Overexpression of a Novel ROP Gene from the Banana (MaROP5g) Confers Increased Salt Stress Tolerance

Rho-like GTPases from plants (ROPs) are plant-specific molecular switches that are crucial for plant survival when subjected to abiotic stress. We identified and characterized 17 novel ROP proteins from Musa acuminata (MaROPs) using genomic techniques. The identified MaROPs fell into three of the four previously described ROP groups (Groups II–IV), with MaROPs in each group having similar genetic structures and conserved motifs. Our transcriptomic analysis showed that the two banana genotypes tested, Fen Jiao and BaXi Jiao, had similar responses to abiotic stress: Six genes (MaROP-3b, -5a, -5c, -5f, -5g, and -6) were highly expressed in response to cold, salt, and drought stress conditions in both genotypes. Of these, MaROP5g was most highly expressed in response to salt stress. Co-localization experiments showed that the MaROP5g protein was localized at the plasma membrane. When subjected to salt stress, transgenic Arabidopsis thaliana overexpressing MaROP5g had longer primary roots and increased survival rates compared to wild-type A. thaliana. The increased salt tolerance conferred by MaROP5g might be related to reduced membrane injury and the increased cytosolic K+/Na+ ratio and Ca2+ concentration in the transgenic plants as compared to wild-type. The increased expression of salt overly sensitive (SOS)-pathway genes and calcium-signaling pathway genes in MaROP5g-overexpressing A. thaliana reflected the enhanced tolerance to salt stress by the transgenic lines in comparison to wild-type. Collectively, our results suggested that abiotic stress tolerance in banana plants might be regulated by multiple MaROPs, and that MaROP5g might enhance salt tolerance by increasing root length, improving membrane injury and ion distribution.


Introduction
Small GTPases (GTP)-binding proteins, present in a wide variety of eukaryotes, are the central regulators of numerous signal transduction processes [1,2]. These proteins are structurally classified into at least five families, including Rat sarcoma (RAS), Ras homolog (RHO), Rat brain (RAB), RAS-related nuclear (RAN), and adenosine diphosphate (ADP) ribosylation factor (ARF) [1][2][3]. In all reported eukaryotes, RAS and RHO families are signaling switches, whereas these proteins in other families are primarily involved in the regulation of vesicle and large molecule movement [1,2]. However, higher plants have a unique RHO subfamily of small GTP-binding proteins known as ROPs (Rho-like GTPases from plants) [4,5]. These proteins are also known as RAS-related C3 botulinum toxin substrates (RACs), due to their sequence similarity to Rac GTPases [6].
ROP expression levels and biological functions are affected by various abiotic stressors. When exposed to cold, the transcriptional expression of apple ROP increases, leading to a decrease in the concentrations of ethylene and reactive oxygen species in the fruits [21]. In A. thaliana, ROP11 expression affected seed germination, seedling growth, stomatal closure, abscisic acid (ABA)-mediated responses, and drought stress responses [22]. The overexpression of ROP1 in tobacco increased salt sensitivity in response to salt stress by increasing H 2 O 2 production [23]. The Na + /K + ratio in transgenic A. thaliana expressing the Medicago falcate small GTPase gene (MfARL1) was lower than that in wild-type (WT) A. thaliana; the transgenic plants consequently had an increased tolerance for salt stress [24]. Knock out of the A. thaliana ROP effector (RIC1) increased the survival rate of plants under salt stress by improving the reassembly of depolymerized microtubules [25]. Taken together, these studies have revealed the many important roles of ROPs in the regulation of plant response to abiotic stresses.
The banana (Musa acuminata) is one of the most intensively produced and globally important fruit crops [26]. As a large monocotyledonous herbaceous annual, banana plants are frequently harmed or destroyed by various abiotic stress conditions during growth and development [27]. In particular, saline soil is a major abiotic stressor which limits banana cultivation worldwide [28,29]. Genome-wide identification of genes involved in the resistance of banana plants to cold, drought, and salt stress increases our knowledge of plant mechanisms for environmental stress tolerance, while the functional identification of relevant genes acts as a framework for future genetic studies focused on increasing the resistance of the banana plant to these stressors [30][31][32][33]. However, genome-wide investigations of the ROP gene family, and thus an integrated assessment of the potential functions of this important molecular switch, are still lacking in banana.
To address this information gap, we identified ROPs genome-wide in M. acuminata, known as MaROPs, and analyzed their phylogenetic relationships, gene structures, protein motifs, and expression changes in response to a number of abiotic stressors, including cold, drought, and salt. More importantly, we noted that the expression of the MaROP5g gene was associated with salt stress in banana. The overexpression of MaROP5g in A. thaliana conferred increased salt tolerance by lengthening roots, improving recovery of membrane injury and ion distributions. This comprehensive study of MaROPs in M. acuminata enhances our understanding of ROPs in response to abiotic stress conditions in banana plants, and provides a foundation for future studies aiming to improve the abiotic stress resistance of crop plants, especially with respect to salt stress.

Gene Structure and Conserved Motif Analysis of Banana MaROP Genes
Evolutionary analysis supported the classification of the 17 MaROP genes into three distinct groups (Groups II-IV), which is consistent with their exon-intron structural divergence within Figure 1. Phylogenetic analysis of Rho-like GTPases from plants (ROPs) from A. thaliana, rice, and bananas. The maximum-likelihood phylogenetic tree was drawn with MEGA5.2, using 1000 bootstraps. Four subgroups were identified (Groups I-IV). Circles, squares, and triangles represent ROP proteins from rice, A. thaliana, and bananas, respectively.

Gene Structure and Conserved Motif Analysis of Banana MaROP Genes
Evolutionary analysis supported the classification of the 17 MaROP genes into three distinct groups (Groups II-IV), which is consistent with their exon-intron structural divergence within families ( Figure 2). Our analysis of the exon-intron structure using the Gene Structure Display Server showed that the MaROP genes contained 8 exons in Group II, 7 exons in Group III, and 6-7 exons in Group IV, suggesting the conservation of the exon-intron structures of MaROPs within the same group.
To explore MaROPs structural diversity and potential functionality, we analyzed the conserved motifs of the identified MaROPs and predicted their functional annotations. We identified 10 conserved motifs across the 17 MaROP proteins with Multiple Em (Motif Elicitation), which are annotated with InterPro ( Figure 2; Table S2). Motifs 1-4 were annotated as a RhoGEF domain (PF00621), the characteristic domain of the ROP protein family. Motifs 1-5 were found across all of the 17 MaROPs, while motif 6 was only present in MaROP-2a, -2b, and -2c; motifs 6-8 were found in MaROP-3a and -3b; and motifs 9 and 10 were found in MaROP4. It is probable that this pattern of conservation and variation across the motifs reflected the evolutionary relatedness and functional divergence of the 17 MaROPs. families ( Figure 2). Our analysis of the exon-intron structure using the Gene Structure Display Server showed that the MaROP genes contained 8 exons in Group II, 7 exons in Group III, and 6-7 exons in Group IV, suggesting the conservation of the exon-intron structures of MaROPs within the same group.
To explore MaROPs structural diversity and potential functionality, we analyzed the conserved motifs of the identified MaROPs and predicted their functional annotations. We identified 10 conserved motifs across the 17 MaROP proteins with Multiple Em (Motif Elicitation), which are annotated with InterPro ( Figure 2; Table S2). Motifs 1-4 were annotated as a RhoGEF domain (PF00621), the characteristic domain of the ROP protein family. Motifs 1-5 were found across all of the 17 MaROPs, while motif 6 was only present in MaROP-2a, -2b, and -2c; motifs 6-8 were found in MaROP-3a and -3b; and motifs 9 and 10 were found in MaROP4. It is probable that this pattern of conservation and variation across the motifs reflected the evolutionary relatedness and functional divergence of the 17 MaROPs. Exon-intron structure analyses were performed with the Gene Structure Display Server (GSDS). Blue boxes indicate upstream/downstream; yellow boxes indicate exons; black lines indicate introns. All of the proteins were identified using the Multiple EM for Motif Elicitation (MEME) database, using the complete predicted amino acid sequences of each MaROP. Motifs 1-4 were annotated as a RhoGEF domain.

Expression Analysis of MaROP Genes in Response to Cold, Salt, and Osmotic Stresses
To investigate the response of the MaROP genes in response to different abiotic stressors, we analyzed the MaROP expression in banana leaves following exposure to cold, salt, and osmotic stress conditions ( Figure 3A; Table S3). Compared to the control, significant differences in the expression of 14 MaROP genes (82%) were detected following the exposure to abiotic stress All of the proteins were identified using the Multiple EM for Motif Elicitation (MEME) database, using the complete predicted amino acid sequences of each MaROP. Motifs 1-4 were annotated as a RhoGEF domain.

Expression Analysis of MaROP Genes in Response to Cold, Salt, and Osmotic Stresses
To investigate the response of the MaROP genes in response to different abiotic stressors, we analyzed the MaROP expression in banana leaves following exposure to cold, salt, and osmotic stress conditions ( Figure 3A; Table S3). Compared to the control, significant differences in the expression of 14 MaROP genes (82%) were detected following the exposure to abiotic stress treatments ( Figure 3A; Table S3). In BaXi Jiao (BX), the expression levels of MaROP-3b, -5a, -5c, -5f, -5g, and -6 were significantly upregulated, as indicated by the fragments per kilobase of exon per million fragments mapped (FPKM) value, which is higher than 2.0 by all three of the abiotic stressors. MaROP5d was upregulated by cold treatment only (FPKM > 2.0), and MaROP2c was downregulated by osmotic treatment only (FPKM < 0.5). In Fen Jiao (FJ), MaROP-3b, -5a, -5c, -5f, -5g, -6 were significantly expressed (FPKM > 8. 9) in the presence of abiotic stressors. Under osmotic treatment, MaROP6 was upregulated (FPKM > 2.3) in FJ, but maintained a low level of expression in BX (FPKM < 0.66). In addition, MaROP-3a and -5h are significantly downregulated in the presence of stress in BX and FJ, compared to control. MaROP5g had a higher level of expression (FPKM > 24) in response to salt stress than any of the other ROPs across both the BX and FJ genotypes, implying MaROP5g might play an important role in the regulation of salt stress tolerance in banana plants.

Validation of Differential Expression of Six MaROP Genes by Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) Analysis
Our RNA-seq analysis indicated that MaROP-3b, -5a, -5c, -5f, -5g, and -6 showed significant expressions by abiotic stressors. Such a feature of these six genes was further verified by quantitative real-time polymerase chain reaction (qRT-PCR) analysis. After normalization, all the examined MaROPs, with the exception of MaROP3b in FJ-salt and MaROP5c in FJ-salt, were well-correlated and generally consistent with our RNA-seq analyses (r = 0.8789-0.9992; Figure 3B; Table S4), indicating the reliability of our transcriptomic results in both banana varieties. Of particular interest was MaROP5g, which showed high expression following salt stress treatment compared to other MaROP genes, as determined by both RNA-seq and qRT-PCR experiments.

Full-Length cDNA, Subcellular Localization, and Expression Pattern of MaROP5g under Salt Stress
Based on the results of the RNA-seq and qRT-PCR analyses, we used PCR to amplify the full-length cDNA of MaROP5g from banana roots. The full-length MaROP5gcDNA had a 591 bp open reading frame (ORF), encoding 196 amino acids. The predicted MaROP5g protein had a typical RhoGEF domain and several additional characteristics of the ROP protein family ( Figure S1; Table S2).
We measured the transcriptional response of MaROP5g in BX and FJ plant roots to salt stress. Compared to 0 h (no stress condition), the roots of the BX plants became black following 6 h of salt stress treatment ( Figure 4A). However, there was no discernible phenotypic change in the roots of FJ plants under the same treatment ( Figure 4C). The expression of MaROP5g quickly increased from 0 h, reached the maximum level at 4 h, and then gradually decreased at 6 h ( Figure 4B). The expression pattern of MaROP5g was similar between FJ and BX ( Figure 4D), but MaROP5g showed lower expression in FJ compared to BX under salt stress. These results suggested that the regulation of MaROP5g expression by salt treatment was genotype-dependent, as the roots of the two tested banana genotypes may have variable sensitivity to salt stress treatments. BX showed more sensitivity than FJ under salt stress treatment.
To localize the MaROP5g protein in the cell, we introduced the MaROP5g ORF into a pCAMBIA1304-GFP vector upstream of the GFP gene to create a MaROP5g-GFP fusion construct, which was used to transform A. thaliana. Co-localization experiments showed that the MaROP5g-GFP (green fluorescent protein) fusion protein was localized to the FM4-64-labeled plasma membrane in A. thaliana root tips ( Figure 4E).   Phenotypes of BX roots exposed to salt stress; (B) expression of MaROP5g in BX roots exposed to salt stress; (C) phenotypes of FJ roots exposed to salt stress; (D) expression of MaROP5g in FJ roots exposed to salt stress. Data are presented as means ± standard deviations of n = 3 biological replicates. Different lowercase letters above bars indicate significant differences at p < 0.05, and different uppercase letters above bars indicate extremely significant differences at p < 0.01, using Duncan's multiple range tests. (E) MaROP5g subcellular localization. GFP fluorescence is green, and FM4-64 is red. Merge was created by merging the GFP and FM4-64 fluorescent images. Scale bars = 10 μm.

MaROP5g Overexpression Enhances Tolerance to Salt Stress
To investigate the role of MaROP5g in response to salt stress, MaROP5g was introduced into the pCAMBIA1304 vector under the control of the 35S promoter. After a floral-dip transformation of A. thaliana, we analyzed three transgenic lines with single-copy transgene (R3, R8, and R42) from the T3 generation, selected through genomic DNA Southern blot analysis ( Figure S2A). The expression level of MaROP5g in three transgenic lines was 89-109 folds compared to WT and empty vector (VC), as revealed by qRT-PCR analysis ( Figure S2B).
Under no-salt (control) and high-salt conditions, the seed germination rate and root growth were greater in the transgenic seedlings as compared to those in the WT seedlings. After salt treatments ranging from 100 to 200 mM, the transgenic seedlings had grown longer primary roots, as compared to WT seedlings ( Figure 5A-C). Furthermore, when adult A. thaliana growing in soil were treated with 350 mM salt daily for 15 days, the transgenic lines grew better ( Figure 5D,E) and were more likely to survive ( Figure 5F), as compared to those of WT. Thus, transgenic A. thaliana lines overexpressing MaROP5g were more tolerant to salt stress than those of WT. Phenotypes of BX roots exposed to salt stress; (B) expression of MaROP5g in BX roots exposed to salt stress; (C) phenotypes of FJ roots exposed to salt stress; (D) expression of MaROP5g in FJ roots exposed to salt stress. Data are presented as means ± standard deviations of n = 3 biological replicates. Different lowercase letters above bars indicate significant differences at p < 0.05, and different uppercase letters above bars indicate extremely significant differences at p < 0.01, using Duncan's multiple range tests. (E) MaROP5g subcellular localization. GFP fluorescence is green, and FM4-64 is red. Merge was created by merging the GFP and FM4-64 fluorescent images. Scale bars = 10 µm.

MaROP5g Overexpression Enhances Tolerance to Salt Stress
To investigate the role of MaROP5g in response to salt stress, MaROP5g was introduced into the pCAMBIA1304 vector under the control of the 35S promoter. After a floral-dip transformation of A. thaliana, we analyzed three transgenic lines with single-copy transgene (R3, R8, and R42) from the T3 generation, selected through genomic DNA Southern blot analysis ( Figure S2A). The expression level of MaROP5g in three transgenic lines was 89-109 folds compared to WT and empty vector (VC), as revealed by qRT-PCR analysis ( Figure S2B).
Under no-salt (control) and high-salt conditions, the seed germination rate and root growth were greater in the transgenic seedlings as compared to those in the WT seedlings. After salt treatments ranging from 100 to 200 mM, the transgenic seedlings had grown longer primary roots, as compared to WT seedlings ( Figure 5A-C). Furthermore, when adult A. thaliana growing in soil were treated with 350 mM salt daily for 15 days, the transgenic lines grew better ( Figure 5D,E) and were more likely to survive ( Figure 5F), as compared to those of WT. Thus, transgenic A. thaliana lines overexpressing MaROP5g were more tolerant to salt stress than those of WT.   ANOVA was used to compare the significance of differences, using Dunnett's tests in the comparison between WT and each overexpression lines. Data are presented as means ± standard deviations of n = 3 biological replicates. Different lowercase letters above bars indicate significant differences at p < 0.05, different uppercase letters above bars indicate extremely significant differences at p < 0.01.

MaROP5g Overexpression Reduced Malonaldehyde (MDA) Content and Ion Leakage (IL), Increased Ca 2+ and K + /Na + Ratio under Salt Stress
Malonaldehyde (MDA) is usually employed as an index of oxidative damages in plants [36]. Ion leakage (IL) is an important indicator of membrane injury [36]. To investigate whether MaROP5g expression influences MDA and IL content, we measured MDA and IL content in the shoots and roots of transgenic lines and WT plants, following high-salt treatments vs. no-salt control ( Figure 6). Following high-salt treatment, MDA content was lower in the shoots and roots of the transgenic plants as compared to those of WT ( Figure 6A-G). Similarly, IL value was lower in the shoots and roots of transgenic plants as compared to those of WT under high salt treatment ( Figure 6B-H). Taken together, these results suggest that the overexpression of MaROP5gin transgenic A. thaliana plants may have prevented or reduced membrane injury under salt stress as compared to WT.

MaROP5g Overexpression Reduced Malonaldehyde (MDA) Content and Ion Leakage (IL), Increased Ca 2+ and K + /Na + Ratio under Salt Stress
Malonaldehyde (MDA) is usually employed as an index of oxidative damages in plants [36]. Ion leakage (IL) is an important indicator of membrane injury [36]. To investigate whether MaROP5g expression influences MDA and IL content, we measured MDA and IL content in the shoots and roots of transgenic lines and WT plants, following high-salt treatments vs. no-salt control ( Figure 6). Following high-salt treatment, MDA content was lower in the shoots and roots of the transgenic plants as compared to those of WT ( Figure 6A-G). Similarly, IL value was lower in the shoots and roots of transgenic plants as compared to those of WT under high salt treatment ( Figure 6B-H). Taken together, these results suggest that the overexpression of MaROP5gin transgenic A. thaliana plants may have prevented or reduced membrane injury under salt stress as compared to WT. ANOVA was used to compare the significance of differences, using Student's t tests in the comparison between control and NaCl treatment. Data are presented as means ± standard deviations of n = 3 biological replicates. Different lowercase letters above bars indicate significant differences at p < 0.05, and different uppercase letters above bars indicate extremely significant differences at p < 0.01.
Under highly saline conditions, plant cells survive by retaining a high cytosolic Ca 2+ concentration and a high K + /Na + ratio [37,38]. Under high-salt treatment, the concentrations of Ca 2+ and K + in the shoots or roots of transgenic A. thaliana plants were greater ( Figure 6C,D), while the Na + concentration was lower in the shoots or roots of transgenic A. thaliana plants as compared to ANOVA was used to compare the significance of differences, using Student's t tests in the comparison between control and NaCl treatment. Data are presented as means ± standard deviations of n = 3 biological replicates. Different lowercase letters above bars indicate significant differences at p < 0.05, and different uppercase letters above bars indicate extremely significant differences at p < 0.01.
Under highly saline conditions, plant cells survive by retaining a high cytosolic Ca 2+ concentration and a high K + /Na + ratio [37,38]. Under high-salt treatment, the concentrations of Ca 2+ and K + in the shoots or roots of transgenic A. thaliana plants were greater ( Figure 6C,D), while the Na + concentration was lower in the shoots or roots of transgenic A. thaliana plants as compared to those of WT ( Figure 6E). Therefore, the shoots or roots of the transgenic lines maintained higher K + /Na + ratios than did those of the WT plants during salt treatment ( Figure 6F). These results suggested that the overexpression of MaROP5g in plants subjected to salt stress increased cellular Ca 2+ and K + accumulation, decreased cellular Na + accumulation, and improved the K + /Na + ratio.

MaROP5g Overexpression Increased the Expression of Salt Overly Sensitive (SOS)-Pathway and Ca 2+ -Sensing Genes
To gain an in-depth understanding of MaROP5g function in response to salt stress, we measured the expression of three Salt Overly Sensitive (SOS)-pathway genes (namely SOS1, SOS2, and SOS3) and several genes encoding calcium-signaling pathway proteins, including calcineurin B-like (CBL) proteins, CBL-interacting protein kinases (CIPKs), and calcium-dependent protein kinases (CDPKs) [39], in both WT and MaROP5g-overexpressing A. thaliana (Figure 7). Under standard growth conditions, we observed no significant differences in the transcription levels of the tested genes in the transgenic lines as compared to those in WT plants. However, under salt stress, the gene expression levels of SOS1, SOS2, SOS3, CBL, CIPK, and CDPK were higher in the transgenic lines as compared to those in WT. This indicated that MaROP5g overexpression in response to salt stress led to the up-regulation of both SOS-pathway genes and calcium-signaling pathway genes.
those of WT ( Figure 6E). Therefore, the shoots or roots of the transgenic lines maintained higher K + /Na + ratios than did those of the WT plants during salt treatment ( Figure 6F). These results suggested that the overexpression of MaROP5g in plants subjected to salt stress increased cellular Ca 2+ and K + accumulation, decreased cellular Na + accumulation, and improved the K + /Na + ratio.

MaROP5g Overexpression Increased the Expression of Salt Overly Sensitive (SOS)-Pathway and Ca 2+ -Sensing Genes
To gain an in-depth understanding of MaROP5g function in response to salt stress, we measured the expression of three Salt Overly Sensitive (SOS)-pathway genes (namely SOS1, SOS2, and SOS3) and several genes encoding calcium-signaling pathway proteins, including calcineurin B-like (CBL) proteins, CBL-interacting protein kinases (CIPKs), and calcium-dependent protein kinases (CDPKs) [39], in both WT and MaROP5g-overexpressing A. thaliana (Figure 7). Under standard growth conditions, we observed no significant differences in the transcription levels of the tested genes in the transgenic lines as compared to those in WT plants. However, under salt stress, the gene expression levels of SOS1, SOS2, SOS3, CBL, CIPK, and CDPK were higher in the transgenic lines as compared to those in WT. This indicated that MaROP5g overexpression in response to salt stress led to the up-regulation of both SOS-pathway genes and calcium-signaling pathway genes.  Data are presented as means ± standard deviations of n = 3 biological replicates. Different lowercase letters above bars indicate significant differences at p < 0.05, and different uppercase letters above bars indicate extremely significant differences at p < 0.01.

Discussion
Despite its economic and social importance, research on banana plants has generally been slower relative to many other crops, especially with respect to the abiotic stress responses [31,32]. ROP is an important molecular switch involved in plant signal transduction processes, which has been suggested to play crucial roles in the regulation of the environmental stress responses in numerous plant species [8,11,21,40]. We have identified 17 MaROPs by searching the M. acuminata genome, which were classified into three groups (II-IV), following the nomenclature derived for OsROPs [18]. Of the 17 MaROPs, none was categorized into Group I, congruent with ROPs in other higher plants, such as O. sativa [18], Z. mays [16], Medicago truncatula [7], and N. tabacum [7]. The recovered phylogenetic relationships were further supported by our analyses of gene structure and conserved motifs. The MaROP genomic sequences in Groups II-IV were found to contain 6-8 exons and 6-7 introns. Similar structural features have been observed in ROPs of other plant species, including N. tabacum [23], A. thaliana [15], and M. truncatula [7]. Moreover, all of the identified MaROPs had the typical RhoGEF domain (PF00621), and MaROP proteins within each group shared similarly conserved motifs (Figure 2), which is consistent with the observations in M. truncatula [7].
Bananas are extremely sensitive to abiotic stress and can suffer severe losses in yield and quality when exposed to cold, salt, or drought conditions [36]. We found that 82% of the 17 MaROPs showed transcriptional changes following cold, salt, and osmotic stress treatments ( Figure 3). Interestingly, except the high expression genes (MaROP-3b, -5a, -5c, -5f, -5g, and -6) under three stress treatments, MaROP-3a and -5h are significantly downregulated in the presence of stress in BX and FJ compared to control. To our knowledge, this is the first report showing that banana MaROPs exhibit extensive and diverse responses to abiotic stressors. The induction of ROP expression by cold, salt, and drought has previously been reported in other plants, such as Malus× domestica Borkh [21], A. thaliana [22], and N. tabacum [23].
It is particularly important to note that the expression of MaROP5g among the 17 MaROP genes was most highly induced following salt stress treatment across both banana genotypes tested (Figures 3 and 4A-D), which may imply its positive role in mediating banana's response to salt stress. Further, although MaROP5g expression in roots of both BX and FJ genotypes can be induced by salt stress treatment, BX showed more sensitivity than FJ under salt stress treatment. This may suggest that BX, with its genome constitution as AAA, is more sensitive to salt treatment in comparison to the B-genome-containing genotype FJ. Such an observation is consistent with previous studies that FJ, with AAB genotypes, exhibited higher tolerance to abiotic stresses relative to BX [27].The MaROP5g protein was located on the plasma membrane ( Figure 4E), consistent with A. thaliana ROP2 [41] and rice OsRac5 [18]. To better understand the function of MaROP5g during salt stress, we generated a number of MaROP5g-overexpressing transgenic A. thaliana lines. Under salt stress, the transgenic seedlings and adult plants exhibited a higher survival rate and increased root length as compared to WT ( Figure 5), suggesting that MaROP5g overexpression might contribute to the maintenance of a healthy growth status, through the improvement of root development and distributions [24,25], and hence enhance salt stress tolerance.
As cell membranes are one of the primary targets of various environmental stresses, MDA is commonly used as an index of oxidative damages [36], and IL is used as an important indicator of membrane injury in plant research [36]. MDA content and IL were measured to assess the role of MaROP5g overexpression in reducing membrane injury under salt conditions. MaROP5g overexpression resulted in decreased IL and MDA content relative to WT, indicating that MaROP5g-overexpressing plants may experience less membrane injury and maintain a healthy physiological status under salt conditions. In plants, high K + and low Na + concentrations are beneficial for the maintenance of physiological processes under salt stress [42]. In recent years, a high cytosolic K + /Na + ratio has become an accepted marker of salinity tolerance [38]. Previous studies have reported that the expression of MfARL1 resulted in a reduced Na + /K + ratio in transgenic A. thaliana as compared to WT, due to a lower accumulation of Na + [24]. Under salt stress, MaROP5g overexpression decreased the accumulation of cellular Na + , increased the Ca 2+ concentration, and improved the K + /Na + ratio in transgenic A. thaliana as compared to WT plants ( Figure 6). Therefore, the increased salt stress tolerance conferred by MaROP5g overexpression may be due not only to the decreased Na + accumulation, but also to the increased Ca 2+ concentration in transgenic lines as compared to WT.
Many different ion transporters and channel proteins, such as SOS, CDPK, CBL, and CIPK, play crucial roles in maintaining ion homeostasis during salt stress [39,42,43]. For example, under salt stress, the SOS1 and SOS2 proteins regulate Na + /K + homeostasis in A. thaliana; once the calcium binding protein SOS3 senses an increase in cytosolic calcium concentration, the SOS3-SOS2 protein kinase complex activates the SOS1 ion transporter [42,43]. In addition, calcium (Ca 2+ ), as a second messenger, plays an important role in salt stress processes [39,44]. Ca 2+ increase can be decoded and recognized by Ca 2+ sensors, including CBLs, CIPKs, and CDPKs [44]. CBLs recognize the increase in cytosolic Ca 2+ concentration triggered by Na + accumulation [39]. CIPKs and CDPKs may unify and coordinate ionic homeostasis at the cellular and organismal level [44]. We examined the expression of these SOSand calcium-signaling pathway genes in the MaROP5g-overexpressing transgenic A. thaliana seedlings in relation to WT seedlings. Following salt treatment, SOS-and calcium-signaling pathway genes were a significant expression in the transgenic seedlings as compared to WT seedlings. This suggested that the MaROP5g-overexpressing transgenic plants were more responsive to SOS-and calcium-signaling compared to WT plants, implying that MaROP5g-overexpressing plants had improved Na + and Ca 2+ ionic homeostasis under salt stress conditions.

Plant Materials
BaXi Jiao (BX; M. acuminata cv. Cavendish; AAA group) is a triploid banana cultivar that is of high yield and high quality and can be stored for an extended period of time [31,32]. Fen Jiao (FJ; M. acuminata; group AAB), another triploid banana cultivar, has good flavor, rapid ripening, and a high tolerance for abiotic stress [31,32] For the salt and drought-simulation experiments, five-leaf stage banana plants of both cultivars were irrigated with 300 mmol·L −1 NaCl or 200 mmol·L −1 mannitol, respectively, for 7 days, as previously described by Hu et al. [27]. For the cold experiments, five-leaf stage banana plants of both cultivars were subjected to 4 • C for 22 h, as previously described by Hu et al. [27]. For the control experiments, five-leaf stage banana plants of both cultivars were irrigated with equal volume water with the stress groups at 28 • C. After the completion of each treatment, we harvested and immediately froze in liquid nitrogen the leaves and root systems of each plant, which were stored at −80 • C. Twelve five-leaf stage banana plants and three biological replicates were performed for each treatment.

Transcriptomic Analysis
We isolated total RNA from the leaf tissues of the banana seedlings subjected to each of the three treatments (salt, osmosis, and cold) and control (no stress conditions), which were constructed into respective cDNA libraries [31,32]. Deep paired-end sequencing was performed with an Illumina GAII according to manufacturer's instructions. There are two replicates for each sample. The sequencing depth was 5.34X on average. Adaper sequences in the raw reads were removed using FASTX-tookit (Illlumina, San Diego, CA, USA). Using Tophat v.2.0.10, clean reads were mapped to the DH-Pahang genome [33]. The transcriptome assemblies were performed by Cufflinks [27]. Gene expression levels were calculated as fragments per kilobase of exon per million fragments mapped (FPKM). DEGseq was used to identify differently expressed genes [55]. Under three stress conditions, the expression level of each gene was compared with the control. A heat map was created based on the FPKM value of the MaROPs, compared to the control by MeV 4.9.0 software (available online: https://sourceforge.net/projects/mev-tm4/).

QRT-PCR Analysis
The gene expression of MaROPs in response to cold, salt, and osmotic (drought) stress was measured with qPCR, using a SYBR Premix ExTaq kit (TaKaRa, Shiga, Japan) on a Stratagene Mx3000P detection system (Stratagene, San Diego, CA, USA). The primer pairs with high specificity and efficiency were selected based on their melting curve and on agarose gel electrophoresis (Table S5). The amplification efficiencies of the primer pairs chosen ranged from 0.9 to 1.1. MaActin (EF672732) and MaUBQ2 (HQ853254) were used as the internal controls. The expression levels of MaROP relative to MaActin and MaUBQ2 were calculated with the 2 −∆∆CT method [56]. Three biological replicates for each sample were performed.

Full-Length cDNA of MaROP5g and Gene Expression During Salt Treatment
Based on our RNA-seq results, we selected the ROP gene MaROP5g for further analysis. The entire coding region of MaROP5g was amplified with PCR, using single-stranded cDNA obtained from the roots of banana plants subjected to salt stress as the source template, using a specific primer pair (5 -gcaccatggagatgagcgcgtcgaggt-3 and 5 -gcgactagtcaatatggagcaacctttc-3 ). The resulting MaROP5g fragment was verified with DNAMAN (available online: http://www.lynnon.com/) and compared to the genome database of DH-Pahang using BLAST [33,35].
Twelve ex vitro banana plants with uniform growth at the five-leaf stage were selected and divided into four groups for salt treatments, which were irrigated with half-strength Hoagland solution, supplemented with 300 mM NaCl for 0, 2, 4, or 6 h (n = 4 per time period), as previously described by Xu et al. [36]. Samples were frozen individually in liquid nitrogen and stored at −70 • C. Compared to the expression of 0 h in BX, the relative expression level of MaROP5g at 2, 4, and 6 h under salt stress in BX was calculated. Compared to the expression of 0 h in FJ, the relative expression level of MaROP5g at 2, 4, and 6 h under salt stress in FJ banana plants was calculated.

Subcellular Localization of MaROP5g
The ORF of MaROP5g was digested with the restriction enzymes Nco I and Spe I and inserted into a pCAMBIA1304-GFP expression vector to generate a MaROP5g-GFP fusion protein, under the control of the cauliflower mosaic virus (CaMV) 35S promoter. The recombinant pCAMBIA1304-MaROP5g-GFP plasmid was transferred to Agrobacterium tumefaciens strain LBA4404, and used to transform A. thaliana through a floral-dip method [57]. Root tips (3-5 mm) of A. thaliana seedlings (5-day old) with a stable expression of MaROP5g-GFP were incubated in 1 mL 1/2 Murashige and Skoog (MS) medium containing 10 µg FM4-64 (Invitrogen, Carlsbad, CA, USA) for 5 min at 25 • C, according to the Riqal et al. [58] methods. The GFP (488 nm emission filter) and FM4-64 (543 nm emission filter) signals were visualized using confocal laser scanning microscopy (CLSM; Nikon, A1, Tokyo, Japan). According to the Protein Subcellular Localization Prediction Tool (PSORT) software (available online: www.genscript.com/psort.html) prediction, the XXRR-like motif in the N-terminus of MaROP5g protein was identified as a membrane retention.

Plant Transformation and Generation of Transgenic Plants
The ORF of MaROP5g was digested with the restriction enzymes Nco I and Spe I, and inserted into a pCAMBIA1304 vector. The recombinant pCAMBIA1304-MaROP5g plasmid was transferred to A. tumefaciens strain LBA4404 [57]. Transgenic A. thaliana plants were generated using the floral dip-mediated infiltration method [57]. Seeds from T 0 transgenic plants were plated in kanamycin selection medium (50 mg·L −1 ). The homozygous T 3 lines were used for further functional investigation of MaROP5g.

Southern Blot Analyses
Genomic DNA isolated from the T 3 generation kanamycin-resistant transgenic lines was digested with the EcoR I restriction enzyme. A 436 bp region of MaROP5g was amplified by PCR, using a pair of specific oligo primers (5 -gtggtggatggtaacacagtta-3 and 5 -aacctttctgttgcttttttttc-3 ). Based on this sequence, we prepared a hybridization probe for use with DIG-dUTP (Roche Applied Science, Mannheim, Germany), following the manufacturer's instructions. After hybridization, the HyBond N + membrane (Amersham) was washed and exposed to X-ray film (Kodak BioMax MS, Kodak Eastman, Rochester, NY, USA), following the method described by Miao et al. [59].

Salt Stress Treatments in WT and Transgenic Plants
The seeds of both transgenic and WT A. thaliana (Columbia ecotype; control) were first vernalized for 2 days at 4 • C in the dark, and surface sterilization in 75% ethanol for 10 min, prior to germination on half-strength MS medium or directly in soil. These A. thaliana plants were maintained at 22 • C with 70% humidity and a 16 h light/8 h dark cycle (Sylvania GRO LUX fluorescent lamps; Utrecht, The Netherlands). To analyze A. thaliana phenotypes in early seedlings under normal conditions, four day-old seedlings were transferred to 1/2 MS medium for 15 days, photos were taken, and root lengths were measured. To test salt stress tolerance in early seedlings, four day-old seedlings were transferred to either 1/2 MS or 1/2 MS supplemented with 100-200 mM NaCl for 15 days, after which photos were taken and the root length was measured. To test salt stress tolerance in adult plants, 4-week-old A. thaliana plants were irrigated with 350 mM NaCl for 15 days, and then photos were taken and survival rates were assessed (leaves fall and roots rot were identified as death). To measure the expression of SOS-and calcium-signal pathway genes in the WT and transgenic lines, 15-day-old seedlings were transferred to 1/2 MS supplemented with 350 mM NaCl for up to 10 h. Whole leaves were used to quantify relative gene expression, using qRT-PCR (see Table S5 for primer sequences).

Measurement of IL and MDA Content
Four-week-old A. thaliana plants were irrigated with 350 mM NaCl for 15 days and leaf samples were collected to examine IL and MDA. IL was detected according to the method described by Xu et al. [36]. Leaf samples were cut into strips and incubated in 10 mL of distilled water at 25 • C for 8 h. The initial conductivity (C1) was determined with a conductivity meter (DDBJ-350). The samples were then boiled for 10 min to yield complete IL. After cooling down, the electrolyte conductivity (C2) was measured. IL was calculated according to the equation: IL (%) = C1/C2:100. MDA content was measured according to the thiobarbituric acid colorimetric method, as described by Xu et al. [36].

Ca 2+ , Na + , and K + Concentrations
We irrigated 4-week-old WT and transgenic plants with 350 mM NaCl for 15 days. We then collected the plant roots and washed them with ultrapure water. Plant roots were heated to 105 • C for 10 min and then dried at 80 • C for 48 h. We dissolved 50 mg of each dried sample in 6 mL nitric acid and 2 mL H 2 O 2 (30%), and then heated the solution to 180 • C for 15 min. The digested samples were diluted to a total volume of 50 mL with ultrapure water, transferred to clean tubes, and analyzed with atomic absorption spectroscopy (Analyst400, Perkin Elmer, Waltham, MA, USA).

Statistical Analysis
Three biological replicates for each sample were performed. Statistical analyses were performed using SPSS 19.0 (Chicago, IL, USA). We used analyses of variance (ANOVAs) to compare the significance of differences based on Dunnett's tests or Student's t tests. Specifically, Dunnett's tests were used to compare between WT and each overexpression line, while Student's t tests were used to compare between control and NaCl treatment. p < 0.05 was considered a significant level and p < 0.01 an extremely significant level.

Conclusions
In this study, for the first time for banana, we identified 17 MaROP genes in the M. acuminata genome, and classified these genes into three groups (II-IV) based on phylogeny, gene structure, and conserved protein motifs. The expression patterns of the MaROP genes in response to abiotic stress as reported here may shed light on the possible involvement of these genes in the regulation of abiotic stress signaling pathways. Of particular interest was MaROP5g, the overexpression of which increased the plant's tolerance for salt stress not only by maintaining a healthy growth status, but also by reducing membrane injury and improving ion distribution. Our results lay a foundation for genetic improvements in the banana plant, increasing resistance to various abiotic stressors, particularly salt. It is necessary to point out that these conclusions were drawn from the heterologous expression of banana MaROP5g in A. thaliana as a model plant system, which may or may not be valid in other plant systems. Further studies are required to characterize the function of MaROP5g in banana.