Members of the plant AP family have been implicated in various physiological and developmental processes, including protein processing and degradation, senescence, stress response, programmed cell death and reproduction. However, virtually nothing is known about this family in woody species. Since grapevine is one of the most important fruit trees worldwide, and various forms of both biotic and abiotic stresses have a serious impact on its production and quality. Further study on stress-related responses in this genus could prove to be a significant asset. Therefore, we have sought to undertake the genome-wide identification of AP genes in grape, and provide clues regarding both their evolutionary histories and expression diversity with respect to stress-related conditions.
Tandem and segmental duplications contributed to the expansion of the grape AP gene family
Gene duplication, including tandem, segmental and whole genome duplications, has played an important role in the evolution of various organisms , and land plants have undergone abundant gene duplication throughout their evolutionary history . Since the grapevine genome has not undergone any recent whole genome duplication events , segmental and tandem duplications would be the two main causes of gene family expansions in grape, although there is debate on the exact nature and timing of these events in grape [12, 13]. In this study, 26 of 46 grape AP genes which could be precisely located on chromosomes were associated with either tandem or segmental duplication events (Figure 1 and Additional file 3), consistent with findings in rice whereby 51 of 93 AP genes were located in either tandemly or segmentally duplicated regions . Taken together, this suggests that tandem and segmental duplications likely played an important role in the expansion of the AP family in plants. Although the duplicated grape AP genes identified here have a common ancestor, we could not conclude from the work conducted here what the ancestral functions and expression patterns may have been, since gene duplication is considered to provide the raw material for evolution and duplicated genes which, if survive, could undergo substantial changes in their structures and/or regulatory mechanisms allowing them to assume novel roles .
The structural conservative and divergence of grape AP genes
Although several models of the genome evolution have been proposed from comparative genomic analyses of model organisms [54–56], little attention has been paid to the structural evolution of duplicated gene families . In fact, exon/intron diversification of gene family members has played an important role in the evolution of multiple gene families through three main types of mechanism: exon/intron gain/loss, exonization/pseudoexonization, and insertion/deletion . It was obvious that grape AP genes within the same phylogenetic clade (Figure 4A) possessed highly similar exon/intron structures (Figure 4C) and most of the grape AP genes that clustered in the same phylogenetic clade were segmental or tandem duplications (Figure 1 and Figure 4A). Based on the results presented here, it is clear that the expansion of the grape AP family was the result of either segmental or tandem duplication. This is consistent with findings in rice , despite the fact that the numbers of exons of the VvAP genes in category C (1–4 or 9–12) were different from those of the OsAP genes in the same category (less than four).
Exon/intron gain/loss and divergence in exon/intron length which were observed within the coding sequences of several grape AP genes may be the result of chromosomal rearrangement and fusions. A good case was VvAP32, which had seven exons, but its paralogous gene, VvAP21, had just two exons. Moreover, VvAP32 contained two other domains (RING and DUF1117) besides the ASP domain, and this might have resulted from a mechanism(s) leading to the additional exons. Divergence either in exon/intron length or exon/intron quantity could potentially lead to the generation of functionally distinct paralogs .
Functional conservation and divergence of tandem and segmental duplicated AP genes in grape
As discussed above, tandem and segmental duplication expanded the grape AP gene family, but it should be noted that only one gene in two of the four tandem duplicated paralogs (VvAP5/VvAP6/VvAP7/VvAP8, VvAP22/VvAP23/VvAP24) could be included in the phylogenetic tree (Figures 1 and 3), implying that sequences of these genes have been altered to a large extent after gene duplication, and may therefore have lost their original functions or gained new ones.
The grape AP genes involved in the tandem and segmental duplication with complete ASP domains, clustered in the same phylogenetic clade (Figure 4A), and had highly similar exon/intron structures (Figure 4C); however, in different tissues and/or under a variety of stress or hormone treatments, the transcript abundance of the two genes within each pair of paralogs varied from each other (Figures 5 and 6). Although the expression patterns of VvAP34 and VvAP35 under each treatment were almost identical, different expression levels could be observed in the same plant organ. The expression of another pair of tandem duplicated paralogs, VvAP3 and VvAP4, were barely detectable under the various treatments, but slight expression of VvAP3 could be detected in the stems and tendrils of ‘Kyoho’. Regarding the segmental duplication paralogs, almost all of the two AP genes within each pair of paralogs showed different transcript levels under three or four of the five treatments, and exhibited similar transcript abundance under the other one or two treatments. One exception was VvAP17/VvAP42, whose transcript levels were totally different under all of the five treatments. These results are similar to findings in rice , which showed that not all OsAP genes in the same category had similar expression patterns, and OsAP genes classified in the same expression pattern might have a distant phylogenetic relationship.
It seems possible that high sequence similarity is not necessarily correlated with similar transcript levels, because proteins with very similar sequences, presumably performing similar biochemical functions, are needed in different tissues and at different periods during growth and development, while at the same time responding to different stresses and hormone treatments. Similar transcript abundance exhibited by different AP genes with dissimilar sequences may perform different biochemical functions, suggesting they may work together in the tissues during growth and development or in response to the same stress or hormone treatment.
It has been reported that duplicated genes rarely diverge with respect to their biochemical function, but instead are limited to alterations in regulatory control . So the different expression profiles between duplicated genes may be caused by varied regulatory network or mutations in the cis-regulatory regions , or mutations affecting the related regulatory network [60, 61]. Epigenetic mechanisms, such as DNA methylation have also been suggested to potentially contribute to the expression divergence of duplicated genes [62, 63], where transcriptional silencing has often been associated with DNA methylation in promoter regions [64, 65].
A large part of expression divergence is considered to arise through duplication in the course of evolution , and functional diversification of the surviving duplicated genes is also considered a major feature of the long-term evolution of polyploids . It has been reported that four types of functional differentiation may follow gene duplication: pseudogenization, conservation of gene function, subfunctionalization and neofunctionalization . Many duplicated genes may be lost from the genome after the duplication events, while neofunctionalization and subfunctionalization contribute to the retention of new genes. The VvAP gene family presents an opportunity to study how expression has diverged following gene duplication. Similar transcript abundance between duplicated genes, such as VvAP34 and VvAP35, suggest that the regulatory mechanism of their expression have been conserved; on the other hand, divergence in expression patterns of the duplicated AP genes (neofunctionalization or subfunctionalization) could reflect the acquisition of novel regulatory mechanisms, while silencing of gene expression after duplication leading to nonfunctionalization of the gene implies drastic alteration of the regulatory mechanism.
Besides the possibilities that have been discussed above, the differences in specificity/catalytic properties and cellular localization among/between the duplicated genes could also contribute to the development of different biological functions, leading to the observed expression divergence [6, 68, 69]. It was found that almost all of the AP genes within each pair of paralogs were located in different parts of the cell, with two exceptions of VvAP3/VvAP4 (both in the nucleus), and VvAP17/VvAP42 (both in the plasma membrane) (Additional file 11). Even if these paralogs shared similar gene structure and cellular localization, the diversity of expression of these genes may be the result of alterations in regulatory sequences occurring shortly after duplication. In addition, alteration of function could also result from the presence or absence of protein-processing enzymes responsible for the activation/deactivation of the enzymes .
In summary, diversity in the transcript levels of the duplicated genes may be affected by different and multiple genetic factors depending on the causal duplication mechanism , and there maybe cross talk between different treatments or regulatory mechanisms. More research is needed to clarify the specifics of any functional divergence between grape duplicated AP genes, and new factors that may affect transcript divergence and how different factors work together are worth investigation.
The evolution of AP proteins in grape and Arabidopsis and functional prediction of grape AP genes
Genomic comparison is a quick way to transfer knowledge acquired in one taxon for which there is a better understanding of genome structure and function to a less-studied taxon . Thus, the richness of gene functional information known for model plants such as Arabidopsis enables one to extrapolate functions of their orthologous genes in other plant taxa. To obtain an overall picture of the grape AP proteins and their relationships with those of Arabidopsis, both syntenic and phylogenetic analyses have been performed, and the evolutionary relationship of this gene family within and among the different species has been systematically studied.
There were 23 grape and 25 Arabidopsis AP genes, as well as the other four Arabidopsis genes that were syntenic orthologous (Figure 2, Additional file 4). Among these, 10 were single grape-to-Arabidopsis AP orthologs, indicating these genes come from a common ancestor. The other genes constituted a more complex situation, including ten cases of two grape AP genes that corresponded to one Arabidopsis AP gene, 10 cases of one grape AP gene corresponding to multiple Arabidopsis AP genes, and three cases of two duplicated grape AP genes that corresponded to multiple Arabidopsis AP genes. Certainly, most of the genes included in the complex situation appeared more than once. For example, VvAP27 correspondence to At1g11910 and At1g62290 located on the Chr1 of Arabidopsis, as well as At4g04460 and At4g022050 located on the Chr4 of Arabidopsis, and VvAP45 correspondence to At1g11910, At1g62290 and At4022050, but not to At4g04460, so it is impossible to elucidate whether divergence of VvAP27 and VvAP45 located in segmental duplications of grape and At1g11910, At1g62290, At4022050 and At4g04460 in Arabidopsis occurred prior to or after the divergence of grape and Arabidopsis from the last common ancestor. Although 27 grape AP genes could not be mapped into any syntenic blocks, we could not conclude that these genes from grape and Arabidopsis did not share a common ancestor. This may be explained by the fact that after the divergence of lineages that led to grape and Arabidopsis, their genomes underwent multiple rounds of significant chromosomal rearrangement and fusions, followed by selective gene loss, which can severely obscure the identification of chromosomal syntenies . In such case, it maybe concluded that some of the AP genes in grape and Arabidopsis come from a common ancestor, while the others do not. Although the evolutionary histories of grape AP genes could not be established for the period prior to the split between grape and Arabidopsis lineages, at least some of the grape genes appeared to share a common ancestor with their Arabidopsis AP counterparts.
In order to improve prediction of the functions of specific grape AP genes based on the reported function of their Arabidopsis homologs, a phylogenetic tree was constructed, and bootstrap support values (1000 re-sampling) exceeding 50% were used to identify possible orthologous pairs (Figure 3). For example, VvAP30 and At4g30030 were clustered together in the phylogenetic tree, but the bootstrap value of their node was no more than 50%. Therefore, VvAP30 and At4g30030 were excluded from the orthologous pairs, as were VvAP13 and At1g08210/At5g22850, VvAP16/VvAP44 and At3g50050, VvAP27 and At4g04460, VvAP17/VvAP42 and At5g10080, VvAP46 and At3g12700. There were 11 orthologous pairs in the phylogenetic tree that could not be detected in the syntenic orthologs, and 21 syntenic orthologs that could not be detected or were not clustered together in the phylogenetic tree (Additional file 4, Additional file 5, Additional file 12). Thus, there were ten orthologs including 8 grape AP genes (VvAP27/VvAP45-At1g11910/At1g62290, VvAP39-At1g49050, VvAP21-At5g02190, VvAP25-At2g36670, VvAP32-At2g39710, VvAP33-At3g54400 and VvAP44-At5g43100) that could be clustered together in the phylogenetic tree and were also contained in the syntenic map (Additional file 5, Additional file 12). Expression patterns of the ten orthologs were more similar than other orthologous pairs that only clustered together in the phylogenetic tree or were syntenic orthologs (Additional file 12). As a result, we can speculate that the functions of the eight grape AP genes are more similar to their Arabidopsis homologs than the other grape APs in the phylogenetic tree and syntenic map. Ling et al.  has used phylogeny-based methods to identify orthologs between Arabidopsis and cucumber, and further analyzed the correlation of roles of orthologous pairs under abiotic stresses. Their results showed that correlative expression profiles in stress-inducible orthologous WRKY genes between cucumber and Arabidopsis and orthologous WRKY genes with different evolutionary patterns displayed a low correlation in their expression patterns . Our study combining synteny analysis with a phylogenetic tree provides new insight for investigating the function of grape AP genes by comparing orthologous genes between two plants, in one of which functional roles for the genes have been identified, in this case, between grape and Arabidopsis.
VvAP proteins play important roles in a range of biological processes
It has been reported that plant APs are implicated in a variety of biological processes [6, 7]. The study on the rice AP gene family showed that 66 genes were presented in at least one of the developmental stages analyzed . Timotijevic et al.  isolated an aspartic proteinase gene FeAP12 from developing buckwheat seeds, and found the gene was seed-specifically expressed. Moreover, the highest levels of FeAP12 expression were observed in the early stages of seed development, suggesting a potential role in nucellar degradation . Our RT-PCR results showed that most of the VvAP genes exhibited diverse expression levels in all six organs, indicating these APs often participate in plant development. Genes which showed higher expression levels in one organ than in others may play key roles in the development process of the corresponding organ. It was worth noting that the expression of VvAP36 could only be detected in stems, indicating its potential role in the stem development.
Evidence is accumulating that AP proteins are involved in plant responses to various abiotic and biotic stresses. Cruz et al.  reported that in drought-susceptible common bean cultivars subjected to water deficit, the expression of an AP gene was shown to be transcriptionally up-regulated and its activity was significantly increased. In recently published reports, Yao et al.  have shown that an Arabidopsis gene, ASPG1 (aspartic protease in guard cell 1), may function in drought avoidance through abscisic acid (ABA) signaling in guard cells. An aspartic protease gene, FeAP9, whose expression was up-regulated in leaves under different abiotic stresses has also been found in developing organs of buckwheat . In the present study, we showed that 18 VvAP genes exhibited differential transcript abundance in response to at least one abiotic stress (Figure 6), indicating that VvAP genes may play an important role in protecting grape from abiotic stresses.
Expression of an extracellular AP gene has also been detected in tobacco and tomato leaves and implicated in the degradation of pathogenesis-related (PR) proteins. It has been suggested that APs may play a role in a conserved mechanism for PR-protein turnover, preventing over accumulation and thereby regulating the biological functions of these stress induced proteins [76, 77]. These APs were also shown to be constitutively expressed either in healthy or infected leaves, which was consistent with our findings in this study. Studies with potato showed that the expression levels of StAPs were associated with the degree of resistance of potato cultivars to Phytophthora infestans, and potato aspartic proteinases were components of the plant defense response . Xia et al.  have also shown the accumulation of an AP gene, CDR1 (Constitutive disease resistance), in response to pathogen attacks. The CDR1 gene in rice has also been studied, and the results suggested that OsCDR1 was implicated in disease resistance signaling . Powdery mildew, caused by the obligate biotrophic fungus, Uncinula necator, has a serious impact on grape productivity and fruit quality . As shown in Figure 6, four of the grape AP genes exhibited increased transcript abundance in the infected leaves, indicating these genes may participate in the plant response to powdery mildew infection. It has been reported that the PSI may take part in defensive mechanisms against pathogens and/or as an effector of cell death , but none of the grape AP genes in group A1 was up-regulated upon powdery mildew infection. However, we cannot conclude that the VvAPs in group A1 have no function in defensive against pathogens, because they may participate in resistance against other pathogens.
Besides the functions of APs in response to abiotic and/or biotic stress, some APs were reported to be involved in PCD (programmed cell death) . In addition, nucellin, an AP belonging to group B subfamily and known to be expressed specifically in nucellar cells during degeneration after ovule fertilization in barely, was suggested to be involved in PCD . So we can speculate that VvAPs in group B is also involved in various types of PCD .
Differences of the cellular localization of AP genes may result in their different biological functions [6, 68, 69], and it has been reported that most plant APs were vacuolar enzymes [82–85], or were secreted to the cell wall[86, 87]. But there were many aspartic proteinases in Arabidopsis and one third of the grape AP genes were predicted to be localized to the chloroplast and chloroplast thylakoid membrane, respectively  (Additional file 11). In tobacco, one chloroplast-located AP gene named CND41 (for 41 kD Chloroplast Nucleoid DNA-binding protein) is involved in degradation of the Rubisco holoprotein during leaf senescence, and the accumulation of CND41 is negatively correlated with chloroplast transcript levels in tobacco cells [88–90]. The homologs of CND41 in Arabidopsis have also been confirmed to participate in the regulation of Rubisco turnover and senescence [91–93]. In a more recently published report, Paparelli et al.  have identified a chloroplast-located AP gene NANA whose misexpression or overexpression not only influences photosynthetic carbohydrate metabolism but also plastid and nuclear gene expression . So the localization of these AP genes to the Arabidopsis or grape thylakoid membranes raises the possibility that they may fulfill roles as specific processing enzymes in this organelle, or participate in maintenance or degradation of photosystem proteins [7, 95].
To get a more complete understanding of the biological functions of the AP gene family, identification of substrates that AP proteins act on and the regulatory network of AP genes participating in response to various pathogens are necessary. The results presented here indicate that the regulatory role of AP proteins under abiotic and biotic stress is complex and more work is needed to understand the regulatory mechanisms.