Transposable elements (TEs) contribute to gene regulation and phenotypic diversity in plants. Miniature inverted-repeat TEs (MITEs) are short, non-autonomous DNA transposons (100–800 bp) that are numerically the most abundant TEs in the rice genome, and tightly associated with at least 58% of rice genes (Lu et al., 2017). MITEs have been shown to be a major driver of gene expression changes (Castanera et al., 2023), and genome-wide association studies using MITE insertion polymorphisms may allow to dissect the underlying causal genes of agronomic traits (Castanera et al., 2021).

As MITEs are an important source of genetic variation, we hypothesized that genome editing (GE) of MITEs might be an efficient approach to generate novel alleles with altered gene expression for tuning crop traits. Two agriculturally important rice genes, growth-regulating factor 4 (GRF4) and stress-responsive NAC1 (SNAC1), were selected for testing this hypothesis. OsGRF4 could positively regulate yield-related traits (Wang et al., 2022) and has a 294-bp PIF/Harbinger superfamily MITE inserted within 1200 bp 3′ to the stop codon (Figure S1). OsSNAC1 can confer salt stress tolerance (Hu et al., 2006), whereas no MITEs were detected in its upstream and downstream untranslated regions (UTRs) (Figure S2).

Since MITEs in the 3′ UTRs of certain rice genes have been revealed to mediate translational repression of target genes (Shen et al., 2017), we proposed that the downstream MITE of OsGRF4 could be excised by CRISPR/Cas9 to generate an overexpression allele, and designed a deletion vector transformed into rice calli. An average mutation frequency (35.4%) was achieved in the T₀ transgenic plants, carrying homozygous (10.4%) or heterozygous (16.7%) deletion mutations. Finally, we obtained two homozygous transgene-free T₂ OsGRF4^mite lines L1 and L2 (Figure 1a; Figure S3). OsGRF4 mRNA levels in the OsGRF4^mite lines were comparable to those of wild type (WT) (Figure 1b). However, OsGRF4 protein levels of OsGRF4^mite lines were higher than that of WT (Figure 1c; Figure S4). We compared agronomical traits between OsGRF4^mite and WT plants grown under field conditions. Plant height of OsGRF4^mite lines decreased significantly compared to WT plants but the productive tiller number (PTN) per plant increased (Figure 1d–f). Thousand-grain weight (TGW) of the OsGRF4^mite lines increased 6.4% on average compared to WT. This increase is accompanied by a slight increase in grain length, but not in grain width (Figure 1g,h; Figure S5). A small decrease in seed setting rate (SSR) was observed in OsGRF4^mite L1 and L2 lines compared to WT, with an average decrease of 8.6% and 9.3%, respectively (Figure 1i). In general, OsGRF4^mite plants slightly increased grain yield per plant (Figure 1j), which was also observed in OsGRF4-overexpressing plants (Wang et al., 2022). These results showed that the MITE deletion in OsGRF4^mite plants could increase OsGRF4 abundance to improve rice agronomic traits.

Some MITEs in the 5′ UTRs of rice genes have previously been reported to act as enhancers, such as the miniature Ping (mPing) TE, which could confer salt stress inducibility on nearby genes in rice (Naito et al., 2009). Therefore, we attempted to insert the 430-bp mPing into salt-tolerance gene OsSNAC1. Recently, an efficient approach to inserting large DNA fragments was developed by combining CRISPR/Cas9 with phosphorothioate-modified 3′-overhang double-stranded oligodeoxynucleotides (dsODNs) (Han et al., 2023). Using the above method to create the OsSNAC1^MITE allele, an sgRNA target site at 53-bp upstream of the OsSNAC1 start codon was designed (sgRNA-1), and the corresponding CRISPR/Cas9 plasmid was constructed (Figure 1k). We synthesized dsODNs containing the mPing with five consecutive phosphorothioate modifications and 10-bp 3′-overhang complementary to the resected overhang induced by the Cas9. The mPing dsODNs were then delivered into rice calli together with the CRISPR/Cas9 vector by particle bombardment. A total of 81 independent T₀ transgenic plants were obtained. We found that five plants (6.2%) had targeted insertions in the intended orientation and two plants (2.5%) with the reverse orientation (Figure 1l). Two independent T₂ homozygous targeted lines, OsSNAC1^MITE L1 and L2, were obtained for further analysis (Figure 1m). Under control conditions, there were no obvious differences in OsSNAC1 mRNA levels between OsSNAC1^MITE and WT plants. However, after 1 h of salt stress, the relative mRNA levels of OsSNAC1 in OsSNAC1^MITE L1 and L2 were 1.9- and 2.3-fold that of WT, respectively (Figure 1n). Consistent with this, OsSNAC1^MITE lines showed higher survival rates than the WT plants under high-salinity conditions (Figure 1o,p). These results suggested that the mPing insertion in OsSNAC1^MITE plants confers enhanced salt-inducible gene expression, thereby increasing salt tolerance.

In summary, we have shown that genetic manipulation of MITEs in rice could create different beneficial alleles to regulate gene expression and improve crop traits. We have been able to engineer CRISPR/Cas9-targeted loci to achieve site-specific MITE deletion or insertion, thus enabling the regulation of target genes by exploiting the ability of MITEs to control gene expression. Given the widespread presence of MITEs in many plant genomes, it is conceivable that this strategy could be used more widely in the future to optimize plant development and improvement.

转座元件（TE）有助于植物的基因调控和表型多样性。微型反向重复 TE (MITE) 是短的非自主 DNA 转座子 (100-800 bp)，是水稻基因组中数量最多的 TE，并且与至少 58% 的水稻基因紧密相关（Lu等，2016） 。 ， 2017 ）。 MITE 已被证明是基因表达变化的主要驱动因素（Castanera等， 2023 ），使用 MITE 插入多态性的全基因组关联研究可能有助于剖析农艺性状的潜在因果基因（Castanera等， 2021） ）。

由于 MITE 是遗传变异的重要来源，我们假设 MITE 的基因组编辑 (GE) 可能是生成具有改变基因表达的新等位基因以调整作物性状的有效方法。选择两个农业上重要的水稻基因，生长调节因子 4 ( GRF4 ) 和胁迫响应 NAC1 ( SNAC1 ) 来检验这一假设。 OsGRF4可以积极调节产量相关性状（Wang等人， 2022 ），并在终止密码子 3' 1200 bp 处插入了 294 bp PIF/Harbinger超家族 MITE（图 S1）。 OsSNAC1可以赋予盐胁迫耐受性（Hu et al ., 2006 ），而在其上游和下游非翻译区（UTR）中没有检测到MITE（图S2）。

由于某些水稻基因 3′ UTR 中的 MITE 已被发现介导靶基因的翻译抑制（Shen等人， 2017 ），因此我们提出可以通过 CRISPR/Cas9 切除OsGRF4的下游 MITE，以产生过表达等位基因，并设计了转化水稻愈伤组织的缺失载体。在T ₀转基因植物中实现了平均突变频率（35.4%），携带纯合（10.4%）或杂合（16.7%）缺失突变。最后，我们获得了两个纯合的无转基因T ₂ OsGRF4^螨系L1和L2（图1a；图S3）。 OsGRF4^螨系中的OsGRF4 mRNA 水平与野生型 (WT) 相当（图 1b）。然而， OsGRF4^螨系的 OsGRF4 蛋白水平高于 WT（图 1c；图 S4）。我们比较了OsGRF4^螨和在田间条件下生长的 WT 植物之间的农艺性状。与 WT 植物相比， OsGRF4^螨系的株高显着降低，但每株植物的分蘖数 (PTN) 增加（图 1d-f）。与 WT 相比， OsGRF4^螨品系的千粒重 (TGW) 平均增加了 6.4%。这种增加伴随着晶粒长度的轻微增加，但晶粒宽度没有增加（图1g、h；图S5）。与 WT 相比， OsGRF4^螨L1 和 L2 系的结实率 (SSR) 略有下降，平均下降 8.6% 和 9%。分别为 3%（图 1i）。一般来说， OsGRF4^螨植物会略微增加每株植物的谷物产量（图1j），这在OsGRF4过度表达的植物中也观察到（Wang等人， 2022 ）。这些结果表明， OsGRF4^螨植物中的MITE缺失可以增加OsGRF4丰度，从而改善水稻农艺性状。

水稻基因 5' UTR 中的一些 MITE 此前已被报道可充当增强子，例如微型 Ping ( mPing ) TE，它可以赋予水稻附近基因盐胁迫诱导能力 (Naito et al ., 2009 )。因此，我们尝试将430 bp的mPing插入到耐盐基因OsSNAC1中。最近，通过将 CRISPR/Cas9 与硫代磷酸酯修饰的 3'-突出双链寡脱氧核苷酸 (dsODN) 相结合，开发了一种插入大 DNA 片段的有效方法 (Han等人， 2023 )。利用上述方法创建OsSNAC1 ^MITE等位基因，设计OsSNAC1起始密码子上游53-bp的sgRNA靶位点（sgRNA-1），并构建相应的CRISPR/Cas9质粒（图1k）。我们合成了含有mPing的 dsODN，该 mPing 具有五个连续的硫代磷酸酯修饰和 10-bp 3'-突出端，与 Cas9 诱导的切除突出端互补。然后通过粒子轰击将mPing dsODN 与 CRISPR/Cas9 载体一起递送至水稻愈伤组织中。总共获得81株独立的T ₀转基因植株。我们发现五株植物 (6.2%) 在预期方向上进行了有针对性的插入，而两株植物 (2.5%) 则进行了相反方向的定向插入（图 1l）。获得两个独立的 T ₂纯合目标系OsSNAC1 ^MITE L1 和 L2，用于进一步分析（图 1m）。 在对照条件下， OsSNAC1 ^MITE和 WT 植物之间的OsSNAC1 mRNA 水平没有明显差异。然而，盐胁迫1小时后， OsSNAC1 ^MITE L1和L2中OsSNAC1的相对mRNA水平分别是WT的1.9倍和2.3倍（图1n）。与此一致的是， OsSNAC1 ^MITE系在高盐条件下表现出比 WT 植物更高的存活率（图 1o，p）。这些结果表明， OsSNAC1 ^MITE植物中的mPing插入增强了盐诱导基因的表达，从而提高了耐盐性。

总之，我们已经证明，对水稻中的 MITE 进行基因操作可以产生不同的有益等位基因来调节基因表达并改善作物性状。我们已经能够设计 CRISPR/Cas9 靶向位点来实现位点特异性 MITE 删除或插入，从而通过利用 MITE 控制基因表达的能力来调节靶基因。鉴于 MITE 在许多植物基因组中广泛存在，可以想象，这种策略在未来可以更广泛地用于优化植物的发育和改良。