AAV-mediated gene editing lights up the lung

Main text

In the current issue, Liang et. al.¹ describe findings from the NIH Somatic Cell Genome Editing consortium tackling the difficult problem of gene editing in the airways for diseases such as cystic fibrosis. They demonstrate efficient CRISPR-based gene editing in both the large and small airways (∼16%–26%) using a dual-recombinant adeno-associated viral (rAAV) vector approach that activates a fluorescent reporter by targeted deletion of a LoxP-STOP-LoxP sequence within transgenic mice. These studies demonstrate clear proof-of-concept for gene editing of cellular targets that are important in cystic fibrosis lung disease with an approach that overcomes the packaging limitations of rAAV vectors. Next steps needed to apply this approach to cystic fibrosis include translation of this technology to homology-directed repair, non-homologous directed mini-gene insertion, or base editors.

CRISPR-based gene editing approaches allow for targeted genetic modifications to a gene of interest (GOI), ranging from substitutions of single/multiple-nucleotides, large deletions (exon skipping), gene disruption (dominant disorders), or targeted insertions of an intact/partial cDNA at a predetermined site. Editing an endogenous gene at a disease-causing genetic locus maintains its physiologic expression and thus is ideal for complex disorders where the GOI has highly regulated and differential expression among distinct cell types of the target organ (e.g., the lung in cystic fibrosis). Applications of gene editing in humans are showing great promise for sickle cell disease using an ex vivo mutation repair strategy in hematopoietic stem cells and for transthyretin amyloidosis by disrupting a dominantly inherited disease-causing allele using an in vivo strategy. While the effort to develop these two gene editing drugs for clinical use in humans has been significant, these examples represent relatively low-hanging fruit and many inherited diseases with more complex etiologies will require more challenging in vivo approaches.

The major challenge for in vivo gene editing of DNA is the delivery of the molecular cargo required to mediate repair. In the case of homology-directed repair, the vector must be capable of delivering both CRISPR components and a homologous DNA fragment to facilitate recombination/repair at the site of a double-stranded DNA break (DSB). Alternative approaches using base editing or prime editing are also attractive since they do not require a homologous DNA fragment to the target site or DSB, but these approaches require nuclease-dead dCas9 fusion proteins that are very large. Both of these approaches require packaging cargo that exceeds the capacity of the most widely used vector system for in vivo delivery (i.e., rAAV). Despite this limitation of rAAV for gene editing, its use in gene editing has merit since the rAAV genome undergoes highly efficient recombination at sites of homology in the host genome and this efficiency is enhanced two orders of magnitude by a DSB at the targeted site.²

With a maximum rAAV packaging capacity of ∼4.9 kb, most CRISPR gene editing applications using this vector system will require a dual-rAAV approach to deliver all the necessary components (Cas9, gRNA, and the donor fragment for homology-directed repair/insertion or homology-independent targeted insertion). The success of such a strategy requires highly efficient transduction of both rAAV vectors in the same cell. The utility of dual-rAAV delivery to circumvent the packaging limitations of this vector system is not new. It emerged two decades ago as a trans-splicing strategy capable of delivering large genes, where intermolecular dimerization occurs between two rAAV genomes in co-transduced cells to reconstitute an intact mini-gene expression cassette from an oversized gene that is split into two-halves flanked by splicing donor or acceptor sequence. While this approach delivers robust transgene expression to mouse skeletal muscle, heart, and liver, it was not effective in reconstituting split transgene expression in the lung.³ The failure of this dual-rAAV split transgene approach to work in the lung could be due to inefficient co-transduction of lung epithelia or the fact that lung epithelial cells do not concatemerize rAAV genomes. Studies by Liang et al. have begun to address critical aspects of using dual-rAAV vector approaches in the lung for gene editing.¹

Liang et al. use an elegant approach to address a very simple question: Can two rAAV vectors express the required components for CRISPR gene editing within the same airway epithelial cell? As a readout for gene editing, they targeted excision of a LoxP-STOP-LoxP cassette that prevents Tomato expression from the Rosa26 locus of Ai9 mice to “turn on the lights” in lung epithelial cells. To ensure that co-transduction of airway epithelial cells was required to flip the switch of the reporter, they used one rAAV5 vector to express SpCas9 and another rAAV5 vector to express two gRNAs and an EGFP reporter. Targeted excision of the LoxP-STOP-LoxP element, followed by nonhomology end joining (NHEJ), led to the activation of Tomato expression.

This well-designed study quantified both gene delivery (EGFP) and gene editing (Tomato) within transduced airway epithelial cells of the large and small airways. Analyses concluded 19%–26% gene editing efficiency in the large and small airways, as well as in alveolar cells in the lung parenchyma. Notably, the percentage of EGFP-positive airway epithelial cells (∼30%) was only slightly higher than Tomato-positive cells (∼23%), demonstrating that the SpCas9 vector co-transduced 77% of cells transduced with the sgRNA/EGFP vector. An equal and random pattern of 30% transduction with the two vectors would only predict 9% gene editing (i.e., co-transduction). Two possibilities could explain these findings: (1) higher rates of transduction with the SpCas9 vector (i.e., the functional titer was higher than that of the sgRNA/EGFP vector), or (2) very efficient rAAV5 transduction of a subset of airway epithelial cells (i.e., a subset of airway cells was refractory to transduction). Given that the percentage of ciliated cells (∼30%) and club cells (∼15%) edited was fairly similar, these findings are unlikely to be explained by a cell-specific tropism. Quantification of SpCas9 expression would have been helpful in differentiating these two possibilities. Regardless of potential differences in the potency of the two rAAV vector preparations, the reported ∼23% efficiency of NHEJ-based gene editing excision is quite impressive. This may also underestimate the efficiency of SpCas9-mediated DSBs and thereby co-transduction of the two rAAV vectors, since a simultaneous cleavage at both ends of the LoxP-STOP-LoxP (Tomato positive) would likely be less efficient than sequential cleavage followed by NHEJ (Tomato negative).

It is generally accepted that homology-dependent repair is more efficient in proliferating airway epithelial cells, whereas repair of DSBs in quiescent, differentiated, cell types, such as ciliated cells, is largely limited to NHEJ.⁴ Thus, applications of the described dual-rAAV approach to gene editing in cystic fibrosis would be most suitable for homology-independent targeted insertion of a partial cDNA.⁴ Given that the half-life of a post-mitotic ciliated cell in the mouse airway is ∼17 months, or about half the life-expectancy of a mouse in captivity,⁵ this approach may lead to durable treatments in CF patients. However, gene editing of multipotent airway progenitors would be most ideal. This approach also efficiently targets club cells, and a subset of club cells is thought to be multipotent progenitors in the mouse intrapulmonary airways. Notably, these progenitors have been shown to be transduced by rAAV5 and clonally expand following airway injury.⁶ Thus, future applications of this dual-rAAV CRISPR approach targeting conditional CRE expression in a Confetti mouse lung injury model would allow for quantification of gene editing efficiencies in airway progenitors using clonal analysis (Figure 1).

Figure 1.

Quantifying the efficiency of CRISPR gene editing in airway stem/progenitor cells using a dual-rAAV vector approach

(A) Dual-rAAV CRISPR approach. Intratracheal delivery of the dual-rAAV CRISPR system includes two separate rAAV vectors that express either the spCas9 or two sgRNAs that target excision of a strong transcriptional termination sequence, which prevents expression of an encoded recombinase in transgenic mice. (B) Activation of CRE^ERT2 expression. The dual-rAAV CRISPR system would be administered to double-transgenic mice bearing a conditional CRE^ERT2 expression cassette, such as ROX-STOP-ROX-CRE^ERT2, and a Confetti Cre-reporter. Activation of CRE^ERT2 expression only occurs in the airway epithelial cells co-transduced with both rAAV vectors, where the CRPISR/Cas9 complex excises the transcriptional termination sequence. (C–E) Confetti labeling and injury-induced clonal expansion of airway stem/progenitor cells. (C) Co-delivery of dual-rAAV CRISPR activates CRE^ERT2 expression in gene-edited cells but not fluorescence. (D) Tamoxifen treatment induces the translocation of CRE^ERT2 into the nucleus of gene-edited airway epithelial cells to enable CRE-mediated recombination of the Confetti reporter and random labeling of cells with one of four possible fluorescent reporter proteins. (E) With cell turnover, which can be accelerated by airway epithelial injury (e.g., with naphthalene, SO₂, polidocanol, etc.) stem/progenitor cells proliferate to repair the damaged epithelium, expand clonally, and differentiate into contiguously labeled groups of cells. The efficiency of dual-rAAV CRISPR editing of stem/progenitor cells can be quantified by the number and size of traced clones in the airways. Clones containing many cells and multiple cell types are typically associated with multipotent stem cells, while clones containing fewer cells of the same cell type are typically associated with transient-amplifying progenitors that have a more limited capacity to proliferate and differentiate.