Cas9 slide‐and‐seek for phage defense and genome engineering

Abstract

How does the Cas9 nuclease locate a specific 20‐nucleotide target sequence in a crowded intracellular environment packed with mega bases of distracting non‐target DNA? Previously, it was shown that Cas9 finds DNA targets via three‐dimensional diffusion. In this issue of The EMBO Journal, Globyte et al (2019) reveal another dimension of the search process, which involves short‐range one‐dimensional sliding. These results have implications for understanding the natural function of Cas9 and its applications in genome engineering experiments.

Cas9 is an RNA‐guided nuclease that is routinely used for targeted genome engineering in eukaryotic cells. However, the roles and responsibilities of genome engineers are profoundly different from the natural selective pressures that optimized Cas9 for protection from invading genetic parasites, like phages. To optimize Cas9 for applications in genome engineering, it is critical to understand the biophysical mechanisms that govern target identification. In this issue of The EMBO Journal, Globyte et al (2019) show that the Cas9 target search process involves a combination of 1D sliding and 3D diffusion.

Unwinding all the double‐stranded DNA (dsDNA) in a cell to find a complementary target would be a slow and energetically expensive process. Instead, target identification by Cas9 first relies on detection of a short, double‐stranded sequence motif known as the PAM (protospacer adjacent motif; Anders et al, 2015). PAM recognition by Cas9 destabilizes the duplex and facilitates RNA‐guided strand invasion for detection of a complementary (protospacer) target (Sternberg et al, 2014). This sequential process facilitates target finding in a crowded intracellular environment. But how does Cas9 locate PAMs, how long does it take to find the PAM with an adjacent complementary (protospacer) target, and how does the landscape of this search change when we move Cas9 from a small bacterial cell containing ~600 μm of DNA to a human nucleus containing over 2 m of DNA (Fig 1)?

Figure 1. Cas9 search mechanisms and PAM distributions.

(A) Schematic representation of Cas9 1‐dimensional diffusion (sliding) with probabilities assigned by Globyte et al (2019) (B) Schematic of Cas9 average dwell times (Δτav) reported by Globyte et al (2019) on DNA substrates with different number of PAMs. (C) Physical lengths and PAMs distributions in genomes of Homo sapiens, Streptococcus pyogenes, and bacteriophage A25.

Many DNA‐binding proteins, such as the Lac repressor LacI, find their targets at speeds that approach, or surpass, the limits of 3‐dimensional diffusion (Halford, 2009). These proteins commonly rely on a combination of 1D sliding and 3D hopping or jumping. Sliding involves short‐range movements along the contour of the DNA, whereas hopping and jumping require dissociation and rebinding events over short (~10‐bps, hopping) or long distances, respectively. The relative contributions of sliding, hopping, and jumping have been optimized by evolution for each target‐protein pair (Halford, 2009).

Single‐molecule tracking of Cas9 has previously shown that Cas9 predominantly uses 3D diffusion (e.g., jumping) to locate DNA targets (Sternberg et al, 2014; Jones et al, 2017; Knight et al, 2015). Cas9 frequently samples off‐target DNA, where it dwells for an average of 750 ms in human cells or 30 ms in Escherichia coli (Jones et al, 2017; Knight et al, 2015). However, previous methods did not have the resolution to detect short‐range sliding, representing a gap in our knowledge of how Cas9 locates targets. In this issue of The EMBO Journal, Globyte et al (2019) address this knowledge gap using single‐molecule Förster Resonance Energy Transfer (smFRET) assays designed to examine the behavior of Cas9 with high spatial and temporal resolution. Their results suggest that Cas9 frequently slides between adjacent PAMs, and that the probability of sliding to a neighboring PAM decreases exponentially as the distance traveled increases (Fig 1). Models of target search by a DNA‐binding protein, which use sliding, hopping, and jumping to find its target rapidly, suggest the optimal sliding length is ~10 times the size of its target (Halford, 2009). Therefore, the max sliding distance measured (i.e., ~25 base pairs) suggests that sliding by Cas9 has been optimized for rapid detection of short PAMs, rather than the longer protospacer. Additionally, Globyte et al show that the average dwell time of Cas9 is nearly 2‐fold larger (2.5 s) on DNA containing 5 PAMs in close proximity, compared to a sequence containing 1 PAM (Globyte et al, 2019; Fig 1). They show that this increased dwell time is due to shuttling of Cas9 between nearby PAMs. These observations help explain why Cas9 bound more strongly to PAM‐rich regions of DNA (Sternberg et al, 2014), and illustrate how PAM‐rich zones may sequester Cas9 in vivo. 86% of the 142,700 PAMs in the Streptococcus pyogenes genome lie within 25 bps of another PAM, which suggests that they will frequently be sampled by sliding. The human genome contains three orders of magnitude more PAMs, and an even larger proportion (i.e., 94%) of these are within 25 bps of another PAM. Therefore, Cas9 variants that rely heavily on sliding during target search are likely to be sequestered in the many PAM‐rich islands present within the human genome.

In the natural context of a bacterial cell, and in the absence of an infection, a certain population of Cas9 may associate with a few high affinity, PAM‐rich islands on the host genome, while the remaining Cas9s are free to search for PAMs and complementary DNA targets. When foreign DNA is introduced to the cell, Cas9 associates with new DNA via long‐range jumping. However, once Cas9 has bound to this new segment of DNA, sliding and hopping behaviors are important for ensuring that Cas9 interrogates the foreign DNA. The Cas9 slide‐and‐seek behavior described by Globyte et al (2019) is an important new dimension to the target search process that protects bacterial cells from infection.

Recently, many strategies have been developed to redesign Cas9 effectors and the RNA guide for improved gene editing specificity (Wilkinson et al, 2018). Mutations of S. pyogenes Cas9 that remove non‐sequence specific interactions with the phosphate backbone of the complementary DNA strand (SpCas9‐HF1) significantly reduced off‐target cleavage events and DNA association rates, while retaining most on‐target activity (Kleinstiver et al, 2016; Singh et al, 2018). Therefore, it is possible that mutations in SpCas9‐HF1 decreased the relative contributions of sliding and hopping in favor of jumping. Thus, mutational manipulation of search mechanisms may re‐optimize Cas9 for target searching in the context of a human genome, where a jump‐and‐seek approach may be preferable to slide‐and‐seek.

The EMBO Journal (2019) 38: e101474