A Precision Medicine Approach to Optimize Modulator Therapy for Rare CFTR Folding Mutants

Trikafta, a triple-combination drug, consisting of folding correctors VX-661 (tezacaftor), VX-445 (elexacaftor) and the gating potentiator VX-770 (ivacaftor) provided unprecedented clinical benefits for patients with the most common cystic fibrosis (CF) mutation, F508del. Trikafta indications were recently expanded to additional 177 mutations in the CF transmembrane conductance regulator (CFTR). To minimize life-long pharmacological and financial burden of drug administration, if possible, we determined the necessary and sufficient modulator combination that can achieve maximal benefit in preclinical setting for selected mutants. To this end, the biochemical and functional rescue of single corrector-responsive rare mutants were investigated in a bronchial epithelial cell line and patient-derived human primary nasal epithelia (HNE), respectively. The plasma membrane density of P67L-, L206W- or S549R-CFTR corrected by VX-661 or other type I correctors was moderately increased by VX-445. Short-circuit current measurements of HNE, however, uncovered that correction comparable to Trikafta was achieved for S549R-CFTR by VX-661 + VX-770 and for P67L- and L206W-CFTR by the VX-661 + VX-445 combination. Thus, introduction of a third modulator may not provide additional benefit for patients with a subset of rare CFTR missense mutations. These results also underscore that HNE, as a precision medicine model, enable the optimization of mutation-specific modulator combinations to maximize their efficacy and minimize life-long drug exposure of CF patients.


Introduction
Cystic fibrosis (CF), one of the most prevalent life-threatening monogenetic diseases, is caused by loss-of-function mutations in the CF transmembrane conductance regulator (CFTR) gene [1]. The CFTR proteins forms an anion channel expressed at the apical surface of various epithelia, which is involved in the ion composition, pH and volume regulation of the luminal fluid in the airway, intestine and biliary, as well as pancreatic duct [1]. CFTR is a member of the ABC transporter superfamily and consist of two homologous halves, each containing a membrane spanning domain (MSD) and a nucleotide binding domain (NBD) that are connected by a unique regulatory domain [2].

Human Nasal Epithelia
HNE cell isolation from tissue collected by scrape biopsy was performed as described [36]. HNE were expanded in presence of irradiated feeder cells in Rock inhibitor containing medium F, a process termed conditional reprogramming [37]. Functional measurements were performed in filter grown differentiated HNE. For this purpose, HNE were seeded at a density of 5 × 10 5 cell/filter on 1.12 cm 2 Snapwell filter supports (Corning) and differentiated under air-liquid interface by culturing in PneumaCult-ALI medium (Stemcell Technologies) for ≥three weeks.

PM Density Measurement
To determine the PM density of 3HA-tagged CFTRs in CFBE, a cell surface enzymelinked immunosorbent assay (ELISA) using mouse monoclonal anti-hemagglutinin (HA) antibody (Biolegend, clone 16B12, order no. 901524, 1:2000) was applied [14]. Cells were treated with compounds or 0.2% DMSO (vehicle control) for 24 h at 37 • C in full medium. PM density values were normalized with cell viability determined by alamarBlue Assay (Invitrogen) and related to WT-CFTR by normalizing with CFTR mRNA abundance measured by qPCR as previously [19].

Peripheral Stability Measurement and Immunoblotting
To measure the stability of the complex-glycosylated form of CFTR variants, CFBE cells were treated with 100 µg/mL cycloheximide (CHX) for 0-7.5 h at 37 • C in full medium. Cells were lysed in RIPA buffer supplemented with protease inhibitors (Complete, Roche) followed by removal of insoluble material, protein concentration determination, immunoblotting and CFTR detection using the mouse monoclonal anti-hemagglutinin (HA) antibody as described before [14].

Short-Circuit Current Measurement
Short-circuit current (I sc ) measurement of polarized HNE was performed as described [19,27]. HNE were treated with CFTR modulators or 0.2% DMSO (vehicle control) for 24 h at 37 • C in serum-free PneumaCult-ALI medium (StemCell Technologies). Snapwell filters were mounted in Ussing chambers (Physiologic Instruments) in Krebs-bicarbonate Ringer (KBR) buffer (140 mM Na + , 120 mM Cl − , 5.2 mM K + , 25 mM HCO 3 − , 2.4 mM HPO 4 , 0.4 mM H 2 PO 4 , 1.2 mM Ca 2+ , 1.2 mM Mg 2+ , 5 mM glucose, pH 7.4), which was mixed by bubbling with carbogen (95% O 2 and 5% CO 2 ). To generate a basolateral-to-apical chloride gradient NaCl was replaced with 115 mM Na + gluconate in the apical buffer and the I sc was determined in the presence of 100 µM amiloride. The transepithelial voltage was clamped at 0 mV (VCC MC8 multichannel voltage/current clamp, Physiologic Instruments) after compensating for voltage offsets and current and resistance were recorded at 37 • C with the Acquire and Analyze package (Physiologic Instruments).

Statistics
Results are presented as mean ± SEM with the number of experiments indicated. Statistical analysis was performed by two-tailed Student's t-test with the means of at least three independent experiments and the 95% confidence interval was considered significant.

Identification of Mutants with High Responsiveness to Type I Correctors
To determine the relative correction efficacy of the type I corrector VX-661 and the type III corrector VX-445, we measured the PM density of 13 CF-causing mutations that are associated with folding defects in the CF bronchial epithelial cell line CFBE41o-(CFBE). These mutations, distributed throughout CFTR, lead to various reductions of the mutant PM densities relative to the wild-type (WT) after normalization for mRNA expression (Supplementary Figure S1A). The CFTR PM density changes ranged from very severe loss (<5% of the WT) for S13F, E92K, S492F, F508del, V520F, L1077P, M1101K, and N1303K to relatively mild reduction (>20% of the WT) for R31C, R352G and S549R, as reported previously [19,27]. Apart from S492F, V520F and M1101K, which were resistant to correction with VX-661, the processing defect of all mutations was partially reversed by both correctors to various extend (Supplementary Figure S1A). To identify mutations that exhibit preferential responsiveness to type I correctors, the ratio of the relative correction efficacy of VX-661 and VX-445 was calculated ( Figure 1A). Mutations in the MSD1 and the proximal N-terminus were more responsive to VX-661 than to VX-445 ( Figure 1A), consistent with their high responsiveness to, and the stabilization of the MSD1 and CFTR N-terminal half by VX-809 ( Figure 1B) [38][39][40][41]. Mutations in the other CFTR domains showed partial selectivity for correction with VX-445, with the exception of the S549R in the NBD1, which could be nearly completely corrected by either corrector ( Figure 1A and Supplementary Figure S1A). Since VX-661 treatment resulted in >50% of WT PM expression, the highest absolute correction of P67L, L206W and S549R variants, we chose these mutants to test whether the addition of VX-445 would provide additional benefit.

Cooperative Correction of the Mature Protein and PM Expression by VX-661 + VX-445 Corrector Combination
Similar to monitoring the PM density, the folding defect and its correction can be assessed by quantifying the abundance of the mature, complex-glycosylated form (band C) of CFTR mutants relative to that of the WT, after normalizing for the mRNA abundance [19]. The type I correctors VX-809 and VX-661 resulted in a substantial increase in the band C abundance of P67L-, L206W-and S549R-CFTR, which for P67L and L206W could be further increased by VX-661 + VX-445 treatment ( Figure 1B). Similar results were obtained by PM density measurement for P67L-and L206W-CFTR, and the small increase in S549R-CFTR PM density upon addition of VX-445 to the type I corrector treatment reached significance in this assay ( Figure 1C). Since the PM density assay allows for higher-throughput measurements, we also determined the effect of the type I correctors ABBV-2222 and FDL-169, which are in clinical development [42,43]. Similar to its effect on E92K-CFTR [27], ABBV-2222 exhibited higher efficacy to correct the folding defect of P67L-and L206W-CFTR in comparison to other type I correctors ( Figure 1C). Co-treatment of all tested type I correctors with VX-445 resulted in a small, but significant, increase in the PM density of P67L-, L206W-and S549R-CFTR ( Figure 1C).
To test whether the cooperative correction can be accounted for by discrete mechanism of actions on distinct structural defects, the additivity or synergy of the two corrector classes was analyzed by combinatorial profiling. To this end, the rescue efficacy of corrector pairs was compared to their theoretical additivity, based on their individual efficacy (Supplementary Figure S1B-D). The profiles of their interaction with other preclinical correctors enabled us to cluster compounds with similar mechanisms. Cluster analysis for P67L-, L206W-and S549R-CFTR indicated that the type I correctors clustered together, VX-445 clustered together with the type III corrector 4172, and the type II corrector 3151 forms its own category with variable distance to the other two clusters, similar to the published data for F508del and some missense mutations [27] ( Figure 1D). These results confirm the mechanistic classification of the correctors as proposed earlier [19,20,27] and show that mutations, which are highly responsive to stabilization of the MSD1 and MSD1-NBD1, can be further corrected by correctors targeting the NBD1 or NBD2. This is in agreement with a recent study reporting that P67L impairs folding of downstream domains including the NBD1 and NBD2 [44].  The folding defect introduced by the P67L, L206W and S549R mutations may reduce the trafficking of the channel to the plasma membrane, stability of the mature protein or combination of both defects. Severely reduced maturation rate of L206W and the VX-809mediated folding correction of P67L and L206W have been shown [38,39,45], therefore we focused on determining the stability of the post-Golgi, complex-glycosylated form of the mutants. As determined by cycloheximide (CHX) chase and immunoblotting, presence of the P67L or L206W mutation reduced the half-life of complex-glycosylated CFTR tõ 2.5 h and~5 h, respectively, while the S549R only introduced a minor stability defect (Figure 2A,B). VX-809 or VX-661 partially correct the P67L-CFTR and almost completely corrected the L206W-CFTR stability defect. Co-treatment with VX-445, however, did not provide significant additional benefit (Figure 2A,B).

Some Patients with the L206W Mutation Exhibit Progressive Lung Function Decline
To analyze the effect of CFTR modulators on the mutant function in patient-derived airway epithelia, we identified and subsequently collected human nasal epithelia (HNE)

Some Patients with the L206W Mutation Exhibit Progressive Lung Function Decline
To analyze the effect of CFTR modulators on the mutant function in patient-derived airway epithelia, we identified and subsequently collected human nasal epithelia (HNE) form nine CF patients carrying the P67L, L206W or S549R mutation. Since these are rare mutations with an allelic frequency of 0.17%, 0.23% and 0.065% for P67L, L206W and S549R (CFTR2 database, www.cftr2.org, accessed on 6 July 2021), respectively, all patients are compound heterozygous with F508del as the most frequent mutation on the second allele.
The L206W mutation leads to a mild CF phenotype [45], which is exemplified by a low incidence of pancreatic insufficiency, rare chronic colonization with Pseudomonas aeruginosa and only moderate elevation of sweat chloride (CFTR2 database, www.cftr2.org, accessed on 6 July 2021). In our small patient cohort, the average age at the time of HNE collection was 54.3 ± 10.3 years (range 33-64 years) and the forced expiratory volume in 1s (FEV1) % predicted 70% ± 29% (range 39-107%), indicative for the variable disease phenotype. However, a subset of these patients had substantial airflow obstruction (FEV1% predicted <50%) and exhibited a progressive loss of lung function over time ( Figure 3A). Consistent with the association of the P67L mutation to a milder phenotype [46], the lung function of the compound heterozygous F508del/P67L patient remained stable (FEV1% predicted >50%) over several years ( Figure 3B). Due to the pronounced gating defect the S549R mutation is associated to a severe CF phenotype [47], however the F508del/S549R patient in this study was able to maintain a FEV1% predicted >80% after the onset of ivacaftor therapy at age 18 ( Figure 3B). These results underscore the need for modulator therapy even in some patients carrying mutations that are generally associated with a milder phenotype.

Functional Correction of P67L-, L206W-and S549R-CFTR in Human Nasal Epithelia
The HNE harboring P67L-, L206W or S549R-CFTR were expanded by conditional reprogramming and differentiated at air-liquid interface as reported [36]. In HNE from seven patients with one allele of L206W the mean forskolin-activated short-circuit current (Isc) was 6.6% relative to that of WT-CFTR in HNE isolated from 10 donors [27], which could be increased to 27.5% and 25.8% by correction with VX-809 and VX-661, respectively ( Figure 4A,B). Under activation conditions with saturating forskolin concentra-

Functional Correction of P67L-, L206W-and S549R-CFTR in Human Nasal Epithelia
The HNE harboring P67L-, L206W or S549R-CFTR were expanded by conditional reprogramming and differentiated at air-liquid interface as reported [36]. In HNE from seven patients with one allele of L206W the mean forskolin-activated short-circuit current (I sc ) was 6.6% relative to that of WT-CFTR in HNE isolated from 10 donors [27], which could be increased to 27.5% and 25.8% by correction with VX-809 and VX-661, respectively ( Figure 4A,B). Under activation conditions with saturating forskolin concentrations, the acute or chronic treatment with VX-770 did not significantly increase the I sc ( Figure 4A,B). For comparison, Trikafta treatment increased the F508del channel function in homozygous HNE to~60% of the WT [27].  A large inter-donor variation in the forskolin stimulated Isc in compound heterozygous L206W-CFTR HNE was observed both before and after correction with VX-809 or VX-661, similar as noted for homozygous ΔF508-CFTR HNE [19]. Concordantly, after expressing the corrected Isc as percentage of the basal channel activity lower variance A large inter-donor variation in the forskolin stimulated I sc in compound heterozygous L206W-CFTR HNE was observed both before and after correction with VX-809 or VX-661, similar as noted for homozygous ∆F508-CFTR HNE [19]. Concordantly, after expressing the corrected I sc as percentage of the basal channel activity lower variance between HNE from individual patients was observed (Supplementary Figure S2A). These observations led us to investigate whether the basal channel activity is correlated to the level of correction in individual HNE. Both parametric (Pearson correlation coefficient) and non-parametric (Spearman's rank correlation coefficient) correlation analysis indicated a highly significant correlation between the forskolin-activated basal I sc and the VX-809 or VX-661 corrected I sc (Figure 4C), similar to the correlation between the VX-809 corrected and basal CFTR activity observed in a panel of rare CFTR genotypes in HNE [48].
Next, we focused on the correction of L206W in compound heterozygous HNE carrying a splice site mutation on the second allele, which leads to no or substantially reduced protein expression. Similar to the global analysis of L206W carrying HNE, VX-809 or VX-661 resulted in a significant correction of the forskolin-activated I sc in these HNE, which could not be further increased by VX-770 in either the partially or completely activated channels ( Figure 4D and Supplementary Figure S2B). Importantly, the addition of VX-445 led to heterogenic effects and only further improved the VX-661 corrected I sc in one of the two tested HNE activated with saturating forskolin concentrations ( Figure 4D). In contrast, VX-445 increased the I sc in both HNE in which CFTR was partially activated with 100 nM forskolin (Supplementary Figure S2B).
We also investigated the corrector response of P67L/F508del and S549R/F508del compound heterozygous HNE. The CFTR function in P67L/F508del HNE was significantly corrected by VX-809 and VX-661 ( Figure 5A,B). The functional correction of the partially or fully activated P67L/F508del CFTR was further augmented by acute, but not by chronic addition of VX-770 ( Figure 5A,B and Supplementary Figure S2B), likely due to destabilization of both, corrected P67L and F508del proteins, by chronic VX-770 exposure as reported before [36,49]. CFTR function in these cells was also partially corrected by VX-445 and VX-661 + VX-445 co-treatment significantly increased the correction efficacy. Single corrector treatment led only to minor correction of the CFTR function in S549R/F508del HNE, which was significantly increased by both acute and chronic VX-770 exposure, consistent with the gating defect of the S549R mutant ( Figure 5A,B) [33,50]. Dual corrector treatment, however, did not further increase the functional correction efficacy ( Figure 5A,B). While the contribution of the F508del allele to the functional correction in these HNE is not established, the responses to correctors were similar to the biochemical results obtained in CFBE cells (Figure 1).
A weak correlation between the FEV1% predicted of F508del homozygous CF patients and the forskolin-induced swelling of corresponding patient-derived intestinal organoids has recently been reported, suggesting that differences in low CFTR residual function may contribute to clinical heterogeneity in F508del homozygous patients [51]. In case of the compound heterozygous P67L-, L206W or S549R-CFTR patients, however, we did not observe significant correlation between the basal channel function in HNE and the FEV1% predicted, which may be explained by the lower sample size in our study or the higher average age that augments the influence of environmental factors (Supplementary Figure S2C). S549R/F508del HNE, which was significantly increased by both acute and chronic VX-77 exposure, consistent with the gating defect of the S549R mutant ( Figure 5A,B) [33,50] Dual corrector treatment, however, did not further increase the functional correction ef ficacy ( Figure 5A,B). While the contribution of the F508del allele to the functional correc tion in these HNE is not established, the responses to correctors were similar to the bio chemical results obtained in CFBE cells (Figure 1).

Discussion
The approval of Trikafta, the combination of the folding correctors VX-661 + VX-445 and the gating potentiator VX-770, was recently expanded to 177 additional CFTR mutations located throughout the channel based on in vitro data in FRT cells, so far without formal publication of these results (Vertex press release: https://investors.vrtx.com/ news-releases/news-release-details/vertex-announces-fda-approvals-trikaftar, accessed on 6 July 2021) [52]. Here we are investigating three of these mutants, which were selected based on their high responsiveness to type I correctors, to analyze their responsiveness to the individual and combinations of CFTR modulators with the goal to achieve the highest functional correction efficacy with the least number of compounds. This goal is desirable since CFTR modulators can have adverse drug-drug interactions as has been shown for VX-809, which induces the cytochrome P450 variant CPY3A4 that is the metabolizing enzyme for VX-770, thus reducing the VX-770 plasma concentration [53]. VX-770 also has been shown to destabilize F508del-CFTR and some rare missense mutants, which attenuates the corrector efficacy [36,49].
In CFBE cells, treatment of P67L-, L206W-and S549R-CFTR with type I correctors corrected the PM density and complex-glycosylated form expression to~30-80% of the WT in agreement with published results [19,27,39]. The notable exception was treatment of P67L-and L206W-CFTR with the corrector ABBV-2222 [42] that resulted in ≥WT PM densities, indicating variable efficacies within the group of type I correctors. Co-treatment with the type III corrector VX-445 substantially increased the correction efficacy of P67Land L206W-CFTR but had a less pronounced effect on S549R-CFTR. Consistent with the tight correlation between corrector effects in CFBE and HNE cells [19], similar trends were observed for the functional measurements in HNE cells. The I sc of compound heterozygous HNE with one allele of P67L-, L206W-or S549R-CFTR was partially corrected by VX-809 or VX-661, which in P67L/F508del and in one of the two L206W/splice mutations containing HNE could be augmented by co-treatment with VX-445. Under conditions of CFTR activation by saturating concentrations of forskolin, VX-770 did not further increase the I sc in L206W-CFTR containing HNE. In P67L containing HNE the increase in I sc by acute VX-770 addition was negated by chronic VX-770 exposure likely due to destabilization of the corrected channel as reported before [36]. Since the level of channel activity and extend of CFTR phosphorylation in the lung tissue are not known, the modulator responses were also examined in HNE partially activated with 100 nM forskolin, which confirmed that VX-661 + VX-445 are sufficient for the functional correction of the P67L and L206W mutants. Consistent with clinical responses [47], the I sc was significantly corrected by VX-770 in S549R-CFTR containing HNE, which could be augmented by type I correctors but not by co-treatment with VX-445. The I sc measurements in L206W-containing HNE also showed that the residual channel function, which depends on individual's genetic and epigenetic determinant factors, is correlated to the corrector response. Thus, assuming that the CFTR function in HNE is a therapeutic marker for clinical responsiveness to CFTR modulators as has been suggested for F508del-CFTR [48,54] and a variety of other mutations [55], measurement of the basal channel function could predict the extend of a patient's clinical responsiveness.
We recently proposed the concept of allosteric corrector combinations, which is based on the posttranslational completion of the CFTR cooperative domain-folding and the coupled domain-misfolding of F508del and other folding mutations [18,20,56,57]. It predicts that the localized stabilization of CFTR domains by correctors of distinct folding defects will be propagated to distant regions of the channel and, thereby, combinations of correctors targeting different CFTR domains may synergistically stabilize a variety of mutants [19,27]. However, with mutants that are highly corrected by type I correctors the fractional benefits of additional correctors targeted to different domains is diminishing, which could explain the low responsiveness of S549R-CFTR to co-treatment with VX-445.
If the results in HNE can be extrapolated to the clinic, VX-661 + VX-770 (Symdeko) or VX-809 + VX-770 (Orkambi) will provide more efficacious restoration of the channel function in patients carrying the S549R mutation in comparison to VX-770 (Kalydeco) treatment alone and comparable to Trikafta. Patients carrying the P67L-and L206W-CFTR mutations could benefit from the VX-661 + VX-445 corrector combination, which is so far not marketed separately, but may not necessarily profit from the addition of VX-770. This study adds to the growing number of publications, in which patient-derived HNE are used to study the modulator responses of rare CFTR mutants [27,[31][32][33][58][59][60][61][62]. Our results suggest that HNE can not only be used to identify modulator responsive mutants, but also to optimize the modulator combinations for rare mutants. Thus, HNE may serve as a precision medicine tool to optimize the modulator therapy regiment for CF patients with rare mutations.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/jpm11070643/s1, Figure S1: The effect of single correctors and dual corrector combinations on the PM density of CFTR mutants in CFBE, Figure S2: Functional correction of L206W-CFTR in HNE.

Author Contributions:
The overall design of the study was by G.V. and G.L.L.; G.V. performed experiments and analyzed the results; H.X. generated the mutant CFTR expressing CFBE cell lines; N.V., L.B. and E.M. collected the patient data and samples for HNE isolation; T.V. provided the ABBV-2222 and VX-445. The manuscript was primarily written by G.V. and G.L.L., with input from all authors. All authors have read and agreed to the published version of the manuscript. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available in the article and supplementary material.

Conflicts of Interest:
The authors declare no conflict of interest.