The RabGAP Gene Family in Tomato (Solanum lycopersicum) and Wild Relatives: Identification, Interaction Networks, and Transcriptional Analysis during Plant Development and in Response to Salt Stress

RabGTPase activating proteins (RabGAP) are responsible for directing the deactivation of vesicular trafficking master regulators associated to plant development, the RabGTPase proteins. Recently, RabGAPs were identified in Arabidopsis and rice, but studies were not yet reported in tomato. Herein, we identified 24 RabGAP-encoding genes in cultivated tomato (Solanum lycopersicum) and its wild relative genomes (Solanum pimpinellifolium and Solanum pennellii). We analyzed them based on their exon-intron structures, conserved protein motifs, putative subcellular localizations, phylogenetic and gene duplications analyses, interaction networks, and gene expression patterns in tomato. Phylogenetic relationship analysis also indicated that RabGAP family is classified into seven subclasses, of which subclasses I and II are plant-exclusive. Furthermore, segmental duplication events and positive evolutionary forces are associated with the maintenance of the number and function of their members. On the other hand, the protein–protein interaction networks on tomato suggested that members of subclasses I, II, and III could be associated to endocytic traffic routes. In addition, the qRT-PCR experiments in S. lycopersicum and Solanum chilense exposed to a salt stress treatment validated the differential expression patterns of 20 RabGAP genes in different tissues, development stages, and stress conditions obtained through extensive microarray-based analyses. This work suggests the critical role of RabGAP family in the context of intracellular vesicular trafficking in tomato, particularly under conditions of abiotic stress. It also contributes to the breeding programs associated with the development of crops tolerant to salt stress.


Introduction
Abiotic stress represents the main environmental challenge in the production of agronomically important crops. Salt, drought, or heat, among other stresses, alter the internal homeostasis of plants, damaging their tissues and organs, as well as reducing their yield [1]. To counteract these negative effects, mechanisms for synthesis and accumulation of compatible osmolytes [2], detoxification of reactive oxygen species [3], mobilization of ion and water transporters [4,5], or synthesis and mobilization of lipids are activated [6]. The performance of these mechanisms depends on the efficient vesicular traffic between the different subcellular compartments [7][8][9]. The small GTPases of the Rab family (RabGTPases) regulate the vesicular traffic, alternating between an "active" state GTP-bound and an "inactive" state GDP-bound, as a molecular switch. The activated state is dependent on guanine
To assign a name to each RabGAP protein and determine the subclass to which it belongs, the complete length of the amino-acid sequence was aligned with its closest homolog of A. thaliana using the software ClustalO [15,33]. The phylogenetic tree was constructed with MEGA7 software using the neighbor-joining method with 5000 iterations bootstrap [34].
Synteny analysis was carried out using the chromosomal locations of homologous and paralogous RabGAP genes in S. lycopersicum and A. thaliana obtained from the PGDD database (http://chibba.agtec. uga.edu/duplication/) [42]. The graphic representation was made using the web-based service ClicO FS [43]. Along with this, the selection pressure of RabGAP genes was determined by the ratio of nonsynonymous to synonymous nucleotide substitutions (Ka/Ks) between paralogs in S. lycopersicum. The approximate date of the duplication events was estimated using T = (Ks/2λ) × 10 −6 million years ago (mya), based on the clock-like rates (λ) in Solanaceae of 1.5 × 10 -8 [44].

RabGAP Gene Expression Profiles in Databases
Tissue-specific expression profiles of the RabGAP genes of S. lycopersicum and S. pennellii were obtained from the Bio-Analytic Resource database from the University of Toronto (http://bar.utoronto.ca). The data were obtained from flowers, shoots, leaves, vegetative meristems, seedling shoots, seedling roots, mature fruit, and developing fruit [45]. To establish a relation between the expression profiles of the RabGAP genes and the capacity of tolerance to abiotic stress among S. lycopersicum and its wild relatives (S. pimpinellifolium, S. habrochaites, and S. pennellii), values normalized in respect to S. lycopersicum were obtained from the "Tomato Expression" database (http://malooflab.phytonetworks. org/apps/tomato-expression/) [45]. In addition, to study the expression profiles of RabGAP genes in response to abiotic stress, we used the PLEX database (http://www.plexdb.org). The data were extracted from L6 (GEO accession GSE16401), L12 and L13 experiments (GEO accession GSE22304). The L6 experiment corresponded to a salt stress assay considering samples of leaves from 6-week-old S. lycopersicum and S. pimpinellifolium treated with NaCl 200 mM for 5 h [46]. The L12 experiment correspond to drought stress considering samples of leaves from four-week-old S. lycopersicum and S. habrochaites, previously non-irrigated for 14 days. Finally, the L13 experiment corresponded to a heat stress assay considering samples of leaves from four-week-old S. lycopersicum and S. habrochaites treated at 40 • C [47]. The profiles are relative to the control without stress.

Plant Material and Gene Expression Analysis
Seeds of S. chilense (Dunal) and S. lycopersicum were germinated in a mixture of perlite, vermiculite and peat moss (1:1:1), grown under greenhouse conditions at 23-25 ºC, with photoperiod of 16/8 h light/dark, irrigated with 400 mL every five days and fertilized with a commercial Hoagland's solution (1/4 strength) every 10 days. At the sixth week, a saline stress of NaCl 300 mM was applied to a group of plants. Leaves and roots were collected at 0, 3, 6, 12, 24, 48, and 72 h for the gene expression analyses. Then, total RNA from both tissues of S. chilense was extracted, treated, and quantified following the protocol of San Martín-Davison et al. (2017) [21]. To obtain the first strand of cDNA, 2 µg of RNA was retro-transcribed in 20 µL of reaction using the First Strand cDNA Synthesis Kit (Thermo Scientific, Waltham, MA, USA) and following the manufacturer's instructions.
Relative levels of gene expression were measured by qRT-PCR, using the Stratagene Mx3000p thermocycler (Agilent Technologies, Santa Clara, CA, USA). Maxima SYBR Green/ROX qRT-PCR Master MIX (Thermo Scientific) was used for all reactions according to the protocol described by the manufacturer. For each sample (three biological replicates), qRT-PCR was performed in triplicate (technical replicates) using 10 µL of Master MIX, 0.5 µL of 250 nM of primers, 1 µL of cDNA and nuclease-free water in a final volume of 20 µL [48]. The amplification was followed by a melting curve with continuous fluorescence measurement at 55 to 95 • C. The data were manually analyzed, and the expression normalized with the Ubiquitin3 gene [49]. All primers are listed in Table S1. The transcript levels of each gene were evaluated using the 2 −∆∆CT method [50]. The qRT-PCR data were analyzed using ANOVA (one-way analysis of variance) tests (with a significance of p-value < 0.05). Data management and standardization were performed with R version 3.2.5 [51].

24 Proteins Organized in 7 Subclasses Constitute the RabGAP Family in Tomato
An analysis of HMM on the proteomes of S. lycopersicum, S. pimpinellifolium, and S. pennellii allowed us to identify 24 putative RabGAP proteins in each species. They were named according to the closest ortholog of A. thaliana [15]. Among them, 21 RabGAPs possessed the TBC domain, two the Rab3-GTPase_cat domain and one the Rab3GAP2_N domain [14,52]. At the amino-acid level, all RabGAP proteins had the conserved residues of arginine and glutamine, critical for catalytic activity. The exceptions were RabGAP23b and those with the Rab3-GTPase_cat and Rab3GAP2_N domains, see in Figure S1. In S. lycopersicum, the RabGAP proteins had a length between 352 (SlRabGAP18) and 942 amino-acids (SlRab3GAP2) and a theoretical isoelectric point ranging from 4.79 (SlRabGAP6) to 8.97 (SlRabGAP18). Similar characteristics were also observed in the RabGAP protein families of S. pennellii and S. pimpinellifolium. On the other hand, the GRAVY index, associated to the hydrophobicity level of a protein, was less than 0, revealing that all RabGAPs are hydrophilic. In this same sense, the prediction of subcellular localization using three different web servers indicated that the main locations for all RabGAP proteins were in the nucleus and cytoplasm, as shown as in Table 1. In respect to the classification of RabGAP proteins, the phylogenetic analysis evidenced that they are grouped in seven subclasses (I, II, III, IV, V, VI-Rab3GAP, and VIII), according to their homologs in A. thaliana, shown in Figure 1 [15]. This analysis revealed that the subclass with the most members was subclass I and those with the fewest members were subclasses III and V. In addition, subclasses II, V, and VIII of S. lycopersicum had fewer members than those of A. thaliana, while subclasses III and IV had more. In respect to the classification of RabGAP proteins, the phylogenetic analysis evidenced that they are grouped in seven subclasses (I, II, III, IV, V, VI-Rab3GAP, and VIII), according to their homologs in A. thaliana, shown in Figure 1 [15]. This analysis revealed that the subclass with the most members was subclass I and those with the fewest members were subclasses III and V. In addition, subclasses II, V, and VIII of S. lycopersicum had fewer members than those of A. thaliana, while subclasses III and IV had more.  [15] are represented in green circles. In red, 24 RabGAP of S. lycopersicum; in light blue, 24 RabGAP of S. pimpinellifolium; and in dark blue, 24 RabGAP of S. pennellii.

Events of Segmental Duplication Have Kept Constant the Number of Members of the RabGAP Family
To evidence the genomic organization and the role of duplication events in the evolutionary history of RabGAP family genes, synteny mapping, intron-exon structure identification and selection pressure analysis were performed, see in Figure 2. First, the genomic analysis revealed that the SlRabGAP genes were differentially distributed on 11 chromosomes of S. lycopersicum: One gene each on Sl-Chr1, -4, -5, -10, and -11; two genes each on Sl-Chr2 and Sl-Chr3; three genes each on Sl-Chr6, -9, and -12; and six genes on Sl-Chr7. Then, the comparative analysis between the exon-intron structures and their chromosome locations allowed us to identify seven events of segmental duplication (SlRabGAP20-SlRabGAP22, SlRabGAP21a-SlRabGAP21b, SlRabGAP1a-SlRabGAP1b, SlRabGAP9a-SlRabGAP9b, SlRabGAP5-SlRabGAP6, SlRabGAP2a-SlRabGAP2b, and SlRabGAP3-SlRabGAP16) and one triplication event between SlRabGAP23a, -23b, and -15. Along with the above, in all these cases the ratios of synonymous and non-synonymous nucleotide substitutions (Ka/Ks) were less than 1, indicating that the function of the genes has not diverged and has been maintained due to a stabilizing selection force. Finally, the estimated time in which these duplications occurred was 15 to 28 mya, see in Table 2.

Events of Segmental Duplication Have Kept Constant the Number of Members of the RabGAP Family
To evidence the genomic organization and the role of duplication events in the evolutionary history of RabGAP family genes, synteny mapping, intron-exon structure identification and selection pressure analysis were performed, see in Figure 2. First, the genomic analysis revealed that the SlRabGAP genes were differentially distributed on 11 chromosomes of S. lycopersicum: One gene each on Sl-Chr1, -4, -5, -10, and -11; two genes each on Sl-Chr2 and Sl-Chr3; three genes each on Sl-Chr6, -9, and -12; and six genes on Sl-Chr7. Then, the comparative analysis between the exon-intron structures and their chromosome locations allowed us to identify seven events of segmental duplication (SlRabGAP20-SlRabGAP22, SlRabGAP21a-SlRabGAP21b, SlRabGAP1a-SlRabGAP1b, SlRabGAP9a-SlRabGAP9b, SlRabGAP5-SlRabGAP6, SlRabGAP2a-SlRabGAP2b, and SlRabGAP3-SlRabGAP16) and one triplication event between SlRabGAP23a, -23b, and -15. Along with the above, in all these cases the ratios of synonymous and non-synonymous nucleotide substitutions (Ka/Ks) were less than 1, indicating that the function of the genes has not diverged and has been maintained due to a stabilizing selection force. Finally, the estimated time in which these duplications occurred was 15 to 28 mya, see in Table 2.  In relation to the characterization of the RabGAP gene structure, the results showed that the number of introns is very diverse, see in Figure 3, Figure 3A. On average, the RabGAP genes had 13 introns, of which RabGAP2a, -20, and -22 had the lowest number (four introns), whereas Rab3GAP2 had the highest number (21 introns). On the other hand, the RabGAP proteins were characterized by having between four or five conserved motifs associated with the "catalytic core" from the TBC domain, as shown in Figure 3, Figure 3B. Exceptionally, the catalytic domain of subclass VI, specifically associated with Rab3, was present in three proteins-Rab3GAP1, Rab3GAP2, and RabGAP10-as seen as in Table 1.  In relation to the characterization of the RabGAP gene structure, the results showed that the number of introns is very diverse, see in Figure 3, Figure 3A. On average, the RabGAP genes had 13 introns, of which RabGAP2a, -20, and -22 had the lowest number (four introns), whereas Rab3GAP2 had the highest number (21 introns). On the other hand, the RabGAP proteins were characterized by having between four or five conserved motifs associated with the "catalytic core" from the TBC domain, as shown in Figure 3, Figure 3B. Exceptionally, the catalytic domain of subclass VI, specifically associated with Rab3, was present in three proteins-Rab3GAP1, Rab3GAP2, and RabGAP10-as seen as in Table 1.  Table S2.

In Silico Analysis of Protein-Protein Interactions Suggests the Association of RabGAP Proteins with Specific Vesicular Trafficking Pathway
In order to suggest specific vesicular trafficking pathways in which the RabGAP proteins could participate, we identified their possible affinities or physical interactions with the RabGTPase subfamilies of S. lycopersicum using the STRING database, see in Table S3. The results showed that in the exocytic pathway, SlRabGAP5, -6, -9a, and -9b could be interacting with RabGTPases of the subfamilies-A and E. In the anterograde pathway, RabGAP4 could be interacting with the subfamily-D, whereas in the retrograde pathway, SlRabGAP2a, -2b, -20, -21a, -21b, -23a, and -5 could be related with subfamily-H. In the endocytic pathway, only RabGAP18 could be interacting with subfamily-F, whereas SlRabGAP3, -7, -15, -16, -20, -21a, -21b, and -22 could be relating with subfamily-G (see in Figure 4).

In Silico Analysis of Protein-Protein Interactions Suggests the Association of RabGAP Proteins with Specific Vesicular Trafficking Pathway
In order to suggest specific vesicular trafficking pathways in which the RabGAP proteins could participate, we identified their possible affinities or physical interactions with the RabGTPase subfamilies of S. lycopersicum using the STRING database, see in Table S3. The results showed that in the exocytic pathway, SlRabGAP5, -6, -9a, and -9b could be interacting with RabGTPases of the subfamilies-A and E. In the anterograde pathway, RabGAP4 could be interacting with the subfamily-D, whereas in the retrograde pathway, SlRabGAP2a, -2b, -20, -21a, -21b, -23a, and -5 could be related with subfamily-H. In the endocytic pathway, only RabGAP18 could be interacting with subfamily-F, whereas SlRabGAP3, -7, -15, -16, -20, -21a, -21b, and -22 could be relating with subfamily-G (see in Figure 4). Genes 2019, 10, 638 10 of 20 Figure 4. Identification of proposed protein-protein interactions between RABs and RabGAPs. The vesicular trafficking routes directed by the RabGTPase subfamilies between the different organelles and membranes are represented by arrows. RabGAP proteins were associated to each RabGTPase using the protein-protein interaction functional networks provided by the STRING database [41]. ER: Endoplasmic Reticulum. MVB: Multi-Vesicular Body. All proposed protein-protein interaction networks are detailed in Table S4.

Differential Expression of RabGAP Genes in Different Tissues and Development Stages of Cultivated and Wild Tomato Species
To explore patterns in the transcriptional regulation of the RabGAP genes, we analyzed the expression profiles in different organs and development stages of S. lycopersicum and S. pennellii (see Figure 5, File S1). We evidenced that the expression patterns of RabGAP genes in flowers, shoots, The vesicular trafficking routes directed by the RabGTPase subfamilies between the different organelles and membranes are represented by arrows. RabGAP proteins were associated to each RabGTPase using the protein-protein interaction functional networks provided by the STRING database [41]. ER: Endoplasmic Reticulum. MVB: Multi-Vesicular Body. All proposed protein-protein interaction networks are detailed in Table S4.

Differential Expression of RabGAP Genes in Different Tissues and Development Stages of Cultivated and Wild Tomato Species
To explore patterns in the transcriptional regulation of the RabGAP genes, we analyzed the expression profiles in different organs and development stages of S. lycopersicum and S. pennellii (see Figure 5, File S1). We evidenced that the expression patterns of RabGAP genes in flowers, shoots, seedling shoots, seedling roots, and mature fruit of S. pennellii were inverse to that of S. lycopersicum. Furthermore, in almost all organs of S. lycopersicum the expression patterns were positive, while in S. pennellii, they were only positive in leaves and vegetative meristems. At the subclass level, no obvious differences were observed. However, we highlight that the most up-regulated RabGAP genes in S. lycopersicum were SlRabGAP23a in flowers, leaves, and vegetative meristems, SlRabGAP18 in stems and developing fruit, SlRabGAP2a in seedling shoots, and RabGAP15 in seedling roots. While in S. pennellii were SpRabGAP18 in seedling shoots, seedling roots, flowers and stems. SpRabGAP23a in leave, SpRabGAP3 in vegetative meristems, SpRabGAP15 in mature fruit, and SpRabGAP2a in developing fruit. It is also interesting to mention that RabGAP18, the only RabGAP possibly associated with RabGTPase of the F subfamily, was ubiquitously expressed in all organs and stages of development of both wild and cultivated tomato.   Table S5.
To evidence differential patterns of regulation between tomato and wild relatives in normal conditions, we analyzed the expression profiles of the RabGAP genes in leaves of S. pimpinellifolium, S. habrochaites, and S. pennellii, see in Figure 6. SpiRabGAP23a, ShaRabGAP3, and SpeRabGAP10 were the most up-regulated genes in each wild species, while SpiRab3GAP1, ShaRabGAP14, and SpeRabGAP16 were the most down-regulated. Studying an expression pattern associated with the degree of tolerance to abiotic stress of each wild species, we observed that RabGAP1a, -10, and -21a had positive correlations, whereas only RabGAP18 had a negative tendency. In addition, the analysis showed that both RabGAP2a and -22 represented about 35% of the number of transcripts of RabGAP  Table S5.
To evidence differential patterns of regulation between tomato and wild relatives in normal conditions, we analyzed the expression profiles of the RabGAP genes in leaves of S. pimpinellifolium, S. habrochaites, and S. pennellii, see in Figure 6. SpiRabGAP23a, ShaRabGAP3, and SpeRabGAP10 were the most up-regulated genes in each wild species, while SpiRab3GAP1, ShaRabGAP14, and SpeRabGAP16 were the most down-regulated. Studying an expression pattern associated with the degree of tolerance to abiotic stress of each wild species, we observed that RabGAP1a, -10, and -21a had positive correlations, whereas only RabGAP18 had a negative tendency. In addition, the analysis showed that both RabGAP2a and -22 represented about 35% of the number of transcripts of RabGAP family (see Table S6).
developing fruit. All values are detailed in Table S5.
To evidence differential patterns of regulation between tomato and wild relatives in normal conditions, we analyzed the expression profiles of the RabGAP genes in leaves of S. pimpinellifolium, S. habrochaites, and S. pennellii, see in Figure 6. SpiRabGAP23a, ShaRabGAP3, and SpeRabGAP10 were the most up-regulated genes in each wild species, while SpiRab3GAP1, ShaRabGAP14, and SpeRabGAP16 were the most down-regulated. Studying an expression pattern associated with the degree of tolerance to abiotic stress of each wild species, we observed that RabGAP1a, -10, and -21a had positive correlations, whereas only RabGAP18 had a negative tendency. In addition, the analysis showed that both RabGAP2a and -22 represented about 35% of the number of transcripts of RabGAP family (see Table S6).   Table S6.

RabGAP Genes Associated with Endocytic and Pre-Vacuolar Trafficking Are Up-Regulated in Roots Subjected to Salt Stress
To obtain more information about the role of RabGAP genes under abiotic stress conditions, we analyzed their expression profiles in response to heat, drought and salt stress in S. lycopersicum and compared them with those results obtained in wild relatives. In this case, the only data available for the RabGAP3, -4, -18, -20, -21a, and -21b were found in different databases (see Figure 7). During heat stress, only RabGAP4 and -20 were as strongly induced in the tolerant species as RabGAP4, -21a, and -21b were in the S. lycopersicum. Interestingly, under conditions of drought stress, unlike heat stress, RabGAP3, -4, -18, -20, and -21b of tolerant species versus RabGAP21a and -21b of S. lycopersicum, were up-regulated. Finally, under salt stress, similar expression profiles were observed in the sensitive and tolerant species. In this case, RabGAP21a and -21b were the most up-regulated genes in both species. stress, only RabGAP4 and -20 were as strongly induced in the tolerant species as RabGAP4, -21a, and -21b were in the S. lycopersicum. Interestingly, under conditions of drought stress, unlike heat stress, RabGAP3, -4, -18, -20, and -21b of tolerant species versus RabGAP21a and -21b of S. lycopersicum, were up-regulated. Finally, under salt stress, similar expression profiles were observed in the sensitive and tolerant species. In this case, RabGAP21a and -21b were the most up-regulated genes in both species.  Table S7.
Considering the low number of genes previously analyzed, and to learn more about the stress response and determine the possible biological effect of RabGAP genes of S. lycopersicum during salt stress, 20 RabGAP genes representing at least one gene of each subclass were selected for expression analysis by qRT-PCR (see Figure 8). Along with the above, we used S. chilense as the wild and halophyte species to contrast the analyses, as shown in Figure 9. The results showed that, in both leaves and roots of S. lycopersicum, four genes were up-regulated (SlRabGAP2b, -7, -9b, and -20), while 10 genes were induced in leaves, and three only in roots. At early times of salt stress, RabGAP2b and -22 in roots and RabGAP9b and -20 in leaves were the most up-regulated. Similarly, at late times, RabGAP3 and -22 in roots and SlRabGAP1a, -4, -10, and -18 in leaves had the highest expression levels. Interestingly, RabGAP3, -16, and -22 in roots and RabGAP1b in leaves managed to maintain a strong and stable induction throughout the experiment. In this context, it should be also noted that pairs of paralogous genes (RabGAP1a, -1b, -2a, -2b, -9a, -9b, -21a, and -21b) showed different expression patterns.  Table S7.
Considering the low number of genes previously analyzed, and to learn more about the stress response and determine the possible biological effect of RabGAP genes of S. lycopersicum during salt stress, 20 RabGAP genes representing at least one gene of each subclass were selected for expression analysis by qRT-PCR (see Figure 8). Along with the above, we used S. chilense as the wild and halophyte species to contrast the analyses, as shown in Figure 9. The results showed that, in both leaves and roots of S. lycopersicum, four genes were up-regulated (SlRabGAP2b, -7, -9b, and -20), while 10 genes were induced in leaves, and three only in roots. At early times of salt stress, RabGAP2b and -22 in roots and RabGAP9b and -20 in leaves were the most up-regulated. Similarly, at late times, RabGAP3 and -22 in roots and SlRabGAP1a, -4, -10, and -18 in leaves had the highest expression levels. Interestingly, RabGAP3, -16, and -22 in roots and RabGAP1b in leaves managed to maintain a strong and stable induction throughout the experiment. In this context, it should be also noted that pairs of paralogous genes (RabGAP1a, -1b, -2a, -2b, -9a, -9b, -21a, and -21b) showed different expression patterns.
The expression patterns of RabGAP genes in S. chilense revealed differences with those of S. lycopersicum, as seen as in Figure 9. In this species, 14 RabGAP genes significantly increased their relative expression in both leaves and roots during stress, whereas RabGAP15, -18, and -23a did it only in roots, and RabGAP21b only in leaves. We want to highlight the high and stable induction levels of RabGAP18 in roots of S. chilense throughout the salt stress, the inductions of RabGAP1a, -4, and -9a in roots at late times, and the large fluctuations in the expression of RabGAP2a in leaves. In addition, similar to S. lycopersicum, the expression levels of paralogous gene pairs also differed from the tissue in which they were expressed and their levels of induction. Figure 8. Spatio-temporal expression patterns of RabGAP genes of S. lycopersicum exposed to salt stress. Relative levels of SlRabGAP gene transcripts in roots (orange bar) and leaves (green bar) of 10-12-week-old S. lycopersicum plants were determined at 0, 3, 6, 12, 24, 48, and 72 h after stress initiation with 300 mM NaCl. The relative levels of the salt-stress marker gene (SlAREB1) were measured. In SlRabGAP14 and -23b, transcripts were not detected. Columns and error bars represent the mean and standard deviation for three biological and three technical replicates. * p-value < 0.05. Figure 8. Spatio-temporal expression patterns of RabGAP genes of S. lycopersicum exposed to salt stress. Relative levels of SlRabGAP gene transcripts in roots (orange bar) and leaves (green bar) of 10-12-week-old S. lycopersicum plants were determined at 0, 3, 6, 12, 24, 48, and 72 h after stress initiation with 300 mM NaCl. The relative levels of the salt-stress marker gene (SlAREB1) were measured. In SlRabGAP14 and -23b, transcripts were not detected. Columns and error bars represent the mean and standard deviation for three biological and three technical replicates. * p-value < 0.05. Figure 9. Spatio-temporal expression patterns of RabGAP genes of S. chilense exposed to salt stress. Relative levels of RabGAP gene transcripts in roots (orange bar) and leaves (green bar) of 10-12 weeks old S. chilense plants were determined at 0, 3, 6, 12, 24, 48, and 72 h after stress initiation with 300 mM NaCl. The relative levels of the salt-stress marker gene (SchAREB1) were measured. In SlRabGAP14, transcripts were not detected. Columns and error bars represent the mean and standard deviation for three biological and three technical replicates. * p-value < 0.05.
The expression patterns of RabGAP genes in S. chilense revealed differences with those of S. lycopersicum, as seen as in Figure 9. In this species, 14 RabGAP genes significantly increased their relative expression in both leaves and roots during stress, whereas RabGAP15, -18, and -23a did it only in roots, and RabGAP21b only in leaves. We want to highlight the high and stable induction levels of RabGAP18 in roots of S. chilense throughout the salt stress, the inductions of RabGAP1a, -4, and -9a in roots at late times, and the large fluctuations in the expression of RabGAP2a in leaves. In Figure 9. Spatio-temporal expression patterns of RabGAP genes of S. chilense exposed to salt stress. Relative levels of RabGAP gene transcripts in roots (orange bar) and leaves (green bar) of 10-12 weeks old S. chilense plants were determined at 0, 3, 6, 12, 24, 48, and 72 h after stress initiation with 300 mM NaCl. The relative levels of the salt-stress marker gene (SchAREB1) were measured. In SlRabGAP14, transcripts were not detected. Columns and error bars represent the mean and standard deviation for three biological and three technical replicates. * p-value < 0.05.

Discussion
Within the Solanaceae family, the commercial tomato, S. lycopersicum, is the main cultivated vegetable in the world, but its sensitivity to abiotic stress negatively affects its productivity. On the other hand, wild species related to it such as S. pimpinellifolium, S. habrochaites, S. pennellii, and S. chilense are able to tolerate different degrees of abiotic stresses. A strategy of tolerance that S. chilense particularly uses could be associated with vesicular trafficking processes such as endocytosis, vacuolar compartmentalization of toxic ions, and damaged element recycling [21]. In this context, we have identified and characterized for the first time the family of RabGAP genes of different Solanaceaes through analyzing of the whole genome, mainly through transcriptional study. This allowed us to indicate the possible relationships between them and the components that direct the vesicular trafficking.

Overview of RabGAP Gene Family in the Tomato Genome
The RabGAP gene family encodes proteins capable of accelerating the hydrolysis of GTP to GDP bound to RabGTPase proteins, allowing them to direct a new round of vesicular trafficking [53]. In this work, 24 RabGAP genes were identified in each tomato species (S. lycopersicum, S. pennellii, and S. pimpinellifolium), a similar number of genes to that of other species such as rice or A. thaliana [15]. However, according to motifs and multiple alignment analyses, the results suggest that four proteins, including RabGAP23b, should not have RabGAP activity due to the absence of a conserved Arginine residue. In this regard, it has been evidenced that a point mutation on this residue in OsGAP1 (R450A) causes a severe loss of this activity against two RabGTPases-OsRab11 and OsRab8a-revealing its importance in catalytic activity [24].
Interestingly, the subclasses I and II of the RabGAP family are composed of six and two RabGAP genes, respectively. In addition, both families are potentially exclusive to plants and it is predicted that their evolution is an adaptation mechanism due to the loss of function related to the RasGAP family [15]. From the above, possible coevolution mechanisms between RabGTPases and RabGAPs, which have sculpted the endomembrane system, have been suggested [54]. On the other hand, the equivalent number of members of the RabGAP family between tomato and evolutionarily distant species, such as A. thaliana or rice, is an important question to raise. In this sense, it has been suggested that duplication and diversification events from five ancestral RabGAP genes have been the mechanisms that have kept the number of genes constant [15,28,55]. The exon-intron structure also provides relevant information to understand the evolution process of the RabGAP family. In this analysis, we found paralogous genes from subclasses I (RabGAP23b and -15), III (RabGAP21b and -21b), and VIII (RabGAP9a and -9b) located on chromosomes 7 and 12. The arrangements and size of exons and introns of these genes supports the hypothesis of an important segmental duplication event between both chromosomes, which has also been reported for members of the B-BOX, LEA, and GRAS families of S. lycopersicum [56][57][58]. Together, these data suggest that events of segmental duplications have played an important role in the gene evolution of the RabGAP gene family.

The New Subfamily VI and Rab3GAP Could Play an Important Role in Autophagy and Stress Tolerance
Regarding the characteristics of other identified genes, we found two new members of subclass VI with a Rab3GAP domain whose functional roles in plants have not yet been determined, see in Figure 1 and Table 1. However, homolog genes in animal cells participate in: (i) the deactivation of Rab3 [14], a homolog of RabD in plants, (ii) interaction with ATG8 in autophagy processes [59,60], (iii) or activation of a RabGTPase of subfamily C, acting as a guanine exchange factor in degradative pathways and macro-autophagy [61]. Intriguingly, a RabD of plants (RABD2A) localizes with proteins associated with salt-stress tolerance capacity, such as EHD1 or SYP41 [62,63]. In a similar way to expression patterns of Rab3GAP1, Rab3GAP2, and RabGAP10 of S. lycopersicum, the autophagy is a mechanism rapidly induced by salt stress [64]. These interesting characteristics suggest that the proteins of subclass VI can probably interact with RabD or ATG8 and play an important role during salt stress in tomato.

Endocytosis-Associated RabGAPs Could Be Mediating the Salt Stress Tolerance in Tomato
In general, the different RabGAP proteins of S. lycopersicum identified in this work were located in the nucleus or cytoplasm. Particularly, the predicted nuclear localization was consistent among RabGAP22 from S. lycopersicum and A. thaliana [65]. The patterns of expression of both genes are also similar. In both species, RabGAP22 is expressed in various tissues and stages of development, such as flowers, roots, or vegetative meristems. Along with this, RabGAP22 from A. thaliana can interact with peroxisome proteins such as AGT1 and be associated with photorespiration processes or adaptation mechanisms against abiotic stress [66]. According to this, the homolog of AtRabGAP22 in S. lycopersicum could have an important role in tomato roots when they are subjected to salt stress. On the other hand, the predicted subcellular location of RabGAP21a from S. lycopersicum, with respect to its rice homolog, OsGAP1 [23], is also coherent. In this case, the data suggest that SlRabGAP21a could be localized in the Trans-Golgi network and the pre-vacuolar compartments. Equally, it is interesting to observe that the expression patterns of OsGAP1 and RabGAP21a of S. lycopersicum are similar in leaves and roots when plants are subjected to salt stress [24]. These data suggest that RabGAP21 could also have an important role in a trafficking pathway that mobilizes transporters or proteins associated with the vacuole. Additionally, SlRabGAP21a could potentially interact with the TVP38 protein, which has been associated with the stability of the membranes [67]. The results for RabGAP18 are also consistent with the fact that the endocytosis mechanism is an inducible process against salt stress and regulated for RabGTPases of the subfamily-F [20,22]. To date, the RabGAP proteins that regulate this subfamily-F are unknown, however, the protein-protein interaction analysis suggested that RabGAP18 could be an exclusive regulator of this family. Interestingly, RabGAP18 is also strongly induced in the roots of S. chilense when they are exposed to salt stress in a similar manner to SchRabGDI1, whose overexpression in A. thaliana can increase its tolerance to salt stress [21]. In general, the various antecedents support the idea that RabGAP have a certain degree of affinity for RabGTPases, but the results equally show a high degree of promiscuity for some of them. In this sense, further analyses and experiments are needed to establish the vesicular trafficking routes in which the members of the RabGAP family of S. lycopersicum participate in order to finally understand their physiological functions during the abiotic stress.

Conclusions
In conclusion, we have identified 24 RabGAP genes in each S. lycopersicum, S. pennellii, and S. pimpinellifolium and classified them into seven subclasses. Two proteins with Rab3GAP domain were identified for the first time. All of them have the TBC domain characteristic of the RabGAP family. Evolutionarily segmental duplication and purifying selection have maintained constant the number and function of these genes in the S. lycopersicum genome. The results presented here show that the RabGAP proteins, such as RabGAP18 or RabGAP22, could be associating to different intracellular vesicular trafficking pathways and that the genes that encode them can change their expression depending on both endogenous and environmental stimuli. In this sense, the study of the regulation mechanisms of gene expressions, as well as the specific interactions with RabGTPase proteins could provide important information to understand their participation in tolerating abiotic stress in tomato species.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4425/10/9/638/s1, Figure S1: Multiple alignment of RabGAP sequences. RabGAP amino-acid sequences of S. lycopersicum and A. thaliana. In green, conserved amino-acids associated with catalytic activity, File S1: Expression patterns of RabGAP genes in different tissues and development stages of tomato (Tomato eFP Browser), Table S1: Primers used in this study, Table S2: Regular expression profile of the conserved motifs defined by MEME of SlRabGAP, Table S3: ID of RabGAP genes in A. thaliana used in phylogenetic analysis, Table S4: Protein-Protein interaction networks predicted with STRING database, Table S5: Relative expression of RabGAP genes in different tissues and development stages of S. lycopersicum and S. pennellii. Log2 (ratio), Table S6: Relative expression of RabGAP genes among different tomato species (normalized counts per million), Table S7: Relative expression of RabGAP genes from contrasting tomato species in different types of abiotic stresses.