Abstract
The cystic fibrosis transmembrane conductance regulator (CFTR) represents the main Cl- channel in the apical membrane of epithelial cells for cAMP-dependent Cl- secretion. Here we report on the synthesis and screening of a small library of 6-phenylpyrrolo[2,3-b]pyrazines (named RP derivatives) evaluated as activators of wild-type CFTR, G551D-CFTR, and F508del-CFTR Cl- channels. Iodide efflux and whole-cell patch-clamp recordings analysis identified RP107 [7-n-butyl-6-(4-hydroxyphenyl)[5H]-pyrrolo[2,3-b]pyrazine] as a submicromolar activator of wild-type (WT)-CFTR [human airway epithelial Calu-3 and WT-CFTR-Chinese hamster ovary (CHO) cells], G551D-CFTR (G551D-CFTR-CHO cells), and F508del-CFTR (in temperature-corrected human airway epithelial F508del/F508del CF15 cells). The structural analog RP108 [7-n-butyl-6-(4-chlorophenyl)[5H]pyrrolo[2,3-b]pyrazine], contrary to RP107, was a less potent activator only at micromolar concentrations. RP107 and RP108 did not have any effect on the cellular cAMP level. Activation was potentiated by low concentration of forskolin and inhibited by glibenclamide and CFTRinh-172 [3-[(3-trifluoromethyl)phenyl]-5-[(4-carboxyphenyl-)methylene]-2-thioxo-4-thiazolidinone]but not by calixarene or DIDS (4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid). Finally, we found significant stimulation of short circuit current (Isc) by RP107 (EC50 = 89 nM) and RP108 (EC50 = 103 μM) on colon of Cftr+/+ but not of Cftr-/- mice mounted in Ussing chamber. Stimulation of Isc was inhibited by glibenclamide but not affected by DIDS. These results show that RP107 stimulates wild-type CFTR and mutated CFTR, with submicromolar affinity by a cAMP-independent mechanism. Our preliminary structure-activity relationship study identified 4-hydroxyphenyl and 7-n-butyl as determinants required for activation of CFTR. The potency of these agents indicates that compounds in this class may be of therapeutic benefit in CFTR-related diseases, including cystic fibrosis.
Cystic Fibrosis (CF) is a worldwide fatal autosomal recessive genetic disease caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene (Riordan et al., 1989; Ratjen and Doring, 2003). CF affects 1/2500 to 1/3500 live births with an estimated global CF population above 70,000 individuals. Mutations in the CFTR gene alter composition of epithelial secretions and lead to chronic airway obstructions and infections, pancreatic failure, male infertility, and elevated level of salt in sweat (Quinton, 1983; Riordan et al., 1989; Ratjen and Doring, 2003). CFTR functions as a cyclic AMP-dependent and ATP-gated Cl- channel (Anderson et al., 1991; Tabcharani et al., 1991; Sheppard and Welsh, 1999) and is thought to be the main cAMP-dependent pathway for Cl- exit and hence for fluid secretion in epithelial cells in the airways, pancreas, intestine, testis, and other fluid-transporting tissues (Gray et al., 1988; Shen et al., 1994; Sheppard and Welsh, 1999; Ratjen and Doring, 2003). Individuals suffering from CF often require frequent hospitalizations and heavy medication. Although these treatments do not cure the disease, they have allowed the median age of survival to rise from 14 years of age in 1969 to ≈35 nowadays and also improved the physical comfort and social life of CF patients (http://www.cff.org). More than 1300 mutations of the CF gene have been so far detected via worldwide chromosome analysis (http://www.genet.sickkids.on.ca/cftr). Because of this large number of gene abnormalities, CFTR mutations were assigned according to the fate of the final product into one of six classes of mutations (Welsh and Smith, 1993). Therefore, based on this classification, it is feasible to predict the strategy for developing a CF therapy according to the class of the CFTR mutation.
Identification of a selective agonist for a particular ion channel without secondary effect on other channels or unrelated proteins is critical for use as a biological tool or pharmaceutical drug but is a high-cost process that usually takes several years. However, a number of compounds have been identified during the last 15 years, activating wild-type and mutated CFTR by mechanisms that do not involve an increase of intracellular cAMP (reviewed in Becq, 2006). Our laboratory develops and synthesizes original small molecules and evaluates them for their ability to activate CFTR using a simple and robust robotic cell-based assay (Marivingt-Mounir et al., 2004). In this report, we present the 6-phenylpyrrolo[2,3-b]pyrazines, a family of derivatives (also named Aloisines) initially described as CDK/GSK-3 inhibitors (Mettey et al., 2003). One of these compounds, RP107 [7-n-butyl-6-(4-hydroxyphenyl)[5H]pyrrolo[2,3-b]pyrazine], activates CFTR-dependent Cl- secretion with submicromolar affinity in various epithelial cell types. Moreover, RP107 activates two CF-associated CFTR mutants (G551D-CFTR and F508del-CFTR mutants). The potency of these agents indicates the possibility that compounds in this class may be of therapeutic benefit in CF.
Materials and Methods
Chemistry. The 6-phenylpyrrolo[2,3-b]pyrazines derivatives shown in Table 1 were synthesized as presented in Fig. 1 from alkylpyrazines and aromatic nitriles, as described previously (Vierfond et al., 1981). Demethylation of methoxycompounds (1, 3, 5, 7, and 9) was obtained by heating in refluxing HBr. Synthesis and spectral data of compounds 1 through 8 and 11 through 12 (Table 1) are described elsewhere (Mettey et al., 2003).
Compounds used are as follows: 6-(4-methoxyphenyl)[5H]pyrrolo-[2,3-b]pyrazine (1, RP11); 6-(4-hydroxyphenyl)[5H]pyrrolo[2,3-b]-pyrazine (2, RP26); 6-(4-methoxyphenyl)-7-methyl[5H]pyrrolo[2,3-b]pyrazine (3, RP95); 6-(4-hydroxyphenyl)-7-methyl[5H]pyrrolo[2,3-b]pyrazine (4, RP96); 6-(4-methoxyphenyl)-7-propyl[5H]pyrrolo[2,3-b]pyrazine (5, RP127); 6-(4-hydroxyphenyl)-7-propyl[5H]pyrrolo[2,3-b]pyrazine (6, RP132); 7-n-butyl-6-(4-methoxyphenyl)[5H]pyrrolo[2,3-b]pyrazine (7, RP106); 8, RP107; 6-(4-chlorophenyl)[5H]-pyrrolo[2,3-b]pyrazine (11, RP14); and 7-n-butyl-6-(4-chlorophenyl)[5H]pyrrolo[2,3-b]pyrazine (12, RP108). Similar chemical synthesis method has been used for compounds 9 and 10. The corresponding elemental analyses are provided below.
Compound 9, RP149. m.p. 144.2°C; IR 3143, 3059, 2959, 2927, 2858 cm-1; 1H NMR (60 MHz, DMSO-d6), δ 11.75 (bs, 1H), 8.40 and 8.20 (2d, 1H each, J = 2.6 Hz), 7.70 and 7.10 (2bd, 2H each, J = 8 Hz), 3.90 (s, 3H), 3.20-2.80 (m, 2H), 2.00-1.1 (m, 6H), 0.9 (t, 3H, J = 7.2 Hz). Analog calculated for C18H21N3O (295.39), calculated for C18H21N3O: C, 73.19; H, 7.17; N, 14.22. Found: C, 73.06; H, 7.26; N, 14.09.
Compound 10, RP150. m.p. 270.7°C. IR 3224, 3150, 3060, 2949, 2924, 2861 cm-1. 1H NMR (60 MHz, DMSO-d6) δ 11.85 (bs, 1H), 9.90 (bs, 1H), 8.35 and 8.15 (2d, 1H each, J = 2.6 Hz), 7.50 and 6.95 (2d, 2H each, J = 7 Hz), 3.10-1.70 (m, 2H), 2.00-1.05 (m, 6H), 0.80 (t, 3H, J = 7 Hz). Analog calculated for C17H19N3O (281.36): C, 72.57; H, 6.81; N, 14.93. Found: C, 72.35; H, 6.78; N, 14.75.
Other Chemicals. TS-TM calix[4]arene (5,11,17,23-tetrasulfonato-25,26,27,28-tetramethoxy-calix[4]arene), an inhibitor of outwardly rectifying Cl- channels (Singh et al., 1995), was generously provided by A. Singh and R. J. Bridges (University of Pittsburgh, Pittsburgh, PA). The specific CFTR inhibitor 3-[(3-trifluoromethyl)-phenyl]-5-[(4-carboxyphenyl)methylene]-2-thioxo-4-thiazolidinone (CFTRinh-172) (Ma et al., 2002) and forskolin were purchased from VWR International (Fontenay/bois, France) and PKC Pharmaceuticals, Inc. (Woburn, MA), respectively. All other chemicals were from Sigma (St Louis, MO). All chemical agents, including pyrrolo[2,3-b]pyrazines compounds, were dissolved in DMSO (final concentration ≤0.1%), with the exception of TS-TM calix[4]arene, which was dissolved in H2O. The currents were not altered by DMSO alone.
Cell Culture. All cell lines were grown at 37°C in 5% CO2 under standard culture conditions as follows. CHO cells stably transfected with pNUT vector alone (mock-CHO) or containing wild-type CFTR (WT-CFTR-CHO) and the mutant G551D-CFTR were provided by J. R. Riordan and X. B. Chang (Mayo Clinic of Scottsdale, Scottsdale, AZ) (Tabcharani et al., 1991; Becq et al., 1994, 1999). They were maintained in α-minimal essential medium-GlutaMAX containing 7% FBS, 50 IU/ml penicillin and 50 μg/ml streptomycin, and methotrexate for cell selection (WT-CFTR-CHO, 100 μM; G551D-CHO, 20 μM; pNUT-CHO, 20 μM) (Tabcharani et al., 1991; Becq et al., 1994). The human pulmonary epithelial cell line Calu-3 cell (American Type Culture Collection, Manassas, VA) (Shen et al., 1994) was maintained in Dulbecco's modified Eagle's medium-Ham's F-12 (1:1) nutritive mix supplemented by 10% FBS and 100 IU/ml penicillin and 100 μg/ml streptomycin (Dérand et al., 2004). The human nasal epithelial JME/CF15 cell line, derived from a F508del homozygous patient (Jefferson et al., 1990), was maintained in Dulbecco's modified Eagle's medium-Ham's F-12 (3:1) nutritive mix supplemented by 10% FBS, 100 IU/ml penicillin and 100 μg/ml streptomycin, 5 μg/ml insulin, 5 μg/ml transferrin, 5.5 μM epinephrine, 180 μM adenine, 1.64 nM epidermal growth factor, 2 nM T3 (3,3′,5-triiodo-l-thyronine sodium salt), and 1.1 μM hydrocortisone (Norez et al., 2006). All culture media and antibiotics were from Invitrogen (Cergy-Pontoise, France), and FBS was from PerbioScience (Brebières, France). Hormones and growth factors were from Sigma. Cells were seeded in 24-well plates for iodide efflux and [cAMP] measurements and in 35-mm plastic dishes for whole-cell patch-clamp recordings. Medium was renewed at 2-day intervals.
[cAMP] Measurements. Calu-3 cells were incubated for 5-10 min at 37°C in culture medium with drugs at indicated concentration. Supernatants were collected, and cells were lysed with 12% trichloroacetic acid. The different collected fractions were homogenized and centrifuged. The cAMP was evaluated with radioimmunoassay 125I-cAMP kit (PerkinElmer Life Sciences, Courtaboeuf, France). The radioactivity was counted by gamma Cobra II counter (PerkinElmer Life Sciences). The cAMP values were determined following the manufacturer's instructions.
Iodide Efflux. Screening of small molecules and concentration-response curves were determined by measuring the rate of 125I efflux with a high-capacity robotic system (MultiProbe II EXT; PerkinElmer Life Sciences) adapted to the determination of iodide efflux as described previously (Marivingt-Mounir et al., 2004). Our protocol of screening is as follows. Four different cell types, WT-CFTR-CHO, G551D-CFTR CHO, Calu-3, and CF-15, were incubated in Multiwell plates at 37°C in Krebs' solution containing 1 μM KI and 1 μCi/ml Na125I (NEN, Boston, MA) for 30 min (CHO cells) or 1 h (Calu-3 and CF15 cells) to permit the 125I to reach equilibrium. The first three aliquots were used to establish a stable baseline in ice-cold Krebs' buffer (from t0 to t2). A medium containing the appropriate drug was then used for the remaining aliquots from t3 to t8. Residual radioactivity was extracted at the end of the experiment, with a mixture of 0.1 N NaOH and 0.1% SDS and determined using a gamma counter (Cobra II; PerkinElmer Life Sciences). The fraction of initial intracellular 125I lost during each time point was determined, and time-dependent rates (k = peak rate, min-1)of 125I efflux were calculated from the following equation: k = ln (125It1/125It2)/(t1 - t2), where 125It is the intracellular 125I at time t and t1 and t2 are successive time points (Marivingt-Mounir et al., 2004). Relative rates were calculated: kpeak (peak rate of efflux) - kbasal (basal rate of efflux) (min-1), i.e., the maximal value for the time-dependent rate (kpeak, min-1) excluding the third point used to establish the baseline (kbasal, min-1). Concentration-dependent activation curves were constructed as percentage maximal activation (set at 100%) transformed from the calculated relative rates. In some experiments, chloride transport inhibitors were present in the loading solution and in the efflux buffer. The activity of CFTR-dependent iodide efflux was stimulated either by 1 μM forskolin or by pyrrolo[2,3-b]pyrazines derivatives or by a cocktail containing 1 μM forskolin + pyrrolo[2,3-b]pyrazines. Other details are elsewhere (Marivingt-Mounir et al., 2004).
Whole-Cell Patch-Clamp Recordings. Whole-cell patch-clamp experiments were performed on Calu-3 and WT-CFTR-CHO cells at room temperature. Currents were recorded with a RK-400 patch-clamp amplifier (Biologic, Grenoble, France). I-V relationships were built by clamping the membrane potential to -40 mV and by pulses from -80 mV to +80 mV (20-mV increments). Pipettes with a resistance of 3 to 4 MΩ were pulled from borosilicate glass capillary tubing (GC150-TF10; Clark Electromedical Inc., Reading, UK) using a two-step vertical puller from Narishige (Tokyo, Japan). They were filled with the following solution: 1 mM NaCl, 113 mM l-aspartic acid, 113 mM CsOH, 1 mM MgCl2, 27 mM CsCl, 1 mM EGTA, 10 mM TES, and 3 mM MgATP (pH 7.2; 285 mOsm). They were connected to the head of the patch-clamp amplifier through an Ag-AgCl pellet. Seal resistances ranging from 6 to 20 GΩ were obtained. Results were analyzed with the pClamp 6.0.2 package software (Molecular Devices, Sunnyvale, CA). The external bath solution contained 145 mM NaCl, 4 mM CsCl, 1 mM CaCl2, 1 mM MgCl2, 10 mM glucose, and 10 mM TES (pH 7.4; 340 mOsm with mannitol). The liquid potential was corrected before seal establishment. Pipette capacitances were electronically compensated in cell-attached mode. To standardize experiments, recordings were performed only when the input resistance had a value ≤10 MΩ. The mean value of access resistance was 7.3 ± 1.1 MΩ (n = 5) for Calu-3 cells and 7.9 ± 0.8 MΩ (n = 5) for WT-CFTR-CHO cells. Membrane capacitances were measured in the whole-cell mode by fitting capacitance currents obtained in response to a hyperpolarization of 6 mV, with a first-order exponential, and by integrating the surface of the capacitance current. Mean values of membrane capacitance were 52.2 ± 12.2 pF (n = 5) for Calu-3 cells and 63.7 ± 11.6 pF (n = 5) for WT-CFTR-CHO cells. For graphic representations, I-V relationship was normalized to 1 pF to remove variability due to differences in cell sizes. For time-course experiments, current amplitude measured at +40 mV was plotted each 15 s.
Short-Circuit Current Measurement. Experiments were carried out on the colon epithelium of wild-type (Cftr+/+) or CFTR knock-out (Cftr-/-) mice B6 129-CFTRtm1Unc (Snouwaert et al., 1992) obtained from CNRS-CDTA (Center de Distribution, Typage et Archivage Animal, Orléans, France). Animals were killed by cervical dislocation, a procedure approved by the local animal ethic committee of the University of Poitiers. We used one animal per experiment (n is the number of animals). The ascending colon was removed, opened longitudinally, and washed in ice-cold phosphate-buffered saline. The muscle layers and connective tissue were dissected away from the colon epithelium. The epithelium was mounted in a vertical Ussing chamber (0.26 cm2 surface area). Luminal and serosal sides were bathed at 37°C, with a nutrient buffer containing 120 mM NaCl, 1.2 mM CaCl2, 1.2 mM MgCl2, 0.8 mM K2HPO4, 3.3 mM KH2PO4, 25 mM NaHCO3, and 10 mM d-glucose. The pH value of this solution was 7.4 when gassed with 95% O2-5% CO2 at 37°C. The upper level of fluids in both luminal and serosal reservoirs was identical. After mounting tissues in Ussing chamber, an equilibration period ≥40 min, was allowed for stabilization of basal electrical properties. Transepithelial potential difference (VTE) was measured by the Ag/AgCl electrodes placed as close as possible to (and on either side of) the epithelium to reduce the magnitude of the solution series resistance. They were connected to the preamplifier headstage of an amplifier (EC-825 Epithelial Voltage-Clamp; Warner Instruments Corporation, Hamden CT). Potential difference values were corrected for the junction potential between the luminal and serosal solutions when no tissue was mounted in the chamber. The polarity of Isc and VTE was referred to serosal side of the epithelium. An apical anion secretion was indicated by a decrease in Isc. Transepithelial resistances (RTE) were determined according to Ohm's law, from the voltage deflection (ΔVTE) due to pulsed current injection (1 s) of 0.5 μA and subtraction of the empty chamber resistance. In our experiments, the range of colonic epithelia RTE was 580 to 2000 Ω · cm-2, and the means were 1113 ± 157 Ω · cm-2 (N = 20) for Cftr-/- mice colon and 1388 ± 251 Ω · cm-2 (N = 26) for wild-type mice colon (no significant difference). Data were collected with the Chart version 4.2.2 package software (ADInstruments Phy Ltd, Castle Hill, NSW, Australia). All experiments were carried out in the presence of amiloride (500 μM) in the apical solution to prevent sodium transport.
Data Analysis. All of the data are presented as mean value ± S.E.M., where n refers to the number of experiments and N to the number of animals. For the Isc experiments and iodide efflux experiments, the unpaired Student's t test was used to compare sets of data, whereas for [cAMP] measurements, the paired Student's t test was used (duplicate experiments). All graphs are plotted with GraphPad Prism 4.0 for Windows (GraphPad Software, San Diego, CA). Values of P < 0.05 were considered as statistically significant: *, P < 0.05; **, P < 0.01; ***, P < 0.001. Nonsignificant difference was P > 0.05.
Results
Our study was carried out with the 12 pyrrolo[2,3-b]pyrazines derivatives shown in Fig. 2A. These agents are named RP derivatives (see Table 1) to facilitate the reading. These compounds were selected from a library of more than 100 derivatives synthesized in our laboratory, because they share structural determinants with other CFTR activators (reviewed in Becq, 2006).
Identification of cAMP-Independent CFTR Activators. We first examined the effects of 100 μM of each compound on CHO cells stably expressing WT-CFTR with our robotic cell-based primary screening assay using iodide efflux measurement. This assay allowed the rapid detection of CFTR channel activity from cells cultured in 24-well plates. Several control experiments have been first performed on resting cells (noted ctrl for control, corresponding to 0%) or on cells stimulated with a submaximal concentration (1 μM) of the adenylate cyclase activator forskolin (hereafter noted fsk, corresponding to 100%, gray bar in Fig. 2B). This minimal concentration of fsk was chosen, because it allows low stimulation of CFTR-dependent iodide efflux as reported previously (Marivingt-Mounir et al., 2004). Among these compounds only RP107 and RP108 activated an iodide efflux above the control (Fig. 2B). The remaining 10 RP compounds did not stimulate iodide efflux (dashed line in Fig. 2B was set at the level of fsk response). To begin to characterize the mechanism of activation, we determined the cellular cAMP level in WT-CFTR-CHO cells in the following experimental conditions, with cells stimulated by 1 μM fsk alone or stimulated by 1 μM fsk plus 100 μM RP107 or plus 100 μM RP108 and with cells stimulated by 10 μM fsk. Figure 2C shows that neither RP107 nor RP108 potentiated the cAMP level measured with 1 μM fsk alone. In addition, the compounds alone have no effect on the basal cellular cAMP level.
We then determined the concentration-response relationships in presence of 1 μM fsk plus increasing concentrations of each RP in WT-CFTR-CHO cells (Fig. 3). The iodide secretion peaked and then fell rapidly. For RP107, the peak rate (kpeak) of iodide efflux occurred within the first 2 min after adding the compound. In control condition (DMSO), kpeak was 0.065 ± 0.003 min-1 in the presence of 1 μM fsk, and kpeak increased to 0.108 ± 0.005 min-1 and was significantly potentiated to 0.22 ± 0.008 min-1 after the addition of 100 μM RP107 (n = 4 for each condition). The decline of the efflux rate could be attributed either to tracer depletion (because of a finite source of 125I at the beginning of each experiment), local ATP depletion, or desensitization. With 1 μM fsk present, the maximal stimulation (plateau) was achieved at 3 μM (Fig. 3, A and B), and the half-maximal effective concentration was EC50 = 152 ± 1.2 nM for RP107 (Fig. 3B). For RP108, the corresponding EC50 value was much higher (24 ± 1.2 μM; Fig. 3C). No stimulation in the presence of either RP107 or RP108 was noted with mock-CHO cells, i.e., nonexpressing CFTR cells (not shown). Thus, RP107 selectively stimulates CFTR-dependent iodide efflux with submicromolar concentrations, whereas its analog RP108 seems approximately 158-fold less potent.
In the following experiments, we determined the pharmacological profile for inhibition of the RP-stimulated iodide efflux in WT-CFTR-CHO cells. The peak of iodide efflux stimulated by 1 μM fsk + 1 μM RP107 (kpeak = 0.207 ± 0.003 min-1, n = 8) was fully abolished by 100 μM glibenclamide (kpeak = 0.063 ± 0.003 min-1, n = 8) and 10 μM CFTRinh-172 (kpeak = 0.055 ± 0.003 min-1, n = 8), two well known CFTR inhibitors (Fig. 4A, right traces). In contrast, fsk/RP107 response was not affected by 100 nM TS-TM calix[4]arene (kpeak = 0.2 ± 0.004 min-1, n = 8) or 500 μM DIDS (kpeak = 0.213 ± 0.011 min-1, n = 8), two non-CFTR channel inhibitors (Fig. 4A, left traces). A summary of the data are presented in Fig. 4B with statistical analysis. We then determined the effect of RP107 in Calu-3 cells, a model for human airway epithelial cells endogenously expressing WT-CFTR (Shen et al., 1994; Dérand et al., 2004). Figure 5A shows stimulation by RP107 of iodide efflux in Calu-3 cells in the absence of fsk. As can be seen in Fig. 5A, the latency to reach the maximal efflux response (kpeak) was reduced at higher RP107 concentrations (t5 at 100-300 nM but t3 at 1-10 μM). The corresponding half-maximal effective concentration was EC50 = 303 ± 1.5 nM (n = 4; Fig. 5B, right). In similar experiment but in the presence of submaximal concentration of fsk (1 μM), RP107 activated the efflux with a greater affinity (EC50 = 140 ± 2.6 nM, n = 4; Fig. 5B, left). These results suggest that the stimulation of CFTR by RP107 depends on the phosphorylation state of the channels.
Activation of G551D- and F508del-CFTR by RP107. To determine whether RP107 could activate CFTR mutants, we performed iodide efflux experiments with cells expressing two of the most common CF mutations, i.e., the class III mutation G551D-CFTR studied with CHO cells stably expressing G551D-CFTR and the class II deletion F508del-CFTR studied in human airway epithelial CF15 cells endogenously expressing F508del-CFTR protein. The mutation G551D-CFTR disrupts activation of CFTR, and phosphorylation alone is not able to achieve sufficient stimulation of G551D-dependent Cl- transport (Welsh and Smith, 1993; Becq et al., 1994; Illek et al., 1999; Marivingt-Mounir et al., 2004). Iodide efflux experiments on G551D-CFTR CHO cells have been performed in the presence of 10 μM fsk. No stimulation of iodide efflux was found with 10 μM fsk alone (Fig. 6A). However, RP107 stimulates an efflux in G551D-CFTR expressing cells with an EC50 of 1.5 ± 1.1 nM in the presence of 10 μM fsk (n = 4; Fig. 6A). The latency to reach the maximal efflux response (t5-t6) was delayed compared with wild-type CFTR (see Fig. 5). This delay in response may be related to the gating defect of the G551D-CFTR channel (Welsh and Smith, 1993) and to its inability to response to high concentration of cAMP agonists (Becq et al., 1994; Illek et al., 1999; Marivingt-Mounir et al., 2004). To study F508del-CFTR, we used the human airway epithelial CF15 cells. Prior to the experiments, CF15 cells were incubated for 24 h at low temperature to rescue F508del-CFTR from endoplasmic reticulum retention (Denning et al., 1992), a maneuver allowing efficient rescue of functional F508del-CFTR in CF15 cells (Norez et al., 2006). RP107 stimulates F508del-CFTR with EC50 of 111 ± 2.2 nM in the presence of 1 μM fsk (Fig. 6B). The latency to reach the maximal efflux response in this case was not delayed as it was for G551D-CFTR but was similar to that of wild-type CFTR (t3 for WT-CFTR and F508del-CFTR). Interestingly, no stimulation of iodide efflux by 100 μM RP107 was obtained in CF15 cells cultured at 37°C (control: kpeak = 0.11 ± 0.010 min-1, n = 4; RP107; kpeak = 0.13 ± 0.011 min-1, n = 4).
Activation of a Linear Cl- Current by RP107 and Inhibition by CFTR Inhibitors in Calu-3 and WT-CFTR-CHO Cells. The iodide efflux data were complemented by whole-cell patch-clamp recordings to characterize the effect of RP107 in Calu-3 and WT-CFTR-CHO cells. To eliminate non-CFTR Cl- currents (for example outwardly rectifying Cl- currents), we performed all experiments in the presence of 100 nM TS-TM calixarene and 200 μM DIDS in the bath. Figure 7 presents typical whole-cell currents and associated current-voltage plots (Fig. 7D) recorded in the presence or absence of RP107 in the bath. In control experiments with unstimulated Calu-3 cells (noted control in Fig. 7A), a small linear current was recorded with a current density of 1.2 ± 0.5 pA/pF at +40mV. The addition of 1 μM RP107 to the bath stimulated a time-independent nonrectifying current (Fig. 7B), with a reversal potential of -40 ± 0.9 mV (n = 5), close to the theoretical Nernst potential equilibrium ECl of -41mV (Fig. 7D). The current density of the linear Cl--selective current activated by RP107 was 11.2 ± 2.7 pA/pF (n = 5, measured at +40 mV). The amplitude of the current was statistically different from the basal current (P < 0.001). In the absence of DIDS/calixarene in the bath, 1 μM RP107 stimulated similar linear Cl--selective current with a current density of 10.2 ± 1.7 pA/pF, not significantly (ns) different from experiments including inhibitors.
To further demonstrate that the chloride current activated by RP107 was due to CFTR, we added 100 μM glibenclamide to the bath after stable and maximal activation of the current (Fig. 7C). The current density significantly decreased in the presence of glibenclamide to 3.1 ± 1.5 pA/pF (n = 5, P < 0.001 compared with RP107 alone and not significant compared with control; Fig. 7D). Whole-cell patch-clamp experiments were also conducted in WT-CFTR-CHO cells. As expected, RP107 also stimulated CFTR currents. An example of the concentration- and time-dependent activation of CFTR Cl- currents (representative of five separate cells) is presented in Fig. 8. As for Calu-3 cells, we incubated the cells in a mixture of calixarene/DIDS (control in Fig. 8A). In this basal condition, the current amplitude was 103 pA at +40mV. Increasing the concentrations of RP107 in the bath from 10 nM to 10 μM activated a linear Cl- current (Fig. 8B), with an amplitude (at +40 mV) of 309 pA at 10 nM, 633 pA at 100 nM, and 1077 pA at 1 μM (Fig. 8A). At 1 μM RP107, the CFTR Cl- current was maximal; with 10 μM of the current amplitude remained unchanged (1064 pA at +40 mV). The current was fully and rapidly (140 pA at +40 mV after 2 min) inhibited after the addition to the bath of 10 μM CFTRinh-172 (in the presence of 10 μM RP107). Figure 8B presents the corresponding time course of activation for the current traces shown in Fig. 8A. These results show that, in WT-CFTR-CHO and Calu-3 cells, RP107 activates CFTR Cl- currents inhibited by CFTRinh-172 and glibenclamide.
Stimulation of the Cl- Secretion in the Proximal Colon of CFTR Wild-Type Mice but Not of CFTR Null Mice. Finally, we measured the effects of RP107 and RP108 on transepithelial ion transport in the mouse colon under short-circuit condition as described under Materials and Methods. In short-circuited colon of Cftr+/+ mice (Fig. 9A), the application of 500 μM amiloride inhibited the resting Na+ current set by the activity of the amiloride-sensitive ENaC channels (De Jonge et al., 2004). After the serosal application of 10 μM fsk, ΔIsc increased by 19 ± 3.6 μA · cm-2 (N = 7), which corresponds to Cl- secretion. After stable stimulation, the current was inhibited by glibenclamide (N = 7). In contrast, with Cftr-/- mice, whereas the application of 500 μM amiloride also inhibited the resting Na+ current in short-circuited colon, the serosal application of 10 μM fsk did not have any effect on Isc (ΔIsc = 0.3 ± 0.11 μA · cm-2, N = 4). No effect of 500 μM glibenclamide was observed (N = 4; Fig. 9B).
The experiments described below were conducted after inhibition of the ENaC current by 500 μM amiloride. The addition of RP107 to either apical and basolateral or basolateral only produced the same response in CFTR wild-type mice colon (Fig. 9C). Increasing concentrations of RP107 induced a change of Isc, as for the fsk-activated Isc, corresponding to an apical Cl- secretion. The maximal RP107-dependent response ΔIsc = 7.27 ± 2.4 μA · cm-2 (N = 8; Fig. 9C) was fully reversed after bilateral application of 1 mM glibenclamide (ΔIsc = 1.16 ± 1.45 μA · cm-2, N = 8, P < 0.001; Fig. 9C) but was not affected by 500 μM DIDS (ΔIsc = 7.47 ± 0.44 μA · cm-2, N = 3; data not shown). In contrast, in CFTR null mice colon, no change of Isc was obtained with RP107 used at the same concentrations (N = 8; Fig. 9D). Figure 9E indicates ΔIsc for each concentration of RP107 tested on the colon of Cftr+/+ and Cftr-/- mice (N = 8 animals for each). We determined an EC50 of 89 ± 6 nM for Cftr+/+ mice (N = 8; Fig. 9F). Similar experiments have been performed in the presence of 1 μM fsk to mimic the protocol of iodide efflux experiments. After the addition of amiloride, the tissue was first stimulated by 1 μM fsk and then by RP107. After stable stimulation by fsk of the apical Cl- secretion, the response was further potentiated by RP107, with an EC50 of 173 ± 36 nM (N = 6). Finally, with RP108, we found stimulation of a glibenclamide-sensitive apical Cl- secretion in Cftr+/+ but not Cftr-/- mice, with an EC50 of 103 ± 9 μM (N = 6).
Discussion
In the present study, we report on the discovery of 6-phenylpyrrolo[2,3-b]pyrazines, a novel family of wild-type CFTR, G551D, and F508del-CFTR activators. These agents have been identified with our CF drug discovery program using a simple and robust robotic cell-based assay, allowing the discovery of small correctors and potentiators of CFTR chloride channels. Two agents that we described here are RP107 and RP108. RP107 is the most potent with an affinity of 140 to 152 nM on wild-type CFTR (Calu-3 and WT-CFTR-CHO cells), 1.5 nM on G551D-CFTR, and 111 nM on temperature-corrected F508del-CFTR. The Cl- secretion was stimulated in proximal colon of mice, with an affinity of approximately 90 nM for RP107 and 103 μM for RP108. The mechanism of action, although still unknown, is independent of cAMP but depends on the phosphorylation state of CFTR, because these agents have a greater affinity for the phosphorylated channels. RP107 and RP108 are selective for CFTR because, in nonexpressing CFTR cells (mock-CHO cells), in F508del cells at 37°C (CF15 cells), and in the colon of Cftr-/- mice, these agents have no effect on iodide efflux, whole-cell Cl- currents, or Isc. Although we have no evidence for a direct activation of CFTR, it is reasonable to conclude that 6-phenylpyrrolo[2,3-b]pyrazines selectively activate CFTR but not other Cl- channels.
6-Phenylpyrrolo[2,3-b]pyrazines (Aloisines). This family of compounds has been first described as CDK and GSK-3 inhibitors (Mettey et al., 2003). Aloisines act by competitive inhibition of ATP binding to the catalytic subunit of the kinases and may be of therapeutic value in Alzheimer's disease because implication of CDK5 and GSK-3 has been suggested in the disease (Knockaert et al., 2002). All of the compounds listed in Fig. 2 and tested on CFTR function are active on CDK1, 2, and 5 at submicromolar concentrations (Mettey et al., 2003), but not all of them are CFTR activators. This indicates that CFTR activation and CDK inhibition are probably not related, which also suggests that the design of a CFTR-specific activator may be achieved after careful determination of the structural determinants of Aloisines.
Structural Determinants for CFTR Activation. To begin to define a structure-activity relationship with 6-phenylpyrrolo[2,3-b]pyrazines tailored to CFTR activation, we compared the chemical structure of RP107 and RP108 with the 10 inactive related derivatives (Fig. 2A) and with several known CFTR activators. First, within the 6-phenylpyrrolo[2,3-b]pyrazines family, we observed that RP108 bearing a 4-chlorophenyl substituent is less potent than RP107 having 4-hydroxyphenyl substituent. In addition, 4-methoxyphenyl compounds (RP11, RP95, RP127, RP106, and RP149) are inactive. Therefore, one favorable determinant for CFTR activation is the phenyl ring substituted at the 4-position, with the following potency: OH>Cl>>OCH3. Interestingly, a hydroxyphenyl substituent appears also for NS004, NS1619, genistein, apigenin, and benzo[c]quinolizinium compounds (reviewed in Becq, 2006). A chlorine atom is also commonly found for NS004, benzo[c]quinolizinium, and 1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid diamides (Gribkoff et al., 1994; Marivingt-Mounir et al., 2004; Hirth et al., 2005). Some agents possess both OH and Cl on the same ring (NS004) or on different rings (MPB-91 and MPB-104). In a previous structure-activity relationship of the CFTR activators, benzo[c]quinolizinium salts (Marivingt-Mounir et al., 2004), we observed that alkyl substituent and especially butyl chain improved the efficacy of the drugs. Interestingly, the substitution of the pyrrolopyrazines core with butyl in R1 (Fig. 1) is also present in RP107 and RP108 (Fig. 2A). Therefore, a second favorable determinant for CFTR activation is the length of the alkyl chain with the following potency: (CH2)3-CH3>> H ≈ CH3 ≈ (CH2)2-CH3 ≈ (CH2)4-CH3 as found in compounds RP107, RP108 >> RP26 ≈ RP96 ≈ RP132 ≈ RP150. Finally, compounds RP26 having the OH-phenyl but not the butyl chain, RP106 having the butyl chain but not the OH-phenyl, or RP14 having the Cl-phenyl but not the butyl chain are not activators. These observations suggest that both determinants (OH-phenyl and butyl chain) are required for CFTR activation. These very preliminary data provide some interesting clues to improve the potency of future 6-phenylpyrrolo[2,3-b]pyrazines.
Toward a New Scaffold Structure for CFTR Activation. The CFTR activators of the first generation were originally identified after investigations of the intracellular signaling pathways (phosphodiesterases and phosphoprotein phosphatases) (reviewed in Becq 2006). This field has profoundly changed the last decade with the achievement of specific and new screening assays developed by universities (Chappe et al., 1998; Galietta et al., 2001; Ma et al., 2002; Sammelson et al., 2003; Yang et al., 2003; Marivingt-Mounir et al., 2004; Szkotak et al., 2004) and biopharmaceutical companies (Hirth et al., 2005; Van Goor et al., 2006).
Several commercial agents, such as phenanthrolines and benzoquinolines, have been reported to activate epithelial Cl- secretion with EC50 from 612 to 34 μM (Duszyk et al., 2001; Cuthbert, 2003; Cuthbert and MacVinish, 2003). The 4-chloro-benzo[f]isoquinoline has an EC50 of 4 μM in Calu-3 cells (Murthy et al., 2005). Introducing a halogen atom (Cl) in position 4 increased the efficacy of the agent compared with other benzoisoquinolines (Szkotak et al., 2004; Murthy et al., 2005). A similar effect after the addition of a halogen atom has been noted earlier for benzoquinoliziniums (Marivingt-Mounir et al., 2004). Other agents were discovered by high-throughput screening (Galietta et al., 2001; Ma et al., 2002; Sammelson et al., 2003; Yang et al., 2003), like for example 3-(2-benzyloxyphenyl)isoxazoles and 3-(2-benzyloxyphenyl)-isoxazolines (Sammelson et al., 2003). Agents with tetrahydrocarbazole, hydroxycoumarin, and thiazolidine core structures reversibly activated CFTR with Kd as low as 200 nM (Ma et al., 2002). It is noteworthy that trifluoromethylphenylbenzamine activated G551D-CFTR channels with a Kd > 10 μM (Ma et al., 2002). Stimulation of F508del-CFTR activity was also reported with tetrahydrocarbazole and N-phenyltriazine with Kd below 1 μM (Ma et al., 2002). Further high throughput screening efforts also identified 1,2,3,4-tetrahydroisoquinoline-3-carboxylic acid diamides as CFTR modulators with nanomolar potency (Hirth et al., 2005) and quinazolinone from a library of 164,000 synthetic compounds (Van Goor et al., 2006).
In conclusion, we have identified 6-phenylpyrrolo[2,3-b]pyrazines as a new scaffold structure for the selective activation of CFTR, with submicromolar affinity on wild-type CFTR, G551D-, and F508del-CFTR, two of the most common CF mutations. Although the mechanism of activation is still unknown, the challenge will be to design more potent CFTR-specific activators and to determine their potential and optimal use for therapeutic application in cystic fibrosis and CFTR-related diseases.
Acknowledgments
We thank Nathalie Bizard, Patricia Léon and James Habrioux for excellent technical assistance, Drs. Singh and Bridges for generous gift of TS-TM calix[4]arene and Laurent Meijer for discussions.
Footnotes
-
This work was supported by the French Association Vaincre la Mucoviscidose (VLM), ABCF Proteins, and CNRS. S.N. was supported by a thesis studentship from MucoVie 66. C.N. was supported by a thesis studentship from VLM. C.F. was supported by a specific grant from CNRS. Part of this work has already been published in an abstract form, in Noel S, Mettey Y, Norez C, Rogier C, and Becq F (2005) Activation of CFTR-dependent chloride secretion by a novel activator with nanomolar affinity. Pediatr Pulmonol39 (Suppl. 28):205.
-
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
-
doi:10.1124/jpet.106.104521.
-
ABBREVIATIONS: CF, cystic fibrosis; CDK, cyclin-dependent kinase; CFTR, cystic fibrosis transmembrane conductance regulator; CFTRinh-172, 3-[(3-trifluoromethyl)phenyl]-5-[(4-carboxyphenyl)methylene]-2-thioxo-4-thiazolidinone; CHO, Chinese hamster ovary; DIDS, 4,4′-diisothiocyanatostilbene-2,2′-disulfonic acid; DMSO, dimethyl sulfoxide; ENaC, epithelial sodium channel; fsk, forskolin; GSK-3, glycogen synthase kinase-3; RP107, 7-n-butyl-6-(4-hydroxyphenyl)[5H]pyrrolo[2,3-b]pyrazine; RP108, 7-n-butyl-6-(4-chlorophenyl)[5H]pyrrolo[2,3-b]pyrazine; RTE, transepithelial resistance; RP11, 6-(4-methoxyphenyl)[5H]pyrrolo[2,3-b]pyrazine; RP26, 6-(4-hydroxyphenyl)[5H]pyrrolo[2,3-b]pyrazine; RP95, 6-(4-methoxyphenyl)-7-methyl[5H]pyrrolo[2,3-b]pyrazine; RP96, 6-(4-hydroxyphenyl)-7-methyl[5H]pyrrolo[2,3-b]pyrazine; RP127, 6-(4-methoxyphenyl)-7-propyl[5H]pyrrolo[2,3-b]pyrazine; RP132, 6-(4-hydroxyphenyl)-7-propyl[5H]pyrrolo[2,3-b]pyrazine; RP106, 7-n-butyl-6-(4-methoxyphenyl)[5H]-pyrrolo[2,3-b]pyrazine; RP149, 7-n-pentyl-6-(4-methoxyphenyl)[5H]pyrrolo[2,3-b]pyrazine; RP150, 7-n-pentyl-6-(4-hydroxyphenyl)[5H]pyrrolo[2,3-b]pyrazine; RP14, 6-(4-chlorophenyl)[5H]pyrrolo[2,3-b]pyrazine; TES, N-tris[hydroxymethyl]methyl-2-aminoethane-sulfonic acid; TS-TM calix[4]arene, 5,11,17,23-tetrasulfonato-25,26,27,28-tetramethoxy-calix[4]arene; VTE, transepithelial potential difference; WT, wild type; FBS, fetal bovine serum; T3, 3,3′,5-triiodo-l-thyronine sodium salt; NS004, (5-trifluoromethyl-(5-chloro-2-hydroxyphenyl)-1,3-dihydro-2H-benzimidazol-2-one; NS1619, (1-(2′-hydroxy-5-trifluoromethylphenyl)-5-trifluoromethyl-1,3-dihydro-2H-benzimidazole-2-one; MPB-91, 5-butyl-10-chloro-6-hydroxybenzo[c]quinolizinium chloride; MPB-104, 5-butyl-7-chloro-6-hydroxybenzo[c]quinolizinium chloride.
- Received March 15, 2006.
- Accepted July 6, 2006.
- The American Society for Pharmacology and Experimental Therapeutics