PR10 proteins belong to the huge family of pathogenesis related (PR) proteins ubiquitous in the plant kingdom. PR proteins were first identified as defence molecules produced in response to pathogen attack and some of them actually display an antimicrobial activity. However, numerous studies have reported their induction under a great variety of abiotic stress conditions as well as possible constitutive or developmentally regulated expression [1]. Sharing common biochemical characteristics (acidic pI, resistance to proteolytic degradation, small molecular mass) PR proteins are divided into seventeen different groups based on their primary structure, serological relationships and biological activity [2]. Most of them are extracellular, but some others are found in the cytoplasm, mainly in the vacuole. PR10 proteins present the specificity to be free in the cytoplasm and are therefore classified as intracellular PR (IPR) proteins. They are closely related to a group of major tree pollen allergens and food allergens, that belong to the Bet v 1-like superfamily [3].

PR10 genes form multigene families with low intraspecific variation and higher interspecific variation that make them interesting phylogenetic markers [4–6]. Some of them were shown to be organized in chromosome clusters [7, 8]. Characterised in a number of plant species, most PR10 genes share an open reading frame (ORF) from 456 to 489 bp interrupted by an intron of 76-359 bp at a highly conserved position [9]. This ORF codes for an acidic small protein with conserved sequence features: three amino acids E96, E148 and Y150 (as positioned in Bet v 1) possibly involved in ribonucleasic activity and two other remarkable domains comprising the motif GXGGXGXXK (aa 47-55) that forms a P-loop supposed to have NTPase activity and the Bet v 1 motif [PS00451] characteristic of proteins from the Bet v 1 superfamily [9].

The significance of multiple close copies of a gene in a single plant species has to be clarified. In birch or yellow lupine, some PR10 genes are constitutively expressed while others are induced under pathogen attack, abiotic stress or during plant development, suggesting functional diversification [10, 11]. A significant common feature of PR10 proteins is a large Y-shaped hydrophobic cavity, as shown by the determination of three-dimensional structure [3, 12–15]. This internal cavity could be responsible for the intracellular transport of apolar ligands, so diverse as fatty acids, flavonoids, cytokinins or brassinosteroids [10, 14, 16, 17]. Slight modifications of the structure and shape of this cavity would allow to bind different ligands, what could account for the diverse roles hypothesised for PR10 proteins in plant defence and development [12, 13].

To date, three different PR10 genes have been described in Vitis vinifera. VvPR10.1 was shown to be up-regulated during a pathogen interaction with Pseudomonas syringae in the cultivar Ugni blanc [18], while the expression of VvPR10.1, VvPR10.2 and VvPR10.3 was detected during somatic embryogenesis (SE) induction in the cultivar Chardonnay [19]. Moreover, several studies showed a strong and specific production of PR10 proteins in V. vinifera under salt or herbicide stress [20, 21], as well as after fungal attack [22, 23].

The recent publication of the whole genome sequence of V. vinifera by the Genoscope [24] and the Instituto Agrario San Michele all'Adige (IASMA) [25] makes genome scale analyses possible. Using these databases, we characterised all sequences of the PR10 gene family of V. vinifera and monitored the expression of nine candidate genes in various tissues and conditions.

We found a total of seventeen PR10 related sequences in the whole V. vinifera genome. Thirteen unique sequences were retained from an automatic search that initially produced ninety regions, reflecting redundancy of the database as well as annotation errors partly due to wrong homology detection. A manual search allowed the recovery of four additional PR10 related sequences. All seventeen sequences were found in a single compact cluster on the chromosome 5. Plant PR10 belong to multigene families. There are at least five PR10 genes in pea [27], eighteen Mal d 1 genes in apple [7], ten Bet v 1 genes in birch [6], eight Fra a 1 genes in strawberry [28], six PR10 genes in Solanum surattense [29], eight in yellow lupine [11], five in rice [30], and eight Pru p 1 and Pru d 1 genes in peach and almond, respectively [8]. They were shown to form physical clusters in apple [7] and peach [8]. Poplar PR10 genes are also supposed to be grouped on chromosomes [26]. Tandem duplicates are frequent in plant genomes and represent up to 16% of Arabidopsis genes [31]. Such gene clusters are thought to be produced by successive single gene or small-scale duplications. We found that thirteen out of the seventeen V. vinifera PR10 sequences are present on the chromosome in direct orientation, suggesting that most copies were produced by unequal crossing over events, as described in Arabidopsis and rice [31].

Following duplication, new copies of a gene may undergo modifications allowing functional diversification, which is a significant source of evolutionary novelty in plants [32]. However, gene duplication mostly creates copies that are rapidly lost through pseudogenisation. As a result, from numerous homologous sequences coexisting in a genome, only a part are functional genes. From the seventeen V. vinifera PR10 sequences, only fourteen have an intact ORF. In birch, the copy number of PR10 genes varies from twelve to twenty-five, depending on the species, and pseudogenes represents as much as one-third of Betula nigra alleles [6]. In V. vinifera, the pseudogenes s7 and s9 share the highest nucleotide similarity with VvPR10.6 (s4), suggesting that they could derive from its duplication. Likewise, s6 is closer to VvPR10.5 (s5) than to all other sequences and could therefore originate from its duplication. Moreover, s6 and VvPR10.5 are also very similar at the intron level (81% of nucleotide similarity), indicative of possible recent duplication or slow pseudogenisation.

Apart from pseudogenes, sequence divergence is generally reduced in PR10 multigene families. PR10 genes within a species are generally very similar and more distant from gene copies of other plant species [4]. Such a low intraspecific variability has been observed in Passiflora [5] and in the Betula genus [6]. The different paralogs are thought to undergo strong concerted evolution which is the tendency of a family of repeated genes to jointly evolve. The close physical proximity of tandem duplicates facilitates gene conversion or unequal crossing-over events leading to concerted evolution [6]. The phylogenetic tree in Figure 7 illustrates probable concerted evolution in the V. vinifera PR10 family. However, V. vinifera PR10 proteins divide into two clades interrupted by PR10 of other plants, indicating partial independent evolution. Interestingly, the seven sequences s10 to s16 corresponding to the very homologous proteins PR10.1-A to PR10.3-C follow one another on the chromosome, what could favor concerted evolution and suggest strong local selection pressure. However, it cannot be ruled out that recent duplication events produced these seven very homologous sequences.

Grapes, as other eudicot species, probably originate from an ancient hexaploid ancestor formed through whole genome duplication (WGD) events [24, 33]. Recurrent polyploidisation is tightly linked to evolution in angiosperms, providing raw materials for gene diversification and genome refinement, and coinciding with species radiation [32]. A WGD is followed by incomplete and asymmetric loss of gene copies and chromosome rearrangements allowing the recovery of a diploid-like state compatible with effective reproduction. In Arabidopsis, it was possible to track gene pairs released by a recent specific WGD event [34]. Only 28.6% of gene pairs were retained from the transient tetraploid genome, the other pairs having lost a copy from one of the homeologous chromosomes. In V. vinifera, PR10 sequences were solely located on the chromosome 5, suggesting loss of some of the triplicate ancestral copies and/or translocation on a unique chromosome. The seventeen present PR10 sequences are most probably due to repetitive small duplications having continuously occurred during V. vinifera evolution. Subsequent mutations and possible positive selection are responsible for the observed divergence within the sequences. However, at least a part of V. vinifera PR10 sequences are probably subjected to concerted evolution, reducing variability and hampering the identification of putative triplicate ancestral copies.

We were able to detect the expression of nine out of the fourteen complete PR10 related sequences in varied non-treated and treated tissues of V. vinifera cv. Chardonnay. Three genes, i.e. VvPR10.1-a, VvPR10.2 and VvPR10.3-a, were already shown to be expressed under pathogen attack or abiotic stress, and during somatic embryogenesis induction [18, 19], whereas expression of VvPR10.4-VvPR10.9 was never studied before. VvPR10.10 transcripts were not detectable in our conditions. This sequence has no corresponding EST in the databases and lacks canonical transcription signals. However, expression could be limited to very specific tissues and/or developmental stages. Although specific ESTs in the databases suggest a probable expression of the four sequences VvPR10.1-b and -c and VvPR10.3-b and -c, the distinction of each transcription product would require more sensitive methods than PCR, such as cDNA-AFLP or SSCP, which were used for discriminate highly homologous sequences in potato, polyploid cotton and barley [35–37]. The expression of VvPR10.5, VvPR10.7 and VvPR10.8 is not surprising because well matching ESTs were found in the databases. On the contrary, no EST was found to correspond to VvPR10.4, VvPR10.6 and VvPR10.9, whose expression is therefore reported for the first time. Interestingly, VvPR10.7, VvPR10.8 and VvPR10.9 are expressed although devoid of canonical transcription signals.

We found transcripts of all nine PR10 genes in varying amounts, in the different non-treated tissues analysed, suggesting a role apart from defence. Specific expression profiles were found in intact tissues, except for VvPR10.2 and VvPR10.3-a. Stems and intact embryos expressed a reduced subset of PR10 proteins. On the contrary, immature flowers expressed a large subset of PR10 genes, suggesting a possible role of some PR10 proteins during sexual reproduction. No V. vinifera PR10 gene was solely expressed in calli derived from 2,4-D-treated embryos. However, expression of several genes was enhanced in these tissues, at least weakly. Somatic embryogenesis is generally obtained from tissues subjected to the auxin 2,4-D, which acts as a stress factor able to trigger the reprogramming of plant somatic cells towards embryogenesis [38]. As a consequence, high amounts of defence proteins are produced following a 2,4-D treatment, as shown in grapevine cultures [39]. Consistent with the results reported here, we previously showed that varied PR genes including VvPR10.1-a and VvPR10.3-a are up-regulated during secondary somatic embryogenesis induction in V. vinifera [19]. VvPR10.1 expression was also previously reported to be induced in whole plant leaves challenged with the incompatible bacterium Pseudomonas syringae [18]. Interestingly, VvPR10.4 and VvPR10.7 were not responsive to 2,4-D treatment.

On the whole, various expression patterns were found for V. vinifera PR10 genes, indicating functional diversification and possible tissue specificity. Likewise, in other plants, PR10 proteins show expression diversity suggesting various biological activities [9]. A major feature of PR10 proteins is their internal hydrophobic cavity with openings at the protein surface. Several three-dimensional modeling studies showed that this structure is suitable for the binding and transport of diverse hydrophobic ligands as brassinolides [14] or homocastasterone [12]. Some birch and yellow lupine PR10 can bind diverse molecules such as cytokinins, brassinosteroids, fatty acids and flavonoids, suggesting that they could interfere with the trafficking of hormones inside the cell [10, 16, 17]. Moreover, overexpression of a PR10 gene in pea led to a change in the ratio between cytokinins and abscisic acid, showing that the PR10 content could be relevant for intracellular hormone regulation [40]. Although all crystallographic models for PR10 proteins share the same canonical fold, their superimposition can reveal structural differences [12]. Remarkably, the volume of the internal cavity and of its openings can show some variability, what could influence the type of transported ligand. In yellow lupine, different shapes for the inducible LlPR10.1A and the constitutive LlPR10.1B could account for their different biological roles [13].

The availability of the complete genome sequence of V. vinifera allowed us to characterise the PR10 gene family. Seventeen different PR10 related sequences including three pseudogenes were identified and located in a single compact cluster on the chromosome 5, most probably reflecting repetitive small duplications during evolution. A phylogenetic analysis showed a characteristic low variability within the different sequences, especially within seven sequences closely located on the chromosome, suggesting probable concerted evolution. We could analyse the expression of nine genes in various tissues. Different expression patterns indicate functional specialisation. Several genes showed a typical stress induced up-regulation. Further experiments will help to elucidate the differential regulation of V. vinifera PR10 gene expression.