Modulation of cAMP metabolism for CFTR potentiation in human airway epithelial cells

Cystic fibrosis (CF) is a genetic disease characterized by CF transmembrane regulator (CFTR) dysfunction. With over 2000 CFTR variants identified, in addition to known patient to patient variability, there is a need for personalized treatment. The discovery of CFTR modulators has shown efficacy in certain CF populations, however there are still CF populations without valid therapeutic options. With evidence suggesting that single drug therapeutics are insufficient for optimal management of CF disease, there has been an increased pursuit of combinatorial therapies. Our aim was to test cyclic AMP (cAMP) modulation, through ATP Binding Cassette Transporter C4 (ABCC4) and phosphodiesterase-4 (PDE-4) inhibition, as a potential add-on therapeutic to a clinically approved CFTR modulator, VX-770, as a method for increasing CFTR activity. Human airway epithelial cells (Calu-3) were used to test the efficacy of cAMP modulation by ABCC4 and PDE-4 inhibition through a series of concentration–response studies. Our results showed that cAMP modulation, in combination with VX-770, led to an increase in CFTR activity via an increase in sensitivity when compared to treatment of VX-770 alone. Our study suggests that cAMP modulation has potential to be pursued as an add-on therapy for the optimal management of CF disease.


Results
In vitro extracellular cAMP assay analysis of ABCC4 inhibitor compounds in human airway epithelial cells. An in vitro extracellular cAMP assay of two commercially available ABCC4 inhibitors, MK-571 and Ceefourin-1 (Fig. 1a,b), was performed in human airway epithelial Calu-3 cells, and human bronchial epithelial cells (HBEC6-KT, Supplementary Fig. 1), to define their efficacy in decreasing extracellular cAMP levels. Concentration-dependent ABCC4 inhibition using MK-571 and Ceefourin-1 (Fig. 1) caused a decrease in extracellular cAMP levels in both cell lines. The half maximal inhibitory concentration (IC 50 ) values of MK-571 and Ceefourin-1 were found to be 2.6 μM and 0.7 μM respectively and defined a role for ABCC4-mediated cAMP transport in human airway epithelial cells that may be leveraged for CFTR potentiation.
Detection of ABCC4 and CFTR in human airway epithelial cell lines. The expression and function of ABCC4 has been confirmed in HBEC6-KT cells ( Supplementary Fig. 1) although CFTR expression and function remain to be defined 41,42 . In contrast, human airway epithelial Calu-3 cells are well documented for CFTR expression and function but not for ABCC4 [53][54][55] .
ABCC4 and CFTR protein expression levels were therefore probed via immunoblot in HBEC6-KT and Calu-3 cell lines with total protein stain performed as a loading control (Fig. 2). Due to reported time-dependent differentiation and polarization of Calu-3 cells, protein was extracted at 4 different time points (0, 7, 14, and 21 days post-confluency) 53,[56][57][58] . ABCC4 was present in both HBEC6-KT and Calu-3 cells at all-time points, indicated by the bands present at 150 kDa. CFTR was present at all-time points in Calu-3 cells. No CFTR was detected in HBEC6-KT cells. Due to co-expression of ABCC4 and CFTR, Calu-3 cells were used for subsequent studies exploring the interrelationships between ABCC4, cAMP modulation, and CFTR.

Receptor-dependent and -independent activation of CFTR activity-impact of VX-770. A
concentration-response analysis was performed with forskolin (FSK) and isoproterenol (ISO), G protein-coupled receptor-independent and -dependent cAMP inducers, respectively (Fig. 3). The half maximal effective concentration (EC 50 ) value for FSK and ISO were determined to be 0.19 μM and 0.07 μM respectively, with both compounds demonstrating ability to induce CFTR activity as measured by a validated membrane potential sensitive assay 14 .
Concentration-response curves with CFTR modulator VX-770, using both receptor-independent FSK and receptor-dependent ISO cAMP inducers (Fig. 4a,e), demonstrated an upward and leftward shift in the curve, indicating an increase in CFTR activity. Area under the curve (AUC) analysis showed a significant increase with the addition of VX-770 with either cAMP inducer (Fig. 4b,f; **P ≤ 0.01 and ****P ≤ 0.0001). VX-770 increased the sensitivity of FSK-induced CFTR activity as measured by EC 50 analysis ( Fig. 4c; **P ≤ 0.01). However, the maximal effect concentration (E max ) remained unchanged (Fig. 4d). Despite changes in AUC, no changes in EC 50 or E max were observed with the combination of VX-770 and ISO (Fig. 4g,h).
Collectively, these results demonstrate that both receptor-dependent and -independent mechanisms of cAMP elevation are able to induce CFTR activity in Calu-3 cells, which in turn is potentiated with VX-770.
Effect of ABCC4 and PDE-4 inhibition on CFTR activity. Following our quantification of CFTR activity with the clinically approved CFTR potentiator, VX-770, we next investigated the effect of ABCC4 inhibition and PDE-4 inhibition in Calu-3 cells.
For pharmacological inhibition of PDE-4 we used Roflumilast and Rolipram with FSK and ISO ( Fig. 6a-d). In contrast to ABCC4 inhibition, the concentration-response curves of PDE-4 inhibitors showed leftward shifts with ISO (Fig. 6c,d). Similarly, AUC analysis revealed significant increases with the addition of either Roflumilast or Rolipram with ISO but not FSK ( Fig. 6e; *P ≤ 0.05). EC 50 , E max , and baseline values for PDE-4 inhibitors with either cAMP inducer were not different ( Fig. 6f-h). In addition to Roflumilast and Rolipram, pharmacological inhibition using non-specific PDE inhibitor IBMX was also tested ( Supplementary Fig. 2). Significant increases in AUC ( Supplementary Fig. 2c,i; *P ≤ 0.05; **P ≤ 0.01), with either cAMP inducer, were seen and significant increases in baseline values for IBMX with ISO only was revealed (Supplementary Fig. 2l; *P ≤ 0.05).
Effect of cAMP modulation with VX-770 on CFTR activity. Due to our observations that ABCC4 and PDE-4 inhibition alone showed minor potentiation of CFTR activity relative to VX-770, we next investigated whether a combinatorial approach could be more efficacious 13 .
Concentration-response curves of VX-770 + MK-571 or Roflumilast combinations with FSK stimulation led to an upward shift of the curve at all concentrations, indicating an increase in CFTR activity (Fig. 7a,b). VX-770 + MK-571 or Roflumilast combinations with ISO also led to shifts in the curves, however only at lower concentrations (Fig. 7c,d). For the AUC analysis of each combination, increases were observed with FSK but not ISO ( Fig. 7e; **P ≤ 0.01, and ***P ≤ 0.001). EC 50 and E max analysis for each combination with both cAMP inducers showed no difference (Fig. 7f,g). Concentration-response analysis of two different cAMP-elevating agents on CFTR activity. Calu-3 cells were exposed to G protein-coupled receptor-independent cAMP inducer (a) forskolin and G proteincoupled receptor-dependent cAMP inducer (b) isoproterenol at increasing concentrations while measuring CFTR activity using an in vitro CFTR membrane potential assay. EC 50 values were analyzed and reported. Each concentration-response curve was normalized to baseline over DMSO vehicle control with data presented as ± standard deviations (n = 8). To compare our series of cAMP modulation interventions with and without VX-770 (Figs. 4, 5, 6, 7), foldchange comparisons of AUC, EC 50 , and E max , relative to control conditions were performed for all experiments (Fig. 8). Fold-change analysis of AUC with FSK stimulation showed that VX-770 was superior to either ABCC4 (MK-571) or PDE-4 (Roflumilast) inhibition alone, and that combinatorial approaches did not potentiate CFTR AUC values (Fig. 8a, **P ≤ 0.01 and ***P ≤ 0.001). ISO stimulation resulted in similar modest trends that were not significant (Fig. 8d). Fold-change EC 50 analysis between VX-770 alone to VX-770 with ABCC4 or PDE-4 inhibition showed a significant decrease with ISO for both combinatorial approaches, observations that were absent with FSK stimulation. (Fig. 8b,e, *P ≤ 0.05). Lastly, fold-change E max analysis performed with either FSK or ISO showed no changes for any combinatorial approach (Fig. 8c,f).

Discussion
There is an increased need for combinatorial therapies for CF subjects to optimally treat all subjects at a personalized level 13,25,27,29 . CFTR potentiation with VX-770 represents a high-water mark for increased chloride conductance in CF patients. In the present study, we quantified the capacity of combinatorial approaches that modulate intracellular cAMP levels to activate CFTR. Specifically, we inhibited ABCC4, a cAMP-efflux pathway, and PDE-4, a cAMP metabolism pathway. Using Calu-3 cells, we demonstrate that interventions targeting intracellular cAMP are unable to potentiate CFTR to levels observed with VX-770. Combinatorial approaches of cAMP modulation with VX-770 suggest that increases in CFTR activity as measured by AUC and EC 50 values, are possible and warrant further exploration in primary human airway epithelial cells expressing loss of function CFTR variants 14 .
The addition of VX-770 to Calu-3 cells induced shifts in the curve with both FSK and ISO, providing context for the amount of CFTR potentiation possible in Calu-3 cells with a clinically approved CFTR potentiator. While there was a significant change to EC 50 with FSK, no changes were seen with ISO, suggesting that mechanisms   www.nature.com/scientificreports/ significant increase in baseline with ISO (data not shown). Collectively, this suggests not only is the mechanism of action in elevating cAMP levels affecting sensitivity, but also efficacy [59][60][61][62][63] . Collectively, these results suggest that it is possible to potentiate CFTR activity in Calu-3 cells by modulating cAMP levels. www.nature.com/scientificreports/ Combinatorial approaches have been suggested as the future for modulation of CFTR activity and may include direct or indirect approaches 13 . Modulation of cAMP levels functions as an indirect mechanism to modulate CFTR and could potentially be combined with direct CFTR modulator approaches. We therefore decided to investigate ABCC4 and PDE-4 inhibitors which are able to differentially modulate cAMP levels by blocking extracellular transport and intracellular breakdown, respectively.
It has been previously demonstrated that pharmacological inhibition of ABCC4 by MK-571, in combination with VX-770, was able to improve CFTR activity in primary nasal epithelial cells from CF patients heterozygous for the G551D-variant 14 . In addition, in primary non-CF HBECs, it has been demonstrated that the administration of Roflumilast led to CFTR activation 38 . However, in this same study, the addition of Roflumilast to Calu-3 cells did not potentiate CFTR activity beyond the max stimulation that resulted from the addition of FSK 38 .
In our concentration-response analysis of ABCC4 and PDE-4 inhibitors in Calu-3 cells, only mild CFTR potentiation was observed alone or in combination with VX-770. Our results for ABCC4 inhibition contrast our previous demonstration of ABCC4 inhibitor MK-571 in potentiating CFTR activity in G551D-variant expressing primary human airway epithelial cells 14 . Conversely, our results of PDE-4 inhibitors align with previous findings in primary HBECs that were exposed to whole cigarette smoke in which Roflumilast, in combination with VX-770, demonstrated an increase in short-circuit current, suggesting this combination leads to an increase in CFTR activity 38 . We highlight that it may be possible that wild-type CFTR expressed in Calu-3 cells is maximally active and difficult to further potentiate with cAMP modulating agents. Our data showing shifts in the left of the curve (increased sensitivity of CFTR activity) without changes in maximal response support this concept. Specifically, the addition of VX-770 with either cAMP inducer to Calu-3 cells led to an upward shift in the curve, www.nature.com/scientificreports/ a significant increase in AUC but no significant changes to E max . However, VX-770 with FSK showed a significant decrease in EC 50 and VX-770 with ISO showed a significant increase in baseline values. When VX-770 was combined with an ABCC4 or PDE-4 inhibitor and cAMP inducer FSK, there were significant increases in AUC, indicating an increase in CFTR activity, but again no change in E max . In order to determine whether ABCC4 and PDE-4 inhibitor with VX-770 had an additive effect beyond VX-770 alone, a fold-change analysis for AUC, EC 50 , and E max values of the independent experiments was performed for indirect comparisons. Combinatorial additions of VX-770 with ABCC4 or PDE-4 inhibitor, in the presence of cAMP inducer ISO, led to a decrease in fold EC 50 values compared to VX-770 alone, suggesting that ABCC4 and PDE-4 inhibitors may increase the sensitivity to VX-770. In addition, with cAMP inducer FSK, VX-770 with Roflumilast had a significant increase in fold baseline values, suggesting that there may also be an increase in efficacy. With these fold-change comparison analyses in mind, further investigation of these combinatorial treatments should be performed. Our interrogation of the role of ABCC4 in airway epithelial cell modulation of cAMP levels is grounded in reports from primary human airway epithelial cells and cell lines, while examining CFTR activity in Calu-3 cells has a strong foundation 35,41,42,53,54 . To determine the optimal cell system to perform ABCC4 interventions with outcomes of CFTR activity, we interrogated both HBEC-6KT and Calu-3 cells for ABCC4 and CFTR protein expression levels. To our surprise, while ABCC4 is present in both cell lines, CFTR is only present in the Calu-3 cells, leading to the selection of Calu-3 cells as the model of study. Our observation highlights the importance of basal expression analysis for CFTR protein in any human airway epithelial cells or lines prior to performing functional experiments.
Elevations in intracellular cAMP can be induced via multiple G protein-coupled receptor-dependent (β-adrenergic receptor and adenosine receptor agonists-ISO and adenosine respectively) and independent mechanisms (direct activation of the enzyme adenylyl cyclase-FSK) [64][65][66][67][68] . The compartmentalization of cAMP signalling may be important for downstream signalling pathways within the cell-with specific implications for CFTR activity 34,69 . To determine whether the mechanism of cAMP production impacts CFTR activity, Calu-3 cells were treated with FSK and ISO at varying concentrations. Sigmoidal shape concentration-response curves were observed for both cAMP-elevating agents with the maximal degree of change for CFTR activity greater with ISO. The interpretation of this observation is that the compartmentalization of cAMP to the plasma membrane may more effectively couple protein kinase A activation to CFTR phosphorylation, which could be masked by global cytosolic increases in cAMP induced by FSK 34,69 .
In our previous studies, we have not performed complete concentration-response analysis experiments for the leukotriene receptor antagonist MK-571, which has off-target ABCC4 inhibition effects, in parallel to the more selective ABCC4 inhibitor Ceefourin-1, thus we wanted to explore both in greater detail 14,42,45 . In our concentration-response analysis experiments utilizing an in vitro extracellular cAMP assay platform, both commercially available Ceefourin-1 and MK-571 were able to decrease extracellular cAMP levels, thus both were used in downstream in vitro CFTR membrane potential assays to investigate the consequences of ABCC4 inhibition on CFTR function. Due to similar findings using either compound in the in vitro CFTR membrane potential assay and previous demonstrations of MK-571, in combination with VX-770, potentiating CFTR activity beyond VX-770 alone, MK-571 was selected for combinational studies 14 .
While our findings suggest that cAMP modulation has the potential to be used as an add-on therapy with existing therapeutics, limitations of the study should be noted. Although Calu-3 cells expressed both ABCC4 and CFTR, it might not have been an appropriate model of study due to its WT-CFTR expression. Potentiation of CFTR activity by ABCC4 or PDE-4 inhibitors in the literature were performed in primary cells, suggesting that a potential add-on combinatorial therapy may only show an effect in primary cell cultures, especially ones with CFTR defects. This emphasizes that further investigation of ABCC4 and PDE-4 inhibitors as a potential add-on combinatorial therapy should be performed in primary human airway epithelial cells from a large subset of CF subjects, covering a wide-range of CFTR variants with compromised CFTR expression or function, in order to determine which patient populations could benefit from cAMP modulation therapy. In addition, further investigations of cAMP modulation with other drugs that modulate CFTR, such as VX-770, VX-809, VX-661, VX-445, and genistein, should be tested as it may be more efficacious.   Fig. 4).

Reagents
In vitro extracellular cAMP assay. Calu-3 and HBEC6-KT cells were cultured as described previously 41,42,71,72 . Cells were pretreated with IBMX (20 μM) for 2 h prior to experimental conditions. After the pre-treatment of IBMX, cells were treated with ABCC4 inhibitors (0.01-100 μM) or DMSO for 30 min. Following these exposures, the cells were treated with forskolin (10 μM) for 6 h and cell-culture supernatants were collected for analysis of extracellular cAMP. The negative control was IBMX alone and the positive control was IBMX and forskolin.
In vitro CFTR membrane potential assay. An in vitro CFTR membrane potential assay was performed on Calu-3 cells and primary human bronchial epithelial cells, from two independent donors, cultured at 37 °C as previously described 14 . All human samples were collected from consented individuals under ethics approval granted by Hamilton Integrated Research Ethics Board (Project Number 5099-T). Calu-3 cells were grown for 21 days on 96-well plates and primary HBECs were grown for 14 days under air-liquid interface conditions on collagen-coated (Sigma-Aldrich) 6.5 mm Transwell Inserts (Corning). All cells were washed with HBSS prior to experimental conditions. After washing with HBSS, cells were loaded with BLUE Membrane Potential Dye (Molecular Devices, #R8042) dissolved in 37 °C chloride-free buffer (NMDG Gluconate Buffer -150 mM NMDG-gluconate, 3 mM KCl, 10 mM HEPES, pH 7.35, 300 mOsm). Immediately after dye loading, CFTR activity measurements were taken. CFTR activity measurements are fluorescent readouts for detecting ion channel activity. CFTR inhibition was performed with CFTRinh-172 (10 μM) for all experiments to attribute changes in fluorescence as CFTR channel activity. This assay has been validated using proof of concept studies where results obtained from this assay were able to recapitulate findings obtained from direct measurements of ion channel activity using Ussing chambers 14 . At the end of this assay, raw data was exported for statistical analysis. Tracings demonstrating the variation in CFTR Activity as a function of time, for key experiments, have been included in the supplementary information ( Supplementary Fig. 5).
For Calu-3 cells, measurements were taken during baseline (40 min), cAMP elevation (30 min), and CFTR inhibition (20 min). cAMP elevation was performed using forskolin or isoproterenol (0.0005-50 μM). For assays with ABCC4 or PDE-4 inhibitors, a pre-incubation with the inhibitors (30 min) in the plate reader was performed prior to cAMP elevation. CFTR modulator VX-770 was used at 1 μM. CFTR activity was determined by dividing a single membrane potential peak measurement after the ABCC4 or PDE-4 inhibition and cAMP elevation additions to the stabilized baseline over DMSO control.
For primary HBECs, measurements were taken during baseline (8 min), drug additions (40 min), and CFTR inhibitions (18 min). cAMP elevation was performed using forskolin (10 μM). The same concentrations used in Calu-3 cells for ABCC4 and PDE4 inhibitors, and VX-770 were used for primary HBECs, however the preincubation was skipped and instead, all drug combinations were added once at the beginning of the drug addition step. CFTR activity was determined by averaging six independent measurements within a single Transwell insert. Measurements were normalized to the averaged baseline over DMSO control.

Statistical analysis.
For the in vitro extracellular cAMP assay, values were normalized to the positive control (IBMX and forskolin) and the half maximal inhibitory concentration (IC 50 ) was determined. SD was calculated using data from biological replicates (n = 4-5).
For in vitro CFTR membrane potential assays, in Calu-3 cells, values were normalized to the averaged baseline over averaged vehicle (DMSO) control of all biological replicates. SD was calculated using data from biological replicates (n = 4-8). An area under the curve (AUC), half-maximal concentration (EC 50 ), maximum concentration (E max ), and baseline analysis was performed for each individual replicate then averaged together. The AUC analysis is an aggregate measure of sensitivity and maximal response, allowing for the observation of the total net responsiveness of the cells. The AUC analysis was done by normalizing the individual replicates to the bottom most value of the averaged vehicle control. In primary HBECs, six independent measurements were taken within a single Transwell insert. The value is the average of at least six measurements. These values were normalized to the averaged baseline over the averaged vehicle (DMSO) control. SD was calculated by treating the six measurements as independent, with additional technical replicates (n = 6-18). An AUC and max peak analysis were performed for each measurement, then averaged together. The max peak analysis was the max CFTR activity achieved during the drug addition step.
In order to compare the independent experiments performed, a fold-change analysis using individual AUC, EC 50 , and E max , determined from their respective concentration-response curves, was performed and was normalized to the averaged DMSO vehicle control. A one-way ANOVA with subsequent post-hoc test or a paired t-test was performed. Statistical analysis was performed using GraphPad Prism 6.

Data availability
The datasets generated and analysed during the course of the study are available from the corresponding author on reasonable request.