High-Efficiency, Selection-free Gene Repair in Airway Stem Cells from Cystic Fibrosis Patients Rescues CFTR Function in Differentiated Epithelia

SUMMARY

Cystic fibrosis (CF) is a monogenic disorder caused by mutations in the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) gene. Mortality in CF patients is mostly due to respiratory sequelae. Challenges with gene delivery have limited attempts to treat CF using in vivo gene therapy, and low correction levels have hindered ex vivo gene therapy efforts. We have used Cas9 and adeno-associated virus 6 to correct the ΔF508 mutation in readily accessible upper-airway basal stem cells (UABCs) obtained from CF patients. On average, we achieved 30%–50% allelic correction in UABCs and bronchial epithelial cells (HBECs) from 10 CF patients and observed 20%–50% CFTR function relative to non-CF controls in differentiated epithelia. Furthermore, we successfully embedded the corrected UABCs on an FDA-approved porcine small intestinal submucosal membrane (pSIS), and they retained differentiation capacity. This study supports further development of genetically corrected autologous airway stem cell transplant as a treatment for CF.

Graphical Abstract

In Brief

Vaidyanathan et al. use Cas9 and AAV to correct the CF causing ΔF508 mutation in >30% of alleles in airway basal stem cells from CF patients and embed corrected cells on a scaffold for engraftment. This method restores physiologic CFTR function and provides an ex vivo strategy to treat cystic fibrosis.

INTRODUCTION

Cystic fibrosis (CF) is an autosomal recessive monogenic disease caused by mutations in the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) gene that encodes a Cl⁻ channel. Although CF is a systemic disease that affects exocrine activity in multiple organ systems, lung disease is the major cause of morbidity and mortality. Over the past decade, small-molecule CFTR protein correctors and potentiators have been developed and represent a significant advancement in CF therapeutics (Van Goor et al., 2009, 2011; Keating et al., 2018; Taylor-Cousar et al., 2017). Although these small molecules have benefitted patients by improving lung function and reducing pulmonary exacerbations, they are expensive, show variable therapeutic responses, and must be administered daily for the life of the patient (Wainwright et al., 2015). Furthermore, these therapies are not applicable in the ~10% of patients with mutations that prevent production of any CFTR protein. Because of these limitations, there is significant interest in developing gene therapy and genome editing strategies to correct CFTR mutations in stem cells and achieve durable restoration of native CFTR function.

The discovery of CF as a monogenic disease caused by CFTR mutations in 1989 prompted several attempts to use gene therapy for treatment (Alton et al., 2015; Knowles et al., 1995; Moss et al., 2007; Wagner et al., 2002). These studies employed various viral and non-viral strategies but failed to show significant benefit (Griesenbach et al., 2015). Thus, there is the need for a further improvement in the efficiency of gene delivery to obtain sustained clinical benefit (Griesenbach and Alton, 2013). Ex vivo expanded autologous gene corrected airway stem cells may offer a potential alternative to in vivo gene therapy because it avoids the thick mucus, immune reactions against Cas9 (Charlesworth et al., 2019; Wagner et al., 2018) and toxic inflammatory environment present in patients that are significant barriers to efficient gene transfer in vivo. Two key factors, among many, that are critical for the development of ex vivo genome edited basal stem cells as an effective, safe, and durable gene therapy approach for CF include (1) an efficient method of correcting the most common disease causing mutation in the CFTR gene in airway basal cells (the stem cells of the airway epithelia), and (2) a strategy to transplant the gene corrected basal cells back into the patient in order to reconstitute the airway epithelia with cells that express wild-type (WT) CFTR protein. While the level of correction needed is not definitively established and is the subject of ongoing research, previous studies suggest that physiologic restoration to 15% of WT levels in an epithelial sheet would have important and significant clinical benefit (Char et al., 2014; Sheppard et al., 1993). Here, we show that we can achieve such levels of gene correction in readily accessible and clinically applicable primary human basal cells from the upper airways of CF patients and that after correction the population of cells can be embedded in an FDA-approved and clinically used membrane as a scaffold while maintaining their differentiation potential. The corrected basal cells may be transplanted back into patients, once transplantation has been optimized in future studies. Chronic upper-airway sinusitis occurs in essentially 100% of CF patients and is a major cause of morbidity and thus remains an important tissue to correct. Moreover, this approach for the upper airways can be the foundation for a similar approach for the lower airways and perhaps even the intestinal tract in the future.

Genome editing using zinc-finger nucleases or CRISPR/Cas9 has been attempted in intestinal stem cells and induced pluripotent stem cells (iPSCs), respectively (Crane et al., 2015; Firth et al., 2015; Schwank et al., 2013). Genome editing using nucleases involves the creation of a double-stranded break (DSB), which is then repaired using either the non-homologous end joining (NHEJ) or the homologous recombination (HR) pathway. NHEJ results in insertions and deletions (INDELs) and therefore cannot be used to precisely correct mutations. The HR pathway can be exploited to correct mutations, but it has been challenging to achieve clinically relevant levels of HR to correct CFTR mutations in primary stem cells. Previous studies focused on the ΔF508 mutation that affects >70% of CFTR patients and employed selectable markers to enrich for cells corrected using the HR pathway. The efficiencies reported in these studies (0.02% [Schwank et al., 2013] before clonal selection and 16% after clonal selection [Crane et al., 2015]) are useful to understand the pathophysiology of different mutations and may enable drug screening but are too low for clinical applications. In addition, subsequent directed differentiation of iPSCs into a clinically relevant airway cell type remains a major challenge, and this approach also carries the additional risk of teratoma formation from residual undifferentiated iPSCs. In contrast, a highly efficient, selection-free, feeder-free, genome editing strategy in endogenous airway stem cells could facilitate the clinical translation of durable autologous cell therapies to treat CF airway disease.

Recent studies have achieved >50% allelic modification in hematopoietic stem cells, T cells, and embryonic stem cells using Cas9 complexed with a single-guide RNA (sgRNA) modified with 2′-O-methyl 3′phosphorothioate (MS) in the 5′ and 3′ terminal nucleotides (MS-sgRNA) and correction templates delivered by adeno-associated virus 6 (AAV6) (Bak et al., 2018; Dever et al., 2016; Hendel et al., 2015; Martin et al., 2018). We modified and adapted this general strategy to airway stem cells.

RESULTS AND DISCUSSION

Isolation and Culture of UABCs

We obtained sinonasal tissue from non-CF and CF patients undergoing functional endoscopic sinus surgery (FESS) and performed pronase digestion and red blood cell lysis. On day zero, 2%–22% (mean ± standard deviation (SD) = 10% ± 8%) of cells were found to express cytokeratin 5 (KRT5⁺), a marker for stem/progenitor cells in upper- and lower-airway epithelia (Figure S1A) (Bravo et al., 2013; Hogan et al., 2014; Wang et al., 2015). We initially cultured UABCs as 3-dimensional organoids without any feeder cells in collagen/laminin basement membrane extract (BME) domes with media containing epidermal growth factor (EGF), the bone morphogenetic protein (BMP) antagonist Noggin, and the transforming growth factor-β (TGF-β) inhibitor A83–01 (EN media). Such UABC organoids submerged under media were KRT5⁺ (Figures S1B–S1E). Under these feeder-free organoid conditions, KRT5⁺ UABCs were enriched after 5 days in culture from 10% ± 8% to 72% ± 18% (Figure S1A). Optimal cell density was empirically determined to be 300–500 cells/μL BME (Figure S1F).

Optimization of Cas9 and sgRNA Delivery into UABCs

Several sgRNAs in the exon carrying the ΔF508 mutation (exon 11) were tested (data not shown), and the most efficient sgRNA (Figure 1A) was used for further experiments. This sgRNA is active even in non-ΔF508 cells since the protospacer sequence is eight base pairs away from the ΔF508 mutation. To electroporate UABCs, we first used the protocols reported for hematopoietic stem and progenitor cells (HSPCs) (Bak et al., 2018; Charlesworth et al., 2018; Dever et al., 2016). Edited UABCs only showed INDELs in 15% ± 7% alleles and did not proliferate (Figure S1G). Strategies to improve editing efficiencies and proliferation were not immediately obvious. Among the multiple strategies tested, resuspension of cells in OPTI-MEM media during electroporation as opposed to the Lonza P2 electroporation buffer improved editing efficiencies, as previously reported in intestinal cells (Fujii et al., 2015). UABCs edited using this protocol showed 31% ± 5% INDELs (Figure S1G) but were still not proliferative. We attempted to improve proliferation by treating cells with the Rho-kinase inhibitor Y-27632 (RI) (Fujii et al., 2015). Treatment of cells with 10 μM RI for 48 h before and after the electroporation improved cell proliferation significantly (Figure S1H). In contrast, addition of RI just at the time of electroporation did not improve proliferation (data not shown). Further experiments used UABCs cultured in EN media supplemented with 10 μM RI.

Figure 1. Editing of Primary Airway Basal Stem Cells.

(A) Schematic describing the Cas9 and AAV-mediated strategy to correct ΔF508. The underlined segment represents the sequence complementary to sgRNA used. The PAM (protospacer adjacent motif) is indicated in italicized gray font for the WT sequence. Silent mutations introduced in the correction template to prevent re-cutting by Cas9 are in lowercase and colored blue.

(B) The region around exon 11 was amplified using IN-OUT PCR to quantify INDELS and HR using TIDER. INDELs were observed in 38% ± 2% (mean ± SD) alleles, and HR events were observed in 43% ± 5% (mean ± SD) alleles. Controls treated with only AAV did not show any INDELs or HR.

(C) On day 4 after editing, the UABCs were stained for KRT5 and P63. A representative FACS plot shows that >95% of edited cells were KRT5⁺P63⁺. In trials using cells from multiple donors, the KRT5⁺P63⁺ population was similar between control and edited UABCs (line corresponds to the mean and error bars correspond to standard deviation).

(D) On day 4 after editing, the UABCs were stained for KRT5 and cytokeratin 14 (KRT14). A representative FACS plot shows that >95% of edited cells were KRT5⁺KRT14⁺. The KRT5⁺KRT14⁺ population was similar between control and edited UABCs from multiple donors.

(E) The UABCs were also stained for Integrin alpha 6 (ITGA6). A representative FACS plot shows that >95% of edited cells were KRT5⁺ITGA6⁺. The KRT5⁺ ITGA6⁺ population was similar between control and edited UABCs from multiple donors.

(F) The UABCs were also stained for nerve growth factor receptor (NGFR). A representative FACS plot shows that 98% of edited cells were KRT5⁺NGFR⁺. The KRT5⁺NGFR⁺ population was similar between control and edited UABCs from multiple donors.

Insertion of Correction Sequence by HR in ΔF508 Locus

We then proceeded to optimize the insertion of exogenous sequences relying on the HR pathway. We attempted gene editing of UABCs using a previously reported control HR template expressing GFP from the CCR5 locus (Hendel et al., 2015) delivered using AAV6. We attempted to optimize HR in UABCs cultured as organoids and as monolayers. Human UABCs were cultured and expanded as KRT5⁺ monolayers using the same EGF/Noggin media (EN media) employed for culturing organoids (Figure S2A). Using cells from the same donors as in Figure S1F, optimal cell density was empirically determined to be ~10,000 cells/cm² for UABCs cultured as monolayer. This method also resulted in a similar fold expansion of KRT5⁺ cells as organoid cultures (Figure S2A). Cell culture at 5% O₂ also increased cell proliferation of KRT5⁺ UABC monolayers compared to 21% O₂ (Figure S2B). Since HR was more efficient in UABCs cultured as monolayers (Figure S2C), we used this approach in subsequent experiments with the EN media optimized from organoids. We also tested the use of a previously reported Cas9 mRNA (Vaidyanathan et al., 2018) but found Cas9 RNP to be more effective for insertion of the GFP HR template (Figure S2C). We also screened several AAV serotypes and AAV6 was found to have the highest transduction among commonly used serotypes (Figure S2D).

HR correction templates were designed with silent mutations to both prevent re-cutting by Cas9 and also to serve as a marker to quantify insertion in non-ΔF508 cells (Figure 1A). We tested the efficiency of two HR correction sequences carrying either 5 or 8 silent mutations (Figures S2E and S2F). In these experiments, the CFTR exon 11 locus was amplified using junction PCR 4 days post-editing, after which insertions and deletions (INDELs) and HR events were quantified using TIDER (Brinkman et al., 2018) (Figure S2G). The insertion of the correction sequence was also verified by cloning the PCR product into TOPO plasmids, which were then transformed into E. coli (Dever et al., 2016). We sequenced 50 E. coli clones and observed the presence of our correction sequence (Figure S2H). Last, we also used a different knockin analysis developed by Synthego termed ICE and obtained similar levels of INDEL and HR events (Figure S2I) (Hsiau et al., 2018). Through these experiments, we found that a correction sequence with 8 silent mutations surrounding the Cas9 DSB was more effective than one with 5 silent mutations (Figures S2E and S2F). Therefore, the correction template that included 8 silent mutations was used in subsequent experiments to edit the ΔF508 locus (Figure 1A). Initial experiments used AAV6 at a multiplicity of infection (MOI) of 10⁵ particles per cell as reported previously for HSPCs (Dever et al., 2016) and resulted in HR rates under 20%. Because further functional titration experiments showed improved HR rates at an MOI of 10⁶ particles per cell, this MOI was used in further experiments (Figure S2J). We also tested the effect of cell passage in culture on gene correction efficiency by editing UABCs at different passages. UABCs from the same donors showed reduced insertion of the HR correction template when edited after passage 3 as opposed to editing at passage 1 (Figure S2K). Hence, further experiments were performed using UABCs edited at passage 1.

Using this optimized template along with optimized culture and transduction conditions, our correction sequence was observed in 41% ± 6% alleles (Figure 1B) and INDELs were observed in 38% ± 3% of the alleles (Figure 1B). This is significantly higher than the HR frequency of 21% ± 7% observed prior to the optimization of cell culture and AAV transduction (Figure S2L). Under the optimal conditions, we observe a ratio of HR:INDEL of >1, which is consistent with prior work showing that high frequency HR-mediated editing is possible when optimal amounts of the correction template (donor DNA) is delivered to dividing cells (Charlesworth et al., 2018; Hendel et al., 2014). On the day of genomic DNA extraction, 95% ± 3% of edited cells were KRT5⁺ and P63⁺, which was not significantly different from control (mock) samples, which had 94% ± 3% KRT5⁺ and P63⁺ cells (Figure 1C). In addition, >80% of corrected cells on average were also positive for cytokeratin 14 (KRT14), integrin-alpha 6 (ITGA6) and nerve growth factor receptor (NGFR) (Figures 1D–1F). Thus, insertion of the correction sequence did not alter the phenotype of the corrected UABCs (Bravo et al., 2013; Hogan et al., 2014; Wang et al., 2015). Starting from the plating of cells from tissue, we achieved 25- ± 10-fold (mean ± SD) total expansion of KRT5⁺ cells after gene editing by passage 3 (Figure S2M) and a total expansion of 34- ± 16-fold by passage 6 (Figure S2M). Thus, the editing performed in passage 1 does not limit the expansion of the edited UABCs. The expanded cells were used to optimize functional assays and embedding of UABCs on the porcine small intestinal submucosal membrane (pSIS) membrane.

Off-Target Activity

Cas9 nucleases can tolerate mismatches between the sgRNA and genomic DNA sequences. This results in unintended off-target DSBs depending on the number, position and distribution of mismatches (Hsu et al., 2013; Lin et al., 2014). Such off-target activity is undesirable as it can result in unanticipated mutations in important genes (e.g., proto-oncogenes). In silico methods predicted over 50 possible off-target sites. The off-target activity of the MS-sgRNA was characterized at the top 47 predicted off-target sites (Table S1). Off-target activity above background levels (0.1%) was observed in only one site. INDELs were observed in 0.17% of alleles in Chr11:111971753–111971775 (OT-41 in Table S1). This region corresponds to an intron of the gene coding for the protein DIXDC1. DIXDC1 is a regulator of Wnt signaling and has been shown to be active in cardiac and neural tissue but not in airway cells (Wang et al., 2006). The intronic target sequence has no known functional significance, and the Cas9-induced DSB did not occur in the putative splice donor or acceptor sequences of the intron.

Correction of the ΔF508 Mutation in KRT5⁺ Stem Cells Expanded from Primary Airway Epithelia of CF Patients

We used the optimized protocol to correct the ΔF508 mutation in both UABCs and HBECs from 6 different homozygous (ΔF/ΔF) patients and UABCs from 4 different compound heterozygous (ΔF/other) patients. In UABCs and HBECs from homozygous patients, we observed allelic correction rates of 28% ± 5% and 41% ± 4% alleles, respectively. We observed 42% ± 15% allelic correction in compound heterozygous UABC samples (Figure 2A). Gene-corrected airway cells cultured in air-liquid interface (ALI) cultures using Transwell inserts retained their ability to form a pseudostratified epithelium comprised of a basal layer of KRT5⁺ cells and a luminal layer of ciliated cells (acetylated α-tubulin⁺) and mucin secreting cells (MUC5AC⁺) cells (Figure 2B). The fraction of edited cells did not change appreciably over the 28–35 days during which cells were cultured in ALI, suggesting equal contributions of corrected and uncorrected basal cells to reconstituting the epithelium (Figure 2C).

Figure 2. Restoration of CFTR Function in Cells Derived from CF Patients.

(A) Summary of percentage of alleles exhibiting HR in CF patient samples (ΔF/ΔF indicates ΔF508 homozygous samples and ΔF/other indicates compound heterozygous samples).

(B) Edited CF cells cultured on ALI differentiate into a sheet with basal cells (KRT5⁺), ciliated cells (acetylated α-tubulin⁺), and mucus-producing cells (MUC5AC⁺), which are all readily detected when ALI cultures are sectioned and stained using immunohistochemistry for expression of bio-markers.

(C) Allelic correction rates in edited cells were assessed at the time of plating on ALI and at the end of the Ussing assay 28–35 days later. Allelic correction rates as measured by TIDER did not change appreciably before and after culture in ALI for Ussing assays, suggesting equal contribution of corrected and uncorrected basal cells to reconstituting the epithelium.

(D) Western blot probing CFTR expression. Non-CF nasal cells (lane 1) showed a clear band corresponding to the higher-molecular-weight mature CFTR (band A). Mature CFTR expression was absent in non-edited ΔF508 homozygous (ΔF/ΔF) cells (lane 2), but a faint, lower-molecular-weight band corresponding to immature CFTR was detected (CFTR band B). ΔF508 homozygous (ΔF/ΔF) cells after correction showed a restored mature CFTR band (lane 3) while also retaining a portion of the immature CFTR band. β-actin was used as a loading control.

(E) Representative traces obtained from epithelial monolayers by Ussing chamber analysis in one control non-CF sample and both uncorrected and corrected ΔF508 homozygous samples from the same donor.

(F) CFTR_inh-172 sensitive short-circuit currents observed in corrected CF samples from 10 different donors as a percentage of non-CF controls (UABCs: ΔF/ΔF: n = 4 donors, ΔF/other: n = 3 donors. HBECs: ΔF/ΔF: n = 3 donors). Studies in patients with milder CFTR mutations suggest that even 15% CFTR function relative to non-CF subjects would be therapeutically beneficial (Char et al., 2014; Sheppard et al., 1993).

(G) CFTR_inh-172-sensitive short-circuit currents observed in non-CF, uncorrected and corrected CF UABC samples from multiple donors as a function of editing (ΔF/ΔF: n = 4 donors, ΔF/other: n = 3 donors and non-CF: n = 3 donors).

(H) CFTR_inh-172-sensitive short-circuit currents observed in non-CF, uncorrected and corrected CF HBECs from multiple donors as a function of editing (n = 3 donors, non-CF: n = 2 donors)

Thus, we achieved correction of the ΔF508 mutation in ~40% of alleles in therapeutically relevant primary airway stem cells obtained from CF patients. This level of correction is a 100-fold improvement over previous studies using CRISPR/Cas9 (Crane et al., 2015; Firth et al., 2015; Schwank et al., 2013) and is within the range necessary for clinically significant restoration of CFTR function. Significantly, our approach achieves this high level of gene correction without the use of any drug-based selection strategy (e.g., puromycin), since puromycin drug selection is not compatible with clinical translation.

Our studies also provide information that may be useful for adapting protocols used for editing HSPCs to edit other cell types (Bak et al., 2018). In our efforts to gene correct airway stem cells, some critical alterations include the treatment of UABCs with RI for 48 h before electroporation, suspension of cells in OPTI-MEM during electroporation, and the delivery of HR templates using AAV at 10-fold higher MOIs. This same protocol is also effective in HBECs. The use of OPTI-MEM and RI was also critical for gene editing other adherent stem cells such as embryonic stem cells and iPSCs as recently reported (Martin et al., 2019). Fujii et al. reported a similar protocol for improving the survival of intestinal stem cells after gene editing, although the editing efficiencies were lower, likely due to the use of plasmid HR templates (Fujii et al., 2015). Thus, it seems likely that a similar protocol using OPTI-MEM and RI combined with optimization of AAV MOI and culture conditions might be necessary to gene edit other adherent stem cells at high efficiencies.

Recovery of CFTR Function in Airway Epithelial Cells Differentiated from Gene-Corrected Airway Basal Stem Cells

We first tested CFTR protein expression in the ALI cultures generated from gene corrected UABCs. Figure 2D shows a representative western blot probing CFTR expression using CFTR Antibody 450 in non-CF, uncorrected and corrected CF samples after differentiation in ALI. Mature CFTR expression was not observed in the uncorrected homozygous sample (lane 2) but was restored in cells corrected using the Cas9 and AAV6 platform (lane 3). Mature CFTR expression in corrected cells was appreciable, although less than that seen in non-CF UABCs (lane 1) as expected. These results demonstrate that gene edited basal cells are able to reconstitute an epithelium showing restored CFTR protein expression in vitro.

To quantify CFTR function, we differentiated UABCs and HBECs (corrected CF, uncorrected CF, and non-CF) in ALI cultures and measured currents in short-circuited monolayers using the Ussing chamber assay. Representative traces from epithelial monolayers derived from one non-CF donor and one CF donor both before and after correction are shown in Figure 2E. Representative traces from epithelial monolayers derived from HBECs of one donor are shown in Figure S3A. Consistent with expectations from western blot results, corrected CF samples showed restored CFTR function, as indicated by increased CFTR_inh-172-sensitive short-circuit current (Figure 2E). The forskolin-stimulated currents were noted to be small for both control and corrected monolayers. Because culture conditions can pre-activate CFTR to various extents as detailed below (Figures S3B–S3H), the magnitude of responses to inhibition by CFTR_inh-172 provides a better indication of CFTR function than does the magnitude of the forskolin response. CFTR_inh-172-sensitive short-circuit currents in corrected samples were lower than corresponding short-circuit currents in non-CF samples. The CFTR_inh-172-sensitive currents in edited UABCs and HBECs from 10 different donors are presented as a percentage of non-CF controls in Figure 2F. Genotype information, percentage of alleles corrected, and restoration of CFTR_inh-172 short-circuit currents relative to average non-CF currents for individual samples from 10 donors are presented in Table 1. The CFTR_inh-172-sensitive currents from corrected UABCs (n = 4 ΔF/ΔF donors; n = 3 ΔF/other; n = 3 non-CF) and HBECs (n = 3 ΔF/ΔF; n = 2 non-CF) are plotted as a function of allelic correction in Figures 2G and 2H, respectively. UABC cultures from compound heterozygotes with higher editing efficiencies showed correspondingly greater restoration of CFTR function. HBEC samples from 3 donors had similar correction rates and showed similar CFTR_inh-172-sensitive currents.

Table 1.

Summary of Percentage of Allelic Correction of (ΔF508) in CF UABC and HBEC Samples and Relative CFTR Function with Respect to Non-CF Controls

Patient	Genotype	Percentage of Editing	Raw Inhibitable CF Current (μA/cm2)	Percentage of Non-CF Inhibitable Current	Cell Type	Sex	Age
Non-CF inhibitable current UABC (mean ± SD) = 42 ± 6 (μA/cm2)
Patient 1	ΔF/Other	22	−2.93	7	UABC	male	22
Patient 2	ΔF/Other	44	−1.55	4	UABC	female	26
Patient 2	ΔF/Other	44	−9.5	23	UABC	female	26
Patient 3	ΔF/Other	66	−18.41	44	UABC	male	44
Patient 3	ΔF/Other	66	−15	36	UABC	male	44
Patient 3	ΔF/Other	54	−8.67	21	UABC	male	44
Patient 3	ΔF/Other	54	−11.27	27	UABC	male	44
Patient 3	ΔF/Other	54	−11.32	27	UABC	male	44
Patient 3	ΔF/Other	54	−8.71	21	UABC	male	44
Patient 4	ΔF/ΔF	30	−17.16	41	UABC	male	19
Patient 4	ΔF/ΔF	30	−16.46	40	UABC	male	19
Patient 4	ΔF/ΔF	30	−18.45	44	UABC	male	19
Patient 5	ΔF/ΔF	33	−4.27	10	UABC	unknown	unknown
Patient 5	ΔF/ΔF	33	−4.29	10	UABC	unknown	unknown
Patient 6	ΔF/ΔF	22	−26.04	63	UABC	male	32
Patient 7	ΔF/ΔF	28	−9.7	23	UABC	unknown	unknown
Patient 7	ΔF/ΔF	28	−7.87	19	UABC	unknown	unknown
Non-cf inhibitable current HBEC (mean ± SD) = 18 ± 3 (μA/cm2)
Patient 8	ΔF/ΔF	33	−7.75	42	HBEC	male	25
Patient 8	ΔF/ΔF	26.7	−6.87	37	HBEC	male	25
Patient 8	ΔF/ΔF	33.1	−7.19	39	HBEC	male	25
Patient 8	ΔF/ΔF	43	−5.51	30	HBEC	male	25
Patient 8	ΔF/ΔF	43	−4.73	26	HBEC	male	25
Patient 8	ΔF/ΔF	43	−8.52	46	HBEC	male	25
Patient 8	ΔF/ΔF	26.7	−9.52	52	HBEC	male	25
Patient 8	ΔF/ΔF	33	−8	43	HBEC	male	25
Patient 9	ΔF/ΔF	40	−9.91	54	HBEC	male	32
Patient 9	ΔF/ΔF	40	−18.36	99	HBEC	male	32
Patient 10	ΔF/ΔF	45	−13.71	61	HBEC	male	44
Patient 10	ΔF/ΔF	45	−11.17	78	HBEC	male	44

Corrected ΔF508 homozygous CF UABC cultures showed CFTR_inh-172-sensitive short-circuit currents of 13 ± 3 μA/cm² (31% ± 5% relative to non-CF), which was significantly higher than currents of 0.78 ± 0.04 μA/cm² seen in uncorrected CF controls (p < 0.0001, t test). Corrected ΔF508 compound heterozygous samples showed CFTR_inh-172-sensitive short-circuit currents of 9 ± 2 μA/cm² (20% ± 5% relative to non-CF), which was significantly higher than uncorrected CF controls (p < 0.0001, t test). Non-CF UABC cultures showed CFTR_inh-172-sensitive short-circuit currents of 42 ± 6 μA/cm².

Corrected ΔF508 homozygous HBECs showed CFTR_inh-172-sensitive Cl⁻ currents of 10 ± 1 μA/cm² (51% ± 3% relative to non-CF) compared to 2 ± 1 μA/cm² seen in uncorrected ΔF508 homozygous HBECs (p < 0.0001, t test) and 18 ± 3 μA/cm² seen in non-CF HBECs.

Thus, epithelial sheets derived from gene corrected UABCs and HBECs showed restored CFTR activity in our experiments. In addition to exhibiting restored CFTR activity, corrected UABCs and HBECs, when cultured in ALI, gave rise to differentiated epithelia containing ciliated and mucus-producing cells. Thus, gene editing did not compromise the ability of the corrected airway stem cells to differentiate into the most commonly observed airway cell types. Different media and culture conditions have been reported for the culture of epithelial sheets in ALI (Gentzsch et al., 2017). To ensure that we measured true CFTR function, we tested two commonly used media conditions (Pneumacult ALI medium and ALI medium reported previously [University of North Carolina {UNC} medium] [Gentzsch et al., 2017]). UABCs differentiated in the UNC medium described by Gentszch et al. showed significantly higher forskolin responses compared to cells cultured in Pneumacult ALI medium (Figures S3C–S3H). Although there were differences in the response to forskolin, the response to CFTR_inh-172 was similar under both conditions in both non-CF and corrected CF cells (Figures S3B–S3H). The data reported in Figure 2 derive from cells cultured in Pneumacult. The differences in the forskolin response is likely due to the presence of factors that increase baseline CFTR stimulation (e.g., cholera toxin), but this cannot be confirmed since the composition of Pneumacult media is proprietary.

We observed restoration of CFTR function in both ΔF508 compound heterozygous as well as homozygous samples. However, since the sgRNA is located a few base pairs away from the ΔF508 mutation site and is active in non-CF cells, half of the replaced alleles did not contain the ΔF508 mutation. Therefore, as expected, a higher level of allelic correction (~2-fold higher) is required in compound heterozygous samples for equivalent restoration of CFTR function. We also tested a previously reported sgRNA specific to ΔF508 (Schwank et al., 2013) but did not observe INDELs above the limit of detection. SgRNAs specific to ΔF508 were also not found for S. Aureus Cas9 and Cas12a. Although engineered Cas9 with altered PAM (protospacer adjacent motif) specificities have been reported (Kleinstiver et al., 2015), readily available RNP formulations of these variants are not available. Since editing using Cas9 mRNA was less efficient (Figure S2C), and, since compound heterozygous ΔF508 samples showed restored CFTR function, we did not pursue these alternatives. Nonetheless, it is likely that in the first clinical trials, patients with homozygous ΔF508 will first be treated because both alleles can be a target for correction thereby converting the cell into a functional heterozygous cell.

Previous studies have attempted to determine the minimal level of correction necessary to restore normal Cl⁻ transport by co-culturing non-CF or corrected CF cells and CF cells in ALI. These reports indicate that 10%–50% normal cells are sufficient to restore non-CF level Cl⁻ transport in ΔF508 homozygotes (Johnson et al., 1992; Shah et al., 2016). Another study used a lentivirus based strategy to overexpress CFTR under the control of an RSV promoter in pig epithelia (Cooney et al., 2016). They reported low genomic integration (<1 copy per 10 cells) but observed high levels of CFTR function. The minimal level of gene correction in the endogenous CFTR locus that can restore CFTR function to non-CF levels has not been previously reported. We detected as much as 2-fold difference in CFTR function in technical replicates from the same CF donor despite the same level of correction in cells (Table 1; Figure 2G). On average, 30%–40% allelic correction restored CFTR function to 50% of non-CF levels in bronchial cells. However, one bronchial sample with 40% allelic correction showed 99% CFTR function relative to non-CF controls. Average CFTR function is approximately 60% higher in corrected HBECs (50% of non-CF levels) than UABCs (30% of non-CF levels). However, since the UABCs and HBECs were not from the same donors, it is unclear whether the differences are due to underlying biology of these related yet distinct airway basal cell populations or due to the distinct genetic backgrounds of the individual donors.

CFTR function has been reported to vary logarithmically in organ outputs measured in vivo (e.g., sweat chloride) and has been shown to be rate-limiting at low levels of CFTR expression (Char et al., 2014). Thus, even augmentation to a low level of CFTR function may provide significant clinical benefit. For example, patients homozygous for the R117H mutation experience infertility and mildly increased sweat chloride but are completely free of any respiratory or pancreatic symptoms (de Nooijer et al., 2011). Of note, R117H and other class IV mutations are associated with significantly lower mortality compared to class II mutations such as ΔF508 (McKone et al., 2003). Patch-clamp and apical conductance measurements on cells expressing exogenous R117H-CFTR showed as little as 15% Cl⁻ conductance relative to cells expressing WT CFTR (Sheppard et al., 1993). By way of contrast, Char et al. estimated <2% CFTR function relative to WT levels in patients with R117H mutations (Char et al., 2014). Thus, the 20%–50% function relative to non-CF CFTR function that we found in our studies would be predicted to provide a meaningful clinical benefit to patients if maintained in vivo.

Gene Edited Airway Basal Cells Can Be Embedded onto an FDA-Approved Porcine SIS Membrane

As an initial step toward transplanting ex vivo gene-corrected autologous cells into patients, we tested their capacity to embed on a pSIS membrane that is already in clinical use and FDA-approved for several indications, including sinonasal repair (Nayak et al., 2018). We optimized the seeding density and culture conditions to obtain a monolayer of corrected UABCs that retained expression of KRT5, P63, and KRT14. We determined that the optimal plating density to achieve 50%–70% primary cell coverage in 4 days was 100,000 cells/cm² (Figure 3A). H&E staining of the pSIS in cross-section showed cells organized as a monolayer (Figure 3B). Cells seeded on the pSIS membrane remained KRT5⁺ (Figure 3C, subject 1, Figures S4A–S4I, subjects 2–4). Manders’ co-localization coefficients were calculated, and the fraction of live (calcein green positive) cells that were simultaneously positive for KRT5 (M1) was determined to be 53% ± 15% (n = 4 from 4 individual donors). The fraction of KRT5⁺ cells corresponded to the KRT5⁺ fraction measured on the day of seeding and thus did not change appreciably despite proliferating on the pSIS membrane (Figure S4J). We further assayed for the expression of P63 and K14 in cells embedded on the pSIS membrane and found that a majority of cells remained positive for P63 and KRT14 (Figures 3D and 3E). Thus, the cells embedded in the pSIS membrane maintained the expression of phenotypic markers of UABCs.

Figure 3. Embedding Gene Edited UABCs in a Scaffold for Engraftment.

(A) Edited UABC cells plated on pSIS membranes at a density of 10⁵ cells/cm² resulted in 50%–70% confluence in 4 days (scale bar, 1000 μm).

(B) H&E staining shows a monolayer of cells on pSIS membranes (scale bar, 50 μm).

(C) Sheets fixed on day 4 after embedding on pSIS membrane were KRT5⁺. Calcein green indicates live cells and KRT5⁺ cells are stained red (scale bar, 100 μm). A few cells are positive for calcein green but not KRT5 (^). Some non-viable cells were still KRT5⁺ (*). Manders’ coefficients were calculated. The fraction of calcein green positive cells also positive for KRT5 was determined to be 78% for the sample presented here (average = 53% ± 15% for n = 4 biological replicates). (D and E) A majority of the UABCs cultured on pSIS membrane were also positive for (D) P63 (scale bar, 100 μm) and (E) KRT14 (scale bar, 100 μm).

(F) UABCs cultured on pSIS for 4 days were reseeded in Transwells and cultured in ALI. Epithelial sheets derived from UABCs cultured on the pSIS membrane showed similar Forskolin and CFTR_inh-172 responses as epithelial sheets derived from UABCs from the same donors (n = 2 donors) cultured on BME.

(G) Average CFTR_inh-172 responses were 19 ± 5 and 12 ± 2 μA/cm² (mean ± SD) for epithelial sheets derived from UABCs cultured on BME and pSIS, respectively. The average responses were not significantly different between the two groups (p = 0.06). Error bars represent SD.

We then tested the differentiation potential of the UABCs embedded in the pSIS membrane. The pSIS membrane has been shown to be degraded within 2 weeks in vivo in other wound healing applications (Carey et al., 2014). Therefore, the ability of UABCs to differentiate on the pSIS is less relevant for their long-term activity than their ability to differentiate after the pSIS membrane has been degraded. To model the differentiation potential of UABCs after degradation of the pSIS membrane, we embedded UABCs on pSIS for 4 days after which we removed them by trypsinization and transferred them to ALI culture. Mock-electroporated cells from the same donors cultured on BME-coated plates served as controls. We did not observe a significant difference in the CFTR_inh-172 response of epithelia derived from UABCs cultured on the pSIS membrane when compared to epithelia derived from control UABCs from the same donor (Figures 3F and 3G). Thus, UABCs cultured on pSIS not only maintain phenotypic markers of basal stem cells but also retain their ability to produce a differentiated epithelium that displays CFTR function independent of the pSIS membrane environment, suggesting that the pSIS membrane is a suitable scaffold for testing the transplantation of corrected UABCs.

Achieving correction of CFTR mutations in vivo, either by transplanting corrected cells or editing in vivo, remains a hurdle for gene therapy. Transplantation of airway stem cells into the lower airways has been reported in animal models but further optimization is necessary for clinical use (Rosen et al., 2015). We focused our experiments on upper-airway basal cells since the ease of access to the upper airways may greatly aid clinical translation by providing readily accessible airway stem cells and by enabling surveillance of the sites using nasal endoscopy after transplantation. Moreover, sinuses of CF patients have been shown to act as a reservoir for drug-resistant pathogens that can lead to chronic lung infections (Hansen et al., 2012). Chronic rhinosinusitis and recurrent infectious disease of the upper airways are poorly addressed medical needs that affect CF patients, since upper-airway manifestations of CF are not ameliorated by lung transplantation. Moreover, AAV-mediated gene therapy in the maxillary sinus was previously unsuccessful (Wagner et al., 2002). Importantly, because the upper airways continue to function as a reservoir for pathogens in immunosuppressed CF patients after lung transplantation, they may confer a significant risk of allograft infection by resistant bacteria (Choi et al., 2018). Thus, upper-airway manifestations of CF present a significant unmet medical need in patients. The surface area of one maxillary sinus has been estimated to be ~13 cm² (de Oliveria et al., 2014). Given that 10⁵ cells/cm² on the pSIS membrane represents an optimal cell density, we estimate that 1.3 million corrected cells will be sufficient to completely cover the surface of one maxillary sinus even without any lateral outgrowth of gene-corrected basal cells following implantation. We have further tested our ability to expand edited UABCs. We can achieve a total expansion of 25- ± 10-fold expansion of edited UABCs within three passages after tissue extraction from the donor (Figure S2M). Thus, a starting yield of 200,000 UABCs will readily provide 5 million edited UABCs for transplantation, well more than the 1.3 million corrected cells needed to treat one maxillary sinus.

Furthermore, our results also show that gene correction does not impair the ability of UABCs to expand for up to 6 passages (Figure S2M). In addition, UABCs still form differentiated epithelial sheets that retain CFTR expression even after they are mock electroporated, expanded, seeded on pSIS, and differentiated after removal from pSIS (Figures 3F and 3G). Thus, our results suggest that the benefit provided by restoring CFTR expression using our approach would be durable if achieved in vivo. Transplantation of embedded cells with in vivo analysis will be the focus of future studies.

Conclusions

Our study shows that a platform consisting of Cas9, MS-sgRNA, and AAV6 can be used to successfully correct the ΔF508 mutation in primary airway stem/progenitor cells obtained from CF patients. This selection-free strategy achieves clinically significant restoration of CFTR function in differentiated epithelial sheets derived from corrected stem cells. Our findings pave a pathway for future experiments to optimize the transplantation of corrected UABCs into the upper airway to treat CF sinus disease. Successful optimization of stem cell transplantation strategies into the sinuses may then enable further investigations into the use of cell therapies to treat CF lung disease.

STAR★METHODS

LEAD CONTACT AND MATERIALS AVAILABILITY

Further information and requests for resource/reagents should be directed to and will be fulfilled by the Lead Contact, Matthew Porteus (mporteus@stanford.edu)

The HR correction templates generated in this study are available for distribution upon request without restrictions.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Subject details

Upper airway tissue was obtained from adult patients undergoing endoscopic surgery after obtaining consent. The protocol was approved by the Institutional Review Board at Stanford University. The CFTR mutations observed in the subjects, age and gender have been recorded. HBECs were obtained from Lonza Inc. or the Primary Airway Cell Biobank at McGill University.

METHOD DETAILS

EN media

ADMEM/F12 supplemented with B27 supplement, Nicotinamide (10 mM), human EGF (50 ng/mL), human Noggin (100 ng/mL), A83–01 (500 nM), N-acetylcysteine (1 mM) and HEPES (1 mM)

Cell Culture of UABCs

Tissue pieces were cut into small pieces (1–2 mm²). Tissue pieces were washed with 10 mL sterile, PBS w/2X antibiotic/antimycotic (Penicillin, Streptomycin, Amphotericin B, GIBCO # 15240062) on ice and digested with pronase (1.5 mg/mL, Sigma #P5147) for 2 h at 37°C or at 4°C overnight. Digestion was stopped using 10% FBS. Digested tissue was filtered through cell strainers (BD Falcon # 352350) into a sterile 50 mL conical tube. The mixture was centrifuged at 600xg for 3 minutes at room temperature. RBC lysis was then performed using RBC lysis buffer (eBioscience) as per manufacturer’s instructions. After RBC lysis, cells were suspended in 1 mL EN media and counted. A small sample was fixed using 2% paraformaldehyde and permeabilized using Tris-buffered saline with 0.1% Tween 20. Cells were stained for cytokeratin 5 (KRT5, Abcam, ab 193895). An isotype control (Abcam, ab 199093) was used to control for non-specific staining. KRT5+ cells were plated at a density of 10,000 cells per cm² as monolayers in tissue culture plates coated with 5% BME. Cells were incubated at 37°C in 5% O₂ and 5% CO₂ in EN media with 10 μM ROCK inhibitor (Y-27632, Santa Cruz, sc-281642A). Organoids were grown as 40 μL BME domes submerged under identical medium at initial seeding density of KRT5+ 20,000 cells/dome. Cells obtained from CF patients were grown in EN media supplemented with additional antimicrobials for two days (Fluconazole – 2 μg/mL, Amphotericin B 1.25 μg/mL, Imipenem – 12.5 μg/mL, Ciprofloxacin – 40 μg/mL, Azithromycin – 50 μg/mL, Meropenem - 50 μg/mL). The concentration of antimicrobials was decreased 50% after 2–3 days and then withdrawn after editing (day 5–6). Cells were also stained for cytokeratin 14 (KRT14, Abcam, ab181595), p63 (Biolegend, 687202), Integrin Alpha 6 (ITGA6, Biolegend, 313608) and Nerve Growth Factor Receptor (NGFR, Biolegend, 345106).

Cell Culture of HBECs

Human bronchial epithelial cells were cultured in Pneumacult Ex-Plus at 3000–10000 cells/mL in tissue culture flasks without any coating.

Gene Editing of UABCs

Cells were cultured in EN media with 10 μM ROCK inhibitor (Y-27632, Santa Cruz, sc-281642A). The presence of ROCK inhibitor (RI) for at least 24 h was critical for cell survival after electroporation. Media was replaced on day 3 and day 4 after plating from tissue. Gene correction was performed 5 days after plating. Cells were detached using TrypLE Express Enzyme (GIBCO^™ 12605010). Cells were resuspended in OPTI-MEM (GIBCO^™ 31985062) at a density of 5 million cells/mL. Other electroporation buffers were also tested but the best results were obtained with OPTI-MEM. A similar observation has been reported in intestinal organoids (Fujii et al., 2015). Electroporation (Nucleofection) was performed using Lonza 4D 16-well Nucleocuvette Strips (Lonza, V4XP-3032). 6 mg of Cas9 (Integrated DNA Technologies, IA, Cat: 1074182) and 3.2 μg of MS-sgRNA (Trilink Biotechnologies, CA) (molar ratio =1:2.5) were complexed at room temperature for 10 minutes. 100,000 cells (20 μL of OPTI-MEM with 5 million cells/mL) was added to Cas9 and MS-sgRNA mixture used per well and transferred to the strip. Cells were electroporated using the program CA-137. 80 μL of OPTI-MEM was added to each well after electroporation. Cells were transferred to a 12 well plate coated with 5% BME (density = 20,000 cells/cm²) and 400 μL of EN media with 10 μM ROCK inhibitor was added. AAV carrying the correction template was added immediately after electroporation to maximize transduction (Bak et al., 2018; Charlesworth et al., 2018). Multiplicity of Infection (MOI) of 10⁶ particles per cell (as determined by qPCR) was optimal. AAV titers can also be determined by droplet digital PCR (ddPCR) but the output results in titers are 10-fold lower for the same sample when compared to qPCR. If AAV titer is measured using ddPCR, the appropriate MOI would be 10⁵ particles per cell. Media was replaced 48 h after electroporation. Gene correction levels were measured at least 4 days after electroporation.

Gene Editing of HBECs

The same protocol used for UABCs was used for HBECs except for the fact that the cells were cultured in Pneumacult Ex-Plus supplemented with 10 μM ROCK inhibitor. We did not test if HBECs could be edited without RI treatment.

Measuring Gene Correction

4 days (or more) after gene correction, genomic DNA was extracted from cells using Quick Extract (Lucigen, QE09050). The cells were suspended by trypsinization and washed once with OPTI-MEM or media. 50–100,000 cells were centrifuged and the supernatant was discarded. 50 μL quick extract (QE) was added to the cell pellet and vortexed. The quick extract suspension was heated at 65°C for 6 minutes, vortexed and then heated at 98°C for at least 10 minutes to inactivate QE. We observed that inactivation for less than 10 minutes often resulted in failed PCR. The ΔF508 locus was amplified using the primers:

• Forward: CCTTCTACTCAGTTTTAGTC

• Reverse: TGGGTAGTGTGAAGGGTTCAT

The PCR product was sanger sequenced (primer: AGGCAAGTGAATCCTGAGCG) and the percent of corrected alleles was determined using TIDER.

Measuring off-target activity

Primary UABCs from non-CF patients were electroporated with Cas9 and sgRNA without the HR template. gDNA was extracted 4–5 days after RNP delivery. Potential off-target sites were identified using the bioinformatic tool COSMID (Cradick et al., 2014) allowing for 3 mismatches within the 19 PAM proximal bases. Predicted off-target loci were initially enriched by locus specific PCR followed by a second round of PCR to introduce adaptor and index sequences for the Illumina MiSeq platform. All amplicons were normalized, pooled and quantified using a Qubit (ThermoFisher Scientific) and were sequenced using a MiSeq Illumina using 2 × 250bp paired end reads. INDELs at potential off-target sites were quantified as previously described (Lee et al., 2016).

Briefly, paired-end reads from MiSeq were filtered by an average Phred quality (Qscore) greater than 20. Single reads were quality trimmed using Cutadapt and Trim Galore prior to merging of paired end reads using Fast Length Adjustment of SHort reads (FLASH).

Alignments to reference sequences were performed using Burrows-Wheeler Aligner for each barcode. Percentage of reads containing insertions or deletions with a ± 5-bp window of the predicted cut sites were quantified.

Air-Liquid Interface Culture of Corrected UABCs and HBECs

Gene corrected cells were plated 4–10 days after editing. 30,000 to 60,000 cells per well were plated in 6.5 mm Transwell plates with 0.4 μm pore polyester membrane insert (Corning Inc., 07-200-154). EN media was used to expand cells for 1–2 weeks. Once cells were confluent in Transwell inserts and stopped translocating fluid, media in the bottom compartment was replaced with Pneumacult ALI media. For comparison, a small batch of cells were also cultured in media obtained from the University of North Carolina (UNC media)(Gentzsch et al., 2017). The need for an additional coating of plates with collagen IV was also tested.

Immunoblot

Immunoblotting methods were used to compare CFTR protein expression pre/post-correction. Non-CF nasal cells, and patient-derived ΔF508 homozygous pre/post-correction were plated and cultured according to the above methods. Lysis was performed by incubating cells for 15 minutes in ice-cold RIPA buffer supplemented with EDTA-free protease inhibitor. Lysates were gathered and rotated at 4°C for 30 min, then spun at 10,000 × g at 4°C for 10 minutes to pellet insoluble genomic material. The supernatant was collected and mixed with 2x Laemmli sample buffer containing 100 mM DTT and subsequently heated at 37°C for 30 min. Approximately 5 μg of total protein was loaded for Calu-3 lane, and approximately 14 μg of total protein was loaded for non-CF nasal and ΔF508 cell lines and fractionated by SDS-PAGE, then transferred onto PVDF membrane. Blocking was performed with 5% nonfat milk in TBST (10 mM Tris, pH 8.0, 150 mM NaCl, 0.5% Tween 20) for 60 min. The membrane was probed with antibodies against CFTR (Ab450, 1:1000), and β-actin (1:10,000). Membranes were washed and incubated with a 1:10,000 dilution of horseradish peroxidase-conjugated anti-mouse for 1 h, then developed by SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific, 34095).

Ussing Chamber Functional Assays

Ussing chamber measurements were performed 3–5 weeks after cells had stopped translocating fluid as described before. For Cl⁻ secretion experiments with UABCs, and HBECs, solutions were as following in mM: Mucosal: NaGluconate 120, NaHCO₃ 25, KH₂PO₄ 3.3, K₂HPO₄ 0.8, Ca(Gluconate)² 4, Mg(Gluconate)² 1.2, Mannitol 10; Serosal: NaCl 120, NaHCO₃, 25, KH₂PO₄ 3.3, K₂HPO₄ 0.8, CaCl₂ 1.2, MgCl₂ 1.2, Glucose 10. The concentration of ion channel activators and inhibitors were as follows:

• Amiloride - 10 μM – Mucosal

• Forskolin - 10 μM – Bilateral

• VX-770 - 10 μM – Mucosal

• CFTR_inh-172 - 20 μM – Mucosal

• UTP – 100 μM – Mucosal

Embedding cells on pSIS membrane

pSIS membranes (Biodesign^® Sinonasal Repair Graft; COOK Medical, Bloomington, IN) were placed in 8 well confocal chambers. UABCs were seeded 4–8 days after electroporation. Four days after seeding, cells were incubated with calcein green for 15 min and the excess washed away. Cells were imaged using a dissection scope to identify densities that provided optimal coverage. pSIS membranes with cells were fixed with 4% paraformaldehyde, permeabilized with TBS-T (0.1% Tween 20) and stained for KRT5, KRT14, p63 and imaged using Leica SP8 confocal microscope.

KEY RESOURCES TABLE

REAGENT OR RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-Cytokeratin 5 antibody [EP1601Y] (Alexa Fluor® 647)	Abcam	Cat#: ab193895; RRID: AB_2728796
Rabbit IgG, monoclonal [EPR25A] - Isotype Control (Alexa Fluor® 647)	Abcam	Cat#: ab199093; RRID: AB_2818935
Alexa Fluor® 488 anti-human/mouse CD49f Antibody	Biolegend	Cat#: 313608 ;RRID: AB_493635
Anti-Cytokeratin 14 antibody	Abcam	Cat#: ab181595; RRID: AB_2811031
Anti-p63 Antibody	Biolegend	Cat#: 687202 ;RRID: AB_2616941
Anti-NGFR antibody	Biolegend	Cat#: 345106; RRID: AB_2152647
Biological Samples
Upper airway basal cells (UABCs)	Stanford University	N/A
Upper airway basal cells (UABCs)	University of Arizona	N/A
Bronchial epithelial cells (HBECs)	Lonza, McGill University	N/A
Reagents
Transwell inserts	Corning Inc.	07-200-154
Adeno-associated Virus Serotype 6 with ΔF508 correction template	Vigene Inc	Custom order
Adeno-associated Virus Serotypes 1,2,5,6,7,8,9, DJ with UBC-GFP	Vigene Inc	Custom Order
Alt-R® S.p. Cas9 Nuclease 3NLS, 500 μg	Integrated DNA technologies	1074182
MS-sgRNA	Trilink Biotechnologies	O-5200
Y-27632 dihydrochloride	Santacruz Biotechnology	sc-281642A
HEPES	GIBCO	15630-080
GlutaMAX-1 (100x)	GIBCO	35050-061
NIC (Nicotinamide)	Sigma	N0636
B-27 supplement (50x)	GIBCO	125870-01
A83-01	Torcis	2939
Human EGF	R&D	236-EG-01M
Human Noggin	R&D	6057-NG/CF
admem F-12	Invitrogen	12634-028
Biodesign™ Sinonasal Graft	Cook Medical	G35950
CellTrace Calcein Green, AM	Invitrogen	C34852
Anti-CFTR antibody 450	University of North Carolina	N/A
Pneumacult ALI media	Stem Cell Technologies	#05001
Pneumacult Ex-Plus	Stem Cell Technologies	05040
Oligonucleotides
TIDER PCR Primer Forward	Integrated DNA technologies	CCTTCTACTCAGTTTTAGTC
TIDER PCR Primer Reverse	Integrated DNA technologies	TGGGTAGTGTGAAGGGTTCAT
TIDER Sequencing primer	Integrated DNA technologies	AGGCAAGTGAATCCTGAGCG
Software and Algorithms
TIDER	Deskgen	https://tider.deskgen.com/
COSMID Off-Target Analysis Tool	N/A	https://crispr.bme.gatech.edu

Highlights.

• Cas9 RNP and AAV can be used to efficiently gene edit human airway basal stem cells

• This method yields >30% allelic correction without selection markers or antibiotics

• Correction of >30% ΔF508 alleles restores CFTR function to near non-CF levels

• Corrected stem cells can differentiate after embedding in the scaffold for engraftment

DATA AND CODE AVAILABILITY

The data and code are not available.