Taming CasRx: When delivery becomes part of the safety mechanism

Main text

RNA-targeting CRISPR systems have transformed the ability to manipulate gene expression without permanently altering the genome. Among them, Cas13 effectors offer programmable RNA knockdown with considerable therapeutic potential, but their biology poses an unusual safety challenge: target recognition can activate collateral cleavage of non-target RNAs. This phenomenon has been demonstrated in mammalian cells and appears to depend on the Cas13 ortholog, target abundance, cellular context, and expression level.^1,2,3,4 In the previous issue of Molecular Therapy Nucleic Acids, Osakada et al.⁵ reported that adenoviral delivery of CasRx can achieve efficient RNA knockdown while markedly reducing detectable collateral activity in cultured cells and in mouse liver. Their findings highlight an important principle for therapeutic RNA targeting: controlling where, when, and how much Cas13 is expressed may be as important as choosing the Cas13 effector itself.^2,3,4,6

The same molecular architecture that makes Cas13 powerful creates a distinctive safety problem. In mammalian systems, collateral activity has been observed with several Cas13 effectors, although its magnitude varies considerably among orthologs and experimental settings.^2,3,4

Importantly, recent studies indicate that collateral activity is not simply an invariant property of Cas13; it can depend on target RNA abundance and on the intracellular abundance of Cas13 and guide RNA^2,3,7 (Table 1).

Table 1.

Key determinants and strategies for controlling collateral activity in therapeutic Cas13 applications

Determinant/strategy	Rationale	Relevance to CasRx safety	Reference
Cas13 effector abundance	excessive effector expression may increase collateral cleavage	controlled expression may preserve on-target activity while limiting collateral effects	Ai et al.2; Tong et al.8
Target RNA abundance	target engagement can influence Cas13 activation and collateral activity	highly abundant targets may alter the therapeutic window	Ai et al.2; Shi et al.3; Li et al.4
Effector engineering	reduce intrinsic collateral activity	high-fidelity Cas13 variants may provide an additional safety layer	Tong et al.8
Guide optimization	improve productive target recognition	may increase on-target efficiency without requiring excessive effector expression	Konermann et al.6
Delivery/expression control	restrict magnitude and duration of Cas13 exposure	adenoviral CasRx provides an example of this approach	Osakada et al.5; Hart et al.9
Tissue-selective delivery	restrict Cas13 activity to intended tissues	may improve therapeutic index	Shimizu et al.10
Transcriptome-wide monitoring	detect collateral effects not captured by selected reporters	necessary for comprehensive safety assessment	Shi et al.3; Li et al.4

This context dependence is particularly important for therapeutic development. A Cas13 system may produce substantial target knockdown while remaining below the level at which collateral cleavage becomes readily detectable. Conversely, high effector expression or prolonged exposure may shift the system into a regime in which collateral activity becomes biologically consequential. Indeed, experimental evidence has linked excessive Cas13 activity to widespread transcript perturbation and, in some in vivo settings, severe toxicity^3,4 Thus, an emerging question is not simply whether Cas13 has collateral activity, but how the therapeutic window between useful on-target activity and harmful collateral activity can be engineered.^2,3,4

Osakada et al. approach this problem from an unconventional yet compelling perspective: focusing on the delivery side rather than redesigning CasRx itself (Figure 1). They initially compared CasRx, PspCas13b, and Cas13X.1 using a dual-reporter system in human and mouse cell lines. Plasmid-mediated CasRx expression resulted in efficient on-target knockdown and substantial suppression of the non-target Gluc reporter in several cell types. PspCas13b showed lower collateral activity, whereas Cas13X.1 displayed weaker and more variable knockdown. These observations established CasRx as the most effective effector for the subsequent adenoviral experiments while also reinforcing the context dependence of Cas13 behavior.^2,5

Figure 1.

Adenoviral control of CasRx expression to manage collateral activity

Unregulated CasRx expression following plasmid transfection is associated with high collateral activity and off-target knockdown (left part of the image). Adenoviral delivery provides controlled CasRx expression while maintaining on-target Pcsk9 mRNA knockdown with undetectable collateral activity under the conditions examined (right part of the image). The vector incorporates regulatory elements, including a miR-122a-targeted sequence (center), illustrating how delivery and expression control can be incorporated into the safety architecture of therapeutic Cas13 RNA targeting.¹⁰

The central experiment was to package CasRx and its guide RNA expression cassette into an adenoviral vector engineered to limit vector-associated liver toxicity. In reporter-expressing cells, Ad-CasRx retained substantial target knockdown while the reduction in the collateral reporter was largely undetectable. The authors then generated a stable SK-HEP-1 cell line expressing both reporters, thereby avoiding differences in reporter abundance between plasmid transfection and adenoviral transduction. Under these controlled conditions, Ad-CasRx reduced Fluc expression to approximately 30% of the non-targeting control without significantly reducing CasRx and Gluc expression.^5,9

Perhaps the most informative observation is that CasRx expression itself was substantially lower after adenoviral transduction than after plasmid transfection. In the stable dual-reporter system, CasRx transcript abundance was approximately 7-fold higher following plasmid delivery, and this higher expression was accompanied by reductions in both the intended and non-target transcripts. These findings provide a plausible quantitative explanation for the different phenotypes and are consistent with previous work showing that limiting Cas13d copy number or expression can mitigate collateral activity.^5,7,9

The study then moves beyond artificial reporters to an endogenous therapeutic target, PCSK9. In cultured murine hepatocytes, Ad-CasRx reduced Pcsk9 mRNA while leaving several highly abundant endogenous transcripts unchanged. This is an important step because reporter assays can overestimate or underestimate collateral effects depending on reporter abundance and cellular localization. The observation that highly expressed endogenous transcripts were preserved strengthens the conclusion that collateral activity was below the detection threshold under these experimental conditions.⁵

The in vivo experiment provides a further translational dimension. Systemic administration of Ad-CasRx targeting PCSK9 reduced hepatic Pcsk9 mRNA to approximately half of control levels and lowered total serum cholesterol to about 60% of control. The authors did not detect major alterations in genes involved in cholesterol metabolism or hepatic function, and serum ALT was not elevated. The magnitude of the metabolic effect is broadly consistent with the biological consequences of PCSK9 suppression observed with RNAi-based approaches, supporting the feasibility of CasRx as an alternative RNA-targeting modality.⁵

The work nevertheless raises an important conceptual question that extends beyond adenovirus: is reduced collateral activity an intrinsic property of adenovirus-mediated CasRx delivery, or does adenovirus simply place CasRx within a safer expression range? The experiments establish an association among adenoviral delivery, lower CasRx expression, and reduced detectable collateral activity, but they do not completely separate the contributions of the delivery platform from that of effector abundance. This distinction matters because the same principle could potentially be implemented using other vectors, inducible systems, tissue-selective promoters, or engineered Cas13 proteins.^3,5,7,9 More broadly, these observations suggest that collateral activity should be considered a property of the entire Cas13 therapeutic system, which includes the effector, guide, target, delivery platform, and the level and duration of expression (Table 1).

A second issue concerns the meaning of “no detectable collateral activity.” The absence of measurable changes in selected reporters or abundant endogenous transcripts does not demonstrate transcriptome-wide absence of collateral cleavage. The authors appropriately acknowledge that the threshold doses for collateral activity remain to be defined and that latent effects should be examined by comprehensive transcriptomic analysis. This caveat is important because tissue-specific tolerance, duration of expression, target abundance, and vector dose could all alter the safety window.^3,4,5

The adenoviral platform also introduces its own translational considerations. Adenovirus offers a large packaging capacity and efficient transduction of hepatocytes, but vector-associated inflammation, leaky viral gene expression, and hepatotoxicity remain relevant limitations. The authors address these concerns by using an E4-122aT vector incorporating hepatocyte-enriched miR-122a regulation (Figure 1), a strategy previously developed to reduce adenovirus-associated toxicity^5,10

Looking forward, the most compelling implication may therefore be a layered approach to Cas13 safety. High-fidelity Cas13 variants can reduce intrinsic collateral activity, whereas negative autoregulation can constrain effector abundance. Controlled viral delivery could provide an additional layer by limiting intracellular exposure to Cas13. Combining effector engineering, guide optimization, quantitative expression control, tissue-selective delivery, and transcriptome-wide monitoring could produce a substantially more robust safety architecture for therapeutic RNA targeting^7,8,9 (Table 1).

The study by Osakada et al. is important because it shifts the discussion from whether Cas13 has collateral activity to a more therapeutically meaningful question: under what conditions does collateral activity become biologically significant? Their data suggest that delivery is not merely a logistical component of RNA-targeting therapy; it can influence the balance between therapeutic activity and collateral risk. Although additional studies will be required to define expression thresholds, dose dependence, tissue specificity, and long-term transcriptomic effects, the work provides a useful framework for engineering CasRx toward clinically relevant applications.

In the broader development of programmable RNA therapeutics, this may prove to be the most durable lesson of the study. Rather than treating the nuclease, guide, vector, and target as independent variables, future platforms may need to optimize them as a single pharmacological system. The therapeutic objective is not maximal Cas13 activity but sufficient activity within a defined biological window. Adenoviral CasRx provides an experimentally tractable demonstration of this principle and may help guide the development of safer RNA-targeting CRISPR technologies across tissues and disease settings.^5,7,8,9

Declaration of interests

The authors declare no competing interests.