Characterization and expression analysis of growth regulating factor ( GRF ) family genes in cucumber

The growth regulating factor (GRF) family is a conserved class of transcription factors involved in various biological processes in plants. However, there have been only a few studies of the GRF family genes in cucumber, Cucumis sativus (Cs). In this study, we identified and characterized 8 CsGRF genes in cucumber. Two highly conserved domains, QLQ and WRC, were identified to be present in all CsGRF proteins. In addition, three less conserved domains (FFD, TQL, and GGPL) were also detected in some CsGRF members. Based on phylogenetic analysis, the GRF genes from cucumber, Arabidopsis, tomato, rice and maize could be classified into 10 groups, and CsGRFs were clustered closer with the GRF genes from dicots (Arabidopsis and tomato) than with those from monocots (rice and maize). Promoter analysis revealed that the CsGRF genes were involved in cucumber growth and development as well as in responses to various hormones and stresses. Transcriptome data showed that the CsGRF genes have distinct expression patterns in different tissues, especially in ovaries and leaves. Expression profiling analysis indicated that all CsGRF genes were responsive to salt and drought stress treatments. These results demonstrate that the cucumber GRF gene family may function in organ development and plant stress responses.

GRF proteins also function in responses to various abiotic stress conditions such as salt, drought, heat and cold [12,30]; however, the roles of plant GRF family genes in stress responses remain elusive.Although a previous study reported 8 GRF family genes in cucumber [6], to date there has been no report about the molecular and expression characterization of the cucumber GRF gene family.In this study, we examined the phylogenetic relationships and promoter sequences of cucumber GRF genes, as well as their expression profiles in different tissues and in response to different abiotic stresses.The results indicated that GRF genes might have differential functions during tissue development and responses to various abiotic stresses, and they provide a foundation for further elucidation of the functions of GRF members in cucumber.

Identification and protein properties of CsGRFs
The sequences of Arabidopsis and rice GRF proteins were downloaded from TAIR (http://www.Arabidopsis.org/)and TIGR (http://rice.plantbiology.msu.edu/), respectively, according to previous studies [2,3].A BLASTP search was performed using these protein sequences in the Cucumber (Chinese Long) Genome Database (http://cucurbitgenomics.org/organism/2) to identify putative GRF proteins.After removal of redundant sequences, the remaining putative GRF protein sequences were subjected to online database searches of SMART (http://smart.embl-heidelberg.de/) and Pfam (http://pfam.sanger.ac.uk/) to confirm the conserved QLQ and WRC domains.Finally, the candidate GRF proteins were named based on their distributions on the seven cucumber chromosomes according to a previous study [6].The ProtParam tool (http://web.expasy.org/protparam/) was used to ana-lyze the theoretical molecular weight (MW), isoelectric point (pI), and grand average of hydropathicity (GRAVY) of the deduced CsGRF proteins.

Multiple sequence alignment and phylogenetic analysis
Multiple sequence alignment of GRF protein sequences from cucumber, Arabidopsis, and rice was performed using Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/) [31] and displayed using the Genedoc software.For a phylogenetic analysis of the plant GRF gene family, the full-length GRF protein sequences from cucumber, Arabidopsis, tomato, rice, and maize were aligned with Clustal Omega and a phylogenetic tree was then generated using the MEGA 7.0 software by the neighbor-joining method with 1,000 bootstrap replicates [32].

Analysis of promoter regions
To investigate the cis elements in the promoter regions of CsGRF genes, 1000-bp upstream sequences of the transcriptional start site of the CsGRFs were chosen to identify the cis elements in the putative promoter regions using the PlantCARE tool (http://bioinformatics.psb.ugent.be/webtools/plantcare/html)[33].

Expression profiling of CsGRFs by transcriptome data
Genome-wide transcriptome data from different tissues were downloaded from a public repository database (https://www.ncbi.nlm.nih.gov/sra/?term=sra046916).The tissue expression profiles of the CsGRF genes were retrieved and analyzed according to previous studies [34,35] and displayed as scaled and centered fragments per a kilobase-exon model per million mapped reads (FPKM) of values by the pheatmap package in R software.

Plant materials and treatments
Two-week-old seedlings of cucumber (Cucumis sativus L. cv.Chinese long No. 9930) were subjected to salt and drought stress conditions, which were described in our previous study [36].The leaf tissues were collected at 0, 3, 6, and 12 h, frozen immediately in liquid nitrogen and stored at -80°C until RNA isolation.

RNA isolation and quantitative RT-PCR (qRT-PCR)
Total RNA from seedlings samples was extracted using Trizol reagent (Tiangen, China) according to the manufacturer's instruction.The first-strand cDNA was synthesized using the SuperscriptIII RNase H-Reverse Transcriptase kit (Invitrogen, USA) following the manufacturer's instructions.qRT-PCR was performed in triplicate using the SYBR Green Master Mix (Tiangen, China) as described previously [36].The cucumber actin gene (CsAct3) was used as an internal control and the relative expression level was calculated using the 2 -ΔΔCT method [37], and the expression of control plants (without treatment, 0 h) was normalized to 1.The primers used for qRT-PCR are listed in Table S1.

Identification and characterization of GRF proteins in cucumber
Eight candidate cucumber GRF proteins (CsGRFs) were identified and named according to a previous study [6].Chromosome 2 and 3 contained 3 CsGRF genes each, whereas one CsGRF gene was observed in chromosomes 1 and 6 (Table 1).The CsGRF genes varied in length from 1129 bp (CsGRF2) to 4403 bp (CsGRF4), and encoded proteins ranged from 319 (CsGRF2) to 672 (CsGRF3) amino acids in length.Analysis of the physical and chemical characteristics of CsGRF proteins showed that the CsGRFs had MW ranging from 36.81 (CsGRF2) to 71.21 (CsGRF3) kDa, the pI ranged from 7.04 (CsGRF3) to 9.59 (CsGRF5), and GRAVY was from -1.052 (CsGRF2) to -0.547 (Cs-GRF5) (Table 1).The GRAVY values of CsGRF proteins were lower than 0, indicating that all the CsGRFs were hydrophilic proteins.
Nearly all the CsGRF proteins contained one QLQ domain and one WRC domain in the N-terminal region, whereas CsGRF5 contained one QLQ domain and two WRC domains (Table 1).To investigate the conserved domains in GRF proteins, we performed a comparative analysis of the full-length deduced amino acid sequences of GRF proteins from cucumber, Arabidopsis and rice.As shown in Fig. 1A, the Q-L-Q residues were highly conserved in nearly all the GRF proteins from cucumber, Arabidopsis and rice, with the exception of AtGRF9, which contained F in place of L. This phenomenon was also observed in the GRF proteins of Solanum lycopersicum [12] and Camellia sinensis [11].A zinc finger motif (CCCH) was present within the WRC domain in all CsGRF, AtGRF and OsGRF proteins.In addition, three short stretches of amino acid residues termed as FFD, TQL and GGPL domains were also found in the C-terminal region of some GRF proteins in cucumber, Arabidopsis and rice (Fig. 1B and C), and these domains might serve as transactivation domains [1], suggesting that they may have specific functions.

Phylogenetic analysis of GRF proteins from different plant species
To further elucidate the phylogenetic relationships of GRF family proteins in cucumber and other plant species, we constructed a phylogenetic tree with the GRF protein sequences from four other plant species including Arabidopsis (9 AtGRFs), rice (12 OsGRFs), tomato (13 SlGRFs), and maize (14 ZmGRFs).Phylogenetic analysis suggested that all the 56 GRF proteins were clustered into ten groups (a-j), and each group had 2-8 members with high bootstrap (Fig. 2).Of these, only two groups (a and j) contained GRFs from both monocot and dicot species.All of the 8 CsGRFs were distributed in 5 groups and clustered together with GRF members from Arabidopsis and tomato.For example, CsGRF3 and CsGRF8, which were clustered with 2 AtGRFs, 2 SlGRFs and 1 OsGRF, belonged to group a (Fig. 2).Group j was comprised of CsGRF5, which was clustered with 1 Arabidopsis, 2 tomato, 1 rice and 2 maize GRF proteins (Fig. 2), while groups d, f and h, which were composed of other CsGRFs, only contained GRFs from dicot species.

Analysis of cis elements in the promoter regions of CsGRF genes
Cis-acting regulatory elements are important molecular switches involved in the regulation of gene transcrip-tion during plant growth and development as well as in the responses to various hormones and stresses [38].
To identify the presence of cis elements in the promoter regions of CsGRF genes, a 1.0-kb promoter region of each CsGRF gene was retrieved and a comprehensive cis element analysis was performed using PlantCARE.The cis elements related to plant stress, hormone responses and developmental processes are shown in Fig. 3 and Table S3.Numerous cis elements related to plant stress and hormone response were identified, implying that CsGRF genes might be involved in response to various stresses and hormones.Essentially, 3 TC-rich repeats and 4 TCA elements, which might account for plant responses to various defenses and salicylic acid (SA), were present in the promoters of CsGRF4 and CsGRF8, respectively.In addition, several cis elements related to developmental processes such as meristem-specific regulation (CAT-box and CCGTCC-box), circadian control (circadian), zein metabolism regulation (O2site), cell cycle regulation (MSA-like), and endospermspecific elements (GCN4_motif and Skn-1_motif) were found in some CsGRF promoter regions (Fig. 3 and Table S3), implying that CsGRF genes have developmentspecific expression.

Expression analysis of CsGRF genes in cucumber tissues
To better understand the functions of CsGRFs in growth and development of cucumber, their FPKM values of transcriptome data in different developmental tissues (ovaries, tendrils, roots, stems, leaves, male and female flowers) were retrieved and analyzed according to our previous study [34].The results showed that some CsGRFs were strongly transcribed in ovaries.For example, CsGRF1 and CsGRF8 had preferential accumulation of transcripts in unexpanded ovaries, and CsGRF5 displayed the highest expression in fertilized ovaries, while CsGRF3 was the most abundantly transcribed in unfertilized ovaries (Fig. 4).In addition, the highest expression of CsGRF2 was observed in both fertilized and unfertilized ovaries, but it was not detected in unexpanded ovaries.Additionally, CsGRF4 and CsGRF6 exhibited the highest expression in leaves, and relatively lower expression in unfertilized and fertilized ovaries.In particular, the expression of CsGRF7 was only detected in roots (Fig. 4).These results indicated that CsGRFs might be involved in the development of specific tissues and growth periods in cucumber.

Expression profiles of CsGRF genes in response to salt and drought stresses
To examine the potential functions of CsGRF genes in response to abiotic stress, the transcript levels of these genes were determined in seedlings under salt and drought stress conditions.Under salt treatment, the expression of CsGRF5, CsGRF6 and CsGRF8 markedly decreased at 3 h, while that of CsGRF7 increased (Fig. 5A).However, nearly all CsGRFs were downregulated at 6 h and 12 h, with the exception of CsGRF6, whose transcription level showed a drastic increase (Fig. 5A).After 3 h of drought treatment, the expression of CsGRF3 and CsGRF8 sharply decreased, while that of other genes had no significant change (Fig. 5B).
In addition, the expression of nearly all CsGRF genes significantly decreased under drought treatment at

Fig. 2.
Phylogenetic tree of GRF proteins from cucumber, Arabidopsis, tomato, rice and maize.The phylogenetic tree was constructed using the neighbor-joining method as implemented in MEGA 7.0 with the full-length GRF protein sequences.The GRF proteins can be classified into 10 groups (a-j).Bootstrap values from 1,000 replicates and the protein IDs are displayed in Table S2.12 h and 24 h, while that of CsGRF6 was remarkably induced (Fig. 5B).These results showed that CsGRF genes were involved in responses to salt and drought stresses.

DISCUSSION
In this study, we conducted a comprehensive analysis of the GRF family genes in cucumber and we identified 8 CsGRFs.The number of GRFs was smaller than that in some other plant species, such as Arabidopsis (9 members) [2], citrus (9 members) [8], pear (9 members) [7], rice (12 members) [3], tomato (13 members) [12], maize (14 members) [4], whereas it was larger than that in tea plant (6 members) [11].This phenomenon may be attributed to gene duplication events of the GRF gene family.For example, one segmental duplication (BdGRF3 and BdGRF6) was identified among the GRF genes in Brachypodium distachyon [39].Four Arabidopsis GRF genes and 8 pear GRF genes were found in the duplication regions of their genomes [7], while no CsGRF genes were located in the duplication region of cucumber genome.
All CsGRF proteins contain one QLQ domain and one WRC domain, except for CsGRF5, which contains one QLQ domain and two WRC domains, similar to GRF12 in Chinese cabbage [5], GRF10 in tomato [12] and GRF6 in tea plant [11].In addition, the FFD and TQL domains are conserved in nearly all CsGRF proteins with the exception of CsGRF5 and CsGRF6 which lack the FFD and TQL domains, respectively.Moreover, CsGRF3, CsGRF7, CsGRF8 and some GRF members from Arabidopsis and rice shared a GGPL domain in their C-terminal regions, which is consistent with the results of previous studies [5,11].These results suggest that CsGRF proteins are evolutionarily conserved in plants.In addition, phylogenetic analysis showed that the GRF proteins from cucumber, Arabidopsis, tomato, maize and rice can be divided into 10 groups, with only 2 groups (group a and j) containing clusters of GRFs from both monocot and dicot species, indicating that these GRF proteins may have evolved either before (groups a and j) or after (groups b-i) the divergence of monocots and dicots.Moreover, CsGRFs seem to have a closer relationship with GRF proteins in dicots (Arabidopsis and tomato) than with those in monocots (rice and maize).For example, most of the CsGRFs were clustered in the groups with dicot species only.Phylogenetic analysis showed that CsGRF6 and AtGRF5 belong to the same group (group d); interestingly, both lack the TQL domain.Although CsGRF3, CsGRF5 and CsGRF8 were located in the groups with monocot species only, they also clustered together with GRFs from tomato and Arabidopsis.
GRF proteins are plant-specific TFs involved in normal plant growth, such as leaf, stem, flower and seed development [5,8,14,19,25].In this study, the expression patterns of the CsGRF genes in different tissues showed some spatial differences, indicating their diverse roles in growth and development of cucumber.For example, the CsGRF genes, except CsGRF4 and CsGRF7, were expressed in different growth periods of ovary development and shared a common cis-element Skn-1_motif, which is required for endosperm expression.Another cis element involved in endosperm expression, the GCN4_motif, was also found in the promoter regions of CsGRF1 and CsGRF2.In Arabidopsis, GRF genes are highly expressed in developing pistils, and overexpression of miR396 results in pistil abnormality through directly mediated cleavage of its GRF targets, which disrupts the formation of the GRF/GIF complex [21].A recent report also demonstrated that the GRF-GIF duo regulates the meristematic and pluripotent competence of carpel margin meristems and the archesporial cell lineage in Arabidopsis [19].These results indicate that some CsGRF genes may play essential roles in ovary development.In addition, CsGRF6 showed the highest expression in leaves, and clustered together with AtGRF5 in the same cluster of group d.AtGRF5 is required for the development of appropriate leaf size and shape and promotes cell proliferation in leaf primordia [16].Hence, CsGRF6 and AtGRF5 may have similar functions in leaf development.
The expression of GRF genes has been reported to be regulated by various abiotic stresses [9,12,30].In this study, all CsGRF genes exhibited differential accumulation or downregulation in response to salt and drought stresses, indicating that they may play essential roles in the response to abiotic stress.However, most CsGRF genes were downregulated, which is different from the responses of SlGRF genes to salt and drought stresses [12].A previous study showed that AtGRF7 acts as a repressor of a wide range of osmotic stress-responsive genes, presumably by preventing growth inhibition under normal conditions [30].Hence, the downregulated CsGRF genes under salt and drought stresses may play negative roles in the stress response.In addition, some closely related genes may have similar expression patterns in response to salt and drought stresses.For example, CsGRF1, Cs-GRF2 and SlGRF4 were in the same cluster of group d; CsGRF1 and CsGRF2 were sharply downregulated by salt and drought stresses at 6 h and 12 h.Similar expression patterns were also observed for SlGRF4 under salt and drought treatments, implying similar functions of these genes in response to salt and drought stresses.It is worth noting that the expression of CsGRF6 was significantly affected by salt and drought treatments, and CsGRF6 harbors a MBS cis element that might be involved in drought-stress tolerance [40].In addition, the promoter in CsGRF6 was also predicted to possess a W-box, which is the cognate cis element for WRKY proteins and is widely present in the promoter region of stress-related genes [41,42].These results suggest that CsGRF6 may act as a positive regulator involved in salt and drought stress.Further functional characterization of the stress-responsive CsGRF genes will lay the foundation for a better understanding of the potential mechanisms against abiotic stresses in cucumber.

CONCLUSIONS
We performed a detailed analysis of the phylogenetic relationships and cis elements in the promoter regions, and of the expression of CsGRF genes in cucumber.Our results provide the foundation for a better understanding the roles of GRF genes in responses to environmental stresses and help reveal the functions of this important gene family in specific developmental processes.

Fig. 4 .
Fig. 4. Expression patterns of CsGRF genes in different developmental tissues of cucumber.The spectrum from blue to red shows the relative expression level of each gene from 0 to 1.

Fig. 5 .
Fig. 5. Expression of CsGRF genes in response to salt (A) and drought (B) stress treatments.Values presented are means of three independent experiments, with error bars indicating standard deviations.

Table 1 .
A complete list of 8 CsGRFs identified in this study.