CRISPR-Cas9 nuclear dynamics and target recognition in living cells
活细胞中的 CRISPR-Cas9 核动力学和靶标识别

The clustered regularly interspaced short palindromic repeats (CRISPR) bacterial acquired immunity system, as redesigned for application in eukaryotic cells, has become a powerful tool for genome editing (Jinek et al., 2012; Cong et al., 2013; Mali et al., 2013), gene expression modulation (Qi et al., 2013), and genomic DNA visualization (Chen et al., 2013; Ma et al., 2015, 2016). The CRISPR-associated protein 9 (Cas9)–guide RNA complex initially locates a protospacer adjacent motif (PAM) sequence and a vicinal 10–12 nt seed sequence for complementarity, both of which are essential for DNA targeting and cleavage (Jinek et al., 2014; Sternberg et al., 2014; Jiang et al., 2015; Richardson et al., 2016). Biochemical studies suggest that structural reorganization of the Cas9–guide RNA complex and R-loop formation at the target are essential for DNA cleavage (Szczelkun et al., 2014; Rutkauskas et al., 2015; Sternberg et al., 2015; Jiang et al., 2016). Single molecule tracking in living cells revealed that the Cas9–guide RNA complex searches DNA by 3D diffusion with a dwell time of <1 s at off-target sites and established a lower limit of ∼5 min for on-target residence time (Knight et al., 2015).
成簇规则间隔短回文重复序列（CRISPR）细菌获得性免疫系统经过重新设计以应用于真核细胞，已成为基因组编辑的强大工具（ Jinek 等，2012 ； Cong 等，2013 ； Mali 等，2013）。 ，2013 ），基因表达调节（ Qi等，2013 ）和基因组DNA可视化（ Chen等，2013 ； Ma等，2015，2016 ） 。 CRISPR 相关蛋白 9 (Cas9)-引导 RNA 复合物最初定位一个原型间隔子相邻基序 (PAM) 序列和一个邻近的 10-12 nt 种子序列以实现互补，这两者对于 DNA 靶向和切割都是必需的（ Jinek 等，2017）。 ，2014 ； Sternberg 等，2014 ； Jiang 等，2015 ； Richardson 等，2016 ）。生化研究表明，Cas9-引导 RNA 复合物的结构重组和靶标处 R 环的形成对于 DNA 切割至关重要（ Szczelkun 等，2014 ； Rutkauskas 等，2015 ； Sternberg 等，2015 ； Jiang 等）等，2016 ）。活细胞中的单分子追踪表明，Cas9-引导 RNA 复合物通过 3D 扩散搜索 DNA，停留时间为 <1 id=175>Knight 等，2015）。

However, the true on-target dwell time of Cas9–guide RNA and its influence on CRISPR activity in living cells remain unknown. Here, we have investigated the intranuclear assembly and target-binding dynamics of the Cas9–guide RNA complex in live cells and have found that target residence time highly correlates with cleavage activity.
然而，Cas9-guide RNA 的真正靶向停留时间及其对活细胞中 CRISPR 活性的影响仍然未知。在这里，我们研究了活细胞中 Cas9-引导 RNA 复合物的核内组装和靶标结合动力学，并发现靶标停留时间与裂解活性高度相关。

In previous studies, either nuclease-dead Cas9 (dCas9) from Streptococcus pyogenes was tagged with fluorescent proteins for live cell tracking (Chen et al., 2013; Ma et al., 2015), or RNA aptamers were inserted into the guide RNA to recruit fluorescent protein-fused binding partners for visualization of DNA targets (Ma et al., 2016). In this study, we directly labeled both dCas9 and the guide RNA to track their assembly and target-site interrogation. The system we designed is diagrammed in Fig. 1 A, and a detailed description of its molecular components and strategic operation is given in the figure legend. The guide RNA carries the aptamer “Broccoli” (Filonov et al., 2014) so that one can convert it to a fluorescent RNA by adding a specific small molecule to the medium, which, after cellular uptake, forms a conjugate with the guide RNA’s aptamer that elicits the small molecule’s fluorescence (see also Fig. S1). The guide-Broccoli RNA is constitutively expressed via the U6 promoter, and its expression is reported by the coexpression of blue fluorescent protein (BFP) downstream via a constitutive phosphoglycerol kinase promoter. In parallel, the system employs a ligand-tunable destabilization domain (DD; Banaszynski et al., 2006) in addition to the Tet-On system to tightly control the expression of dCas9-mCherry. As the genomic target for interrogation by Cas9–guide RNA, we chose a repeated sequence uniquely situated in the subtelomeric region of the long arm of chromosome 3 (C3; Ma et al., 2016), which facilitated visualization of the complex at the target with a favorable ratio of foci signal/nuclear background. As shown in Fig. 1 B, DD-dCas9-mCherry and the C3-targeting guide RNA–2XBroccoli were colocalized on distinct foci (C3).
在之前的研究中，要么用荧光蛋白标记来自化脓性链球菌的核酸酶死亡 Cas9 (dCas9) 以进行活细胞追踪（ Chen et al., 2013 ； Ma et al., 2015 ），要么将 RNA 适体插入指导 RNA招募荧光蛋白融合结合伴侣以实现 DNA 靶点的可视化（ Ma et al., 2016 ）。在本研究中，我们直接标记 dCas9 和引导 RNA，以跟踪它们的组装和靶位点询问。我们设计的系统如图1A所示，图例中给出了其分子组成和策略操作的详细描述。指导RNA携带适体“Broccoli”（ Filonov等，2014 ），因此可以通过向培养基中添加特定的小分子将其转化为荧光RNA，在细胞摄取后，与指导RNA的结合物形成缀合物引发小分子荧光的适体（另参见图S1 ）。引导西兰花 RNA 通过 U6 启动子组成型表达，其表达是通过组成型磷酸甘油激酶启动子下游蓝色荧光蛋白 (BFP) 的共表达来报告的。同时，除了 Tet-On 系统之外，该系统还采用配体可调去稳定结构域（DD； Banaszynski 等人，2006 ）来严格控制 dCas9-mCherry 的表达。作为 Cas9-guide RNA 询问的基因组靶标，我们选择了位于 3 号染色体长臂亚端粒区域的重复序列（C3； Ma et al., 2016 ），这有利于靶标处复合物的可视化具有良好的焦点信号/核背景比。如图所示。 如图 1 B 所示，DD-dCas9-mCherry 和 C3 靶向指导 RNA–2XBroccoli 共定位于不同的病灶 (C3)。

To directly measure the Cas9 and guide RNA levels in living cells, we used FACS (Fig. 2 A). The results showed that the DD-dCas9-mCherry level was tightly controlled by the addition of doxycycline (Dox) and Shield1, whereas C3–guide RNA–2XBroccoli was not fluorescent until the cells were exposed to DFHBI-1T. Intriguingly, in the absence of DD-dCas9-mCherry, the level of fluorescent guide RNA was too low to be detected by either FACS or microscopy (Fig. 2, A and C) under similar BFP expression (Fig. 2, B and C), indicating that the guide RNA is extremely unstable in the absence of dCas9, which was confirmed by RT-PCR analysis (Fig. 2 D).
为了直接测量活细胞中的Cas9并指导RNA水平，我们使用了FACS（图2A ）。结果表明，DD-dCas9-mCherry 水平受到强力霉素 (Dox) 和 Shield1 的添加的严格控制，而 C3-引导 RNA-2XBroccoli 直到细胞暴露于 DFHBI-1T 后才发出荧光。有趣的是，在没有 DD-dCas9-mCherry 的情况下，在类似的 BFP 表达（图 2，B 和C）下，荧光向导 RNA 的水平太低，无法通过 FACS 或显微镜检测（图 2，A 和 C ） ），表明在没有dCas9的情况下指导RNA极其不稳定，这通过RT-PCR分析得到证实（图2D ）。

To investigate the stability of the dCas9–guide RNA complex, cell lines expressing C3–guide RNA–2XBroccoli and DD-dCas9-mCherry were treated with actinomycin D at a concentration that inhibits transcription by all three RNA polymerases. Approximately 50% of the guide RNA disappeared within 15 min, and the remainder declined over the subsequent 4 hours, whereas both BFP and Cas9 levels remained stable, confirming their expression and that of the guide RNA (Fig. 3, A and B). In additional experiments, we examined the effect of guide RNA–2XBroccoli with shorter guide sequences (i.e., 11, 6, or 0 nt in length) on the stability of the dCas9–guide RNA complex. As shown in Fig. 3 C (middle), stability decreased with the shorter guide RNA lengths.
为了研究 d​​Cas9-guide RNA 复合物的稳定性，用放线菌素 D 处理表达 C3-guide RNA-2XBroccoli 和 DD-dCas9-mCherry 的细胞系，其浓度可抑制所有三种 RNA 聚合酶的转录。大约 50% 的指导 RNA 在 15 分钟内消失，其余部分在随后的 4 小时内下降，而 BFP 和 Cas9 水平保持稳定，证实了它们的表达和指导 RNA 的表达（图 3，A 和 B ）。在其他实验中，我们检查了引导序列较短（即长度为 11、6 或 0 nt）的引导 RNA–2XBroccoli 对 dCas9–引导 RNA 复合物稳定性的影响。如图3 C （中）所示，稳定性随着引导 RNA 长度的缩短而降低。

To understand how the stability of the guide RNA affects DNA interrogation, we compared the guide RNA containing a full-length tracrRNA (trans-activating CRISPR RNA) component (+85 nt) to guide RNAs with truncated tracrRNA components (+54 or +67 nt), which had been shown to be less stable (Hsu et al., 2013; Mekler et al., 2016). Reduced nuclear guide RNA–Broccoli signal and reduced foci signals were evident with guide RNAs with truncated tracrRNA components compared with the full-length one (Fig. 4 A). The expression levels of BFP and dCas9-mCherry were comparable in each condition, indicating that Cas9–guide RNA complexes with truncated tracrRNAs are less stable and result in inefficient DNA targeting. The level of nuclear C3–guide RNA–Broccoli, presumably reflecting the assembled Cas9–guide RNA complex, positively correlated with target binding as shown by foci brightness (Fig. 4 B and Fig S2). Hence, DNA interrogation by Cas9–guide RNA is limited by the guide RNA stability and assembly with Cas9 and is driven by the concentration of the assembled Cas9–guide RNA complex. Interestingly, with the two shorter guide RNAs, we observed a nucleolar localization of Cas9 (Fig. 4 A), raising the possibility that some dCas9 remains unassembled with these shorter guide RNAs and enters nucleoli where there is a high concentration of hundreds of small noncoding RNA species (Jorjani et al., 2016).
为了了解引导 RNA 的稳定性如何影响 DNA 询问，我们将包含全长 tracrRNA（反式激活 CRISPR RNA）组件（+85 nt）的引导 RNA 与包含截短 tracrRNA 组件（+54 或 +67）的引导 RNA 进行了比较。 nt），已被证明不太稳定（ Hsu 等人，2013 ； Mekler 等人，2016 ）。与全长的相比，具有截短的tracrRNA成分的引导RNA明显减少了核引导RNA-西兰花信号和焦点信号（图4A ）。 BFP 和 dCas9-mCherry 的表达水平在每种条件下都相当，表明 Cas9-guide RNA 与截短的 tracrRNA 复合物稳定性较差，导致 DNA 靶向效率低下。核 C3-引导 RNA-西兰花的水平，大概反映了组装的 Cas9-引导 RNA 复合物，与靶点结合呈正相关，如焦点亮度所示（图 4 B和图 S2 ）。因此，Cas9-指导RNA对DNA的询问受到指导RNA稳定性和与Cas9的组装的限制，并且由组装的Cas9-指导RNA复合物的浓度驱动。有趣的是，通过这两个较短的向导RNA，我们观察到Cas9的核仁定位（图4A ），这提高了一些dCas9与这些较短的向导RNA未组装并进入核仁的可能性，其中有数百个小非编码的高浓度RNA 种类（ Jorjani 等人，2016 ）。

To understand how CRISPR complex interrogates true or mismatched target, we then used FRAP of GFP-dCas9 to measure the on-target residence time (Fig. 5, A and B). The residence time and the off-rate of the dCas9/C3-11–guide RNA complex on the C3 target were estimated to be 206 ± 4.5 min and 2.9 ± 0.1 × 10⁻⁴ s⁻¹ (Fig. 5 C and Fig. S3). Complexes with mismatched guide RNAs at the seed sequence all displayed shorter residence times from one third to one hundredth on the target in an identity- and position-dependent manner (Fig. 5 C and Fig. S3). These results, obtained in live cells, show that even a single mismatch between the guide RNA and the target DNA significantly reduces the dwell time of the dCas9–guide RNA complex on the target. A mismatched guide RNA that had shorter target residence was also reflected in reduced overall foci brightness (Fig. 5, D and E). Purine-to-purine mismatches at position -5 away from the PAM previously were shown to result in loss of cleavage activities in all 46 guide RNAs examined (Doench et al., 2016). Here, we show that -5 (C) to G or A mutations, resulting in G to G or G to A mismatches, resulted in reduction of residence times to 7.4 ± 3.6 min or 1.4 ± 0.6 min, respectively (Fig. 5C). These results reveal that mismatches on guide RNA seed sequence reduce target residence time in an identity- and position-dependent manner, but not with respect to their proximity to the PAM site, at least as compared with the mismatches at the -5 or -3 positions.
为了了解 CRISPR 复合物如何询问真实或不匹配的靶标，我们随后使用 GFP-dCas9 的 FRAP 来测量在靶标上的停留时间（图 5，A 和 B ）。 dCas9/C3-11-向导RNA复合物在C3靶标上的停留时间和解离速率估计为206±4.5分钟和2.9±0.1×10 ^-4 s ^-1 （图5C和图5C）。 S3 ）。种子序列处引导RNA不匹配的复合物都以身份和位置依赖的方式在靶标上显示出从三分之一到百分之一的较短停留时间（图5C和图S3）。这些在活细胞中获得的结果表明，即使引导 RNA 与目标 DNA 之间存在单一错配，也会显着缩短 dCas9-引导 RNA 复合物在目标上的停留时间。具有较短目标停留时间的不匹配引导RNA也反映在总体焦点亮度降低上（图5、D和E ）。先前显示，远离 PAM 的 -5 位的嘌呤与嘌呤错配会导致检查的所有 46 个引导 RNA 中裂解活性丧失（ Doench 等人，2016 ）。在这里，我们表明-5（C）到G或A的突变，导致G到G或G到A错配，导致停留时间分别减少到7.4±3.6分钟或1.4±0.6分钟（图5C ） 。这些结果表明，向导RNA种子序列上的错配以身份和位置依赖性方式减少了靶标停留时间，但与它们与PAM位点的接近程度无关，至少与-5或-3处的错配相比职位。

The correlation between shortened residence time and lower target-binding efficiency was also observed when the guide RNA length truncated from 11 to 8 nt, and the residence time decreased from 206.0 ± 4.6 min to 25.3 ± 7.3 min (Fig. S4). This finding aligns with recent studies that guide RNAs 12 and 20 nt in length were comparable in transcription activation, whereas 1 of 8 nt had almost negligible activation (Kiani et al., 2015) because the residence time of transcription factors on the promoters has been shown to be correlated with transcription activation (Lickwar et al., 2012).
当向导RNA长度从11 nt截短至8 nt时，也观察到停留时间缩短与靶点结合效率降低之间的相关性，并且停留时间从206.0±4.6分钟减少到25.3±7.3分钟（图S4 ）。这一发现与最近的研究一致，即长度为 12 和 20 nt 的指导 RNA 在转录激活方面具有可比性，而 8 nt 中的 1 个的激活几乎可以忽略不计（ Kiani 等，2015 ），因为转录因子在启动子上的停留时间已显示与转录激活相关（ Lickwar 等人，2012 ）。

We then asked how the residence time on the target correlates with CRISPR activity. To directly visualize cleavage activity, target foci were labeled using pairs of guide RNAs, and loss of foci brightness caused by cleavage was monitored (Fig. 6 A). Truncated (11 nt; C3-11 for labeling only; Ma et al., 2016) and full-length (20 nt; C3-20 as cleavage competent) guide RNAs were simultaneously used together with nuclease-active Sp Cas9-3XGFP. Because the target is a repeated sequence, cleavage will be manifest by signal diminution and dispersal. Mutations in the C3-20 guide RNAs are predicted to result in variable degrees of cleavage. Intermediate degrees of cleavage will result in an increase in the number of foci with lower intensity, whereas higher levels of cleavage will result in loss of detectable foci altogether. When only C3-11 guide RNA (labeling competent, cleavage noncompetent) was used, four distinct target foci were detected, whereas no distinct foci were observed when the C3-20 guide RNA was coexpressed with C3-11 (Fig. 6, B and C). Mismatched guide RNAs that displayed shorter target residence times (Fig. 5 C) displayed less cleavage activity (Fig. 6, B and C; Fig. S5). The degree to which cleavage activity was impaired by each of the mismatched guide RNAs aligns with previously reported studies (Hsu et al., 2013). These results suggest that radical alterations of residence time occur when there are mismatches in the seed sequence, implying that in living cells, the Cas9–guide RNA complex acutely detects a true target DNA before executing the cleavage step.
然后我们询问目标上的停留时间与 CRISPR 活性有何关联。为了直接可视化切割活性，使用引导RNA对标记目标焦点，并监测切割引起的焦点亮度损失（图6A ）。截短的（11 nt；C3-11 仅用于标记； Ma et al., 2016 ）和全长（20 nt；C3-20 作为切割活性）引导 RNA 与核酸酶活性 Sp Cas9-3XGFP 同时使用。由于靶标是重复序列，因此切割将通过信号减少和分散来体现。 C3-20 指导 RNA 中的突变预计会导致不同程度的切割。中等程度的裂解将导致强度较低的焦点数量增加，而较高水平的裂解将导致可检测焦点的完全丢失。当仅使用C3-11指导RNA（标记能力，切割非能力）时，检测到四个不同的靶点，而当C3-20指导RNA与C3-11共表达时，没有观察到明显的靶点（图6，B和C ）。显示较短目标停留时间的错配引导RNA（图5C ）显示较低的切割活性（图6、B和C ；图S5 ）。每个不匹配的向导RNA对切割活性的损害程度与之前报道的研究一致（ Hsu等人，2013 ）。这些结果表明，当种子序列不匹配时，停留时间会发生根本改变，这意味着在活细胞中，Cas9-引导RNA复合物在执行切割步骤之前能够敏锐地检测到真正的目标DNA。

In the type II CRISPR-Cas system, Cas9 and guide RNAs assemble into a ribonucleoprotein complex and act as a single-turnover enzyme to cleave target DNA highly efficiently in vitro (Jinek et al., 2012; Sternberg et al., 2015). However, the cell’s microenvironment of Cas9, guide RNA, and the target might alter target recognition efficiency (Farasat and Salis, 2016; Fu et al., 2016). Here, we examined Cas9 and guide RNA’s intracellular location, level, and stability in living cells and their effects on RNP assembly for target recognition. We found that guide RNAs are extremely unstable in the absence of Cas9 or when the guide RNA has a truncated tracrRNA component. In the presence of Cas9, the guide RNA displayed two nuclear populations with half-lives of ∼15 min and ∼2 h, respectively (Fig. 3). We speculate that the former represents Cas9–guide RNA complexes roaming the nucleus, and the latter represents ones that are stabilized by DNA binding, either to targets or nontargets. The fact that the guide RNA is so unstable in the absence of Cas9 emphasizes the importance of expressing it at stoichiometric levels with respect to Cas9 in gene-editing studies. Conversely, the nucleolar localization of Cas9–guide RNA complexes under some conditions, as we have observed (Fig. 4 A) and also reported previously (Chen et al., 2013), suggests that attention must be paid in gene-editing endeavors to the possible sequestration of CRISPR machinery in this RNA-rich nuclear body.
在II型CRISPR-Cas系统中，Cas9和引导RNA组装成核糖核蛋白复合物，并作为单周转酶在体外高效切割靶DNA（ Jinek等，2012 ； Sternberg等，2015 ）。然而，Cas9、引导RNA和靶标的细胞微环境可能会改变靶标识别效率( Farasat and Salis, 2016 ; Fu et al., 2016 )。在这里，我们检查了 Cas9 和引导 RNA 在活细胞中的细胞内位置、水平和稳定性及其对目标识别的 RNP 组装的影响。我们发现，在缺少 Cas9 的情况下或当指导 RNA 具有截短的 tracrRNA 成分时，指导 RNA 极其不稳定。在Cas9存在的情况下，指导RNA显示出两个核群体，半衰期分别为～15分钟和～2小时（图3 ）。我们推测前者代表在细胞核中漫游的 Cas9-引导 RNA 复合物，后者代表通过 DNA 结合（无论是与靶标还是非靶标）而稳定的复合物。事实上，指导 RNA 在缺少 Cas9 的情况下非常不稳定，这一事实强调了在基因编辑研究中相对于 Cas9 以化学计量水平表达它的重要性。相反，正如我们所观察到的（图 4 A ）以及之前报道的那样（ Chen 等，2013 ），Cas9-guide RNA 复合物在某些条件下的核仁定位表明，在基因编辑工作中必须注意CRISPR 机制可能被隔离在这个富含 RNA 的核体中。

It has not been well understood how the Cas9–guide RNA complex reaches targets within structurally organized chromatin environments. The genome contains a vast number of off-target binding sites for any given guide RNA (Kuscu et al., 2014; Wu et al., 2014), and these would be expected to attenuate free intranuclear diffusion of the Cas9–guide RNA complex, thus slowing the kinetics of the search process for the bona fide target (Farasat and Salis, 2016). These model studies suggested that the on-target cleavage activity would be increased and off-target binding would be substantially reduced when the genome size increases from the ∼50,000 base pairs of λ phage to ∼3 billion base pairs in the human genome. We find that increased nuclear Cas9 and guide RNA concentrations significantly enhance DNA targeting. Moreover, our finding that guide RNAs are extremely unstable in the absence of Cas9 means that achieving high and equal levels of both nuclear Cas9 and guide RNA will be essential for CRISPR efficiency, whether for editing, gene regulation, or chromosomal locus labeling.
目前尚不清楚 Cas9-引导 RNA 复合物如何在结构组织的染色质环境中到达目标。基因组包含大量针对任何给定向导 RNA 的脱靶结合位点（ Kuscu 等人，2014 年； Wu 等人，2014 年），预计这些位点会减弱 Cas9-引导 RNA 复合物的自由核内扩散，从而减慢了真正目标的搜索过程的动力学（ Farasat 和 Salis，2016 ）。这些模型研究表明，当基因组大小从 λ 噬菌体的约 50,000 个碱基对增加到人类基因组中的约 30 亿个碱基对时，中靶切割活性将增加，脱靶结合将大幅减少。我们发现核 Cas9 和指导 RNA 浓度的增加显着增强了 DNA 靶向性。此外，我们发现引导RNA在没有Cas9的情况下极其不稳定，这意味着无论是编辑、基因调控还是染色体位点标记，核Cas9和引导RNA的高水平和同等水平对于CRISPR效率至关重要。

In the type I CRISPR-Cas system, in vitro kinetics studies have shown that target recognition occurs through directional R-loop zipping from the PAM and that intermediate R-loops stall at mutations and collapse in a PAM proximity–dependent manner, resulting in shorter residence times at targets with mismatches closer to the PAM (Rutkauskas et al., 2015). Here, our in vivo measurements, with a type II CRISPR-Cas9 system, revealed that on-target residence time radically changes with single mismatches in the guide RNA seed sequence in a base identity- and position-dependent manner, but independent of PAM proximity. Specifically, the transversion mutation (C to A) reduced the residence time to a greater extent than the transition mutation (C to U), which agrees with the reported higher residual CRISPR activity of seed transition mutations than transversions (Hsu et al., 2013; Tsai et al., 2015). Moreover, we found that the distance of a mismatch from the PAM is not directly correlated with residence time and cleavage activity. We found that C to G or A mismatches -5 away from the PAM were more detrimental than ones at position -3.
在 I 型 CRISPR-Cas 系统中，体外动力学研究表明，目标识别是通过从 PAM 定向 R 环压缩而发生的，并且中间 R 环在突变时停滞并以 PAM 邻近依赖性方式崩溃，从而导致更短的失配目标的停留时间更接近 PAM（ Rutkauskas 等，2015 ）。在这里，我们使用 II 型 CRISPR-Cas9 系统进行的体内测量表明，引导 RNA 种子序列中的单个错配会以碱基身份和位置依赖性方式彻底改变目标停留时间，但与 PAM 接近度无关。具体来说，颠换突变（C 到 A）比过渡突变（C 到 U）更大程度地减少了停留时间，这与报道的种子过渡突变比颠换具有更高的残余 CRISPR 活性一致（ Hsu et al., 2013蔡等人，2015 ）。此外，我们发现 PAM 错配的距离与停留时间和裂解活性不直接相关。我们发现远离 PAM -5 的 C 到 G 或 A 不匹配比位置 -3 的不匹配更有害。

In summary, our measurements of target residence times revealed the critical negative impact of certain mismatches in the seed sequence and their inverse correlation with cleavage activity (Fig. 7). Measurements of target residence time for mismatches at each position can thus be used in the rational design of optimal guide RNAs. It will be intriguing to examine whether new Cas orthologues (Hou et al., 2013; Ran et al., 2015; Zetsche et al., 2015) or S. pyogenes Cas9 variants with increased specificity (Kleinstiver et al., 2016; Slaymaker et al., 2016) sensitize target discrimination by altering residence times.
总之，我们对目标停留时间的测量揭示了种子序列中某些错配的严重负面影响及其与裂解活性的负相关性（图7 ）。因此，每个位置上错配的目标停留时间的测量可用于最佳指导 RNA 的合理设计。检查新的 Cas 直系同源物（ Hou 等人，2013 ； Ran 等人，2015 ； Zetsche 等人，2015 ）或化脓性链球菌Cas9 变体是否具有更高的特异性（ Kleinstiver 等人，2016 ； Slaymaker ）将是很有趣的。 et al., 2016 ）通过改变停留时间来提高目标歧视的敏感性。

The expression vector for dCas9-mCherry, 3XGFP from S. pyogenes was described previously, originally constructed from pHAGE-TO-DEST (Ma et al., 2015), and a DD was inserted into the N-terminal region. The guide RNA expression vector was based on the pLKO.1 lentiviral expression system (Addgene), in which TetR-P2A-BFP (Addgene) was inserted right after the phosphoglycerate kinase (PGK) promoter, with sequences coding for the desired guide RNA, guide RNA–1XBroccoli, or guide RNA–2XBroccoli inserted immediately after the U6 promoter. To place C3-20 and C3-11 guide RNAs in the same plasmid, DNA sequences for the U6 promoter-C3-11 guide RNA were amplified by PCR and cloned downstream of the U6–guide RNA cassette, and then C3-20 or the other guide RNA mutants were cloned into the same vector. The rapid guide RNA expression plasmid construction protocol was described previously (Ma et al., 2015). Details on the Cas9s and guide RNAs used in this study are given in Table S1.
之前描述了来自化脓性链球菌的 dCas9-mCherry、3XGFP 表达载体，最初是从 pHAGE-TO-DEST 构建的（ Ma 等人，2015 ），并且将 DD 插入 N 端区域。指导RNA表达载体基于pLKO.1慢病毒表达系统（Addgene），其中TetR-P2A-BFP（Addgene）插入到磷酸甘油酸激酶（PGK）启动子之后，其序列编码所需的指导RNA，引导 RNA–1XBroccoli 或引导 RNA–2XBroccoli 立即插入 U6 启动子之后。为了将 C3-20 和 C3-11 指导 RNA 置于同一质粒中，通过 PCR 扩增 U6 启动子-C3-11 指导 RNA 的 DNA 序列，并克隆到 U6 指导 RNA 盒的下游，然后将 C3-20 或 C3-11 指导 RNA 置于同一质粒中。其他指导RNA突变体被克隆到同一载体中。先前描述了快速引导RNA表达质粒构建方案（ Ma等人，2015 ）。表 S1中给出了本研究中使用的 Cas9 和向导 RNA 的详细信息。

Human osteosarcoma U2OS cells (ATCC) were cultured on 35-mm glass-bottom dishes (MatTek Corporation) at 37°C in DMEM (Thermo Fisher Scientific) containing high glucose and supplemented with 10% (vol/vol) FBS. For transfection, typically 200 ng of dCas9-mCherry or dCas9-3xGFP plasmid DNA and 1 µg of total guide RNA plasmid DNA indicated were cotransfected using Lipofectamine 2000 (Thermo Fisher Scientific), and the cells were incubated for another 24–48 h before imaging. For actinomycin D (Thermo Fisher Scientific) treatment, cells were seeded in 6-well plates in medium containing 2 µg/ml doxycycline (Sigma-Aldrich) and 0.5 µM Shield1 (Takara Bio Inc.). After 2 d, actinomycin D (Sigma-Aldrich) was added at 10 µg/ml at the indicated times before FACS analysis. For Fig. 6 B and Fig. S5, cells were transfected for 24 h and fixed in 4% formaldehyde for 10 min before capturing images.
人骨肉瘤 U2OS 细胞 (ATCC) 在 35 毫米玻璃底培养皿 (MatTek Corporation) 上于 37°C 下在含有高葡萄糖并补充有 10% (vol/vol) FBS 的 DMEM (Thermo Fisher Scientific) 中培养。对于转染，通常使用 Lipofectamine 2000 (Thermo Fisher Scientific) 共转染 200 ng dCas9-mCherry 或 dCas9-3xGFP 质粒 DNA 和 1 µg 总指导 RNA 质粒 DNA，并在成像前将细胞再孵育 24-48 小时。对于放线菌素 D (Thermo Fisher Scientific) 处理，将细胞接种到 6 孔板中，培养基中含有 2 µg/ml 多西环素 (Sigma-Aldrich) 和 0.5 µM Shield1 (Takara Bio Inc.)。 2 天后，在 FACS 分析前的指定时间添加 10 µg/ml 放线菌素 D (Sigma-Aldrich)。对于图 6 B和图 S5，细胞转染 24 小时并在捕获图像之前在 4% 甲醛中固定 10 分钟。

HEK293T cells were maintained in Iscove’s Modified Dulbecco’s Medium (Thermo Fisher Scientific) containing high glucose and supplemented with 1% GlutaMAX (Thermo Fisher Scientific), 10% FBS (Hycolne FBS; Thermo Fisher Scientific), and 1% penicillin/streptomycin (Thermo Fisher Scientific). 24 h before transfection, ∼5 × 10⁵ cells were seeded in 6-well plates. For each well, 0.5 µg of pCMV-dR8.2 dvpr (Addgene) and 0.3 µg of pCMV-VSV-G (Addgene), each constructed to carry HIV long terminal repeats, and 1.5 µg of plasmid containing the gene of interest were cotransfected by using TransIT transfection reagent (Mirus) according to the manufacturer’s instructions. After 48 h, the virus was collected by filtration through a 0.45-µm polyvinylidene fluoride filter (Pall Laboratory). The virus was immediately used or stored at -80°C. For lentiviral transduction, U2OS cells maintained as described in Cell culture and transfection were transduced by spinfection in 6-well plates with lentiviral supernatant for 2 d and ∼2 × 10⁵ cells were combined with 1 ml lentiviral supernatant and centrifuged for 30 min at 1,200 g.
HEK293T 细胞维持在含有高葡萄糖的 Iscove's Modified Dulbecco's Medium (Thermo Fisher Scientific) 中，并补充有 1% GlutaMAX (Thermo Fisher Scientific)、10% FBS (Hycolne FBS；Thermo Fisher Scientific) 和 1% 青霉素/链霉素 (Thermo Fisher Scientific)科学）。转染前24小时，将~5×10 ^{5 个}细胞接种到6孔板中。对于每个孔，共转染 0.5 µg pCMV-dR8.2 dvpr (Addgene) 和 0.3 µg pCMV-VSV-G (Addgene)，每个都构建为携带 HIV 长末端重复序列，并共转染 1.5 µg 含有目的基因的质粒根据制造商的说明使用 TransIT 转染试剂 (Mirus)。 48小时后，通过0.45微米聚偏二氟乙烯过滤器（颇尔实验室）过滤收集病毒。病毒立即使用或保存在-80°C。对于慢病毒转导，按照细胞培养和转染中所述维持的 U2OS 细胞通过在 6 孔板中用慢病毒上清液旋转转染转导 2 d，并将 ∼2 × 10 ^{5 个}细胞与 1 ml 慢病毒上清液混合，并在 1,200 ℃ 下离心 30 分钟。克。

Cells expressing the desired fluorescent Cas9 and/or guide RNA were selected by fluorescence-activated sorting (FACSAria cell sorter; BD) or analyzed on an LSR II cytometer (BD). All data were processed with FlowJo software (Tree Star). Both the FACSAria cell sorter and LSR II cytometer were equipped with 405-, 488-, and 561-nm excitation lasers, and the emission signals were detected by using filters at 450/50 nm (wavelength/bandwidth) for BFP, 530/30 nm for Broccoli or GFP, and 610/20 nm for mCherry. Single cells were sorted onto individual wells in 96-well plates containing 1% GlutaMAX, 20% FBS, and 1% penicillin/streptomycin in chilled DMEM medium. For the Broccoli signal analysis, 5 µM DFHBI-1T was added before FACS.
通过荧光激活分选（FACSAria细胞分选仪；BD）选择表达所需荧光Cas9和/或向导RNA的细胞，或在LSR II细胞计数器（BD）上进行分析。所有数据均使用 FlowJo 软件（Tree Star）进行处理。 FACSAria细胞分选仪和LSR II细胞仪均配备405、488和561 nm激发激光器，并使用BFP、530/30的450/50 nm（波长/带宽）滤光片检测发射信号nm 对于西兰花或 GFP，610/20 nm 对于 mCherry。将单细胞分选到含有 1% GlutaMAX、20% FBS 和 1% 青霉素/链霉素的冷冻 DMEM 培养基中的 96 孔板的各个孔中。对于西兰花信号分析，在 FACS 之前添加 5 µM DFHBI-1T。

A microscope (DMIRB; Leica Biosystems) was equipped with an EMCCD camera (iXon-897D; Andor Technology), mounted with a 2× magnification adapter and 100× oil objective lens (NA 1.4), resulting in a total 200× magnification equal to a pixel size of 80 nm in the image. The microscope stage incubation chamber was maintained at 37°C. A custom-built FRAP module was inserted between the fluorescent lamp and the microscope body, with a motorized translating mirror (Leica Biosystems) used to switch between imaging and FRAP mode. The FRAP module used a 488-nm laser (Obis; Coherent) set to 20 mW output power and created a diffraction limited spot of roughly 800 nm in diameter. The laser was fiber coupled (Pointsource; Qioptiq) and collimated out of an FC connector using a 30 mm focal length lens and projected into the sample with a 200 mm focal length lens and the microscope’s excitation tube lens and objective (Leica Biosystems). The FRAP module was controlled through a custom-built mechatronic assembly consisting of a fast linear actuator (Everest), DC motor driver (MCP Technology Systems), custom electrical circuitry, and 3D-printed adapters. The guide RNA used was 11 nt in length to achieve the necessary high signal/noise ratio for live-cell imaging as compared with one of 20 nt in length (Ma et al., 2016). Our use of a guide RNA with an 11-nt seed was also based on the facts that a guide RNA with a 10-nt seed sequence forms a stable Cas9–guide RNA–DNA target ternary complex in vitro (Richardson et al., 2016) and that a guide RNA with a 12-nt seed region determined the Cas9–guide RNA targeting specificity (Larson et al., 2013). The C3 target and its immediate periphery were bleached for 5 s, and the postbleach images were acquired up to 4 h as necessitated by the recovery times being observed in experiments with the various guide RNAs. GFP was excited with an excitation filter at 470/28 nm (Chroma Technology Corp.), and its emission was collected using an emission filter at 512/23 nm (Chroma Technology Corp.). Imaging data were acquired by MetaMorph acquisition software (Molecular Devices). Thresholds were set on the basis of the ratios between nuclear focal signals to background nucleoplasmic fluorescence. For fixed cells in Fig. 5, cells were imaged by using a custom-built, single-molecule, real-time microscope as previously described (Grünwald and Singer, 2010; Ma et al., 2016). In brief, imaging was performed on a custom-built, dual-channel setup housing a 150× 1.45 NA oil immersion objective (Olympus) combined with 200-mm focal length tube lenses (LAO-200, cvi; Melles Griot). This produces an effective magnification of 167× and a 95.8-nm pixel. The emission is split in the primary beam path onto three electron-multiplying charge-coupled devices (Andor Technology): iXon3-897E, iXon-897D, and iXonUltra-897U. Emission filters (Semrock) are 460/60 for BFP, 534/20 for mNeongreen, and 600/37 for mCherry. Excitation of fluorescent proteins was with a 405-nm and 515-nm diode laser (Obis, Coherent) and a solid-state 561-nm laser (SE; Cobolt), and the intensity and on/off switching were controlled by an acousto-optic tunable filter (AA Opto-Electronics). In Fig. 6 B and Fig. S5, the step size in z-stacks was 160 nm, and the exposure time was 150 ms. Each z-stack series was 31 frames, and each image was obtained by a maximum projection of a z-stack series using ImageJ.
显微镜（DMIRB；Leica Biosystems）配备 EMCCD 相机（iXon-897D；Andor Technology），安装有 2 倍放大适配器和 100 倍油物镜（NA 1.4），总放大倍数为 200 倍，等于图像中的像素大小为 80 nm。显微镜载物台孵育室保持在37°C。在荧光灯和显微镜主体之间插入定制的 FRAP 模块，并使用电动平移镜（Leica Biosystems）在成像和 FRAP 模式之间切换。 FRAP 模块使用 488 nm 激光器（Obis；相干公司），输出功率设置为 20 mW，并产生直径约为 800 nm 的衍射极限光斑。激光采用光纤耦合（Pointsource；Qioptiq），并使用 30 mm 焦距透镜从 FC 连接器准直，并使用 200 mm 焦距透镜以及显微镜的激发管透镜和物镜（Leica Biosystems）投射到样品中。 FRAP 模块通过定制机电组件进行控制，该组件由快速线性执行器 (Everest)、直流电机驱动器（MCP 技术系统）、定制电路和 3D 打印适配器组成。与长度为 20 nt 的指导 RNA 相比，使用的指导 RNA 长度为 11 nt，以实现活细胞成像所需的高信噪比（ Ma et al., 2016 ）。我们使用具有 11 nt 种子序列的向导 RNA 也是基于以下事实：具有 10 nt 种子序列的向导 RNA 在体外形成稳定的 Cas9-向导 RNA-DNA 靶三元复合物（ Richardson et al., 2016） ）并且具有 12 nt 种子区域的向导 RNA 决定了 Cas9 向导 RNA 的靶向特异性（ Larson 等，2013 ）。 C3 靶标及其直接外围被漂白 5 秒，并且根据在各种引导 RNA 的实验中观察到的恢复时间的需要，获得长达 4 小时的漂白后图像。 GFP 用 470/28 nm 的激发滤光片 (Chroma Technology Corp.) 激发，并使用 512/23 nm 的发射滤光片 (Chroma Technology Corp.) 收集其发射光。成像数据通过MetaMorph采集软件(Molecular Devices)采集。根据核焦点信号与背景核质荧光之间的比率设置阈值。对于图 5中的固定细胞，如前所述，使用定制的单分子实时显微镜对细胞进行成像（ Grünwald 和 Singer，2010 ； Ma 等人，2016 ）。简而言之，成像是在定制的双通道装置上进行的，该装置装有 150×1.45 NA 油浸物镜 (Olympus) 和 200 毫米焦距管透镜 (LAO-200，cvi；Melles Griot)。这可产生 167 倍的有效放大倍率和 95.8 纳米像素。发射在主光束路径中被分成三个电子倍增电荷耦合器件（Andor Technology）：iXon3-897E、iXon-897D 和 iXonUltra-897U。发射过滤器 (Semrock) 对于 BFP 为 460/60，对于 mNeongreen 为 534/20，对于 mCherry 为 600/37。荧光蛋白的激发采用 405 nm 和 515 nm 二极管激光器（Obis，Coherent）和固态 561 nm 激光器（SE；Cobolt），强度和开/关切换由声学控制光学可调滤波器（AA Opto-Electronics）。在图 6B和图 S5 中，z 堆栈的步长为 160 nm，曝光时间为 150 ms。 每个 z-stack 系列为 31 帧，每张图像都是使用 ImageJ 通过 z-stack 系列的最大投影获得的。

The time series FRAP images were registered and analyzed by Fiji and then plotted using Origin 9.0 software (OriginLab). The images were first corrected for cellular movements by using the StackReg plugin (Thévenaz et al., 1998). All fluorescence intensities were corrected by background subtraction. Double-bleach correction was performed by using the fluorescence level from the whole nucleus as the indication of photobleaching. The first post bleach image was set to time 0, and the prebleach image was set to intensity 1. We fitted the FRAP curves based on the Diffusion-Uncoupled model (McNally, 2008) with a minor modification: $FRAP\left(t\right)=A_1-{\Phi }_{app}{e}^{-k_{off}t},$ where t is time, A₁ is a constant to represent the highest recovery level, Φ_app is the apparent bleached fraction, and k_off is the dissociation rate.
时间序列 FRAP 图像由 Fiji 配准并分析，然后使用 Origin 9.0 软件（OriginLab）进行绘图。首先使用 StackReg 插件对图像进行细胞运动校正（ Thévenaz 等人，1998 ）。所有荧光强度均通过背景扣除进行校正。通过使用整个细胞核的荧光水平作为光漂白的指示来进行双漂白校正。第一个漂白后图像设置为时间 0，漂白前图像设置为强度 1。我们根据扩散非耦合模型（ McNally，2008 ）拟合 FRAP 曲线，并进行了较小的修改： $FRAP\left(t\right)=A_1-{\Phi }_{app}{e}^{-k_{off}t},$ 其中t是时间， A ₁是表示最高回收水平的常数，Φ _app是表观漂白分数， k _off是解离率。

Mining of chromosome-specific repeats was described previously (Ma et al., 2015). The chromosome 3–specific locus is situated in the subtelomeric region q29, having the identifier Chr 3: 195199022–195233876 in the human reference genome hg19 at the University of California Santa Cruz genome browser (http://genome.ucsc.edu).
先前描述了染色体特异性重复的挖掘（ Ma et al., 2015 ）。 3 号染色体特异性位点位于亚端粒区 q29，在加州大学圣克鲁斯分校基因组浏览器 ( http://genome.ucsc.edu ) 的人类参考基因组 hg19 中具有标识符 Chr 3: 195199022–195233876。

Cells were treated as described in Fig. 2 D, and RNA was extracted with an RNeasy Plus Mini Kit (QIAGEN) and then subjected to RT-PCR using the following primers and probe (Integrated DNA Technologies) for C3–guide RNA–2XBroccoli: Forward primer: 5′-TGGGCTCTAGCAAGTTCAAATAA-3′; complementary to nt 55–78 of the RNA; Probe: 5′-ACTTGAGACGGTCGGGGTCCAGATA-3′; complementary to nt 94–118 of the RNA; Reverse primer: 5′-CCCACACTCTACTCGACAAGATA-3′; complementary to 144–122 nt of the RNA.
如图 2D所示处理细胞，并使用 RNeasy Plus Mini Kit (QIAGEN) 提取 RNA，然后使用以下引物和探针 (Integrated DNA Technologies) 对 C3–guide RNA–2XBroccoli 进行 RT-PCR：正向引物：5'-TGGGCTCTAGCAAGTTTCAATAA-3'；与 RNA 的 nt 55-78 互补；探针：5′-ACTTGAGACGGTCGGGGGTCCAGATA-3′；与 RNA 的 nt 94-118 互补；反向引物：5'-CCCACACTCTACTCGACAAGATA-3'；与 RNA 的 144-122 nt 互补。

Figs. S1–S5 include the sequence and structure of C3–guide RNA–2XBroccoli, target interrogation related to the intranuclear Cas9 and guide RNA levels, target residence times from FRAP analysis, effects on residence time of shorter guide RNAs, and images of live-cell cleavage assay. Table S1 lists all guide RNA sequences used in this study.
无花果。 S1–S5 包括 C3–guide RNA–2XBroccoli 的序列和结构、与核内 Cas9 和指导 RNA 水平相关的目标询问、FRAP 分析的目标停留时间、对较短指导 RNA 停留时间的影响以及活细胞图像裂解测定。表 S1 列出了本研究中使用的所有指导 RNA 序列。

We thank Rita Strack and Samie Jaffrey for the Broccoli aptamer. We thank Aviva Joseph for sharing her lentiviral transduction protocol, Bonginkhosi Vilakati for performing the RT-PCR analyses, and Ankit Gupta, Mehmet Fatih Bolukbasi, Yu-Chieh Chung, and Feng He for valuable discussions. We are also grateful to Erik Sontheimer and Scot Wolfe for critical input and comments on the manuscript.
我们感谢 Rita Strack 和 Samie Jaffrey 提供西兰花适体。我们感谢 Aviva Joseph 分享了她的慢病毒转导方案，感谢 Bonginkhosi Vilakati 进行了 RT-PCR 分析，感谢 Ankit Gupta、Mehmet Fatih Bolukbasi、Yu-Chieh Chung 和 Feng He 进行了宝贵的讨论。我们还感谢 Erik Sontheimer 和 Scot Wolfe 对手稿的重要意见和评论。

This work was supported in part by the Vitold Arnett Professorship Fund, National Institutes of Health grant U01 DA-040588-01 and Amyotrophic Lateral Sclerosis Association grant 16-LGCA-300 (H. Ma and T. Pederson), and National Institutes of Health grants 5 U01 EB021238-02 to D. Grunwald and R01 GM102515 to S. Zhang.
这项工作得到了 Vitold Arnett 教授基金、国立卫生研究院拨款 U01 DA-040588-01 和肌萎缩侧索硬化症协会拨款 16-LGCA-300（H. Ma 和 T. Pederson）以及国立卫生研究院的部分支持将 5 U01 EB021238-02 授予 D. Grunwald，将 R01 GM102515 授予 S. Zhang。

The authors declare no competing financial interests.
作者声明不存在竞争的经济利益。