Genome-Wide Analysis of the Expansin Gene Superfamily Reveals Grapevine-Specific Structural and Functional Characteristics

Background Expansins are proteins that loosen plant cell walls in a pH-dependent manner, probably by increasing the relative movement among polymers thus causing irreversible expansion. The expansin superfamily (EXP) comprises four distinct families: expansin A (EXPA), expansin B (EXPB), expansin-like A (EXLA) and expansin-like B (EXLB). There is experimental evidence that EXPA and EXPB proteins are required for cell expansion and developmental processes involving cell wall modification, whereas the exact functions of EXLA and EXLB remain unclear. The complete grapevine (Vitis vinifera) genome sequence has allowed the characterization of many gene families, but an exhaustive genome-wide analysis of expansin gene expression has not been attempted thus far. Methodology/Principal Findings We identified 29 EXP superfamily genes in the grapevine genome, representing all four EXP families. Members of the same EXP family shared the same exon–intron structure, and phylogenetic analysis confirmed a closer relationship between EXP genes from woody species, i.e. grapevine and poplar (Populus trichocarpa), compared to those from Arabidopsis thaliana and rice (Oryza sativa). We also identified grapevine-specific duplication events involving the EXLB family. Global gene expression analysis confirmed a strong correlation among EXP genes expressed in mature and green/vegetative samples, respectively, as reported for other gene families in the recently-published grapevine gene expression atlas. We also observed the specific co-expression of EXLB genes in woody organs, and the involvement of certain grapevine EXP genes in berry development and post-harvest withering. Conclusion Our comprehensive analysis of the grapevine EXP superfamily confirmed and extended current knowledge about the structural and functional characteristics of this gene family, and also identified properties that are currently unique to grapevine expansin genes. Our data provide a model for the functional characterization of grapevine gene families by combining phylogenetic analysis with global gene expression profiling.


Introduction
Expansins are cell wall-loosening proteins that regulate cell wall expansion and cell enlargement in a pH-dependent manner [1]. The first expansins were found in cucumber (Cucumis sativus) [2] but many others have subsequently been identified throughout the plant kingdom [3] and related proteins exist in bacteria, nematodes and mollusks, probably helping to digest plant biomass [4][5][6][7]. Genome-wide analysis has revealed a large expansin superfamily (EXP) in plants comprising four families [8] defined as expansin A (EXPA), expansin B (EXPB), expansin-like A (EXLA) and expansin-like B (EXLB) according to a standardized nomenclature [9]. Only the largest families (EXPA and EXPB, the latter originally identified as group I pollen allergens) have been shown empirically to loosen cell walls, whereas the expansin-like proteins (EXLA and EXLB) are recognized only by their conserved sequences and their precise functions remain unknown [10,11].
Expansins are typically 250-275 amino acids in length and comprise a 20-residue N-terminal signal peptide that targets the protein to the cell wall and two major domains, one homologous to glycoside hydrolase family 45 although its activity has yet to be confirmed (domain 1) and another thought to bind cell-wall polysaccharides (domain 2) [8]. The mechanism of action is still largely uncharacterized but may involve the acid-induced disruption of hydrogen bonds linking cellulose and hemicellulose fibers, promoting slippage between the polymers and allowing the cell to absorb water, ultimately leading to cell wall expansion [1,12]. The shortage of functional information reflects the scarcity of informative structural data, given that the only solved crystal structures represent one bacterial expansin [4] and the pollenspecific EXLB in maize [13]. Although no structural data exist for plant EXPA and EXPB proteins, the developmental and physiological functions of expansins have been studied in many plant species, organs and developmental stages, showing that these proteins are involved in processes such as root and shoot apical meristem development [14,15], the definition of plant architecture [16], xylem formation [17], abscission [18,19], seed germination [20], several types of plant-microbe interactions [21][22][23] and fruit softening [24][25][26][27].
Grapevine (Vitis spp., family Vitaceae) is an important perennial woody species because of its economically-valuable role in fruit and wine production, thus it was the first fruit crop selected for genome sequencing [28,29]. The relatively small grapevine genome (475 Mb) and the absence of recent whole genome duplication (WGD) events means it is also ideal for the genomewide characterization of gene families. Recent studies have identified many gene families and superfamilies in grapevine, including stilbene synthases [30], dehydrins [31], pathogen-related proteins (PR-10) [32], aldehyde dehydrogenases [33] and transcription factors such as TIFY [34] and NAC [35]. There has been no systematic analysis of grapevine expansins thus far, although the expression profiles of several expansin genes have been studied during berry development [36,37] and one expansin was identified as minor grape allergen [38]. Recently, certain expansin transcripts have been identified by microarray analysis as markers of berry development [39][40][41].
We screened the V1 annotation of the 12X grapevine genome (PN40024 genotype; http://srs.ebi.ac.uk/) and identified 29 expansin genes representing all four families. This is the first genome-wide study of the grapevine expansin superfamily and we propose a systematic nomenclature based on chromosome position and existing guidelines [9]. We carried out a global analysis of differential expansin gene expression by mining publicly-available microarray data from the recently-published grapevine transcriptome atlas [42] and combining this with data from quantitative real-time RT-PCR experiments. Finally, we focused on the role of expansins in post-harvest withering, a process used in the manufacture of straw wines. Our data provide a basis for the further evolutionary and functional characterization of expansin genes and our exhaustive genome-wide expression analysis has helped to determine the precise role of expansins in grapevine, and thus their potential functions in other fruit crops.

Identification of expansin genes in grapevine by genome-wide analysis
The genome of the near-homozygous PN40024 genotype of V. vinifera cv Pinot Noir was screened for EXP gene sequences. Previously-identified expansin sequences from Arabidopsis thaliana, rice (Oryza sativa) and poplar (Populus trichocarpa) were used as BLASTP queries against the 12X V1 assembly using the JIGSAW software platform [43] at the CRIBI in Padova, Italy (http:// genomes.cribi.unipd.it/grape/). This led to the identification of 29 expansin or expansin-like sequences, representing all four expansin families [8]. We identified 24 genes containing the HFD motif which is found only in the EXPA and EXPB families, including 20 in which domain 1 included the characteristic large insertion (ainsertion) and deletion (a-deletion) that is unique to the EXPA family, suggesting that the grapevine genome contains 20 EXPA genes and four EXPB genes ( Figure S1). Among the remaining sequences, one was classified as an EXLA gene, based on the absence of the HDF domain and the presence of a characteristic EXLA extension at the C-terminus, and the other four were classified as EXLB genes ( Figure S1). All 29 genes were named based on family identity and, within each subgroup, their chromosomal position. The proposed nomenclature for grapevine expansin genes, together with identifiers in the 12X V1 annotation and the length of each open reading frame (ORF), are listed in Table 1. We compared the number of grapevine genes in each expansin family to the other available angiosperm genomes, i.e. Arabidopsis, rice and poplar ( Table 2). The EXPA family is the largest, with 20 members in grapevine (VvEXPA1-VvEXPA20), 26 in Arabidopsis [6], 34 in rice and 27 in poplar [44,45]. The EXPB family is similar in grapevine, Arabidopsis and poplar, but has undergone a grass-specific expansion in rice [8]. EXLA was the smallest family with only one member in each species, whereas the EXLB family appeared to have expanded specifically in woody species (grapevine and poplar) but not in Arabidopsis and rice.
Experimental evidence has shown that EXPA and EXPB proteins can loosen cell walls [11,46] whereas EXLA and EXLB have been identified by homology and their activity has not been demonstrated. The similar phylogenetic profile of EXLA and EXLB genes in grapevine and poplar may suggest they perform similar roles in these two perennial woody species.
The chromosomal distribution of grapevine expansin genes is presented in Figure 1, which shows that paralogs are present on all chromosomes except 5, 10, 11 and 16. The last released grapevine genome assembly did not allow us to assign the positions of three expansin gene models (VvEXLB2-VvEXLB4) so these could potentially be located on the chromosomes listed above. The distribution of expansin genes is similar in grapevine and poplar, with no more than three expansin genes present on a single chromosome ( Figure S2). In contrast, expansin genes tend to be clustered in Arabidopsis and rice, including the presence of 18 expansin genes on rice chromosome 3, reflecting the large number of tandem duplications present in the rice genome ( Figure S2).
The expansins do not appear to have undergone such extensive phylogenetic expansion and diversification as other superfamilies, e.g. stilbene synthases, terpene synthases and other genes related to wine characteristics [28]. As proposed for Arabidopsis, the grapevine expansin superfamily may result from massive gene loss following genome duplication.

Phylogenetic and structural analysis of the grapevine expansin genes
We constructed a phylogenetic tree including the 29 grapevine expansin proteins based on their deduced amino acid sequences, clustering those in the same families ( Figure 2A). To gain more insight into the structural evolution of the expansin gene superfamily, the previously-defined exon-intron boundaries [8] were investigated, revealing that genes from the same family were generally characterized by similar exon-intron structures ( Figure 2B). The grapevine EXPA genes generally contained two introns at positions A and B, except VvEXPA9, VvEXPA15 and VvEXPA10 (which lacked the B intron), VvEXP4 (which featured an unusually long B intron), and VvEXPA6, which contained a third intron at position H, downstream from the BOX 2 site ( Figure 2B). The grapevine EXPB genes generally contained three introns at positions A, C and B, except VvEXPB2 which featured an intron at position F instead of C. Finally, it is noteworthy that the most phylogenetically-divergent EXLB gene (VvEXLB1) displayed an exon-intron structure dissimilar to the other EXLB genes, but identical to that of the only grapevine EXLA gene (VvEXLA1) with four introns at positions A, C, B and F. The highly-conserved sequences and similar exon-intron structures within each family suggest that the families may have undergone independent gene duplications to generate paralogs with partially or completely overlapping functions.
The four expansin gene families in angiosperms were previously subdivided into 17 clades based on a reconstruction of the last common ancestor of monocots and eudicots, which was predicted to contain 10-12 EXPA genes, two EXPB genes, one EXLA gene and two EXLB genes, making 15-17 in total [8]. Each clade would therefore include all of the genes descending from a single gene present in the last common ancestor, constituting a monocoteudicot orthologous group. We therefore constructed a phylogenetic tree containing all the grapevine expansins and additional Arabidopsis, rice and poplar expansins that represent all the proposed monocot-eudicot clades [8] ( Figure S3). We found that all 17 clades were present in the grapevine genome, including EXLB-II and EXPA-IX, which were lost in Arabidopsis but retained in rice and poplar. Based on the minimal ancestral exonintron structures for each clade [8], we found that most of the grapevine expansins had maintained the ancestral pattern ( Figure  S3). The exceptions were VvEXPA12 (clade EXPA-XI), which featured two introns (A and B) instead of a single intron at position B as reported for poplar [45], suggesting the ancestral exon-intron structure of clade EXPA-XI may need to be revised. The EXPA-IX clade includes VvEXPA4, which has lost intron A, potentially representing a grapevine-specific intron loss as also reported for some members of clade EXPA-VIII in poplar [45]. Furthermore, an extra intron was found in VvEXPA6 (clade EXPA-IV) whereas VvEXPA10 (also clade EXPA-IV) had lost the intron at position B ( Figure S3), suggesting either a unique exon-intron structure for certain grapevine expansins or the presence of an independent monocot-eudicot clade which has been lost in Arabidopsis, rice and poplar. We identified five pairs of grapevine expansin genes apparently reflecting a recent lineage-specific gene duplication event, given their short branches (VvEXPA1:VvEXPA18, VvEXPA13:VvEXPA6, VvEXPA9:VvEXPA15, VVEXB4:VvEXPB3 and VvEXLB2:V-vEXLB3). The duplication of EXLB genes from clade EXLB-I (VvEXLB2, VvEXLB3 and VvEXLB4) is particularly remarkable because this appears to be a grapevine-specific event. The majority of grapevine expansin genes appear more closely related to poplar expansins than those from Arabidopsis and rice, which is consistent with the more recent divergence of the grapevine and poplar genomes from a common ancestor than from the lineages leading to monocots and Arabidopsis [28]. Tandem and segmental duplication across different plant lineages has led to the species-dependent expansion of expansin gene families in rice, Arabidopsis and poplar [45]. Grapevine has not undergone a recent whole genome duplication event, but it is unclear whether the haploid grapevine genome (which originated from the contribution of three ancestral genomes) is the product of a true hexaploidization event or successive genome duplications [28]. Recently, it was proposed that the extant triplicate state of the grapevine genome could have arisen from the fusion of two chromosome groups approximate 60 million years ago or by allopolyploidy following eudicot radiation [47]. Therefore, it will be necessary to investigate microsyntenic relationships in order to better clarify the evolution of the grapevine expansins.

Global analysis of grapevine expansin gene expression
The recently-published V. vinifera cv Corvina gene expression atlas [42] allowed us to characterize the expression of the expansin superfamily in different grapevine organs and developmental stages. We analyzed the fluorescence intensity values of the 29 expansin transcripts, generating a biclustered heat map in which the data were normalized based on the mean center genes/rows adjustment ( Figure 3A). The different expansin families shared a common expression profile, with the EXPA, EXPB and EXLA transcripts detected in a wide range of tissues, organs and developmental stages without any particular correlation with the phylogenetic tree. The exception was the EXLB family, whose expression was restricted to organs such as the winter dormant bud (E-L 1), swelling bud (E-L 2), woody stem, root and mature stages of the rachis (V, MR and R).
Pearson's distance correlation analysis was used in a previous study to show that 54 diverse grapevine samples clustered The Expansin Superfamily in Grapevine PLOS ONE | www.plosone.org predominantly in relation to temporal dynamics, distinguishing particularly between mature/woody and vegetative/green samples, whereas the organ identity appeared to be less important [42]. We therefore used the same method to cluster the 29 sets of fluorescence data representing the grapevine expansin genes, and the resulting dendrogram was strikingly similar to the one produced in the earlier study ( Figure 3B). Temporal-dynamic clustering was preserved in the expansin-specific correlation analysis, highlighting the distinction between mature and vegetative/green samples. These observations indicate that different expansins are predominantly expressed in one or the other grapevine temporal-dynamic cluster and that different organs are characterized by the expression of more than one expansin during its development. Indeed, when grapevine expansin gene expression was analyzed in individual organs, it was still possible to detect the separation between vegetative/green and mature phases ( Figure S4). However, the expansin-specific dendrogram also contained a third branch representing woody organs (plus pollen and senescing leaf), which was well-separated from the mature samples. This discrepancy between the expansin dendrogram and the whole-transcriptome atlas reflects the strong co-expression of the EXLB family in woody tissues.
Certain expansin genes were characterized by unique expression profiles in particular tissues, as confirmed by real time RT-PCR ( Figure 3C). VvEXPA9 was expressed uniquely in the seed after fruit setting ( Figure 3C, upper panel) and this sample also represents an outlier in the co-expression analysis (Figure 3B), suggesting a highly-specific expansin gene expression profile in seeds at this stage. VvEXPA8 was expressed predominantly in the root ( Figure 3C, middle panel), a phenomenon also observed in other species because expansins are specifically involved in processes such as root hair initiation in Arabidopsis [14] and root elongation in soybean (Glycine max) and rice [48,49]. Finally, VvEXPA12 was expressed strongly in pollen ( Figure 3C, lower panel) and to a lesser extent in the flower at the onset of flowering (10% caps off; Flower -FB) and during flowering (50% caps off; Flower -F), indicating a role in pollen development. No EXPB genes were expressed in grapevine pollen, whereas this family plays a prominent role in monocot pollen development [10]. Our data suggest that only the EXLB family shows tissue-dependent functional specialization in grapevine.

EXLB genes are expressed in woody tissues
EXLA and EXLB genes contain both domains that characterize the expansin superfamily but the sequence has diverged substantially from the EXPA and EXPB families and no function has yet been established for EXLAs and EXLBs [9].
The EXLB genes showed a unique expression profile in the grapevine gene expression atlas, characterized by little or no expression in the tissues that express EXPA and EXPB genes, such as berries, tendrils, flowers and leaves ( Figure 3). Although VvEXLB1 was not modulated in any of the tissues we analyzed, the remaining EXLB genes (VvEXLB2-4) were expressed strongly during the mature stages of rachis development (V, MR and R), in winter and swelling buds (bud-W and bud-S), in woody stems (stem-W) and, to a lesser extent, in roots and seeds ( Figure 4A). VvEXLB2 was strongly expressed in all the woody tissues listed above, VvEXLB3 was induced in the seed and rachis-MR samples, and VvEXLB4 was induced in woody stems and swelling buds. All these tissues are characterized by various degrees of lignification: by the time berries reach full maturity, the rachis is already brown, dehydrated and lignified, the seed contains 44% lignin (as well as 7% cellulose and 31% hemicellulose) [50], the winter and swelling buds are still enclosed within protective lignified tissues known as perulae, and the woody stem and roots have secondary cell walls.
The microarray expression profiles were validated by real time RT-PCR analysis of individual EXLB genes (VvEXLB3 and VvEXLB4) in rachis, bud and stem tissues ( Figure 4B). To emphasize the relationship between EXLB gene expression and secondary cell wall lignification, VvEXLB3 and VvEXLB4 transcript accumulation was compared in vegetative/green tissues (green stem, latent bud and rachis at fruit setting) and the same tissues during the mature/woody phase (winter stem, winter bud and rachis at the ripening stage), confirming that VvEXLB3 and VvEXLB4 are strongly induced in all woody tissues ( Figure 4B).
Expansins are thought to be required for cell wall loosening during primary wall growth [46], which occurs during many different developmental processes in growing and non-growing organs. However, certain expansin genes are expressed specifically in the xylem secondary wall, suggesting a role in secondary wall formation. In Chinese fir (Cunninghamia lanceolata), two recentlyidentified EXPA genes (ClEXPA1 and ClEXPA2) are strongly expressed in wood-forming tissues and preferentially in the cambium [51]. Expansins are also expressed during primary xylem formation in zinnia (Zinnea elegans) [52]. In hybrid aspen (Populus spp.), three EXPA genes (PttEXPA1, PttEXPA2 and PttEXPA8) and one EXPB gene (PttEXPB1) are expressed in the cambium and adjacent radial zone of mature stem tissues, supporting a role for expansins in secondary cell wall formation [17]. A recent transcriptome-wide analysis of the transition from primary to secondary stem development in Populus trichocarpa identified four EXPA genes and one EXLA gene whose expression profiles were linked to secondary growth [53]. The lack of functional data for EXLB genes means their precise role is difficult to establish. However, an intriguing observation is that the expansin superfamily in the moss Physcomitrella patens is similar in number to that in Arabidopsis, but lacks EXLA and EXLB genes [54]. Generally, bryophytes are land plants that do not have vascular tissues and do not contain lignin. This observation supports a relationship between the presence/action of expansinlike sequences such as EXLA and EXLB and the range of processes linked to lignification. Furthermore, a recent investigation of vascular overgrowth in Arabidopsis following treatment with an auxin transport inhibitor identified two EXLB sequences induced in vascular tissues and hydathodes [55]. Taken together, these data clearly indicate that EXLA and EXLB are specifically involved in secondary wall formation and support our proposal that at least VvEXLB2, VvEXLB3 and VvEXLB4 play specific functional roles in grapevine woody tissues and organs.

Berry growth and ripening require the activities of different expansin genes
The economical importance of grapevine depends largely on the quality of its berries, which undergo a long process of development and ripening involving a series of dramatic physical and biochemical changes. These include several cell wall structural modifications which are required for cell expansion and softening [56]. During berry development, there are two distinct periods of rapid growth [57]. The first growth phase involves an increase in cell number and volume, whereas the second involves an increase in volume only and is also characterized by softening.
We investigated the role of expansins in berry development by focusing on the transcriptional changes during fruit-set (FS) and post-fruit set (PFS) in the first growth phase, and during veraison (V), mid-ripening (MR) and ripening (R) in the second growth phase. The expression profiles of 14 grapevine expansin genes that show significant modulation of expression in at least one of the five samples (encompassing all expansin families except EXLB) are shown in Figure 5A. We identified 10 EXPA genes in clades EXPA-I and EXPA-II, suggesting these are particularly relevant during fruit development, as well as three EXPB genes with berryspecific expression (VvEXPB2-4) and also VvEXLA1, whose global expression profile suggests a role in berry ripening and rachis development ( Figure 3A). These expression profiles divide the samples in two clear groups: berries before veraison and berries during and after veraison ( Figure 5A). This is consistent with the contrast between vegetative/green and mature samples on the basis of expansin expression profiles ( Figure 3B).
The expansins expressed before veraison were VvEXPA4 and VvEXPA16 (validated by real-time RT-PCR as shown in Figure 5B), as well as VvEXPA6, VvEXPB2 and VvEXPA18. Interestingly, VvEXPA11 was expressed in the FS and V samples and only at low levels, suggesting a marginal involvement in berry development ( Figure 5A). We identified eight expansin genes whose expression was upregulated during the second growth phase. The majority of these (including VvEXPA19, VvEXLA1, VvEXPB4, VvEXPA14 and VvEXPA17) had similar expression profiles albeit with different degrees of induction. VvEXPB3 and VvEXPA1 showed single expression peaks in the MR and R samples, respectively, whereas VvEXPA5 was expressed only at veraison, suggesting a specific role during the second phase of berry growth. The expression profiles of VvEXPA19 and VvEXPA5 were also validated by RT-PCR ( Figure 5B).
We noted that some members of the expansin superfamily were expressed specifically either in the berry skin or pulp. For example, VvEXPA1 was expressed only in the skin whereas VvEXPA17 was found predominantly in the flesh ( Figure 3A). This suggests that common expansin substrates are modified specifically within each tissue and that different expansins contribute to the fine tuning of cell wall metabolism in the berry flesh and skin.
Previous studies have shown that some cell wall-modifying enzymes, including expansins, are expressed in a bimodal fashion in which high expression levels coincide with the two periods of rapid berry growth and cell expansion [37]. Transcriptome analysis has also demonstrated that the abundance of most expansin transcripts increases or decreases steadily during berry development [39,58] although some show characteristic expression profiles matching the second phase of fruit enlargement and softening [39,59,60]. One expansin transcript was also shown to be a potential negative marker (i.e. conspicuous by its absence) of the pre-ripening/ripening phases [41]. Expansin genes expressed during the first rapid period of berry growth period may be required for primary cell wall loosening and the incorporation of new cell wall material, whereas those expressed after veraison may regulate contact between cell wall polymers and the enzymes that degrade them, facilitating cell wall disassembly and fruit softening [36,37]. The classification of grapevine expansin genes involved in the modulation of berry cell wall metabolism, which is a key molecular event in berry development and ripening, could be used to improve berry quality traits and thus the economic value of grapevine.

Grapevine expansins are involved in water-deficit stress responses
A link has been established between expansins and water-deficit stress, as demonstrated by the induction of several expansin genes by dehydration stress [61,62] (as well as the induction of others by heat, shading, anoxia and pathogens [63][64][65][66][67]) and the ability of expansin transgenes to confer drought tolerance, e.g. in transgenic tobacco (Nicotiana tabacum) expressing TaEXPB23 [68,69] and in transgenic rose (Rosa spp.) expressing RhEXPA4 [70]. In this context, expansins may regulate adaptive changes in the cell wall structure to increase flexibility and thus prevent damage caused by the folding of rigid cell walls as the cell volume is reduced.
Samples representing controlled biotic and/or abiotic stress were not included in the grapevine gene expression atlas, but three stages of post-harvest berry withering were included [42]. This is a controlled form of post-harvest dehydration which is used to improve certain berry quality traits for the production of premium wines. During withering, the berries undergo several chemical and physical changes representing metabolic processes such as the modification of cell wall polymers in the skin [71,72]. A recent transcriptomic survey of withering in Raboso Piave berry skins [73] showed that VvEXPA1 transcript levels declined during this period. We therefore used the grapevine expression atlas data to investigate the role of grapevine expansins during post-harvest withering ( Figure 6A). In Corvina berries, 12 expansin genes were differentially modulated during withering and we analyzed their expression profiles separately in the flesh and skin. VvEXPA14, VvEXPB3 and VvEXPB4 were expressed in both tissues and were downregulated during withering. We validated the VvEXPA14 expression profile by real-time RT PCR ( Figure 6B). The other expansins showed a more complex expression profile, e.g. VvEXPA17 was upregulated at the second withering time point (PHWII) specifically in the berry skin, as confirmed by real-time RT PCR ( Figure 6B). VvEXPA19 and VvEXPA1 were modulated from the onset of the stress treatment (PHWI) and their expression was also skin-specific ( Figure 6A, 6B). Interestingly, VvEXLA1 expression was induced only in the berry flesh at the last stage of the dehydration process, when weight loss approached 30% (Figure 6A, 6B). The expansins that were induced during withering represent candidate dehydration stress response markers. The coordinated activity of pectin methylesterase (PME) and polygalacturonase (PG) has recently been shown to degrade pectins in Corvina berry skin [74]. Therefore, the skin-specific expansins induced during dehydration (e.g. VvEXPA19 and VvEXPA1) may be involved in pectin metabolism, probably by regulating contact between pectins and pectin-degrading enzymes. Our data suggest that expansins contribute to post-harvest dehydration stress responses by regulating cell wall metabolism in berries.

Conclusion
We carried out a genome-wide survey of the expansin superfamily in grapevine to gain insight into the chromosomal organization, phylogenetic relationships and expression profiles of these genes. The grapevine genome contains 20 EXPA genes, four EXPB genes, one EXLA gene and four EXLB genes, representing all 17 monocot-eudicot clades proposed for Arabidopsis and rice as well as a potential novel clade not identified thus far in other species, represented by VvEXPA10. Most of the grapevine expansin genes retained the ancestral exon-intron structure, although some showed unique structures that may be grapevine specific. A comparison of the grapevine, Arabidopsis, rice and poplar expansin genes revealed that grapevine and poplar expansins were the most closely related and that the grapevine expansins had not undergone significant phylogenetic expansion. However, integrated phylogenetic characterization and gene colinearity analysis within and between plant genomes will be necessary to understand the evolution of the grapevine expansin superfamily in more detail.
Genome-wide gene expression analysis indicated that members of the same expansin family did not necessarily share common expression profiles although most grapevine EXLB genes were expressed strongly in woody tissues and organs, suggesting a specific role in the formation of secondary cell walls. Correlative analysis of expansin gene expression profiles suggested temporaldynamic clustering in which mature samples and vegetative/green samples were clearly separated.
We also identified the expansins whose expression was significantly modulated during berry development, particularly in the context of the two distinctive phases of berry growth. All expansin families except EXLB were found to be involved in berry development, supporting previous reports that cell wall modification is a critical component of fruit growth and maturation.
Finally, we investigated expansin expression during post-harvest withering, which showed that some expansins are induced by dehydration, indicating a potential role in the control of adaptive cell wall modification under stress. Our data lay the groundwork for the further structural, functional and phylogenetic characterization of expansin genes and their developmental role in commercially valuable fruit crops.

Materials and Methods
Database searches, analysis of gene structure and chromosome mapping Previously-identified Arabidopsis and rice expansin protein sequences (AtEXPA1, AtEXPB1, ATEXLA1, ATEXLB1, OsEXPA1, OsEXPA24, OsEXPB2, OsEXPB15, OsEXLA1 and OsEXLB1) were used as BLAST queries against the Genoscope Grape Genome Database (http://www.genoscope.cns.fr) representing the 8.4X and 12X V0 assemblies of the PN40024 genotype. The search was extended by matching expansin genes and their chromosomal coordinates against an uploaded version of the PN40024 12X assembly V1 (http://genomes.cribi.unipd.it/), which merges the 12 V0 prediction based on GAZE software [75] and a gene prediction performed with JIGSAW software [43] at the CRIBI Genomics in Padova, Italy. Sequences were processed using Vector NTI v9 (Invitrogen) and the gene structure was deduced from the Genoscope and CRIBI annotations or by manual editing based on comparisons with corresponding expressed sequence tags and deduced protein sequences from paralogous expansin genes. The exon-intron structures were presented using the online tool FancyGene [76]. The chromosomal loci were inferred from the genome browsers listed above and rendered using NCBI MapViewer (http://www.ncbi.nlm.nih. gov/).

Phylogenetic analysis
Multiple alignments of the 29 deduced expansin protein sequences from grapevine, together with 16 from rice, 21 from poplar and 21 from Arabidopsis, were created using the E-INSI tool in the MAFFT v6.0 software suite, which takes into account the possibility of large gaps in the alignments [77]. An unrooted phylogenetic tree was generated using the neighbor joining method [78] in MEGA v5.0 [79]. The reliability of the tree was tested by bootstrapping with 1000 replicates, and the resulting tree was edited and modified with MEGA v5.0 and Treedyn software (http://treedyn.org). An additional tree was produced with the same tools considering only the grapevine expansin gene sequences predicted from the 12X V1 assembly of the PN40024 genotype.

Analysis of the V. vinifera cv Corvina expression atlas
The expression profiles of grapevine expansin genes were analyzed in the global V. vinifera cv Corvina (clone 48) gene expression atlas of different organs at various developmental stages [42]. The gene expression microarray data were obtained by hybridization to a NimbleGen microarray 090818 Vitis exp HX12 (Roche, NimbleGen), which contains probes targeting 29,549 predicted grapevine genes (http://ddlab.sci.univr.it/ FunctionalGenomics/), representing ,98.6% of the genes predicted from the V1 annotation of the 126 grapevine genome sequence (http://srs.ebi.ac.uk/) and 19,091 random probes as negative controls. The Corvina expression atlas comprises the following tissues: in vitro roots, green stem, buds after bud burst (rosette of leaf tips visible), young leaves (leaves collected from shoots with only five leaves), senescing leaves (leaves at the beginning of leaf-fall), berry rachis (from fruit-set to ripening), flowers (50% cap-fall). In particular, berry pericarp was collected at five developmental stages: the first stage (15 days (20.0u Brix). The atlas also contains a set of expression data obtained from berries that have undergone post-harvesting withering for 1-3 months after harvest: At the first withering stage (WI), berry weight was 76.4% the ripe value and the sugar content was 24.5u Brix. The second stage (WII) was characterized by 69.7% berry weight and 25.9u Brix, and the last stage (WIII) was characterized by 67.3% berry weight and 26.7u Brix. The expression data were analyzed using T-MeV v4.81 [80], normalized based on the mean center genes/ rows adjustment and (in the analysis of individual organs) were also variance filtered with a 50% cut-off. Abbreviations after organ names indicate the developmental stage. FS, fruit set; PFS, post fruit set; V, veraison; MR, mid-ripening; R, ripening; PHWI, postharvest withering (1 st month); PHWII, post-harvest withering (2 nd month); PHWIII, post-harvest withering (3 rd month), Bud -L, latent bud; Bud -W, winter bud; Bud -S, bud swell; Bud -B, bud burst; Bud -AB, bud after burst; Inflorescence -Y, young inflorescence with single flowers separated; Inflorescence -WD, well developed inflorescence; Flower -FB, flowering begins; Flower -F, flowering; Tendril -Y, young tendril; Tendril -WD, well developed tendril; Tendril -FS, mature tendril; Leaf -Y, young leaf; Leaf -FS, mature leaf; Leaf -S, senescing leaf; Stem -G, green stem; Stem -W, woody stem.

Correlation analysis
A correlation matrix was prepared using R software and Pearson's correlation coefficient as the statistical metric to compare the values of the 29 members of grapevine expansin genes in all 54 samples [42], using the average expression value of the three biological replicates. Correlation values were converted into distance coefficients to define the height scale of the dendrogram.
RNA extraction, cDNA synthesis and real-time RT-PCR Total RNA was isolated using the Spectrum TM Total RNA kit (SIGMA-Aldrich, St. Louis, MO), with sample-specific modifications of the extraction protocol, as previously described [42]. First strand cDNA was synthesized using 1 mg of total RNA as the template and the Improm-II TM Reverse Transcription system (Promega), according to the manufacturer's instructions. The total RNA was derived from Vitis vinifera cv. Corvina organs harvested at the same phenological phase as described in the grapevine gene expression atlas [42], and all RNA samples were first treated with DNase I (Promega). Gene-specific primers were targeted, when possible, to the 39-UTR, using ubiquitin (VIT_16s0098g01190) as the reference gene (Table S1). Primers and cDNA were mixed with the Power SYBRH Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) and the reaction was carried out on a Stratagene MX 3000 P TM QPCR System (Agilent, Technologies, CA) using the following cycling conditions: 95uC hold for 10 min followed by 45 cycles at 95uC for 30 s, 55uC for 30 s and 72uC for 20 s. Nonspecific PCR products were identified by inspecting dissociation curves. In addition, the specificity of EXLB PCR products was verified by cloning the relative amplicons in the pGemH-T Easy Vector (Promega), sequencing and aligning them onto the reference genome. Amplification efficiency was calculated from raw data using LingRegPCR software [81]. Standard error (SE) values were calculated as described [82]. Figure S1 Alignment of grapevine expansin protein sequences. Alignment of all grapevine expansin deduced protein sequences representative of the VvEXPA, VvEXPB, VvEXLA and VvEXLB families. The alignment was created using MAFFT v6.0 and edited with GeneDoc. The family-specific conserved motifs and active sites are indicated. (TIF) Figure S2 Gene frequency per chromosome of expansins in Arabidopsis, poplar, rice and grapevine. Arabidopsis data were obtained from TAIR, whereas poplar and rice data were deduced by consulting the expansin center site (https://homes.bio.psu.edu/ expansins/index.htm). (TIF) Figure S3 Phylogenetic tree of grapevine expansins together with a set of sequences obtained from Arabidopsis, rice and poplar that represent all of the monocot-eudicot clades present in all the three species proposed [45]. The sequences were aligned with CLUSTALW and a neighbor-joining tree was constructed with MEGA v4.0. Clades are indicated on the right with black bars, numbered using the conventional nomenclature [45]. (TIF) Figure S4 Expression of expansin genes in individual organs. The expression of expansin genes was considered separately in (A) bud, (B) leaf and (C)rachis. Genes were variance-filtered eliminating 50% of the expansin genes showing low variation between samples. (TIF) Table S1 List of real time RT PCR primers.