Cyclic Peptidyl Inhibitors against CAL-CFTR Interaction for Treatment of Cystic Fibrosis

Abstract

Cystic fibrosis (CF) is caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene, encoding for a chloride ion channel. Membrane expression of CFTR is negatively regulated by CFTR-associated ligand (CAL). We previously showed that inhibition of the CFTR/CAL interaction with a cell-permeable peptide improves function of rescued F508del-CFTR. In this study, optimization of the peptidyl inhibitor yielded PGD97, which exhibits a K_D value of 6 nM for the CAL PDZ domain, ≥ 130-fold selectivity over closely related PDZ domains, and a serum t_1/2 of >24 h. In patient-derived F508del homozygous cells, PGD97 (100 nM) increased short-circuit currents by ~3-fold and further potentiated the therapeutic effects of small-molecule correctors (e.g., VX-661) by ~2-fold (with an EC₅₀ of ~10 nM). Our results suggest that PGD97 may be used as a novel treatment for CF, either as a single agent or in combination with small-molecule correctors/potentiators.

Graphical Abstract

INTRODUCTION

Cystic fibrosis (CF) is a devastating autosomal recessive disorder which is the most common genetic disease amongst Caucasians, with a prevalence of approximately one in 3400 births in the United States.¹ The molecular pathology of CF can be traced to mutations in the CFTR gene, which encodes for the cystic fibrosis transmembrane conductance regulator (CFTR), a cAMP-regulated transmembrane Cl⁻ ion channel.² In the thirty years since the identification of the CFTR gene locus,^3–5 over 2000 different mutations within the CFTR gene have been uncovered, each of which confers a distinct phenotype with variable disease severity.⁶ The CFTR mutations that give rise to CF have been classified into six categories, including mutations that produce destabilized, misfolded CFTR (Class II), those that cause dysregulated channel gating with reduced Cl⁻ channel open time (Class III), and those that reduce the channel conductivity (Class IV).⁷ While CF manifests as a multi-organ, systemic pathology, the primary cause of mortality stems from reduced ion channel activity in lung epithelia. Impaired clearance of normal exocrine secretions leads to the deposition of a thick, neutrophil-rich mucus that promotes chronic bacterial infections and inflammation. Mucosal obstructions in other organs can present as pancreatitis,⁸ gastro-intestinal issues,⁹ and cirrhosis¹⁰ resulting in increased morbidity.

The severity and life-altering nature of CF has made the search for therapeutic interventions of the utmost importance. Symptomatic treatment options, such as antibiotics and inhaled corticosteroids, are helpful but do not address the underlying pathophysiology.¹¹ Direct modulation of aberrant CFTR biochemical function has been challenging because CFTR mutants are nontraditional drug targets and highly heterogeneous; each therapeutic is likely to be only effective against a specific class of CFTR mutations. To date, three different classes of modalities have been developed to rescue the function of CFTR mutants, namely potentiators, correctors, and stabilizers (Figure 1a). Potentiators, as exemplified by the small molecule Ivacaftor (VX-770),¹² act by increasing the probability for the ion channel to stay in the open state and are most effective against Class III mutations (e.g., G551D). Correctors such as Tezacaftor (VX-661)¹³ and the newly FDA approved Elexacaftor (VX-445)¹⁴ function to promote proper folding and stabilize Class II CFTR mutants (e.g., F508del). They generally lack single-agent efficacy and must be used in combination with potentiators. The third class, stabilizers, function to increase the membrane expression of CFTR by preventing it from lysosomal trafficking and degradation.¹⁵ Stabilizers represent a very attractive modality for the treatment of CF as they should be effective across the entire CFTR mutational landscape and synergize with correctors and/or potentiators. Unfortunately, such a compound is not currently available to CF patients.

Figure 1.

(a) Intracellular trafficking, endocytosis, and recycling of CFTR among the different membranous compartments and the effect of correctors (1), potentiators (2), and stabilizers (3) on mutant CFTR biogenesis and function. (b) Design strategy for disulfide cyclized CPP-cargo conjugates which have improved proteolytic stability and cell-permeability when outside the cell, but are converted into the linear, biologically active form inside the cytosol/nucleus upon reduction by intracellular thiols. (c) Structure of peptide 1.

The most common mutation, corresponding to a deletion of phenylalanine 508 (F508del), is present in an estimated 90% of CF patients.¹⁶ As a Class II mutant, F508del-CFTR is conformationally unstable and upon rescue subject to rapid lysosomal degradation. While combination therapies of correctors and potentiators have substantially improved lung function of CF patients with one or two F508del alleles, they cannot address the underlying issue of rapid membrane turnover conferred by the F508del mutation. The membrane expression of CFTR is controlled through interactions between its C-terminus and two different classes of PDZ domain-containing proteins.^17,18 While binding to PDZ domain proteins Na⁺/H⁺-exchanger regulatory factor isoform-1 and 2 (NHERF-1 and 2) stabilizes CFTR on the plasma membrane,^19,20 interaction with CFTR-associated ligand (CAL) targets CFTR for lysosomal degradation (Figure 1a).^21,22 These observations suggest that selective inhibitors against the CAL PDZ-CFTR interaction should provide a novel class of CFTR stabilizers. This approach was initially validated by the discovery of a peptidyl inhibitor of the CAL PDZ domain, iCAL36, which enhanced CFTR membrane half-life in cultured airway epithelial cells.²³ Since then, a number of other peptidyl inhibitors have been reported, with CAL binding affinities typically in the low μM to high nM range.²⁴ There have also been efforts in developing covalent small-molecule inhibitors of this interaction.²⁵ Peptides are well suited to target the CAL-CFTR interaction because of their potential for high affinity and specificity, but they generally lack cell-permeability and metabolic stability. To overcome these limitations, we previously designed a cell-permeable, reversibly cyclized peptidyl inhibitor, Ac-cyclo(CRRRRFWQC)TRV (peptide 1, Table 1), by fusing a cell-penetrating peptide (CPP) with a CAL PDZ domain-binding sequence (Figure 1b,c).²⁶ When outside the cell, peptide 1 exists as a disulfide-cyclized macrocycle, which has improved proteolytic stability and cell-permeability. Upon entering the cell, peptide 1 is converted into the linear, active form by intracellular thiols [e.g., glutathione (GSH)]. Peptide 1 increased the ion channel activity of F508del-CFTR in CF bronchial epithelial (CFBE) cells by 77%, but required high concentrations (50 μM) owning to its modest binding affinity for the CAL PDZ domain (K_D = 0.5 μM), limited cell-permeability, as well as proteolytic stability (serum t_1/2 ~5 h). In this work, we carried out a modeling-guided medicinal chemistry campaign and greatly improved the potency, specificity, cell-permeability, and metabolic stability of the parent peptide. The optimized inhibitor greatly improved the ion channel activity of F508del-CFTR in CFBE cell line and CF patient-derived primary lung epithelial cells at low nanomolar concentrations.

Table 1.

Sequences of peptides used in this work

Peptide	Sequence[a]	IC50 (μM)	Cellular Entry (%)[d]
1	Ac-cyclo(Cys-Arg-Arg-Arg-Arg-Phe-Trp-Gln-Cys)-Thr-Arg-Val-OH	ND	ND
2	Ac-Arg-Arg-Phe-Trp-Gln-Cys-Thr-Arg-Val-OH	4.3 ± 0.2[b]	ND
3	Ac-Arg-Arg-Phe-Trp-Gln-Cys-Thr-Arg-Ile-OH	3.7 ± 0.5[b]	ND
4	Ac-Arg-Arg-Phe-Trp-Gln-Cys-Thr-Arg-Tle-OH	1.6 ± 0.2[b]	ND
5	Ac-Arg-Arg-Phe-Trp-Gln-Cys-Thr-Arg-Cha-OH	7.9 ± 0.7[b]	ND
6	Ac-Arg-Arg-Phe-Trp-Gln-Cys-Thr-hArg-Tle-OH	2.4 ± 0.1[b]	ND
7	Ac-Arg-Arg-Phe-Trp-Gln-Cys-Thr-Lys(Nic)-Tle-OH	2.2 ± 0.1[b]	ND
8	Ac-Arg-Arg-Phe-Trp-Gln-Cys-Thr-Lys(Tfa)-Tle-OH	2.5 ± 0.3[b]	ND
9	Ac-Arg-Arg-Phe-Trp-D-Pro-Cys-Thr-Arg-Tle-OH	7.4 ± 0.4[b]	ND
10	Ac-Arg-Arg-Phe-Trp-D-Cys-Cys-Thr-Arg-Tle-OH	10 ± 2[b]	ND
11	Ac-Arg-Arg-Phe-Trp-Cys-Cys-Thr-Arg-Tle-OH	8.0 ± 3.0[b]	ND
12	Ac-Cys-Arg-Arg-Arg-Phe-(1-Nal)-Gln-Cys-Thr-Arg-Tle-OH	1.8 ± 0.4[c]	ND
13	Ac-Cys-Arg-Arg-Arg-Phe-(2-Nal)-Gln-Cys-Thr-Arg-Tle-OH	1.5 ± 0.3[c]	ND
14	Ac-Cys-Arg-Arg-Arg-Phe-Trp-Gln-Pen-Thr-Arg-Tle-OH	0.68 ± 0.02[c]	ND
15	Ac-Cys-Arg-Arg-Arg-(1-Nal)-Trp-Gln-Pen-Thr-Arg-Tle-OH	0.15 ± 0.03[c]	ND
16	Ac-Cys-Arg-Arg-Arg-(2-Nal)-Trp-Gln-Pen-Thr-Arg-Tle-OH	0.15 ± 0.01[c]	ND
17	Ac-Cys-Arg-Arg-Arg-(D-2-Nal)-Trp-Gln-Pen-Thr-Arg-Tle-OH	0.37 ± 0.03[c]	ND
18	Ac-Cys-Arg-D-Arg-Arg-(2-Nal)-Trp-Gln-Pen-Thr-Arg-Tle-OH	0.40 ± 0.06[c]	17 ± 2
19	Ac-Cys-D-Arg-Arg-Arg-(2-Nal)-Trp-Gln-Pen-Thr-Arg-Tle-OH	0.17 ± 0.02[c]	28 ± 1
20	Ac-Cys-D-Arg-D-Arg-Arg-(2-Nal)-Trp-Gln-Pen-Thr-Arg-Tle-OH	0.27 ± 0.01[c]	23 ± 4
21	Ac-Cys-D-Arg-Arg-Arg-(2-Nal)-Trp-Pen-Tle-Thr-Arg-Tle-OH	0.38 ± 0.03[c]	92 ± 12
22	Ac-Cys-D-Arg-Arg-Arg-(2-Nal)-Bta-Pen-Tle-Thr-Arg-Tle-OH	0.24 ± 0.03[c]	518 ± 74
23	Ac-Cys-D-Arg-Arg-Arg-(2-Nal)-Bta-Pro-Pen-Thr-Arg-Tle-OH	ND	36 ± 20
24	Ac-Cys-D-Arg-Arg-Arg-(2-Nal)-Bta-Pip-Pen-Thr-Arg-Tle-OH	0.020 ± 0.001[c]	61 ± 3

^[a]Ac, acetyl; D-Arg, D-arginine; D-Cys, D-cysteine; 1-Nal, 3-(1-naphthyl)-L-alanine; 2-Nal, 3-(2-naphthyl)-L-alanine; D-2-Nal, 3-(2-naphthyl)-D-alanine; D-Pro, D-proline; Tle, tert-butyl-L-alanine; Pen, L-penicillamine; Bta, 3-(3-benzothienyl)-L-alanine; Cha, 3-cyclohexyl-L-alanine; hArg, L-homoarginine; Lys(NIC), nicotinyl-L-lysine; Lys(TFA), trifluoroacetyl-L-lysine. See Figure S1 for detailed structures.

^[b]Values determined by FP method A.

^[c]Values determined by FP method B.

^[d]All values are for the corresponding disulfide-cyclized peptides labeled at the N-terminus with naphthofluorescein and relative to that of CPP9 (100%). All values reported are mean ± SD of n = 2–4 independent experiments.

RESULTS AND DISCUSSION

Peptide 1 required substantial improvement in four key areas: target binding affinity, selectivity over closely related NHERF1/2 PDZ domains (vide infra), proteolytic stability, and cellular entry efficiency. Additionally, we intended to keep the size of the peptide (i.e. molecular weight) to a minimum, reducing both the cost of production and the number of potential proteolytic sites. This approach necessitates dual use of at least some of the amino acid residues for both target binding and cell penetration. We therefore opted to optimize all four parameters concurrently through iterative cycles of in silico modeling, chemical synthesis, and experimental testing.

Enhancing CAL Binding Affinity.

PDZ domains recognize specific C-terminal sequences (typically the last 3–6 residues) of their partner proteins and require a free C-terminal carboxylate for binding affinity and specificity.²⁷ Docking of a linear version of peptide 1 (peptide 2, Table 1 and Figure S1) into a previously reported crystal structure for the CAL PDZ domain²⁸ revealed that the C-terminal valine (defined as position 0) occupies a hydrophobic pocket, but leaves significant surface area unburied (Figure S2). We replaced Val⁰ with a focused library of nine alternative hydrophobic residues and evaluated them for binding to CAL PDZ domain in silico (Table S1). Peptides containing the top three residues (based on the calculated binding energies), isoleucine (Ile), tert-leucine (Tle), and β-cyclohexylalanine (Cha) were chemically synthesized and experimentally tested for binding using a fluorescence polarization (FP)-based competition assay. Substitution of Tle for Val⁰ increased the binding affinity by 2.7-fold, whereas the other two residues did not significantly improve binding (Table 1, peptides 3–5). Examination of the docked pose suggested that the tert-butyl side chain of Tle interacts with a greater hydrophobic surface area on CAL while maintaining the hydrogen bonding between the peptide C-terminal carboxylate and CAL residues Leu²⁹⁹, Gly³⁰⁰, and Ile³⁰¹ located along the hydrophobic pocket (Figure S2). Tle should have the additional benefit of improving the proteolytic stability against carboxypeptidases and endopeptidases thanks to its bulky tert-butyl side chain. Tle was selected as the P0 residue in all further studies.

At the −1 position, docking studies suggested that substitution of homoarginine (hArg) for Arg might facilitate a cation-π interaction between the guanidinium group and His³¹⁹. However, FP analysis of the resulting peptide (Table 1, peptide 6) showed no gain in binding affinity, likely due to an increased entropic penalty from the additional methylene unit present in hArg. Inspired by previous work using chemically modified Lys residues at the −1 position,²⁹ we also replaced Arg with N^ε-nicotinoyllysine [Lys(NIC)] or N^ε-trifluoroacetyllysine [Lys(TFA)] (Table 1, peptides 7 and 8). Unfortunately, neither substitution improved the CAL PDZ binding affinity. We elected to retain Arg at the −1 position during future rounds of optimization.

Since the Thr at the −2 position is highly conserved and essential for effective CAL engagement while the Cys residue at the −3 position is required for cyclization, we next replaced the Gln at P-4 position with D-proline, D-cysteine, or L-cysteine (Table 1, peptides 9–11). A D-amino acid at this position would increase the proteolytic stability, while D/L-cysteine would provide an alternative site of cyclization. Cyclization at position −4 (instead of P-3) would generate a smaller and more conformationally constrained ring, which is expected to have greater metabolic stability as well as cell-permeability.³⁰ These modifications resulted in significant losses of binding affinity and were not carried forward.

At the −5 position, replacement of Trp with other hydrophobic aromatic residues [1- or 2-naphthylalanine (1- or 2-Nal)] were essentially neutral with respect to CAL binding affinity (Table 1, peptides 12 and 13), so the Trp-Gln motif was retained at this stage.

Upon further examination of the docked pose of peptide 2, we realized that the side chain of Cys at P-3 makes minimal contacts with CAL (Figure S2). Therefore, we substituted L-penicillamine (Pen) for the Cys, anticipating that the gem-dimethyl groups at Cβ would enhance proteolytic stability through steric occlusion of the neighboring backbone amides, yield a more conformationally defined disulfide bond, and/or potentially engage in additional surface contacts with CAL. Satisfyingly, incorporation of Pen resulted in a ~2-fold improvement in CAL binding affinity (Table 1, peptide 14). Pen was retained as the P-3 residue moving forward.

The phenylalanine at position −6 was originally incorporated as a part of the CPP sequence.²⁶ We previously found in another study that replacement of Phe with a larger hydrophobic residue (e.g., 2-Nal) greatly increases the cellular entry efficiency of cyclic CPPs.³⁰ We thus replaced Phe of peptide 14 with 1-Nal, 2-Nal, or D-2-Nal (Table 1, peptides 15–17), with the goal of improving the cell-permeability of the peptidyl inhibitor. Gratifyingly, substitution of 1- or 2-Nal improved the CAL binding affinity by ~5-fold, presumably as a result of increased hydrophobic interaction with CAL. Substitution of D-2-Nal also improved CAL binding, but to a lesser degree (~2-fold). 2-Nal was selected as the P-6 residue in subsequent studies and the corresponding peptide (16) had an IC₅₀ value of 150 nM in the FP-based competition assay (Table 1).

Optimization of Cell-Permeability.

Having obtained an inhibitor of respectable CAL binding affinity, we next set out to optimize the cell-permeability. Cyclic CPPs enter the cell by endocytosis followed by endosomal escape.^31,32 Earlier studies have shown that smaller rings, greater conformational rigidity, and proper stereochemistry of the arginine and hydrophobic residues lead to more efficient cellular entry.³² With these in mind, we reduced the number of arginine residues in the CPP sequence from four (as in peptide 1) to three, to minimize the size of the CPP ring. The arginine at P-8, P-9, or both positions of peptide 16 was (were) then replaced with D-arginine to give peptides 18-20 (Table 1). Incorporation of D-arginine residues should also improve proteolytic stability. We kept the stereochemistry of the P-7 arginine unchanged, as highly efficient cyclic CPPs generally have the same stereochemical configuration for this arginine and the adjacent hydrophobic residue (2-Nal in peptide 16).³² The disulfide-cyclized forms of peptides 18–20, which were designated as peptides 18c-20c (where “c” for “cyclic”), respectively, were labeled at the N-terminal amine with a pH-sensitive dye, naphthofluorescein (NF), and their cytosolic entry efficiencies into HeLa cells were determined by flow cytometry.³³ Among the three peptides, peptide 19c (with a D-arginine at P-9) had the highest cytosolic entry efficiency, which was 28% of that of cyclic CPP9,³² one of the most active cyclic CPPs (Table 1). Peptide 19 also retained most of the CAL binding affinity (IC₅₀ = 0.17 μM), whereas incorporation of D-arginine at the −8 position resulted in significant loss of CAL binding affinity (Table 1, peptides 18 and 20).

To further enhance the cellular entry efficiency of peptide 19c, we replaced Gln at the −4 position with Pen and Pen at the −3 position with Tle and formed a smaller macrocycle (7 aa instead of 8 aa) (Table 1, peptide 21). Indeed, although these changes decreased CAL binding by ~2-fold, the disulfide cyclized peptide 21c showed 4-fold improvement in cell-permeability (92% relative to CPP9). Further replacement of Trp at the −5 position with 3-(3-benzothienyl)-L-alanine (Bta) gave peptide 22, which demonstrated a ~1.5-fold improvement in CAL binding affinity (IC₅₀ = 0.24 μM). The corresponding cyclic peptide (22c) showed a 5-fold improvement in cytosolic entry efficiency. Bta is isosteric with Trp but is more hydrophobic and less prone to oxidative degradation than Trp. Presumably, the more hydrophobic benzothienyl group inserts more deeply into the cell membrane, generating greater membrane binding affinity and cell-permeability. Overall, peptide 22c represents an 18-fold improvement in cell-permeability over peptide 19c with only a small reduction in CAL binding affinity, providing a desirable candidate for cellular studies.

Selective Binding to CAL Is Required for Improving the Function of Rescued F508del-CFTR.

We evaluated peptides 19c and 22c for their efficacy in improving the chloride ion channel activity of primary CF human bronchial epithelial (CF-HBE) cells from a patient homozygous for the F508del mutation.³³ In this assay, CFTR function is assessed by measuring the short-circuit current (I_SC) flowing across differentiated primary CF-HBE cultures grown at air/liquid interface using Ussing chambers.^34,35 CF-HBE cells were incubated with 5 μM of VX-809, a small-molecule corrector,³⁶ and 100 nM peptide before measuring CFTR function. Peptide 19c increased the I_SC value by ~50% relative to VX-809 alone, suggesting effective rescue of F508del-CFTR on the apical membrane (Figure 2). Much to our surprise, peptide 22c had no significant effect on I_SC despite its similar CAL binding affinity and 18-fold higher cellular entry efficiency. We reasoned that the lack of efficacy by peptide 22c might be caused by off-target effects. Specifically, the C-terminus of CFTR interacts with the PDZ domain(s) of several different proteins, including CAL, NHERF1, and NHERF2.²² While binding to CAL targets CFTR for lysosomal degradation, interaction with NHERF1/2 facilitates CFTR trafficking to the plasma membrane and stabilizes CFTR on the membrane. Thus, simultaneous inhibition of NHERF1/2 PDZ domains would offset the beneficial effects of inhibiting the CAL-CFTR interaction. To test this hypothesis, we determined the binding affinities of fluorescein (FAM)-labeled peptides FAM-19 and FAM-22 to CAL PDZ, NHERF1 PDZ domain 1, and NHERF2 PDZ domains 1 and 2 by FP. Our attempt to produce soluble NHERF1 PDZ domain 2 was unsuccessful. Peptide FAM-19 showed modest selectivity for CAL-PDZ (14- to 36-fold) over NHERF1/2 PDZ domains (Table 2). In comparison, peptide FAM-22 exhibited lower selectivity for CAL vs NHERF1/2 domains (6.7- to 57-fold) and significant binding to NHERF1 PDZ1 domain (K_D = 330 nM). These results suggest that selective inhibition of CAL PDZ domain is critical for increasing F508del-CFTR function.

Figure 2.

Comparison of peptides 19c and 22c for their efficacy in increasing CFTR-mediated short-circuit currents in CF-HBE cells (from one donor). The cells CF-HBE cells were treated with 5 μM VX-809 for 48 h, followed by 100 nM peptide for 2 h before measuring short-circuit current (I_sc) in an Ussing chamber. Values are scaled relative to treatment with VX-809 alone (100%) and represent the mean ± SD of 2–3 replicates/condition. *, p <0.05.

Table 2.

Binding affinity of selected peptides to CAL and NHERF1/2 PDZ domains (K_D, nM)^[a]

Peptide	CAL PDZ	NHERF1 PDZ1	NHERF2 PDZ1	NHERF2 PDZ2
FAM-19	113 ± 2	1640 ± 60	1870 ± 20	4070 ± 160
FAM-22	49 ± 3	330 ± 7	560 ± 10	2820 ± 30
FAM-23	87 ± 2	3970 ± 840	6040 ± 870	7660 ± 720
FAM-24	6 ± 3	780 ± 10	1370 ± 380	7480 ± 580

^[a]Values reported represent the mean ± SD of n = 3 independent experiments.

Enhancing Inhibitor Selectivity for CAL PDZ Domain.

Since replacement of Gln⁻⁴-Pen⁻³ with Pen⁻⁴-Tle⁻³ was the primary difference between peptides 19 and 22 and likely the cause of selectivity loss, we reverted back to Gln⁻⁴-Pen⁻³ at these two positions and explored alternative strategies to enhance the inhibitor selectivity. Previous studies had suggested that a Pro at P-4 enhances selectivity for CAL over NHERF1/2 PDZ domains.²⁴ We thus replaced Gln at P-4 with Pro or Pip, producing peptides 23 and 24, respectively (Table 1). Incorporation of proline had relatively minor effect on CAL binding affinity (K_D = 87 nM for peptide FAM-23), but substantially improved its selectivity for CAL over NHERF1/2 PDZ domains (46- to 88-fold; Table 2). Peptide 23c also had slightly improved cellular entry efficiency relative to 19c (Table 1). Remarkably, substitution of Pip for Gln⁻⁴ produced an exceptionally potent and selective CAL PDZ ligand. Peptide FAM-24 exhibited a K_D value of 6 nM for the CAL PDZ domain and 130- to 1240-fold selectivity over NHERF1/2 PDZ domains (Table 2 and Figure S3). Peptide 24c (Figure 3), which is the disulfide-cyclized form of peptide 24, also showed robust cellular entry efficiency into HeLa cells (61% relative to that of CPP9) and was named as “PGD97” thereafter. The absolute cytosolic entry efficiency (defined as the ratio of cytosolic over extracellular concentration) of PGD97 was estimated to be 37%, based on the previously reported value of 62% for CPP9.³²

Figure 3.

(a) Structures of PGD97 (24c) and its reduced, biologically active form (24). The value next to each residue of peptide 24 (in red) represents the fold of reduction in CAL binding affinity upon substitution of alanine for that residue. (b) Docked pose of peptide 24 in complex with the CAL PDZ domain. The structure for peptide 24 was generated in Schrödinger Maestro³⁹ before molecular dynamics refinement using Desmond⁴² employing a previously reported crystal structure of CAL-PDZ (PDBID: 4K75).²⁸ Peptide 24 is depicted with carbon atoms in light blue, polar hydrogens in white, nonpolar hydrogens hidden, nitrogens in deep blue, oxygens in red, and sulfurs in yellow while the van der Waals surface of CAL PDZ is shown in tan. (c) Diagram showing the key interactions between peptide 24 and the CAL PDZ domain in (b). Hydrogen bonds are indicated by dashed red lines, while a π-π stacking interaction is indicated by a dashed blue line.

Structural Basis of Inhibitor 24 Binding to CAL PDZ Domain.

We assessed the contribution of each residue in peptide 24 to CAL binding by performing an “alanine scan”, i.e., replacement of each residue of peptide 24 with an alanine (or D-alanine). The C-terminal tert-leucine and Pip at P-4 are most critical for CAL PDZ binding. Substitution of Ala for the tert-leucine almost completely abolished CAL binding, while replacement of Pip with Ala reduced the binding affinity by 537-fold (Figures 3a and S4). Thr at P-2, Bta at P-5, and Arg at P-1 also contribute greatly to CAL PDZ binding, as removal of their side chains decreased the binding affinity by 78- to 9.3-fold. Interestingly, the three arginine residues of the CPP motif contribute significantly to CAL binding as well (2- to ~4-fold each). In contrast, the side chains of Pen at P-3 and 2-Nal at P-6 do not contribute significantly to CAL binding. Note that the latter observation (lack of CAL binding by the P-6 residue) conflicts with our earlier data on peptides 14 and 15 (Table 1). It appears that the Pip residue at P-4 alters the conformation of the peptide (presumably by generating a kink in the peptide backbone), allowing the entire peptide to engage in productive interactions with the PDZ domain surface. This may explain the dramatic effect on CAL binding (537-fold) when the Pip residue was replaced with Ala (Figure 3a).

To gain insight into the structural basis for the exceptional potency and selectivity of peptide 24, we modified the previously docked structure of peptide 1 in complex with CAL and performed multiple short (~100 ns) MD simulations. Over the duration of the simulations, peptide 24 remained associated with CAL, suggesting a stable binding mode was being sampled (Figure 3b). Analysis of the resulting binding mode provides additional support for the rationale employed during the optimization phase. The predicted binding mode maintains critical hydrogen bonds between the C-terminal carboxylate of peptide 24 and the backbone amides of Leu²⁹⁹, Gly³⁰⁰, and Ile³⁰¹ while demonstrating enhanced occupancy of the adjacent hydrophobic pocket formed by Leu¹⁶, Ile¹⁸, Ile²⁰, and Val⁷⁰ (Figure 3c). Introduction of Pip at position −4 results in a well-defined conformation that optimally positions Bta⁻⁵ for π-π stacking between the thienyl ring and His³⁰⁹, underscored by the 24-fold loss in affinity upon mutation of Bta⁻⁵ to Ala. The contribution of the cell-penetrating motif is also facilitated by the Pip-induced conformation. While Nal⁻⁶ is predominately solvent exposed, Arg⁻⁷ has productive charge-charge interactions with Glu³¹⁷ and N-terminal Cys⁻¹¹ acts as a hydrogen donor/acceptor to Val³⁷⁰. Taken together, these data support the value of introducing a distinct conformation to enhance CAL binding affinity and specificity.

Metabolic Stability of PGD97.

The stability of PGD97 (both 24 and 24c) in human serum was first assessed and compared with that of peptide 1 (reduced, linear form), the starting point of this medicinal chemistry campaign. The disulfide-cyclized form (24c) was highly stable in human serum; ~77% of the peptide remained intact after incubation for 24 h at 37 °C (Figure 4a). The reduced form (24) was slightly less stable, but still exhibited a serum t_1/2 >24 h at 37 °C. Under the same condition, peptide 1 was completely degraded within 30 min.

Figure 4.

Proteolytic stability of PGD97. (a) Stability of peptides 1 (linear form), 24 (linear form), and 24c (cyclic form) in 25% human serum at 37 °C as a function of time, as determined by analytical HPLC. (b) Structures of three potential proteolytic fragments of peptide 24. (c) Stability of peptide 24 in HCT116 cell spheroids. Heatmaps for peptide 24 and the three representative proteolytic fragments in (b) were generated by MALDI imaging at 0, 4, and 24 h and scaled to the intensity of the highest signal.

We also examined the intracellular degradation of PGD97 by mass spectrometric imaging. Colon cancer HCT116 cell spheroids were incubated with 24c (5 μM) for 4 or 24 h, washed, quickly frozen, and imaged by MALDI-TOF-TOF mass spectrometry and analyzed for the presence of proteolytic degradation fragments. Heatmaps, scaled relative to the most intense signal, were generated for the intact peptides (24c and 24) as well as ~20 potential proteolytic fragments at each time point. The heatmaps for peptide 24 and three representative fragments are shown in Figure 4b,c. High concentrations of peptide 24, but not 24c, was present in the spheroids after either 4 or 24 h of incubation. A degradation fragment derived from proteolytic cleavage between Thr⁻² and Arg⁻¹ (#1; Figure 4b) was observed, but none of the other potential degradation products were detected. These results indicate that PGD97 (24c) efficiently entered the cytosol of HCT116 cells and was rapidly reduced by intracellular thiols into its linear, biologically active form (24), which subsequently underwent partial degradation inside the cytosol. Importantly, high concentrations of intact peptide 24 was still present inside the cells after 24 h.

PGD97 Is Non-Cytotoxic.

To test whether PGD97 exerts any cytotoxicity to mammalian cells, we treated several commonly available human cell lines, including CFBE41o-, HCT116, and H358 (lung cancer) cells, with increasing concentrations of PGD97 and measured the viability of the treated cells by the MTT assay. No significant cytotoxicity was observed with any of the cell lines after 72 h of incubation at up to 50 μM PGD97 (Figure S5).

PGD97 Increases Surface Expression of Rescued F508del-CFTR.

Previous studies have shown that knockdown of CAL expression by siRNA increases the plasma membrane population of F508del-CFTR by 4.4-fold.³⁷ Inhibition of the CAL-CFTR interaction by PGD97 is therefore expected to reduce lysosomal degradation of plasma membrane CFTR, thereby increasing the overall cellular level and cell surface expression of CFTR (Figure 1a).^23,26 To test this hypothesis, we treated CFBE41o- cells,³⁸ which stably overexpress F508del-CFTR, with vehicle (DMSO), VX-661 (a corrector, 5 μM), or VX-661 (5 μM) in combination with PGD97 (100 nM) and immunostained the cells with an antibody against CFTR. In vehicle treated cells, the F508del-CFTR fluorescence was weak and predominantly perinuclear (Figure 5a). Treatment with VX-661 alone slightly increased the overall fluorescence intensity (by ~20%), but the fluorescence remained mostly perinuclear (Figure 5a,b). In contrast, cells treated with both VX-661 and PGD97 showed a 400% increase in CFTR fluorescence, which was also more diffused throughout the cytoplasmic region, with some staining extending to the surface of the cells. These observations indicate that PGD97 greatly increases the surface expression of F508del-CFTR.

Figure 5.

PGD97 improves CFTR function in CFBE41o- cells stably overexpressing F508del-CFTR. (a) Immunofluorescence staining of CFBE41o- cells after treatment with vehicle (DMSO), 5 μM VX-661, or 5 μM VX-661 + 100 nM PGD97. CFTR is stained green, the tight junction marker Zona Occludins-1 (ZO-1) is stained red, and nuclei are stained blue. (b) Quantification of CFTR staining in (a). (c) Change in short-circuit currents (I_sc) in CFBE41o- cells after treatment with VX-661 (5 μM) or VX-661 (5 μM) + PGD97 (100 nM). (d) Fraction of CFTR remaining in CFBE41o- cells pretreated with VX-661 (5 μM) or VX-661 (5 μM) + PGD97 (100 nM) after 2 h of cycloheximide treatment as assessed by western blot quantification. Data shown are relative to that of cells which were not treated with cycloheximide. *, p<0.05; **, p<0.01; and ***p<0.001 from n = 3 independent experiments.

PGD97 Improves Function of Rescued F508del-CFTR.

We next evaluated the capacity for PGD97 to restore F508del-CFTR function in CFBE41o- cells. Co-treatment with corrector VX-661 (5 μM) and PGD97 (100 nM) resulted in ~50% increases in I_SC over VX-661 alone as measured with either CFTR activation or inhibition (Figure 5c and data not shown). In order to confirm that PGD97 improves function of rescued F508del-CFTR by preventing its degradation, CFTR protein stability was evaluated. CFBE41o- cells were pretreated with VX-661 (5 μM) or a combination of VX-661 (5 μM) and PGD97 (100 nM) followed by a cycloheximide chase to arrest de novo protein synthesis. The fraction of extant CFTR was quantitated by western blotting. Co-treatment with VX-661 and PGD97 decreased the amount of F508del-CFTR degradation by ~60% relative to VX-661 alone (Figure 5d). Taken together, these results demonstrate that PGD97 increased the stability of the rescued CFTR population resulting in improved channel function.

Encouraged by the results obtained with the immortalized CFBE41o- cells, we next tested PGD97 in patient-derived primary lung epithelial cells from multiple CF donors homozygous for F508del-CFTR (CF-HBE). In the presence of a fixed concentration of VX-661 (5 μM), PGD97 dose-dependently increased the ion channel activity (I_SC), with an estimated EC₅₀ value of ~10 nM, and reached the maximal activity at ~100 nM concentration (Figure 6a). Importantly, at the saturating concentration (100 nM), PGD97 increased the ion channel activity by ~2-fold, relative to the VX-661 only control as observed using a CFTR specific inhibitor (Inh-172) (Figure 6b and S6). A greater negative change in current upon the addition of Inh-172 signifies more chloride transport by CFTR. Interestingly, PGD97 also demonstrated efficacy as a single agent (in absence of a corrector or potentiator); treatment of primary CF-HBE cells with 100 nM PGD97 alone increased the ion current by ~3-fold, relative to the control (no treatment) (Figure 6c). The magnitude of improvement by 100 nM PGD97 was comparable to that achieved by 5 μM VX-661.

Figure 6.

Ex vivo efficacy of PGD97 in CF patient-derived primary HBE cells (CF-HBE). (a) Dose response of PGD97 in increasing the short-circuit currents (I_sc) in the presence of 0 or 5 μM VX-661. (b) Comparison of CFTR-mediated I_sc using the specific CFTR inhibitor Inh-172 (100 μM) in CF-HBE cells treated with vehicle (CT), VX-661 alone (5 μM), or VX-661 (5 μM) + PGD97 (100 nM). (c) Change in I_sc after treatment with 100 nM PGD97 alone. Primary CF-HBE cells from 3 CF donors were used in the experiments. Data shown were from five independent experiments (n = 5 filters). *, p<0.05, **, p<0.01, and ***, p<0.001 via ANOVA with Tukey’s posthoc multiple comparison.

CONCLUSION

A medicinal chemistry campaign assisted by in silico modeling and focused on enhancing affinity, selectivity, cellular uptake, and stability of a previously reported disulfide-cyclized peptidyl inhibitor led to the discovery of PGD97 as a potent, selective, cell-permeable, and proteolytically stable inhibitor against the CAL-CFTR interaction. To our knowledge, PGD97 represents the most potent CAL ligand reported to date and demonstrates excellent selectivity over other proteins that also recognize the C-terminus of CFTR. Biological evaluation of PGD97 indicates that it is capable of stabilizing F508del-CFTR at the cell membrane and improving CFTR function required for proper fluid homeostasis in the lung. PGD97 provides a potential novel treatment for CF patients, regardless of the specific types of mutations, and presents an opportunity for significant synergy with existing therapeutic options, such as potentiators and correctors. In addition, this study demonstrates that peptides can be developed into drug-like molecules by overcoming their inherent limitations in metabolic stability and cell-permeability while retaining their high affinity and specificity.

EXPERIMENTAL SECTION

Materials.

Fmoc-protected amino acids for peptide synthesis were purchased from Advanced ChemTech (Louisville, KY), NovaBiochem (La Jolla, CA), or Aapptec (Louisville, KY). Rink amide resin (100–200 mesh, 0.3–0.6 meq/g), Wang resin (100–200 mesh, 1.0–1.4 meq/g), O-benzotriazole-N,N,N’,N’-tetramethyluronium hexafluorophosphate (HATU), and 1-hydroxybenzotriazole hydrate (HOBt) were purchased from Chem-Impex (Wood Dale, IL). Fluorescein succinimidyl ester (FAM), 5-(and-6)-carboxynaphthofluorescein succinimidyl ester (NF), and human serum were purchased from Sigma Aldrich (St. Louis, MO). Cell culture media, fetal bovine serum, penicillin-streptomycin, Dulbecco’s phosphate-buffered saline (DPBS) (2.67 mM potassium chloride, 1.47 mM potassium phosphate monobasic, 137 mM sodium chloride, 8.06 mM sodium phosphate dibasic), and 0.25% trypsin-EDTA were purchased from Invitrogen (Carlsbad, CA). Cell proliferation kit (MTT) was purchased from Roche (Indianapolis, IN). H358 and HCT116 were purchased from ATCC (Manassas, VA). CFBE410- was purchased from EMD Millipore (Burlington, MA). All solvents and other chemical reagents were obtained from Sigma-Aldrich, VWR (West Chester, PA), or Fisher Scientific and were used without further purification unless noted otherwise.

Peptide Synthesis, Cyclization, and Purification.

Peptides were synthesized by standard solid-phase peptide synthesis on Wang resin. After swelling the resin in DCM, the first C-terminal residue was coupled using 5 eq. Fmoc-protected amino acid, 5 eq. diisopropylcarbodiimide and 0.05 eq. 4-(N,N-dimethylamino)pyridine in DCM for 1 h at RT with mixing (repeated twice). For all subsequent coupling steps, the Fmoc protecting group was removed using 20% piperidine in DMF twice for 5 min at RT with mixing. Peptide couplings were performed using 5 eq. Fmoc-protected amino acid, 4.9 eq. HATU, 5 eq. HOBt and 10 eq. DIPEA in 70:30 DMF/DCM (v/v) for 1 h at RT with mixing. After completion of the linear sequence, peptides were acetylated at the N-terminus using 10 eq. Ac₂O, 10 eq. DIPEA in DCM for 15 min at RT with mixing. For peptides bearing a terminal fluorophore, the N-terminal Fmoc group was removed and the resin was incubated overnight with 2 eq. NHS-fluorophore and 2 eq. DIPEA in DMF with mixing. Excess unreacted dye was removed by thoroughly washing the resin with DCM/DMF before cleavage. For peptides containing a functionalized lysine residue, the side-chain Mtt protecting group was removed by repeated treatment with 97:2:1 DCM/TFA/triisopropylsilane (TIPS) in DCM (v/v/v), until no yellow color was observed (requiring approximately 50 mL of deprotection cocktail per 100 mg of resin). The resin was thoroughly washed with DCM, DMF, and then incubated with 1 M HOBt in DMF. N^ε modifications were introduced by incubating the resin with either 2 eq. nicotinic acid succinimidyl ester (Nic-OSu) in DMF or 2 eq. trifluoroacetic anhydride in DCM and reacted for 2 h at RT with mixing, after which the resin was thoroughly washed with DCM and DMF.

Peptide cleavage and deprotection were accomplished by treating the resin with 92.5:5:2.5 (v/v) TFA/2,2’-(ethylenedioxy)diethanethiol/TIPS (1 mL of cocktail/50 mg resin) for 3 h at RT with mixing. The crude peptide solution was concentrated under nitrogen, triturated with chilled diethyl ether and the precipitate was isolated by centrifugation. The precipitate was resuspended in chilled diethyl ether and centrifuged (repeated twice). Following trituration, the crude peptide was dried under inert atmosphere, dissolved in a minimal volume of DMF, diluted 1:10 into ddH₂O/acetonitrile containing 0.05% TFA, and purified by reversed-phase HPLC equipped with a Waters XBridge C18 semi-preparative column using a linear gradient of 5–60% acetonitrile in ddH₂O with 0.05% TFA. Fractions were collected based on absorbance at 280 nm, analyzed by MALDI-MS and lyophilized.

For disulfide-cyclized peptides, purified lyophilized peptide was dissolved in DMSO, diluted 1:10 (v/v) in 10x PBS (for a final concentration of 10% DMSO), and incubated overnight at RT with mixing. The cyclization reaction mixture was fractionated by RP-HPLC using a Waters XBridge C18 semi-preparative column with a linear gradient of 5–60% acetonitrile in ddH₂O containing 0.05% TFA to separate the linear and disulfide-cyclized peptides. Cyclization was confirmed by analytical HPLC through co-injection with the linear species using a Waters X-Bridge C18 analytical column and high-resolution MALDI-FT-ICR mass spectrometry. All peptides used in this work were judged to have ≥95% purity by analytical HPLC or UPLC.

Protein Expression and Purification.

Plasmids encoding for CAL,²² NHERF1 and NHERF2 PDZ domains²² were a generous gift from D. R. Madden (Dartmouth College). Recombinant CAL, NHERF1, and NHERF2 PDZ domains were produced in E. coli BL21 cells with a C-terminal His₆ tag. In brief, 100 mL of LB broth containing 75 mg/L ampicillin was inoculated with a single colony and incubated overnight at 37 °C with shaking (180 RPM). The next day, 10 mL of the seed culture was diluted into 1 L of LB broth supplemented with 75 mg/L ampicillin and incubated at 37 °C until OD₆₀₀ reached 0.6–0.8, at which point the culture was induced with 0.1 mM IPTG and incubated for 16 h at 18 °C. After 16 h, the cells were pelleted by centrifugation and frozen at −80 °C for further processing.

Cell pellets (from 1 L of culture) were lysed in 50 mL of lysis buffer [50 mM Na₂HPO₄, pH 8.0, 300 mM NaCl, 20 mM imidazole, 0.2 mg/mL lysozyme, 2 mM DTT, and two EDTA-free protease inhibitor cocktail tablets (Roche)] by sonication. After centrifugation, the clear cell lysate was loaded onto a Ni-NTA affinity column and the bound protein was eluted with an eluting buffer (50 mM Na₂HPO₄, pH 8.0, 300 mM NaCl, 2 mM DTT, and 250 mM imidazole). Fractions containing the desired protein were pooled, concentrated, and exchanged into PBS (pH 7.4) containing 10% glycerol in an Amicon Ultra-15 centrifugal concentrator (10K MWCO, EMD Millipore). Approximately 30 mg of CAL PDZ protein was obtained from each litre of cell culture.

Docking.

Compounds for screening were constructed in Maestro and then prepared for docking using LigPrep,⁴⁰ with all ligands ionized to pH 7.4 with all tautomers generated when applicable and stereochemistry retained from the input structures. The receptor for docking (PDBID: 4K75)²⁸ was prepared using the Grid Generation tool with no constraints specified and the grid box defined based on the geometrical center of the extant docked peptide ligand.

Docking was performed using flexible ligand docking in Glide with XP precision with modified settings.⁴¹ Each compound was docked with nitrogen inversions and ring conformations sampled, amides penalized for nonplanar conformations and the planarity of conjugated π groups enhanced. For conformer generation, enhanced sampling was used and during initial pose selection 200,000 poses per ligand were retained with the top 5000 poses per ligand kept for energy minimization combined with expanded sampling. Energy minimization was performed over 5000 steps and halogens were counted as hydrogen bond acceptors. Final docked compounds were ranked by their docking score and visualized using Maestro or UCSF Chimera.

Binding Affinity.

For direct binding affinity by fluorescent polarization, 50 nM fluorescein-labelled peptide was incubated with varying concentrations of PDZ protein (0–2.5 μM for CAL-PDZ, 0–25 μM for NHERF1/2 PDZ domains) in PBS (pH 7.4) containing 0.01% Triton-X100 and 2 mM tris(carboxyethyl)phosphine (TCEP) for 1 h at room temperature. After 1 h, 20 μL of the solution was transferred into non-binding 384-well black-on-black microplates (Greiner Bio-One) and fluorescence polarization was measured using a TECAN Infinite M1000 plate reader. Data was processed using GraphPad PRISM ver. 8.0. K_D values reported are the mean ± SD of n = 3 independent experiments.

For competitive binding experiments, the following two different assay methods were employed. Method A (for peptides 2–11): Fluorescein-labeled PDZ probe (FAM-ANSRWPTSII-OH²⁴; 100 nM) was incubated with PDZ domain (5.0 μM) in PBS (pH 7.4) containing 5 mM TCEP for 1 h. Serial dilutions of each competitor peptide (0–35 μM) were prepared from DMF stock solutions into PBS containing 0.01% Triton-X100 and 2 mM TCEP and added to aliquots of the equilibrated PDZ/probe solution. After incubation for 1 h in the dark and with gentle mixing, 20 μL of each sample was transferred into 384-well black-on-black microplates and fluorescence polarization as measured using a TECAN Infinite M1000 plate reader. Data was analysed using GraphPad PRISM ver. 8.0 and fit to a four-parameter variable slope equation to determine IC₅₀ values. Method B (for peptides 12–24): Same as method A, except that the assay reaction contained 50 nM FAM-22 as probe, 100 nM PDZ domain, and 0–20 μM peptide competitor. IC₅₀ values reported are the mean ± SD of n = 2–4 independent experiments.

Human Serum Stability.

Whole human serum was diluted 1:4 (v/v) in sterile DPBS and equilibrated at 37 °C for 15 min. The indicated compounds (100 μM final concentration) were added to the diluted serum and incubated at 37 °C with gentle mixing. At designated time points, 100 μL aliquots were withdrawn and mixed with 100 μL of 15% trichloroacetic acid (TCA) in MeOH (w/v) and 100 μL of MeCN and stored at 4 °C for 24 h to effect complete de-proteinization. After protein precipitation was complete, each aliquot was centrifuged (15000 g, 5 min, 4 °C) and the supernatant was analyzed by reversed-phase HPLC using a Waters XBridge C18 analytical column and a linear gradient of 5–60% MeCN in ddH₂O containing 0.05% TFA and a UV/Vis detector monitoring at 214 nm. The chromatograms were integrated and the area of the parent compound peak at t = 0 and each time point was used to determine the percentage of the parent compound remaining. Aliquots of the peak used for integration and comparison purposes were collected and analyzed by MALDI-TOF to ensure that the desired, intact parent compound was present and contributing to the observed absorbance at 214 nm.

Intracellular Stability.

HCT116 cells were grown in McCoy’s 5A media supplemented with 1% L-glutamine and 10% fetal bovine serum. Cells were cultures in 5% CO₂ at 37 °C. HCT116 spheroids were seeded with 7,000 cells in agarose-coated plates and grown for 12 days with half volume media changes every 2 days. They were prepared for imaging mass spectrometry analysis as previously described.⁴³ PGD97 was added to HCT116 spheroids to a final concentration of 5 μM. After specified incubation times, media was aspirated and the spheroids were washed with 1x PBS. They were then embedded in gelatin. The spheroids were harvested and sectioned into 16 μm-thick slices by using the gelatin-assisted sectioning method previously described.⁴³ Sections were thaw-mounted on ITO coated slides for MALDI MSI analysis. A TM sprayer nebulizer (HTX Technologies, Carrboro, NC) was used to apply the matrix for MALDI MSI analysis. The matrix, 2,5-dihydroxybenzoic acid (DHB), was dissolved in 50:50 acetonitrile/water with 0.1% TFA (EMB, Billerica, MA) to yield a final concentration of 10 mg/mL. Matrix was applied at 75 °C for 8 passes over the sample. The flow rate of matrix was 0.1 mL/min at a velocity of 1000 mm/min, track spacing of 2 mm, pressure 10 psi, gas flow rate of 3 L/min, nozzle height 40 mm, and a drying time of 20 s between each pass. After matrix was applied, the sample was dried in a desiccator before MALDI MSI analysis.

An UltrafleXtreme MALDI-TOF-TOF spectrometer (Bruker Daltonics, Billerica, MA) was operated in positive ion mode with the reflector on and set to acquire a mass range of 360–1400 m/z. Laser raster was set to 40 μm along both the x and y axes. Each MALDI mass spectrum for each pixel is a result of 200 consecutive laser shots. External calibration was performed using a standard peptide mixture on one spheroid. The images were processed using FlexAnalysis 3.0 and FlexImaging 4.1 (Bruker Daltonics, Billerica, MA). All spectra were normalized against total ion count, defined as the sum of all intensities in the mass range analyzed to reduce influences by matrix hot spots. All samples were analyzed in biological triplicate and technical triplicate. Heatmaps, scaled relative to the most intense signal, were generated for each time point and fragment.

Cellular Uptake.

HeLa cells were maintained in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin and incubated at 37°C, 5% CO₂. The day before the experiment, cells were washed with DPBS and harvested using 0.25% trypsin/EDTA, before centrifugation and resuspended in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin. Cell density was determined using a hemocytometer (Reichert) and were seeded into 12-well cell-culture treated plates (VWR) at a final density of 15 × 10⁴ cells/well in DMEM supplemented with 10% FBS and 1% penicillin/streptomycin and incubated overnight at 37 °C in the presence of 5% CO2. The next day, the media was aspirated and the cells were washed three times with warm DPBS. Peptides were diluted from DMSO stock solutions to a final concentration of 5 μM, 0.5% DMSO in DMEM containing 10% FBS with 1% penicillin/streptomycin, added to each well, and then incubated for 2 h at 37 °C with 5% CO2. After 2 h, the compound-containing media was aspirated and the cells were washed three times with ice-cold DPBS. Cells were removed from the plate by treatment with 300 μL of 0.25% trypsin/EDTA for 5 min at 37 °C, 5% CO₂, neutralized with 600 μL of ice-cold DPBS, and centrifuged at 300 g, 4 °C for 5 min. Cells were resuspended in ice-cold DPBS and immediately analyzed by flow cytometry using a BD Biosciences LSR II flow cytometer and gated using FlowJo. Mean fluorescence intensity (MFI) values are reported as percentages relative to that of the positive control, CPP9 (cyclo-[fΦRrRrQ]-miniPEG-K[NF]). Values reported are the mean ± SD of n = 3 independent experiments.

MTT Viability Assay.

Immortalized human cell lines CFBE41o-, HCT116 and H358 were maintained in complete growth media supplemented with 10% FBS and 1% penicillin/streptomycin at 37°C, 5% CO₂. The day before treatment, cells were washed with DPBS and harvested using 0.25% trypsin/EDTA for 5 min at 37°C, followed by neutralization with DPBS and centrifugation. Cells were resuspended in complete growth media, the density was determined using a hemocytometer and then seeded in cell culture-treated clear 96-well microplates at a final density of 5 × 10³ cells/well and incubated overnight at 37°C, 5% CO₂. The next day, a serial dilution of 25c was prepared from the DMSO stock in DPBS and supplemented with additional DMSO to yield a constant concentration of 0.5% DMSO (v/v) at the indicated concentrations. Control wells were treated with 0.5% DMSO in DPBS. Compounds were added to each well and incubated for 72 h at 37°C, 5% CO₂. Following compound treatment, 10 μL of MTT stock solution was added to each well and incubated for 4 h at 37°C, 5% CO₂. After 4 h, 100 μL of SDS-HCl solubilizing solution was added to each well and the plate was incubated overnight at 37°C, 5% CO₂. The next day, the plate was removed from the incubator and absorbance of the solubilized formazan product was measured at 565 nm on a Tecan Infinite M1000 Pro plate reader. Absorbance in control wells was defined as 100% viable and viability in treatment wells was calculated relative to this value. Values reported are the mean ± SD of n = 3 independent experiments.

Tissue Culture.

CFBE41o- is an immortalized human bronchial epithelial cell line from a CF patient (homozygous for F508del mutation) overexpressing F508del-CFTR and was purchased from EMD Millipore (Burlington, MA). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing L-glutamine, 10% fetal bovine serum and penicillin (100 U/mL) and streptomycin (100 μg/mL), and hygromycin B (300 μg/ml) on plates coated with human fibronectin (1 mg/mL), bovine collagen I (3 mg/mL), and bovine serum albumin (1 mg/mL).

Primary human bronchial epithelial cells (CF-HBE) were isolated from CF patients receiving lung transplant following a protocol approved by Nationwide Children’s Hospital (Columbus Ohio). Primary CF-HBE cells were isolated and grown as previously described.^34,35

CFTR Function using Ussing Chambers.

Primary CF-HBE cells were incubated with a CFTR corrector. After 24 h, a peptide or vehicle (DMSO) was added apically in Hank’s balanced salt solution and incubated for 1 h, removed to keep the cultures at the air-liquid interface, and the cells were incubated for another 24 h. CFTR-mediated ion transport was measured using Ussing chambers as previously described.^34,35 Briefly, cells were mounted into an Ussing chamber system VCC MC6 (Physiologic Instruments, San Diego, CA) with U2500 self-contained Ussing chambers. Cells were bathed in Ringer’s solution (25 mM NaHCO₃, 2.4 mM KH₂PO₄, 1.24 mM K₂HPO₄, 115 mM NaCl, 1.2 mM MgCl₂, 10 mM D-glucose, and 1.2 mM CaCl₂, pH 7.4). After allowing transepithelial currents to stabilize, the apical buffer was removed and replaced with low-chloride Ringer’s solution (1.2 mM NaCl and 115 mM sodium gluconate, and otherwise identical to Ringer’s solution). One hundred μM Amiloride (Sigma Aldrich, St. Louis, MO) was added to the apical chamber to inhibit epithelial sodium channels (ENaC). Following stabilization of current, CFTR was activated by the addition of 10 μM forskolin (Abcam, Cambridge, MA) to the apical and basal chambers. Finally, CFTR was inhibited by the addition of 10 μM Inh-172 (Sigma Aldrich, St. Louis, MO), a specific CFTR inhibitor, to the apical chamber. Amiloride was not added to experiments involving CFBE41o- cells, since they do not express functional ENaC. Changes in short-circuit currents (Isc) were measured after activation of CFTR by forskolin (10 μM) and inhibition by Inh-172. Data reported were generated with primary CF-HBE cells derived from three F508del/F508del patients and at least three independent experiments.

CFTR Protein Stability.

CFBE41o- cells stably expressing F508del-CFTR were incubated with 5 μM VX-661 (Selleckchem, Houston, TX). After 24 h, PGD97 (100 nM) or vehicle (DMSO) was added and cells were incubated for another 24 h. After that, cells were treated with 100 μg/mL cycloheximide (Sigma Aldrich, St. Louis, MO) for 2 h prior to lysis and processing for immunoblotting with antibodies against CFTR (R&D Systems, Minnesota, MN) and GAPDH (Santa Cruz Biotechnology, Dallas, TX). The fraction of remaining CFTR was calculated by normalizing the ratio of CFTR to GAPDH after 2 h to the that of cells untreated with cycloheximide. Cycloheximide inhibits de novo protein synthesis and can be used to measure the biological half-life of proteins.

Immunofluorescence.

Cells were treated as described previously (CFTR protein stability). Cells seeded on glass coverslips were washed with ice-cold PBS and then fixed with methanol at −20 °C for 20 min. Following fixation, cells were rinsed with PBS and blocked with 1% bovine serum albumin (BSA). Cells were incubated with primary antibody solutions to stain for CFTR (mouse monoclonal clone 24–1, R&D Systems, Minnesota, MN) and ZO-1 (rabbit polyclonal clone H-300, Santa Cruz Biotechnology, Dallas, TX). Following primary staining, cells were washed with PBS and blocked again with 1% BSA. Cells were then incubated with secondary antibody solution containing Alexa Fluor 488-conjugated mouse antibody and Alexa Fluor 594-conjugated rabbit antibody. Coverslips were then rinsed with PBS, and mounted on glass slides using ProLong^™ Gold antifade media containing 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen, Waltham, MA) and imaged the next day using an Olympus IX-51 fluorescent microscope (Olympus Corporation, Tokyo, Japan). Quantification of the fluorescence was performed using NIH Image J software. Three independent experiments were performed and fluorescence was quantified in at least 12 cells/condition. Statistical analysis was performed using t-test.

Statistical Analysis.

Statistical calculations were performed using GraphPad Prism^® v8.02 (GraphPad Software, Inc.) and data are expressed as mean ± SD. Statistical significance was determined using t-test or One-way ANOVA with Tukey’s posthoc multiple comparison. P values ≤0.05 were considered significant.

ABBREVIATIONS

CF

cystic fibrosis

CF-HBE

CF patient-derived primary human bronchial epithelia cells

CPP

cell-penetrating peptide

FAM

5(6)-carboxyfluorescein

FP

fluorescence polarization

NF

5(6)-carboxynaphthofluorescein

PPI

protein-protein interaction

TMR

tetramethylrhodamine