There are several ways by which plant scientists have attempted to generate gain-of-function alleles using CRISPR (Fig. 1). The first and most widely adopted approach is unbiased, multiplex, CRISPR-mediated, mutagenesis of a target promoter with multiple guide RNAs (gRNAs) in hope of generating gain-of-function alleles via fortuitous mutagenesis of upstream regulatory sequences. Since the first demonstration of Cas9-driven promoter mutagenesis by Rodríguez-Leal et al. (2017), several groups have utilized this approach to upregulate their genes of interest (Song et al., 2022; Zhou et al., 2023; Karavolias et al., 2024; Patel-Tupper et al., 2024). However, this method tends to result in mostly loss-of-function (knock-out and knock-down) alleles, and some gain-of-function alleles isolated using this method may involve larger-scale rearrangements such as chromosomal inversions (Patel-Tupper et al., 2024). Additionally, the underlying mechanism driving overexpression of the target gene is frequently not understood in this approach, limiting its potential for rational design and engineering of promoters. A second and more recently attempted approach involves the insertion of enhancers identified via genome enrichment or STARR-seq assays followed by subsequent introduction of the enhancer into the target promoter through CRISPR-mediated knock-in. This method is one of the first to demonstrate systematic engineering of promoter overexpression. However, the mechanism by which these enhancers elicit overexpression also remains elusive. In several cases, these enhancers were either found to barely increase expression above endogenous levels for certain genes (Claeys et al., 2024) or result in excessive overexpression that led to stunted growth and sterility (Yao et al., 2024). The third approach involves CRISPR-targeted mutagenesis of upstream open reading frames (uORFs) present in the 5′-UTR to increase translation of the primary ORF (pORF; Zhang et al., 2018). This is a powerful technique for increasing protein expression but cannot be applied to genes lacking uORFs. In the current study by Wang et al., the authors introduce a fourth method to overexpress an endogenous target gene. Unlike most previous methods, this approach is informed by a solid mechanistic understanding of repressive CREs within promoters, presenting a more rational and systematic approach to achieving endogenous gene overexpression (Fig. 1).

Recognizing the importance of improving protein content in major crops, Wang et al. overexpressed NF-YC4, a conserved transcription factor known to enhance leaf and seed protein content (Li et al., 2015), via targeted deletion of repressive RAV1A and WRKY binding motifs from the rice and soybean NF-YC4 gene promoters. Here, the systematic analysis incorporated into the experimental design deserves particular mention: the authors did not blindly assume that all copies of a putative repressor-binding site were functional. While no previous study has provided satisfying conclusions on how to distinguish functional transcription factor binding sites (TFBS) from nonfunctional ones, many studies have highlighted the critical role of DNA shape and flanking sequences in conferring affinity and specificity to transcription factor binding in plants (Zhang et al., 2019; Sielemann et al., 2021; Li et al., 2023). Notably, a study by Zhang et al. (2019) investigating G-box function in soybean promoters has revealed that the presence of a core TFBS motif is not sufficient for function, and that motifs with consequential effects on expression are often bordered by appropriate flanking sequences, which may be highly motif- and context-dependent. Simply deleting all possible copies of putative repressive CREs without first obtaining data about which elements are actually functional may not lead to the desired outcome, resulting in lost time and effort. Wang et al. used a combination of orthogonal luciferase reporter assays and in vitro electrophoretic mobility shift assay (EMSA) experiments, which allowed them to focus on consequential motifs, thereby facilitating the design of suitable gRNAs for CRISPR-based editing in a targeted and efficient manner. It is worth noting here that the authors successfully used a transient luciferase reporter assay in Nicotiana benthamiana leaves to obtain initial readouts of gene expression driven by various NF-YC4 promoter variants. Jores et al. (2021) have previously revealed limitations of studying dicot transcriptional responses in monocots, and vice versa; however, the results of Wang et al. indicate that well-established model systems such as N. benthamiana retain the power to provide accurate readouts on monocot (rice) promoter activity as long as the transcription factors and their cognate TFBS are known to be well conserved across monocots and dicots. The strong correlation between luciferase readouts and leaf transcript levels, as shown for rice NF-YC4 expression, strongly supports this and may spare future researchers the laborious task of isolating protoplasts from their species of interest for initial validation experiments. Thus, researchers aiming to rationally engineer promoters for overexpression may want to consider focusing their efforts on targeting putative repressive sites within promoters that are conserved across plant species. By examining transcript levels as well as protein and carbohydrate content across different tissues, the authors also reveal differences in gene expression enhancement in leaves and seeds, raising important questions of tissue specificity and prompting researchers to consider how one may go about achieving tissue-specific endogenous overexpression of target genes in the future.

Another highlight of the Wang et al. study is the utilization of site-directed nuclease (SDN)-1 type edits to overexpress NF-YC4 in rice and soybean. There are three major types of edits used to categorize CRISPR modification outcomes in plants (Podevin et al., 2013): (1) SDN-1 edits, which include all editing outcomes of double-strand breaks resulting from the plant's native nonhomologous end joining (NHEJ) repair mechanism; (2) SDN-2 edits that utilize a repair template with only a few bases differing from the original sequence; and (3) SDN-3 edits, which involve incorporation of a foreign or exogenous piece of DNA, typically large and considered as GM (Ahmad et al., 2023). Wang et al. generated their NF-YC4 promoter deletions via simultaneous introduction of gRNAs that target promoter segments harbouring functional repressive CREs, resulting in double-strand breaks that were presumably repaired by NHEJ, so the edits are considered SDN-1. After editing occurred in the initially transformed T0 generation, transgene-free plants were recovered in subsequent generations by Mendelian segregation, and these plants are more likely to be regulated as non-GM crops by most countries according to current product-based regulations (Ahmad et al., 2023). Such an approach is not only simple and generalizable but may also accelerate the practical impacts of the research, as these newly generated plants would be exempt from lengthy GM regulatory approval processes. Additionally, targeted deletion of repressive CREs may also minimize unintended pleiotropic effects and help preserve most of the native transcriptional regulation. Although current CRISPR tools may not allow researchers to achieve precise deletion of only the repressive CREs of interest, the results of Wang et al. demonstrate that the deletions need not necessarily be 100% precise for a physiologically relevant and desirable phenotype to be attained. New gene-editing tools are being developed for ever-more precise endogenous DNA sequence manipulation, and this study provides plant scientists with a roadmap for implementing a powerful strategy for targeted promoter engineering and endogenous gene overexpression.

植物科学家尝试使用 CRISPR 生成功能获得性等位基因的几种方法（图 1）。第一种也是最广泛采用的方法是用多个向导 RNA （gRNA） 对靶启动子进行无偏倚、多重、CRISPR 介导的诱变，以期通过上游调节序列的偶然诱变产生功能获得性等位基因。自 Rodríguez-Leal 等 人首次证明 Cas9 驱动的启动子诱变以来。（2017 年），几个小组已经利用这种方法来上调他们感兴趣的基因（Song等 人，2022 年;周 et al.， 2023;Karavolias 等 人，2024 年;Patel-Tupper 等 人，2024 年）。然而，这种方法往往导致大多数功能丧失（敲除和敲除）等位基因，并且使用这种方法分离的一些功能获得性等位基因可能涉及更大规模的重排，例如染色体倒位（Patel-Tupper等 人，2024 年）。此外，在这种方法中，驱动靶基因过表达的潜在机制经常不被理解，这限制了其对启动子进行合理设计和工程的潜力。第二种也是最近尝试的方法涉及插入通过基因组富集或 STARR-seq 测定鉴定的增强子，然后通过 CRISPR 介导的敲入将增强子引入靶启动子中。这种方法是最早证明启动子过表达系统工程的方法之一。然而，这些增强子引发过表达的机制仍然难以捉摸。 在几种情况下，这些增强子要么被发现几乎不会将某些基因的表达增加到内源水平以上（Claeys等 人，2024 年），要么导致过度表达，导致生长迟缓和不育（Yao等 人，2024 年）。第三种方法涉及 5'-UTR 中存在的上游开放阅读框 （uORF） 的 CRISPR 靶向诱变，以增加初级 ORF （pORF;Zhang et al.， 2018）。这是一种增加蛋白质表达的强大技术，但不能应用于缺乏 uORF 的基因。在 Wang 等 人目前的研究中，作者介绍了第四种过表达内源靶基因的方法。与以前的大多数方法不同，这种方法以对启动子内抑制性 CRE 的扎实机制理解为基础，为实现内源基因过表达提供了一种更合理和系统的方法（图 1）。

认识到提高主要作物蛋白质含量的重要性，Wang 等 人。过表达 NF-YC4，一种已知可增强叶片和种子蛋白含量的保守转录因子 （Liet al.， 2015），通过靶向删除水稻和大豆 NF-YC4 基因启动子中的抑制性 RAV1A 和 WRKY 结合基序。在这里，特别值得一提的是，纳入实验设计的系统分析：作者并没有盲目地假设推定的阻遏物结合位点的所有拷贝都是功能性的。虽然以前的研究没有就如何区分功能性转录因子结合位点 （TFBS） 和非功能性转录因子结合位点提供令人满意的结论，但许多研究强调了 DNA 形状和侧翼序列在赋予植物转录因子结合的亲和力和特异性方面的关键作用（Zhang等人 ，2019 年;Sielemann等 人，2021 年;Li et al.， 2023）。值得注意的是，Zhang 等 人的一项研究。（2019） 研究大豆启动子中的 G 盒功能表明，核心 TFBS 基序的存在不足以实现功能，并且对表达有相应影响的基序通常与适当的侧翼序列接壤，这可能高度依赖于基序和上下文。在没有首先获得有关哪些元素实际起作用的数据的情况下，简单地删除所有可能的推定压迫性 CRE 副本可能不会导致预期的结果，从而导致时间和精力的损失。Wang 等 人。 结合使用了正交荧光素酶报告基因测定和体外电泳迁移率变化测定 （EMSA） 实验，这使他们能够专注于相应的基序，从而促进以有针对性和有效的方式为基于 CRISPR 的编辑设计合适的 gRNA。值得注意的是，作者成功地在本氏烟草叶片中使用了瞬时荧光素酶报告基因测定，以获得由各种 NF-YC4 启动子变体驱动的基因表达的初步读数。Jores 等 人。（2021） 之前揭示了研究单子叶植物双子叶植物转录反应的局限性，反之亦然;然而，Wang 等 人的结果。表明，只要已知转录因子及其同源 TFBS 在单子叶植物和双子叶植物中非常保守，诸如本氏烟草（N. benthamiana）等成熟的模型系统仍然能够提供单子叶植物（水稻）启动子活性的准确读数。如水稻 NF-YC4 表达所示，荧光素酶读数与叶片转录水平之间的强相关性强烈支持这一点，并且可能使未来的研究人员免于从其感兴趣的物种中分离原生质体以进行初始验证实验的艰巨任务。因此，旨在合理设计过表达启动子的研究人员可能希望考虑将他们的工作重点放在靶向启动子内在植物物种中保守的假定抑制位点。 通过检查不同组织的转录水平以及蛋白质和碳水化合物含量，作者还揭示了叶子和种子中基因表达增强的差异，提出了组织特异性的重要问题，并促使研究人员考虑未来如何实现靶基因的组织特异性内源性过表达。

Wang et al.研究是利用定点核酸酶 （SDN）-1 型编辑在水稻和大豆中过表达 NF-YC4。有三种主要类型的编辑用于对植物中的 CRISPR 修饰结果进行分类（Podevin等 人，2013 年）：（1） SDN-1 编辑，其中包括由植物的天然非同源末端连接 （NHEJ） 修复机制引起的双链断裂的所有编辑结果;（2） SDN-2 编辑使用修复模板，只有几个碱基与原始序列不同;（3） SDN-3 编辑，涉及掺入外来或外源性 DNA 片段，通常很大，被认为是 GM（Ahmad等 人，2023 年）。Wang 等 人。通过同时引入靶向含有功能性抑制性 CRE 的启动子片段的 gRNA，产生其 NF-YC4 启动子缺失，导致可能由 NHEJ 修复的双链断裂，因此这些编辑被认为是 SDN-1。在最初转化的 T0 代发生编辑后，无转基因植物通过孟德尔分离在随后的世代中被回收，并且这些植物更有可能根据当前基于产品的法规被大多数国家作为非转基因作物进行监管（Ahmad等 人，2023 年）。这种方法不仅简单易行，而且可能加速研究的实际影响，因为这些新产生的植物将免于漫长的转基因监管审批程序。此外，抑制性 CREs 的靶向缺失也可能最大限度地减少意外的多效性效应，并有助于保留大部分天然转录调控。 尽管目前的 CRISPR 工具可能不允许研究人员仅实现对感兴趣的抑制性 CRE 的精确删除，但 Wang 等 人的结果。证明缺失不一定是 100% 精确的才能获得生理相关和理想的表型。正在开发新的基因编辑工具，以实现更精确的内源性 DNA 序列操作，这项研究为植物科学家提供了实施靶向启动子工程和内源性基因过表达的强大策略的路线图。