Generation of non-transgenic rice lines deficient in BEI and BEIIb
Rice is one of the top three crops with the highest production in the world, and serves as the major dietary source of carbohydrates in Asian and African countries (Fairhurst and Dobermann 2002; Awika 2011). Given the recent increase in the number of diabetic patients worldwide, new rice lines allowing the restriction of carbohydrate intake are being developed (Wang et al. 2017). Among the various starch biosynthetic enzymes, it is clear that reduction in the level of amylopectin biosynthesis enzymes, BEIIb and/or SSIIIa, leads to high amylose content (Wang et al. 2017; Fujita 2014). Importantly, suppression of multiple BE isozymes allows the accumulation of amylose and RS to high levels in maize (Li et al. 2008), wheat (Regina et al. 2006; Corrado et al. 2020), barley (Carciofi et al. 2012; Regina et al. 2010), and rice (described later). In a rat feeding trial, intake of RS improved gastrointestinal health (Zhu et al. 2012; Regina et al. 2006). and prevented sudden increase in blood glucose levels (Zhu et al. 2012). In rice, the expression of BEI and BEIIb genes has been suppressed using anti-sense RNA (Wei et al. 2010), artificial microRNA (Butardo et al. 2011), and RNA interference (RNAi) technology (Sawada et al. 2018), and the resulting transgenic lines showed drastic starch phenotypes compared with the WT. However, there are no information about the seed weight or agricultural traits in these transgenic rice lines, except that the materials produced by RNAi technology, which suppressing BEI and BEIIb, significantly reduced it seed weight (Sawada et al. 2018). Unlike these previous studies, we generated a non-transgenic be1 be2b double mutant rice line for more practical use by crossing single mutant lines that are completely deficient in BEI or BEIIb.
The be1 be2b double mutant showed higher seed weight than the be2b single mutant (Fig. 2) over three consecutive years (2016, 2017, and 2018; data not shown). There are at least three reasons why the seed weight of be1 be2b was heavier than that of be2b single mutant. The first possibility is that the absence of both BEI and BEIIb diminished amylopectin biosynthesis but enhanced amylose synthesis. Loss of BEIIb is accompanied by the reduction in SSI activity to 50%, which may be due to the absence of branches that serve as primers for SSI. SSI activity in be1 be2b was also lower than that in the WT (Miura et al., in preparation). Less amylopectin branches in be1 be2b result in less non-reducing ends to serve as a substrate for SS. This leads to less consumption of ADP-glucose by SS isozymes including SSI. Then, an abundance of ADP-glucose in amyloplasts enhances amylose biosynthesis by granule-bound starch synthase I (GBSSI). The second possibility is that be1 be2b may have inherited some seed weight traits from Taichung 65. The seed weight of Taichung 65, the parental WT genotype of be1, was slightly (1.1-fold) heavier than that of Kinmaze, the parental WT genotype of be2b (Fig. 2), and this tendency was observed in all three years (2016, 2017, and 2018; data not shown). The third possibility is the difference in flowering time. Plants of be2b (Kinmaze background) flowered in late August to early September, whereas those of be1 (Taichung 65 background) and be1 be2b flowered in mid-August. Therefore, it is possible that the be1 be2b double mutant inherited flowering loci from be1 (Taichung 65 background). Earlier flowering time guarantees higher temperature during the seed development, which results in higher activity of starch biosynthetic enzymes, thus enhancing starch synthesis (Fig. 2).
In addition, the growth and fertility rate of be1 be2b plants was similar to those of the WT (unpublished data). These data suggest that the presence of BEIIa alone is sufficient for plant growth and the accumulation of up to 60% endosperm starch compared with the WT. The agricultural traits of be1 be2b such as seed weight and growth are beneficial for commercializing the double mutant as an ultra-high RS rice cultivar, although further breeding such as a backcrossing with elite rice cultivars is necessary.
Effects of loss of both BEI and BEIIb on amylopectin structure and thermal properties
It is clear that the loss of BEIIb, among the three BE isozymes, drastically affects starch properties by decreasing amylopectin short chains (Nakamura 2002; Nishi et al. 2001; Takahashi and Fujita 2017). The be1 be2b double mutant synthesized less amylopectin chains with DP 10–20 and more amylopectin chains with DP > 20 than the be2b single mutant (Fig. 4), resulting in a higher ratio of long to short amylopectin branches (Table S1). These results are consistent with the GEMS-0067 lines in maize, which are thought to be be1 be2b double mutants (Li et al. 2008), and with the down-regulation of BEI and BEIIb genes in transgenic rice (Zhu et al. 2012; Sawada et al. 2018; Lin et al. 2019). Comparison of the theoretical curve of chain length distribution, obtained by the addition of the subtraction curves of be1 and be2b, with the actual curve revealed that the actual curve showed less chains with DP 10–20 and more chains with DP ≥ 40 (Fig. 4). This indicates a synergistic effect due to the loss of both BE isozymes, suggesting that in the absence of BEIIb, the remaining BE enzymes (BEI and/or BEIIa) may generate amylopectin branches with DP 10–20; however, when both BEI and BEIIb are absent, BEIIa alone may not be able to compensate for their absence, and the number of branches with DP 10–20 may also be diminished. The average and median values of amylopectin chain length distribution (Table S1) were DP 26–27 and DP 14, respectively, in WT cultivars, and DP 33.8 and DP 17 in be2b; however, both values were greater in be1 be2b (average, DP 36.7; median, DP 19), suggesting that amylopectin possessed extremely long double-helical branch chains in the absence of BEI and BEIIb.
These unique starch structures greatly influence the gelatinization temperature. High temperature and energy are required to unwind double-helical structure (Hizukuri 1986); therefore, the longer the amylopectin branches, the higher the gelatinization temperature. Most of double helices in amylopectin molecules consist of amylopectin branches with DP ≤ 24 within a cluster of amylopectin. The greater the number of short amylopectin chains (DP ≤ 12) and the lower the number of intermediate chains (13 ≤ DP ≤ 24), the lower the gelatinization temperature (Fujita 2014; Hayashi et al. 2015). The increase in the gelatinization temperature of be2b by 16 °C compared with the WT may be because be2b showed a significant decrease in amylopectin with DP ≤ 12 and a great increase in amylopectin with DP 13–24 (Table S1). Although be1 be2b had slightly less branches with DP ≤ 12 compared with be2b, branches with DP 13–24 in be1 be2b were significantly less than those in be2b (Table S1). However, the peak gelatinization temperature of be1 be2b was significantly higher than that of be2b by 7.5 °C. The amylopectin chain length responsible for gelatinization temperature is essentially DP ≤ 24 in the WT as described above. However, formation of even longer double helices in be2b (EM10) and be1 be2b likely resulted in an extremely high gelatinization temperature due to the significant increase in DP 25–36 in be1 be2b compared with that of be2b (Table S1). These trends are in agreement with the results of previous studies on maize GEMS-0067 lines (Li et al. 2008) and transgenic rice lines down-regulated for the expression of BEI and BEIIb genes (Lin et al. 2019).
Effects of loss of both BEI and BEIIb on amylose content
The apparent amylose content of be1 be2b was 51.7%, as measured by gel filtration chromatography. This value was higher than the current highest record of 45.1% in the non-transgenic ss3a be2b double mutant japonica rice (#4019) (Asai et al. 2014). The amylose content of a transgenic indica rice line, in which the expression of BEI and BEIIb was down-regulated, was 44.8%, based on gel filtration chromatography (Zhu et al. 2012), while that of maize GEMS-0067 lines was 83.1–85.6%, based on the iodine method (Li et al. 2008). Suppression of both BEI and BEIIb genes leads to higher amylose content compared with the suppression of BEIIb alone, which is in agreement with previous studies (Li et al. 2008; Wang et al. 2018). In these cases, loss of two major branching enzymes, BEI and BEIIb, involved in amylopectin biosynthesis likely resulted in enhanced amylose biosynthesis.
Rice cultivars with high amylose content can be divided into two types: high true amylose content cultivars and high extra-long amylopectin chain content cultivars (Horibata et al. 2004). Extra-long chain is a long linear glucan, similar to amylose but connected to amylopectin, synthesized by GBSSI which is encoded by Waxy (Wx) gene (Takeda et al. 1987). The difference in extra-long chain content is thought to be dependent on single nucleotide polymorphisms (SNPs) in the Wx gene (Crofts et al. 2019). The be1 be2b double mutant used in this study was low in extra-long chain (1.3%) (Table 2, Fig. S1) but high in true amylose content (50.5%). This was because be1 be2b harbors the Wxb (gbss1L) gene; this gene was inherited from either Kinmaze or Taichung 65 and contains a thymine instead of a guanine at the first nucleotide position of intron 1, resulting in a low level of GBSSI protein (Crofts et al. 2019).
The maize GEMS-0067 lines accumulate higher levels of amylose and intermediate components than the amylose-extender (ae) lines, in which intermediate components, i.e., glucan molecules, are smaller than normal amylopectin but are enriched with longer average branch chains (Li et al. 2008). The elution profile of debranched purified starch obtained by gel filtration chromatography showed that amylose in be1 be2b was eluted as a broad peak at 110–120 min, which did not return completely to the baseline, unlike that observed in be2b (Fig. S1). On the other hand, the elution profile of purified amylopectin in be1 be2b showed almost no peak around 110 min, indicating that glucans eluted from the purified starch at approximately 110–120 min may be low molecular weight amylose. Thus, starch in the rice be1 be2b double mutant may have some different characteristics from the intermediate chains in maize GEMS-0067 lines. The apparent amylose content (83–86%) and extra-long chain content (20–26.8%) of maize GEMS-0067 lines were much higher than those of rice be1 be2b, as measured by the gel filtration method (Li et al. 2008). Furthermore, suppression of all three genes encoding BE isozymes (BEI, BEIIa, and BEIIb) in transgenic barley resulted in a further increase in amylose content (99.1%), as measured by the iodine method (Carciofi et al. 2012). Considering these cases, amylose content of be1 be2b was relatively low; however, this is because be1 be2b harbors the Wxb gene that encodes low levels of GBSSI (Sano 1984), whereas maize and barley lines harbor the normal Wx gene, which encodes high levels of GBSSI. In indica rice harboring the Wxa gene, which encodes high levels of GBSSI, while both BEI and BEIIb activities were suppressed, the amylose content was 64.8%, as determined by the iodine method (Zhu et al. 2012). Although the amylose content measured by different methods cannot be directly compared, rice be1 be2b seems to contain lower amylose content than other plant species, regardless of the level of GBSSI. This suggests that additional factors prevent excess biosynthesis or accumulation of amylose in rice.
Effects of loss of both BEI and BEIIb on RS content
Amylopectin, the major component of starch, forms double helices using adjacent branches in its native state. When starch absorbs water and gets heated, it undergoes gelatinization. The double helices are then unwound, and gelatinized starch is easily broken down by digestive enzymes. When the gelatinized starch is cooled, nearby branches re-form double helices to produce retrograded starch, although those double helices may be imperfect. When the amylopectin branch chains are long, starch is harder to gelatinize, quicker to retrograde, and therefore less degradable. By contrast, when the amylopectin branch chains are short, starch is easier to gelatinize, slower to retrograde, and therefore more degradable. Previously, we showed that the loss of BEIIb increases the amount of long amylopectin chains, which form long double helices, leading to a significant increase in RS content compared with high amylose rice lines such as indica rice and SSIIIa deficient mutant rice (Zhou et al. 2016; Tsuiki et al. 2016).
In this study, the be2b single mutant showed significantly higher RS content than WT cultivars and be1, regardless of the product type and the method of preparation (raw or gelatinized rice flour and mashed or un-mashed steamed rice). Strikingly, the be1 be2b double mutant showed 2.6–6.8-fold higher RS content than be2b (Table 3), possibly because of the high apparent amylose content as well as high amylopectin long branches in be1 be2b. These findings in non-transgenic japonica rice, be1 be2b, are consistent with those in maize GEMS-0067 lines (Jiang et al. 2010) as well as in transgenic japonica rice lines suppressed from the expression of BEI and BEIIb genes (Lin et al. 2019).
Rice is utilized in many different food applications, in addition to the ordinary steamed rice. To reflect those differences in applications, the RS contents were analyzed using samples prepared by four different methods. Although chewing of foods greatly varies among individuals, the RS content of steamed rice would be intermediate between that of un-mashed and completely mashed steamed rice. Additionally, although the degree of gelatinization may vary with the application, such as bread, cookies, and noodles, the RS content of rice flour would be intermediate between that of raw and completely gelatinized rice flour. Comparison of RS content of rice samples prepared by different methods revealed that the RS content of all analyzed rice lines was the highest in un-mashed steamed rice (Table 3). This was perhaps because the digestive enzymes could not function effectively in un-mashed steamed rice, given the small surface area, and this is thought to be a typical feature of RS1. In the WT and be1 mutant, the second highest RS content was detected in mashed steamed rice and gelatinized rice flour, while the lowest RS content was detected in raw rice flour (Table 3), possibly because the digestive enzymes functioned effectively in raw flour, given its large surface area. It has been suggested that the majority of starch in the raw rice flour can be digested by α-amylase, even though it has A-type crystallinity. The reason why gelatinized rice flour showed higher RS content than raw rice flour may be because grinding with pestle and mortar after gelatinization resulted in the samples becoming sticky and slightly lumpy, leading to lower surface area for digestive enzymes. On the other hand, in be2b and be1 be2b, the second highest RS content was detected in raw rice flour, followed by mashed steamed rice, and the least in gelatinized starch. The reason why raw rice flour showed higher RS content than mashed steamed rice and gelatinized rice flour may be explained by two possibilities. The loss of BEIIb in be2b and be1 be2b mutants probably increased long amylopectin chains, which formed longer double helices with B-type crystallinity and were more resistant to degradation by digestive enzymes. The other possibility is that gelatinized rice flour and mashed steamed rice prepared from be2b and be1 be2b were rapidly retrograded; this is thought to be a typical feature of RS3. Although some lumps appeared during the grinding procedure, the lumps of be2b and be1 be2b were less sticky than that of the WT and be1 and could be easily suspended in the digestive enzyme solution; therefore, samples had a larger surface area for digestive enzymes to work effectively.
Effects of loss of both BEI and BEIIb on the structure of RS
Detailed analyses of RS structure have been performed only in maize GEMS-0067 and ae lines (Jiang et al. 2010). RS samples of GEMS-0067 lines treated with thermostable α-amylase at 95–100 °C contained two components: high molecular weight glucans (DP 840–951), including amylose and slightly branched glucans derived from intermediate components, and linear, low molecular weight glucans (DP 59–74) (Jiang et al. 2010). In this study, RS samples were prepared by treating rice samples with digestive enzymes at 37 °C. The remaining RS materials were debranched and analyzed by gel filtration chromatography. The RS structure of be2b rice flour showed almost no amylose peak, while that of be1 be2b rice flour showed a clear amylose peak, although the peak area of be1 be2b RS from raw rice flour was 30–40% of the peak area of be1 be2b purified starch. This may be because be1 be2b has a very high true amylose content (50.5%), and the amylose may have formed long double helices. Of the true amylose content, 30–40% could not be degraded by digestive enzymes and therefore maintained the original molecular weight of amylose, while 60% was partially degraded to lower molecular weight amylose; hence, the peaks at 145 and 160 min, corresponding to amylopectin long chains and amylopectin short chains, respectively, were higher than those of purified starch. It is also possible that the partially degraded amylose, which was of a similar molecular weight as the long amylopectin chains, formed double helices, thus avoiding degradation by digestive enzymes. The RS sample of be1 be2b rice flour showed a small peak at 160 min. This peak is thought to represent the remaining long amylopectin chains within the size of one amylopectin cluster, and corresponds to RS2. The first peak of the RS structure of mashed steamed rice from be1 be2b was diminished similar to that of be2b. This suggests that almost all gelatinized amylose can be degraded to lower molecular weight amylose. The peak at 120–130 min corresponding to low molecular weight amylose from purified starch was lower in RS prepared from mashed steamed rice of be1 be2b than those of purified starch. It can be speculated that the low molecular weight amylose can also be degraded to an even lower molecular weight. The RS prepared from mashed steamed rice of be2b showed a clear peak at 158 min, which corresponds to long amylopectin chains within the size of one amylopectin cluster, as described above. By contrast, the RS prepared from mashed steamed rice of be1 be2b showed a large peak at 130–170 min, with the tallest peak at 144 min. We speculate that this peak represents the degradation products of amylose, long amylopectin chains spanning two or more clusters, and chains within one cluster. The gelatinized starch was incubated with digestive enzymes at 37 °C. It is possible that starch retrograded and formed double helices with a variety of different sized glucans, thus transforming into RS3, which is more resistant to degradation by digestive enzymes. The type of glucan molecules that form double helices in raw rice flour and retrograded starch can only be speculated at this moment, and further detailed analyses of RS structure are required. In addition, the actual RS content and structure in processed foods should also be determined in future studies.