Switching the activity of Cas12a using guide RNA strand displacement circuits

The CRISPR effector protein Cas12a has been used for a wide variety of applications such as in vivo gene editing and regulation or in vitro DNA sensing. Here, we add programmability to Cas12a-based DNA processing by combining it with strand displacement-based reaction circuits. We first establish a viable strategy for augmenting Cas12a guide RNAs (gRNAs) at their 5′ end and then use such 5′ extensions to construct strand displacement gRNAs (SD gRNAs) that can be activated by single-stranded RNA trigger molecules. These SD gRNAs are further engineered to exhibit a digital and orthogonal response to different trigger RNA inputs—including full length mRNAs—and to function as multi-input logic gates. We also demonstrate that SD gRNAs can be designed to work inside bacterial cells. Using such in vivo SD gRNAs and a DNase inactive version of Cas12a (dCas12a), we demonstrate logic gated transcriptional control of gene expression in E. coli.

In this study, the authors showed strand displacement-mediated activation of gRNA can regulate the DNA cleavage by Cpf1 by trigger-RNA-dependent manner. Activity of AsCpf1 gRNA can be repressed by attaching switch domain that abolishes secondary structure of gRNA. They succeeded in activating the repressed gRNA via RNA strand displacement by trigger RNA, and they termed this gRNA as SD gRNA. They further engineered SD gRNA to be activated by orthogonal multiple RNAs and built AND logic gate in vitro cell-free system. However, the strand displacement principles used in this article is very similar to the previously published methods (e.g. Green, A. A. et al. Nature, 2017). Also, authors did not show that SD gRNA could work in living cells, leaving broad gap between their results and the problem they wanted to address (in line 48 to 51). Overall, their experiments clearly showed that SD gRNA could work as they designed in vitro, but significance and novelty are limited as pointed above and the current manuscript is not suitable for publication in Nature Communications.
Major points: 1. In the introduction (in line 48 to 51), the authors said the advantage of this system against the previously reported strand displacement system is its regulatability of non-engineered parts (such as genomic mRNAs) and availability in eukaryotic cells. However, these two advantages have not been demonstrated yet. The authors should show the data to demonstrate these advantages, which would increase the impact of this study. Fig. 2d, the authors claim that "one molecule of trigger leads to the activation of approximately one SD gRNA and the cutting of one target molecule." However, in Fig. 2c, they use the 125 nM of SD gRNA and 375 nM of trigger RNA, in which they observed the unprocessed SD gRNA. This suggests that one trigger RNA molecule cannot activate the same amount of SD gRNA. Also, they used 40 nM of SD gRNA and 10 nM of trigger RNA, greatly different condition from the Fig. 2c, which does not use 1 to 1 ratio of SD gRNA and trigger RNA. Therefore, there is no convincing experimental evidence to support author's claim. Besides, the possibility that one molecule of activated SD gRNA can cut multiple target DNA should also be examined.

In
3. In line 169, the authors say "we deliberately left the spacer domain fixed for our designs." However, the reason why they left the spacer domain fixed are not specified. Explaining the reason why they fixed or experimentally showing the programmability of this domain would be needed since they discuss "Randomizing the spacer would increase the sequence space" in line 274 to 275 and this is the potential merit of this device. 4. To claim the generality of the SD gRNA design, I think that seemingly sequence-dependent leak activity shown in Fig. S5b and Fig S4d should be investigated. Discussing the different characteristics of t3 sequence from t1 and t2 would help, but designing other sequences would be needed to understand the mechanism. 5. In Fig. S5c, while "AND2 p1 + AND2 p2" is shown, there is no "AND1 p1 + AND1 p2". As we can see in Fig. 4e, the activities of "AND2 p1 + AND2 p2" and "AND1 p1 + AND1 p2" are apparently different, which imply the difference of kinetics between these two conditions. Therefore, the author should also show the kinetics of "AND1 p1 + AND1 p2". 6. In Fig. S5c, the maximum fluorescence of "AND2 p1 + AND2 p2" is higher than that of the positive control (regular trigger), although the kinetics of the former is slower than the latter. Some explanations about this phenomenon are needed.
Minor points: 1. The term "spacer" they used to describe the sequence between handle and switch in SD gRNA was confusing since DNA-targeting sequence is also called spacer in CRISPR system and I recommend them to use other term. Also, I could not figure out what does "spacer" mean in the text "For single-stranded extensions, we tested a random, unstructured single-stranded sequence (ss v1), the full spacer of AsCpf1 (ss v2) and the full spacer taken from Francisella novicida Cpf1 (FnCpf1) (ss v3)." I recommend to use widely accepted terms and not to use the same term for different purposes in an article to help readers understand.
2. There is no description about the target of gRNAs used in Fig. 1. If t1 was the target of these gRNAs, the authors should describe it in page 4 or the legend of Fig. 1. 3. In line 190 to 191, they say "Finally, orthogonality is introduced even for fixed sequences by varying the usage of GU wobble base pairs," but it was not easy to immediately understand how they introduced orthogonality. Thus, it may be better to have figure showing the example.