WANG Xi-cheng,WU Wei-min,ZHOU Bei-bei,WANG Zhuang-wei,QIAN Ya-ming,WANG Bo,YAN Lichun

Institute of Pomology,Jiangsu Academy of Agricultural Sciences/Jiangsu Key Laboratory for Horticultural Crop Genetic Improvement,Nanjing 210014,P.R.China

Abstract Serine carboxypeptidase-like (SCPL) proteins are a group of acyltransferase enzymes that have important roles in plant growth,development,and stress responses. Although SCPL proteins have been studied in many plants,the biological functions of SCPL genes in grape are still unknown. In this study,59 putative SCPL proteins were identified from the grape genome. A bioinformatics analysis,including chromosomal locations,exon/intron structures,phylogeny,ciselements,and conserved motifs,was performed for the gene family. The phylogenetic analysis revealed that VvSCPL proteins could be classified into three groups,with the gene motifs in each group showing high similarity levels. The number of exons in the VvSCPL genes ranged from 1 to 19,suggesting significant variations among grape SCPL genes.The expression of the VvSCPL genes,as assessed by RNA sequencing (RNA-seq) and quantitative real-time PCR,showed that most VvSCPL genes responded to drought-and waterlogging-stress treatments,which indicated their roles in abiotic stress responses. The results provide useful information for further study of SCPL genes in grape.

Keywords:serine carboxypeptidase-like (SCPL) protein,grape,phylogenetic relationships,abiotic stress,gene expression

1.Introduction

The serine carboxypeptidase-like (SCPL) protein and serine carboxypeptidase (SCP) are similar in function and structure,and both of them belong to the S10 carboxypeptidase family,Clan SC (Feng and Xue 2006;Liet al.2016). The S10 family contains many proteolytic enzymes that are comprised of a conservative α/β hydrolase tertiary structure,and the catalytic centers have peculiar topological structures (Zhuet al.2018).Generally,all SCPL proteins contain a SCP (PF00450) domain. All SCPL proteins contain a conserved catalytic triad formed by a serine,an aspartate,and a histidine,which forms the hydrogen-bonding network for the stabilization of the reaction intermediate (Milkowski and Strack 2004;Stehleet al.2006). SCPL protein is a novel and particular plant acyl enzyme,which can be acylated by the glucose ester (Stehleet al.2006). In some plants,SCPL protein is not only involved in the synthesis of glucose polyester,but also related to the synthesis of a ultraviolet ray protective agent (Lehfeldtet al.2000;Fraseret al.2007). For microbes,SCPL proteins is also necessary for the bacterial acylation,and plays pivotal roles in the synthesis process. Additionally,it aids in soilborne pathogens resistance (Mugfordet al.2009). OtherAtSCPs have also been involved in the acylation process,such asAtSCT(Shirleyet al.2001;Leeet al.2012),AtSMT(Hauseet al.2002;Stehleet al.2008a) andAtSST(Stehleet al.2008b).

SCPL proteins play important roles in plant stress responses and disease resistance. Previous research found that the expression ofOsBISCPL1was induced by jasmonic acids,salicylic and was also up-regulated after rice (Oryzasativa) plants suffered a blast fungus infection. Transgenic plants ofOsBISCPL1became less sensitive to abscisic acid and were partially resistant to oxidative stress (Liuet al.2008). Furthermore,the protein Brassinosteroid insensitive 1 is closely related to the sinapate ester biosynthesis. Thus,SCPL proteins may be involved in biochemical pathways and crucial for plant growth and resistance to adverse stresses (Milkowski and Strack 2004).Arabidopsisover-expressing theExtraCarpelsandSeeds(ECS1) gene was also sensitive to brassinolide (Wenet al.2012),and the expression ofSad7enhanced soil-borne pathogen resistance.

In addition,SCPL proteins have a multitude of vital functions in plant growth and development. The studies demonstrated that theGS5gene was crucial for grain size,and theSCP46gene was an important regulator for seed growth and development,which is possibly involved in abscisic acid signaling (Liet al.2011;Xuet al.2015).The grain weight ofArabidopsis,overexpressing theECS1,increased by～33% compared with the wild type (Wenet al.2012). A pea (Pisumsativum)SCPgene,PsCP,whose expression was induced by gibberellins,was involved in the early steps of reproductive and vegetative development (Cercoset al.2003). Additionally,in tabacco (Nicotianatabacum),SCP1andSCP2control cell development (Bienertet al.2012). SCPL proteins are also involved in damage responses,seed germination,and cell programmed death (Domínguezet al.2002;Granatet al.2003;Mugford and Milkowski 2012).

Many genome-wide analyses ofSCPLgenes have been performed in rice (Feng and Xue 2006),poplar (Populustrichocarpa) (Zhuet al.2018) andArabidopsis(Feng and Yu 2009). Grape (Vitisvinifera) is an economically important fruit crop. Although mainly used for wine production,grapes are also consumed fresh and processed to produce raisins,juices,dyes,tannins,and antioxidants. Studies about SCPL proteins in grape have seldom been reported,although grape genomic data has accumulated over the past years (FPCGGC 2007;Bontpartet al.2018). In this study,59SCPLgenes from grape were identified,analyzed,and classified. Their conserved motif composition and expression patterns were assessed under drought and waterlogging-stress conditions. Furthermore,potential transcription factor (TF) binding sites in the promoters ofSCPLgenes were also investigated. This comprehensive study of theSCPLfamily will help shed light on its functions in grape and other related species.

2.Materials and methods

2.1.ldentification of SCPL genes in grape

To identifySCPLgenes in theVitisviniferagenome,the SCPL protein sequences of rice andArabidopsiswere downloaded from the rice genome database (http://rice.plantbiology.msu.edu/downloads_gad.shtml) and TAIR (https://www.arabidopsis.org/). The protein sequences of grape were obtained from theV.viniferaproteome 12× database (http://www.genoscope.cns.fr/externe/GenomeBrowser/Vitis/). The Hidden Markov Model (HMM) of rice andArabidopsisSCPL protein sequences were constructed by the use of HMMER 3.0 Software (HMMER:http://hmmer.janelia.org/) and the HMM was used to search for the grape protein sequences using the BLASTP program. TheE-value was set as 1×e-50to predict the grapeSCPLgene family. To confirm the predictedVvSCPLgenes from grape genome sequences,the candidate sequences were further annotated using Pfamscan and Pfam (https://www.ebi.ac.uk/Tools/pfa/pfamscan/). The genes were identified asVvSCPLgenes only when they contained the PF00450.21 domain.

2.2.Bioinformatics analysis of SCPL genes

The protein molecular weight and theoretical pI of SCPL proteins were calculated using Compute pI/MW (http://au.expasy.org/tools). The subcellular localization predictor (http://linux1.softberry.com/berry.phtml) was used to predict subcellular localizations. Multiple sequence alignments of amino acid sequences were generated using ClustalW (Hung and Weng 2016;Wanget al.2018). The phylogenetic tree was generated using the sequences of SCPL proteins fromArabidopsis,rice,andV.vinifera. The tree was visualized using MEGA5 Software with the neighbor-joining method (Tamuraet al.2011). The chromosomal location images ofVvSCPLgenes were generated using MapInspect Software (http://www.softsea.com/review/MapInspect.html). Gene Structure Display Server (GSDS) was used to analyze the exon/intron structural of theVvSCPLgenes (Guoet al.2007). The online tool MEME (http://meme.nbcr.net/meme) was used to search for conserved motifs within the VvSCPL proteins (Baileyet al.2006).

2.3.Analysis of the cis-regulatory elements in the promoters

2.4.Expression analysis of VvSCPL genes

To investigate the expression profiles of grapeSCPLgenes,RNA-seq data (SAMN04914490) were retrieved from published sources (Haideret al.2017). Bowtie2,TopHat2,cufflinks,and R packages were used to calculate the expression levels and create a heatmap of grapeSCPLgenes (Wickham 2009;Robinsonet al.2010;Langmead and Salzberg 2012;Trapnellet al.2012;Kimet al.2013). On the heatmap,the gene expression levels,presented by reads per kilobase per million mapped reads (RPKM) values,were converted to log2RPKM values.Contributing to the large variations in the expression levels of significantly expressedVvSCPLgenes.

2.5.Plant materials,treatments,RNA isolation,and cDNA synthesis

One-year-old grapevine rootstock ‘SO4’ was grown in pots under standard cultivation conditions in a rainout shelter at the Jiangsu Academy of Agricultural Sciences,Jiangsu,China. To study the expression profiles of grapeSCPLgenes under waterlogging stress,‘SO4’ was placed in water for 30 days,and 16VvSCPLgenes were selected for a qRT-PCR analysis. Grapevine leaves were collected 0,15,and 30 d after the treatment and immediately frozen in liquid nitrogen. Total RNAs were extracted from the samples using the SDS method (Zhanget al.2010),then reverse transcribed into cDNA by the Prime Script™ RT Reagent Kit (TaKaRa,Japan).

2.6.qRT-PCR-based expression analysis

The expression levels of 16 selectedVvSCPLgenes were examined by qRT-PCR using the KAPA SYBR FAST qPCR Kit Master Mix (2×) Universal (KAPA BIOSYSTEMS,Boston,MA,USA) on an Applied Biosystems®7500 Real-Time PCR machine (Applied Biosystems,Foster City,CA,USA). Specific primers were designed using Primer5.0 (Table 1). Grapeactinwas used as the housekeeping gene (AB073011) to monitor cDNA abundance (Shangguanet al.2018). The 20-μL PCR volumes contained 1 μL of cDNA,10 μL of KAPA SYBR,0.5 μL of forward primer,0.5 μL of reverse primer,and 8 μL of ddH2O. PCR was performed using the following cycling conditions:2 min at 95°C,followed by 40 cycles of 5 s at 95°C,and 30 s at 60°C. For each sample,three biological repeats were performed. Relative quantitative analysis of 16 target genes were calculated using the 2-∆∆Ctmethod (Livak and Schmittgen 2001;Wanget al.2014).

Table 1 qRT-PCR primer sequences used in this study

2.7.Statistical analyses

All the expression data were analyzed using Excel andSPSS Statistical Software (version 17.0;Chicago,IL,United States),by ANOVA,followed by Tukey’s significant difference test atP＜0.05.

3.Results

3.1.ldentification of SCPL family members in grape

To identifySCPLgenes from the grape genome,54A.thaliana,60O.sativaSCPL protein sequences,as well as 26 346V.viniferaprotein sequences were downloaded,respectively. The comparisons of the candidate protein sequences from BLAST and HMM hits were performed using anE-value cutoff of 10-5. After screening and validating conserved domain,a total of 59 putativeVvSCPLgenes were identified,and the VvSCPL proteins were named VvSCPL1-VvSCPL59,in accordance with the Gene ID from theV.viniferagenome (http://www.genoscope.cns.fr/externe/GenomeBrowser/Vitis/).VvSCPLgenes exhibited significant variations in protein size and physicochemical properties,with protein sizes ranging from 61 (VvSCPL1) to 997 (VvSCPL10) amino acids,molecular weights (MWs) ranging from 6.95 kDa (VvSCPL1) to 110.76 kDa (VvSCPL10),and protein isoelectric points (pIs) ranging from 4.74 (VvSCPL44) to 9.8 (VvSCPL53). The subcellular localization prediction demonstrated that the putative VvSCPL proteins were mainly located in extracellular and vacuolar compartments. Detailed parameters for each gene,including sequence data,sequence ID,gene locus,coding-sequence length,and protein chemical characterization,are presented in Table 2 and Appendix A.

The chromosomal distributions of the identified 59VvSCPLgenes were investigated. They were assigned to chromosomes 1,3-6,8,10-14,17,18,7_random,10_random,and Un (Fig.1),and their distribution among the chromosomes was unequal. In total,13VvSCPLgenes were clustered on chromosome 3,which exhibited the highest density ofVvSCPLgenes,while 8VvSCPLgenes were clustered on chromosome 11. Six chromosomes (1,6,8,13,14,and 18),each with three to sevenVvSCPLgenes,harbored 28VvSCPLgenes in total,whereas 2 chromosomes (4 and 12) harbored 2VvSCPLgenes each. The other 6 chromosomes each had a singleVvSCPL(Table 2;Fig.1;Appendix B).

Fig.1 Chromosomal locations of 59 VvSCPL genes. Their names are located on the right side of each chromosome,in accordance with the approximate physical location of each VvSCPL gene. No VvSCPL genes were found on chromosomes 2,9,15,16,and 19,which are not illustrated here.

Table 2 Characteristics of the putative SCPL genes in grape

Table 2 (Continued from preceding page)

3.2.Structural analysis of SCPL genes in grape

Gene structure analysis is an important method used to study genetic evolution. The numbers of introns and exons inSCPLfamily members were calculated inV.viniferaand theSCPLstructure in grapevine was created. As shown in Fig.2,all identifiedVvSCPLgenes could be divided into three groups according to the exon/intron distribution. Among the 59VvSCPLgenes,16 had 10 exons,9 had 14 exons,7 had 8 exons,6 had 9 exons,4 had 15 exons,and 4 had 3 exons,while the remaining genes had 1 exon (VvSCPL24),2 exons (VvSCPL1),5 exons (VvSCPL50),6 exons (VvSCPL14),11 exons (VvSCPL7andVvSCPL13),13 exons (VvSCPL22andVvSCPL40),16 exons (VvSCPL34),17 exons (VvSCPL10andVvSCPL37),18 exons (VvSCPL47),and 19 exons (VvSCPL58). Further analyses indicated thatVvSCPLgenes within the same group usually had a similar exon/intron structure,but the gene structures ofVvSCPLmembers in different groups differed significantly in the grapevine. For example,theVvSCPLgenes in Group I had 8 to 17 exons,while genes in Group III had 8 to 11 exons. However,the number of exons in Group II varied considerably,ranging from 1 to 19 (Fig.2).

Fig.2 Phylogenetic relationships and gene structures of SCPL genes in Vitis Vinifera. The phylogenetic tree was constructed using the MEGA5.0 Software with the neighbor-joining method. Gene structures of the SCPL genes are illustrated on the right panel. The exons,UTRs,and introns are marked with green boxes,orange boxes,and single lines,respectively. The gene models were drawn to scale,as indicated above. UTR,untranslated regions.

3.3.Phylogenetic relationship analysis of SCPL proteins in grape

To characterize the phylogenetic relationship between the grape SCPL family members and those from rice andArabidopsis,the amino acid sequence alignment of all members from those species was conducted (Fig.3).The phylogenetic analysis showed that the 59 putative VvSCPL proteins could be divided into three distinct groups (I,II,and III) together with their SCPL orthologs fromArabidopsisand rice,and these were the same as those identified in the previous reports (Feng and Xue 2006;Bontpartet al.2018).

We all walked back and boarded the bus. Nobody kidded Robbie about the flowers anymore. For a long time, even the girls in the back were strangely silent.

Fig.3 Phylogenetic analysis of proteins encoded by the putative SCPL genes in Vitis vinifera,Oryza sativa,and Arabidopsis thaliana. The phylogenetic tree was constructed using the MEGA5.0 software with the neighbor-joining method. The bootstrap values are 1 000 replicates. Three different groups of SCPLs are represented by I,II,and III,respectively.

3.4.Conserved motif and promoter region analyses of SCPL genes in grape

To better understand the similarity and diversity ofVvSCPLgene motifs,the conserved motifs of SCPL proteins were investigated using MEME online software,and 20 motifs were identified (Fig.4). All the putative SCPL proteins contained the PF00450.21 domain,indicating that all the grape SCPL proteins were typical members of the SCPL family. Moreover,most of the VvSCPL proteins contained 13 varieties of motifs (motifs 1-10,13,14,and 16) (Fig.4).

The SCPL proteins identified in the same subgroup had identical or similar motifs. For example,most SCPL proteins in Groups I and III had the conserved motifs 15,17,and 20 at the C-terminal and middle regions,but motif 19 only appeared in some SCPL proteins in Group II. All the SCPL proteins in Groups I and III contained specific motifs 1 and 2,but these motifs were lacking in short SCPL proteins in Group II (Fig.4).

To more accurately predict the function of VvSCPL proteins and further evaluate the evolutionary relationships ofVvSCPLgenes,a phylogenetic tree was constructed (Fig.4). The VvSCPL proteins could be classified into three clades. Equal representation was not noted among these clades. Clades I and III only had 17 and 12 members,respectively,while Clade II contained 30 members. The SCPL proteins that shared similar motif compositions also exhibited close phylogenetic relationships.

Fig.4 The conserved motifs of putative grape SCPL proteins based on the phylogenetic relationships. A total of 20 motifs were identified using the online MEME program. Different colored boxes represent different conserved motifs of VvSCPL proteins. Details of the 20 motifs are shown in Appendix C.

We identified thecis-regulatory elements in the regions 2 000-bp upstream of the transcriptional start site. A total of 1 914 TF-binding sites were recorded,including 337 MIKC_MADS-related,290 Dof-related,247 AP2-related,201 BBR_BPC-related,122 C2H2-related,120 MYB-related,and some other relatedcis-elements,in the promoter regions of the identifiedVvSCPLgenes in grape. The number ofcis-regulatory elements varied among the genes (Fig.5;Appendix D).

Fig.5 The cis-elements in the promoter sequences of SCPL genes in Vitis vinifera. Different colored boxes represent different promoter elements and their position in each VvSCPL gene. Detailed information for the 1 914 promoter elements is illustrated in Appendix D.

3.5.Expression profiles of VvSCPL genes in response to drought stress

IndividualSCPLgene functions respond to various biotic and abiotic stresses (Mugfordet al.2009;Jianget al.2018). To reveal the expression levels ofVvSCPLgenes under drought conditions,we collected RNAseq data from grapes subjected to drought stress for 20 d (SAMN04914490) (Haideret al.2017). Using the RNA-seq data,a heatmap of the 59VvSCPLgenes,represented by RPKM values from different drought-stress treatments,was established (Fig.6). Among the 59 genes,16VvSCPLgenes (VvSCPL1/ 2/5/7/9/13/15/17/18/22/23/26/27/35/52/54) were upregulated,and 32VvSCPLgenes (VvSCPL3/6/8/11/12/14/16/19-21/25/28-31/33/34/36-38/40-43/45-49/56/58/59) were down-regulated,but the remaining 11 genes (VvSCPL4/10/24/32/39/44/50/51/53/55/57) were stable or slightly influenced by the drought treatment. Among the up-regulated genes,VvSCPL26showed the highest transcript abundances,while the expression level ofVvSCPL15had no significant difference. In contrast,among the down-regulated genes,12VvSCPLgenes (VvSCPL20/30/31/33/34/36/37/42/45/46/58/59) showed the same transcript abundance as in the treated samples.

Fig.6 Heatmap representing expression profiles of VvSCPL genes in leaves of Vitis vinifera under drought-stress conditions for 20 d.

3.6.Expression profiles of grapevine SCPL genes in response to waterlogging stress

To further determine whether the expression levels ofVvSCPLgenes were influenced by a waterloggingstress treatment,16VvSCPLmembers were randomly selected from 59 grapeSCPLgenes using a qRT-PCR experiment (Fig.7). Most of theVvSCPLgenes were significantly induced or repressed by the stress. For instance,VvSCPL1,-7,-15,-21,-31,-36,-40,and-54significantly responded to the waterlogging treatment at 15 and 30 d. Expressions ofVvSCPL9,-25,-45,and-48were up-regulated only at 15 or 30 d of stress,while in contrast,expressions ofVvSCPL12andVvSCPL28were down-regulated significantly at 15 and 30 d after the waterlogging treatment.VvSCPL51was down-regulated only at 30 d. However,the expression level ofVvSCPL58did not change significantly at 15 or 30 d. Moreover,several genes showed opposite expression patterns after different stress treatments. For instance,VvSCPL21,-25,-31,-36,-40,and-48were significantly induced by the waterlogging stress,but were repressed by the drought stress.

Fig.7 Expression profiles of VvSCPL genes in leaves of Vitis vinifera under waterlogging conditions. T0 d,T15 d,and T30 d represent 0,15,and 30 d after the waterlogging treatment,respectively.(＊,P＜0.05;＊＊,P＜0.01).

4.Discussion

Characterizations of gene families are useful for studying their functions. SCPL proteins fromArabidopsis(Zhuet al.2000),rice (Ciarkowskaet al.2018),tomato (Lycopersiconesculentum) (Mouraet al.2001),persimmon (Diospyroskaki) (Ikegamiet al.2007),and tea (Camelliasinensis) (Chiuet al.2016) have been purified and characterized to investigate their roles in regulating plant growth and development. Identification and analysis of the gene families must be built on genomic sequences. With the completion of grape genome sequencing,it is possible to identify all SCPL proteins in the grape. Thus,we performed a genome-wide analysis ofSCPLgenes in grape and examined gene expression profiles to investigate their regulations during drought and waterlogging stress in this study.

A total of 59SCPLgenes were identified in theV.viniferagenome and divided into three subfamilies using a comprehensive phylogenetic tree (Fig.3). The number of subfamily members are comparable to those in the 54,60,and 59 identified inA.thaliana,rice,and poplar genomes,respectively. The protein size of the identified VvSCPL proteins ranged from 61 to 997 amino acids,and the variations may be associated with the differences in the numbers of introns (Cakir and Kilickaya 2015). Here,most genes in the same subgroup contained identical or similar exon/intron structures,which was consistent with the structures in some model plants (Feng and Xue 2006;Feng and Yu 2009;Zhuet al.2018).

In general,genes in the same subgroup show similar exon/intron structures (Duet al.2012). Alterations in exon/intron structure or conserved domains may change the gene or protein function (Xuet al.2012). Our phylogenetic analysis agreed with these observations.Interestingly,fiveVvSCPLgenes contained more than 15 exons respectively (Appendix E),while none of theSCPLgenes contained as many exons in rice,Arabidopsis,and poplar (Feng and Xue 2006;Feng and Yu 2009;Zhuet al.2018). This analysis indicated that these fiveVvSCPLgenes went through great variation during evolution. Additionally,genes in the same group had closer evolutionary relationships than genes in different groups. Moreover,the differentiation of conserved motifs in the same genes in paralogs represented the functional differences. Large numbers of stress-,pathogen-,and hormone-relatedcis-elements were found in the putative promoter regions of theVvSCPLgenes in grape. Our results suggested that theseVvSCPLgenes may play important roles in response to various abiotic and biotic stresses. Meanwhile,the same or similarcis-elements were also identified in theSCPLgenes of rice andArabidopsis(Feng and Xue 2006;Feng and Yu 2009).

Interpretation of gene expression patterns can assist in exploring gene functions. To further investigate the functions ofVvSCPLgenes,their expression patterns in response to waterlogging and drought stresses were examined. The expression ofSCPLgenes is easily induced by biotic and abiotic stresses (Liuet al.2008;Chiuet al.2016),and this was consistent with our results. Studies on riceSCPLgenes suggested thatOsBISCPL1plays an important role in defense responses against multiple biotic and abiotic stresses (Liuet al.2008). In diploid oat (Avenastrigosa),Sad7encodes a functional SCPL acyltransferase that is required for the synthesis of antimicrobial triterpene glycosides and for broadspectrum disease resistance (Mugfordet al.2009).

Thus,SCPL proteins may have roles in the synthesis of diverse secondary compounds that confer resistance to biotic and abiotic stresses. According to the RNA-seq data of grape,the expression levels of～81.4% (48/59) ofVvSCPLgenes were affected by the drought-stress treatment,while only 18.6% (11/59) remained unchanged. Among 16 up-regulated genes,VvSCPL26had the highest expression following the drought-stress treatment.Moreover,a qRT-PCR analysis illustrated that eachVvSCPLgene was differentially expressed in response to waterlogging stress,with 87.5% (14/16) of theVvSCPLgenes being significantly affected at 15 and/or 30 d. The transcript levels ofVvSCPLincreased in response to the waterlogging but decreased in response to the drought treatment (Figs.6 and 7). Different transcriptional responses ofVvSCPLto drought and waterlogging indicated the functional differentiation ofVvSCPLgenes under different environmental stresses.

5.Conclusion

SCPL protein is one of the largest groups of enzymes catalyzing proteolysis during functional protein maturation.To date,little is known about the functions of SCPL proteins in grape. The phylogenetic tree grouped the 59VvSCPLgenes into three subgroups (I,II,and III),and most genes in the same subgroup contained similar protein motifs. The majority of these genes were induced by drought or waterlogging stress treatments. The results provided important information for a better understanding of the grapeSCPLgene family. Furthermore,the study laid a foundation for fully elucidatingVvSCPLfunctions and the molecular mechanisms of plant growth,stress resistance,and other important traits in grape and related plant species.

Acknowledgements

This research was supported by the National Key Research and Development Program of China (2018YFD0201300),the Natural Science Funds of Jiangsu Province,China (BK20160587) and the China Agriculture Research System of MOF and MARA (CARS-29-11).

Declaration of competing interest

The authors declare that they have no conflict of interest.

Appendicesassociated with this paper are available on http://www.ChinaAgriSci.com/V2/En/appendix.htm