Severity and disease progression in people with Cystic Fibrosis (CF) is typically dependent on their genotype. One personalised therapeutic option is the use of exon skipping antisense oligonucleotides (AO) that are based on a specific genotype (Radpour et al., 2007; Shteinberg et al., 2017; Gaikwad et al., 2018). A review of the literature led to two observations that have suggested partial redundancy of CFTR exon 9: 1) splice mutations 1209 + 1 G > C and 1209 + 2 T > G were documented to cause CFTR exon 9 skipping and these variants were reported to manifest as a milder form of disease caused by mutations in the CFTR gene; congenital bilateral absence of vas deferens (CBAVD) (Li et al., 2012) and 2) CFTR exon 9 is in-frame and is reported to carry relatively few CF-causing mutations, when compared to other CFTR exons (Toronto 2011; Henrie et al., 2018; Molinski et al., 2018). CFTR exon 9 is an in-frame exon therefore the exclusion of this exon would not alter the reading frame of the remaining mRNA. This strategy could be beneficial for people with class I “null mutations” in exon 9 such as p.Trp401X that prevent synthesis of a functional protein. While the impact of exon 9 skipping on gene expression and cellular pathways can be studied in cells in vitro, trace amounts of full-length normal or mutated material could confound the testing. To overcome this limitation, the impact of CFTR exon 9 skipping on disease phenotype and severity should be more effectively evaluated in a small animal model.

Typically, mice homozygous for a Cftr mutation or disrupted Cftr gene display many features common to a young person with CF caused by two severe mutations (ie p.Phe508del), including failure to thrive, meconium ileus, alteration of mucus and serous glands, and obstruction of gland-like structures with thickened eosinophilic material (Lavelle et al., 2016). Genetically modified mouse models of CF disease were initially created by homologous recombination in embryonic stem cells through targeting a mutation to a specific site in the murine genome, each with varied outcomes (Clarke et al., 1994; Gosselin et al., 1998) (Davidson and Rolfe 2001). (Davidson and Rolfe 2001). Typically, CF-associated death in mouse models with specific Cftr mutations was from intestinal obstruction, usually before 40 days of age (Snouwaert et al., 1992). Some CF mouse models used gene insertions without loss of genomic sequence therefore, potentially allowing a small amount of normal Cftr mRNA to be expressed that may contribute to a reduced severity of disease (Davidson and Rolfe 2001; Scholte et al., 2004). There are ‘two residual function’ CF mouse models, Cftr ^tm/Hgu and Cftr ^tm/Bay that show 90 and 40% survival respectively, but also have 10 and 2% of wild type (WT) Cftr mRNA present that would contribute to their higher survival rates (O'Neal et al., 1993; Wilke et al., 2011). Furthermore, three mouse models with the p.Phe508del mutation exist, showing variable WT CFTR mRNA levels and survival rates (5–90% survival at weaning) (Colledge et al., 1995; Zeiher et al., 1995; Rozmahel et al., 1996; Davidson and Rolfe, 2001).

Some transgenic CF mouse models used a gene replacement strategy to disrupt the Cftr gene creating absolute nulls. These complete Cftr KO models have no WT mRNA detected, such as Cftr ^tm/Unc; Cftr ^tm/Cam; Cftr ^tm/Hsc; Cftr ^tm3Bay; Cftr ^tm3Uth and have also shown variable survival rates after weaning (5–40%) (Colledge et al., 1995). Cftr mouse models generated with various mutations that result in nonsense mediated decay led to an absence of the functional Cftr protein (Snouwaert et al., 1992; O'Neal et al., 1993; Ratcliff et al., 1993; Hasty et al., 1995; Rozmahel et al., 1996; Wilke et al., 2011). One of these CF KO mouse models was generated with a premature termination codon in Cftr exon 2 (Cftr ^m3/Bay) and resulted in intestinal obstruction, which led to death in 60% of the mutant animals within 1 month after birth (Hasty et al., 1995).

The β-ENaC mouse model was the first to recapitulate the pathophysiology of the CF lung. This model over-expresses the β subunit of the ENaC with increased sodium ion absorption. Mice present with reduced airway surface liquid, reduced mucociliary clearance with airway mucus obstruction, goblet cell hyperplasia, chronic lung inflammation, and high mortality related to the lung disease (Zhou et al., 2011). Although these differences reinforce the notion that mouse models do not accurately recapitulate the human disease phenotype or progression, they can provide important mechanistic data.

We hypothesised that CFTR exon 9 could be partially redundant, meaning the encoded amino acids might not be essential for full CFTR function, therefore enabling an exon skipping AO strategy to be therapeutic for people with amenable mutations in that exon. Skipping of CFTR exon 9 would not disrupt the reading frame and if the resultant isoform retained some function, that could result in a “milder mouse CF phenotype” when compared to complete KO models. To investigate this possibility, a Cftr exon 9 deleted mouse model (Cftr ^Δ9/Δ9 ) was generated using homologous recombination that has no residual WT mRNA. Survival rates were compared to other CF mouse models and lung and intestinal organ histology examined for any pathologies. Primary airway epithelial cells (pAECs) were also harvested from Cftr ^Δ9/Δ9 mice and cultured at the air liquid interface (ALI) for CFTR functional assessment using Ussing Chamber analysis.

Here, the literature was interrogated for an exon that harboured multiple CFTR mutations that presented as a milder form of CF disease (ie CBAVD) indicating that the exon could be partially redundant, with CFTR exon 9 being identified. To investigate the in vitro and in vivo functionality of a CFTR isoform missing the domain encoded by exon 9, a Cftr ^Δ9/Δ9 mouse model was generated. Although Cftr mouse models do not develop CF lung disease, this model was used to ascertain if the level of disease severity correlated with Cftr genotype. It was hypothesised that mice lacking Cftr exon 9 alone would present with less severe pathological disease, when compared to other models, including Cftr gene KO or mutations resulting in premature termination of translation and nonsense mediated decay (Ratcliff et al., 1993; Colledge et al., 1995; Hasty et al., 1995). The Cftr ^Δ9/Δ9 mouse model generated here by homologous recombination presented with intestinal obstructions that were not prevented or improved by a liquid diet. At time of weaning (21 days) Cftr ^Δ9/Δ9 mice had a survival rate of 83% that dropped to 38% by day 50, with most mice eventually reaching a humane endpoint as described in the approved ethics protocol due to intestinal obstruction or malnutrition. Finally, airway epithelial cell cultures established from Cftr ^Δ9/Δ9 mouse airways were not responsive to the CFTR stimulator drug forskolin. Collectively, data generated in this study suggest that Cftr exon 9 is not partially redundant in mice, although airway cells from Cftr ^Δ9/+ littermates failed to culture due to contamination, poor adherence to the culture surface therefore this vital control group is lacking.

Survival of the Cftr ^Δ9/Δ9 mice generated here was initially high at 83% in the first 21 days, compared to the 21 days survival of published Cftr KO mouse models with no residual WT mRNA, such as the Cftr ^tm1Unc null model with ∼25% survival (Snouwaert et al., 1992), and the Cftr ^tm1Cam null with only 17% (Ratcliff et al., 1993). Cftr mouse models with missense mutations and residual WT Cftr mRNA, such as Cftr ^tm/Bay has 40% and Cftr ^tm/Hgu, has 90% survival at 21 days. This could suggest that the disease severity of the Cftr ^Δ9/Δ9 mouse model generated here was marginally less severe with respect to intestinal complications. Homologous recombination mouse models as developed here, overcome any confounding issues by having no residual WT Cftr mRNA present that could potentially mitigate the disease presentation. As the Cftr exon 9 deletion occurred in the genome and is the only source of the Cftr gene expressed in this homozgygous mouse model, there will be no residual full-length Cftr that could influence the survival rate. In addition to specific genetic differences, variations in husbandry and housing environment such as the use of Peptamen® and ALPHA-dri® bedding as used here may also influence survival rates between mouse models.

There are many phenotypic differences between different mouse models of CF, due to background strain, environment and the specific Cftr mutation. Like many other animal models (e.g., the mdx mouse and Duchenne muscular dystrophy), CF mouse models do not accurately reflect the lung disease that occurs in humans, and our model was no different. Histological examination of organs from Cftr ^Δ9/Δ9 mice and heterozygous control littermates revealed few differences. There were no notable changes in the lungs, kidneys or liver evident on Hematoxylin and Eosin stained sections, which is comparable with many Cftr mouse models (Scholte et al., 2004). The organs, including the small intestines were also investigated for structural differences using Hematoxylin and Eosin staining of sections and examining mucus accumulation via Alcian blue staining. Examination of these sections from Cftr ^Δ9/Δ9 mice appeared to have an increase in the number of goblet cells with a swollen lumen and dilated Crypts of Lieberkühn, when compared to heterozygous littermates although quantification could not be done. Alcian blue staining showed dilated goblet cells and Crypts of Lieberkühn that were mucus filled. These findings reflect those from other CF mouse models presenting with intestinal complications and no lung disease (Wilke et al., 2011).

Airway epithelial cells were successfully isolated from homozygous Cftr ^Δ9/Δ9 mice and differentiated ALI cultures were established. The ALI cultures developed beating cilia, produced mucous and differentiated, as supported by high TEER values. Ussing chamber studies on ALI cultures revealed that these Cftr ^Δ9/Δ9 ALI cultures had a good response to amiloride, indicating sodium channels were present and blocked by the addition of this drug. The Cftr ^Δ9/Δ9 ALI cultures did not respond to forskolin stimulation, indicating that no CFTR channels were opening in response to the drug. There was minimal response to the CFTR inhibitor-172 as expected, and the small change seen could be due to volume-sensitive outwardly rectifying chloride ion conductance (Melis et al., 2014). These results indicate that the CFTR protein produced from the Cftr mRNA lacking exon 9 did not have the functional capacity to respond to forskolin, and therefore was a non-functional CFTR channel. When investigating this finding it was noted that electrophysical responses differ depending on the mouse strain, since it is known that other factors such as genetic background can influence this response (Davidson and Rolfe 2001; Wilke et al., 2011). For example, Cftr ^tm1Cam C57Bl/6 mice have a forskolin response that is comparable to WT mice, whereas Cftr ^tm1Cam mice on an FVB background have no forskolin response (Wilke et al., 2011). The difference between our Cftr ^Δ9/Δ9 C57Bl/6 cell model lacking forskolin response and the Cftr ^tm1Cam C57Bl/6 model that had a forskolin response could be due to differences in the functional measurements. Here, pAEC ALI cultures were established in vitro from bronchial epithelium to measure the forskolin response, whereas Wilke et al (2011) utilised excised nasal epithelial sheets for Ussing chamber analysis (Wilke et al., 2011). This confounding issue reflects some of the problems encountered when utilising animal models to represent human disease and different methods to capture functional data.

Given the intermediate survival rate seen with diet supplemented (from 21 days) Cftr ^Δ9/Δ9 mice, Cftr exon 9 could have some minor level of redundancy, at least in this mouse model. Duchenne muscular dystrophy is an example of genotype/phenotype correlations; and the DMD gene contains a number of partially dispensable exons, as we and others have observed some functional protein is preferable to no functional protein (Mann et al., 2001). This redundancy may be of therapeutic potential and future work should utilise human tissue to investigate this further. This mouse model like others could be used to bridge the gap between cell culture-based studies and human trials for evaluation of novel drug efficacy and dosing. This Cftr ^Δ9/Δ9 mouse model demonstrates that exon skipping could be a feasible strategy for selected CFTR mutations to reduce a severe CF phenotype, although further investigation in human-derived tissue models is required as has been done for CFTR exon 23 skipping for mutations p.Trp1282X (Kim et al., 2022) (Oren et al., 2021; Michaels et al., 2022).