Suppression of starch synthase I (SSI) by RNA interference alters starch biosynthesis and amylopectin chain distribution in rice plants subjected to high temperature

Qin Zho, Xioxi Du,Zhnyu Hn,Yu Ye,Gng Pn,Muhmmd-Asd-Ullh Asd,Qif Zhou, Fngmin Cheng,*

aInstitute of Crop Science,College of Agriculture and Biotechnology, Zhejiang University,Hangzhou 310058,Zhejiang, China

bJiangsu Collaborative Innovation Center for Modern Crop Production,Nanjing 210095,Jiangsu,China

Keywords:Rice Starch synthase I RNA interference Grain quality Amylopectin High temperature

ABSTRACT Based on known cDNAs of rice starch synthase isoforms, we constructed dsRNA interference vectors for starch synthase I (SSI) to produce transgenic plants containing starch with a moderately high amylose content. We investigated the effect of SSI suppression on grain quality traits, starch biosynthesis, and amylopectin chain distribution in rice plants exposed to two different temperature regimes. The activities and transcripts of BEs, DBEs,and other SS isoforms were further investigated to clarify the effect of SSI suppression on these key enzymes and their specific isoforms under different temperature treatments. Suppression of SSI by RNAi altered grain starch component and amylopectin chain distribution,but it exerted only a slight effect on total starch content(%)and accumulation amount (mg kernel-1) and on starch granule morphology and particle size distribution. Under normal temperature (NT), insignificant differences in kernel weight, chalky kernel proportion, chalky degree, and starch granule morphology between SSI-RNAi line and its wild type (WT) were observed. However, amylose content (AC) level and granule-bound starch synthase (GBSS) activity in rice endosperms were markedly increased by SSI-RNAi suppression.The chalky kernel proportion and chalky degree of SSIRNAi lines were significantly higher than those of WT under high temperature (HT)exposure at filling stage. Inhibition of SSI by RNAi affected amylopectin chain distribution and raised starch gelatinization temperature (GT) in two ways: directly from the SSI deficiency itself and indirectly by reducing BEIIb amounts in an SSI-deficient background.The deficiency of SSI expression led to an alteration in the susceptibility of grain chalkiness occurrence and starch gelatinization temperature to HT exposure, owing to a pleiotropic effect of SSI deficiency on the expression of other genes associated with starch biosynthesis.

1. Introduction

Rice (Oryza sativa L.) is a starch-rich staple food that provides 35%-60% of caloric intake for about half of the world’s population [1]. Starch reserve in rice endosperm accounts for about 70%-80% of final endosperm weight [2,3]. Rice endosperm starch consists of two classes of polymer: amylose, a lightly branched linear molecule with a degree of polymerization of 1000-5000 glucose units, and amylopectin, a much larger polymer containing frequent α-1,6 branching linkages[4]. Normally, non-waxy starch in rice endosperm consists of 15%-25%amylose and 75%-85%amylopectin[4,5].The ratio of amylose to amylopectin and the fine structure of amylopectin play a critical role in determining the physicochemical properties of starch, including gelatinization temperature(GT) and gel consistency (GC) [5,6], and the palatability of cooked rice [6]. In rice endosperm tissue, starch biosynthesis from adenosine diphosphoglucose (ADP-glucose) is catalyzed by multiple enzymes. Amylose is synthesized by granulebound starch synthase (GBSS) encoded by the Waxy gene,whereas amylopectin biosynthesis is catalyzed by soluble starch synthases(SSs;EC 2.4.1.21),starch branching enzymes(BEs; EC 2.4.1.18), and starch debranching enzymes (DBEs;EC3.2.1.68)[7-10].

Ambient temperature during kernel filling is one of the most important environmental factors affecting starch biosynthesis and its composition in rice endosperm [5,10]. High temperature (HT) during the grain-filling period causes the formation of loosely packed starch granules with larger air space [11] and also reduces starch deposition in discrete granules [11], thereby leading to a reduction in kernel weight together with increased numbers of chalky and abnormal kernels [11-13]. More importantly, the palatability of cooking rice cultivars may be compromised by the alteration of starch composition in rice endosperm after rice plant exposure to high temperature during grain filling [14-16]. However, our understanding of the physiological processes underlying the effect of high temperature on starch composition and palatability is limited.

SS catalyzes the chain-elongation reaction of α-1,4-glucosidic linkage by transferring a glucose moiety from ADPglucose to the non-reducing end of the linkage [7,17]. In rice,there are eight distinct SS isoforms, which generally are categorized [18,19] into four types: one SSI isoform of the SSI type, three SSII isoforms (SSIIa, SSIIb, and SSIIc) of the SSII type, two SSIII isoforms (SSIIIa and SSIIIb) of the SSIII type,and two SSIV isoforms (SSIVa and SSIVb) of the SSIV type.These distinct SS isoforms share a highly conserved K-X-G-GL motif in the C-terminal domain that is responsible for substrate binding, but there is large variation in the N terminus upstream of the catalytic core [18,20]. Among the isoforms, SSI is the largest component of total soluble SS activity,accounting for ＞60%of this activity in developing rice endosperm [21]. Normally, SSI are ubiquitously expressed in endosperm and several other cereal tissues, while SSIIa and SSIIIa are expressed preferentially or solely in endosperm and reproductive tissues [20]. The other SS isoforms, including SSIIb, SSIIc, SSIIIb, and SSIV, are rarely detectable in the sink organs of many plant species [17-19]. Typical japonica rice lacks active SSIIa isozyme in consequence of two amino acid substitutions [18]. In maize, an SSI-deficient mutant shows only 25% of the total SS activity of its wild type, leading to a marked alteration in amylopectin chain-length distribution[22]. An in vitro experiment using recombinant rice SSI suggested that SSI generates chains with degree of polymerization (DP) 8-12 from very short (DP 6-7) chains during amylopectin biosynthesis in rice endosperm [23]. However,recent studies have also suggested that the function of SSI in starch biosynthesis extends beyond its catalytic activity [24],and that it participates in the regulation of other starch biosynthetic enzymes in cereal endosperm via proteinprotein interaction and/or other functions [25]. Furthermore,distinct SS isoforms play a synergistic role in the synthesis of amylopectin clusters [19], and a deficiency of an individual isoform may exert a pleiotropic impact on grain development and starch accumulation[26].For example,the loss of SSIIIa in maize endosperm leads to a deficiency of other enzymatic proteins(DBEs)in starch granules and to the sugary mutation phenotype [27]. Studies in rice and maize [19,22] have suggested that SSIIIa co-operates with SSI and/or SSIIa in the elongation of intermediate and long length chains of amylopectin. Recently, Liu et al. [28] have suggested that SSI is a central component of a protein complex interacting with SBEIIb and/or SSIIa in starch granule.But to date,the effect of HT exposure at the filling stage on kernel starch quality and amylopectin biosynthesis in the background of SSI deficiency remains poorly understood in rice and other cereal species.

In this study, we constructed a dsRNA interference vector for the OsSSI gene and introduced it into the rice genome to produce transgenic plants. Transgenic rice and wild-type plants were exposed to different temperature treatments at filling stage.Our objectives were 1)to investigate the effect of altering SSI expression on grain quality,starch accumulation,and amylopectin components in response to HT exposure during grain filling; and 2) to assess the accompanying alteration in the response of other starch synthesisassociated enzymes to HT exposure in the SSI deficiency background. The results were expected to illuminate the contribution of varying SSI expression to cooking and eating qualities in relation to HT exposure during grain filling.

2. Materials and methods

2.1. Plasmid construction and plant transformation

Fig.1-Generation and identification of SSI-RNAi transgenic lines.

DNA manipulations followed standard procedures as described by Cao et al. [15]. OsSSI (Os06g0160700) was cloned into pTCK303 at the BamH I, Kpn I, Sac I, and Spe I restriction sites (PCR primers: OsSSI-F: Tggtacc (Kpn I) actagt (Spe I)ATGGATGTGAAGGAGCAAGC;OsSSI-R:Tggatcc(BamH I)gagctc(Sac I)CGCATAGCACAGGAGTGTGT)(Fig.1-A).The vector was then transformed into a japonica rice cultivar(Nipponbare)via the Agrobacterium tumefaciens-mediated transformation method [21]. Twelve independent T0SSI-RNAi suppressed lines were screened for hygromycin(50 mg L-1)resistance and confirmed by PCR using hyg-specific primers (Fig. 1-B-D). To obtain homozygous SSI-RNAi generations (T2-Line 2, T2-Line 5, and T2-Line 8), transgenic plants were selected on 1/2 MS medium containing 50 mg L-1hygromycin by screening for a 3:1 segregation of the hygromycin-resistance phenotype in the T1and 100% hygromycin-resistance phenotype in the T2generation.

(A) Structure map of pTCK303-RiOsSSI expressing framework. (B) Electrophoregram of double digestion of recombinant final vector pTCK303-RiOsSSI by BamH I and Kpn I. “M”represents a DNA marker.“1”and“2”represented two repeats of the pTCK303-RiOsSSI vector. (C) Electrophoregram of the final vector pTCK303-RiOsSSI after transformation to EHA105 by PCR. “+” indicates a plasmid of recombinant final vector pTCK303-RiOsSSI (positive control). “-” represents a negative control(no template).“1”and“2”represent recombinant final vectors in Agrobacterium tumefaciens. (D) PCR confirmation of T0transgenic plants. “+” is a positive control (plasmid of pTCK303-RiOsSSI),lane 2 contains a negative control(untransformed Nipponbare). L1-L5 represent different lines of T0transgenic plants. (E, F) Total activity and Native-PAGE staining of SS in developing endosperm of three SSI-RNAi transgenic lines and wild type (Nipponbare) 14 days after heading. (G) Comparison of plant morphology and kernel shapes of SSI-RNAi transgenic lines and wild type.

2.2. Plant materials and temperature treatments

The T3generation of SSI-RNAi-Line 2 and its wild type(Nipponbare) were used as experimental materials. Temperature treatments were imposed in 2017 at the experimental station of Zhejiang University, Hangzhou, China. Rice seeds were sown in April and 30-day old seedlings were transplanted to different pots,with 12 pots for each genotype and 2-3 seedlings in each pot.Rice plants grown in pots were managed under normal water supply and fertilizer conditions in a greenhouse with natural light cycles and moderate temperatures (28 °C daytime/22 °C nighttime) until heading.At full heading stage,30-40 panicles with uniformity anthesis day were randomly selected and tagged. The pots were then moved into two phytotrons (Model PGV-36; Conviron, Winnipeg, Canada) to undergo different temperature treatments until maturity. One phytotron was designed for a hightemperature (HT) regime and the other for the normaltemperature (NT) control regime, with putatively optimum temperature for achieving optimal palatability. The daily mean temperatures were 32 °C for HT and 22 °C for NT,respectively.The diurnal change of temperature was designed by a simulation of daily temperature fluctuation based on natural climate. The daily maximum and minimum temperatures were set at 2:00 PM and 5:00 AM, with 36 °C and 28 °C for HT, 24 °C and 20 °C for NT, respectively. No climatic conditions but temperature treatment were varied between the phytotrons.The photoperiod was from 5:30 AM to 7:00 PM with 150 to 180 J m-2s-1of light intensity, and the relative humidity was maintained around 75%-80%with a wind speed of 0.5 m s-1.After the initiation of temperature treatment(the day of full heading), tagged panicles were sampled at 7-day intervals with 3-5 tagged panicles in each time sample.Samples were immediately frozen in liquid nitrogen and stored at -80 °C until further use. At maturity, the other tagged panicles were harvested for the determination of starch physicochemical properties.

2.3. Enzyme activity assays

Twenty dehulled rice grains, sampled at the 14th day after heading, were hand-homogenized at 4 °C in a chilled mortar and pestle with 5 mL extraction buffer(100 mmol L-1Tricine-NaOH(pH 7.5),8 mmol L-1MgCl2,2 mmol L-1EDTA,12.5%(v/v)glycerol;1%(w/v)PVP-40,50 mmol L-1β-mercaptoethanol).After centrifugation at 10,000 ×g at 4 °C for 20 min, the supernatant was used directly to determine the activities of SS, SBE, and DBE [21]. The pellet was resuspended in 2 mL extraction buffer for assay of GBSS activity [6]. GBSS activity was measured in the same manner as SS by monitoring the change in absorbance of NADPH at 340 nm [15]. All enzyme activity assays were repeated 3-4 times.

2.4. Activity staining of different isoforms on native-PAGE

Crude enzymes from immature rice kernels (14th day after heading) were used to identify different isozymes of SS, DBE,BE, and phosphorylase, and 10 μL of supernatant was subjected to Native-PAGE[21,29].For the identification of distinct SS isozymes, Native-PAGE was performed on a slab gel prepared with 7.5% (w/v; resolving gel) containing 0.8% (w/v)oyster glycogen (type II; Sigma, St. Louis, USA) and 3.3% (w/v;stacking gel) acrylamide as previously described by Fujita et al. [21]. Native-PAGE/activity staining of BEs and phosphorylase followed Crofts et al. [29]. The Native-PAGE/activity staining of DBEs followed Lin et al.[27].

2.5. RNA isolation, cDNA preparation, and quantitative RTPCR

RNA extraction and cDNA preparation followed Cao et al.[15].Quantitative real-time PCR was performed with a SYBR Green Real-time PCR Master Mix reagent Kit(Toyobo,Osaka,Japan).Reactions were performed on the Bio-Rad CFX96 Real-time system (Bio-Rad, California, USA) following the protocol provided by the manufacturer. The gene-specific primer pairs used in this study are listed in Table 1.The amplification of various genes was normalized by ACTIN(LOC_Os10g36650)expression and their relative expression levels were determined by the 2-ΔΔCTmethod [18]. The mean and standard error of experimental data were calculated from three independent biological replicates.

2.6. Determination of starch physicochemical properties and chain-length distribution of amylopectin

Measurement of chain-length distribution of amylopectin was performed by the FACE method [30]. A 5-mg sample of isolated amylopectin powder was digested with Pseudomonas amyloderamosa isoamylase (sigma-Alarice) and then labeled with 5 μL of a solution of 0.1 mol L-18-aminopyrene-1,3,6-trisulfonic acid (APTS) in 15% acetic acid and 10 μL of 0.5 mol L-1NaH3BCN. The mixture was then incubated at 65 °C for 3.5 h and diluted to total volume of 2 mL with anelectrophoresis buffer (50 mmol L-1Tris-H3PO4, pH 2.5). FACE was conducted with duplicate measurements for each sample,following a previously reported method[15,30].

Table 1-Primers used in the study.

Amylose content was measured by iodine colorimetry [5].Starch gelatinization properties were determined with a differential scanning calorimeter (DSC, Model DSC-7, Perkin-Elmer, Norwalk, CT, USA) equipped with an intra-cooling II system. The onset temperature (To), peak temperature (Tp),and conclusion temperature (Tc) were recorded and the gelatinization enthalpy(ΔH)was normalized to the dry weight of milled flour [14]. Granule size distribution of starches was determined with a Horiba LA-920 laser-scattering particle-size distribution analyzer(Malvern Instruments Inc.,UK)following Cao et al.[15].Granule size was expressed as number-average diameter automatically calculated from the instrument software.

2.7. Bimolecular fluorescence complementation (BiFC) assay and luciferase imaging observation for protein-protein interaction

For a BiFC assay, the N-terminal 173 amino acids and Cterminal 155 amino acids of yellow fluorescent protein (YFP)were fused to the pBWA(V)HS vector(provided by the BioRun company, Wuhan, China) to yield products named pBWA(V)HS-vn173 (YN) and pBWA(V)HS-vc155 (YC), respectively. The full-length coding sequence of OsSSI was inserted into the YN vector to generate YN-OsSSI, and the coding sequence of OsBEIIb was cloned and inserted into the YC vector to generate YC-OsBEIIb.Two proteins,AtToc159(NCBI,accession number 827934)and AtToc33(accession number 839248)were also inserted into the YN and YC vectors as positive controls[31]. Both the YN-OsSSI + YC and the YC-OsBEIIb + YN combinations of YN and YC represented negative controls.These vectors were subsequently transformed into A.tumefaciens strain EHA105 and co-expressed in N.benthamiana leaves.YFP and chloroplast fluorescence signal were detected by confocal microscopy (LSM710; Zeiss) after 2-3 days’ infiltration following Ueguchitanaka et al.[32].

2.8. Statistical analysis

All analyses were performed in triplicate unless stated otherwise. Data analysis was performed with SPSS 16.0(SPSS, Inc., Chicago, IL, USA). The data were subjected to variance analysis and the means were tested by least significant difference(LSD)at P ＜0.05.

3. Results

3.1. Generation, identification, seed phenotype, and grain quality of SSI-RNAi lines

SSI expression was effectively suppressed using RNAi technology, and the total SS activity of SSI-RNAi transgenic lines was markedly lower than that of its wild type,in which the SS activities of homozygous T2generations of three SS-RNAi lines(L2,L5,and L8)were reduced to 33.1%-67.8%of wild-type levels by the 14th day after heading (Fig. 1-E). Native-PAGE analysis of various SS isoforms indicated that the transgenic lines differed markedly from the wild type in the intensity of the protein band corresponding to SSI, with only the faint band for the three homozygous SSI-RNAi lines(L2,L5,and L8).In contrast,the intensity of the protein band corresponding to SSIIIa did not vary substantially between the three transgenic lines and the wild type(Fig.1-F).This result indicated that the marked reduction in total SS activity in SS-RNAi lines resulted from the reduced amounts of SSI isozyme and OsSSI transcript.

Under normal field growth, SSI-RNAi transgenic rice was visually similar to Nipponbare (wild type) in plant height,growth period,panicle morphology,and seed shape(Fig.1-G).There was no significant difference in rice yield traits,including available panicles per plant, kernel number per panicle, kernel weight and seed-setting rate, between SSIRNAi rice and its wild type (Table 2). However, the chalky degree of SSI-RNAi rice was slightly higher than those of its wild type. SS-RNAi rice also showed significantly higher AC than the wild type (Table 2). We further investigated the impact of SS deficiency on the chain length distribution of amylopectin and particle-size distribution of starch granules in mature endosperm (Fig. 2). The result showed that SSI suppression reduced the proportion of chains with DP 8-13,but it increased those of very short chain with DP 6-7 and chains with DP 16-21 (Fig. 2-A, B). Interestingly, scanning electron microscopy (SEM) imaging revealed that SSI-RNAi rice appeared similar to its wild type in morphology and space-packed structure of starch granules (Fig. 2-C, D). The insignificant difference in particle-size distribution and meandiameter of starch granules was also observed between SSIRNAi rice and its wild type (Fig. 2-E). These results suggested that SSI-RNAi strongly influenced the starch composition of rice endosperms,such as the ratio of amylose to total starch,and the chain length distribution of amylopectin, but it did not disturb the starch granule morphology and particle-size distribution in rice endosperm.

Table 2-Comparisons of some important agronomic traits and grain qualities between SSI-RNAi lines and their wild type.

3.2. Differential response of starch properties to environment temperature for SSI-RNAi rice and its wild type

High temperature exerted a strong effect on the starch composition and also starch granule structure of rice kernels,in which the AC level of HT-ripened kernels was significantly lower than that in NT-ripened ones,while the opposite trend was true for the chalky kernel proportion and chalky degree of HT-ripened kernels (Table 3). This result is in accord with previous reports[6,9].Interestingly,the chalky kernel proportion and chalky degree of SSI-RNAi were more susceptible to HT exposure than those of its wild type,as reflected by the Dvalues and relative differences between the two temperature regimes (Table 3). HT exposure resulted in a significant increase in starch GT and ΔH for both SSI-RNAi and its wild type.However,the extent of increase in GT and ΔH induced by HT exposure was more substantial for SSI-RNAi relative to its wild type (Fig. 3-A, B). In contrast, HT exposure narrowed the range of AC variation between SSI-RNAi and its wild type.For example,the AC level of SSI-RNAi varied from 19.6%under NT to 17.2% under HT, while the AC level of its wild type ranged from 17.3% under NT to 14.1% under HT, with approximately 15.6%of reduction in the AC level of HT-ripened grains(Table 3). Thus, the AC level of SSI-RNAi under HT exposure (17.2%)was similar to that of its wild type under the NT regime(17.3%).

Fig.2-Difference in the chain length distributions of amylopectin(A and B),scanning electron microscopy(SEM)of the starch granules(C and D),and particle-size distribution of starch granules(E)in mature endosperm of wild type and SSI-RNAi.Panel A illustrates the genotypic difference in chain length profile. Panel B shows the difference between wild type and SSI-RNAi in values of chain length fractions.Bar = 20 μm.

Table 3-Difference in chalky kernel proportion (%),chalky degree (%), and amylose content (%) between SSIRNAi and its wild type (WT) as affected by high temperature during grain filling.

HT exposure resulted in an increase in the mean diameters of starch granules, with an increased proportion of larger starch granules for both SSI-RNAi and its wild type (Fig. 3-C,D). However, only slight differences in starch granule morphology and particle-size distribution were observed between the two genotypes under the same temperature regime (Fig. 3-C-E). Thus, the effect of HT exposure on starch granule morphology and particle-size structure was much larger than the effect of SSI mutation. HT exposure significantly reduced the proportions of amylopectin short chains with DP 6-7 and short-intermediate chains with 10 ≤DP ≤18,but led to an increase in the proportions of the intermediate and long chains with DP ＞21 for both SSI-RNAi and its wild type (Fig. 3-F, H, I). The effect of HT exposure on the chainlength distribution of amylopectin varied somewhat between SSI-RNAi and its wild type. For example, HT-induced decline in the proportion of the short chains with DP ＜9 became much larger for SSI-RNAi relative to its wild type,as shown by their varying range of DP molar (%) between the two temperature regimes(HT and NT)(Fig.3-G,I).

3.3. Pleiotropic effects of SSI-RNAi silencing on other starchsynthesizing enzymes and their major isoforms under different temperature treatments

As shown in Table 4, SSI-RNAi showed significantly higher GBSS activity than its wild type under the same temperature regime, coinciding with the increased AC level for SSI-RNAi.However, the GBSS activities of both SSI-RNAi and its wild type were significantly reduced by HT exposure.Interestingly,SSI-RNAi showed a marked reduction in BE activity,accompanied by lowered SSI expression and total SS activity.Under HT exposure, the activities of SS and BE in SSI-RNAi were still significantly lower than those of its wild type. In contrast,SSI-RNAi showed a higher level of DBE activity than the wild type, with an elevated change in DBE activity under HT exposure. These results indicated that SSI-RNAi mutation could cause simultaneous alteration in the activities of various key enzymes responsible for amylose and amylopectin biosynthesis in rice endosperm,under which the activities of SS and BE were suppressed by SSI-RNAi, but those of GBSS and DBE were up-regulated by SSI-RNAi.

Fig.4 depicts the activities of distinct isozymes of SSs,BEs,and DBEs induced by the suppression of SSI-RNAi and their response to HT exposure at filling stage.No visible difference in the intensity of protein bands corresponding to SSIIIa was detected between SSI-RNAi and its wild type under the same temperature regime(Fig.4-A).However,the amounts of SSIIIa isozyme under HT exposure were substantially lower than those under NT exposure, a response similar to that of SSI isozyme to HT exposure (Fig. 4-A). This result suggested that the deficiency of SSI isozyme exerted little influence on the expression amount of SSIIIa isozyme in filling rice endosperm, without a visible increase in the SSIIIa amount in compensation for SSI deficiency. Indeed, HT exposure significantly suppressed the expression of SSIIIa isozyme, even if the expression of SSI isozyme was severely inhibited by RNAi(Fig.4-A).

Among different BE isozymes, SSI-RNAi differed evidently from its wild type in the amount of BEIIb isozyme, and HT exposure led to a marked decrease in BEIIb level for SSI-RNAi(Fig. 4-B). In contrast, negligible difference in the amounts of BEI and BEIIa was observed between SSI-RNAi and its wild type under the same temperature regime and also for the two rice genotypes subjected to HT exposure (Fig. 4-B). Notably,the amount of BEIIb isozyme in SSI-RNAi decreased more dramatically than that in its wild type when rice plants were exposed to HT (Fig. 4-B). This result indicated clearly that a deficiency of SSI could cause a concomitant decline in the expression amount of BEIIb isozyme in filling endosperm,with a very low level of BEIIb isozyme for SSI-RNAi under HT exposure. We speculated that the markedly lowered BEIIb activity derived from the inhibition of SSI by RNAi might play an important role in SSI-RNAi induced alteration in amylopectin chain distribution under different temperature regimes, in addition to the direct effect of SSI deficiency on amylopectin biosynthesis.

Two distinct DBE, including isoamylase (ISA) and pullulanase(PUL),are crucial in the amylopectin biosynthesis in cereal endosperm [6]. SSI-RNAi showed relatively higher amounts of ISA and lower amounts of PUL isozymes than its wild type under the same temperature regime (Fig. 4-C). HT exposure increased ISA expression but reduced PUL expression (Fig. 4-C). Compared to the effect of SSI-RNAi on the expression of ISA isozyme, HT exposure had a stronger influence on the expression amount of ISA isozyme and PUL for SSI-RNAi (Fig. 4-C). These results suggested that the HTinduced increase in the total activity of DBE for SSI-RNAi(Table 4) was due mainly to the elevated expression of ISA isozyme,and also that the elevated ISA expression might play a complementary role in the maintenance of total DBE activity, because the suppressed expression of PUL isozyme was obviously shown for both HT exposure and SSI-RNAi.

Fig.3- Differential response of starch properties to environmental temperature in SSI-RNAi rice and its wild type.(A,B)differences in DSC profiles of rice starch between different temperature regimes (HT and NT).Tp(WT-NT) = 69.7 °C, Tp(WTHT) = 74.3 °C,Tp(SSI-RNAi-NT) = 70.4 °C,Tp(SSI-RNAi-HT) = 75.8 °C;ΔH(WT-NT) = 5.83 J g-1,ΔH(WT-HT) = 7.21 J g-1,ΔH(SSIRNAi-NT) = 6.31 J g-1,ΔH(SSI-RNAi-HT) = 8.36 J g-1.(C,D)Differences in mean diameter and size distribution (volume-based proportion)of starch granules between HT and NT.(E) SEM of starch granules.(F-I)difference in chain length distribution of amylopectin between HT and NT.HT and NT denote high-temperature and normal-temperature regimes,respectively.F and H indicate temperature regime differences in chain length profile.G and I show chain length fraction differences between HT and NT.

Table 4-Difference between SSI-RNAi and its wild type(WT) in the enzymatic activities (mmol·min-1 endosperm-1) of SS, GBSS, BE, and DBE in developing endosperm as affected by high-temperature exposure of kernels at the filling stage.

The transcripts of several important genes that encoded GBSS, SSI, SSIIa, SSIIIa, BEI, BEIIa, BEIIb, ISA1, and PUL in rice endosperm and their temporal pattern in response to HT exposure were further investigated by quantitative real-time(qRT)-PCR (Fig. 5). Suppression of SSI-RNAi reduced the transcripts of BEIIb, with a downregulation pattern of transcripts under HT exposure over the whole sampling stage,while the transcript amounts of the Wx gene encoding GBSS were markedly increased by SSI-RNAi and downregulated under HT exposure.The effects of SSI-RNAi on the transcripts of BEI, BEIIa, ISA1, and PUL were somewhat variable, depending on sampling stage,and no apparent trend in the variation of SSIIa and SSIIIa transcripts was observed between SSI-RNAi and its wild type under the same temperature regime. These results indicated that the effect of SSI-RNAi and/or HT exposure on the transcripts of Wx and BEIIb was similar to that on the expression of GBSS and BEIIb enzymes at the protein level. We accordingly speculated that the increased GBSS activity and lowered BEIIb isozyme expression induced by SSI-RNAi might result from the alteration of Wx and BEIIb at the transcription level.The BiFC assay for protein and protein interaction confirmed that a strong YFP fluorescence signal was detected for the combinations of YN-OsSSI + YC-BEIIb and positive control, with green fluorescence and yellow fluorescence being shown for the merged image, whereas no green fluorescence signal was detected for the negative control (Fig. 6-A). This finding strongly suggested that SSI and BEIIb co-operated closely with each other at the protein level, possibly participating in the formation of SSI/BEIIb protein complexes with other starch biosynthetic enzymes.Considering the formation of protein complexes as generally being controlled by protein phosphorylation, we further investigated the activity of phosphorylase in filling endosperm by Native-PAGE. HT exposure markedly inhibited the amount of phosphorylase in rice endosperm (Fig. 6-B). These findings suggested that there was a close interaction between SSI and BEIIb during amylopectin biosynthesis in rice endosperm and that formation of protein complexes containing SSI/BEIIb might be severely impeded by HT exposure in the SSI-RNAi background. In this scenario, the deficiency of SSI isozymes possibly altered the binding of BEIIb to starch granules and simultaneously lowered BEIIb on amylopectin biosynthesis and chain distribution.

4. Discussion

Fig.4-SS isoenzyme patterns(A),BE isoenzyme patterns(B),and DBE isoenzyme patterns(C)in developing endosperm of from SSI-RNAi and wild type at 14 days after heading,following exposure to different temperature regimes(HT and NT).

Fig.5-Temporal patterns of transcript expression of GBSSI,SSI,SSIIa,SSIIIa,BEI,BEIIa,BEIIb,ISA1,and PUL in developing rice endosperm of WT and SSI-RNAi exposed to different temperature regimes(HT and NT).

Previous studies[4,7]have shed light on the specific functions of individual enzyme isoforms of SSs in starch biosynthesis metabolism. SSI is primarily responsible for the synthesis of the shorter chains of amylopectin.Further elongation to form DP ＞20 chains is performed by other SS isozymes, including SSIIa and SSIIIa [8,9].However,the functions of distinctive SS isoforms (SSI, SSIIa, and SSIIIa) may be overlapping and/or cooperative in the synthesis of amylopectin cereal endosperm[19,22,29]. For example, studies [27,28] in maize revealed that SSIIIa could compensate for most of SSIIa function in longerchain elongation in the presence of SSIIa deficiency. In rice,loss of SSIIIa led to increased SSI activity in developing endosperms [29].In the present study, the suppression of SSI isozyme in the RNA-SSI mutant significantly reduced the abundance of amylopectin chains of DP 8-13 and enriched the amylopectin chains of DP 6-7 and DP 16-21, while altering only slightly the proportion of chains with DP ＜22(Fig.2-A,B).A largely similar impact of SSI suppression by RNAi on the amylopectin chain profile was previously reported in rice[21,23], maize [27], and wheat [33], although the decrease in DP ＞22 chains was less evident in SSI mutant lines than in their control lines as previously described by Fujita et al. [21].However, our present result indicated clearly that the deficiency of SSI exerted only a slight effect on kernel weight and total starch accumulation in rice endosperm (Table 2),and that the suppression of SSI by RNAi did not alter starch granule morphology and particle-size distribution under the same temperature regime (Fig. 2-C-E). In a previous study in wheat, suppression of SSI using RNAi technology altered granule morphology, inducing a marked decline in the small B-granule population and an increase in the large-A granule population [33]. In an Arabidopsis SSI mutant [34], SSI deficiency led to a compensatory increase in the activities of the other SS isoforms SSIIa and SSIIIa.However,in the present study, the SSI-RNAi rice mutant did not differ markedly from its wild type in the amounts of other SS isozymes(Fig.1-F).We conclude that the effect of SSI deficiency on granule morphology and particle-size distribution, as well as the pleiotropic effects of SSI deficiency on other SS isozymes, varies widely among plant species. In contrast to Arabidopsis, rice did not show a marked increase in the amount of SSIIIa expression in developing endosperm in the SSI-deficient background. The differences among plant species in the influence of SSI-RNAi on starch particle size distribution and granule morphology could be partly explained by species-dependent differences in these traits, with a wider size distribution in wheat than in rice under normal growth [4,33]. In previous studies [35,36],the gloomy and somewhat dull appearance,referred to as the dull phenotype,of the endosperm starch of the du1 mutants of maize and rice was caused by SSIIIa deficiency. In other studies [37-39], rice and barley mutants for isoamylase 1(sugary-1) showed striking changes in starch granule morphology and particle-size distribution, in addition to altered chain length. In the present study, the chalky kernel proportion and chalky degree of SSI-RNAi rice lines were slightly different from those of its wild type under NT (Table 2),without the marked alterations in granule morphology and particle-size distribution shown by SSI-RNAi (Fig. 3-C-E).However, HT-induced increase in chalky kernel proportion and chalky degree was more pronounced for SSI-RNAi relative to its wild type.These findings suggested that the suppression of SSI by RNAi might alter the susceptibility of kernels to chalkiness under HT exposure.

Fig.6-Interaction between OsSSI and OsBEIIb in tobacco leaves(A)and PHO isoenzyme patterns in developing endosperm at 14 days after heading in SSI-RNAi and wild-type exposed to HT and NT(B).GF,green fluorescence;DIC,differential interference contrast bright field;Merged,overlap of GF and DIC.

High temperature (HT) reduced AC levels in endosperm of non-waxy rice, a phenotype closely associated with HTinduced decline in GBSS activity and Wx transcript in developing rice endosperms [15,40]. Complete deficiency of GBSSI led to glutinous rice grains containing starch composed exclusively of amylopectin [29]. In contrast, absence of the BEIIb gene in maize, rice and other cereal species, referred to as the amylose-extender (ae) mutation, led to an increased-AC phenotype, with a moderate increase in the ratio of amylose to amylopectin in cereal endosperm [41-44]. In the present study,inhibition of SSI by RNAi resulted in a marked increase in AC level in rice endosperm,in addition to the altered chain length distribution of amylopectin in SSI-RNAi(Table 2).These might be indirect effects of the SSI deficiency on other genes of starch synthesis for SSI-RNAi rice, given that SSI catalyzes the initial short chain elongation of amylopectin[4,7],and this biochemical process was not directly associated with the altered AC level in cereal endosperms[35].However,SSI-RNAi also showed a significant increase in Wx transcripts and higher GBSS activity over the Nipponbare wild type under the same temperature regime. (Table 4, Fig. 5). These findings agree with those of Li et al.[45],who reported that knockdown of the SSI gene increased the AC level of rice grains [45]. The markedly reduced BEIIb expression at both transcriptional and enzymatic protein levels was accompanied by SSI-RNAi suppression (Figs. 4-B, 5). We accordingly propose that the effects of SSI deficiency on amylose synthesis and AC level in rice endosperm may be attributed to two factors: a marked increase in GBSS activity and/or a marked decrease in BEIIb expression, given that both elevated GBSS activity and repressed BEIIb activity could increase the ratio of amylose to total starch in cereal endosperms[39,41].Interestingly,both the activity of GBSS and the expression amount of BEIIb isozyme were downregulated by HT exposure(Table 4,Figs.4 and 5). By contrast, SSI-RNAi displayed a more severely suppressed amount of BEIIb isozyme than its wild type (Figs.4-B,5).Hence,the HT-induced decrease in AC level as a result of the lowered GBSS activity under HT exposure might be partly counteracted by the opposite effect of the lowered BEIIb expression on AC level, given that severe BEII inhibition generally leads to a moderate increase in AC level in cereal endosperm [42,46]. For this reason, the AC level of SSI-RNAi was less sensitive to HT exposure than that of Nipponbare(wild type).

The gelatinization property is one of the most important indicators of the cooking quality and processing characteristics of rice starch. Studies of the relationship between amylopectin chain distribution and GT behavior have suggested that a fraction of A chains with DP ≥10 is necessary for the formation of the crystalline structure [4,13] and that a large proportion of short chains with DP 6-7 reduces the crystallinity of starch [18,26]. In the present study, SSI-RNAi displayed higher values for peak temperature(Tp)and ΔH than its wild type under the same temperature regimes(Fig.3-A,B).Tpand ΔH increased from 69.7 °C and 5.83 J g-1for Nipponbare to 70.4 °C and 6.31 J g-1for SSI-RNAi, respectively. This result suggests that the extent of alteration in Tpand ΔH for SSIRNAi under NT growth was much smaller than those of the rice mutants for isoamylase1 (sugary-1) and SSIIa (alk) [37,38].Interestingly, the effect of HT exposure on starch gelatinization behaviors (To, Tp, Tc, and ΔH) was also more profound than that of SSI-RNAi, with significant increases in gelatinization temperature and ΔH under HT exposure (Fig. 3-A and B). The effect of HT exposure on starch GT behavior might be explained by their alteration in the chain-length distribution profile of amylopectin (Fig. 3-F-I), given that the lowered abundance ratio of short A chains (DP ＜7) and shortintermediate chains (10 ＜DP ＜18) to intermediate B chains(DP 20-25) was previously [13,15] considered the major factor controlling GT differentiation between different temperature regimes. In a previous study [44], deficiency and inactivation of BEIIb reduced the short chains of DP 8-14 and increased GT,particularly at the onset temperature (To) in starch gelatinization resistance. Studies[41,43] in the ae mutant of rice and maize showed that BEIIb was mainly responsible for the formation of starch crystalline structure. BEIIb activity was positively correlated with the proportion of short (DP 6-7)chains of amylopectin, but negatively associated with the proportion of intermediate (DP ＞18) B chains and the GT behavior of starch crystallinity [43,44]. The rice ae mutation with BEIIb deficiency has been shown [42] to cause a 50%decrease in SSI activity. In the present study, the extent of increases in GT and ΔH induced by HT exposure was more substantial for SSI-RNAi than for its wild type (Table 3). The expression of BEIIb was more severely repressed by HT exposure in SSI-RNAi than in its wild type (Fig. 4-B). We conclude that the increase in GT and ΔH in SSI-RNAi under HT exposure was not caused only by the direct effect of SSI deficiency and that the severely suppressed BEIIb in SSI-RNAi rice also was strongly responsible for HT-induced increase in GT and ΔH for SSI-RNAi. Indeed, our use of the BiFC assay in conjunction with fluorescence signal observation provided evidence that SSI and BEIIb co-operate closely with each other at the protein level(Fig.6-A).Protein-protein combinations in starch synthesis are assembled into high-molecular-weight complexes in a phosphorylation-dependent manner [45-47].The interaction between SSI and BEIIb in rice endosperm,which may participate in the formation of SSI/BEIIb protein complexes in cereals [28,48], could be impeded by HT exposure, owing to HT-induced inhibition of protein phosphorylation in the SSI-RNAi background (Fig. 6-A). Hence, the binding of BEIIb to starch granules might be also affected by SSI deficiency. More work is needed to elucidate the mechanism by which the precise regulation network of starchsynthesizing isozymes affects starch gelatinization resistance and amylopectin chain length in relation to HT exposure during grain filling.

5. Conclusions

SSI-RNAi suppression strongly influenced the starch composition of rice endosperms,expressed as the ratio of amylose to total starch,and the chain length distribution of amylopectin.However,seed phenotype,starch granule morphology,starch accumulation, and particle-size distribution in the rice endosperms of SSI-RNAi rice were almost identical to those of the wild type. Inhibition of SSI by RNAi altered susceptibility to chalky kernel and AC level under HT exposure. Compared with Nipponbare (wild type), SSI-RNAi rice was more susceptible to chalkiness and less susceptible to AC under HT exposure during grain filling. The inhibition of SSI by RNAi affected amylopectin chain distribution, starch gelatinization behaviors(GT and ΔH),and their susceptibility to HT exposure in two ways:directly by the SSI deficiency itself and indirectly by the synergistic reduction of BEIIb abundance in the SSIdeficient background.

Acknowledgments

The authors are deeply indebted to the National Key Research and Development Program of China(2017YFD0300103)and to the National Natural Science Foundation of China (31571602,31871566) for its financial support to this research project.