Genome-wide identi cation and expression analysis of tomato ADK gene family during development and stress

Adenylate kinase (ADK) is widely distributed in organisms and plays an important role in cellular energy homeostasis. In plants, ADK has important functions in plant growth and development regulation as well as in adaptation to the environment. However, little information is available about the ADK genes in tomato (Solanum lycopersicum), an important economic crop. To investigate the characteristics and functions of ADK genes in tomato, a total of 11 ADK genes were identified and named according to their chromosomal locations. The ADK family in Arabidopsis, tomato, potato, and rice was divided into six groups, and motif analysis revealed that each SlADK protein contained five to eight conserved motifs. A total of 4 to 19 exons were identified in tomato ADK gene family members, and interestingly, most members possessed 4 exons. Several stress response elements were identified in the promoter regions of SlADKs. The 11 SlADKs were randomly distributed on 9 of the 12 tomato chromosomes. Three duplication events were observed between tomato chromosomes, and a high degree of conservation of synteny was demonstrated between tomato and potato. The online TomExpress platform prediction revealed that SlADKs were expressed in various tissues and organs, basically consistent with the data obtained from real-time quantitative PCR (qPCR). The qPCR verification was also performed to determine the expression level of SlADKs and demonstrated that the genes responded to multiple abiotic stresses, such as drought, salt, and cold. Besides, the qPCR results showed that SlADK transcription was responsive to most of the applied hormone treatment. For correlation network analysis under 44 global conditions, the results showed that the number of 17, 3, 4, and 6 coexpressed genes matched with SlADK5, 8, 9, and 11, respectively. For specific gene function analysis, expression of SlADK10 was inhibited using virus-induced gene silencing (VIGS). Compared to wild-type plants, plants with silenced SlADK10 gene had poor drought resistance, indicating SlADK10 regulated drought tolerance of tomato positively. In summary, the information provided in the present study will be helpful to understand the evolutionary relationship and their roles of tomato ADK gene family in further research.

To investigate the characteristics and functions of ADK genes in tomato, a total of 11 ADK genes were identi ed and named according to their chromosomal locations. The ADK family in Arabidopsis tomato, potato and rice was divided into six groups and motif analysis revealed that each SlADK protein contained ve to eight conserved motifs. Sequence analysis revealed 4-19 exons in all SlADKs and most members possessed four. Cis-element analysis inferred that several stress response elements were found on the promoters of SlADKs. The 11 SlADKs were randomly distributed on nine of the 12 tomato chromosomes. Three duplication events were observed between tomato chromosome, and a high degree of conservation of synteny was found between tomato and potato. The online TomExpress platform prediction revealed that SlADKs were expressed in various tissues and organs, basically consistent with the data obtained from real-time quantitative PCR (qPCR). The qPCR veri cation was also used to determine the expression level of SlADKs and demonstrated that the genes responded to multiple abiotic stresses, such as drought, salt and cold. Besides, the qPCR results showed that SlADK transcription was responsive to most of the applied hormone treatment: methyl jasmonate, ethylene, salicylic acid, indole 3acetic acid and abscisic acid. Notably, SlADK2 and 4 exhibited signi cant changes under multiple stress treatments. Furthermore, correlation networks analysis revealed co-expressed genes between SlADKs and other tomato functional genes.
Conclusions: These results provide valuable information for clarifying the evolutionary relationship of the tomato ADK family and in aiding functional characterization of SlADKs in further research.

Background
The monophosphate of adenosine, also referred to as adenylate (AMP), is one of the four main mononucleotides making up ribonucleic acid in cells. Moreover, the formation of AMP is often accompanied by the release of energy in organisms [1]. Therefore, adenylate metabolism is an important part of primary metabolism and the change in adenylate content is the main factor affecting cell metabolism [2]. There are three important adenylate forms in organisms: AMP, adenosine diphosphate (ADP) and adenosine tri-phosphate (ATP). The ratio of AMP, ADP and ATP determines the energy charge ratio and carbohydrate metabolism, which directly affects plant growth and development as well as the ability to resist stress [3,4].
Adenylate kinase (ADK, EC 2.7.4.3) is a ubiquitous and abundant enzyme found in virtually all living organisms [5]. It catalyzes a reversible transphosphorylation reaction (ATP+AMP ↔2ADP) and is considered to be a crucial enzyme in maintaining energy metabolism and the pool sizes of various adenylates at equilibrium [6,7]. Usually, ADKs have three domains: a large central CORE domain, nucleoside monophosphate binding domain and an ATP-binding domain [8,9]. In plants, ADK has been found in a wide range of species with different subcellular locations, such as cytosol, mitochondria, and plastids, and its enzyme activity demonstrated in the cases of maize, rice, and potato [10][11][12][13]. Potato is a solanaceous plants abundant in starch, and suppression of StADK expression in potato plastids enhanced content of adenylic acid and signi cantly improved starch production [12]. Arabidopsis thaliana is a good model plant for studying plant growth and development. In Arabidopsis, T-DNA insertion mutants disrupted in one ADK gene At2g37250 showed increased amino acid levels and enhanced root growth [13]. Subsequently, another relevant study revealed that disruption of Arabidopsis ADK gene At5g47840 leads to loss of chloroplast integrity, causing a bleached phenotype from early embryo to seedling development [14]. Very recently, researchers found that ADK3 in the chloroplasts of a green alga could interact with the chloroplast glyceraldehyde-3-phosphate dehydrogenase to form a stable complex, and the interaction might be a mechanism to regulate the crucial ATP: NADPH ratio in the Calvin-Benson cycle [15].
Besides regulation of growth and development, ADK is also widely involved in abiotic stress responses in plants. When roots and stems of maize were treated with solutions of two different ratios of Ca 2+ /Na + , results showed that ADK content had an important relationship with plants salt stress [16]. In tomato, microarray analysis of genes revealed that an ADK homolog (SGN-U214214) was repressed in salt-treated tissues [17]. Other microarray data revealed that ADK gene (SGN-U232826) expression in drought-tolerant tomato was induced by drought stress [18]. Using pea seeds as a model to research the adenylate balance in dehydrating and imbibing seeds driven by mitochondria, results indicated that ADK played a crucial role in building and later using the huge AMP pool which appears as a signature of the dry state in seeds [19].
Tomato is one of the most important agricultural products worldwide, as well as an important model for studying eshy fruit development and ripening [20]. Currently, the ADK family in tomato has not been studied systematically. Due to the importance of the ADK genes in regulating plant growth and stress resistance, it would be of interest to make a systematic investigation of the ADK family in tomato. In the present study, we used bioinformatics methods to identify ADK genes from the tomato genome, and analyze the phylogenetic relationships, sequence features, gene location, chromosomal locations, evolutionary relationship and cis-elements in promoters. The comprehensive transcriptomic pro ling of the ADK family in various tissues and organs of tomato during different developmental stages were carried out using the online TomExpress platform. Also, the dynamic expression patterns of the ADK family in response to different plant hormones methyl jasmonate (MeJA), ethylene (Eth), salicylic acid (SA), indole 3-acetic acid (IAA) and abscisic acid (ABA) and abiotic stresses (drought, cold and salt stress) were systematically studied in detail using quantitative real-time PCR (qRT-PCR). Combined expression data of SlADKs in tomato different developmental stages and stress treatments groups obtained from TomExpress platform, co-expression and correlation networks between SlADKs and other tomato functional genes were further investigated. In brief, the present results will provide useful information for further functional and regulation mechanism investigations of the ADK family in tomato.

Plant materials and growth conditions
Solanum lycopersicum cv. Micro-Tom was used as wild type plant material. The plant samples used in this study were collected from Key Laboratory of Plant Hormones and Development Regulation of Chongqing, School of Life Sciences, Chongqing University, Chongqing, China. Collection of plant materials complied with the institutional, national and international guidelines. No speci c permits were required. To assess potential roles of SlADK family genes throughout tomato development by experiments, the tissues of roots (R); stems (S); leaves (L); bud (B); owers (F); immature green fruit (IMG); mature green fruit (MG); breaker stage fruit (BR); orange fruit (O); red fruit (R) and over ripe fruit (OR) were collected from wild type tomato, frozen in liquid nitrogen immediately and stored at -80 °C for RNA extraction lately. For stress and hormone treatment, germinated seeds were cultivated in greenhouse with suitable conditions: 16/8 h light/dark cycle, 25/18°C day/night temperature, 80% relative humidity and 250 μmol/m 2 /s light intensity. The seeds and subsequently growing plants were watered daily. In addition, the plants were irrigated with nutrient solution once per week. One-month old tomato plants with good growth status were selected and transplanted from soil to hydroponic box for hydroponic culture (pH 5.8), and the Hogland nutrient solution was renewed regularly. Hydroponics were adapted for 5-6 days to eliminate the damage caused by transplantation, then abiotic stresses and hormone treatments experiments were performed with these plants.

Identi cation of tomato ADK genes
The ADK genes in tomato were comprehensively identi ed by two methods. The rst method is using 'adenylate kinase' as keywords inputted in the annotation text to search unigene families in the SOL Genomics Network (http://solgenomics.net) [21]. The second method is screening ADK family genes with bio-linux system. First, we downloaded tomato genome information les in Ensembl database (http://plants. ensembl.org/index.html), including the gff, cds, pep, fasta su x les. Then, gunzipping these les and acquiring corresponding relationship between gene and mRNA. Subsequently, the hidden markov model (HMM) le corresponding to the ADK domain (PF00406) was downloaded from Pfam protein family database (http://pfam.sanger.ac.uk/) [22]. Hmmsearch command was then used to search the ADK genes from tomato genome database. The default parameters were adopted and the cutoff value was set to 0.01 [23]. All candidate genes that may contain ADK domain based on HMMER results were further examined by con rming the existence of the ADK core sequences using the Pfam and SMART program [24]. In a word, the tomato ADK gene family was identi ed with two above mentioned methods, then the identi ed members were further veri ed combined with gene description in SOL Genomics Network and gene sequence blast in National Center for Biotechnology Information (NCBI) database.

Characteristics, phylogenetic relationship and sequence analysis of ADK proteins
Among other species, twelve ADK family genes in potato were also identi ed according to the above methods. In rice, seven ADK family genes has been screened in China Rice Data Center (http://www.ricedata.cn/gene/) and RGAP (http://rice.plantbiology.msu.edu/). Also, seven adenylate kinase isoforms has been found with high sequence homology in Arabidopsis genome [14]. The fulllength ADK proteins in Arabidopsis and identi ed in tomato, potato, and rice were aligned using ClustalW. Phylogenetic analysis of ADK proteins was performed using MEGA 7.0.26 with the neighbor-joining (NJ) method based on the Poisson model [25], the bootstrap method was used to test the tree with 1000 replicates, and paired deletion was performed [26]. Multiple sequences alignments of the amino acid sequences of tomato ADK proteins were analyzed by ClustalX 1.83. and DNAMAN software. The MEME online program (http://meme.nbcr.net/meme/intro.html) provides a uni ed portal for online discovery and analysis of sequence motifs representing features, such as protein interaction domains [27]. Here the MEME Suite Web server for protein sequence analysis was used to identify conserved motifs in the identi ed tomato ADK proteins, with the number of found motifs as ten and the other parameters as default values. Then sequences of the motifs were blast in Pfam (http://www.pfam.org/) and SMART (http://smart.embl.de/) database. The ADK protein sequences in tomato were analyzed by the online ProtParam tool of ExPASy (http://weB.expasy.org/protparam/) for physical and chemical characteristics.

Gene structure, chromosome location and CREs of SlADK genes
The visualization of gene structure and annotation features helps to determine the function and evolution of family genes intuitively. Physical locations of all SlADK genes on each chromosome were obtained from tomato genome database: SOL Genomics Network (http://solgenomics.net). The exon-intron structure of each SlADK was determined by aligning the full-length cDNA sequence with the genomic DNA sequence. The Gene Structure Display Server 2.0 (GSDS2.0; http://gsds.cbi.pku.edu.cn//index.php) program was used to display the gene structures on the basis of the coding sequences (exons), intron and untranslated regions (UTR) composition information [29]. With the strength of the genome annotation, MapChart software was used for mapping the genomic location and relative distances of SlADK genes in the chromosome [30]. The 1500 bp upstream sequences of the SlADK-coding sequences were retrieved from the SOL Genomics Network and then submitted to PlantCARE (http://bioinformatics. psb.ugent.be/webtools/plantcare/html/) to identify stress-and hormone-related cis-regulatory elements (CREs) [31].

Synteny analysis of SlADK genes
For analysis of inter-chromosomal relationships of SlADK genes, the gene duplication landscape was obtained using the MCScanX [32]. Each duplicate segment with SlADK genes was selected, and the syntonic map was generated using CIRCOS [33]. The putative duplicated genes were linked by the connection lines. To analysis of synteny relationships of SlADK genes between different species, the genome sequences, genome annotation information and ADK coding sequences for rice (http://www.ricedata.cn/gene/), tomato (https://solgenomics.net/) and potato (https://solgenomics.net/) were obtained. Fragment lengths in the alignment that exceeded 75% of the length were con rmed as tandem duplications. The synteny relationships between the ADK family member in tomato and these species were determined using MCScanX.
Expression data mining of tomato SlADK genes Expression patterns of identi ed tomato SlADK family genes during vegetative and reproductive development were carried out with TomExpress bioinformatics platform (http://gbf.toulouse.inra.fr/tomexpress). TomExpress provides a uni ed and standard method to judge tomato gene expression from released RNA-Seq data sets. Expression data represent normalized counts per base and mean values of multiple cultivars for each tissue and stage, as well as different biotic and abiotic stress treatment. The expression value is appropriately associated with corresponding experimental annotations. Various forms of data output such as line diagram, heat map and other graphics types have been utilized to make the web pages more user-friendly. For co-expression and correlation networks analysis, co-expression tool of TomExpress platform was used to identify genes that displayed similar or opposite expression pro les, the comprehensive visualization of co-expression results were based upon the calculation of the correlation values [34,35].

Hormone and abiotic stress treatments
One-month-old tomato plants which have been transferred to hydroponic box from the soil were subjected to hormone and abiotic stress treatments. For hormone treatment, 200 μM ethephon (Eth), 100 μM abscisic acid (ABA), 500 μM salicylic acid (SA) and 50 μM Methyl Jasmonate (MeJA) have been prepared. Then tomato leaves were sprayed with above solution, respectively. When the solution drops, stop spraying and quickly seal the leaves with transparent plastic lm for moisture retention. After treatment for 0, 0.5, 1, 2, 3, 6 and 12 h, 24 h, samples of leaves were harvested separately. Plants leaves sprayed with ddH 2 O were used as control. For salt stress treatment, salt of NaCl was added into hydroponic medium to ensure the nal salt concentration was 150 mM, it was necessary to submerge root adequately with salt solution. Samples were collected after 0, 3, 6, 9, 12, 24, 48 and 72 h treatment. For cold stress treatment, tomato plants for hydroponic culture were transferred to a cold chamber maintained at 4 ± 1 °C. Leaves were sampled at 0, 3, 6, 9, 12, 24, 48 and 72 h post treatment and untreated plants were used as controls. For drought treatment, PEG6000 was used for simulating drought. The nal concentration of PEG6000 in hydroponic medium was 12% and it was necessary to submerge root adequately with PEG6000 solution. Samples were collected after 0, 3, 6, 9, 12, 24, 48 and 72 h treatment. Three individual plants in good status were used for each treatment. After treatment, tissue of leaves in each biological replicate was collected and mixed thoroughly, then frozen in liquid nitrogen immediately and stored at -80°C.

RNA isolation and quantitative real-time PCR analysis
Total RNA was extracted using TRIzol reagent (Invitrogen, USA) according to the manufacturer's instructions. After RNA integrity detection and RNA concentration measurement, DNase I (Thermo Scienti c) was used to remove any genomic DNA. About 2 μg of total RNA from each sample was used for rst-strand cDNA synthesis following the manufacturer's protocol. DNAMAN software was used to design primers speci c to the SlADK family genes (Table S5) and SlActin (Solyc03g078400) was used as internal control. Quantitative real-time PCR was conducted on a CFX96 Touch™ Real-Time PCR Detection System (BIO-RAD, USA) using the SuperReal PreMix Plus (SYBR Green) (TIANGEN, China). Each reaction mixture was 20 μL sample volume in total contained 1.2 μL of cDNA, 1.2 μL of primers mix, 10 μL of 2×SYBR Mix Taq and 7.6 μL of sterile distilled water. The PCR ampli cation cycle was performed as follows: 95 °C for 15 min, 40 cycles at 95 °C for 10 s, and 60 °C for 30 s. Melting curve analysis was performed ranging 60 to 95 °C to verify amplicon speci city for each primer pairs. Relative expression levels of the SlADK genes were calculated using the standard curve and normalized by the control's expression. The results were displayed with heatmap to displaying gene clustering based on expression patterns.

Identi cation of the ADK family in tomato
To identify the ADK family in tomato, unigenes were searched in the SOL Genomics Network and a hidden Markov model search was used to scan probable proteins. After con rming the existence of the ADK core sequences according to the description in tomato genome database and the Pfam and SMART programs, a total of 11 ADK genes were identi ed. Based on their chromosomal locations, the genes were assigned as SlADK1-11. Gene information of each ADK gene and physicochemical properties of matched proteins were predicted with tomato gene database and ExPASy online tool, respectively. The names and IDs of genes, chromosomal and strand locations, open reading frame lengths, exon numbers, amino acid numbers, molecular weights (MWs), isoelectric points (pIs) are listed in Table 1. Predicted subcellular location of SlADK proteins are listed in Table 2. The coding sequence (CDS) and gene sequence of the SlADK family are found in Table S1. The lengths of the SlADK proteins ranged from 630 (SlADK1) to 1989 amino acids (SlADK11) and the corresponding range for MWs was 22872.10-74110.62 Da ( Table 1). The 11 SlADK genes were distributed on nine tomato chromosomes, with SlADK3 and 4 located in forward strands and the others in reverse strands. The predicted pI values of SlADKs ranged from 5.76 (SlADK1) to 8.83 (SlADK4). Prediction results of subcellular localization based on four different online tools revealed that most SlADK proteins were presumably located in mitochondria (e.g. SlADK2, 3, 4 and 6); SlADK1 and 5 may located in cytoplasmic; and SlADK8 and 11 may located in chloroplast. Interestingly, SlADK9 may located in cytoplasmic or nuclear, and SlADK10 may located in chloroplast or mitochondria.
Phylogenetic analysis and multiple sequence alignment of SlADK genes The phylogenetic relationship of tomato SlADK proteins, together with seven Arabidopsis AtADKs, twelve potato StADKs and seven rice OsADKs were examined by multiple sequence alignment with full lengths of the amino acid sequences. Then the phylogenetic analysis was conducted by MEGA7 based on the aligned results, and the neighbor-joining method was used with bootstrap replications of 1000. The ADK family was divided into six groups (Fig. 1A). SlADK6 and 10 together with two potato homologs (Sotub09g006620 and Sotub04g013920), two Arabidopsis homologs (AT2g39270 and AT2g37250) and one rice homolog (Os03t0130400) were phylogenetically distinct and formed the predicted rst branch of plant ADKs. The second branch contained two isoforms of potato (Sotub02g037180 and Sotub03g005270), one isoforms of Arabidopsis (AT3g01820), two isoforms of rice (Os08t0118900 and Os07t0412400) and two of tomato (SlADK2 and 3). The third branch contained three evolutionarily similar subbranches: SlADK1 and Sotub01g028550; SlADK4 and Sotub03g020180; and SlADK9 and Sotub08g022760. The fourth branch contained two isoforms of potato (Sotub06g024300 and Sotub11g015570), one isoforms of Arabidopsis (At5g47840), one isoforms of rice (Os08t0109300) and one of tomato (SlADK8). The fth branch contained four subbranches: At5g50370 and At5g63400, SlADK5 and Sotub03g023880; SlADK7 and Sotub05g016010; and Os12t0236400 and Os11t0312220. At last, SlADK11 together with Sotub12g00740, Os08t0288200 and AT5g35170 were formed as branch VI. Multiple sequence alignment of SlADK proteins was performed using ClustalX 1.83 and DNAMAN software (Fig. 1B). Because the amino acid sequence of SlADK11 was much longer than other proteins, the last comparison part which only contained ADK11 C-terminal amino acids is not shown in Fig. 1B.

Motif analysis
According to the amino acid sequences, the MEME web server was used to search the conserved motifs that were shared with the SlADK proteins. A total of 10 distinct conserved motifs were found and the number of the motif residues ranged within 38-50 amino acids; details of the conserved motifs are shown in Fig. 2A. Each SlADK protein contained ve to eight conserved motifs. It is worth noting that motifs 1, 2 and 4 were fundamental in the ADK domains because they were shared by all SlADKs (Fig.  2B). Among the three motifs, motif 2 has been marked in Fig. 1A because the conserved domain belonged to it. As shown in Fig. 2B, some SlADKs (e.g. SlADK1 and 9, SlADK2 and 3, and SlADK5 and 7) shared common motif compositions which supported the grouping results. In addition, motif 7 was unique to SlADK5 and 7 at the N-terminal, and motif 9 was unique to SlADK2 and 3 at the C-terminal. The speci c motifs may contribute to the functional divergence of SlADK genes. In order to better annotate function of genes, the sequence of each motif has been performed blast in pfam and SMART database, the prediction result showed that except motif 7 to10 were too short to predict, motif 1 to 6 all contained ADK domain (Table S3).
Gene structure and chromosomal location analysis of SlADKs Gene organization plays a vital role in the evolution of multiple gene families [36]. The neighbor-joining phylogenetic tree constructed with MEGA7 is shown in Fig. 3A, which was consistent with the result in Fig. 1A. Corresponding to each gene, the genomic sequence and cDNA sequence information were submitted to the Gene Structure Display Server together to show the gene structure. Among these genes, the average gene length was 1610-9218 bp. SlADK 11 had the maximum number of exons and the longest gene length, while SlADK2 was the shortest genes. Each gene had 4-19 exons and the majority of SlADKs harbored four exons (SlADK2, 3, 6 and 10) (Fig. 3A, Table 1). The result also revealed that genes closed to each other in phylogenetic tree had similar gene structure, such as SlADK2 and 3, SlADK6 and 10, SlADK5 and 7, and SlADK1 and 9. Tomato contained 12 chromosomes, the 11 SlADKs were distributed on nine of them randomly (Fig. 3B). The majority of SlADKs were located on the proximate or distal ends of the chromosomes. Chromosome 03 had the greatest number of predicted SlADKs, with three (SlADK3-5), and no SlADK existed on chromosomes 07, 10 and 11. Only one SlADK existed on each of the other eight chromosomes.

Cis-regulatory elements in SlADK promoters
To pave the way for further study of potential gene function and regulatory mechanism of the SlADK family, especially during abiotic stress responses and hormone treatment, the 1500-bp upstream sequences from the translation start sites of SlADKs were submitted to PlantCARE to detect cis-regulatory elements (CRE). After screening, CRE related to stress and hormone were retained and detailed information could be found in Table S4. Also, the location and number of representative 17 CRE (ABRE, ARE, AuxRE, Box4, CGTCA-motif, G-box, GATA-motif, LTR, MRE, MYB, MYC, P-box, TATC-box, TC-rich, TCAelement, TGA-element, TGACG-motif) were visualized on each gene with GSDS software ( Figure 3C). The upstream regulatory sequence of promoters contained multiple elements that respond to hormones (such as Box 4 and G-box, TGACG-motif) and stress signals (such as TC-rich, LTR, MYB-and MYC-binding sites), which indicated that expressions of SlADKs were associated with abiotic stresses and hormone signal transduction response. It is worth noting that from the promoter region of -1500 bp to -700 bp, cisregulatory elements of SlADK11 and 1 were less distributed and almost nonexistent, respectively.

Synteny analysis of SlADK genes
Synteny analysis of SlADK genes were conducted to investigate the duplication event occurring in the tomato ADK family (Fig. 4A). Three duplication events were observed between chr01 and chr08 (SlADK1 and SlADK9), chr02 and chr03 (SlADK2 and SlADK3), and chr04 and chr09 (SlADK6 and SlADK10), which evolved from segment duplication. Interestingly, each gene pair existed duplication events belonged to the same subfamily in the phylogenetic tree (Fig. 3A). To further infer the phylogenetic mechanisms of tomato ADK gene family, we constructed two comparative syntenic maps of tomato associated with two representative species, rice and potato (Fig. 4B). The homology between tomato and potato are closer because they both belong to solanaceae while rice belongs to gramineae. Our result revealed that ten SlADK genes showed syntenic relationship with potato genes (SlADK1 and Sotub01g028550, SlADK2 and Sotub02g037180, SlADK3 and Sotub03g005270, SlADK5 and Sotub03g023880, SlADK6 and Sotub04g013920, SlADK7 and Sotub05g016010, SlADK8 and Sotub06g024300, SlADK9 and Sotub08g022760, SlADK10 and Sotub09g006620, SlADK11 and Sotub12g007490), while no collinear relationship existed of ADK genes between rice and tomato.

Expression patterns analysis of SlADKs
Comprehensive transcriptomic pro ling of 11 SlADKs in tomato vegetative and reproductive tissues was carried out using the online TomExpress platform and associated data mining tools (http://gbf.toulouse.inra.fr/tomexpress) ( Fig. 5A and Fig. S1). Gene SlADK1 was seldom expressed in all tomato tissues. Genes SlADK5 and 10 showed higher expression in seed and root than other genes; SlADK10 showed increased expression in ower and fruit, peaking at bud in 3 mm and stage of mature green fruit (35 DPA). Expression of SlADK5 was enhanced at the onset of fruit development and reached its peak at mature green fruit (35 DPA); however, during ripening, its expression decreased until increasing again during the red fruit stage. Some genes such as ADK2 and 6 exhibited relative moderate expression in all tissues. Of particular interest, SlADK3, 8 and 9 were the most highly expressed during late fruit ripening, displaying a net up-regulation at the onset of ripening starting after the mature green stage. This pattern of expression suggests a potential role of these genes in regulating the ripening process.
To assess the potential roles of SlADKs throughout tomato development, we conducted detailed quantitative real-time PCR (qRT-PCR) to examine transcription in different tissues ( Fig. 5B and C). In nonfruit tissues, including root, stem, leaf, bud and ower, SlADK1 and 4 showed similar expression patterns. Notably, the expression levels of most SlADKs was drastically enhanced in bud compared to other tissues and expression of SlADK2 was much higher in bud. Genes SlADK2, 3, 6, 7, 8 and 10 also showed similar expression patterns, possibly indicating redundancy of gene function. Interestingly, SlADK11 expression was higher in leaves than other tissues. Additionally, expression of SlADK5 and 9 showed little difference among tissues ( Figure 4B). During the critical stages of fruit development, including immature green stage (IMG), mature green stage (MG), breaker stage (BR), orange stage (O), red ripe stage (RR) and overripe stage (OR), the expression levels of SlADK4, 7 and 10 were relatively high at BR stage and those of SlADK3 and 6 were relatively high at O stage. It was remarkable that the mRNA level of SlADK9 was signi cantly up-regulated at RR stage, and interestingly, expression of almost all detected genes was relatively low in OR stage. Due to the quite low expression abundance and ampli cation e ciency of SlADK1, 2 and 11, it was di cult to get satisfactory qPCR results for analysis of gene expression (Fig.  5C).

Expression characteristics of SlADKs under diverse abiotic stresses
To identify potential functions of SlADKs in response to different abiotic stresses, their transcript pro les were assayed under drought, salt and cold treatments (Fig. 6).
Following 3 h of treatment with PEG6000, the expression of most SlADKs was obviously up-regulated except for SlADK1, 3 and 8. Interestingly, expression of most SlADKs decreased rapidly during 9-12 h after treatment, but increased rapidly during 24-48 h. Especially, the transcript levels of SlADK1 and 11 almost linearly increased with time under PEG6000 treatment during 12-72 h. Under salt treatment, expression levels of most SlADKs were obviously down-regulated at the early stage. It should be noted that almost all SlADKs contained two expression peaks at 9 and 48 h; however, the two expression peaks for SlADK7 were at 6 and 24 h. Under cold stress, there was obvious up-and down-regulation in expression levels of SlADK3 during 12-24 h and 24-48 h, respectively. For SlADK7, expression showed no signi cant differences for all time points before 24 h of treatment, but expression rose rapidly during 24-72 h. Notably, SlADK7 exhibited unique changes under multiple stress treatments, suggesting that it may have a unique role in stress responsiveness. Moreover, the result of gene clustering showed that homologous genes always had similar expression patterns especially under PEG6000 treatment, such as SlADK2 and 3, and SlADK5 and 7 ( Fig. 3A and Fig.  6).
Expression pro les of SlADKs in response to diverse hormone treatments Previous evidence indicated that different hormones play important roles in stress signal transduction and cell responses [37][38][39]. Here, we investigated the expression pro les of SlADKs in response to Eth, IAA, ABA, SA and MeJA treatments (Fig. 7). With Eth treatment, in general, the expression of most SlADKs showed little difference, but that of SlADK1, 2 and 7 increased gradually to different levels during 0-2 h. With IAA treatment, the expression peaks of SlADK1, 2 and 4 were at 2, 0.5 and 6 h, respectively. Interestingly, SlADK11 showed decreased transcription at the early stage and almost no expression at later time points. At 0.5 h after ABA treatment, expression of SlADK5, 7 and 10 showed no signi cant differences; however, expression of SlADK3, 4, 6, 8, 9 and 11 decreased, and that of SlADK1 and 2 increased at the rst time point. Moreover, most SlADKs did not change signi cantly after 1-12 h of treatment. SlADK6 showed decreased transcription at the early stage and almost none at following time points. With SA treatment, transcription of most SlADKs was repressed in the rst 0.5 h. Expression of most SlADKs showed mild change under SA treatment, except for SlADK2, 6 and 7. Especially, SlADK2 and 7 showed similar expression patterns throughout all time points, with peak expression at 12 h. The plant regulator MeJA mediates diverse developmental processes and defense responses. It rapidly induced upregulation of SlADK4 and 2 during 1-6 h and 3-12 h time points, respectively. Expression of SlADK6 increase gradually during 1-6 h, with opposite results for expression of SlADK9 during 0-3 h. Similar to IAA treatment, SlADK11 also showed decreased transcription at the early stage and almost no expression for later time points. Hence, transcription of SlADKs was responsive to most of the applied stress treatments. Interestingly, SlADK2 and 4 exhibited signi cant changes under multiple hormone treatments, suggesting that they may have unique roles in hormone regulation. Notably, the result of gene clustering showed that the homologous genes of SlADK6 and 10 had similar expression patterns with the treatment of Eth and IAA; and SlADK8 and 11 had similar expression patterns with the treatment of Eth and ABA (Fig. 3A and  Fig. 7).
SlADKs expression patterns under stress based on RNA-seq data As introduced above, TomExpress provides a uni ed and standard method to judge tomato gene expression from released RNA-Seq data sets. Here, SlADKs expression patterns under treatment with different plant hormone and multiple hormones were analyzed (Fig. 8 and Fig. S2). In general, the expression of most SlADKs (SlADK3, 5, 6, 7 and 10) were lower in leaves (C10 to C17) than in roots (C1-C2, C4-C5, C7-C8) whether these tissues were treated with cytokinin or not. Also, the expression of SlADK1 was lower in many tissues, and even could not be detected in roots (C2, C4, C5, C7) and leaves (C10, C13, C16) under special stress conditions. With the treatment of auxin in tip of roots, the expression of SlADK7 and 10 were down-regulated signi cantly. In fruit, several SlADK genes such as SlADK3, 5, 8 and 10 displayed higher expression levels. Compared with treatment of ACC in fruit, multiple hormones treatment of ACC+ IAA could up-regulated the expression of SlADK3 and 5 (Fig. S2). Heatmap can be very convenient to show gene clustering. As shown in Fig. 8, under treatment with different plant hormone and multiple hormones, SlADK3, 4, 5, 7 and 10 showed similar expression patterns in many tissues, so were the group of SlADK2, 6 and 11 and the group of SlADK1, 8 and 9 (Fig. 8).

Co-expression and correlation networks analysis.
Based on expression data of SlADKs under 44 global conditions of different development and stress treatment derived from TomExpress platform, co-expression and correlation networks were analyzed (Table S6 and Fig. 9). The correlation values of co-expressed genes pairs were calculated and the correlation threshold was set as 0.92. Then, the pairs of co-expressed genes of which the correlation coe cient was more than |0.92| were displayed (Table S6). The results showed that SlADK5, 8, 9 and 11 possessed the number of 17, 3, 4 and 6 co-expressed genes, respectively. Among the 17 genes existed correlation with SlADK5, 16 showed positive correlation and 1 showed negative correlation; Among the 3 genes existed correlation with SlADK8, 1 showed positive correlation and 2 showed negative correlation. However, the 6 co-expressed genes of SlADK11 and the 4 co-expressed genes of SlADK9 showed only positive and negative correlation, respectively. Furthermore, these correlation data were visualized as a heatmap after a hierarchical clustering to highlight the positively and negatively correlated groups. The result showed that SlADK8 (Solyc06g065270) and 9 (Solyc08g077300) showed similar co-expression patterns, while SlADK5 (Solyc03g111200) and 11 (Solyc12g010380) showed quite different co-expression patterns (Fig. 9).

Discussion
ADK is ubiquitous in the kingdoms Animalia and Plantae and it is found in the cytosol and also in many organelles such as mitochondria and chloroplasts. So far, ADK-encoding genes have been cloned from a wide variety of plant species [13]. However, genome-wide analysis of the ADK gene family has not been pursued in tomato, a model plant for studying plant fruit ripening. In the current study, 11 SlADKs in tomato were identi ed, and designated SlADK1-11 on the basis of their chromosomal location ( Table 1). The phylogeny, motif, gene structure, chromosomal location, cis-elements and expression patterns in different tissues and under stress treatments were analyzed. Synteny analysis was also performed in genome of tomato and among several related plant species. Combined with expression data sets from TomExpress platform, co-expression and correlation networks were investigated between SlADKs and other tomato functional genes. This study provides comprehensive information on the SlADK family and will aid understanding of the function of SlADKs.
Previous research revealed seven ADK isoforms with high sequence homology in the Arabidopsis genome [14]. The lower number of ADK genes in Arabidopsis may be related to its small genome [40]. Potato belongs to solanaceae and has strong homology with tomato. In this plant, 12 ADK genes were identi ed. And also, rice belongs to gramineae which contains 7 ADK genes ( Table S2). The ADK proteins from the four plant species were classi ed into six groups, with genes in the same group showing a closer evolutionary relationship. For example, SlADK6 and 10 belonged to group I; SlADK2 and 3 belonged to group II; and SlADK1, 4 and 9 belonged to group III. It is worth noting that SlADK1, 4, and 9 were already identi ed as being closely related to UMP-CMP kinases, with highest homology to the respective three genes in Arabidopsis. Much closer evolutionary relationships existed in the same subbranch. Interestingly, each ADK family member of tomato has a potato ADK family member with high homology, such as SlADK1 with Sotub01g028550, SlADK2 with Sotub02g037180, and so on (Fig. 1A). Subcellular localization prediction showed that SlADKs were distributed in mitochondria, chloroplast and other plastids in cells, with the greatest occurrence in chloroplasts ( Table 2). This is consistent with previous reports that ADK activity in plants is mainly distributed in the chloroplast matrix and mitochondrial membrane space [41][42][43].
Motif analysis revealed a total of 10 motifs ( Fig. 2A), with motifs 1, 2, and 4 shared by all SlADKs. In addition, motif 7 was unique to SlADK5 and 7 at the N-terminal, and motif 9 was unique to SlADK2 and 3 at the C-terminal (Fig. 2B). Common motifs imply functional redundancy and the speci c motifs may contribute to the functional divergence [44]. For the evolution of multiple gene families, the model of gene organization is very important [36]. Gene structure analysis revealed 4-19 exons in each SlADK (Fig.  3A). All above mentioned genes in the same subbranch ( Fig. 1A and Fig. 3A), such as SlADK1 and 9, SlADK2 and 3, and SlADK5 and 7 shared common motif compositions (Fig. 2B) and similar gene structure (Fig. 3A). This correlation between gene structure and motif arrangement further con rmed the classi cations of the SlADKs.
17 CRE related to hormone regulation and stress response were analyzed (Table S4 and Fig. 3C). When plants are exposed to abiotic stresses such as salt, drought or low temperature, ABA-dependent andindependent pathways are simultaneously activated [45,46]. Genes involved in the ABA-dependent pathway not only induce ABA biosynthesis, but also regulate the expression of genes containing ABREs [47,48]. The ABREs mainly occurred in SlADK3 and 6 and the G-box element was mostly distributed in SlADK3 (Table S4, Fig. 3C). In the barley HVA22 gene and the Lea gene promoter, the core sequence ACGT of G-box and other regulatory sequences (CE1 and CE3) constitute an ABA response complex to facilitate the transcription strength regulation of ABA-regulated genes [49]. The MYB elements are found in the promoters of several stress-resistance genes in Arabidopsis [50]. Our results showed MYB elements distributed in all SlADKs, especially SlADK2 and 3 (Fig. 3C). The MYC element is a cis-acting element in response to drought and ABA, and exists in a variety of anti-stress gene promoters, with reports related to soybeans and Arabidopsis [51,52]. Our results revealed that MYC existed in almost all SlADKs except SlADK1 and 7, and was distributed frequently in SlADK6 (Fig. 3C).
Gene duplications play important role in the evolution of plant genome and genetic system [53]. Duplicated genes promote the generation of new genes and their corresponding new functions. Three principal evolutionary patterns contain segmental duplication, tandem duplication and transposition events, the former two patterns can often lead to gene family expansion [54,55]. Our results revealed that in tomato genome, three duplication events were observed between chromosomes, which evolved from segment duplication (Fig. 4A). Compared with other related species, except SlADK4, ten other SlADKs showed syntenic relationship with potato StADK genes (Fig. 4B). It was found that gene pairs existed collinear relationship were also closed to each other in the phylogenetic tree ( Fig. 1A and Fig. 4B). These results indicated whole-genome duplication (WGD) event occurred and SlADK4 may have new gene function.
The expression patterns of ADK genes in different tissues have been described in many species, including Arabidopsis [14] and rice [10]. In Arabidopsis, expression of AtADKs was detected in leaves, roots, and 16d-old seedlings and AtADK1-5 were much more expressed than AtADK6, while AtADK7 was at the detection limit [14]. However, there was no uniform gene expression pattern for SlADKs in tomato. Our qPCR results for SlADKs in tomato were basically consistent with those predicted by the online TomExpress platform. For example, the prediction expression peak of SlADK2 and 11 was in bud of 3 mm and leaves, respectively, which was highly consistent with the qPCR result (Fig. 5A, B). Also, it was di cult to get satisfactory qPCR results for gene expression analysis of SlADK1, 2 and 11 due to quite low expression abundance and ampli cation e ciency, which was consistent with the software prediction that these three genes had very low expression levels in fruit (Fig. 5A, C). For qPCR detection, SlADK2, 3, 6, 7, 8 and 10 also showed similar expression patterns, which hint at similarities in structures, redundancies in functions as well as shared induction mechanisms (Fig. 5B).
Abiotic stresses, such as drought, high salinity, extreme temperature and ooding are major causes of crop loss worldwide, reducing average yields for most major crop plants by more than 50% [56]. In addition to regulation of growth and development, previous study showed that ADKs are widely involved in abiotic stress response in plants [16][17][18][19]. In our study, transcript pro les of the tomato ADK family were assayed under drought, salt and cold treatments. With the increased time of PEG6000 treatment, expression of most SlADKs was up-regulated, especially of SlADK1 and 11. Microarray data revealed that the expression of an ADK gene (SGN-U232826) in drought-tolerant tomato was induced by drought stress [18]. The sequence of SGN-U232826 was consistent with the SlADK10 identi ed in our study, which was induced during 24-72 h following PEG6000 treatment (Fig. 6). The maintenance of mitochondrial ATP synthesis during water stress is essential for preserving plastid function, and increased ADK gene expression may indicate the ability to provide more ATP for maintaining cellular activities under drought stress [18,57]. With PEG6000 treatment, gene clustering results showed that SlADK6 and 10, SlADK5 and 7, and SlADK2 and 3 possessed similar expression patterns which supporting the gene sequence homology. Several enzymes, such as ADK and catalase, were specially induced by drought but repressed under salt stress in tomato [17]. In tomato, microarray analysis of genes revealed that an ADK homolog (SGN-U214214), which is the same gene of SlADK10 named here, was repressed in salt-treated tissues [17]. Also, in the present study, almost all SlADKs contained two expression peaks at 9 and 48 h; however, the two expression peaks for SlADK7 were at 6 and 24 h. With Nacl treatment, gene clustering results showed that SlADK2 and 3, SlADK8 and 11, possessed similar expression patterns which supporting the gene sequence homology. With cold treatment, SlADK6 and 10 displayed quite different expression patterns although their sequences were highly homologous. In addition, SlADK3 and 7 responded strongly to cold treatment, indicating that they may play a role in cold stress. (Fig. 6).
Previous evidence indicated that different hormones play important roles in cell responses and stress signal transduction [37][38][39]. For example, IAA is involved in almost all aspects of plant growth and development, from embryogenesis to senescence, from root tip to shoot tip [58]. Eth is a key regulator during eshy fruit ripening [59]. ABA is a crucial phytohormone induced by biotic or abiotic stress, and plays important roles in plant tolerance to abiotic stresses [60]. MeJA play an important role in alleviating biotic (pathogens and insects) and abiotic stresses in plants [61]. Our study showed that the transcripts of these SlADKs were responsive to most hormone treatments ( Fig. 7 and Fig. 8). In plants, many hormones need to cross function. For example, two plant hormones, ABA and Eth, play an important role in the complex story of abiotic stress and, consequently, cross-talk between these two has been reported [62]. Also, both Eth and SA play important roles in biotic stresses [63]. Notably, SlADK2 and 4 exhibited signi cant changes under these hormone treatments, suggesting that they have unique roles in hormone responsiveness (Fig. 7). So, understanding the response of SlADKs to hormones can lay a foundation for further elucidating their functions in plant growth and stress response.
Gene co-expression network analysis (GCNA) is a genetic approach for investigating correlations between genes using large-scale gene expression pro ling data which is especially useful for nding relationships between phenotypic traits and functional modules [64,65]. Besides a lot of ribosomal proteins owing high correlation relationship with SlADK under 44 global conditions, other proteins such as threonine-protein kinase and fructose-1 6-bisphosphatase also possessed higher correlation with SlADK family members. High positive correlation coe cient (0.92) existed between SlADK8 and one resistance protein (Nbs-lrr), which hinted that SlADK8 may have potential role in stress resistance (Table S6, Fig. 9).

Conclusions
A total of 11 ADK genes were identi ed here and named according to their chromosomal locations. The phylogeny, motif, gene structure, chromosomal location, cis-elements, evolutionary relationship, coexpression analysis and expression patterns prediction in different tissues were analyzed with bioinformatics method. The qPCR veri cation results revealed the expression level of SlADKs in different tomato tissues was basically consistent with prediction results. Additionally, the qPCR revealed that the SlADKs responded to multiple abiotic stresses and plant hormones. Notably, SlADK7 exhibited unique changes under multiple stress treatments, and SlADK2 and 4 exhibited signi cant changes under multiple plant treatments. Analysis of co-expression and correlation networks between SlADKs and other tomato functional supply new idea for exploring gene function. In general, the study provides comprehensive information on the SlADK gene family tomato and will aid in determining the SlADK gene function in further research.
LY collected the public dataset, perform bioinformatics analysis and also drafted the manuscript. HHC and LXG performed the experiments. XPZ and JXX contributed to data collection. QC and GQ contributed to bioinformatics analysis. ZGL conceived this study and reviewed the manuscript. All of the authors read and approved the nal manuscript.