Introduction

Abstract Introduction Materials and Methods Results Discussion References

Cystic fibrosis (CF) is an inherited disease in which mutations in the gene encoding the CF transmembrane conductance regulator (CFTR) (1) lead to inflammation, bacterial colonization of the lung, and progressive deterioration in lung function. One function of the CFTR protein is as a cyclic adenosine monophosphate (cAMP)-regulated chloride channel. The link between this and the worsening cycles of infection and inflammation in the lungs in CF is unclear (2), but the disease carries a current life expectancy of approximately 30 yr. Both genetic and pharmacologic strategies are being investigated (7, 8) that either separately or in combination might restore sufficient CFTR activity, although the degree of correction required for clinical benefit remains speculative (9).

Several strategies for the pharmacologic augmentation of CFTR function are being investigated. One of these is the treatment of respiratory epithelia with phosphodiesterase (PDE) inhibitors, already in use in humans for the treatment of asthma. Currently, seven families of PDEs have been identified (10). Broad-spectrum inhibition can be achieved with isobutylmethylxanthine (IBMX), and this agent has been shown in some studies to induce chloride secretion in CF cells in vitro (11). However, we have been unable to demonstrate induction of chloride secretion in the cftr^TgHm1G551D mutant mouse (12) in response to IBMX applied to the nasal epithelium in vivo. This is in keeping with the findings in another in vivo study with human nasal epithelium (13).

Type III PDEs have been shown to be active in human airway epithelium (14), and milrinone, a specific inhibitor of type III PDEs (15) and an agent already in clinical use (16), has been reported to induce chloride secretion in cells (Calu-3 and 16HBE) derived from non-CF human airways (17). Interestingly, in a cell line (CF-T43) derived from a CF nasal polyp carrying the F508 mutation in CFTR, milrinone also induced chloride secretion when administered in combination with a -agonist, although not when applied alone (18). Furthermore this effect of milrinone was inhibited by antisense oligonucleotides to CFTR. The investigators who made this finding have also reported that milrinone (in combination with a -agonist) induced chloride secretion in the nasal epithelium of F508-mutant CF mice when applied in vivo (19).

We further investigated the therapeutic potential of this selective PDE inhibitor, both in mice and in human subjects. We assessed the effect of milrinone on two types of CFTR mutations: F508, both because of its frequency of occurrence and as an example of a nontrafficking mutation, and G551D, in which substantial amounts of CFTR protein reach the cell surface (20). Our studies were done in vivo on murine and human CF nasal epithelium, and in the lungs of human subjects with CF.

	Materials and Methods

Abstract Introduction Materials and Methods Results Discussion References

Solutions

The standard perfusing solution used in the murine in vivo measuring system was 4-(2-hydroxyethyl)-1-piperazine- N'-2-ethanesulfonic acid (Hepes)-containing Kreb's solution (HK) consisting of: 10 mM Hepes, 140 mM Na⁺, 152 mM C1, 6 mM K⁺, 1 mM Mg²⁺, 2 mM Ca²⁺, and 10 mM glucose in deionized water (pH 7.4). When an additional driving force was required, the perfusate was changed to a low-chloride (6 mM) HK by equimolar substitution with gluconate. HK was also used for nasal measurements made on human subjects. For lung measurements, Ringer's solution of composition 147 mM Na⁺, 155 mM Cl, 4 mM K⁺, 2 mM Ca²⁺, or with gluconate substitution was used.

Drugs

All drugs were obtained from Sigma Aldrich (Poole, UK), except for amiloride, which was a gift from Merck, Sharp, and Dohme (Hoddesdon, Herts, UK), and milrinone (Sanofi Winthrop, Guildford, UK). Drugs were used at the following final concentrations: amiloride, 100 µM; forskolin, 10 µM; adenosine triphosphate (ATP), 100 µM; milrinone, 100 µM; and isoproterenol, 100 µM.

Murine Studies

Animals. Non-CF mice (MF1, Harlan, UK; male, weighing 21-35 g, cftr^+/+) and CF mice (cftr^TgHm1G551D [21] and cftr^tm2Cam [22]) ranging from 15-25 g were housed at 16°C and allowed food (Special Diet Services, Witham, Essex, UK) and water ad libitum. No special conditions were provided for the CF mice besides corncob bedding. Animals were anesthetized intraperitoneally with one part fentanyl fluanisone (Hypnorm; Janssen Animal Health, Oxford, UK), one part midazolam (Hypnovel; Roche, Welwyn Garden City, UK), and two parts water for injection (10 ml/ kg). The temperature of the animals was maintained with an electrically heated plate, and was monitored throughout the experiments. After the study procedure, the animals were maintained in a heated cage until recovery.

Measuring systems. Potential difference (PD) measurements were made with a fine, double-lumen polyethylene catheter (outer diameter = 0.5 mm). One lumen conveyed perfusate via a peristaltic pump (Pharmacia, Cambridge, UK) at a rate of 21 µl/min, and the other served as an exploring electrode connected via a calomel electrode (Russell pH Ltd., Auchtermuchty, Scotland, UK) to a hand-held computer (Psion, London, UK) containing a low-pass signal-averaging filter with a time constant of 0.5 s (Logan Research Ltd., Sussex, UK). A reference electrode was installed subcutaneously in the hindlimb of the mouse undergoing PD measurement and was similarly connected to the computer. The circuit was validated with a measurement of buccal PD prior to insertion of the catheter, acceptable values being 10 to 20 mV. The junctional potential between HK and low-chloride HK was ignored because of the paired comparisons used in this part of the study. According to the experiment, the sequence of drug administration was: (1) amiloride, low-chloride HK, and milrinone; (2) amiloride, low-chloride HK, milrinone, and forskolin; (3) amiloride, milrinone plus forskolin in low-chloride HK, and ATP; and (4) amiloride, low-chloride HK, milrinone plus forskolin, and ATP.

Studies of Human Subjects

Subjects. Healthy nonsmoking subjects with no history of respiratory disease were recruited as normal controls. All subjects with CF fulfilled accepted criteria for diagnosis of the disease, including characteristic clinical findings and an abnormal sweat test. The protocols were approved by the local hospital ethics committees.

Nasal PD measurements. Measurement of nasal PD has been described previously (23). Briefly, a size 8 modified double-lumen Foley catheter was connected via a silver/ silver chloride wire to a high-impedance voltmeter. Values were measured with reference to a silver/silver chloride electrode placed on the abraded skin of the forearm. Measurements were made along the floor of the nasal cavity, with the maximum PD being recorded and drugs perfused at this site at a rate of 4 ml/min. The effects of perfusion of low-chloride solution and drugs were assessed after 5 min (23, 24). The protocols were: (1) milrinone, amiloride, and low-chloride HK-isoproterenol; and (2) amiloride, low-chloride HK, and milrinone plus isoproterenol. All PD measurements are reported as absolute values.

Lower airway PD measurements. Subjects were studied under general anesthesia (without the use of lignocaine). PD was measured with a double-lumen catheter of outer diameter 2.8 mm, passed through the biopsy channel of a bronchoscope; the reference electrode consisted of an intravenously placed cannula. Drug perfusion was done via a peristaltic pump, at a rate of 700 µl/min. Responses were measured at two sites, first in the segmental airways and subsequently at the carina. Drugs were perfused in the following sequence: low-chloride Ringer's solution, amiloride, isoproterenol, ATP, and milrinone.

Statistics

Paired comparisons were made with the Mann-Whitney U test, and comparisons within subjects with the Wilcoxon's rank-sum test. The null hypothesis was rejected at P < 0.05. Data are shown as means ± SEM for convenience.

	Results

Abstract Introduction Materials and Methods Results Discussion References

Murine Nasal Measurements

cftr^+/+ mice. Milrinone induced a small hyperpolarization of the in vivo nasal PD of cftr^+/+ mice of 1.6 ± 0.6 mV (n = 8, P < 0.05) in the presence of amiloride and a low-chloride driving force, but before addition of forskolin (Figure 1a). Amiloride produced a depolarization of 12.8 ± 1.6 mV, and the subsequent low-chloride solution a hyperpolarization produced of 16.0 ± 1.1 mV with n = 8 in both cases.

View larger version (12K): [in this window] [in a new window]   Figure 1.   Typical responses of nasal epithelium of (a) wild type, (b) cftrtm2Cam, and (c) cftrTgHm1G551D mice to stimulation with milrinone (100 µM, MIL) in vivo. Animals were pretreated with amiloride (100 µM, A) and subsequently perfused with a low-chloride (6 mM, LC) solution. Forskolin (10 µM, F) was added subsequently (a) and simultaneously with milrinone (b and c). In the CF traces, the responses to ATP (100 µM) are also shown.

cftr^tm2Cam mice. No hyperpolarization was seen in the nasal PD of cftr^tm2Cam mice (six measurements on three animals) in response to milrinone when the latter was added simultaneously with forskolin in the presence of amiloride and a low-chloride driving force. A representative trace is shown in Figure 1b.

cftr^TgHm1G551D mice. As with cftr^tm2Cam mice, no hyperpolarization was observed in mice with the G551D trafficking mutation. This was the case whether milrinone was added concomitant with (n = 4) or subsequent to (n = 5) perfusion of a low-chloride solution. In both cases milrinone was added with forskolin in the presence of amiloride. A representative trace is shown in Figure 1c.

Human Nasal Measurements

Non-CF subjects. Milrinone, when perfused in the absence of other agents (n = 6), did not significantly alter baseline PD (preperfusion PD, 14.7 ± 4.0 mV; postpersuion PD, 15.3 ± 4.6 mV), nor did it alter the subsequent depolarization induced by amiloride (without milrinone, PD 8.1 ± 1.3 mV; with milrinone, PD 10.0 ± 3.6 mV). However, milrinone significantly (P < 0.05) increased the hyperpolarization resulting from perfusion with a low-chloride solution, but without further effect on the subsequent response to isoproterenol (Figure 2).

View larger version (16K): [in this window] [in a new window]   Figure 2.   Responses of nasal epithelium of non-CF subjects in vivo to perfusion with a low-chloride (6 mM, LC) solution and isoproterenol (100 µM, ISO) in the presence of milrinone (100 µM, squares) or without milrinone (circles). All subjects (n = 6 in each group) were pretreated with amiloride (100 µM). Error bars indicate SEM.

F508 CF subjects. Two separate protocols were assessed. First, to allow direct comparison with non-CF subjects, the protocol described previously was used (n = 6). Milrinone significantly (P < 0.05) increased the baseline PD (preperfusion PD, 52.2 ± 3.3 mV; postperfusion PD, 57.4 ± 4.2 mV). There was a similar increase in the response to amiloride (without milrinone, 26.1 ± 2.4, mV; with milrinone, 35.0 ± 3.9 mV), although this did not reach significance. However, in the presence of milrinone there was no significant effect on the responses to a low-chloride solution or isoproterenol (Figure 3). In a second protocol milrinone was added together with isoproterenol following amiloride and low-chloride perfusion; again no induction of chloride secretion was seen (Figure 4a).

View larger version (17K): [in this window] [in a new window]   Figure 3.   Responses of nasal epithelium of CF subjects in vivo to perfusion with a low-chloride (6 mM, LC) solution and isoproterenol (100 µM, ISO) in the presence of milrinone (100 µM, squares) or without milrinone (circles). All subjects (n = 6 in each group) were pretreated with amiloride (100 µM). Error bars indicate SEM. Values at 5 min following addition of either LC or ISO provide the optimal discrimination between CF and non-CF subjects (see Figure 2).

View larger version (13K): [in this window] [in a new window]   Figure 4.   Milrinone (100 µM, MIL) and isoproterenol (100 µM, ISO) perfused simultaneously following amiloride (100 µM) and a low-chloride (6 mM, LC) solution into the nasal epithelium of (a) F508 subjects (n = 6) and (b) G551D subjects (n = 4). Error bars indicate SEM.

G551D CF subjects. CF subjects with at least one G551D allele (n = 4) were also assessed, using the same protocol as for subjects with the F508 mutation, since with increased amounts of CFTR reaching the apical membrane, chances of detecting any effects of milrinone in these patients should be maximized. However, no responses to the drug were seen (Figure 4b). Interestingly, a response to a low-chloride solution of approximately 5 mV was noted in these patients.

Human Lung Measurements

The effect of milrinone perfusion was assessed in vivo at two levels in the lower airways of CF subjects. Following the sequential addition of a low-chloride (6 mM) solution, amiloride, isoproterenol, and ATP, milrinone did not produce any changes in PD at either the carina (n = 6; Figure 5a) or in segmental airways (n = 6; Figure 5b).

View larger version (16K): [in this window] [in a new window]   Figure 5.   Effects of milrinone (100 µM, MIL) on (a) carinal lower airway potential difference (PD) (n = 6) and (b) segmental airway PD in vivo in CF subjects (n = 6). Prior perfusion included a low-chloride (6 mM) solution, amiloride (100 µM), isoproterenol (100 µM), and ATP (100 µM). Error bars indicate SEM.

	Discussion

Abstract Introduction Materials and Methods Results Discussion References

Our data indicate that milrinone has no significant effect on the induction of chloride secretion in CF airways in vivo. This was the case for F508 and G551D mutations in both mice and human subjects. The latter mutation is of relevance, given that milrinone is postulated to activate apically located CFTR, and G551D allows normal trafficking of this protein to the apical membrane (20) .

A number of in vitro studies have suggested the potential use of milrinone to induce chloride secretion in CF tissues (17). In addition, one recent study, by Kelley and colleagues, demonstrated this effect in vivo in the nasal epithelium of cftr^tmKth mice (19). Although the reason for the difference between the latter report and our own data is not clear, a number of variables should be considered as possible sources of this difference. Inevitably, a number of technical differences existed in the in vivo measurements in animals in the two studies. We considered a number of these, including the rate of drug perfusion and the volume of dead space in the perfusion system, that will influence the initial drug concentration applied to the mucosa. We altered both of these to parallel the conditions described in the previous study, without being able to induce chloride secretion (data not shown).

A further difference in the two studies relates to the effects of low-chloride perfusion and subsequent administration of forskolin in wild-type animals. Our own data indicate a hyperpolarization of approximately 15 mV for the former, in comparison with approximately 5 mV reported by Kelley and colleagues. Furthermore, following this low-chloride stimulus, subsequent administration of forskolin in our hands produced approximately 3 mV of hyperpolarization, whereas Kelley and colleagues reported an approximately 10 mV response to forskolin in their cftr^+/+ animals. A number of groups, including our own, are currently investigating the effects of murine strain on electrophysiologic responses. It is clear that this variable can produce large differences in the patterns of drug stimulation seen, and may in part explain some of the differences between the two reports. However, we would emphasize the concordance between our murine and human data in the present report.

The mechanism of action of milrinone is currently unclear (25). Although studies have suggested a pathway of its action involving protein kinase A (14), there is clearly a discrepancy between its effects on chloride secretion and its effects on intracellular cAMP levels. In keeping with this, we saw an effect of milrinone on the low-chloride but not on the isoproterenol response of non-CF human subjects. Moreover, in the murine nose, even "classic" cAMP-mediated agents such as forskolin may not necessarily produce their action through CFTR (26). In our hands, the 3 mV hyperpolarization described previously (following a low-chloride response) did not discriminate CF from non-CF genotypes. Although patch-clamp studies in cultured airway epithelial cells have indicated an effect of milrinone on CFTR, further investigations on the single-channel basis of milrinone responses in native mammalian tissues may be warranted.

One obvious caveat is that we failed to detect a response below the sensitivity of our assay systems, given the inherent noise in such in vivo measurements. This is of relevance, given that studies have suggested that even 5% of wild-type CFTR function within every cell may be sufficient to prevent intestinal pathology in CF mice (9). Some indication of the sensitivity of the in vivo nasal PD assay may be provided by our study correlating cftr mRNA levels with CFTR function measured with the techniques described earlier. cftr^tm1Hgu/cftr ^tm1Unc compound heterozygotes have a predicted wild-type mRNA level of 5%, and showed an in vivo nasal response to low-chloride perfusion that was approximately 40% of normal. We therefore suggest that had milrinone induced the equivalent level of wild-type CFTR function, it would very likely have been detected with our assay. However, we were able to demonstrate the effect of milrinone in non-CF airways, suggesting both intracellular access of the drug to the cell surface and validation of our measurement system with respect to the magnitude of the effect produced by milrinone. We were also able to show a significant effect of milrinone on the baseline PD in the nasal epithelium of the CF subjects. However, despite providing the airways with a driving force many times greater than is likely to be seen under physiologic conditions, we were still unable to detect a chloride-secretory response. If the establishment of chloride secretion is critical in the treatment of CF, our studies do not provide support for milrinone as a therapeutic agent for this disease.