Vacuolar invertase genes SbVIN1 and SbVIN2 are differently associated with stem and grain traits in sorghum(Sorghum bicolor)

Yunhu Chi,Kimni Wilson,Zhiqun Liu,Xioyun Wu,Li Shng,Limin Zhng,Hichun Jing,c, Huiqing Ho,*

aKey Laboratory of Plant Resources,Institute of Botany,Chinese Academy of Sciences,Beijing 100093,China

bUniversity of Chinese Academy of Sciences,Beijing 100049,China

cEngineering Laboratory for Grass-based Livestock Husbandry,Chinese Academy of Sciences,Beijing 100093,China

A B S T R A C T

Keywords:Sorghum (Sorghum bicolor)Vacuolar invertase SbVIN1 SbVIN2 Enzymatic properties Association analysis In higher plants, vacuolar invertases play essential roles in sugar metabolism, organ development,and sink strength.In sorghum(Sorghum bicolor),two vacuolar invertase genes,SbVIN1 (Sobic. 004G004800) and SbVIN2 (Sobic. 006G160700) have been reported, but their enzymatic properties and functional differences are largely unknown. We combined molecular,biochemical and genomic approaches to investigate their roles in sorghum stem and grain traits. SbVIN1 and SbVIN2 showed different expression levels in internodes,leaves, and panicles, indicating that their importance in each organ was different. In an in vitro sucrose hydrolysis assay, proteins of both genes heterologously expressed in Pichia pastoris displayed similar enzyme properties including the same optimum reaction pH (5)and similar Km for sucroe (49 mmol L-1 and 45 mmol L-1 for SbVIN1 and SbVIN2,respectively). The optimum reaction temperatures of SbVIN1 and SbVIN2 were 45 °C and 65 °C, respectively. SbVIN2 showed higher tolerance to high temperature than SbVIN1. We characterized the sequence variation of these two vacuolar invertase genes in a panel of 216 diverse inbred lines of sweet and grain sorghum and performed gene-based association analysis. SbVIN1 showed significant associations with stem traits including stem length,stem diameter, internode number, stem fresh weight, and Brix, as well as grain traits including hundred-grain weight and grain width. Significantly associated variation sites were mainly in 5′ upstream and intron regions. SbVIN2 only associated with grain width and stem water-soluble carbohydrates (WSCs) content. We conclude that the vacuolar invertase genes SbVIN1 and SbVIN2 are differently associated with stem and grain traits in sorghum.

1. Introduction

Vacuolar invertases (EC 3.2.1.26, VINs) which are active in acid cellular environments and are located in the vacuoles,belong to glycoside hydrolase family GH32(http://www.cazy.org/Glycoside-Hydrolases.html). VIN irreversibly hydrolyzes one sucrose molecule into two hexose molecules, glucose and fructose, thus changing the contents of and ratio between sucrose and hexoses. Owing to the central role of sugar signals in plant metabolism,growth and development,VIN is considered to play crucial roles in carbohydrate metabolism and in growth and development of higher plants,and has been characterized in a wide range of species[1-3].

VINs are often highly expressed in young organs such as hypocotyls of seedlings, developing fruits, and other expanding tissues,suggesting that they function primarily in initial cell differentiation, division and expansion [1-6],controlling the size of plant organs.For instance,in transgenic tomato (Lycopersicon esculentum) fruit, antisense silencing of a VIN gene (TIV1) was proposed to reduce fruit size [7].Antisense suppression of a VIN gene (MAI1) in muskmelon(Cucumis melo) resulted in smaller leaves and fruits and thinner stems in sucrose-accumulating transgenic plants [8].RNAi-mediated suppression of a VIN gene (GhVIN1) caused significant reduction of VIN activity and led to a short-fiber phenotype in a dosage-dependent manner in cotton(Gossypium hirsutum) [9]. Vacuolar invertase (OsINV3) (the ortholog gene of SbVIN1) gene-disruption mutants of rice(Oryza sativa) had shorter panicles with lighter and smaller grains[10].In addition to its regulation at the gene expression level, VIN in plants was also regulated at the protein level by invertase inhibitors[2].VIN activity was much higher in guard cells than in other epidermal cells of Arabidopsis, and specific expression of VIN inhibitors in guard cells reduced VIN activity in guard cells and stomatal aperture and conductance[11]. Taken together, these results demonstrate that downregulation of vacuolar invertase gene expression leads to small organs. VINs also influence sugar composition in plant organs. Antisense silencing of a VIN gene (TIV1) led to an increase of sucrose and decrease of hexose sugar concentrations in tomato fruit[7].Higher VIN expression was associated with higher hexose content in tomato fruit [12,13], litchi fruit[14] and sugarcane stems [15]. The importance of vacuolar invertases in many aspects of the reproductive development of plants has long been recognized. Male sterility caused by drought stress in wheat was accompanied by a decrease in vacuolar invertase activity [16]. Grain yield reduction caused by drought stress in maize was due to ovary development deficiency [17] and insufficient grain filling [18], both were accompanied by the reduction of expression and activity of vacuolar invertase in kernel. Kernels terminating development under high temperature also showed much lower pedicel vacuolar invertase activity than normal kernels [19].In cotton, GhVIN1-RNAi plants showed significantly reduced numbers of viable seeds as a consequence of pollination failure and reduced male and female fertility [20]. Thus,vacuolar invertase is crucial for seed formation and development.

Sorghum(Sorghum bicolor),the fifth most important cereal crop in the world,is the staple food for half a billion people as well as providing feed and fodder for animals[21].Sorghum is regarded as a model bioenergy crop owing to its high tolerance to abiotic stresses, high photosynthetic efficiency,low-input cost and relatively small diploid genome [22].Because both the grain and the sugar-rich stems of sorghum can be used as feedstock for the biofuel and livestock industries, these traits are the targets for breeding improvement.It is thus desirable to investigate the molecular genetic control of grain and stem traits.

As critical sugar-metabolizing enzymes, VINs have been investigated for their roles in sugar metabolism in sorghum.As early as 1987, Lingle et al. [23] found that sucrose accumulation was related to a decline in VIN activity.Tarpley et al. [24] proposed that declines in the activities of soluble sucrose-degrading enzymes are a prerequisite for sucrose accumulation in the stem. Liu et al. [25] found that VIN activity was correlated positively with hexose content and negatively with sucrose content,and recently,the expression level of SbVIN1 was reported to be negatively correlated with sucrose content in 20 sorghum cultivars [26]. A small repeat marker of SbVIN1 was associated with stem Brix in a sweet sorghum population[27].Besides its regulatory role in sucrose accumulation, VIN may function in stem and grain development in sorghum. Lingle et al. [23] observed that the two uppermost elongating internodes showed the highest activities of acid invertases and other sucrose-metabolizing enzymes. Also, Bhatia et al. [28] found that high activity of soluble acid invertases was present in the apical stem parts,where rapid cell division and cell elongation occur. They proposed that the higher VIN activity would enable more sucrose to be decomposed into hexoses to meet the needs of cell division and elongation in sorghum. SbVIN1 expression was found in the early stage of seed development and reached its highest level six days after pollination[29].Thus,VINs may influence both stem and grain traits.

Previous studies identified two vacuolar invertases in sorghum(Sobic.004G004800 and Sobic.006G160700)[26,30],henceforth

SbVIN1 and SbVIN2.The association of SbVIN1 with Brix of sweet sorghum had been suggested[27].Although SbVIN2 is co-located with a meta QTL as defined by Mace et al.[31],and a major QTL(～51.8 Mb) for midrib color, sugar yield, juice volume, and moisture revealed by GWAS [32], which is located 60 kb from SbVIN2,there have been no reported functional studies of SbVIN2.

In the present study,the molecular,biochemical and genomic approaches were combined to investigate the function of SbVIN1 and SbVIN2. We verified the sucrose-cleaving activities of both SbVIN proteins by in vitro enzymatic assays, examined the genomic sequence variation of the two genes, and performed association analysis for grain and stem traits.

2. Materials and methods

2.1. Plant material and field experiments

For gene expression analysis, the grain sorghum cultivar BTx623 and the sweet sorghum cultivar E-tian were grown in the experimental field at the Institute of Botany, Chinese Academy of Sciences, Beijing, China. The uppermost fully expanded leaves at jointing stage, leaves at milking and maturation stage, the upper elongating internodes with length ＜2 cm at jointing stage, the sixth internodes counted from panicle at milking and maturation stages, panicles at booting, flowering and milking stages were sampled. Taking account of diurnal changes in VIN expression [33], samples were always collected between 3 and 4 PM. Samples were frozen in liquid nitrogen immediately after excision from plants and then stored at -80 °C until use.

A panel of 216 accessions collected worldwide was assembled for association analysis. Field trials were conducted at the experimental station (40°N, 116°E, altitude 112 m) of the Institute of Botany, Chinese Academy of Sciences, Beijing, China in 2015, 2016, and 2017. Each accession was planted in single-row plots,with a row spacing of 65 cm and plant spacing of 25 cm.

2.2. cDNA isolation and sequences analyses of SbVINs

To clone the cDNAs of SbVIN1 and SbVIN2, reverse transcription-nested polymerase chain reaction (RT-nested PCR) was performed.The amplified products with blunt ends were subjected to an A-tailing reaction before insertion into a pGEM-T easy vector(Promega,Madison,WI,USA)to generate recombinant plasmids for sequencing. The primers used are listed in Table S1.

A neighbor-joining phylogenetic tree was constructed from the full protein sequences of SbVIN1, SbVIN2 and acid invertase genes cloned from other species using MEGA7 [34]with default parameters and 1000 bootstrap replications.

2.3. Expression analysis of SbVIN1 and SbVIN2

Total RNAs of stem,leaf,and panicle at various developmental stages of BTx623 and E-tian were prepared. Quantitative real-time PCR (qRT-PCR) was performed on an Eco Real-Time System (Illumina Inc., San Diego, CA, USA) with SYBR Green Real-time PCR Master Mix(Beijing CoWin Biosciences,China).Each biological sample had three technical replications, with β-Actin (Sobic.001G112600) as an internal control. The relative gene expression level was calculated by the 2-ΔΔCTmethod[35].The gene-specific primers used for qRT-PCR are described in Supplemental Table S1.

2.4. Heterologous expression and in vitro biochemical characterization of SbVIN1 and SbVIN2

Vacuolar invertases are prone to N-glycosylation, which is required for their activity and stability [36]. The heterologous expression of SbVIN1 and SbVIN2 was accordingly performed in Pichia pastoris GS115, a eukaryote capable of glycosylation modification. SbVIN1 and SbVIN2 proteins were predicted to contain a hydrophobic N-terminal vacuolar-targeting peptide,which could interfere with secretion of the protein in yeast.Thus,PCR amplification yielded two truncated SbVINs,lacking 333 bp and 369 bp from the 5′-terminals of the SbVIN1 and SbVIN2 genes,respectively,according to Ji et al.[37].First,two truncated SbVIN1, and SbVIN2 genes without a sequence encoding the N-terminal vacuolar-targeting region were prepared by PCR using gene-specific primers (Table S1). The resulting amplicons were digested with restriction endonucleases SnaBI and AvrII and then ligated into SnaB I/Avr IIdigested pPIC9k (Invitrogen, Carlsbad, CA, USA), thereby generating two pPIC9K-derived plasmids, pPIC9KSbVIN1 and pPIC9KSbVIN2. The P. pastoris GS115 strain (Invitrogen) was transformed with the Dra I-digested pPIC9KSbVIN1 or pPIC9KSbVIN2.

Selected transformants of the two genes were cultured in 200 mL of BMGY medium(1%yeast extract,2%peptone,1.34%yeast nitrogen base,1.64 μmol L-1biotin,100 mmol L-1potassium phosphate, 1% glycerol, pH 6.0) at 30 °C in a shaker incubator for 24 h. At OD600= 2-4, the cells were collected by centrifugation and resuspended in 50 mL BMM medium(100 mmol L-1potassium phosphate, 1.34% yeast nitrogen base, 1.64 μmol L-1biotin, 0.5% methanol, pH 6.0) for continued induction for 4-5 days at 30 °C. To sustain induction,methanol was added every 24 h to maintain a final concentration of 0.5%. The induced invertases were thus secreted into the medium. The methanol-induced cultures were centrifuged at 12,000 r min-1for 30 min,and the supernatant concentrated and partially purified with Amicon Ultra-15 centrifugal filter devices (50 K MWCO) (Merck KGaA, Darmstadt, Germany) with further purification through ConASepharose. The purified proteins were visualized on 8% (v/v)SDS-PAGE gels and quantified on a NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE, USA). Other purified invertase proteins were stored at 4 °C for enzyme activity measurements.

The optimal pH was determined at 40 °C in McIlwaine buffer (with 8% sucrose) at varying pH values (2.5-8.0). For optimal temperature determination, the reaction was performed at the optimal pH of each SbVIN protein at temperatures varying from 30 °C to 80 °C with intervals of 5 °C. An enzyme protein inactivated by incubation at 98 °C for 30 min was used as control. Yeast invertase, Grade VII, I4504 from Sigma-Aldrich (https://www.sigmaaldrich.com/), was used as the reference (here abbreviated as Sigma-inv). McIlwaine buffer, 95 μL, was mixed with 5 μL enzyme solution in an ice bath followed by incubation for 10 min on a pre-warmed PCR machine and then incubation at 98 °C for 30 min to stop the reaction. Each reaction was performed in triplicate. The released glucose was measured with a YSI 7100 multiparameter bioanalytical system(YSI Inc.,Yellow Springs, OH,USA). The activity relative to the maximum activity of each enzyme was calculated.

To test their thermal stability,the enzymes were incubated at 25-75 °C for 10,20,30,and 60 min and allowed to recover at 4 °C for at least 1 h. The activity of these treated enzymes relative to that of untreated enzymes was then calculated.The vacuole is an ion buffer pool in plant cells. The concentration of ions in vacuoles varies with the environment. So, the influence of divalent metal ions on invertase activity was investigated in McIlwaine buffer with metal cations (CaCl2, MgSO4, CuSO4, MnSO4, CdSO4) concentrations ranging from 0 to 20 mmol L-1under optimal pH and temperature. The enzyme activity at each metal ion concentration was calculated as a function of activity at zero concentration.

Reactions for determination of the Michaelis constant(Km)for sucrose of each enzyme were conducted at the respective optimal pH and temperature of each enzyme, with sucrose concentration ranges of 0-160 mmol L-1. The Kmwas calculated by the Lineweaver-Burk plot method.

2.5. Phenotyping and data analysis of grain and stem traits

Three mature representative plants of each accession were sampled for measuring stem and grain traits (Table 1). Brix was measured with a handheld refractometer using juice squeezed from the internode of the middle section of a sorghum stem. The fresh weight (FW) of about 50 cm of the middle internode was weighed on an electronic balance. The stems were dried in an oven at 105 °C for 2 h and then 80 °C for one week and weighed to obtain dry weight(DW).The WC was estimated as[(FW - DW)/FW] × 100.

Stem sugar content was determined by soaking 50 mg of stem powder in 1.5 mL double-distilled water(ddH2O),boiling the mixture in a water bath for 30 min, and then cooling to room temperature. The mixture was centrifuged at 12,000 r min-1for 10 min and the supernatant collected for sugar measurements. WSCs were estimated by the anthrone colorimetric method. Glu and Suc were directly measured in the YSI 7100 multi-parameter bioanalytical system using the corresponding enzyme membrane.

Fully developed sorghum grains of 216 accessions were used for grain size measurements. One hundred grains randomly selected from each plant and weighed on an electronic balance to determine HGW. To measure GL and GW, 20 grains of each accession were spread on the bed of a Microtek Scanmaker i800 (Microtek, Shanghai, China) and images made using transmitted light. The images were analyzed with a WSeen Measuring System for Rice Grain Quality (Hangzhou WSeen Detection Technology Co. Ltd.,Hangzhou, Zhejiang,China).

Pearson correlations were calculated and visualized using the chart.Correlation() function in the “PerformanceAnalytics”package in R. Best linear unbiased prediction (BLUP) values were calculated in the R package“lme4” as:

Y ＝lmer（y ～ （1jline）＋ （1jyear）

＋（1jreplicate:year）＋ （1jline:year））.

Broad-sense heritability (H2)was calculated as:H2＝

where σ2Gis genotype variance, σ2G×yis genotype-by-year interaction,σ2Eis residual variance,and y is number of years.

2.6. SbVIN sequence polymorphism detection

For all 216 lines, at least 10 seedlings were pooled, ground to powder in liquid nitrogen,and subjected to DNA extraction by the CTAB method.

Before primers to amplify the fragments of the two genes were designed, their variation was investigated, especially in exons from the Phytozome database (https://phytozome.jgi.doe.gov/) [38]. Appropriate detection methods were used todetect the variation. The primers and detection methods are summarized in supplemental Tables S2 and S3.

Table 1-The description of the phenotypes measured in this study.

Distribution of variation sites on genes was drawn with GSDS[39].Pairwise linkage disequilibrium(LD)of markers was calculated with TASSEL[40].

2.7. Candidate gene association analysis

Association analyses of the two SbVINs with stem and grain traits in three years and the BLUP values of each trait were performed using the MLM + Q + K model in TASSEL [40].Population structure (Q) was analyzed using USEPOPINFO model in Structure 2.3.4 [41] with 51 presence/absence variants (PAVs) markers [42] and 148 resequenced accessions that were included in this population panel as “prior known population”.In particular,the Q value was calculated,and an optimal K was chosen for 148 accessions using resequencing data in Admixture software [43]. This prior information was then combined with the 51 PAVs markers to calculate the structure of all 216 accessions using Structure. A kinship matrix was calculated with SPAGeDi [44] using the PAVs markers. Association analysis was performed following the instructions of the TASSEL MLM + Q + K model workflow.Before GWAS, polymorphic loci were filtered using a missing rate of 0.5 and a minor allele frequency (MAF) of 0.05. A stringent Bonferroni correction(0.05/number of markers)was used to determine the threshold of significant associations.Trait differences among major haplotypes (＞5%) were compared in R using one-way ANOVA.

Fig.1- Expression profiles of SbVIN1 and SbVIN2 in grain sorghum BTx623 and sweet sorghum E-tian.Expression of SbVIN1 and SbVIN2 in internodes(A),leaves(B),and panicles (C)of BTx623;in internodes(D),leaves (E) and panicles(F)of E-tian.Expression levels are mean ± SE of triplicate bioassays.Actin(Sobic.001G112600)was used as a control.For the leaf of BTx623,“new mature” represents the uppermost new fully expanded leaves at jointing stage;“milking”and“maturation” represent the sixth leaves counted from the panicle at milking and maturation stages,respectively.The leaves of E-tian were new fully expanded flag leaves and flag leaves at milking and maturation stages.Internode tissues of both cultivars were collected from the upper internode with length ＜2 cm at the jointing stage(elongating)and the sixth internodes counted from panicle at milking and maturation stages.The panicles were sampled at booting,flowering,and milking stages.Because there was only one internode sample of E-tian at elongating stage,there is no bar for both genes.The panicle samples of E-tian at booting and milking stages were lost.

3. Results

3.1. cDNA isolation and expression analyses of SbVIN1 and SbVIN2

The full CDS sequences of SbVIN1 and SbVIN2 were successfully isolated from grain sorghum BTx623 by RT-nested PCR.The CDS of SbVIN1 was identical to those of SAI-1b(GenBank accession number JX535516) and SAI-1c (GenBank accession number JX535517)[27].The CDS of SbVIN2 was identical to the CDS deposited in GenBank(XM_002446812.2).

The phylogenetic relationship of SbVINs with other cloned acid invertases was reconstructed based on their full protein sequences. The two SbVIN proteins clustered with VINs from other plant species.VINs could be further subdivided into two subgroups:monocots and dicots.Both SbVIN1 and SbVIN2 fell into the monocot group. The genomic regions surrounding the two SbVIN genes show strong micro-synteny with those of maize and rice (data not shown). These ortholog genes also had the same gene structure(Fig.S1).

Fig.2-Effects of pH and temperature on enzyme activity,and enzyme reaction velocity at different sucrose concentrations.(A)Effects of pH on enzyme activity;(B)Effect of temperature on enzyme activity;(C)Enzyme reaction velocity at different sucrose concentrations,used for calculating the Km of each enzyme.Results are mean ± SE of three replicates. The invertase enzyme from Sigma-Aldrich(Sigma-inv)was used as a reference.

Expression profiles of SbVIN1 and SbVIN2 were characterized by qRT-PCR (Fig. 1). The elongating internode of E-tian was represented by only one biological sample the panicles at booting and milking stages having been lost. SbVIN1 showed lower expression in panicles at the flowering stage of BTx623(Fig. 1-C) than in E-tian (Fig. 1-F). SbVIN2 showed lower expression in newly expanded leaves and leaves at grain filling stage in BTx623(Fig.1-B)than in E-tian(Fig.1-E).SbVIN1 showed higher expression in elongating and milking stage internodes than SbVIN2(Fig.1-A,D).In leaves,the expression of SbVIN1 was significantly lower than that of SbVIN2(Fig.1-B,E).

3.2. Heterologous expression and biochemical properties of recombinant SbVIN1 and SbVIN2

Fig.3- Thermostability of SbVIN1 and SbVIN2.Remaining enzyme activity of SbVIN1(A),SbVIN2(B), and Sigma-inv(C)after incubating at different temperature for various times.Values are means±SE of triplicate tests.

The two chimeric plasmids were transformed into Pichia pastoris GS115 for methanol-inducible expression. With the α-factor signal sequence of pPIC9K, the expressed recombinant proteins of SbVIN1 and SbVIN2 were secreted into the medium. The purified SbVIN1 and SbVIN2 proteins were examined on SDS-PAGE and revealed a specific band of～70 kDa (Fig. S2), while the control strain with an empty vector (pPIC9K) did not show the corresponding band. The theoretical molecular weights of the amino acid skeleton of the two truncated proteins expressed in P.pastoris were about 62 kDa. The higher molecular weight of protein after purification through ConA-Sepharose suggested glycosylation modification of both proteins.

The two SbVINs showed a similar response to the pH gradients (Fig. 2-A). Activities of both enzymes increased sharply from pH 4 to 5 and peaked at pH 5, decreased significantly with further increase in pH, and vanished at pH 8.0. The optimum reaction temperatures of SbVIN1 and SbVIN2 were 45 °C and 65 °C, respectively (Fig. 2-B). Both SbVIN enzymes showed steep declines in activity above their respective optimum reaction temperatures (Fig. 2-B). Under the optimal reaction conditions of each enzyme and the range 0-160 mmol L-1sucrose(Fig.2-C),the Kms of SbVIN1,SbVIN2,and Sigma-inv were 49 mmol, 45 mmol, and 47 mmol L-1,respectively.

Fig.4- Effects of metal ions on activity of SbVIN1 and SbVIN2.Effects of metal ions at various concentrations on enzyme activity of SbVIN1(A),SbVIN2(B), and Sigma-inv(C).The reactions were performed in McIlwaine buffer(pH 5)containing various concentrations of Ca2+,Cu2+,Mg2+,Mn2+,and Cd2+,and at the optimum reaction temperature of each enzyme for 10 min.Values are means± SE of triplicate tests.

The two enzymes showed differing thermostabilities (Fig.3-A, B). The activity of SbVIN1 persisted for 60 min at a temperature ≤35 °C. Upon incubation at 45 °C for 10 min, its activity dropped to about 60% of the control, decreasing linearly with incubation time and approaching zero after 60 min (Fig. 3-A). At temperatures above 55 °C, the activity of SbVIN1 was entirely lost on incubation for 10 min (Fig. 3-A).The activity of SbVIN2 decreased to ～70%after incubation for 10 min and remained unchanged with the prolongation of incubation time at 25-55 °C, except at 55 °C, where with incubation time increasing from 30 min to 60 min,the activity decreased from 56% to 43%. At 65 °C, the activity of SbVIN2 dropped to zero after incubation for 10 min (Fig. 3-B). The thermostability of Sigma-inv showed a similar trend to that of SbVIN2(Fig.3-C).

Fig. 4 shows the effects of five metal ions including Ca2+(CaCl2), Mg2+(MgSO4), Cu2+(CuSO4), Mn2+(MnSO4), and Cd2+(CdSO4)on enzyme activity.Ca2+and Cu2+showed little effect on SbVIN1 activity in the 0-20 mmol L-1range (Fig. 4-A), but slightly promoted the activity of SbVIN2 (Fig. 4-B). Mg2+increased the activities of both enzymes, but showed a less stimulating effect on SbVIN1 than on SbVIN2.With 5 mmol L--1Mn2+,the activity of SbVIN1 decreased by nearly 20%(Fig.4-A), while the activity of SbVIN2 increased by 20% (Fig. 4-B). A further increase of Mn2+concentration showed little effect on the activity of SbVIN2, but when it reached 15 mmol L-1, the activity of SbVIN1 increased. Cd2+inhibited activities of both invertases.When the Cd2+concentration increased from 15 to 20 mmol L-1,the activity of SbVIN1 decreased more than that of SbVIN2. The activity of Sigma-inv (Fig. 4-C) showed responses to different metal ions similar to those of SbVIN1,but showed less activity reduction at 20 mmol L-1Cd2+than SbVIN1 or SbVIN2.

3.3. Phenotypic analysis of stem and grain traits in sorghum inbred lines

Stem and grain traits displayed wide variation (Table 2). SW and Suc showed the highest variation(CV ＞0.6),and WC and GL the lowest (CV = 0.08 and 0.1, respectively). All traits showed high heritability(H2＞0.7),with SL having the highest(H2= 0.95) and WC the lowest (H2= 0.74). Based on the Shapiro-Wilk statistic (W), most phenotypes showed a nonnormal distribution (P ≤ 0.05) (Table 2). There were high correlations between all the traits (Fig. 5). The stem traits were positively correlated with one another except for WC,which was negatively correlated with the other stem traits.WC was negatively correlated with Brix, as also reported byMorey et al. [26]. SL, SD, and IN, showed high positive correlations with SW (r ＞0.7). Positive correlations were also found among grain traits. Significant negative correlations between grain and stem traits were observed, suggesting a potential trade-off between stem and grain traits in sorghum(Fig.5).

Table 2-Descriptive statistics for grain and stem traits in the association panel.

3.4. Variant sites of genes and their linkage disequilibrium(LD) analysis

After filtering, 28 polymorphic markers (including InDels and SNPs) were retained for association analysis (Fig. 6-A, C). For SbVIN1, one synonymous mutation (CDS 102 bp, C to T) and one nonsynonymous mutation (CDS 181 bp, ACC to GCC,T61A) were identified in exons. Marker 21 of SbVIN1 was the short repeat sequence ATTGA reported by Liu et al.[27].In our population,this short sequence was repeated 4,5,and 6 times in different accessions.For SbVIN2,one synonymous SNP and one 3-bp InDel (CDS 859 bp) were detected in the third exon.None of the variations of either gene were located in regions encoding conserved domains of acid invertase proteins [45].

Fig.5- Variation and Pearson correlations of grain and stem traits in 216 diverse sorghum accessions. The histograms along the main diagonal represent the frequency distributions of phenotypic traits.The lower diagonal contains pairwise scatter plots of all traits with the red line representing the correlations’line of best fit.The pairwise Pearson correlation coefficients are shown on the upper diagonal.The printed sizes of coefficient values correspond to the strength of the correlation.The correlation significance levels are*P =0.05, **P =0.01,and ***P = 0.001.SL,stem length;SD,stem diameter; IN,internode number;SW,stem fresh weight;WC,stem water content;Suc,sucrose content in the stem;Glu,glucose content in the stem;WSCs,water-soluble carbohydrate content in the stem;HGW,hundred-grain weight;GW,grain width;GL,grain length.

Fig.6-Distribution and pattern of pairwise LD of variant loci in SbVIN1 and SbVIN2.(A)and(C)Gene structures and variant site distribution of SbVIN1 and SbVIN2.Green boxes represent exons and red bars represent variant sites.(B) and(D)Pairwise LD blocks of variant loci in SbVIN1 and SbVIN2.LD is indexed by R2 values.R2 values are shown by the legend of the right of the LD block.

3.5. Gene-based association analysis

Candidate gene association analysis showed significant associations between polymorphisms in SbVIN1 and stem and grain traits including SL,SD,IN,SW,Brix,HGW and GW in at least one of three years and the BLUP values of each trait(Fig. 7). In contrast, WC (Fig. 7), Glu, Suc, and WSCs contents were not significantly associated with variations in SbVIN1(Fig. S3). Among the significant loci in SbVIN1, four were poly(dA/dT):markers 14,15,20,and 27(Fig.8-B).

Four major haplotypes (＞ 5% of the accessions) (here named haplotypes 1,2,3.and 4)were identified by combining all significant loci, with 88, 15, 14, and 13 accessions,respectively (Fig. 8-B). Traits showed significant differences among the four haplotypes (Fig. 8-A). Accessions of a given haplotype with lighter and smaller grains had stronger stems(Fig. 8-A). For example, most of those accessions harboring haplotype 1 had strong and juicy stems but small grains.

Variations within SbVIN2 were associated only with WSCs and GW measured in 2015 (Fig. 9-A). Most of the significant loci were located in the 5′ upstream region (Fig. 9-A), which formed a perfect LD block (Fig. 6-D). There were two main haplotypes in this block, haplotypes 1 and 2, represented by 121 and 60 accessions,respectively(Fig.9-C).WSCs and grain width showed significant differences among the two haplotypes(Fig.9-B).

4. Discussion

4.1. Enzyme properties of two SbVIN proteins

The optimal reaction pH of both SbVIN enzymes was 5(Fig.2-A), similar to that of most acid invertases(with a pH range of 4.5-5.0)[3,4].In contrast to Sigma-inv,the two SbVIN enzymes retained high activity only within a narrow pH range(Fig.2-A).The optimal temperature was 45 °C for SbVIN1 and 65 °C for SbVIN2, which were within the reported optimum reaction temperature region of vacuolar invertase, such as VIN of sugarcane (55 °C) [46], rice (40 °C) [47], Arabidopsis (45 °C) [48],mango (60 °C) [49], and Elsholtzia haichowensis (70 °C) [50].SbVIN2 was more tolerant to high temperature than SbVIN1(Fig. 3). The Kmvalues of SbVIN1 and SbVIN2 were 49 and 45 mmol L-1, respectively, within the region of previous reports for vacuolar invertases such as VINs purified from sugarcane mature stem juice(160.6 mmol L-1)[46],sugarcane stem storage tissue (50.0-83.3 mmol L-1) [51], maize seedling roots (4 mmol L-1) [52], and leaves of Arabidopsis(25.5 mmol L-1) [48], rice and wheat VINs expressed in Pichia pastoris (22.6, 13.2, and 7.4 mmol L-1for OsVIN1, OsVIN2, and TaVIN2, respectively) [37]. Ca2+, Mg2+, Cu2+, Mn2+had no or slightly stimulating effect on enzymatic activity, but Cd2+inhibited the activity of two SbVIN enzymes (Fig. 4), behavior similar to that of most invertases (EC 3.2.1.26) in Enzyme Database-BRENDA(https://brenda-enzymes.org/)[53].

Fig.7-Association analysis of SbVIN1 with stem and grain traits.Phenotype data include the data from 2015,2016,and 2017 and the best linear unbiased prediction(BLUP)values of each trait.The horizontal coordinate displays the distribution of variant loci in the gene SbVIN1.

Fig.8- Traits significantly associated with SbVIN1 showed significant differences among four main haplotypes. (A)Comparison of traits among four main haplotypes.H1 to H4 denote haplotypes 1 to 4.Box widths correspond to numbers of accessions harboring the corresponding haplotypes.(B) Significantly associated markers and the four main haplotypes they contain.

Fig.9-Association results of SbVIN2.(A)Manhattan plots of association of SbVIN2 with sugar and grain width.(B)Comparisons of stem WSCs and grain width between two main haplotypes in the 5′upstream region of SbVIN2.(C)Significantly associated markers representing the two main haplotypes.

Thus, the two SbVIN enzymes are acid invertases, hydrolyzing sucrose to glucose and fructose at an optimum reaction pH of 5. Compared with most reported plant vacuolar invertases, they showed moderate affinity for sucrose, and SbVIN2 had a higher optimum reaction temperature.

4.2. Association analysis results

In leaves,the expression of SbVIN2 was much higher than that of SbVIN1 at all three tested stages and in both cultivars. These findings are not consistent with the results of GeneAtlas in the Phytozome database (https://phytozome.jgi.doe.gov) [38] and transcriptome database (http://sorghum.riken.jp/morokoshi/Home.html) [54]. In both databases the expression of SbVIN1 is higher than SbVIN2 in leaves.This inconsistency may be due to the circadian rhythm of VIN gene expression. For instance, in sugar beet, the expression of two VIN genes was highest at 8 h after the onset of light[55].The highest transcript level of Maize Ivr2 was at the beginning of the light period[33].In a pathogenic bacterial infection experiment in sorghum [56], seedlings were sprayed with 0.01% Tween-20 as control. In these control samples,the expression of SbVIN1 was highest at 12 h and then decreased, whereas the expression of SbVIN2 began to increase after 12 h.As the authors did not mention the time of starting the treatment,the only information was that the expression of two SbVIN genes showed differing changes with time. Further research on the impact of circadian on SbVIN genes would be a future priority.

The physical length of the perfect LD blocks in SbVIN1 was smaller than that in SbVIN2(Fig.6).Almost all the variant loci of SbVIN2 were in one LD block(Fig.6-D).This finding may be accounted for by the location of SbVIN1 at the end of chromosome 4, where there are higher recombination rates,whereas SbVIN2 is located in the middle part of chromosome 6 where recombination rates are presumably lower.

Variations in SbVIN1 showed significant associations with SL, SD, IN, and SW (Fig. 7), in agreement with the significant differences in stem phenotypes among different haplotypes(Fig. 8-A). VIN regulation of plant organ size and weight has been reported in tomato [7], muskmelon [8], cotton [9], and rice [10]. VINs hydrolyze one sucrose to two hexoses, which not only supply energy and carbon for cell division and expansion but also act as signals to modulate the expression of several hormones[2].For example,in GhVIN1-RNAi cotton,the expression of several MYB transcription factors and auxin signaling components changed [9]. SbVIN1 may also be involved in a profound network in plants and exert its regulating function on phenotypes.

At this her tongue broke loose again, as she abused him for his folly18 in taking the little one, and she never rested till her husband had described the exact way which led to the sparrow-princess s house

VIN genes determined the sucrose content and sucrose/hexose ratio in fruits that accumulated a large amount of sugar such as tomato[12,13],litchi[14],and strawberry[57].In sugarcane,a close relative of sorghum,VIN plays a crucial role in determining sucrose concentration in the stem [15,58,59].In sorghum, VIN activity declined to a low level at the maturation stage [23,24]. Still, differences in VIN activity were observed among different cultivars, and VIN activity was correlated with sucrose and hexose content [25]. Differences in SbVIN1 expression level among cultivars and the relationship between its expression level and sucrose content have also been reported[26].Thus,the VIN gene plays a role in determining sucrose and hexose content among sorghum cultivars. However, we did not find a significant association between the two genes and sucrose or glucose (Fig. S3).Instead,we found that SbVIN1 was associated with Brix(Fig.7)and SbVIN2 was associated with WSCs(Fig.9).

In this study,a significant association between SbVINs and grain traits was detected,especially for SbVIN1,which showed association with both grain weight and grain size.Mutation of a rice VIN gene (OsINV3), the ortholog gene of SbVIN1 in rice,led to smaller seeds[10].Detailed analysis of OsINV3 mutants showed that smaller grains were caused mainly by smaller cell size and not cell number [10]. The effect of SbVIN1 on grain-size traits may depend on mechanisms similar to those of rice homologs.

Declaration of competing interestThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.AcknowledgmentsWe thank all the other members of Hai-Chun Jing's lab for assistance in the fieldwork. We thank Prof. Xiu-Qing Li (Fredericton Research and Development Centre, Agriculture and Agri-Food Canada,Fredericton,NB E3B 4Z7,Canada)for his constructive suggestions and revisions to this article. This work was supported by the National Natural Science Foundation of China(No. 31471570, No. 31461143023) and Ministry of Science and Technology of the People's Republic of China(2015BAD15B03).Appendix A.Supplementary dataSupplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2019.06.012.