Genome-wide identification and analysis of GDSL-type esterases/lipases in watermelon (Citrullus lanatus)


 Background

The GDSL esterase and lipase families play important roles in abiotic stress, pathogen defense, seed development and lipid metabolism. Identifying the lipase activity of a putative GDSL lipase is necessary to determine its function. Systematic analysis of the GDSL gene family is still lacking in Citrullus lanatus.

Results

In this study, we identified 65 watermelon GDSL-type esterase/lipase genes and divided these genes into 6 clades based on phylogeny. The phylogenetic relationship of watermelon GDSL genes compared with Arabidopsis thaliana GDSL esterases/lipases was also determined, and these genes were divided into four groups related to morphological development, abiotic stress response, pathogen defense, and secondary metabolism. The chromosomal location of these genes revealed that they are distributed unevenly across all 11 watermelon chromosomes. Analysis of duplication events suggested that segmental duplication and tandem duplication were the major driving forces of GDSL family evolution. Synteny analysis indicated that GDSLs in watermelon were highly homologous to those in Arabidopsis thaliana, melon and cucumber. Transcriptome analyses showed the tissue-specific and common expression of the GDSL genes in leaf and root tissues and identified nitrogen-related genes under low nitrogen (N) stress compared with optimal N conditions.

Conclusions

Our results provide a basis for selecting candidate watermelon GDSL genes for further studies to determine the biological functions of the GDSL genes in watermelon.

2 Abstract Background The GDSL esterase and lipase families play important roles in abiotic stress, pathogen defense, seed development and lipid metabolism. Identifying the lipase activity of a putative GDSL lipase is necessary to determine its function. Systematic analysis of the GDSL gene family is still lacking in Citrullus lanatus.

Results
In this study, we identified 65 watermelon GDSL-type esterase/lipase genes and divided these genes into 6 clades based on phylogeny. The phylogenetic relationship of watermelon GDSL genes compared with Arabidopsis thaliana GDSL esterases/lipases was also determined, and these genes were divided into four groups related to morphological development, abiotic stress response, pathogen defense, and secondary metabolism. The chromosomal location of these genes revealed that they are distributed unevenly across all 11 watermelon chromosomes. Analysis of duplication events suggested that segmental duplication and tandem duplication were the major driving forces of GDSL family evolution. Synteny analysis indicated that GDSLs in watermelon were highly homologous to those in Arabidopsis thaliana, melon and cucumber. Transcriptome analyses showed the tissue-specific and common expression of the GDSL genes in leaf and root tissues and identified nitrogen-related genes under low nitrogen (N) stress compared with optimal N conditions.

Conclusions
Our results provide a basis for selecting candidate watermelon GDSL genes for further studies to determine the biological functions of the GDSL genes in watermelon.

Background 3
The family of GDSL lipases/esterases is a large conserved family and is widely present in all plants, animals, and microorganisms. Unlike the lipases/esterases with the GXSXG motif family, whose active serine site is located near the center of the conserved sequence, the catalytic triad in the GDSL-motif-like family was constituted by three highly conserved amino acid residues (i.e., serine, aspartic acid and histidine) and is located near the N-terminus [1][2][3][4].
Watermelon [Citrullus lanatus (Thunb) Mansfeld] of the Cucurbitaceae plant family is one of the most economically important vegetable crops in the world. Since the GDSL-motiflike family of lipases was first reported [3], considerable progress on GDSL esterases/lipases has been made in various plant species, including Arabidopsis thaliana [24], Oryza sativa [25,26], Brassica napus [27], and Rosaceae species [28], but no GDSLlipase members have been identified and functionally characterized in watermelon. The availability of whole-genome watermelon sequences offers an opportunity to search for GDSL-type lipase genes in watermelon. In this study, we performed a genome-wide analysis of GDSL-type lipase genes, including genomic locations, chromosomal distributions, and evolutionary divergence. Additionally, transcriptome analysis also provided information on the identification of tissue-specific expression in response to low N stress. Taken together, these findings and analysis will provide a strong foundation for further studies on the roles of GDSL-type esterase/lipase genes in watermelon, and the 4 comparative analysis between the GDSL-type esterase/lipase gene family from Arabidopsis thaliana and the other Cucurbitaceae crops will help to characterize the evolution of the GDSL-type esterase/lipase gene family species.

Results
Identification and characterization of the GDSL-type lipase genes in watermelon Based on a hidden Markov model (HMM) search, a total of 65 GDSL-type esterase/lipase genes containing the GDSL domain were identified (Table 1) in the watermelon genome.
Among these genes, the ClCG09G018950 gene was identified as the smallest protein with 211 amino acids (aa), whereas the largest one was ClCG02G019240 (1516 aa). The MW of the proteins ranged from 23.5 to 171.9 kDa, and the pI ranged from 4.99 (ClCG07G011520) to 40.8 (ClCG02G015390). The length of the watermelon GDSL coding sequence ranges between 633 and 4548 bp. The characteristics of all 65 watermelon GDSLs are listed in Tables 1 and S1.
Phylogenetic analysis of GDSL-type lipase genes Based on the protein sequences, the phylogenetic analysis indicated that the 65 GDSL members were divided into six clades, corresponding to clades I, II, III, IV, V and VI (Fig.   1a). Among the 65 GDSL members, 16 belong to clade I, 4 to clade II, 4 to clade III, 6 to clade IV, 2 to clade V and 33 to clade VI. For comparative purposes, we further included comparatively well-characterized GDSL genes from model plant species Arabidopsis thaliana into a second phylogenetic tree, and the combined phyto tree could be divided into four groups (Fig. 1b). Group I (blue) of the combined phyto tree harbored almost half (32 genes) of the total watermelon GDSL genes, and most of the Arabidopsis GDSL genes (65 genes) grouped inside this group as well. Group II consists of 22 GDSL genes, including 8 from watermelon and 14 from Arabidopsis. Group III, containing 15 watermelon GDSL genes, clustered with 24 GDSL genes from Arabidopsis. Group IV contains the most 5 GDSL genes, including 32 from watermelon and 67 from Arabidopsis. Most of the watermelon GDSL genes grouped the same as the first phylogenetic tree with only watermelon GDSL genes, but two genes grouped separately from their original clades; for example, CICG02G019240 and CICG02G019240 were grouped in group VI but were grouped originally in clade I and clade II, respectively. There are 3 phylogenetic subgroups in group I, designated I-a, I-b and I-c. The detailed subgroups in each group are shown in Fig.   1.
Gene structure and motif analysis of the watermelon GDSL gene family The MEME results indicated that exon-intron organizations of all the identified GDSL genes were considerably diverse. As shown in Fig. 2b, all watermelon GDSL genes possessed three to thirty exons, and fifty-eight (89.3%) of the family contained more than four exons. Seven genes (10.7%) had three exons. Genes with only one exon were not observed. The detailed genomic locations of GDSL genes are shown in Fig. 2b. The length of the motifs ranged from 15 amino acids to 34 amino acids. The details of the conserved motifs are shown in Fig. 2a.

Chromosomal distribution and gene duplication analysis
According to the physical locations of the GDSL genes, we constructed a map on the distribution of the GDSL genes on the 11 watermelon chromosomes. Fig. 3 shows that the 65 GDSL genes were unevenly distributed, and most of the GDSL genes (41/65) were concentrated on chromosomes 1, 2, 9, and 10. Chromosome 2 had the highest number of GDSL genes (14 genes, 22% of mapped genes), whereas chromosome 6 had the lowest number (2 genes, 3% of mapped genes).
To better understand the evolutionary constraints of the duplicated watermelon GDSL family, the Ka/Ks ratios of the GDSL gene pairs were calculated. The results showed that 6 duplicated gene pairs had Ka/Ks < 1, with 4 of them being even less than 0.5, suggesting that these watermelon GDSL genes might have been subject to strong purifying selective pressure during evolution. The duplication dates for the 6 duplication events were estimated to have occurred approximately between 8 and 60 Mya (Fig. 3 and 4 and Table   S2).

Evolutionary analysis of GDSL genes in watermelon and other species
To further infer the phylogenetic mechanisms of watermelon GDSL family genes, three comparative syntenic maps of watermelon with three representative species, including Arabidopsis, melon and cucumber, were constructed (Fig. 5). There are 35 orthologous GDSL gene pairs obtained between watermelon and Arabidopsis, 46 between watermelon and melon and 48 between watermelon and cucumber. Some Arabidopsis GDSL genes (19 genes) were found to be syntenic to the same two or three watermelon GDSL genes, only two were syntenic for cucumber and none was syntenic for melon ( Fig. 5 and Table S3).

Expression profiling of watermelon GDSL genes under low N stress
To predict the possible functions of watermelon GDSL family genes, we analyzed the expression of the GDSL genes in leaves and roots treated with 0.2 mM and 9 mM N, respectively. The results revealed that the GDSL genes had diverse expression patterns in the leaf and root. Among the 65 GDSL gene members, twenty-seven genes were expressed in both the leaf and the root tissue, and some members, including ClCG09G001270, ClCG07G004140, ClCG01G023580, ClCG02G019240 and ClCG10G019120, showed the highest transcription level ( Fig. 6 and Table S4), implying that these GDSL genes might play important roles in leaf and root development. Conversely, twenty-three genes displayed very low or could not be detected in either of the two tissues, suggesting that these genes might not play roles in the leaf and root tissues, although they might be primarily expressed in other tissues of watermelon not tested or under some special conditions. Some GDSL genes exhibited tissue-specific expression. For instance, the genes ClCG08G000570, ClCG09G000290, ClCG07G013470 and ClCG10G005280 were only expressed in leaves, while ClCG04G009930, ClCG04G009920 and ClCG05G011430 were expressed specifically in roots ( Fig. 6 and Table S4).
Under the treatment of low concentrations of N, the results showed that the expression of some GDSL genes was significantly induced/repressed compared to the optimal treatment of N. In the leaves, fourteen GDSL genes (ClCG06G003270, ClCG10G013760, ClCG02G016030, ClCG10G009690, ClCG01G024110, ClCG02G001070, ClCG03G007300, ClCG05G025850 and ClCG02G001050) were repressed by the low concentration of N treatment. Interestingly, the transcript levels of many GDSL genes, such as ClCG07G014350, ClCG01G020480, ClCG01G023600, ClCG10G005260 and ClCG01G023580, were upregulated by the low concentration of N treatment. In the root, the expression levels of seven genes (ClCG06G003270, ClCG10G013760, ClCG09G001270, ClCG02G016030, ClCG10G009690, ClCG02G001050 and ClCG10G005260) were downregulated, whereas two genes (ClCG07G014350 and ClCG07G004140) were upregulated ( Fig. 6 and Table S4). The overall expression data analysis suggested that GDSL genes showed diverse expression patterns and might play crucial roles in leaf and root development in watermelon.

Discussion
The GDSL lipase/esterase family has been demonstrated to play multiple functional roles in developmental processes and in responses to abiotic and biotic stresses in plants. In the present study, a comprehensive set of 65 GDSL family genes was identified, and these genes were divided into 6 clades. As a model plant, extensive efforts have been made to functionally characterize the genes of A. thaliana. To speculate the possible functions of the GDSL genes identified in this study, we additionally performed a phylogenetic analysis together with the GDSL genes in the model plant species, Arabidopsis, which can provide useful information regarding the possible roles of GDSL genes in watermelon based on their similarities between syntenic genes. The combined phylogenetic tree showed that the GDSL genes from Arabidopsis and watermelon were grouped into four main groups ( Fig. 1b), which is consistent with the results of earlier studies conducted in Arabidopsis [40]. Group 1 contains 10 Arabidopsis and 10 watermelon GDSL genes, and most of these genes have no known function, except AT2G38180, which was reported to be involved in 9 ethylene (ET) defense signaling pathways [41]. Accumulating evidence indicates that the plant GDSL esterase/lipases are also involved in secondary metabolism in plants.
According to the phylogenetic tree, some of the genes in group 2 were also related to secondary metabolism; in Arabidopsis, for example, AT1G54790 (seed fatty acid reducer, SFAR1) was reported to be involved in fatty acid (FA) metabolism in Arabidopsis seeds [9].
It has been reported that AT3G48460 ( SFAR4) is a GDSL-type esterase involved in fatty acid metabolism by reducing the fatty acid content during post germination and seedling development in Arabidopsis [8]. AT1G67830 ( AtGELP33) was reported to be related to xyloglucan metabolism and cell wall composition [9,14]. These findings indicate that several genes in clade 2 are involved in some stress responses. Eleven members of group 3 were reported to be involved in plant resistance/immunity responses, namely, AT1G53920 ( GLIP5), AT1G71120 (GLIP6) and AT1G54030, AT1G54020, AT1G54010, AT1G54000 and AT3G14210 [21,[42][43][44][45]. Among these genes, AT5G40990 ( GLIP1) is reported to regulate plant immunity through regulation of ethylene signaling, and regulation is mediated by its activity to accumulate a systemic signal(s) in the phloem [18,46,47]. The gene expression of AT5G40990 ( GLIP1), as well as AT1G53990 ( GLIP3) and AT3G14225 ( GLIP4), was regulated by two pathogen-responsive MAPKs, MPK3 and MPK6 [48]. AT1G53940 ( GLIP2) plays a role in plant immune responses and pathogen defense and is involved in the resistance to Erwinia carotovora via negative regulation of auxin signaling [19]. AT1G54030 could cause organizational defects in the endoplasmic reticulum (ER) and aberrant protein trafficking in the plant secretory pathway [42]. The gene AT3G14210 (Epithiospecifier modifier 1, ESM1) has been reported to suppress nitrile formation, increase isothiocyanate production, and correlate with plant resistance against herbivores [35]. In addition, it has been reported that the genes AT5G40990 and AT3G14210 are also related to the biotic stress response [18,19,35,46,47]. Moreover, AT3G14210 and AT3G14220 are tandem neighbors of AT3G14225 [35]. These findings suggested that the watermelon GDSL genes classified into group 3 might be involved in plant resistance or immunity. For the genes in group 4, Arabidopsis genes AT3G11210, AT2G38180 and AT5G45920 are homologs of watermelon genes CICG07G004140, CICG10G022120 and CICG01G023570, respectively. AT3G04290 was first reported to play a role in salt tolerance and may also be involved in defense reactions against pathogens [15]. In 2017, the gene AT3G04290 was retrieved by a yeast two-hybrid screen using VACUOLELESS GAMETOPHYTES (VLG, AT2G17740) as bait, which is essential for the development of female and male gametophytes in Arabidopsis [49]. AT1G58430 ( SFAR2), AT2G42990 ( SFAR3) and At4g18970 ( SFAR5) have been demonstrated to act downstream of the GA signaling pathway and are also involved in fatty acid degradation in Arabidopsis seeds [9]. Moreover, AT1G58430 ( SFAR2) and AT2G42990 ( SFAR3) are also involved in important functions in plant development, morphogenesis, and glucose stress tolerance [9]. AT5G45670 ( LIP1) has been reported to be specifically expressed in the epidermis and highly induced by GA and repressed by DELLAs during seed imbibition [50]. AT5G45670 (LIP1) functions as a negative factor through its L1 box present in the LIP1 element for seed germination [51]. At4g30140 ( CDEF1) has cutinase activity, being secreted from cells and directly degrading the polyester in the cuticle, and it is also involved in the penetration of the stigma by pollen tubes and facilitating the emergence of the lateral roots [52,53]. It has been reported that overexpression of the AT1G29670 gene enhances seed germination and seedling establishment, suggesting that the gene could be a promising target to achieve the features of increased germination and higher oil content in plant breeding [54]. It has been reported that the gene AT1G29660 is involved in phloem-mediated long-distance signaling regulating responses to biotic and abiotic stress [55][56][57]. It has been reported that the genes AT5G18430 and AT5G33370 had approximately 70% identity and over 80% similarity of their amino acid sequences with LTL1, which functions as a GDSL-motif lipase and was associated with salt resistance [15,49]. AT1G75910 ( AtGELP42) functions as an extracellular lipase to facilitate pollen hydration on the stigma in the early pollination stage of Arabidopsis [13]. It has been reported that the gene AT1G75930 plays a role in efficient pollination [13] and that the gene AT1G75930 ( EXL6) is a target of a key transcription factor that coordinates pollen wall development and sporopollenin biosynthesis in Arabidopsis [58]. It has been reported that the gene AT1G75930 plays a role in pollen exine formation and is essential for pollen development in Arabidopsis [27].
In view of the importance of nitrogen (N) as the primary inorganic nutrient in plant growth and development, especially for crops requiring large quantities of fertilizers, such as watermelon, and the key roles of GDSL lipases in regulating plant growth and development, the expression patterns of GDSL genes in the leaf and root of watermelon under optimal nitrogen (ON) and low nitrogen (LN) conditions were investigated in this study based on the available transcriptome data published previously [39]. According to the analysis of gene expression profiling, the watermelon GDSL genes showed diverse expression patterns (Fig. 6). The transcriptome data showed that five GDSL genes (CLCG04G009920, CLCG02G001070, CLCG04G009930, CLCG01G024110 and CLCG11G010220) are expressed only in leaf tissue, four GDSL genes (CLCG05G006600, CLCG01G023460, CLCG04G009910 and CLCG10G005280) are expressed only in root tissue, and approximately 21 GDSL genes are highly expressed in both leaf and root tissue.
Among these genes, the Arabidopsis homolog AT2G23540 for ClCG02G013150 was reported to be highly expressed in leaf and root tissues [59], which was also observed for the tomato homolog of Solyc02g090210 [60]. The Arabidopsis homolog gene AT3G04290 for the gene CLCG11G010220 has similar expression patterns and is expressed mainly in leaf and flower tissues [61]. Functionally, studies have demonstrated that the GDSL lipase plays a role in salt tolerance [15] and is also involved in defense reactions against pathogens [15,62]. It has been reported that the gene AT3G04290 may also play a role in cell wall differentiation and plant growth in the Arabidopsis response to ionizing radiation [63]. In contrast, the Arabidopsis homolog gene AT5G55050 for the gene CLCG10G005280 was also mainly expressed in the root tissue and significantly (P = 0.03) induced by > 3fold (normalized) after 6 h of exposure of plants to allelochemicals identified in buckwheat (fagomine, gallic acid, or rutin) in the aquaculture medium [64].
Many researchers have reported that low nitrogen stress has comprehensive impacts on genes involved in various biosynthetic, catabolic and regulatory processes and thus severely inhibits plant growth and development [65][66][67]. Previous transcriptome data revealed that GDSL genes were also related to low nitrogen stress. For example, under LN stress, the GDSL gene GRMZM2G034958 was only detected in cobs, and GRMZM2G046306 and GRMZM2G015708 were only detected in florets [68], suggesting that these three GDSL genes have negative roles in nitrogen-related metabolic processes. In the present study, the expression profiles of genes from group 2 did not show a significant change in their expression fold under low N stress, both in the leaf and root tissues. However, many members of group 4 show differential expression under the low N stress treatment (Fig.   6), implying the possible role of the genes from group 4 in plant growth and development.
Notably, 5 GDSL genes, ClCG02G013150, ClCG11G010220, ClCG02G006480, ClCG01G023460 and ClCG10G005280, had a significant change in their expression fold in the leaf and/or root under LN, which suggested that the genes played important roles in responses to low nitrogen stress. The expression of the GDSL gene ClCG02G013150 in leaves was downregulated by low N, and its homolog in Arabidopsis, AT2G23540, was also 13 downregulated by the stress of 2,4,6-trichlorophenol (2,4,6-TCP) [69]. A functional study demonstrated that the Arabidopsis homolog AT2G23540 plays an important role in cell expansion and cuticle deposition in response to stresses [70]. These findings suggested that the watermelon ClCG02G013150 negatively regulates low nitrogen tolerance and thus has a potential value in watermelon stress-resistance improvement. The Arabidopsis homolog GDSL genes in other groups were also reported to be involved in low-nitrogen stress. For example, for the GDSL gene AT1G54010 in group 3, the expression level was significantly upregulated in response to both short-and long-term N availability increases

Conclusions
In summary, the present study identified 65 GDSL-type esterase/lipase genes in C.
lanatus. Their gene structure, chromosomal location and phylogenetic analyses were performed, which will provide basic information for the functional characterization of GDSL genes in watermelon. RNA-seq data revealed that tissue-specific and common expression of the GDSL genes in leaf and root tissues, suggesting that the GDSL genes had clear function differentiation watermelon. The expression profiling under low N and optimal N conditions showed that some GDSL genes were significantly upregulated or downregulated, indicating their important roles in nitrogen related growth and development of watermelon. Overall, these data are useful for the follow-up study of the functional characteristics of GDSL genes in watermelon.

Identification of the GDSL-type lipase gene family in watermelon and chromosomal distribution
The genomic data of watermelon (C. lanatus) were downloaded from CuGenDB (Version 2.0, ftp://cucurbitgenomics.org/pub/cucurbit/genome/watermelon/WCG/). HMMER searches were first carried out (1e-3 as E-value cut-off.) in watermelon protein sequences using the GDSL domain Hidden Markov Model (HMM) profile (PF00657) downloaded from Pfam (http://pfam. xfam.org/) with a default e value threshold of 0.1 [29], then the complete GDSL family genes in watermelon were identified (0.1 as e-value cut-off.) with the new watermelon-specific HMM file as query using the "hmmbuild" module by HMMER V3.0 program. The GDSL family genes were mapped to watermelon chromosomes based on their physical location information from the watermelon genome database using Circos [30]. Gene characteristics, including the length of the coding sequence (CDS), the protein molecular weight (MW), and the isoelectric point (pI), were calculated by ExPASy (http://www.expasy.org/). The subcellular localization was predicted using the CELLO v2.5 server (http://cello.life. nctu.edu.tw/).

Conserved motifs, gene structures and phylogenetic analysis
The Multiple Expectation Maximization for Motif Elicitation program (MEME, http://meme.nbcr.net/meme/intro.html) [31] was used to identify conserved motifs based on the protein sequences of the identified genes. The parameters employed in the analysis were set with the minimum motif width, 6; maximum motif width, 50; and maximum number of motifs, 10. The online program Gene Structure Display Server (GSDS: http://gsds.cbi.pku.edu.cn) [32] was used to display the exon-intron organization of the 15 GDSL genes based on the data from the genome annotation file. Multiple sequence alignments of the watermelon GDSL protein sequences were performed using the ClustalW program [33]. Phylogenetic trees were constructed using the Molecular Evolutionary Genetics Analysis (MEGA 7.0) with the maximum-likelihood (ML) method, 1000 repetitions of bootstrap value and Poisson model [34].

Identification of duplicated GDSL genes and nonsynonymous/synonymous substitution (Ka/Ks) ratios of gene pairs in watermelon
Gene duplication events were determined on the basis of multiple sequence alignments using ClustalW with the following criteria: the shorter sequences cover > 75% of the longer sequence after alignment, and the similarity of aligned regions is > 75%. Gaps in the alignments were manually removed by Bioedit. The nonsynonymous (Ka) and synonymous (Ks) values of the duplicated GDSL gene pairs were calculated by the program KaKs_Calculator [35]. The Ks values were used to estimate the approximate date of the duplication time (T = Ks/ (2 × 6.5 × 10−9) × 10−6 Mya), and the Ka/Ks ratio was used to show the selection pressure for the duplicate gene pairs [36].

Analysis of syntenic relationships of GDSL genes between watermelon, A. thaliana and other major cucurbits crops
To understand the evolutionary relationships of the orthologous GDSL family genes between watermelon, A. thaliana, melon and cucumber genomes, the MCscan program [37] was employed to identify orthologous regions with default parameters. The genomic and annotation data of melon (version 3.6.1) and cucumber (version 3) were downloaded from the Cucurbit Genomics Database (CuGenDB) (http://cucurbitgenomics.org/), and those of Arabidopsis were downloaded from the Arabidopsis thaliana Plant Genome Database (AtPGD; http://plantgdb.org/AtGDB/). The identification of the GDSL family genes was performed following the same procedures described above. The synteny relationship of the orthologous GDSL genes obtained between watermelon and other selected species was visualized using Circos [30] and TBtools software [38].

Expression analysis of GDSL family genes using transcriptome data
To explore the expression profiles of GDSL family genes, one set of RNA-Seq data that included 12 samples was utilized to draw heat maps according to fragments per kilobase per million mapped reads (FPKM). The RNA-seq experiment measured the transcriptome response of leaves and roots in response to low (0.2 mM) and high (9 mM) concentrations of nitrogen (N) in watermelon. Three biological replicates were performed, and the RNAseq was run using an Illumina HiSeq 2000 paired end sequencing platform. The RNA-seq data were downloaded in "fastq" format from the public database (https://www.ebi.ac.uk/), and the accession number of the study was PRJNA422970, which included 12 run accessions (SRR6389278-SRR6389289). Details about the transcriptome data and analysis of watermelon leaves and roots under low nitrogen were described and carried out in a previous study [39].

Declarations Acknowledgments
All the authors are grateful for the raw data download from the public database. We sincerely thank the Department of Biological Sciences, 385 Serra Mall, Stanford University, for providing access to the Arabidopsis thaliana data. We sincerely thank the Centre for   Table S1. Protein and CDS sequences of the 65 GDSL genes identified in this study. Table S2. Segmental and tandem duplications of GDSL gene pairs in watermelon (C. lanatus) and inference of duplication time. Table S3. Orthologous relationships of GDSL gene pairs between watermelon and three plant species of (ATH, Arabidopsis thaliana; MEL, Melon and CUM, Cucumber). Table S4. RNA-seq data of 65 GDSL genes that were used in this study.