Dear editor,

RNA interference (RNAi) is a promising approach for developing insect‐resistant crops. In the first two proof‐of‐concept studies, DNA fragments derived from essential insect genes were constructed into plant expression cassettes as inverted repeats, enabling long double‐stranded RNAs (dsRNAs) to be transcribed in host plants (Baum et al., 2007; Mao et al., 2007). After ingestion, dsRNAs overexpressed in plants suppressed target gene expression via small interference RNA (siRNA)‐mediated RNAi in western corn rootworm (Diabrotica virgifera virgifera; Baum et al., 2007) and cotton bollworm (Helicoverpa armigera) (Mao et al., 2007), and thereby reduced their viability. Subsequent studies have mostly employed a similar approach and attempted to develop RNAi crops against various insect species (Liu et al., 2020). However, except for coleopterans that are generally susceptible to RNAi, most insects (e.g. lepidopterans, dipterans, hymenopterans and hemipterans) exhibit unpredictable responses to dsRNA‐induced RNAi, and this has become a hurdle for ubiquitous adaption of this strategy (Cooper et al., 2019). Potential factors influencing RNAi efficacy in insects involve dsRNA stability, cellular dsRNA absorption, core RNAi machinery integrity, systemic RNAi spread and target gene amenability (Cooper et al., 2019).

Although most studies on RNAi‐mediated insect‐resistant plants are based on dsRNA (Liu et al., 2020), limited studies have suggested that microRNA (miRNA)‐mediated RNAi is also practically applicable for insect‐resistant plant development. MiRNA‐mediated RNAi plants can be classified into four types based on the origin of miRNA effectors: artificially designed miRNAs (Guo et al., 2014), natural insect miRNAs (Jiang et al., 2017), natural plant miRNAs (Wamiq and Khan, 2018) and modified insect pre‐miRNAs (Bally et al., 2020).

Our previous study demonstrated that striped stem borer (SSB; Chilo suppressalis), a destructive insect pest of rice (Oryza sativa), is refractory to dsRNA‐induced RNAi but relatively susceptible to artificial miRNA (amiRNA)‐induced RNAi. Transgenic rice expressing SSB endogenous miRNA candidates with unknown functions retarded the growth of SSB larvae, without showing significant lethality (Jiang et al., 2017).

In this study, we successfully engineered highly SSB‐resistant rice (named csu260) expressing amiRNA of SSB endogenous miRNA csu‐novel‐miR260 through amiRNA expression technology. Csu‐novel‐miR260 negatively regulates ecdysteroid biosynthesis in SSB by suppressing the expression of disembodied (dib) that encodes a cytochrome P450 enzyme catalysing the C₂₂‐hydroxylation of 2, 22‐dideoxy‐3‐dehydroecdysone (Figure 1a; He et al., 2017). Two representative homozygous csu260 rice lines (csu260‐16 and csu260‐18) with a single‐copy insertion were obtained from 28 independent primary transformants.

A five‐day stemcutting feeding assay was performed in triplicate with 15 first‐instar SSB larvae for each replicate. It showed that csu260‐16 and csu260‐18 exhibited significant lethality against and retarded the growth of feeding SSB larvae compared with those fed the wild‐type (WT) recipient variety Zhonghua11 (Zh11) (Figure 1b). Then, a consecutive stemcutting feeding assay was performed in triplicate with 30 first‐instar SSB larvae for each replicate. It confirmed that csu260‐16 and csu260‐18 rice caused 55.6% and 53.3% larval mortality, respectively, at 35 days after feeding, whereas the mortality of the control larvae was less than 20% (Figure 1c). Correspondingly, the stem–loop quantitative polymerase chain reaction analysis showed consistently lower dib expression in the csu260‐fed larvae than in the control larvae throughout the feeding assay (Figure 1c).

We used small RNA sequencing to characterize the amiRNA expression in csu260‐16 and csu260‐18 (Figure 1d). The sequenced amiRNA reads were normalized as reads per million, and the normalized amiRNA abundance was 1.8‐fold higher in csu260‐16 than in csu260‐18. Consistently, csu260‐16 showed higher and more stable insect resistance than csu260‐18, and dib expression in SSB larvae fed csu260‐16 was lower than that in larvae fed csu260‐18 (Figure 1b, c). These results suggest that amiRNA dose is crucial for insect resistance of amiRNA plants. We used the osa‐MIR528 precursor as a backbone for constructing amiRNA expression vectors (Jiang et al., 2017). Small RNA sequencing showed that amiRNAs in csu260 rice are preferentially processed into 20‐ and 21‐nt long sequences, similar to mature osa‐MIR528 (21 nt), although the amiRNA sequence of csu‐novel‐260 originally constructed into amiRNA expression vector is 27‐nt long. The predominant 20 and 21 nt amiRNAs in cus260 rice possess the intact 5′ end of original csu‐novel‐260, but they lacked 7mer and 6mer at the 3′ end, respectively (Figure 1d). MiRNAs generally function effectively as long as the ‘seed sequence’ (positions 2–8 at the 5′ end of miRNA) matches between miRNA and the 3′ untranslated region (UTR) of messenger RNA (Brennecke et al., 2005). An in vitro feeding assay was performed using 100 nmol synthetic miRNA mimics (agomirs) with the same sequences as the predominant 20 (Ago20) and 21 nt (Ago21) sequences. The agomirs were delivered via five‐day germinated Zh11 seeds, which soaked in agomir solutions for 30 min before the assay. The result confirmed the efficacy of these trimmed amiRNAs (Figure 1e), indicating that the length of natural miRNAs would not be a major consideration when constructing the amiRNA expression vectors.

Moreover, we evaluated SSB resistance under field conditions, resembling practical rice production. Because SSB infestation causes ‘deadheart’ in the tillering stage and ‘whitehead’ in the heading stage, we conducted two independent field assessments of SSB resistance in these two stages. In both field assessments of SSB resistance, each of csu260‐16, csu260‐18, and Zh11 was grown in three plots as three replicates. The deadheart rate of csu260‐16 and csu260‐18 decreased by 62.9% and 50.4%, respectively, compared with that of the WT control, 35 days after the manual inoculation of 30 second‐instar SSB larvae per plant in the early tillering stage (Figure 1f). While the whitehead rate of the two csu260 lines decreased by 74.6% and 54.0%, respectively, compared with that of the WT control, 40 days after manual infestation with 1.4 SSB adult moths per plant in the early heading stage (Figure 1f, g). These results indicate that csu260 rice substantially prevented yield loss even under severe SSB infestation.

To evaluate agronomic performance in the field, each of cus260‐16, csu260‐18 and Zh11 was grown in three plots as three replicates. The result demonstrated that the most important trait yield per plant of two csu260 lines was not significantly different from that of the control. However, panicle length of cus260‐16 and plant height, panicle length and grains per panicle of csu260‐18 showed significant differences from those of the control (Figure 1h). Agronomic variations in transgenic lines are usually caused by somatic mutations occurring in transformation procedure. Therefore, most agronomic variations may be recovered via backcrosses with the WT and subsequent self‐pollinations.

Our results demonstrated that amiRNA‐induced RNAi is more applicable for controlling SSB than dsRNA‐induced RNAi. Because endogenous miRNAs naturally regulate their gene targets in insects, their use theoretically avoids certain factors that cause inefficient RNAi, such as refractory target genes and relevant RNAi machinery deficiency (Cooper et al., 2019). Moreover, critical siRNA and miRNA components associated with their biogenesis and function are significantly different (Cooper et al., 2019), potentially causing variations among insects in terms of small RNA stability, cellular uptake and intercellular transport between siRNA and miRNA. These factors could explain why SSB is resistant to dsRNA‐induced RNAi but relatively susceptible to miRNA‐induced RNAi. Taken together, amiRNA‐induced RNAi can be an alternative approach for controlling these insect pests insensitive to dsRNA‐induced RNAi, which would broaden the applicability of RNAi‐mediated insect pest management.

亲爱的编辑，

RNA干扰（RNAi）是开发抗虫作物的一种有前途的方法。在前两个概念验证研究中，源自必需昆虫基因的DNA片段以反向重复序列的形式构建到植物表达盒中，从而使长双链RNA（dsRNA）可以在宿主植物中转录（Baum et al。，2007）。 ; Mao et al。，2007）。摄入后，植物中过量表达的dsRNA通过小干扰RNA（siRNA）介导的RNAi抑制了西部玉米根虫（Diabrotica virgifera virgifera； Baum等人，2007）和棉铃虫（Helicoverpa armigera）（Mao）的靶基因表达。等人，2007），并因此降低了它们的生存能力。随后的研究大多采用了类似的方法，并尝试开发针对各种昆虫物种的RNAi作物（Liu等，2020）。但是，除了通常对RNAi敏感的鞘翅目外，大多数昆虫（例如鳞翅目，双翅目，膜翅目和半翅目）对dsRNA诱导的RNAi表现出不可预测的反应，这已成为普遍采用该策略的障碍（Cooper等。 ，2019）。影响昆虫RNAi效力的潜在因素包括dsRNA稳定性，细胞dsRNA吸收，核心RNAi机械完整性，系统性RNAi扩散和靶基因适应性（Cooper等，2019）。

尽管大多数关于RNAi介导的抗虫植物的研究都是基于dsRNA（Liu等人，2020年），但有限的研究表明，microRNA（miRNA）介导的RNAi也实际上可用于抗虫植物的发育。MiRNA介导的RNAi植物可根据miRNA效应子的来源分为四类：人工设计的miRNA（Guo等，2014），天然昆虫miRNA（Jiang等，2017），天然植物miRNA（Wamiq和Khan）。 ，2018）和修饰的昆虫pre-miRNA（Bally等，2020）。

我们之前的研究表明，条纹状茎bore（SSB; Chilo inhibitoralis）是水稻的毁灭性害虫（Oryza sativa），对dsRNA诱导的RNAi具有抵抗力，但对人工miRNA（amiRNA）诱导的RNAi相对敏感。表达具有未知功能的SSB内源性miRNA候选基因的转基因水稻可延缓SSB幼虫的生长，而无明显杀伤力（Jiang et al。，2017）。

在这项研究中，我们通过amiRNA表达技术成功设计了能表达SSB内源性miRNA csu-novel-miR260的amiRNA的高度抗SSB的水稻（命名为csu260）。Csu-novel-miR260通过抑制编码细胞色素P450酶的过氧化物酶（dib）的表达来负调节SSB中的蜕皮甾类生物合成，该酶催化2、22-二脱氧-3-脱氢蜕皮激素的C ₂₂羟基化（图1a; He等。 ，2017）。从28个独立的一级转化子中获得了两个具有单拷贝插入的代表性纯合csu260水稻品系（csu260-16和csu260-18）。

每重复一次，一式三份，每组重复15天的初生SSB幼虫，进行为期5天的stem割喂养试验。结果表明，与野生型（WT）受体品种中华11（Zh11）相比，csu260-16和csu260-18表现出对饲喂SSB幼虫的显着杀伤力，并阻碍了它们的生长（图1b）。然后，对于每个重复实验，一式三份地进行连续的stem割饲喂试验，每组重复30只初龄SSB幼虫。它证实csu260-16和csu260-18水稻在喂食后第35天分别引起55.6％和53.3％的幼虫死亡率，而对照幼虫的死亡率小于20％（图1c）。相应地，茎-环定量聚合酶链反应分析表明始终较低DIB 在整个饲喂试验中，csu260喂养的幼虫中的表达水平均高于对照幼虫（图1c）。

我们使用小RNA测序来表征amiRNA在csu260-16和csu260-18中的表达（图1d）。测序的amiRNA读数被标准化为百万分之一的读数，并且在csu260-16中标准化的amiRNA丰度比csu260-18中的1.8倍高。一致地，csu260-16表现出比csu260-18更高和更稳定的昆虫抗性，饲喂csu260-16的SSB幼虫的dib表达低于饲喂csu260-18-18的幼虫（图1b，c）。这些结果表明，amiRNA剂量对于amiRNA植物的抗虫性至关重要。我们将osa‐MIR528前体用作构建amiRNA表达载体的骨架（Jiang等，2017）。小RNA测序表明，尽管最初构建于amiRNA表达载体中的csu-novel-260的amiRNA序列是27nt长。cus260水稻中主要的20和21 nt amiRNA具有原始csu-novel-260的完整5'端，但在3'端分别缺少7mer和6mer（图1d）。只要“种子序列”（miRNA 5'端的2-8位）与miRNA与信使RNA的3'非翻译区（UTR）匹配，MiRNA通常就能有效发挥作用（Brennecke et al。，2005）。）。使用具有与主要20（Ago20）和21 nt（Ago21）序列相同的序列的100 nmol合成miRNA模拟物（agomirs）进行体外喂养试验。苦杏仁通过五天发芽的Zh11种子运送，在测试前将其浸泡在苦杏仁溶液中30分钟。结果证实了这些修饰的amiRNA的功效（图1e），表明在构建amiRNA表达载体时，天然miRNA的长度不是主要考虑因素。

此外，我们在田间条件下评估了抗SSB的能力，类似于实际的水稻生产。由于SSB的侵染会导致分causes期的“丧心”和抽穗期的“白头”，因此我们在这两个阶段进行了两次独立的SSB抗性实地评估。在两个对SSB抗性的现场评估中，csu260-16，csu260-18和Zh11分别在三个样地中作为三个重复样生长。在分till初期，每株植物人工接种30只二龄SSB幼虫后35天，与野生型对照相比，csu260-16和csu260-18的死心率分别降低了62.9％和50.4％（图1f）。在手动侵染40天后，两条csu260品系的白头率分别比WT对照降低了74.6％和54.0％。在抽穗初期，每株植物有4个SSB成年飞蛾（图1f，g）。这些结果表明，即使在严重的SSB侵染下，csu260水稻也基本上防止了产量损失。

为了评估该田间的农艺性能，将cus260-16，csu260-18和Zh11中的每一个在三个样地中作为三个重复样进行种植。结果表明，两个csu260品系的每株植物最重要的性状产量与对照无显着差异。但是，cus260-16的穗长和csu260-18的穗高，穗长和每穗粒数与对照相比有显着差异（图1h）。转基因品系的农艺变异通常是由转化过程中发生的体细胞突变引起的。因此，大多数农艺变异可通过与野生型回交和随后的自花授粉而得以恢复。

我们的结果表明，与dsRNA诱导的RNAi相比，amiRNA诱导的RNAi更适用于控制SSB。由于内源性miRNA在昆虫中自然调节其基因靶标，因此从理论上讲，它们的使用避免了某些导致RNAi效率低下的因素，例如难治性靶标基因和相关的RNAi机械缺陷（Cooper等人，2019）。此外，与它们的生物发生和功能相关的关键siRNA和miRNA组分也存在显着差异（Cooper等人，2019），可能会导致昆虫之间在小RNA稳定性，细胞摄取以及siRNA和miRNA之间的细胞间运输方面发生变化。这些因素可以解释为什么SSB对dsRNA诱导的RNAi有抵抗力，而对miRNA诱导的RNAi相对敏感。综上所述，amiRNA诱导的RNAi可能是控制这些对dsRNA诱导的RNAi不敏感的害虫的替代方法，这将扩大RNAi介导的害虫管理的适用性。