Correction of Airway Stem Cells: Genome Editing Approaches for the Treatment of Cystic Fibrosis

Abstract

Cystic fibrosis (CF) is an autosomal recessive disease caused by variations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. Although CF affects multiple organs, the primary cause of mortality is respiratory failure resulting from poor clearance of hyperviscous secretions and subsequent airway infection. Recently developed CFTR modulators provide significant therapeutic benefit to the majority of CF individuals. However, treatments directed at the underlying cause are needed for the ∼7% of CF patients who are not expected to be responsive to these modulators. Genome editing can restore the native CFTR genetic sequence and function to mutant cells, representing an approach to establish durable physiologic CFTR correction. Although editing the CFTR gene in various airway cell types may transiently restore CFTR activity, effort is focused on editing airway basal stem/progenitor cells, since their correction would allow appropriate and durable expression of CFTR in stem cell-derived epithelial cell types. Substantial progress has been made to directly correct airway basal cells in vitro, theoretically enabling transplantation of autologous corrected cells to regenerate an airway with CFTR functional cells. Another approach to create autologous, gene-edited airway basal cells is derivation of CF donor-specific induced pluripotent stem cells, correction of the CFTR gene, and subsequent directed differentiation to airway basal cells. Further work is needed to translate these advances by developing effective transplantation methods. Alternatively, gene editing in vivo may enable CFTR correction. However, this approach will require robust delivery methods ensuring that basal cells are efficiently targeted and corrected. Recent advances in gene editing-based therapies provide hope that the genetic underpinning of CF can be durably corrected in airway epithelial stem cells, thereby preventing or treating lung disease in all people with CF.

Cystic Fibrosis Overview

Cystic fibrosis (CF) is an autosomal recessive condition resulting from alterations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. Over 2,000 variants, with ∼350 clearly pathogenic sequences, have been identified, the most common being the deletion of the 3 bases coding for the 508th amino acid residue: phenylalanine (F508del). Pathogenic variants ultimately result in reduced CFTR ion channel activity that normally mediates anion transport, principally Cl⁻ and HCO₃⁻, across epithelial barriers. The CFTR function is essential for the normal hypotonicity of sweat, and to prevent dehydration and acidification of luminal secretions in tubular structures, including the vas deferens, pancreatic ducts, and lung airways, leading to hyperviscous secretions and the characteristic CF pathology. CF disease affects multiple organs, including the lungs, pancreas, intestines, sinuses, and reproductive tracts. CF originally received its name from fibrocystic pancreatic changes noted in young children succumbing to the disease in the mid-20th century. However, with improved therapies, it is the pulmonary manifestations that are now most consequential for CF patients. In healthy lungs, transport of Cl⁻ and HCO₃⁻ through the CFTR channel in the apical membrane of airway epithelial cells draws water into the lumen, which hydrates and neutralizes the particle-trapping mucus and facilitates mucociliary clearance. Loss of CFTR function in the lungs leads to thick mucus secretions that impede pathogen clearance; this impairment results in chronic infection, inflammation, and progressive airway remodeling culminating in ectasis of bronchi and bronchioles and end-stage lung disease.

Historically, CF carried a grim prognosis. Therapies targeting CF lung disease, such as approaches to enhance airway clearance, infection prevention and treatment, and anti-inflammatory interventions, slow the rate of declining lung function and, together with an earlier diagnosis of CF, led to a dramatically extended life span.¹ However, patients invariably either succumbed to poor lung function or underwent life-extending lung transplantation. More recently, the development of CFTR modulator therapies, which target molecular processing and activity of abnormal CFTR proteins, represents a transformative advancement in CF therapy. For example, the recently available highly effective triple drug combination (elexacaftor/ivacaftor/tezacaftor) affords significant therapeutic benefit to the vast majority of CF patients.^2,3 However, questions have arisen as to whether modulators represent a lifelong cure since lung function, improved with ivacaftor treatment alone, returned to the pretreatment baseline by 5 years.⁴ Furthermore, the use of modulators is limited by the nature of the causative CFTR mutation since only those that result in the CFTR protein can be targeted by these agents. Approximately 7% of CF individuals in the United States are not expected to benefit from the triple drug combination because they express little or no CFTR protein due to either nonsense mutations, splice-site mutations, or major deletions, or they produce protein variants that are nonresponsive. For these people, restoration of CFTR function will require novel and potentially more challenging approaches.

Numerous CFTR-directed therapeutic approaches are currently in development, including small molecules for premature termination codons (PTCs), oligo therapies for splice mutations, and mutation-agnostic CFTR mRNA or protein delivery. Approaches to durably compensate or reverse the CFTR mutation, shown in Fig. 1, include the following: (1) site-nonspecific integration of CFTR-expressing cassettes into genomic DNA, (2) site-specific genome editing resulting in integration of a CFTR-expression cassette into a safe-harbor site, or (3) directly editing the CFTR genomic DNA to correct or compensate for the mutation. Virus-mediated, site-nonspecific integration (e.g., via lentivirus or foamy virus vectors) is capable of restoring the missing gene product, depending upon the strength of the promoter directing CFTR cDNA expression and in what location(s) the vector integrates (Fig. 1A). Integration of the cDNA expression cassette into a targeted safe-harbor site has the theoretical advantage of providing a consistent level of cDNA expression per cell (Fig. 1B). We note that both the aforementioned “gene addition” approaches leave the endogenous mutant CFTR loci untouched. Directly editing the endogenous CFTR locus to correct specific mutations (Fig. 1C), in principle, is able to restore the expression of the wild-type (wt) or normal CFTR protein to levels of expression appropriate for that particular cell type (i.e., high-level expression in normally high expressing cells). Finally, directly targeting either a partial or full-length CFTR cDNA into the endogenous CFTR locus (Fig. 1D) has the potential advantage of correcting or compensating for all CFTR mutations downstream of the integration site. If this can be done while retaining the native CFTR chromatin structure and regulatory sequences, it also has the possibility of restoring appropriate cell-type-specific expression. In this review, we focus on the use of genome editing with the objective of restoring physiological levels of CFTR expression and function.

Figure 1.

Approaches to durably compensate or reverse CFTR mutations. (A) Site-nonspecific integration of a wt CFTR-expression cassette into genomic DNA to compensate for CFTR mutations; the red star represents F508del as example of a CFTR mutation. The expression cassette includes a viral or cellular promoter (shown in green) directing expression of the full-length CFTR cDNA (exons 1–27) (shown in blue). (B–D) Approaches for site-specific genome editing. (B) Targeted integration at safe-harbor locus (e.g., AAVS1) of a wt CFTR-expression cassette to compensate for CFTR mutations; again, the expression cassette includes a viral or cellular promoter directing expression of the full-length CFTR cDNA. (C) Site-specific correction of CFTR mutations directly in the endogenous CFTR locus. (D) Targeted integration at CFTR locus of a partial CFTR cDNA (shown in blue). For example, site-specific integration of the partial CFTR cDNA exons 3–27, preceded by an SA sequence and followed by a polyadenylation signal (pA) in intron 2 would, in principle, restore wt CFTR expression irrespective of the specific CFTR mutation occurring downstream of the integration site. CFTR, cystic fibrosis transmembrane conductance regulator; SA, splice acceptor; wt, wild type.

Basal Cells as Stem/Progenitor Cells of the Proximal Airway

Although CF is a multiorgan disease, this review focuses on efforts to edit airway epithelial cells. As we consider how one might apply the aforementioned editing approaches to restore CFTR function to the CF airway, it is important to consider the nature of the airway epithelium: its architecture and cellular composition, its cell-type-specific expression of CFTR, and how it is affected by CF disease. There exist excellent reviews on these subjects.^5,6 Our intent here is to highlight those aspects most relevant to ex vivo or in vivo genome editing. Given our therapeutic interest, we primarily focus on the human airway, but draw insights from other animal models where helpful.

The architecture and cellular composition of the human airway epithelium vary as one moves down the conducting airways from the trachea to the bronchi, bronchioles, and finally, the distal alveolar region.⁷ The surface airway epithelium of the large airways comprises a pseudostratified epithelium primarily consisting of multiciliated cells, secretory cells (both serous and goblet cells), and basal cells, as well as the less abundant ionocytes, pulmonary neuroendocrine cells (PNECs), and brush cells. Whereas all epithelial cell types at this airway location contact the basement membrane, basal cells are unique in that they have minimal if any contact with the airway lumen.⁸

Single-cell RNA sequencing (scRNA-seq) studies have provided clarification regarding the identity of CFTR mRNA-expressing cell types in the airway epithelium.^9–11 Although ciliated cells were previously thought to be significant reservoirs of CFTR based upon immunofluorescence detection methods,¹² only extremely low levels of CFTR mRNA expression per cell were observed in a small fraction of ciliated cells.^9–11 The highest level of CFTR expression per cell was found in ionocytes, based upon analysis of both transcriptome and proteome; whereas lower levels of expression per cell were seen in secretory cells. There is still much that is unknown regarding the relative importance of these CFTR-expressing cell types in the human surface airway epithelium. In particular, the amount of CFTR channel activity and overall function in the epithelial barrier context attributable to the rare but high CFTR-expressing ionocytes versus the amount due to the far more abundant but lower CFTR-expressing secretory cells or even ciliated or basal cells (Fig. 2A).^6,9,10 Also, further experimental studies will be required to determine whether restoration of CFTR expression in a particular cell type (e.g., ionocyte) versus all CFTR-expressing cell types is both necessary and sufficient to restore CFTR channel function to the airway. We note that this highly regulated cell-type-specific expression of CFTR mRNA suggests that editing the endogenous CFTR locus might be necessary to restore appropriate levels of corrected CFTR per cell.

Figure 2.

Basal cells as stem/progenitor cells of the proximal airway. (A) Schematic of proximal airway architecture, including basal cells, secretory cells, ciliated cells, and ionocytes; CFTR channels are shown in schematic form for ionocytes and secretory cells (green rectangles). A rough measure of relative CFTR mRNA expression level per cell is shown for each cell type. We note that scRNA-seq analyses^6,9–11 demonstrate a wide range of CFTR mRNA expression for each cell type, with some cells in each cell type, for example, showing negligible expression. +++, High; ++, moderate; −/+ absent or low. (B) Multipotent differentiation of airway basal cells in in vitro culture. Airway basal cells were differentiated in ALI culture conditions for 4 weeks. Differentiated epithelia were vertically sectioned and assessed by H&E staining or assessed by immunostaining for basal cells (keratin 5, KRT5⁺), secretory cells (goblet: mucin 5AC MUC5AC⁺, club: secretoglobin 1A1 SCGB1A1⁺), ciliated cells (acetylated tubulin, ACT⁺), and ionocytes (forkhead box I1 FOXI1⁺). In our immunostaining for FOXI1, it will be noted that we observe some cells with the expected nuclear-localized expression, but additional cells with expression were also detected in the cytoplasm.³⁰ Both these patterns have been reported in the Human Protein Atlas. Scale bar: 50 μm. (C) Multipotent differentiation of airway basal cells in vivo. Airway basal cells were seeded in denuded NIH-Foxn1^rnu rat tracheas, transplanted with open-ended tubing into the flank of a Foxn1^nu mouse, and maintained for 4 weeks. Sectioning and staining were performed as described above. Scale bar: 50 μm. ALI, air–liquid interface; H&E, hematoxylin and eosin; scRNA-seq, single-cell RNA sequencing.

There is an emerging consensus that basal cells are capable both of self-renewing cell division and differentiation to other specialized cells present, namely the abundant secretory and ciliated cells,^13–16 as well as rare cell types such as ionocytes, PNECs, and brush cells.^9,10,17 Although several of the seminal observations identifying the basal cells as stem/progenitor cells were made in mice (e.g., via lineage tracing), either under homeostatic conditions or in response to transient injury, in vitro assays (e.g., growth and differentiation as bronchospheres or multipotential differentiation in air–liquid interface [ALI] cultures) suggest that human basal cells play a similar role in the surface airway epithelium. Thus, the proximal airway basal cells are likely to serve as a major class of stem/progenitor cells within the proximal airway and, as such, are presumed to be the preferred target cells for long-term efficacious CFTR editing.

Recent scRNA-seq analyses of the human surface airway epithelium have provided important insights into the diversity of cell types present in vivo¹¹ (also G. Carraro, pers, comm.). For example, based on transcriptomic profiles, basal cells were classified into five distinct clusters: (1) high expression of canonical basal cell markers; (2) proliferating basal cells; (3) possible basal cells transitioning to a secretory phenotype; (4) the highest expression of the AP-1 family members JUN and FOS; and (5) expression of β-catenin (CTNNB1).¹¹ An important question yet to be answered is to what extent does heterogeneity exist within the basal cell population in vivo, with some cells truly functioning as stem cells (long lived, capable of significant self-renewal and multipotent differentiation) and others cells having characteristics associated with progenitors or transit amplifying cells (limited self-renewal yet retaining multipotent differentiation). Also, if there are distinct subpopulations, are these properties (i.e., stem vs. progenitor) cell intrinsic or rather are they conferred by the particular microenvironmental niche? If it is possible to identify unique surface markers that allow for the specific identification of stem cell-like subpopulations, such distinguishing antigens could be used either for isolation and ex vivo expansion or for in vivo targeting.

Interspersed in the submucosa of the human trachea and bronchi are glands that serve as a source of mucus strands that extend into and are released into the luminal space. Cell types present in the submucosal gland acini include serous, mucus, and myoepithelial cells, while gland tubules contain ionocytes, basal, ciliated, and secretory cells. It has recently been shown in mice that as well as maintaining the gland, myoepithelial cells can regenerate surface basal and luminal cells in response to injury and loss of the surface airway epithelium.^18,19 These results suggest that cell types in the airway, other than basal cells, may also function as stem and progenitor cells under certain injury and repair conditions. Adding to the complexity of stem and progenitor cell populations and plasticity, studies in mice suggest that secretory cells in the pseudostratified epithelium can not only undergo limited self-renewal and differentiation into ciliated cells but may also have the potential to replace depleted basal cells.²⁰

As one moves from the large airways to the smaller airways, in particular the distal bronchi and bronchioles, the height of the pseudostratified epithelium gradually decreases. The epithelium is still composed of secretory cells (now predominantly club cells), ciliated cells, and basal cells, but fewer ionocytes are present. This region is also of significant interest in CF since there is clinical evidence that it represents a critical site of early mucus obstruction and disease.²¹ In the mouse, basal cells are eventually absent as one moves even more distally into domains in which club cells and alveolar type II (ATII) cells serve as stem cells; there are still significant gaps in knowledge for human both with respect to the cell types that are present and their role in epithelial maintenance. Although ATII cells express CFTR, its role in these cells does not seem to be necessary for normal pulmonary surfactant or stem and progenitor cell function.²²

Cell-based therapy to restore CFTR function to the airways would ideally use an autologous source of cells to overcome complications of immune rejection and immunosuppression. In principle, there are at least two overall approaches that may be considered for reconstitution of the airway with CFTR-edited airway basal cells. The first approach involves the preparation ex vivo of a clonal or mixed population of corrected autologous airway basal cells, either as primary airway basal cells or derived from corrected induced pluripotent stem cells (iPSCs), with subsequent transplantation back into the airways of the affected CF individual. Autologous airway epithelial stem cell transplantation represents a potential method to deliver durable, functional CFTR expression to the airways of CF patients. Genomic editing of the CFTR locus in stem cells would ensure that the correction is permanently encoded while multilineage differentiation of stem cell-derived cells will ensure long-lasting restoration of normal CFTR function. Being a patient-derived, cell-based approach, this therapy would avoid the concerns of immune rejection and complications of immunosuppressive therapies. The second approach would involve direct in vivo editing of airway cells, with a focus on targeting epithelial stem cells: basal cells of the tracheobronchial airways and an as-yet poorly defined epithelial stem cell of small airways.

Culture and Expansion of Human Airway Basal Stem Cells

It is possible to directly isolate airway epithelial cells of CF and non-CF patients from various anatomical locations, such as the bronchus, trachea, or nose, via bronchoscopic brushings or nasal curettage. In addition, it is possible to obtain epithelial cells in large numbers from explanted lungs. Recently developed methodologies are capable of significantly expanding in vitro the number of tissue-resident airway basal cells present in such primary human samples. Of note, this has now been demonstrated for CF as well as non-CF samples.²³ One can readily expand basal cells obtained from primary airway sources, such as airway brushings, to the estimated therapeutic dose.^23,24 Some of these methodologies, such as the frequently used “conditionally reprogrammed cell” (CRC) method, rely upon the use of fibroblast feeders together with the ROCK inhibitor (Y27632) to culture and expand the airway epithelium basal cells.^17,23–28 Other methods for culture/expansion are feeder free^29,30; for example, a methodology including inhibitors to both TGFβ and BMP4 was recently shown to expand airway basal cells for up to 80 population doublings (PDs).²⁹ Under either feeder-containing or feeder-free conditions, cells can be maintained as a pooled population of cells, including cells of differing proliferative potential and multipotency. Alternatively, they can be grown as clonal populations,^23,26,28,29 through which such differences in stem/progenitor potential can be more fully examined.²³ These studies indicate that the life span of basal cells is limited to ∼50 PDs and that primary cell isolates contain both short- and long-lived subsets. Ongoing studies indicate that short-lived basal cells (those that can be passaged <13 PDs) had expended many of their PDs before culture. In contrast, long-lived basal cells can be passaged >13 PDs and these cells are candidates for use as cell replacement therapy or a target for in situ correction of CFTR mutations.²³

Expanded cells in culture frequently retain expression of markers/proteins characteristic of resident airway basal cells (keratin 5 [KRT5], TP63, integrin α₆, nerve growth factor receptor [NGFR]), and podoplanin. There are various functional assays for cultured basal cells, including clonogenic growth and differentiation in 3D conditions (e.g., as bronchospheres^17,31 or tracheospheres³²), differentiation in ALI conditions (Fig. 2B), and differentiation in vivo as tracheal xenografts (Fig. 2C).³² For example, when placed in ALI culture conditions, these cells yield a pseudostratified epithelium consisting of basal, secretory, and ciliated cells, as well as ionocytes.¹⁷ Frequently observed in culture/expansion protocols is evidence for loss of basal cell functionality with an extended number of PDs. For example, decreased ability to generate ciliated cells, decreased thickness of ALI-derived epithelium, and decreased CFTR functional activity have all been reported.^27,29,30 Significant effort is currently being devoted to either optimize culture methods capable of retaining basal cell function upon extended passage,²⁶ or perhaps to prospectively isolate, based on surface markers, subpopulations of basal cells, which preserve stem cell-like properties upon extended expansion.

Nonetheless, based on current culture methods, it is very likely that a therapeutic dose of basal cells can be generated without loss of functionality as measured ex vivo. However, what ultimately matters is how autologous cells, with an acceptable safety profile after correction and expansion, will function after transplantation into the airways. Criteria for long-term therapeutic benefit would include the following: (1) that they are retained as a long-lived, self-renewing, population of basal cells (i.e., are not exhausted over the lifetime of the recipient); (2) that they maintain the ability to yield differentiated airway progeny of similar number and phenotype as before correction and expansion; (3) that they generate an airway epithelium expressing therapeutically relevant levels of functional CFTR channel activity; and (4) that they do not disrupt normal airway function or acquire tumorigenic potential. It is very possible that while some of this functionality would be intrinsic to the basal cells, some of it may be determined by the state of the niche into which the cells are transplanted. There is evidence that basal cells isolated from different areas of the lung exhibit regiospecificity in their differentiation properties²⁸; how this will guide therapeutic use remains to be determined. Furthermore, how the intrinsic ability of airway basal cells to create their own niche will influence their persistence and function remains an open question.³³

Editing of the CFTR Gene

There has been significant progress over the past decade to develop methods for editing genomic DNA. There are now several classes of sequence-specific nucleases that are available for DNA sequence-specific genome editing, including clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated 9 (Cas9), zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and meganucleases. The introduction, by the sequence-specific nuclease, of a targeted break in the genomic DNA locally recruits components of cellular DNA repair machinery to repair the lesion. Nuclease-mediated genome editing typically exploits either non-homologous end joining (NHEJ) or homology-directed repair (HDR) mechanisms. NHEJ is typically an error-prone process that rejoins the separated ends of cleaved double-strand DNA (dsDNA). This rejoining is frequently associated with small insertions/deletions/duplications (indels) at the original site of cleavage. By introducing cuts in the DNA at two sites flanking a given sequence, it is possible, utilizing NHEJ, to excise that sequence. In addition, NHEJ can be highly proficient in some cell types at capturing dsDNA donors at the site of dsDNA breaks.^34,35 The other mechanism, HDR, following introduction of a sequence-specific dsDNA break, utilizes homology between portions of the donor template and the sequences targeted for substitution/insertion to facilitate precise genome editing. Whereas NHEJ is active in both noncycling and cycling cells, HDR requires cells to be cycling.

Genome editing of the endogenous CFTR locus reflects the intent to achieve expression of the corrected gene at physiological levels that are appropriate to various types of CFTR-expressing cells. In principle, such editing could be achieved via site-specific correction of selected CFTR mutations (Fig. 1C). Such site-specific correction typically requires a distinct set of sequence-specific nuclease and donor for each CFTR mutation; given that ∼2,000 different CFTR variants have been reported, a sequence-specific approach would only seem justified for the most frequently occurring mutations such as F508del. An alternative mutation-agnostic approach would be targeted integration (TI) of a partial or complete CFTR cDNA into the endogenous CFTR locus (Fig. 1D). This approach, in principle, is capable of providing correction for all CFTR mutations downstream of the targeted partial CFTR cDNA integration site.

Editing of the human CFTR locus has been reported in various immortalized cell lines and non-airway primary human cells. Editing frequencies were generally low before the development of sequence-specific nucleases; nonetheless, the F508del mutation was either introduced or corrected in airway-derived cell lines.^36,37 Utilizing ZFNs, Lee et al. corrected the F508del mutation in a homozygous F508del tracheal epithelial cell line.³⁸ An important proof of concept came in 2013 when Schwank et al. reported successful CRISPR/Cas9 correction of F508del mutations in human homozygous F508del intestinal stem cell-derived organoids³⁹; incorporation of a puromycin resistance (puro^R) expression cassette into the donor facilitated selection of corrected organoids. Significantly, correction resulted in restoration of forskolin-induced swelling in the puro^R selected organoids.³⁹ To establish a model for examining CFTR mutations in the hematopoietic lineage, Jennings et al. introduced, via CRISPR/Cas9 editing, homozygous F508del mutations into HL-60 cells.⁴⁰ Valley et al. recently reported development of an efficient editing pipeline, utilizing CRISPR/Cas9 and single-strand DNA (ssDNA) oligo, to generate cell lines homozygous for CFTR mutations of interest (i.e., F508del, G542X, W1282X).⁴¹ Sanz et al. demonstrated elimination of CFTR deep-intronic splicing mutations in a minigene model via NHEJ-induced deletions, utilizing two guide RNAs (gRNAs) spanning the region of sequence to be deleted.⁴² Maule et al. similarly utilized NHEJ-mediated indels effected by AsCas12a but with a single gRNA to correct, in an allele-specific manner, CFTR intronic splicing mutations in CF patients' intestinal organoids, enabling them to demonstrate restoration of forskolin-induced organoid swelling.⁴³ Whereas the above examples reflected CFTR mutation-specific editing, Bednarski et al. demonstrated a more universal approach in CFBE41o⁻ cell lines, utilizing ZFNs to target integration of the CFTR exons 11 to 27 cDNA sequence into exon 11 of the endogenous CFTR.⁴⁴ This would, in principle, provide correction for all mutations in exon 11 and downstream. Although CFTR was normally expressed at a very low level in this cell line, gene expression was activated by 5-aza-cytidine, and thus, CFTR channel activity was demonstrated.⁴⁴

Recently developed methods of editing genomic DNA, such as adenine or cytidine base editors, which effect single base-pair transitions at the target site, or PRIME editors that are able to perform base-pair transitions and transversion as well as insertions and deletions, neither involve a nuclease-mediated DNA break nor require a separate donor.^45–47 Also, the base and PRIME editors still retain activity in noncycling cells, although at different levels depending on the specific editor.^46–48 Geurts et al. recently demonstrated the ability of adenine editors, SpCas9-ABE and xCas9-ABE, to repair CFTR mutations in intestinal organoids; both genetic and functional repairs were demonstrated.⁴⁹ Another editing method, effected by the formation of triplex DNA structures by specially modified peptide-containing oligonucleotides, has been shown capable of correcting the F508del CFTR mutation in an immortalized airway epithelial cell line and in vivo in mouse airways.⁵⁰

CFTR Genome Editing of Human Airway Basal Stem Cells

Initial reports of CFTR editing in airway basal cells utilized NHEJ-mediated indels to introduce or correct mutations. Peters-Hall et al., first having demonstrated the ability to expand non-CF airway basal cells for 60–120 PDs in modified CRC conditions, utilized CRISPR/Cas9 to introduce indels into CFTR exon 11 in the immediate vicinity of F508.²⁶ After plating cells at low density, single-cell-derived clones were expanded and molecularly characterized. Clones either with indel/indel or wt/wt genotypes were placed into ALI culture and demonstrated retention of multipotency.²⁶ The indel/indel and wt/wt clones demonstrated the absence and presence of functional CFTR, respectively. This demonstrated that edited and expanded clones retained the desired biological activity. Maule et al., as described above for intestinal organoids, also demonstrated that NHEJ-mediated indels repaired CFTR intronic splicing mutations in CF patients' primary airway epithelial cells through the lentiviral transduction of AsCas12a.⁴³

In an important demonstration, Vaidyanathan et al. reported high-efficiency sequence-specific correction of the F508del mutation, without selection, in airway basal cells expanded from both the human nasal epithelium (upper airway basal cells [UABCs]) and bronchial epithelium (human bronchial epithelial cells [HBECs]).⁵¹ Cas9/gRNA ribonucleoprotein complexes were delivered via electroporation followed by AAV-6 delivery of the DNA donor template. Editing of homozygous F508del basal cells yielded allele correction efficiencies averaging 28% for UABCs and 41% for HBECs. When placed in ALI cultures, the pool of edited cells from homozygous F508del UABCs and HBECs demonstrated restoration of the glycosylated mature form of CFTR (band C); furthermore, CFTR currents averaged 31% and 51% of non-CF for UABCs and HBECs, respectively—levels that are of potential therapeutic benefit. The multipotential differentiation capability of the edited basal cells was retained as assayed in ALI cultures. They successfully incorporated the edited pool of UABCs into a clinically approved scaffold with the goal of autologous transplantation into the nasal epithelium of CF patients to reduce CF sinus-related symptoms.⁵¹ We note that even when donor DNAs are codelivered to cells together with efficient nucleases, it is not unusual that a significant frequency of alleles exhibit nuclease-induced indels rather than the desired correction.⁵² For example, in Vaidyanathan et al., the authors observed 38% of indels at the target site in edited HBECs.⁵¹ For CFTR mutations occurring within exon sequences (e.g., F508del), such residual indels would need to be examined to ensure that there are no unintended consequences, for example, the potential loss of response to modulators in case one wishes to continue use of modulators after editing.

In a recent publication,³⁰ Suzuki et al. evaluated both a mutation-specific and a mutation-agnostic approach for CFTR correction in HBECs. For mutation-specific correction of F508del, we delivered ZFNs, designed to recognize the F508del mutation, via electroporation to homozygous F508del airway basal cells; donors were either an electroporated ssDNA oligo yielding an average 11% allele correction frequency or an AAV-6 donor yielding average 31% allele correction efficiency. Editing with either donor did not affect the frequency of basal cells or basal cell-derived secretory cells, ciliated cells, or ionocytes in ALI culture. CFTR function was restored to levels ∼15% and ∼40% of non-CF, respectively. With a view to developing a more universal gene editing approach, we targeted the integration of a partial CFTR cDNA (CFTR_9–27), preceded by a splice acceptor, to intron 8. This targeted genome editing strategy is designed to serve as a therapeutic tool capable of treating a majority of CF patients carrying a diverse range of CFTR mutations. This approach, capable of providing correction for all CFTR mutations downstream of the targeted partial-CFTR cDNA integration site, benefits in principle from the endogenous CFTR promoter activity and native chromatin structure, with the objective being physiologically regulated levels of CFTR expression. Intron 8 TI would provide correction for ∼89% of identified CF-causing CFTR alleles, including F508del, common PTCs (e.g., G542X, R785X, W1282X), and splicing variants (e.g., 3,849 + 10 kb C>T).³⁰ We note that by shifting the target site up from exon 11⁴⁴ to intron 8 would, in principle, enable correction for a slightly greater number of mutations, while avoiding the introduction of indels in exon sequences. Without selection, we obtained ∼50–60% of alleles with TI. We demonstrated successful restoration of mature CFTR protein and CFTR channel function for homozygous F508del, F508del/R553X, G542X/R785X to levels ∼33–56% of non-CF. Since our objective was to retain physiological expression of CFTR, importantly our intron 8 TI had minimal impact on the native CFTR locus open chromatin profile.³⁰

Vaidyanathan et al. recently reported CRISPR/Cas9-mediated TI of the CFTR cDNA into exon 1 of the CFTR locus in airway basal cells.⁵³ Importantly, this would provide correction for nearly all CFTR mutations. Due to the packaging size constraints of AAV, the CFTR cDNA was split into two separate AAV vectors with the CFTR cDNA sequences recombining in the target cells. They also incorporated the truncated CD19 (tCD19) coding sequences into one of the AAV vectors. Although the initial TI efficiencies were relatively low (∼5–10%) in the basal cells, possibly due to the requirement for recombination, the surface expression of tCD19 allowed selection of TI cells to a 50–80% purity. After differentiation in ALI cultures, restoration of CFTR function was demonstrated.⁵³

A significant concern with application of genome editing techniques for the therapeutic correction of somatic cells carrying disease-causing mutations is the potential for introduction of harmful mutations resulting from imprecise editing of the desired locus or through off-target base-pair substitutions, indels, or genomic rearrangements. Each editing methodology brings with it the potential for on-target or off-target adverse events. Unless head-to-head comparisons are made in targeting the same gene sequence and utilizing the same detection methodology, it may be difficult to conclusively determine that one editing platform is preferable to another in this regard. However, ever improving technologies that enhance the fidelity of site-specific gene alterations (e.g., high-fidelity variants of Cas9,⁵² ZFNs of increased specificity,³⁰ high-specificity PRIME editing⁴⁷) have the potential to significantly mitigate these safety concerns. Nonetheless, irrespective of the method of editing, it will be important to verify the genomic integrity of such corrected basal cells. Such interrogations of potential mutations, extensive deletions, or rearrangements in the genomic DNA are challenging, particularly when analyzing a mixed population of edited cells, since events of interest (e.g., off-target indels or base-pair substitutions) may be present at very low frequency and thus not easily identified by standard methodologies. In this regard, analysis of clonal single-cell derived populations typically offers greater sensitivity of detection. In addition to the risk of aberrant DNA alterations resulting from the gene editing technique used, sustained culture could also lead to DNA alterations acquired from nonphysiologic stress and genomic instability. While most mutations would be expected to be silent, a small minority could increase the risk of malignant transformation either in vitro or after in vivo transplant. Thus, the safety and genomic integrity of edited basal cells would need to be validated before human use.

CFTR Genome Editing of Human iPSCs

In addition to the possible transplantation of airway-resident basal cells as considered above, there is a strong rationale to also consider future cell therapeutic approaches for CF, in which patient-specific iPSCs are site specifically corrected for the CFTR mutation, and subsequently differentiated into proximal airway basal cells for subsequent transplantation. iPSCs are highly similar to embryonic stem cells (ESCs) and are of major interest to regenerative medicine approaches in view of their capacity to differentiate into almost any cell type. iPSCs can be routinely generated by overexpressing key transcription factors in somatic cells such as skin fibroblasts or peripheral blood mononucleated cells and reprogramming those cells into a pluripotent state. iPSCs have several favorable properties relevant to airway regeneration, including their capacity for indefinite expansion and ability to differentiate into airway epithelial cells, including tissue-resident basal cells, that are autologous to the potential recipient.

Several groups first reported generation of iPSCs from CF patients carrying various CFTR mutations. Somers et al. reported the derivation of several homozygous F508del iPSC lines using a CRE-excisable lentivirus expressing all four Yamanaka factors (OCT4, SOX2, KLF4, c-MYC).⁵⁴ Wong reprogrammed homozygous F508del fibroblasts from CF patients into iPSCs via four retroviral vectors each expressing one of the four Yamanaka factors.⁵⁵ Furthermore, they reported differentiation of both non-CF and CF iPSCs toward airway epithelium; non-CF iPSC-derived epithelium exhibited CFTR function via the iodide efflux assay, whereas the homozygous F508del iPSC-derived epithelium did not unless treated with an analogue of VX809. Mou et al. utilized iPSCs reprogrammed via synthetic modified mRNAs from CF patients with F508del/G551D or homozygous F508del genotypes.⁵⁶ These iPSC lines were utilized in differentiation protocols to derive NKX2.1⁺ lung cells (see DERIVATION OF PROXIMAL AIRWAY BASAL STEM CELLS FROM iPSCS).

The first reports of correction of CFTR mutations in CF iPSCs used TALENs,⁵⁷ ZFNs,⁵⁸ or CRISPR/Cas9.⁵⁹ Sargent et al. delivered TALENs targeting CFTR exon 11 together with small dsDNA fragments to homozygous F508del iPSCs.⁵⁷ After several rounds of subcloning and allele-specific PCR (distinguishing between F508del and wt), they identified iPSC clones exhibiting correction of one F508del allele per cell. Notably, this methodology did not require any drug selection.

Crane et al. utilized ZFNs specific for CFTR exon 11, together with a donor encoding wt exon 11, to selectively correct either I507del or F508del mutations in I507del/F508del iPSCs or to correct F508del in homozygous F508del iPSCs.⁵⁸ A loxP-flanked puroTK selection cassette, also included in the donor, enabled selection of targeted clones and subsequent Cre-mediated excision. Following directed differentiation toward lung, evidenced by upregulation of NKX2.1, SOX9, TP63, and FOXP2, restoration of mature, fully glycosylated CFTR protein was observed for corrected iPSCs. Furthermore, after seeding of corrected differentiated cells onto semipermeable membranes, restored CFTR channel activity was demonstrated by forskolin-induced short circuit current in Ussing chamber assays; loss of current upon treatment with CFTR inhibitor 172 confirmed that it was CFTR-specific.⁵⁸ Restoration of CFTR function was also confirmed via the iodide efflux assay. Subsequently, in McCauley et al.,⁶⁰ the heterozygous corrected wt/F508del iPSCs, described above, were differentiated to proximal airway epithelial organoids in 3D culture. Significantly, the CFTR corrected organoids exhibited restored forskolin-induced swelling similar to that of non-CF iPSC-derived organoids.

Firth et al. reported successful CRISPR-mediated correction of F508del in homozygous F508del iPSCs utilizing drug selection and piggyBac excision to eliminate residual selection cassette or donor sequence.⁵⁹ This approach not only achieved a more precise editing without residual exogenous sequences but exhibited increased editing efficiency. Corrected cells were then differentiated toward lung progenitors and further matured in ALI culture toward proximal airway epithelium.⁶¹ At day 45, EpCAM (CD326)-positive cells were isolated and patch clamped to reveal restoration of inducible CFTR-specific chloride current. Western blots revealed restoration of mature CFTR protein.⁵⁹

Suzuki et al. utilized the TALEN/small dsDNA fragment and enrichment methodology previously described⁵⁷ to obtain corrected homozygous F508del iPSCs.⁶² This methodology had the distinct advantages of not requiring selection and providing for seamless correction. To demonstrate functional correction, they differentiated corrected iPSCs toward airway cells, and mixed them together with CF mutant CFBE41o⁻ airway epithelial cells in ALI culture; with increased time in culture they observed CFTR 172 inhibited short circuit current. They also reported incorporation of an improved airway differentiation protocol, which enabled demonstration of restored CFTR current without the requirement for the CFBE41o⁻ cells. Merkert et al. also reported the combined use of TALENs together with ssDNA to correct the F508del mutation in homozygous F508del iPSCs.⁶³

In Ruan et al.,⁶⁴ CRISPR/Cas9 together with ssDNA were utilized to efficiently introduce specific CFTR mutations (F508del, G542X, G551D) into non-CF iPSCs. They then applied this methodology to correct the F508del mutation in a homozygous F508del iPSC line. When they analyzed single-cell derived clones, they found 18.6% to be wt/F508del and 3.4% to be wt/wt. Furthermore, they demonstrated that homozygous corrected iPSCs, when differentiated into proximal lung organoids, exhibited forskolin-induced swelling, consistent with restoration of CFTR channel activity. A CRISPR/Cas9-based pipeline utilizing ssDNA for introducing CFTR mutations (G542X, G551D, R1162X, W1282X) into iPSCs has also yielded homozygous edited clones (as a % of clones screened) in the range of 1.4–8.7%; homozygous correction of both alleles in an I507del/F508del line, utilizing two gRNAs and a single ssDNA, occurred at 9.2% frequency (H. Valley, J. Mahoney, M. Mense, pers. comm.).

An alternative, safe-harbor TI approach (e.g., Fig. 1B) for expression of exogenous CFTR was demonstrated in non-CF iPSCs.⁶⁵ ZFNs were used to target into the CCR5 locus the integration of an expression cassette containing the CAG promoter driving the complete CFTR cDNA; also included was a tdTomato fluorescent expression cassette enabling identification of TI clones. They demonstrated strong expression of the mature CFTR protein in the iPSCs even before differentiation.⁶⁵

Derivation of Proximal Airway Basal Stem Cells from iPSCS

There has been considerable progress over the past several years in the ability to direct the differentiation of human pluripotent stem cells (hPSCs), both human ESCs and hiPSCs, to lung epithelial cells of increasing maturity and functionality. Studies by various groups have led to the identification of experimental protocols that recapitulate the major developmental milestones that lead to lung specification, including the induction of definitive endoderm, followed by anteriorization to yield anteriorized foregut endoderm, and subsequent ventralization to yield early lung progenitors identified by the expression of the earliest known maker of the developing lung, NKX2.1. Briefly, lineage tracing studies in mice have shown that basal cells are derived from Nkx2.1 expressing lung progenitors in vivo and are first identified in the developing airways by the expression of the transcription factor Tp63 and subsequently express several canonical markers, including Krt5 and Ngfr.⁶⁶ These developments have been discussed in depth elsewhere.⁵ Here we highlight the advances particularly responsible for derivation of proximal airway epithelium, in general, and airway basal cells, in particular. Evidence that NKX2.1⁺ iPSC-derived lung progenitors were capable of giving rise to cell types and structures characteristic of proximal airway epithelium came both from in vivo and in vitro assays. Initial reports suggesting that iPSC-derived lung progenitors were competent to differentiate into TP63⁺ basal cells lacked knowledge of precise signaling pathways and instead relied on a period of in vivo transplantation. For example, Mou et al. reported that subcutaneous engraftment of a mixed population of hiPSC-derived cells, some of which expressed NKX2.1, gave rise to cells coexpressing NKX2.1 and TP63.⁵⁶ Huang et al. identified the emergence of airway cells, including a population of cells coexpressing TP63 and NGFR after transplantation of hPSC-derived NKX2.1⁺ early lung progenitors under the kidney capsule of immunocompromised mice.⁶⁷ These results demonstrated the intrinsic ability of in vitro hPSC-derived NKX2.1⁺ lung progenitors, when placed in the appropriate in vivo conditions, to yield mature NGFR-expressing airway basal cells. In addition, several groups described the ability of iPSC-derived cells to generate airway epithelium, including cells expressing a subset of basal cell markers, when placed in vitro in ALI conditions.^55,61 For example, Wong et al. described the upregulation of mRNA transcripts of secretory, ciliated, and basal cell markers, while Firth et al. demonstrated multiciliated cells (FOXJ1⁺), and cells expressing secretory (MUC5AC⁺ or SCGB1A1⁺) and basal (TP63⁺) markers in the ALI conditions.^55,61 Other groups reported the ability of iPSC-derived lung progenitors when cultured in 3D culture conditions to generate lung organoids, including proximal airway-like structures containing ciliated (FOXJ1⁺, acetylated tubulin [ACT⁺]) cells and cells expressing secretory (SCGB1A1⁺) or basal (NKX2.1⁺/SOX2⁺/TP63⁺) markers.⁶⁸

It is to be noted that some of these aforementioned reports also described the presence of mesenchymal cells together with the airway epithelium.^55,61,68 To isolate epithelial cells away from the mesenchymal cells that can emerge during differentiation protocols, several groups moved toward isolation of lung progenitors based on surface expression of CPM,^69,70 CD47^hi/CD26^{neg 71} phenotype, or the expression of a fluorescent knockin NKX2.1-GFP reporter.⁷¹ Sorting of CPM⁺ lung progenitors allowed the development of proximal airway epithelial spheroids, which, based on the expression of canonical markers, contained multiciliated cells, club cells, PNECs, basal cells, and mucus-producing cells.⁷⁰ As many of the markers that identify airway cells are not specific to the lung, Hawkins et al. generated an NKX2.1-GFP reporter that was used to prospectively isolate lung progenitors and characterize their differentiated progeny. By plating NKX2.1^GFP+ lung progenitors into 3D culture conditions, it was possible to generate epithelial-only organoids that expressed both airway (basal cells, secretory cells, and PNECs) and alveolar (ATII cells) markers.⁷¹

Many of these protocols yielded both distal (e.g., ATII cells) and airway cells, suggesting imprecise knowledge of the signaling pathways required to direct the fate of immature lung progenitors toward desired fates. Importantly, an examination of potential factors involved in controlling the proximal airway versus distal lung fate of hPSC-derived NKX2.1⁺ progenitors revealed the crucial role of the WNT signaling pathway.⁶⁰ It was observed that withdrawal of WNT from day 15 NKX2.1⁺ progenitors resulted in strong upregulation of SOX2 and TP63 expression, evidence for commitment to proximal airway differentiation; on the other hand, maintenance of WNT predisposed the lung progenitors to distal lung cell development.⁶⁰

Hawkins et al. have recently reported success in efficiently deriving a population of airway basal cells that can be purified and expanded from iPSCs (“iBCs”).⁷² Use of a dual fluorescent reporter system (NKX2.1-GFP;TP63-tdTomato) allowed iPSC-derived cells to be tracked over time, as they first emerge from the iPSC-derived foregut endoderm as developmentally immature NKX2.1^GFP+ lung progenitors, which then augment a TP63 program during subsequent proximal airway epithelial patterning. These cells clonally proliferate as NKX2.1^GFP+/TP63^tdTomato+ immature airway progenitors. In response to a medium developed for primary basal cell expansion, these NKX2.1^GFP+/TP63^tdTomato+ cells mature and display a molecular and functional phenotype similar to mature human airway basal cells, including expression of NGFR, as well as the capacity to clonally self-renew or undergo multilineage ciliated and secretory epithelial differentiation. When differentiated on semipermeable membranes in ALI culture conditions, they formed a pseudostratified, well-differentiated airway epithelium exhibiting CFTR-specific anion channel activity corresponding to the mutant (e.g., homozygous F508del) or corrected (wt/F508del) CFTR genotype.⁷² Provided that these putative airway basal cells retain their biological properties upon extended passage and can be shown to be safe, they represent a candidate cell source for autologous transplantation.

While iPSC-derived basal cells hold several advantages over expanded primary basal cells, they come with additional issues that must be addressed. First, undifferentiated iPSCs have the potential to form teratomas and inefficient or incomplete differentiation could allow some immature cells to be intermixed in the predominately lung basal cell population. Some of these concerns may result from incomplete or inappropriate epigenetic regulation of iPSC-derived basal cells compared with their adult lung counterparts. Second, off-target effects during reprogramming, genome editing, or genomic alterations selectively accumulated during sustained culture pose potential oncogenic risks. The malignant risk of these cells after transplantation remains to be determined and further work is needed to quantify and mitigate these risks.

Engraftment of Corrected Airway Basal Stem Cells

If patient-specific corrected basal cells (derived and expanded either from primary tissue or from iPSCs) can be generated in sufficient numbers, there still remain significant technical and clinical issues to be resolved before therapeutic application. We note that the objective in transplanting the corrected basal cells to the airway is to have them take up residence at their normal basolateral location within the airway, to undergo multipotential differentiation to replenish the airway with CFTR-corrected secretory and ciliated cells as well as ionocytes, and retain the capacity for self-renewing division in order that the corrected basal cell population at least be maintained if not expanded. In Fig. 3 we show, in schematic form, a comparison between normal (Fig. 3A) and CF airway (Fig. 3B), as well as the predicted restoration of CFTR function anticipated to result from integration of corrected basal cells into the airway epithelium (Fig. 3C). Later in this section we consider what percentage of corrected basal cells might need to be incorporated into the CF airway epithelium to restore CFTR function to the airway.

Figure 3.

Restoration of CFTR function to CF airways via edited airway basal cells. (A) Normal airway epithelium with functional CFTR expression facilitates mucociliary clearance. (B) CF airway epithelium with loss of CFTR function exhibits a thick, dehydrated mucus layer with impaired mucociliary clearance. (C) Predicted correction of CF airway epithelium. In the hypothetical case shown, a fraction of basal cells have been corrected (either via ex vivo editing followed by transplantation, or alternatively via in vivo editing, blue cells) and have given rise to corrected secretory cells, ciliated cells, and ionocytes (blue color). Restoration of CFTR function to the airway is predicted to result in rehydration of the mucus layer and restored mucociliary clearance. CF, cystic fibrosis.

Given the strong barrier properties of the airway epithelium, reflected, for example, in its tight junctions and ceiling of mucus, it will be difficult in principle to deliver the transplanted basal cells through the epithelial tissue from the luminal surface to the basolateral surface. Thus, it is anticipated that some form of preconditioning or transient injury will be required. This is not a new concept since preconditioning of the bone marrow, for example, via chemotherapeutic agents, before hematopoietic stem cell transplantation has been known for decades to be a requirement for successful engraftment. However, preconditioning of the human airway to create “space” or a niche for the transplanted basal cells to take up residence presents a very significant challenge. In principle, after removal of the CF mucus layer, one would seek to achieve a highly controlled, localized, and transient depletion of luminal cells, to create some space at the basolateral surface into which basal cells could be deposited, as isolated cells, 2D or 3D encapsulated structures, or perhaps as an organoid. Transplantation, engraftment, and persistence of airway basal cells have been demonstrated in mice^24,73 following naphthalene conditioning alone, or naphthalene followed by sublethal irradiation.^74,75 Other agents or methods under investigation include mild detergents, bronchial thermoplasty, and cryospray.^76–78 Recently, less toxic methods (c-kit-specific antibodies) have been shown to be effective for depletion of endogenous hematopoietic stem cells from the marrow⁷⁹; thus, it is possible that analogous methods may be useful in selective depletion of luminal or basal cells. Targeted preconditioning will require intense investigation before human use as injury of the pulmonary tree may be dangerous, particularly in the presence of impaired innate immunity as in CF. If done in cases of advanced CF disease, there would be significant concern about airway-resident pathogens, such as antibiotic-resistant bacteria, gaining access to the systemic circulation. Thus, there is a strong rationale for performing such procedures earlier in the disease process. Advances in extracorporeal membrane oxygenation could serve as a bridge therapy and provide respiratory support during this period; however, the risk will need to be carefully evaluated before deployment.

Another significant challenge for stable integration of introduced corrected basal cells is competition from the endogenous CFTR mutant basal cells. In ex vivo studies, when ALI cultures were transiently injured, the existing epithelium, including basal cells, rapidly mobilized to fill in the vacated space.⁸⁰ If this process were similarly to take place in vivo, it would perhaps make the integration of delivered cells more difficult. One possibility would be to confer a transient selective advantage upon the corrected airway basal cells that would enable them to compete more successfully with their uncorrected neighbors for proliferation and differentiation. It is unlikely that such a selective advantage will result from simple correction of the CFTR mutation, and so, this capability would need to be conferred upon the cells. A possible analogy for selective advantage is seen in the phenomenon of revertant somatic mosaicism in epidermolysis bullosa, a serious inherited blistering skin condition resulting from mutations in various genes such as COL17A1 or LAMB3. In cases where the mutations are corrected in the skin stem cells by spontaneous mutation, patches of healthy skin are able to compete successfully with the uncorrected tissue.⁸¹

Extremes of preconditioning and engraftment might not be necessary to derive therapeutic benefit. Studies of various CFTR mutations suggest that even modest levels of gene correction may yield therapeutic benefit. For example, 4.7% of the normal level of wild-type CFTR mRNA results in a milder form of CF⁸²; furthermore, 8% of wild-type CFTR transcripts are sufficient for normal lung function.⁸³ In mixing experiments of human airway cells in ALI cultures, even 20% of non-CF cells mixed into a background of homozygous F508del cells yielded CFTR chloride current at levels 70% of non-CF cells.⁸⁴ Recent in vitro ALI culture studies show that seeding chimeric cultures with as few at 1% non-CF cells results in measurable restoration of CFTR function and seeding 5–10% CFTR-competent cells created near wild-type correction.¹⁷ These data support the concept that, given the opportunity to engraft, expand, and functionally integrate, introduction of a small fraction of cells expressing CFTR from the endogenous locus would be sufficient to correct the chloride transport defect in CF, providing clinical relief (Fig. 3C).

In Vivo CFTR Genome Editing

There are clearly numerous obstacles that will need to be addressed and overcome to develop an effective in vivo genome editing therapeutic for CF patients. One of the primary challenges will be the efficient delivery of the editing reagents, typically a sequence-specific nuclease and a donor containing the corrective sequences, to the cell types requiring correction for both short/intermediate-term and long-term restoration of functional CFTR activity.

As highlighted previously, the proximal airway basal cells are presumed to be preferred target cells for long-term efficacious CFTR genome editing. However, due to significant uncertainty regarding the rate of cell turnover in the CF lung, it is unknown how quickly airway basal cells, once successfully edited, will restore CFTR function to the lung by differentiation to functional CFTR-expressing cells. If the substantial replacement of mutant CFTR-expressing cells (be they secretory cells, ciliated cells, or ionocytes) by the differentiated progeny of corrected basal cells were to take several months to years, this would perhaps result in an unacceptably long delay, for example, to assess benefit in clinical trials. Consequently, it is possible that an effective gene editing therapeutic for CF patients will have to provide for both short/intermediate-term (involving successful direct editing of secretory, ciliated, and/or ionocyte cells) and long-term (successful editing of basal cells) restoration of CFTR function. In this regard, the requirement for a successful gene/cell therapeutic is somewhat similar to that shown effective for hematopoietic cell gene therapy, in which correction, ex vivo, of both short-term repopulating cells (for rapid production of corrected neutrophils, e.g.) and long-term repopulating cells (for durable reconstitution of all blood lineages) is required.

To pursue a genome editing strategy applicable to the majority of CFTR mutations and CF patients, we described earlier the TI of a CFTR partial cDNA, providing correction of all mutations downstream of the TI site. For efficient nuclease-mediated editing utilizing a donor template for TI (or similarly, donor sequences for sequence-specific correction), there are at least two primary requirements: (1) that there be robust sequence-specific cleavage at the target site; and (2) that there be sufficient delivery of donor DNA template to drive the TI event, rather than the default pathway resulting in NHEJ-induced indels. In applying corrective editing to an intact tissue, it is particularly critical that both the nuclease and donor be delivered to the same cell in amounts sufficient for editing and with the required timing. This latter requirement is more easily satisfied if the nuclease and donor are codelivered within the same vector, but more difficult to satisfy if split between two vectors or delivery vehicles, for example. Otherwise, efficient nuclease delivery without donor will result in indels, possibly rendering the site no longer targetable by the same sequence-specific nuclease. Thus, there likely needs to be a fine-tuning between the dose of nuclease and dose of donor and the relative timing of delivery. In our editing experiments with primary airway basal cells, highly efficient correction or TI required nearly simultaneous delivery of nuclease (via electroporation of mRNA or protein) and donor (via AAV-6 transduction), together with an experimental optimization of the amount of donor relative to nuclease. In in vivo experiments targeting mouse hepatocytes, the donor was delivered only once via AAV-8, followed by several rounds of systemic lipid nanoparticle (LNP)-delivered ZFNs. In this case, the level of editing and of corrected iduronate-2-sulfatase enzyme increased with each iterative administration of LNP-ZFN.⁸⁵ Thus, it is possible that the AAV donor present within nondividing airway cells may still serve as a template for TI upon subsequent delivery of nuclease. We note that the above discussion regarding the challenge of codelivering two vectors to the same cell also may apply for base and PRIME editors due to their larger size. For example, in vivo AAV-mediated delivery of base editors, either cytosine or adenine, required spitting the editor between two AAV vectors, with the two parts being joined via inteins in the recipient cells.⁴⁶

Among viral vectors, there has been continued strong interest in utilizing AAVs for delivery to the airway due to their demonstrated ability to efficiently transduce noncycling cells, their safe use in human trials, and an extensive choice of natural, evolved, and engineered serotypes. There have been several studies of naturally occurring AAV serotypes for their tropism for human airway epithelium, for example, as assayed in ALI cultures. For example, AAV-6, but not AAV-1, has been reported capable of penetrating through the CF mucus.⁸⁶ Furthermore, either through guided evolution or site-specific-directed modification of naturally occurring AAV serotypes, investigators have identified specific AAV variants that exhibit significantly increased ability to transduce the human airway epithelium when exposed apically to ALI cultures.^87–89 As such, they represent very promising vectors for luminal delivery to the airway.

Due to the pseudostratified architecture of the proximal airway, it remains to be demonstrated that a given vector administered luminally, for example, is capable of efficiently delivering editing reagents both to the apically accessible cell types (e.g., ciliated, secretory, and ionocyte cells) and those cell types located more basolateral (e.g., basal cells). This differential apical versus basolateral accessibility is governed not only by the tight junctions present within the airway epithelium but also due to the differential location (apical vs. basolateral) of specific virus receptors on a given cell. Treatment of in vitro models of the human airway epithelium by transiently disrupting tight junctions (ethylene glycol tetraacetic acid, sodium caprate), inhibiting proteosomal processing (doxorubicin, calpain inhibitor I) and/or inhibiting histone deacetylase activity (sodium butyrate), enhances AAV transduction.^90–93 We note the recent development of chimeric AAV/human bocavirus-based vector systems, which in addition to the ability to package a larger genome than standard AAVs (likely beneficial for gene delivery of the complete CFTR cDNA), will perhaps also enable different cell specificity of delivery.^94,95

However, it remains possible that delivery of AAV vectors to basal cells via luminal delivery will not be as efficient as required. Thus, it has been suggested that systemic delivery of editing reagents via the bloodstream may be a more tractable method for accessing the airway epithelium in general, and the basal cell compartment, in particular. Various AAV serotypes have already been evaluated for their ability to target various organs when delivered systemically to mice⁹⁶ and have been utilized to deliver editing reagents to specific organs, such as the liver.³⁴ Nonetheless, there is good reason to consider vectors other than AAV. For example, preexisting antibodies to specific AAV serotypes may preclude use in some patients; furthermore, AAVs may induce an immune response limiting readministration of the same vector. There have recently been significant advances in the development of various nonviral vectors, including novel formulations of nanoparticles, for in vivo targeting of specific organs via systemic administration. For example, it has been shown that it is possible, via systemic delivery, to successfully target delivery of Cas9 to the lung.⁹⁷ Systemic delivery of editing reagents would potentially extend in vivo editing of CFTR mutations to affected organs other than the lung (e.g., pancreas, intestine). Irrespective of whether the reagents are delivered to the lung via luminal or systemic administration, it will be critical to carefully interrogate for potential editing-mediated adverse events. Finally, it may also be possible to correct CFTR mutations already in utero; promising data in this regard have already been shown in mouse models, both in editing of CFTR with triplex-forming oligonucleotides⁹⁸ and for correction of surfactant protein C mutations with CRISPR/Cas9.⁹⁹

Another important issue relevant to genome editing of the CF airways is the cycling status of relevant target cells. For example, although only a low frequency of airway basal cells is generally cycling at any given time in the non-CF human lung, the frequency of cycling airway basal cells in CF lungs was reportedly higher, which the authors attributed as likely being due to ongoing inflammation and repair of damage.¹⁰⁰ Whether this higher percentage of cycling cells in explanted diseased lungs is characteristic of CF airways in vivo at earlier stages, when interventions would be of highest utility, remains to be determined. In contrast to these findings, however, Carraro et al. recently determined via scRNA-seq that the frequency of basal cells in CF airways that are cycling (∼7.8%) was less than in control non-CF airways (∼14.3%).¹¹ The reason for this discrepancy is not clear at this time. There are insufficient data on the mitotic index of secretory or ionocyte cells in CF airways. Since only actively dividing cells are amenable to HDR, limiting oneself to this particular editing mechanism would likely render the majority of airway cells, including basal cells, non-editable at any given time. As mentioned previously, since the NHEJ repair mechanism, as well as base or PRIME editors, is active in noncycling cells, such editing methods may be required.^{34,35,46–48} Typically, NHEJ-mediated TI integrates donor sequences equally in forward or reverse directions. To maximize the desired directionality of integrated sequences via CRISPR/Cas9, gRNA and protospacer adjacent motif recognition sequences may be directionally incorporated in the donor to favor the correct orientation (known as Homology Independent TI³⁵). This strategy is similar to the previously developed directional approach for NHEJ-mediated TI utilizing obligate heterodimer ZFNs.^101,102

Highly regulated, cell-type-specific expression of CFTR perhaps provides a strong rationale for restoring CFTR via editing of the endogenous CFTR locus. For example, correction of a CFTR mutation in ionocytes, secretory cells, and ciliated cells would be expected to result in high, low, and negligible levels of corrected CFTR mRNA expression per cell, respectively. It is not known whether this tightly regulated cell-type-specific expression is, in fact, necessary for physiological restoration of CFTR function to the airway. If required, then it is difficult to imagine a mechanism by which other methods such as site-nonspecific integration or even safe-harbor integration of an expression cassette could achieve the normal pattern of endogenous expression and function.

One interesting possibility would be specifically targeting delivery of editing reagents to basal cells, for example, to maximize the efficiency of editing in these cells. One surface marker known to be expressed on all airway basal cells is NGFR and would represent a reasonable target for delivery to all basal cells. However, the recent scRNA-seq data¹¹ indicate that not all basal cells are equal and suggest that subsets of basal cells may fulfill distinct functions and assume different fates; thus, there may be value in targeting the editing to particular subsets of basal cells defined by specific surface expressed molecules, or minimally screening to score targeting efficiency within the “stem” cell subset. Furthermore, whether basal cells represent the only cellular target to effect long-term correction of CFTR function or whether targeting of other cell types, such as luminal secretory cells in the large airways, may confer similar long-term retention of corrective genome editing, for example, via dedifferentiation remains to be determined.²⁰ Luminal cells in the small airways are important progenitor cells. Given uncertainties regarding contributions made by distinct epithelial stem/progenitor cells in normal maintenance of the human airway and how rapidly CFTR corrected basal cells would give rise to a fully differentiated airway epithelium, a strong argument could be made for a more general delivery of editing reagents to all epithelial cell types present within the airway. EpCAM would potentially be a useful surface marker for such pan-epithelial cell-specific targeting, although the membrane location of these markers within polarized epithelia (i.e., basal, lateral, apical) may be an additional consideration.

Future Directions/Conclusion

Although this review has primarily focused on correction of CF, some of the same methodologies and approaches may find application for correction of other inherited pulmonary diseases. For example, since multiciliated cells in the surface airway epithelium are derived from airway basal cells, it may be that various ciliopathies, such as primary ciliary dyskinesia, may be amenable to the ex vivo and/or in vivo editing strategies outlined here.⁷² In addition, inherited mutations responsible for diseases primarily affecting alveolar cells in the distal lung, for example, surfactant deficiencies, may also be treatable via editing.^99,103 As emphasized in this review, long-term therapeutic benefit of editing for any such inherited pulmonary disease will likely require correction of the mutation in the relevant stem cell population.

There are clear challenges to the successful deployment of genome editing to correct CF in stem cells of the airway, either ex vivo followed by transplant, or directly in vivo. However, provided that such methods could be therapeutically applied safely and efficiently in those patients expressing little-to-no CFTR protein, the methods could subsequently be utilized for those with sufficient but nonfunctional CFTR. If successful, this could potentially result in a one time, long-term functional “cure” for all those affected by CF, irrespective of the particular mutation.

Author Disclosure

No competing financial interests exist.

Funding Information

We acknowledge funding support from the Cystic Fibrosis Foundation (DAVIS17XX0 and DAVIS19XX0 to B.R.D., HAWKIN19XX0 to F.J.H., SUZUKI19XX0 to S.S., RANDELXX017 to S.H.R., REYNOL17XX0 and REYNOLXX019 to S.D.R., and STRIPP15XXO to B.R.S.) and from the National Institutes of Health (R01HL139876 to B.R.D., and U01HL148692 and R01HL139799 to F.J.H.).