Genome-wide analysis of ATP binding cassette (ABC) transporters in tomato

ATP binding cassette (ABC) transporters are proteins that actively mediate the transport of a wide range of molecules, such as organic acids, metal ions, phytohormones and secondary metabolites. Therefore, ABC transporters must play indispensable roles in growth and development of tomato, including fruit development. Most ABC transporters have transmembrane domains (TMDs) and belong to the ABC protein family, which includes not only ABC transporters but also soluble ABC proteins lacking TMDs. In this study, we performed a genome-wide identification and expression analysis of genes encoding ABC proteins in tomato (Solanum lycopersicum), which is a valuable horticultural crop and a model plant for studying fleshy fruits. In the tomato genome, a total of 154 genes putatively encoding ABC transporters, including 9 ABCAs, 29 ABCBs, 26 ABCCs, 2 ABCDs, 2 ABCEs, 6 ABCFs, 70 ABCGs and 10 ABCIs, were identified. Gene expression data from the eFP Browser and reverse transcription-semi-quantitative PCR analysis revealed their tissue-specific and development-specific expression profiles. This work suggests physiological roles of ABC transporters in tomato and provides fundamental information for future studies of ABC transporters not only in tomato but also in other Solanaceae species.


Introduction
ATP binding cassette (ABC) proteins are proteins harboring an ATP binding domain, called nucleotide binding domain or fold (NBD/NBF), which contains highly conserved motifs, such as the Walker A and Walker B motifs, the ABC signature, the H loop and the Q loop [1]. ABC proteins are universally found in all organisms, including fungi, plants and animals [2]. Some members of the ABC proteins are soluble proteins and do not contain any transmembrane domain (TMD). The ABC proteins harboring TMDs are called ABC transporters and function as ATP-driven primary transporters for active transport of various molecules [3]. A typical functional ABC transporter contains 2 NBDs and 2 TMDs. The two NBDs synergistically bind and hydrolyze ATP to generate energy, which eventually causes conformational changes in the a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 TMDs to create a pore for substrate transport, whiles the TMDs serve as a pathway for unidirectional transport of the substrate [1]. ABC transporters harboring two TMDs and two NBDs are called full-size ABC transporters. On the other hand, ABC transporters harboring only one TMD and one NBD are called half-size. ABC transporters encoded by four genes, two for TMDs and two for NBDs are so-called quarter-size ABC transporters [3,4].
ABC transporters are grouped into eight subfamilies, namely ABCA to ABCI. Plants do not have any ABCH subfamily. Generally, plants possess twice as many as ABC transporters as not in animals. It is assumed that this is due to the sessile nature of plants for growing under various biotic and abiotic stresses [5]. ABC transporters of plants are engaged in numerous functions, including secondary metabolite transport [6,7], heavy metal detoxification [8], antibiotic transport [9] and phytohormone transport [10,11]. ABC transporter counterparts in animal are also shown to function as ion channels, channel regulators [12,13] and in protein targeting [14].
A genome-wide analysis is the comprehensive identification of all genes of the respective family including their family members and organization of their information. This approach provides essential information, such as evolutionary history, diversity and relationship among genes and proteins, which serves as useful fundamental resources for further investigations. Genome-wide analyses of ABC transporters in Arabidopsis [15], rice [16], maize [17], Lotus japonicus [18], grape [19], pineapple [20], and Hevea brasiliensis [4] have already been performed. Whereas little is known about ABC transporters in Solanaceae, including tomato.
In this study, a genome wide analysis was performed to provide information of ABC proteins in tomato. A total of 154 genes putatively encoding ABC proteins were identified in tomato genome. Among these ABC proteins, 47 proteins are soluble ABC proteins lacking any TMDs, while 107 proteins contain TMDs and they are considered to function as ABC transporters. Phylogenetic analysis revealed the evolutionary relationships of tomato ABC proteins. In addition, protein structure, in silico and reverse transcription-semi-quantitative PCR gene expression analyses were performed to provide fundamental information for further ABC protein studies not only in tomato but also in other Solanaceae species.

Identification of ABC proteins in tomato
The BLAST tool of Sol Genomics Network (SGN, http://www.solgenomics.net/) [21] was used for genome-wide identification of genes encoding ABC proteins in tomato. Known ABC proteins of tomato reported by Andolfo et al. [30] and some members of the Arabidopsis ABC subfamilies [15] were used as queries for BLAST search in the tomato genome (SL3.0 and ITAG3.10) [26]. Identified proteins with at least 30% similarity to the query sequence or Evalue less than E-20 were selected. Presence of ABC signature, Walker A and Walker B motifs was confirmed by using the Conserved Domain Database of NCBI (https://www.ncbi.nlm.nih. gov/cdd/) [31]. The predicted genes encoding ABC proteins from SL3.0 of SGN were confirmed by comparing with another tomato genome database TMCSv1.2.1 from TOMA-TOMICS (http://plantomics.mind.meiji.ac.jp/tomatomics/download.php) [23,24].
Phylogenetic, in silico gene expression and protein structure analyses. Phylogenetic analysis was conducted to classify the identified ABC proteins into their respective subfamilies. Entire protein sequences of ABC proteins were aligned using the multiple sequence alignment tool of ClustalW program (http://www.genome.jp/tools/clustalw/) [32] and subjected to cluster analysis by the distance with the neighbor-joining method using MEGA6.06 software (Molecular Evolutionary Genetics Analysis, https://www.megasoftware.net/) [33]. Gene expression data of ABC proteins in various tomato tissues were obtained from the Tomato eFP Browser (http://bar.utoronto.ca/efp_tomato/cgi-bin/efpWeb.cgi) [25,26]. The Pfam web server (http://pfam.xfam.org/) [34] was used to characterize the topology of ABC proteins comprising TMD and NBD.
Plant materials. Tomato (Solanum lycopersicum) 'Micro-Tom' was used for gene expression analysis. The Micro-Tom strain used in this study was obtained from the National Bioresource Project (NBRP)-Tomato (http://tomato.nbrp.jp/browseSearchEn.html) with an accession number TOMJPF00001. Plants were grown in growth chamber (Biotron LPH-350S, NK Systems) adjusted to 25˚C, 16 h light/8 h dark period and 60% relative humidity. Tap water was supplied twice a week. Half concentration of Otsuka liquid fertilizer (Otsuka Chemicals Co., Ltd.) was applied weekly. Young and mature leaves, root, stem, flower, developing fruit tissues at 3, 7, 14, 21, 28 days after pollination (DAP), breaker, orange and red stages were sampled, frozen in liquid nitrogen and stored at -80˚C.
RNA extraction and RT-semi-quantitative PCR (RT-sqPCR) expression analysis. Extraction of total RNA from developing fruits at 14 and 21 DAP was performed using the RNA Suisui-R kit (Rizo). RNA of other tissues was isolated using TRIzol reagent (Life Technologies). PrimeScript RT reagent kit (Takara) was used to synthesize the cDNA. RT-sqPCR was conducted using SYBR Premix Ex Taq kit (Takara) and the ubiquitin gene, SlUBQ (Solyc01g056940) was used as an internal control. Primer sequences and PCR conditions are shown in S1 Table. Results and discussion

Genome-wide identification of ABC proteins in tomato
To clarify the gene family of ABC proteins in tomato, BLAST search on tomato genome database Sol Genomics Network (SGN, http://www.solgenomics.net/) [21] was performed. We searched all the tomato ABC proteins using SL3.0 of SGN database. As a result, 154 genes potentially encoding ABC proteins were found (Table 1). Phylogenetic analysis of the tomato ABC proteins was performed and the obtained phylogenetic tree is shown in Fig 1. In a previous study, Andolfo et al.
[30] identified 180 ABC proteins in the tomato genome, whiles we found 154 ABC proteins. So we compared non-overlapping candidates between our study and Andolfo et al.
[30] (S2 Table). In this study, 3 non-overlapping putative tomato ABC proteins were identified whereas 29 ABC proteins were identified only in Andolfo et al.
[30] (S2 Table). All the 3 ABC proteins identified in this study have NBDs. On the other hand, the 29 ABC proteins found only in Andolfo et al. [30] have no NBD. Thus, we concluded that the 29 candidates without NBD in Andolfo et al. [30] are not ABC proteins and may be mispredicted. Therefore, we did not include them in our list (Table 1).
In addition, since some of the genes may not be computationally annotated in SL3.0 of SGN database, we confirmed the gene prediction of SL3.0 by comparing this database with    Table). Wider research coverage on ABC transporters has caused emergence of several naming schemes. In most cases, they were named based on the mutant characteristics. This eventually resulted in assigning different names to the same subfamily or selected members with common characteristics [35]. To conform to plant and animal ABC communities, the Human Genome Organization (HUGO) nomenclature system [35] was adopted to designate all putatively ABC proteins into their diverse subfamilies (Fig 1). A unified ABC nomenclature proposed by Verrier et al. [35] was also used to assign ABCA-ABCG and ABCI to all the eight subfamilies ( Table 1).
The 154 ABC proteins identified in the tomato genome were grouped into 9 ABCAs, 29 ABCBs, 26 ABCCs, 2 ABCDs, 2 ABCEs, 6 ABCFs, 70 ABCGs and 10 ABCIs (Table 1, Fig 1). The most abundant subfamily members were ABCB, ABCC and ABCG; while ABCD and ABCE were the least abundant. This characteristic is similar to the distribution of ABC proteins in human [36] and other plants, such as Arabidopsis [15], rice [16], L. japonica [18] and H. brasiliensis [4]. At least one EST in the SGN database (http://www.solgenomics.net/) [21] was found for 78 genes. The reason for the absence of ESTs for the 69 genes could be that they are either expressed only under certain conditions or in specific cell types. Alternatively, they could represent pseudogenes as suggested in genome-wide analysis of tomato aquaporins and sugar transporters [37,38]. A typical full-size of ABC protein has >1,200 amino acid residues [39]. The sizes of the 154 ABC proteins of tomato ranged from 50 to over 1,910 amino acid residues, although all of them possess at least one NBD as shown in Table 1. Some of the tomato ABC proteins with shorter sequences might be pseudogene or misannotation as suggested in the genome-wide analysis of tomato aquaporins and sugar transporters [37,38]. Among the 154 tomato ABC proteins, 47 members are lacking a TMD and are considered as soluble ABC proteins (Table 1). On the other hand, the other 107 members possess TMDs and are considered as ABC transporters.
One of the unique features of ABC proteins is their topological diversity. Structural orientation and conserved domains for each protein predicted by the Pfam web server is shown in Table 1. Fifty-four ABC proteins are full-size proteins possessing (TMD-NBD)x2. Among these members, 32 exhibit a forward, while 22 have a reverse topology orientations. Fifty-three ABC proteins were half-size having (TMD-NBD)x1 or (NBD-TMD)x1. Among the half-size ABC proteins, 18 exhibit a forward and 35 a reverse domain orientations. Forty-seven ABC proteins are considered as quarter-size ABC transporter proteins. SlABCB19 and SlABCC13 were uniquely characterized with NBD-TMD-NBD and TMD-NBD-TMD orientations, respectively. Similar topological patterns were reported in ABC proteins of rice [16], maize [17] and L. japonica [18]. Such characteristics might have resulted from gene duplication or evolved to render specicific physiological functions [40].
The tomato ABC protein subfamilies ABCA subfamily. The plant ABCA subfamily is made up of one full-size ABCA and several half-size ABCAs. In Arabidopsis, AtABCA1, also known as ABC one homologue (AOH), is the only full-size ABCA protein and is the largest ABC protein, consisting of 1,882 amino acid residues [15,16]. The remaining are half-size ABCAs are also called ABC two homologues (ATH). In tomato genome, 9 members of the ABCA subfamily were found (Table 1, Fig 2). SlABCA1 was the only full-size ABCA and the largest ABC protein identified, consisting of 1,910 amino acids residues (Table 1). On the other hand, 6 half-size and 2 quarter-size ABCAs were found in tomato genome. A major feature of the ABCA subfamily is the presence of one AOH full-size ABCA in dicots, including tomato (Table 1), Arabidopsis [15], L. japonicas [18] and grape [19], that so far has not been identified in monocots, such as rice [16] and maize [17]. This suggests that the function of this full-size ABCA is specific to dicots.
The functions of ABCAs in plants are currently almost unknown, although mammalian ABCAs have been shown to be involved in numerous functions, such as lipid metabolism, cholesterol homeostasis, intracellular trafficking, pulmonary surfactant secretion and retinal transport [41]. AtABCA1 was reported to be related in pollen germination, seed germination and seed maturation [18,19]. Transcriptome analysis in Arabidopsis roots has revealed that AtATH14 and AtATH15 expressions are responsive to salt stress [42]. Among the 9 SlABCAs, ESTs of 5 members were available. The gene expression profiles from the eFP Browser revealed that SlABCA1 and SlABCA2 are preferentially expressed in the root (Table 1) and they might be involved in secretion activity of roots. SlABCA4-7 are expressed specifically in the flower, suggesting a specific functions in floral organs (Table 1).
ABCB subfamily. The ABCB subfamily is the second largest subfamily. Full-size ABCBs are known as multidrug resistance protein (MDR) or P-glycoprotein (PGP) and the half-size ABCBs are characterized with names such as transporter associated with antigen processing (TAP), ABC transporter of mitochondria (ATM) and lipid A-like exporter putative (LLP) [35].
In tomato, only 10 ESTs out of 29 the SlABCBs were available ( Table 1, Fig 3). Based on the eFP Browser gene expression data, SIABCB7, SIABCB13, SIABCB14, SIABCB18, SIABCB20, SIABCB21, SIABCB24, SIABCB25 and SIABCB29 are ubiquitously expressed in all organs and tissues (Table 1), suggesting their responsibilities for basic cellular maintenance. Most of SlABCBs are highly expressed in the root. This may suggest an involvements of these SlABCBs in ion and heavy metal transports in roots.
In the tomato genome, 26 members of the ABCC subfamily were found and this comprises 12 full-size, 6 half-size and 8 quarter-size ABCCs. SlABCC13 shows a unique protein structure, i.e. TMD-NBD-TMD (Table 1, Fig 4), however as for the non-typical ABCBs this might reflect a prediction error for the CDS or the presence of a pseudogene. SlABCC18 shows reverse orientation (NBD-TMD), which is different from other SlABCCs (TMD-NBD). ESTs for 11 ABCCs were available ( Table 1). The gene expression profile of the tomato eFP Browser shows that SlABCC1, SlABCC7, SlABCC10, SlABCC11, SlABCC13, SlABCC19, SlABCC20 and SlABCC21 are preferentially expressed in the later stages of fruit development (Table 1). These SlABCCs might play important roles in fruit ripening, such as chlorophyll degradation and secondary metabolite accumulation in the vacuole.
ABCD subfamily. ABCDs are also known as peroxisomal membrane proteins (PMPs) and are localized in the peroxisomal membrane [70,71]. In humans, they are exclusively known to be half-size proteins with TMD-NBD orientation, whereas, in plants, both half-and full-size ABC proteins exist [15]. AtABCD1 is implicated in benzoic (BA) synthesis [72], transport of 12-oxophytodienoic acid (OPDA) [73] and jasmonic acids (JA) [74]. The AtABCD1 ABC transporters in tomato mutant is impaired in seed germination [75] and fertility [76]. The tomato genome contains one full-size and one half size ABCDs were found (Table 1, Fig 5). The gene expression profile of the tomato eFP Browser shows constitutive gene expression of both SlABCDs (Table 1). It is likely that these transporters exhibit similar functions as their Arabidopsis counterparts and that they are involved in peroxisomal import of long chain fatty acids.
ABCE subfamily. ABCEs, also called RNase L inhibitor (RLI), possess an N-terminal Fe-S domain, which interacts with nucleic acids [30]. All ABCE subfamily members are soluble ABC proteins harboring two conserved NBDs (NBD-NBD) [17]. In humans, only one ABCE exists and it is involved in ribosome biogenesis and control of translation [77]. There are 3 ABCEs present in Arabidopsis and two each in rice [16], maize [17], grape [19], L. japonicas [18], H. brasiliensis [4] and also in tomato (Table 1, Fig 5). In Arabidopsis, AtABCE1 and ABCE2 are involved in RNA interference (RNAi) regulation [78,79]. Among the two tomato SlABCEs, only one EST of SlABCE1 was available ( Table 1). The tomato eFP Browser revealed that both SlABCE1 and SlABCE2 are expressed constitutively in all organs and tissues (Table 1) and may play roles in ribosome biogenesis, control of translation and gene silencing regulation.
ABCF subfamily. ABCFs are also called general control non-repressible homologs (GCN). The ABCF subfamily is similar to the ABCE subfamily [17], because ABCFs are also soluble ABC proteins containing two fused NBDs (NBD-NBD). In yeast and humans, ABCFs are involved in gene expression regulation [16,80]. In Arabidopsis, 5 ABCFs are present and AtABCF3 is implicated in root growth [81]. In tomato, 6 ABCFs were identified and ESTs were available for 5 ABCFs (Table 1, Fig 5). The Tomato eFP Browser showed constitutive expressions for all 6 SlABCFs (Table 1).
ABCG subfamily. The ABCG subfamily is the largest subfamily in plants while only 5 ABCGs are present in humans [17]. The ABCG subfamily is made up of full-size and half-size ABC proteins, also called pleiotropic drug resistance (PDR) or white-brown complex (WBC), respectively [35]. All full-size and half-size ABCGs have two, respectively one NBD-TMD, respectively, and function as ABC transporters. In the tomato genome, 70 ABCGs were found, which are made up of 22 full-size, 32 half-size and 16 quarter-size ABC proteins (Table 1). This number is larger than the 44 ABCGs reported for Arabidopsis [15]. In humans, ABCGs function as transporters of cholesterol, urate, haem, and other pharmaceutical compounds [82]. On the other hand, in plants, ABCGs have been reported to transport various phytohormones, including abscisic acid (ABA), cytokinin, strigolactone and auxin derivatives [10].
One of the most widely studied ABC protein subfamily in plants are the full-size ABCGs, also called PDRs. A detailed review on plant full-size ABCGs is available [83,84] and a highlight on their functions is shown in Fig 6. The subcellular localization of full-size ABCGs is the plasma membrane [84]. Full-size ABCGs of Arabidopsis AtABCG32 [85], rice OsABCG31 [86], barley HvABCG31 [86] are involved in cuticle formation. The N. plumbaginifolia NpPDR1 [87] and duckweed SpTUR2 are known to participate in sclareol transport [88].
Half-size ABCGs are also called WBCs, have been reported to be localized in the plasma membrane, mitochondrial membrane, chloroplast membrane and cytoplasm [17]. The physiological roles of half-size ABCGs are summarized in Fig 7. In Arabidopsis, half-size ABCGs, i.e. AtABCG11-13 are implicated in cuticle formation [89][90][91]. On the other hand, AtABCG19 confers kanamycin resistance [9]. AtABCG25 has been reported to act as an ABA exporter [92] and AtABCG26 is involved in pollen development [93]. In cotton, GhWBC1 is involved in cotton yarn expansion [94].
The tomato eFP browser shows specific expressions of SlABCG12, SlABCG16, SlABCG31, SlABCG32, SlABCG44, SlABCG45, SlABCG51, SlABCG52, SlABCG55 and SlABCG58 (Table 1), suggesting their importance in root. SlABCG25, SlABCG27, SlABCG29, SlABCG30, SlABCG43, SlABCG65, SlABCG68 and SlABCG70 are expressed specifically in bud. Interestingly, only SlABCG59, which encodes a quarter-size ABCG, shows specific expression in mature fruit, although other SlABCGs are also expressed in fruits. Although we cannot guess the function of SlABCG59, it may play an important roles in tomato fruit maturation.
ABCI subfamily. ABCIs are also called non-intrinsic ABC proteins (NAPs). ABCIs are soluble ABC proteins possessing a single ATP binding domain [35]. In Arabidopsis, AtABCI1 and AtABCI2 are reported to be involved in cytochrome c maturation (CCM) [95]. AtABCI6-8 are implicated in biosynthesis of Fe/S cluster [96,97]. AtABCI13-15 are responsible for plastid lipid formation [97]. On the other hand, AtABCI16 and AtABCI17 confer tolerance to aluminum [8]. In the tomato genome, 10 SlABCIs have been identified and ESTs for 8 SlABCIs were available (Table 1, Fig 8). The gene expression profiles from the tomato eFP Browser showed that SlABCI4, SlABC16 and SlABC18 are constitutively expressed in roots and floral organs, respectively, and SlABCI5, SlABCI6, SlABCI9 and SlABCI10 in developing fruits (Table 1), suggesting their specific functions in these organs and tissues.
https://doi.org/10.1371/journal.pone.0200854.g006 sqPCR (Fig 9). These genes were chosen because their full length cDNA sequences were available in TOMATOMICS database (http://plantomics.mind.meiji.ac.jp/tomatomics/). Therefore, we requested for their full length cDNA clones from National Bioresource Project (NBRP)-Tomato (http://tomato.nbrp.jp/indexEn.html) to sequence and then performed RT-sqPCR to identify their expression patterns. Gene expression was detected in various organs of 'MicroTom', i.e. leaf, stem, root, flower and developing fruits. In addition, to obtain a detailed gene expression profile in fruits, gene expressions in fruit peel and flesh at 10 DAP, breaker and red stages were investigated. Although most SlABCs were ubiquitous expressed, some SlABCs exhibited a characteristic gene expression patterns (Fig 9).
SlABCB4 showed ubiquitous expression, but its transcript level was lower in mature fruits (Fig 9). The closest orthologue of SlABCB4 in Arabidopsis is AtACB19, and has been reported to transports auxin [98]. This suggests that SlABCB4 might be responsible for auxin transport in various organs of tomato. SlABCC11 expression was high in mature leaf and fruits after 21 DAP (Fig 9). Although the function of SlABCC11 is unclear because no close orthologue of Arabidopsis exists (Fig 3), it may play important roles in the later part of tomato fruit development.
Functions of half-size SlABCGs, SlABCG7, SlABCG8, SlABCG9, SlABCG12, SlABCG13, SlABCG17, SlABCG22 and SlABCG28 are unclear, because no characterized orthologue exists (Fig 7). SlABCG7, SlABCG8, SlABCG9, SlABCG12, SlABCG13, SlABCG17, SlABCG22 and SlABCG28 showed different expression patterns and SlABCG9, SlABCG13, SlABCG17, SlABCG22 and SlABCG28 showed relatively higher expression levels in fruits (Fig 9), suggesting that they may play some their roles in fruit development and/or ripening. SlABCG36, which encode a full-size SlABCG, showed ubiquitous expression in all organs (Fig 9). SlABCG36 is likely to transport metabolites involved in cuticle formation, because its closest orthologue of Arabidopsis, AtABCG32 is responsible for cuticle formation (Fig 6) [85]. Therefore we expected high SlABCG36 expression in fruit peel. However, the differences in SlABCG36 expressions between in fruit peel and flesh were not pronounced, although it was slightly higher in the peel than in flesh of red fruit (Fig 9).

Conclusion
This study revealed the presence of 154 putative ABC proteins in the tomato genome. Based on the phylogenetic analysis, the ABC proteins were grouped into their respective subfamilies, ABCA through to ABCI, except ABCH. Members of ABCG, ABCB and ABCC subfamilies were the most abundant, whiles ABCD and ABCE subfamilies were less abundant. Among the 154 tomato ABC proteins, 47 members are soluble ABC proteins, while 107 members encode for ABC transporters with TMDs. As far as we know, this study is the only genome-wide analysis of ABC proteins in the Solanaceae species. In this study, we provided the fundamental and exhaustive information about tomato ABC proteins, i.e. the list of all ABC proteins in tomato with their locus numbers (gene IDs), protein topology, best hit ESTs, gene expression data (Table 1) and phylogenetic trees of subfamily members and orthologues in other plants, showing the reported physiological functions (Figs 2-8). This information is indispensable for further studies of ABC proteins not only in tomato but also in other Solanaceae species. We hope this study will be useful to many researchers studying plant ABC proteins.  Table. Comparison of genes putatively encoding ABC proteins in two tomato genome databases, SL3.0 and ITAG3.10 from Sol Genomics Network and TMCSv1.2.1 from TOMATOMICS. Genes putatively encoding ABC proteins in TMCSv1.2.1 from TOMA-TOMICS (http://plantomics.mind.meiji.ac.jp/tomatomics/download.php) were obtained by blasting using the protein sequences (Table 1)