Potentiation of BKCa channels by cystic fibrosis transmembrane conductance regulator correctors VX-445 and VX-121

Cystic fibrosis results from mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) anion channel, ultimately leading to diminished transepithelial anion secretion and mucociliary clearance. CFTR correctors are therapeutics that restore the folding/trafficking of mutated CFTR to the plasma membrane. The large-conductance calcium-activated potassium channel (BKCa, KCa1.1) is also critical for maintaining lung airway surface liquid (ASL) volume. Here, we show that the class 2 (C2) CFTR corrector VX-445 (elexacaftor) induces K+ secretion across WT and F508del CFTR primary human bronchial epithelial cells (HBEs), which was entirely inhibited by the BKCa antagonist paxilline. Similar results were observed with VX-121, a corrector under clinical evaluation. Whole-cell patch-clamp recordings verified that CFTR correctors potentiated BKCa activity from both primary HBEs and HEK cells stably expressing the α subunit (HEK-BK cells). Furthermore, excised patch-clamp recordings from HEK-BK cells verified direct action on the channel and demonstrated a significant increase in open probability. In mouse mesenteric artery, VX-445 induced a paxilline-sensitive vasorelaxation of preconstricted arteries. VX-445 also reduced firing frequency in primary rat hippocampal and cortical neurons. We raise the possibilities that C2 CFTR correctors gain additional clinical benefit by activation of BKCa in the lung yet may lead to adverse events through BKCa activation elsewhere.


Introduction
Cystic fibrosis (CF) affects approximately 40,000 individuals in the United States and approximately 100,000 people worldwide (1,2).The pathogenesis of CF is the result of mutations to the cystic fibrosis transmembrane conductance regulator (CFTR) protein, an apical membrane anion channel in epithelia.In the human airway, dysfunctional anion secretion leads to a reduction in airway surface liquid (ASL) height, resulting in reduced mucociliary clearance and an increase in the risk of infection (3)(4)(5)(6).CF can be caused by one of roughly 1,700 known mutations to the CFTR gene, which are separated into 6 unique categories based on how they result in a loss of CFTR expression/function (7).The 2 most common categories of mutations include those that affect the trafficking and gating of CFTR.F508del is the most common folding mutation, occurring in approximately 85% of patients with CF, and results in an anion channel that fails to correctly traffic to the apical membrane.
Therapeutics that target folding/trafficking mutations are known as "correctors."Corrector compounds can restore plasma membrane expression of F508del; however, this mutated CFTR protein is still incapable of gating correctly, so additional therapeutics are required, termed "potentiators," which restore CFTR function (8,9).
In addition to apical CFTR, transepithelial anion secretion requires the coordinated regulation of multiple conductances, including basolateral K + channels, which are responsible for potassium recycling and maintenance of the electrochemical driving force for apical Cl -efflux.Our group demonstrated that pharmacological activation of a basolateral Ca 2+ -activated K + conductance (KCa3.1)stimulates Cl -secretion across a wide array of epithelia, Cystic fibrosis results from mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) anion channel, ultimately leading to diminished transepithelial anion secretion and mucociliary clearance.CFTR correctors are therapeutics that restore the folding/trafficking of mutated CFTR to the plasma membrane.The large-conductance calcium-activated potassium channel (BK Ca , K Ca 1.1) is also critical for maintaining lung airway surface liquid (ASL) volume.
Here, we show that the class 2 (C2) CFTR corrector VX-445 (elexacaftor) induces K + secretion across WT and F508del CFTR primary human bronchial epithelial cells (HBEs), which was entirely inhibited by the BK Ca antagonist paxilline.Similar results were observed with VX-121, a corrector under clinical evaluation.Whole-cell patch-clamp recordings verified that CFTR correctors potentiated BK Ca activity from both primary HBEs and HEK cells stably expressing the α subunit (HEK-BK cells).Furthermore, excised patch-clamp recordings from HEK-BK cells verified direct action on the channel and demonstrated a significant increase in open probability.In mouse mesenteric artery, VX-445 induced a paxilline-sensitive vasorelaxation of preconstricted arteries.VX-445 also reduced firing frequency in primary rat hippocampal and cortical neurons.We raise the possibilities that C2 CFTR correctors gain additional clinical benefit by activation of BK Ca in the lung yet may lead to adverse events through BK Ca activation elsewhere.
Potentiation of BK Ca channels by cystic fibrosis transmembrane conductance regulator correctors VX-445 and VX-121 to amiloride, VX-121 (10 μM) also stimulated a large, paxilline-sensitive K + secretory current.As above, VX-121 decreased R te from 453 ± 49 Ω•cm 2 in the presence of amiloride to 248 ± 17 Ω•cm 2 , while addition of paxilline increased R te to 691 ± 68 Ω•cm 2 (n = 29), consistent with BK Ca activation and inhibition, respectively.We next determined whether other components of ETI would stimulate K + secretion across HBEs.Neither the C1 corrector VX-661 (Figure 1D) nor the CFTR potentiator VX-770 (Figure 1E) induced K + secretion, whereas subsequent addition of VX-445 or VX-121 stimulated K + secretion, respectively.To further confirm that this effect was due to apical BK Ca activation, we determined whether the currents induced by VX-445 and VX-121 could be inhibited by an additional selective BK Ca inhibitor, IBTX.As a 37-amino acid peptide, IBTX cannot cross the apical membrane.As shown, both the VX-445-induced (Figure 1F) and VX-121-induced (Figure 1G) K + currents were inhibited by IBTX (300 nM), verifying activation of BK Ca .Similar results were observed in 6 experiments each for VX-445 and VX-121.Given these surprising results, we determined the effect of a known BK Ca opener, NS1619, on K + secretion across HBEs.As shown in Figure 1H, 10 μM NS1619 failed to stimulate K + secretion, though the subsequent addition of 50 μM NS1619 stimulated a substantial K + secretory response.Further addition of VX-445 (10 μM) stimulated an additional K + secretory response that was completely inhibited by paxilline.The average responses are shown in Figure 1I, having a rank order of VX-121 > VX-445 > VX-659 (*P < 0.01).
While the most parsimonious explanation for our results is that C2 correctors potentiate apical BK Ca , we cannot rule out a potential role for the basolateral membrane in this response.Therefore, the effects of VX-445 and VX-121 on apical membrane K + currents were assessed after permeabilization of the basolateral membrane with nystatin (200 μM), as previously described (37).Formation of the nystatin pore is visualized in Figure 2, A and B, as a positive shift in baseline current to a new stable plateau.Subsequent addition of either VX-445 (Figure 2A) or VX-121 (Figure 2B) stimulated paxilline-sensitive K + currents that were indistinguishable from those in the absence of permeabilization.Consistent with our intact monolayers, the response to VX-121 was significantly greater than the response to VX-445 (Figure 2C; *P < 0.01).These results verify that C2 CFTR correctors potentiate apical BK Ca in WT CFTR-expressing HBEs.
We next determined the effects of varying the concentration of VX-445 and VX-121 on K + secretion to assess concentration dependence.As shown in Figure 3A, VX-445 stimulated K + secretion at low-micromolar concentrations, and exhibited a steep concentration dependence between 1 and 10 μM (Figure 3C).Likewise, VX-121 stimulated K + secretion at low-micromolar concentrations.However, VX-121 appeared more potent, producing visible responses at 0.3 and 1 μM (Figure 3B).As above, paxilline inhibited the K + currents stimulated by VX-445 and VX-121 (Figure 3, A and B).Summary data for these studies are provided in Figure 3C.As the response to VX-121 approached saturation at 10 μM, we were able to fit these data to the Hill equation (Figure 3D), and obtained an apparent EC 50 of 4.4 μM with a Hill coefficient of 3. In contrast, VX-445 did not saturate at 10 μM, and thus we were not able to obtain a reliable estimate of the EC 50 for this molecule.We did not routinely go to the next higher half-log concentration including human bronchial epithelial cells (HBEs) (20)(21)(22)(23)(24)(25).In the apical membrane, the BK Ca (KCa1.1)channel has been shown to promote Cl -secretion, and hence regulate ASL volume in HBEs.Indeed, pharmacological inhibition of BK Ca channels significantly reduces ASL volume (26).Importantly, the CF-associated inflammatory mediators IFN-γ and TGF-β decrease BK Ca channel expression, which correlates with a reduction in ASL volume (27,28).Based on these data, it has been proposed that pharmacological activation of BK Ca may be therapeutically beneficial in CF.
Given the role of BK Ca in modulating transepithelial Cl -secretion, we determined the effect of CFTR correctors on K + secretion across HBEs.We demonstrate, for the first time to our knowledge, that the C2 correctors VX-659 (bamocaftor), VX-445, and VX-121 (vanzacaftor) potentiate K + secretion across WT and F508del CFTR-expressing HBEs via Ussing chamber short-circuit current experiments.This C2-mediated K + secretion was entirely abrogated by the BK Ca channel antagonists paxilline and iberiotoxin (IBTX).Patch-clamp studies verified that C2 correctors potentiate BK Ca both heterologously expressed in HEK cells and endogenously expressed in primary HBEs via an increase in channel P o .Furthermore, we show that C2 correctors induce vasorelaxation of microvascular arteries and significantly alter neuronal excitability -two effects consistent with activation of BK Ca channels.Thus, while C2 corrector-dependent potentiation of BK Ca may be of benefit in airway epithelia, the cross-reactivity with BK Ca in other tissues may contribute to the adverse events reported by patients with CF upon initiation of ETI (29)(30)(31)(32)(33)(34).
Initially, we determined the effect of the current standard of care (SOC) C2 CFTR corrector, VX-445, on BK Ca function in primary WT CFTR HBEs grown at an air-liquid interface on Transwell filters.Studies were carried out in Ussing chambers using a 125:5 mM K + gradient (basal to apical) to measure K + secretion across the epithelia, as previously described (20).Note that our solutions result in a large apical-to-basolateral Na + gradient.As shown in Figure 1A, amiloride was used to inhibit the basal Na + absorption, resulting in an inwardly directed current, consistent with K + secretion across the apical membrane.Subsequent addition of VX-445 (10 μM) stimulated a large, slowly developing increase in inward current that was completely inhibited by the specific BK Ca blocker paxilline (10 μM).This result demonstrates that VX-445 stimulates K + secretion across the apical membrane of HBEs.Consistent with activation of BK Ca , VX-445 decreased transepithelial resistance (R te ) from 521 ± 57 Ω•cm 2 in the presence of amiloride to 283 ± 18 Ω•cm 2 , while addition of paxilline increased R te to 1,155 ± 109 Ω•cm 2 (n = 30).As shown in Figure 1B, a distinct C2 corrector, VX-659 (10 μM), similarly stimulated a paxilline-sensitive K + current.More recently, Vertex Pharmaceuticals developed a next-generation C2 CFTR corrector, VX-121, which is currently undergoing clinical trials (36).As shown in Figure 1C, subsequent (10 μM), stimulated a sustained K + secretory current, akin to the response of WT CFTR HBEs.Further, initial addition of VX-445 (10 μM) induced a K + secretory current that was not further increased by VX-661 (Figure 4B).Consistent with activation of an ionic conductance, VX-445 decreased R te from 221 ± 7 Ω•cm 2 in the presence of amiloride to 146 ± 6 Ω•cm 2 (P < 0.001, n = 28), and this was increased to 491 ± 24 Ω•cm 2 (P < 0.001, n = 28) following paxilline addition.As in WT CFTR HBEs, VX-659 (10 μM) stimulated K + secretion across F508del CFTR HBEs (Figure 4C).Similarly, the next-generation C2 CFTR corrector VX-121 (10 μM) stimulated a large, paxilline-sensitive K + secretory current (Figure 4D).Again, this was accompanied by a decrease in R te from 203 ± 9 Ω•cm 2 in the presence of amiloride to 140 ±4 Ω•cm 2 (P < 0.001, n = 28).Inhibition of BK Ca by paxilline increased R te to 441 ± 29 Ω•cm 2 (P < 0.001, n = 28), as expected.Finally, we determined whether VX-809 (lumacaftor), (30 μM), as this is above the concentration of VX-445 achieved in plasma (C max = 15 μM; ref. 38).Further, in 3 experiments in which we applied 30 μM VX-445, the subsequent addition of paxilline resulted in an increase in R te of only 154 Ω•cm 2 , whereas in 8 experiments carried out on the same day in which 10 μM VX-445 was added, the subsequent addition of paxilline increased R te by 654 Ω•cm 2 .As one possibility to explain this result is an overall decrease in R te , we did not pursue these higher concentrations further.Importantly, our results demonstrate that VX-445 stimulates K + secretion in the range of concentrations known for VX-445 in both plasma and cells of patients with CF (see Discussion).
Next, we determined whether CFTR correctors stimulate K + secretion across homozygous F508del CFTR HBEs.As shown in Figure 4A, following amiloride, the current SOC C1 corrector, VX-661 (10 μM), had no effect on K + secretory current.However, subsequent addition of the current SOC C2 corrector, VX-445 As CFTR is activated by cAMP/PKA, which also modulates BK Ca , it seems likely that CFTR and BK Ca are simultaneously activated during cAMP-mediated agonist addition (39,40).Thus, we determined whether forskolin stimulates K + secretion across F508del CFTR HBEs and whether this affects the ability of VX-445 to stimulate K + secretion.Initially, we verified that forskolin stimulates K + secretion across F508del CFTR HBEs.As shown in Figure 5A, subsequent to amiloride, forskolin (10 μM) stimulated a rapid inward current followed by a sustained paxilline-sensitive plateau.As HBEs express Kv7.1, Kv7.3, and Kv7.5 (KCNQ) channels (35,41,42), which are also activated by cAMP/ PKA, we determined whether these channels were contributing to the K + secretory response elicited by forskolin.The pan-Kv7.X a first-generation C1 corrector, would affect transepithelial K + currents across F508del CFTR HBEs.As shown in Figure 4E, VX-809 (10 μM) had no effect on K + current, while the subsequent addition of VX-121 (10 μM) again stimulated a paxilline-sensitive K + current.We used the peak response to thapsigargin (1 μM) as the gold standard for activation of BK Ca (Figure 4F), as we have previously shown that thapsigargin stimulates K + secretion across HBEs under the conditions used here (20).The average response of K + secretion to each CFTR corrector across F508del CFTR HBEs is shown in Figure 4G.Based on these results, we conclude that C1 correctors fail to activate BK Ca , whereas the C2 CFTR correctors activate BK Ca with a relative potency of VX-121 > VX-445 >> VX-659.important, as we could not maintain the filters in VX-445 during the experiment since this would simply recapitulate our demonstrated acute effects, even in the absence of a chronic effect.On the other hand, if VX-445 is readily reversible, any chronic effects may be lost when the filters are bathed in our apical/basolateral solutions.As shown in Figure 6, following stimulation of K + current with VX-445 (10 μM), we carried out 6 complete solution exchanges of the apical and basal chambers (during a break in recording).Following this wash, BK Ca short-circuit currents rapidly returned to pre-potentiated levels.Subsequent addition of paxilline (10 μM) inhibited the remaining current, as above.In 3 separate experiments, washout of VX-445 resulted in a reduction in current averaging 90% ± 6%.Thus, any effects of chronic exposure would be difficult to interpret, using these methods, and were not further pursued.
The simplest explanation for our HBE results is that the C2 CFTR correctors VX-659, VX-445, and VX-121 directly activate apical membrane BK Ca .To assess this, we determined whether VX-445 and VX-121 potentiate BK Ca during whole-cell patchclamp recordings from HEK cells (HEK-BK) heterologously expressing the pore-forming α subunit of BK Ca (αBK Ca ).For these studies, the cell was clamped at -80 mV and pulsed in 20 mV inhibitor XE-991 (10 μM) produced only a modest decrease in K + secretory current induced by forskolin, suggesting that Kv7.X channels are not responsible for the K + secretory current observed.To verify that the initial transient increase in K + current was also due to BK Ca activation, we used paxilline to inhibit the baseline K + current induced by amiloride (Figure 5B), thereby validating that the inward current revealed by amiloride block of Na + absorption was indeed due to K + secretion.As shown in Figure 5B, preaddition of paxilline completely eliminated the forskolin response, verifying that both the peak and plateau currents were due to BK Ca activation.Finally, we determined whether the VX-445 or forskolin responses were affected by the prior addition of the other compound.As shown in Figure 5C, VX-445 (10 μM) stimulated a further increase in K + current subsequent to forskolin.However, in comparison with preaddition of VX-445 (Figure 5D), this response was decreased in magnitude (Figure 5E).Similarly, while prior addition of VX-445 (10 μM) did not affect the peak response to forskolin, the extent of the plateau phase was reduced (Figure 5, D and E).
We next considered whether chronic exposure to VX-445 would affect BK Ca currents.First, however, we determined whether the effect of VX-445 was readily reversible.This is increments to +80 mV.As shown for a single cell (Figure 7, A-C), following establishment of a stable baseline (Figure 7A), VX-445 (5 μM) induced a significant increase in outward current (Figure 7B) that was completely inhibited by paxilline (Figure 7C).Similar studies were carried out using VX-121.As shown for a single cell (Figure 7, D-F), VX-121 (5 μM) potentiated αBK Ca currents, which were blocked by paxilline.The average current-voltage (I-V) relationships for control and 1 μM, 5 μM, and 10 μM VX-445, as well as paxilline, are shown in Figure 7G.Both 5 and 10 μM VX-445 significantly increased BK Ca current density across the range of voltages tested.The average I-V relationships for 1 μM, 5 μM, and 10 μM VX-121 are shown in Figure 7H.Similarly to VX-445, our results show that 5 and 10 μM VX-121 significantly potentiated αBK Ca current density.
To demonstrate that the potentiation of αBK Ca by VX-445 during whole-cell recording is a direct effect on the channel, we performed excised, inside-out patch-clamp recordings on HEK-BK cells.As shown in Figure 8A, excised patches often contained many αBK Ca channels.In this case, mean currents were determined in the absence and presence of VX-445.As shown in Figure 8A, 1 μM VX-445 produced a small increase in total current, and this was further increased by 10 μM VX-445.Channel activity was dramatically reduced following perfusion of 0 Ca 2+ .As shown in Fig- ure 8B, both 1 and 10 μM VX-445 produced a significant increase in mean current (P < 0.05).In additional inside-out patches, small numbers of channels were observed (<5), such that individual opening and closing events could be realized, thereby allowing us to determine both single-channel current amplitude (i) and open probability (P o ).As shown for one experiment in Figure 8C, both 1 and 10 μM VX-445 increased channel P o , as evidenced by the increased frequency of discrete channel opening events (Figure 8, C and D).The average change in P o for 4 patches is shown in Figure 8H, with 10 μM VX-445 inducing a 10-fold increase in P o .All-point histograms for the recording shown in Figure 8C during control (Figure 8E) and 1 μM (Figure 8F) and 10 μM (Figure 8G) VX-445 demonstrated no significant change in i (Figure 8I).In total, our patch-clamp studies demonstrate that the current SOC C2 corrector, VX-445, directly potentiates αBK Ca via an increase in P o , likely explaining the K + secretion observed across HBEs.
Having verified potentiation of αBK Ca in HEK-BK cells, we determined whether C2 correctors potentiate BK Ca currents in primary undifferentiated, nonpolarized HBEs via whole-cell patch clamp.As shown in Figure 9A, HBEs exhibited an outwardly rectified current that was potentiated by VX-445 (10 μM; Figure 9B) and subsequently completely inhibited by paxilline (Figure 9C), verifying expression of BK Ca and potentiation by VX-445.To further validate the identity of these VX-445-potentiated currents, we used another canonical blocker of the BK Ca channel, IBTX.As shown in Figure 9F, IBTX inhibited the VX-445-potentiated K + currents (Figure 9, D and E).The average I-V relationships for 9 experiments are shown in Figure 9G.Interestingly, some primary HBEs, surveyed by whole-cell patch clamp, had low-level BK Ca expression, in which single-channel activity was observed (Figure 9, H and I).In Figure 9H, under control conditions (left array of traces), little or no channel activity was observed at +20 or +40 mV, while at +60 mV clear channel activity was seen.This is consistent with the voltage dependence of BK Ca channels.However, in the presence of VX-445 (10 μM; right array of traces) multiple channel openings were observed at all voltages.Paxilline completely abrogated channel activity in this cell (Figure 9H, right bottom trace).In a separate cell (Figure 9I), no channel activity was observed at +40, +60, or +80 mV under control conditions (left array of traces).Subsequent addition of VX-445 (10 μM) induced individual channel events, which were silenced with 300 nM IBTX.These data validate BK Ca potentiation in HBEs, which consist of several cell types that can potentially express BK Ca (43,44), including CFTR-expressing ionocytes (45).
Our data demonstrate that C2 CFTR correctors directly potentiate BK Ca channels, resulting in K + secretion across WT and F508del CFTR-expressing HBEs.As BK Ca is widely expressed throughout the body (46,47), it is important to determine whether the current SOC C2 corrector, VX-445, modulates the function of additional tissues where BK Ca is expressed.Initially, we determined the effect of VX-445 on vascular reactivity, as activation of BK Ca hyperpolarizes vascular smooth muscle, resulting in vasorelaxation (48,49).To assess the effect of VX-445 on vasoreactivity, mouse mesenteric arteries were preconstricted with the prostaglandin mimetic U46619 (1 × 10 −7 to 5 × 10 −7 M), and the ability of VX-445 to induce vasorelaxation was assessed.As shown in Figure 10A for a single mesenteric artery, following U46619-induced vasoconstriction, VX-445 induced vasorelaxation in a concentration-dependent manner.The average response is shown in Figure 10B (blue), with near-complete vasorelaxation achieved at 10 μM VX-445.This effect was partially attenuated by paxilline (Figure 10B, red), demonstrating VX-445 alters vasoreactivity in a BK Ca -dependent manner.
In the nervous system, BK Ca channels are a key component of the fast afterhyperpolarization, which is an important contributor to neuronal firing frequency (50,51).Indeed, channelopathies involving both gain and loss of function of BK Ca have been reported (52,53).Thus, we determined whether VX-445 and VX-121 alter neuronal excitability.To accomplish this, current-clamp patchclamp recordings were performed on primary E18 rat hippocampal and cortical neurons, and action potential firing frequency was monitored.As shown for 2 separate recordings from spontaneously firing hippocampal neurons (Figure 11, A and B), VX-445 induced a concentration-dependent decrease in action potential firing frequency, which was reversible upon washout.Indeed, as shown in Figure 11B, action potential firing frequency could be repeatedly inhibited by 10 μM VX-445.The average changes in firing frequency (in hertz) for 2.5 μM (n = 5), 5 μM (n = 9), and 10 μM (n = 5) VX-445 are shown in Figure 11C.As shown in Figure 11D, VX-121 (5 μM) similarly reduced action potential firing frequency in a primary hippocampal neuron, with the average change for 5 experiments shown in Figure 11E.Finally, we determined whether VX-445 would similarly alter the action potential firing frequency in primary cortical neurons to begin to assess the generalizability of our results.In contrast to hippocampal neurons, current injection was required to induce action potential firing in cortical neurons, under our recording conditions.As shown in Figure 11F, after current injection (delineated by the step change in voltage at the initiation of the trace), action potentials were observed.Subsequent addition of VX-445 (2.5 and 5 μM) induced a significant reduction in action potential firing frequency that was poorly washed out.The average change in firing frequency for 11 separate neurons is shown in Figure 11G.These results clearly demonstrate that VX-445 and VX-121 alter neuronal excitability, the implications of which are discussed below.

Discussion
Over the past 15 years, the most significant advancement in CF therapeutics has been the development of CFTR potentiators and correctors (54)(55)(56).ETI has proven to be highly efficacious for patients with CF, resulting in diminished morbidity as well as vast improvements to their quality of life (18,19).Key components of ETI are the C1 (VX-661; tezacaftor) and C2 (VX-445; elexacaftor) CFTR correctors, which partially restore the folding and hence trafficking of misfolded CFTR to the apical plasma membrane (15).VX-445 has also been shown to potentiate CFTR (57,58).These C1 and C2 correctors exhibit distinct binding sites on CFTR, resulting in unique mechanisms of action (10,11,15,55).Importantly, we demonstrate that C1 correctors fail to modulate BK Ca activity (Figure 1D and Figure 4, A, B, and E), whereas C2 correctors strongly potentiate BK Ca , suggesting a binding site on BK Ca .VX-445 and VX-659 were developed simultaneously by Vertex Pharmaceuticals and have a similar scaffold (59), whereas VX-121 is notably distinct.Our results demonstrate that VX-121 is a more potent potentiator of BK Ca than VX-445 and VX-659 in HBEs, and the potentiation response exhibits a faster onset (Figures 1-4).Interestingly, as clinical development has improved the efficacy of C2 correctors to rescue misfolded CFTR, so has the potency to activate the BK Ca channel (VX-121 > VX-445 >> VX-659).
We recently demonstrated that potentiation of basolateral KCa3.l by DCEBIO (5,6-dichloro-1-ethyl-1,3-dihydro-2H-benzimidazol-2-one) further stimulates transepithelial Cl -secretion across VX-445/VX-661-corrected primary F508del HBEs (35).This indicates the potential for a K + conductance to be rate-limiting for Cl -secretion.It has been proposed that potentiation of BK Ca would similarly increase the electrochemical driving force for Cl -exit across the apical membrane, such that BK Ca may be an alternate Figure 6.Effect of VX-445 on K + current can be completely washed out in Ussing chambers.Following stimulation of K + current by VX-445 (10 μM), both membranes were washed via 6 complete solution exchanges (noted by a break in recording).Following washout, the K + current rapidly returned toward the pre-VX-445 current level.Subsequent addition of paxilline (10 μM) completely inhibited the remaining current.In 3 experiments, the average reduction in K + current following washout of VX-445 was 90% ± 6%.additional clinical benefit.As CFTR and BK Ca are both regulated by cAMP/PKA (Figure 5), it seems likely that these channels act in a concerted fashion to stimulate KCl secretion.In this way, the positive regulation of BK Ca by VX-445 may help to negate the dehydrating effects of inflammation or CFRD at the air-liquid interface by increasing the P o of available plasma membrane BK Ca .However, as reported in preliminary studies (64), our attempts to directly determine whether VX-445 potentiates cAMP-mediated Cl -secretion across HBEs paradoxically demonstrated that VX-445 inhibits forskolin-mediated Cl -secretion across both WT and F508del CFTR HBEs.This inhibition of Cl -secretion was a result of VX-445 directly inhibiting the basolateral membrane KCa3.1 channel (our unpublished observations).Thus, any effects of BK Ca potentiation may be masked by this additional effect on KCa3.1.
Our data demonstrate that the current SOC C2 corrector, VX-445, and the next-generation C2 CFTR corrector, VX-121, potentiate BK Ca in the low-micromolar range (1-10 μM).Thus, pharmacological target in CF (26).Indeed, BK Ca is important in maintaining ASL volume in the airway (26,27).In airway epithelia, BK Ca coassembles with its γ subunit (gene symbol: LRRC26) (60), as well the β 2 /β 4 subunits.The cytokines IFN-γ and TGF-β, which are elevated in CF, both reduce functional apical expression of BK Ca through downregulation of LRRC26 (27,28).Furthermore, CF-related diabetes mellitus (CFRD) has been linked to a reduced expression of LRRC26 (61), suggesting that one possible explanation for the worse lung outcomes observed in CF patients with CFRD ( 62) is a reduction in apical BK Ca expression.Previous modeling by Sandefur and colleagues (63) suggested a role for apical BK Ca in K + secretion, predicting an increased K + secretion across the CF airway to balance Na + absorption.In this regard, it has been shown that K + is elevated above plasma in the ASL, being about 20-30 mEq, suggesting that human airways secrete K + (6,42).
Given the role BK Ca plays in maintaining ASL volume, it is possible that the previously unrecognized ability of C2 CFTR correctors to potentiate BK Ca may be providing CF HEMTs with  Both symptomatic and asymptomatic hypertension has been a feature of clinical trials involving CFTR correctors, and in some cases therapy has been discontinued as a result (31,32).In addition, individuals expressing the F508del allele have reported headache on ETI, which has also been reported in clinical trials of the clinically evaluated BK Ca agonist BMS-204352 (73)(74)(75)(76).As VX-445 alters vasoreactivity in our experiments, the effects of VX-445 on BK Ca could potentially explain these AEs.However, it should be noted that we observed only a partial reversal of this effect with paxilline, suggesting that other ion channels may play a role in the observed effect.For example, in preliminary studies (64), we demonstrate that VX-445 inhibits members of the KCNN gene family, including KCa3.1 and KCa2.x channels.These channels are known to play a critical role in maintaining vascular reactivity (77,78).Further, CFTR mutations have also been shown to affect smooth muscle contractility (79,80), suggesting that the known potentiation effect of VX-445 on CFTR may also play a role in the effects observed.While we demonstrate a direct it is important to consider whether these are clinically relevant concentrations in patients with CF.In this regard, clinical trial data have shown maximal (C max ) and minimal (C min ) plasma elexacaftor (VX-445) concentrations of 8.4-9.2 μg/mL (about 15 μM) and 4.0-5.4μg/mL (about 6-9 μM), respectively (38,65,66).More recently, mean concentrations of VX-445 in cell lysates from CF patient nasal brushings have been found to range from 0 to 5,454 ng/mL (9 μM) (67).Thus, the low-micromolar effects we observe in HBEs and microvascular arteries may be directly relevant in a clinical setting.However, to our knowledge, the concentrations of CFTR correctors achieved in the brain are unknown.Given our results demonstrating effects of VX-445 and VX-121 on action potential firing frequency, this is an important unresolved question.
While the current HEMT has proven to be highly efficacious for improving lung function as well as quality of life for most patients with CF, it is important to point out that in a subset of patients, adverse events (AEs) have been reported (31,(68)(69)(70)(71). Going forward, it may be both prudent and challenging to design the next class of CFTR correctors to avoid activation of the BK Ca channel.In addition to being regulated by Ca 2+ and voltage, the pore-forming α subunit of BK Ca can also associate with 4 unique β subunits (β 1 -β 4 ) as well as 4 γ subunits (γ 1 -γ 4 ).Indeed, the association of αBK Ca with β 2 /β 4 and γ 1 in HBEs results in the functional apical membrane K + channel (26,27).Further, the BK Ca α subunit exhibits alternate splicing.In this regard, the stress axisregulated insert (STREX) variant of BK Ca is expressed by HBEs grown in depleted medium, such that its regulation by phosphorylation is affected (86).In this study, we verified activity of VX-445 and VX-121 on the α subunit of BK Ca .However, it remains to be determined whether regulatory subunits or splice variants play a role.With regard to accessory subunits, the β 1 subunit is highly expressed in smooth muscle (87).In the CNS, expression of auxiliary subunits β 1 , β 2 , β 3b , β 3c , β 3d , γ 1 , γ 3 , and γ 4 has been reported (60).We demonstrate effects of C2 CFTR correctors on lung, arterial, and brain tissue, a tissue set encompassing most BK Ca auxiliary subunits.Thus, while we cannot say how auxiliary subunits might affect the response of BK Ca to CFTR correctors, we are confident that the presence of these subunits does not preclude this effect.
CFTR HEMTs represent a critical component of the CF therapeutic armament, allowing patients with CF to live well into adulthood.Herein, we demonstrate, for the first time to our knowledge, that the current SOC C2 CFTR corrector, VX-445, directly potentiates a distinct ion channel other than CFTR -BK Ca .Given the clinical benefit of these lifesaving drugs, we would emphasize 2 major points with respect to cross-reactivity with the BK Ca channel.First, the potentiation of BK Ca by VX-445 and VX-121 in F508del CFTR HBEs may represent a previously unrecognized mechanism of action resulting in clinical benefit due to potentiated K + , and hence Cl -secretion.Second, while there is a correlation between the physiological effects of C2 correctors (Figures 10 and 11) and the AEs reported by patients with CF, a great deal of work remains to link our in vitro data presented herein to the AEs reported clinically.Nevertheless, given our data, the wide expression of BK Ca , and its clear link to disease following over-or underexpression (85), the proposal to target this effect of VX-445 on BK Ca and vascular reactivity, further studies are required to clarify the role of each of these conductances in the overall vascular response.
Critically, mental-status changes, which may result from changes in neuronal activity, have been reported in response to ETI (29-31, 33, 34, 81).Several patients have described their symptoms as "mental fogginess" and reported deficits to several aspects of cognition.In most of these patients, the onset of symptoms appeared 1 month after initiation of treatment.In one study, 2 patients discontinued treatment because of their reported AEs (33).Furthermore, as attempted suicide has been reported in patients on ETI, clinicians have recommended close monitoring (29).As patients with CF are predisposed to anxiety and depression (82), limiting these adverse mental-status effects relating to mood, motivation, or cognition is of paramount clinical importance.A recent meta-analysis of ETI trials failed to show a causal relationship between ETI therapy and depression-related symptoms, which the authors concluded were commensurate with the background epidemiology of patients with CF (82).These findings are not uncontested, however, and currently the implications of ETI therapy for mental-status AEs are highly disputed (83) and deserving of serious research effort (84).
While it has been speculated that CFTR correctors and/or potentiators may affect neuronal function (30,31), this has not been directly demonstrated.As noted, BK Ca channels play a critical role in regulating action potential firing via the fast afterhyperpolarization (51).Either pharmacological or genetic manipulation of BK Ca function dramatically affects action potentials (51,85).Herein, we demonstrate, for the first time to our knowledge, that the C2 CFTR correctors VX-445 and VX-121 directly affect action potential firing frequency in primary cultures of hippocampal and cortical neurons (Figure 11).Whether this effect of CFTR correctors can account for the mental-status changes reported by people with CF on ETI is purely a matter of speculation.However, given the critical role BK Ca plays in neuronal action potential firing, we believe our results lay the critical foundation for future studies designed to further test this hypothesis.

Methods
Sex as a biological variable.Sex was not considered a variable in this work, as the authors were unaware of the sex of human donor tissue.
Cell culture.HEK293 cells expressing αBK Ca were provided by Heike Wulff (University of California, Davis, Davis, California, USA) and cultured as in our previous work (90).
Primary HBEs were provided by the University of Pittsburgh human airway cells and tissue core and cultured using the Vertex method (91).Our studies were carried out on WT CFTR HBEs from 9 donors and homozygous F508del/F508del CFTR HBEs from 6 channel in airway epithelia may be fraught with difficulties.Finally, given that VX-445 and VX-121 represent novel BK Ca activators, these compounds could prove useful in dissecting the role of BK Ca in a host of settings.Indeed, as VX-445 is an FDA-approved drug, it may be useful in a host of conditions in which BK Ca potentiators have been proposed to have therapeutic utility (53, 73-76, 88, 89).Therefore, we feel it prudent, in the iterative spirit of drug development, that the structure of existing C2 correctors be interrogated to identify the precise pharmacophore responsible for BK Ca versus CFTR regulation, in order to design more selective compounds for CF.O 2 /5% CO 2 at 37°C.MAs were mounted on a wire myograph (Multiple Myograph Model 620 M, Danish Myotechnology) using 25 μm-diameter tungsten wire and allowed to rest for 30 minutes in PSS.MAs were incrementally stretched to a tension equivalent to 80 mmHg of physiological pressure.MA viability was tested via the addition of potassium (60 mM) in PSS for 5 minutes, followed by 3 washes with PSS.After a 30-minute rest period, MAs were incubated for 15 minutes with either paxilline (10 μM) or control buffer (0.1% DMSO).Subsequently, MAs were constricted with the prostaglandin mimetic U46619 (1 × 10 −7 to 5 × 10 −7 M) for 4 minutes per concentration to induce maximal constriction before vasodilator treatment.After vessels reached maximal constriction, a cumulative concentration-response curve for VX-445 was conducted (1, 2.5, 5, 7.5, and 10 μM for 15 minutes per concentration).Subsequently, Ca 2+ -free PSS containing 1 × 10 −6 sodium nitroprusside was added to determine maximal relaxation.Data were recorded on Lab Chart Software (AD Instruments), and relaxation percentage was normalized to the change in maximal constriction via U46619 and maximal dilation via Ca 2+ -free PSS.
For patch-clamp studies, an Axon 200B amplifier (Axon Instruments) in conjunction with Clampex data acquisition software (version 9.2, Axon Instruments) was used to capture recordings with lowpass Bessel filtering set at 2 kHz and a digitization rate of 10 kHz (1.48 MB/min).Recordings were analyzed using the relevant tools within Clampfit (version 9.2, Axon Instruments).Borosilicate glass electrodes (1.65 mm outer diameter; World Precision Instruments) were pulled with a Narishige puller (model PP-830).After fire polishing with a World Precision Instruments microforge (MF-200), pipettes had a resistance of 2-3 MΩ.A peristaltic pump (Minipuls 3, Gilson) was used to continuously perfuse bath solution (2.3 mL/min) to which various pharmacological agents were added.I-V relationships were determined via a pulse protocol involving a 400-millisecond voltage pulse from -80 to +80 mV flanked by a 50-millisecond pulse to a holding potential of -80 mV.The interpulse interval was 100 milliseconds.The middle 300 milliseconds of each trace was selected to avoid any influence of capacitive transients, and average currents were calculated for each voltage via the I-V tool in Clampfit.To control for cell size, whole-cell currents were normalized to cell capacitance.
Ussing chamber short-circuit current measurements.Costar Transwell inserts were mounted in a modified Ussing chamber (P2300, Physiologic Instruments) and the monolayers continuously short-circuited (VCC MC8, Physiologic Instruments) by forcing of the transepithelial voltage to 0 mV.R te was monitored by application of a 2 mV pulse every 90 seconds.For measurements of transepithelial K + secretion (I K ) the basolateral solution contained (in mM): 120 K-gluconate, 25 NaHCO 3 , 3.3 KH 2 PO 4 , 0.8 K 2 HPO 4 , 1.2 MgCl 2 , 4 CaCl 2 , and 10 glucose.For the apical solution, Na-gluconate was substituted for K-gluconate, thereby creating a 125:5 mM basolateral-to-apical K + concentration gradient.CaCl 2 was used at 4 mM to account for the Ca 2+ buffering capacity of gluconate.The pH of the solution is 7.4 when gassed with 95% O 2 /5% CO 2 .In a subset of experiments, the basolateral membrane was permeabilized with nystatin (200 μM), as previously described (37).The limited Cl -in these solutions (~10 mEq) relieves cell swelling associated with the limited permeability of the nystatin pore to Cl - (92).
Experiments were carried out at 37°C, and compounds were added cumulatively following establishment of a new stable current response.Paxilline, iberiotoxin, and amiloride were added to the apical membrane, while all other compounds were added to both membranes owing to their lipophilic nature.In all experiments, 10 μM amiloride was used to inhibit sodium absorption.ΔI K was calculated as the difference between the baseline current after amiloride inhibition and the peak response to the agonist.
Rat hippocampal and cortical neuron dissection and primary neuron culturing.Hippocampal and cortical neurons were dissected from E18 Long-Evans rat embryos as previously described (93).Dissociated neurons (1 × 10 5 cells per well) were plated on acid-washed 12 mm coverslips coated overnight with poly-d-lysine (high molecular weight 20 mg/mL) and laminin (3.4 mg/mL).Neurons were cultured in Neurobasal medium (Invitrogen) supplemented with 2% B27 (Invitrogen), penicillin and streptomycin (100 U/mL and 100 mg/mL, respectively), and 2 mM glutamine.For hippocampal neurons, 40% of the medium was replaced every 4 days, whereas for cortical neurons, 50% of the medium was replaced daily.Patch-clamp studies were carried out 1-3 weeks after plating.All animal studies were approved by the University of Pittsburgh Institutional Animal Care and Use Committee (protocol 22051190).

Figure 1 .
Figure 1.VX-445, VX-659, and VX-121 stimulate BK Ca currents across WT CFTR HBEs.Currents were recorded with a 125:5 mM K + gradient (basolateral to apical).(A-C) Subsequent to amiloride, short-circuit current (I sc ) was increased by the C2 CFTR correctors VX-445 (A, 10 μM), VX-659 (B, 10 μM), and VX-121 (C, 10 μM).(D and E) In contrast, the C1 CFTR corrector VX-661 (D, 10 μM) and the CFTR potentiator VX-770 (E, 10 μM) failed to increase I sc .Subsequent addition of either VX-445 (D) or VX-121 (E) stimulated I sc .In all experiments, the current was completely blocked by the specific BK Ca inhibitor paxilline (10 μM).(F and G) Additional studies verified that both the VX-445-induced (F) and the VX-121-induced (G) currents were inhibited by the additional specific BK Ca blocker IBTX (300 nM).(H) Subsequent to amiloride, 10 μM NS1619 failed to stimulate I sc , while further addition of 50 μM NS1619 induced a marked increase.This response was further increased by VX-445 (10 μM) and inhibited by paxilline.(I) Average responses (mean ± SEM, *P < 0.01; 1-way ANOVA) are represented as the change in K + current (ΔI K ).The magnitude of the K + current was calculated as described in Methods.Nine donors were used in these studies.Experimental replicates are indicated in parentheses above each data set.

Figure 2 .
Figure 2. VX-445 and VX-121 stimulate BK Ca currents across WT CFTR HBEs following permeabilization of the basolateral membrane with nystatin.Currents were recorded with a 125:5 mM K + gradient (basolateral to apical).Subsequent to amiloride, nystatin (200 μM) was added to the basolateral membrane.Following establishment of a new stable current, both VX-445 (A, 10 μM) and VX-121 (B, 10 μM) stimulated an increase in I K .(C) Average ΔI K (mean ± SEM) for VX-445 and VX-121 (*P < 0.01; unpaired t test).ΔI K was calculated as the change in I K between the current in the presence of nystatin and the peak response to VX-445 or VX-121.Three donors were used in these studies.Experimental replicates are indicated in parentheses above each data set.

Figure 3 .
Figure 3. Concentration dependence of VX-445 and VX-121 stimulation of K + current across WT CFTR HBEs.(A) Subsequent to amiloride, VX-445 induced a concentration-dependent increase in K + current that was inhibited by paxilline.(B) Subsequent to amiloride, VX-121 induced a concentration-dependent increase in K + current that was inhibited by paxilline.(C) Average increase in K + current for each concentration of VX-121 and VX-445 (mean ± SEM).Note that for VX-445 there are different numbers of experiments for each concentration, as not all concentrations were used in each experiment.(D) Data from C for VX-121 were fit to the Hill equation, giving an EC 50 of 4.4 μM with a Hill coefficient of 3.0 (R 2 = 0.8).Three donors were used in these studies.Experimental replicates are indicated in parentheses above each data set.

Figure 5 .
Figure 5.Effect of forskolin and VX-445 on BK Ca currents in F508del CFTR HBEs.(A) Subsequent to amiloride, forskolin (10 μM) stimulated a paxilline-sensitive increase in K + secretion, which is recognized as an initial transient spike followed by a sustained increase in I sc .Addition of XE-991 (10 μM) had little effect on the remaining current.(B) Subsequent to amiloride and paxilline, forskolin (10 μM) failed to stimulate K + secretion.(C) Following forskolin, VX-445 (10 μM) induced a further increase in K + secretion, which was paxilline sensitive.(D) VX-445 (10 μM) stimulated a sustained increase in K + secretory current, which was further increased by forskolin.Subsequent addition of paxilline completely inhibited this K + secretory current.(E) Average responses (mean ± SEM) to forskolin and VX-445 either alone or after addition of the previous agonist ( # P < 0.01; unpaired ANOVA).Three donors were used in these studies.Experimental replicates are indicated in parentheses above each data set.

Figure 8 .
Figure 8.Effect of VX-445 in excised patches.Excised patch-clamp recordings were carried out in symmetric K + and voltage-clamped to +40 mV.(A) Both 1 μM and 10 μM VX-445 increased current in a patch expressing many channels.The current was reversed by addition of 0 Ca 2+ .(B) Average mean current (pA, mean ± SEM) for excised patches containing numerous BK Ca channels for control (n = 5) and 1 μM (n = 4) and 10 μM (n = 5) VX-445.(C) Both 1 μM and 10 μM VX-445 increased current in a patch expressing αBK Ca such that individual channel openings could be observed.(D) Expanded view of trace shown in C, where individual BK Ca channel open and closed events can be observed in magnified current traces corresponding to the parent trace (dashed lines and i, ii, and iii).(E-G) All-point histograms from the recording shown in C such that single-channel current amplitude can be determined, as indicated by the distance between the dotted lines placed at the peak of each curve, which represent a given open state.(H) Average P o (mean ± SEM) calculated for 4 experiments, as shown in C. (I) Average single-channel amplitude for 4 experiments (*P < 0.05; paired ANOVA).

Figure 10 .
Figure 10.Effect of VX-445 on vasoreactivity in mouse mesenteric artery.(A) Recording of force in millinewtons (mN) over time from a single mesenteric artery showing preconstriction with the prostaglandin mimetic U46619 (1 × 10 −7 to 5 × 10 −7 M), after which the ability of increasing concentrations of VX-445 to induce vasorelaxation was assessed.We added 0 Ca 2+ at the end to determine maximal vasorelaxation.(B) Average responses to VX-445 under control conditions (blue line, n = 10) and following preincubation with paxilline (10 μM) for 15 minutes (red line, n = 6).The effect of VX-445 was partially reversed by paxilline, verifying a role for BK Ca .*Statistical difference between VX-445 and VX-445 + paxilline by 2-way ANOVA with P < 0.003 by post hoc Holm-Šidák multiple-comparison test.Data are shown as mean ± SEM.