument growth during grape ovule development has been demonstrated, functional studies in grape can further elucidate the mechanism for the asymmetric growth and the impacts on fruit and seed formation and their potential implications for table grape breeding purposes.

Materials and methods
In silico analysis BLASTP searches (e-value < 1×e −5 and identity >40%) against the PN40024 Vitis vinifera genome 12X V2 prediction, available on the CRIBI Biotech website (http:// genomes.cribi.unipd.it/) 29 , with protein sequences of the Arabidopsis thaliana YABBY family, retrieved from the TAIR database (The Arabidopsis Information Resource, http://www.arabidopsis.org/ gene ID: AT1G08465, AT1G23420, AT1G69180, AT2G26580, AT2G45190, AT4G00180), confirmed the previously identified grapevine YABBY genes as the most likely orthologous candidates 28 . Positional and structural information for each member of the family on the latest genome assembly 12X. v2 30 , for the V2 annotation or other available annotations, were downloaded from https://urgi.versailles.inra.fr/ Species/Vitis/Annotations. Sequence alignments were performed using the MSA tool MUSCLE in MEGA7 software or at http://www.ebi.ac.uk/Tools/msa/ muscle/ using default settings. Grapevine protein sequences corresponding to the V2 prediction were downloaded from the CRIBI website. Only the longest peptide sequence of each gene was used. ngLOC tool (http://genome.unmc.edu/ngLOC/index.html) was applied to predict the subcellular localization of the YABBY proteins. All proteins were submitted to Pfam (http://pfam.xfam.org/search/sequence) to verify the presence of the YABBY domain.

Orthologous relationship from genetic distances and phylogenetic analysis
To establish orthologous relationships for grapevine YABBYs with Arabidopsis, genetic distances and trees were estimated by using the MEGA7. All ambiguous positions were removed for each sequence pair, and the number of base pairs or amino acid differences per site between CDS or protein sequences was used for estimating distance matrices. Orthologous relationships and nomenclature are based on the comparison of the longest CDS with the six Arabidopsis YABBY CDS sequences according to rules established by the Grapevine Super Nomenclature Committee 31 . An unrooted UPGMA tree (p-distance method) was constructed from CDS distances. Branch tree support values were obtained from 1000 bootstrap replicates and branches with values below 70% were condensed.
The evolutionary history of INO proteins identified in different species was inferred using the Neighbor-Joining method in MEGA7. The evolutionary distances were computed using the p-distance method and are in the units of the number of amino acid differences per site. The analysis involved 30 amino acid sequences (Supplementary Table S4). All positions with less than 85% site coverage were eliminated. The optimal tree is shown and the percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) are shown next to the branches.

Comparative 3D modeling
The 3D comparative model of the wild-type VviINO.1 protein (protein length 176 aa) was prepared by performing a multi-template modeling session by using Modeler 45 . 3D crystallized structures used for the multitemplate modeling, predicted by using fold recognition methods analysis were 2lef.pdb and 3cmv.pdb showing 30% of identical amino acids with VviINO.1 sequence 46 . The 3D comparative model of the shorter VviINO.2 isoform (protein length 131 aa) was built by removing the last 45 residues by using PyMOL. For the modeling of the protein-DNA complex, the obtained VviINO.1 3D comparative model was superimposed to 2lef.pdb for docking the DNA molecule (duplicated from 2lef.pdb) within the VviINO.1/2 3D comparative models.

Plant materials
For expression analysis, samples from stems, tendrils, flowers, berries, and leaves were collected from the V. vinifera cv. Italia from an experimental vineyard located in Valenzano, Bari (Italy). In details, flowers were collected at the pre-bloom stage corresponding to 10% caps off (E-L 19) (Coombe, 1995) and at the bloom stage of 50% caps off (E-L 23); berries were sampled at the pea-size stage of 7-mm diameter (E-L 31); small leaves were collected when the shoot bear was approximatively five separated leaves (E-L 12), medium leaves corresponded to 16 separated leaves (E-L 19), and large leaves represented leaves before senescence (E-L 31). For the expression profiles in seedless and seeded cultivars flowers from the same developmental stages of pre-bloom and bloom as previously described were collected also from the variety Big Perlon, located in the same experimental field as Italia, and the varieties Baresana, Vittoria, Fiammetta, and Crimson seedless from an experimental field located in Adelfia, Bari (Italy). Three biological replicates were independently collected for each tissue/stage. For functional analysis, the previously described mutant ino-1 (CS3881) of A. thaliana (Landsberg erecta) 13 and the corresponding wild type were used. Since the mutant produces few seeds it is maintained in heterozygous status. All plants were grown in a growth chamber under controlled conditions (8-h light/16-h dark photoperiod, 24°C/21.5°C, 70% relative humidity) and watered weekly. The light was provided by warm white fluorescent tubes, 120 to 160 µmolphotons m −2 s −2 .
For expression studies in Arabidopsis, several prebloom flowers (up to 100 mg) were collected from different plants belonging to the same homozygous T3 line.

Expression studies
The total RNA was extracted from grapevine tissues (400 mg for berries and 100 mg for other tissues) or Arabidopsis flowers (100 mg) using Spectrum TM Plant Total RNA kit (Sigma-Aldrich, St. Louis, MO) following the manufacturer's instructions. To remove genomic DNA, an optional on-column step with RNase-Free DNase I Set (Qiagen, Hilden, Germany) was included immediately after the binding step. After extraction, RNA was further purified in 3 M LiCl (Sigma-Aldrich), precipitated at 4°C overnight, and centrifuged at 15,000 × g for 20 min at 4°C. The pellet was rinsed with 70% cold ethanol, centrifuged at 13,000 × g for 5 min, and then eluted in 50 µl of nuclease-free water.
The first-strand cDNA for quantitative real-time PCR was synthesized with Superscript ® III First-Strand Synthesis System (Invitrogen, Carlsbad, USA) starting from 1 µg of RNA and primed using the Oligo (dT) 20 following the manufacturer's instructions. cDNAs were diluted ten times in pure water. Quantitative RT-PCR of YABBY genes was conducted as described by Symons et al. 47 in triplicate for each sample using 3 µl cDNA in 1× SYBR Select Master Mix (Applied Biosystems, Foster City, USA) and with 0.4 µM of forward and reverse primer in a total volume of 20 µl. Gene-specific primer pairs were designed with OligoExplorer 1.1.2 avoiding regions of cross-homology for each gene and, for normalization of cDNA levels, for Actin2 from grape (Supplementary Table  S5). CFX96™ Real-Time PCR Detection System (Bio-rad, Hercules, USA) was used, and the data were analyzed with CFX Manager™ software. Copy number for each YABBY gene was assessed according to Bottcher et al. 48 , and reaction specificity was confirmed by melt curve analysis, by agarose gel, and by sequencing (Macrogen, Meibergrdeef, The Netherlands).
To measure the expression levels of the two mRNA VviINO.1 and VviINO.2, we developed TaqMan™ Gene Expression Assays specific for each of the forms, which makes use of a pair of unlabeled PCR primers and the TaqMan probes with a FAM™ dye label on the 5'-end and a minor groove binder (MGB) and a non-fluorescent quencher (NFQ) on the 3'-end (Supplementary Table S5). The cDNA levels were normalized with TaqMan™ Gene Expression Assays for VviActin2 or AthActin2 in grapevine and Arabidopsis, respectively (Supplementary Table  S5). Real-time PCR conditions were as suggested by the assay and Real-Time PCR Detection System (BioRad, USA). In grape, copy number was estimated in tissues/ organs or varieties as previously indicated. Relative quantification to actin was calculated instead for expression analysis in Arabidopsis flowers from transgenic lines.

DNA constructs and plants transformation
The grapevine VviINO.1 (531 bp) and VviINO.2 (612 bp) cDNAs were isolated from V. vinifera cv. Italia flowers at the pre-bloom stage by RT-PCR using primers VviINO-1 forward and VviINO-4 reverse (Supplementary Table S5) and cloned into pJET 1.2/blunt cloning vector system (Thermo Fisher Scientific). Several clones were randomly chosen and sequenced in order to verify the presence of the alternative forms. Both cDNA coding regions were modified by PCR using primers BamVviI-NOF containing a BamHI site and XbaVviINOR (Supplementary Table S5) containing an XbaI site. BamHI/ XbaI fragments were used to replace the Arabidopsis INO cDNA into the previously described pRJM33 chimera vector carrying the Arabidopsis INO cDNA flanked by the corresponding 5'-(2.3 kb) and 3'-(2 kb) genomic regions 13 . These regions were previously shown to be sufficient to enable complementation of the ino-1 mutant phenotype 20,35 . The new Arabidopsis genomic::grape cDNA chimera was inserted as NotI fragment into the pMLBART plant transformation vector 49 to obtain pIM4 and pIM5 plasmids and transferred into the Agrobacterium tumefaciens GV3101 strain.
To evaluate the competence of VviINO.1 and VviINO.2 to complement the ino-1 phenotype in A. thaliana pIM4 and pIM5 constructs were transformed into an Arabidopsis segregating population for the ino-1 mutant previously indicated by the Agrobacterium-mediated floral dip method. Transformants were selected by germinating seeds on Murashige and Skoog (MS) medium, 3% (w/v) sucrose, 0.8% (w/v) agar, and 10 μg ml -1 phosphinothricin (BASTA). The presence of the transgene was confirmed by PCR using the primers 4CKfor and 4CKrev (Supplementary Table S5) that amplify different sized fragments from VviINO.1 and VviINO.2 transgenes. The ino-1 homozygous background was selected genotyping by PCR amplification and sequencing the A. thaliana INO gene using primers ino-1-genfor and ino-1-genrev (Supplementary Table S5). T3 homozygous lines for each of the two transgenes were then selected.

Optical, stereo, and CRYO-SEM microscopy
Complemented Arabidopsis plants were grown under long-day conditions until flower bud formation and flowers sampled. For microscope observation, samples were prepared immediately before use from flowers fixed in FAA solution (3.7% formaldehyde, 5% acetic acid, 50% ethanol) or from isolated ovaries dissected and stored in 70% ethanol at room temperature. Ovule phenotypes evaluation was performed using bright-field optical microscopy or in dark-field under a stereomicroscope for all transgenic lines in ino-1 background as well as ino-1 and wild-type plants. At least four independent flowers were observed for each line by dissecting stereomicroscope, and pictures were taken by bright-field optical microscopy. Representative lines of each phenotypic class were fixed and prepared for CRYO-Scanning Electron Microscopy (CRYO-SEM). Before observation, the ovaries were rinsed in water, opened with a fine needle and tweezers in order to take the ovules that were immediately observed under CRYO-SEM.

Seed set evaluation
The average number of seeds per silique was estimated by counting seeds from four siliques per plant and three plants per each line and a t test was applied to evaluate significant differences.