Boosting, Not Breaking: CRISPR Activators Treat Disease Models

Main Text

Genome editing with DNA-targeting tools, such as CRISPR/Cas9, zinc finger proteins, and transcription activator-like effectors, has emerged as an exciting new approach to gene and cell therapy.¹ Genome editing technologies have become most well-known through the use of these tools as engineered DNA-targeting nucleases that create breaks in the genomic DNA to create changes to gene sequences. This includes repairing genes that are mutated in inherited disorders or knocking out genes that facilitate unwanted functions, such as viral infection. However, an alternative and potentially even more impactful application of these DNA-targeting technologies is not to cut DNA, but rather to change the level or regulation of gene targets to achieve therapeutic outcomes.²

The manipulation of gene expression is a powerful approach to treating disease. For example, RNA interference has been widely used to reduce gene expression for therapeutic applications. However, like RNA interference, most tools for perturbing gene expression are designed to repress target genes, while tools for gene activation have lagged behind. Nevertheless, there are many examples where therapeutic benefit could be conferred by producing an excess of a particular gene.³ In fact, the first clinical trial to test one of these DNA-targeting tools was an engineered zinc finger protein designed to activate the expression of vascular endothelial growth factor (VEGF)-A for diabetic neuropathy (ClinicalTrials.gov: NCT00110500). In a recent study, Liao et al.⁴ extend this concept by using the CRISPR/Cas9 system in conjunction with modified guide RNAs (gRNAs) to achieve gene activation in mice. With this approach, they demonstrate therapeutic efficacy in three different mouse models of disease, including reduced kidney damage and death following acute kidney injury, increased muscle mass in models of muscular dystrophy, and recovery of glucose balance in a model of type I diabetes. This toolset further opens the door for the development and testing of gene therapies based on targeted gene activation. Moreover, it lays the path to a new class of therapeutic targets and underscores the potential for CRISPR-based genetic screens to further uncover novel mechanisms to combat disease.

The CRISPR/Cas9 system has been adapted as a tool for generating precise genome modifications by generating DNA breaks. However, the use of a nuclease null, catalytically inactive Cas9 (dCas9) as a modular DNA-binding domain is a powerful tool for other types of genome manipulations, including gene activation, gene repression, or targeted changes to epigenetic marks.² A variety of protein domains have been identified that alter gene regulation or directly catalyze reactions to change epigenetic marks. By directly fusing these protein domains to dCas9 or recruiting them via protein-protein or protein-gRNA interactions,⁵ it is possible to effectively alter gene expression and chromatin structure.² Several different dCas9-based activator platforms have been described, including first-generation systems based on fusion to the VP64 activator domain,6, 7, 8 next-generation versions based on fusions to or recruitment of more robust engineered activator domains,5, 9, 10 and, finally, fusions to enzymes that catalyze the addition of activating histone marks¹¹ or removal of repressive DNA methylation.¹²

In their study, Liao et al.⁴ designed an in vivo system for gene activation by creatively integrating several previous technological innovations. First, they used a Cas9 transgenic mouse that expresses Cas9 from the Rosa26 locus.¹³ Second, they virally delivered gRNAs that contain an MS2-binding aptamer loop together with a MS2-p65-HSF1 (MPH) fusion protein previously shown to achieve potent activation of target genes when recruited to dCas9/gRNA binding sites in gene promoters.⁵ Finally, rather than use the full-length gRNAs that typically contain 20 nucleotide (nt) targeting sequences, they used “dead” gRNAs (dgRNAs) that only contain 14–15 nt targeting sequences. These shorter targeting sequences facilitate Cas9 binding to a target sequence but do not allow Cas9 to undergo the conformational change necessary to activate nuclease activity and cut DNA.14, 15 Consequently, Liao et al.⁴ were able to use the transgenic mouse that expressed the Cas9 nuclease, but only for targeting the MPH activator, not for DNA cutting, similar to the function of dCas9.

A luciferase reporter mouse crossed to the Cas9 mouse was first used to determine the best combination of modified dgRNA and activator domain. Next, they enhanced muscle mass in the hind limbs of neonatal Cas9 mice through the local delivery of an adeno-associated virus (AAV) encoding a dgRNA targeting the promoter of follistatin, a known inhibitor of the negative regulator of muscle mass myostatin.

After this initial in vivo experiment, the authors examined three mouse models of disease that can be treated through the activation of endogenous genes. The first model was acute kidney injury, to which they systemically delivered AAV encoding gRNAs that activate the expression of Klotho or IL10 prior to cisplatin-induced injury, resulting in prolonged survival. Next, they used a model for the treatment of type I diabetes in which activation of PDX1 in liver cells can induce glucose-responsive insulin production. In addition to single-gene activation experiments, the authors showed that two genes can be activated simultaneously when two dgRNAs are provided. This is important as manipulation of multiple genes may be required for effective therapeutic benefit in some diseases.

Finally, activation of either follistatin or utrophin was explored as a treatment for the mdx mouse model of muscular dystrophy. Importantly, this work used AAV delivery of the dgRNA and the Cas9 gene, rather than using the Cas9 transgenic mice, providing proof-of-concept of a therapeutic approach in which all necessary components are delivered to the animal (Figure 1). These experiments clearly show that delivery of Cas9 coupled with dgRNAs is capable of sufficient gene activation to show a phenotypic change in muscle mass and physiologic benefit to the mdx mice as compared to non-targeting dgRNA controls.

Figure 1.

Schematic of Target Gene Activation through Co-delivery of AAV-Cas9 and AAV-dgRNA-MPH

Co-delivery of AAV-Cas9 and AAV-dgRNA-MPH into the fore and hind limbs of 2-day-old mdx mice leads to expression of spCas9, dgRNA, and MS2-p65-HFS1 in co-transduced cells, resulting in histone acetylation adjacent to the dgRNA target site and subsequent upregulation of either utrophin or follistatin mRNA (depending on the dgRNA sequence). Increased gene expression leads to increased muscle mass and grip strength, providing proof-of-principle that delivery of their targeted gene activation (TGA) system can be used for potential therapeutic applications.

While these experiments show an impressive ability to ameliorate a variety of disease models in different tissues, there are still significant challenges to overcome for preclinical development of this approach. In particular, it is unclear how long the gene activation will persist. Because AAV can be maintained as an episome in post-mitotic tissues for many years, it is possible that Cas9 and dgRNA expression and the downstream effects will be long-lived. However, continued overexpression of target genes may be desirable for some disease indications, but not others. If continued expression of Cas9 and the dgRNA is necessary in vivo, then immune responses to the bacterial Cas9 protein may become a concern.¹⁶ In some cases, activation of target genes may be epigenetically sustained and heritable even in the absence of the original activator.¹² For such applications, the use of non-viral, transient delivery vehicles may be pursued.¹⁷ In fact, efficacy will ultimately be determined by both the efficiency and tropism of in vivo delivery, and future work will be necessary to explore to what extent different genes can be activated in different tissues.

Activating the expression of endogenous genes opens up a whole new class of potential drug targets. For example, this approach could be used to increase the levels of genes that would otherwise be too large to fit into typical gene therapy vectors, as the authors’ demonstrate nicely via the example of activating the large utrophin gene. This general strategy also underscores the power of activating perturbations in the genome and the utility of CRISPR-based gain-of-function screens to discover new therapeutic targets among known genes5, 18 and within the non-coding genome.¹⁹ In fact, extending the current study to activation of non-coding RNAs and distal regulatory elements will be an exciting future area of research.

The findings presented by Liao et al.⁴ represent a clear step forward for the use of CRISPR-based activation tools for in vivo therapeutic and research applications. These impressive results across multiple tissues and disease models highlight the need for continued work on understanding and developing more robust systems and delivery vehicles for in vivo gene activation and repression as well as targeted epigenetic modulation.