In Silico, Ex Vivo and In Vivo Studies of Roflumilast as a Potential Antidiarrheal and Antispasmodic agent: Inhibition of the PDE-4 Enzyme and Voltage-gated Ca++ ion Channels

The aim of the present study was to evaluate the possible gut inhibitory role of the phosphodiesterase (PDE) inhibitor roflumilast. Increasing doses of roflumilast were tested against castor oil-induced diarrhea in mice, whereas the pharmacodynamics of the same effect was determined in isolated rabbit jejunum tissues. For in silico analysis, the identified PDE protein was docked with roflumilast and papaverine using the Autodock vina program from the PyRx virtual screening tool. Roflumilast protected against diarrhea significantly at 0.5 and 1.5 mg/kg doses, with 40% and 80% protection. Ex vivo findings from jejunum tissues show that roflumilast possesses an antispasmodic effect by inhibiting spontaneous contractions in a concentration-dependent manner. Roflumilast reversed carbachol (CCh, 1 µM)-mediated and potassium (K+, 80 mM)-mediated contractile responses with comparable efficacies but different potencies. The observed potency against K+ was significantly higher in comparison to CCh, similar to verapamil. Experiments were extended to further confirm the inhibitory effect on Ca++ channels. Interestingly, roflumilast deflected Ca++ concentration–response curves (CRCs) to the right with suppression of the maximum peak at both tested doses (0.001-0.003 mg/mL), similar to verapamil. The PDE-inhibitory effect was authenticated when pre-incubation of jejunum tissues with roflumilast (0.03-0.1 mg/mL) produced a leftward deflection of isoprenaline-mediated inhibitory CRCs and increased the tissue level of cAMP, similar to papaverine. This idea was further strengthened by molecular docking studies, where roflumilast exhibited a better binding affinity (-9.4 kcal/mol) with the PDE protein than the standard papaverine (-8.3 kcal/mol). In conclusion, inhibition of Ca++ channels and the PDE-4 enzyme explains the pharmacodynamics of the gut inhibitory effect of roflumilast.


Introduction
Gastrointestinal (GI) motility plays an important role in the digestive and absorptive processes of the gut, essential for pushing intestinal material, mixing it with digestive juices, and preparing undigested foods for excretion. Irritable bowel syndrome (IBS) is a gastrointestinal disease related to intestinal motility problems and can be associated with pain, constipation, or diarrhea [1]. Diarrhea, characterized by an increased frequency of bowel movements, wet stool, and abdominal cramps, is a Table 1. Antidiarrheal activity of roflumilast on castor oil (10 mL/kg)-induced diarrhea in mice.

Molecular Docking Analyses
Furthermore, to understand the PDE-4 inhibition, roflumilast and papaverine were docked into the active pockets of PDE-4B and PDE-4D proteins having PDB ID 5WH5 and 5LAQ, respectively. Roflumilast exhibited better binding affinity (−9.4 and −9.3 Kcal/mol) in comparison to papaverine (−8.3 and −8.2 Kcal/mol) in both the isoforms, as shown in Table 2. Roflumilast made two significant hydrogen bonds with Gln615 and Asn567, and halogen bonds with Gln615 and Trp576 while papaverine interacted with the protein by two distinct hydrogen bonds with Tyr405 and Asn567, as shown in Figure 5A,B, respectively. The notable difference in the binding affinities of roflumilast and papaverine can be attributed to the hydrogen-halogen bonding of the former. The hydrogen bond apart from the halogen bond in roflumilast represents an interaction of halogen and hydrogen bond donors on the side chain of the PDE protein, which is considered substantially more favorable in comparison to a canonical hydrogen bond.

Discussion
In view of the wide therapeutic potential of roflumilast, we previously reported studies showing its effectiveness in the treatment of inflammatory conditions and related diseases such as cardio and renal toxicities [20,21]. The current study was therefore planned to test the additional possible therapeutic potential of roflumilast in hyperactive disease states of the gut in mice, with detailed investigation of the mechanism(s) using rabbit jejunum in ex vivo experiments. Induction of diarrhea in normal mice was made possible using castor oil, which after hydrolysis into ricinoleic acid, led to the evocation of spasms in the gut [22]. Roflumilast pre-administration protected mice from diarrhea in a dose-mediated way, similar to the positive control drug, loperamide, a typical antidiarrheal agent [23].
After we observed the antidiarrheal response, our second objective was to determine the possible pharmacodynamics involved in this effect. For this purpose, the method of Bashir et al. [24] was followed by testing cumulative concentrations of roflumilast in an isolated, spontaneously contracting jejunum. Interestingly, roflumilast exhibited complete inhibition, thus showing an antispasmodic effect and further supporting our in vivo observations. Based on our previous observations that antispasmodic agents mediate gut inhibitory effects by blockade of Ca++ channels [23], potassium channel activation [25] and/or PDE inhibition [26], we further tested roflumilast on provoked contractions in rabbit jejunum with the respective contractile agents of high K+, low K+ and carbachol [27]. Critical analysis of the patterns of the inhibitory CRCs of roflumilast against high K+ and CCh shows that it exhibited statistically higher potency against high K+ in comparison to CCh, a typical property of Ca++ channel blockers (CCBs) [28]. Our positive control drug, verapamil, a known voltage-gated Ca++ inhibitor [29], also reversed high K+ with higher potency than CCh, thus further strengthening our findings. No significant difference was observed in the potency of roflumilast against high K+ and low K+ (data not shown), thus ruling out the possible role of K+ channel activation as one of the antispasmodic mechanisms of roflumilast. A test sample that selectively reverses low K+ (<30 mM)-mediated spasms depicts a K+ channel opener, while CCBs relax both low and high K+-mediated spasms with equal potencies, hence these investigations differentiate K+ channel openers from CCBs [30,31]. To further support and confirm the CCB-like action, roflumilast was pre-incubated in jejunal tissues stabilized in Ca++-free buffer solution. Contractile CRCs of Ca++ were measured in jejunal tissues in the absence and presence of pre-incubation with increasing doses of roflumilast. Roflumilast deflected Ca++ CRCs to the right with suppression of the maximum peak, similar to verapamil, thus confirming its CCB-like effect. Roflumilast, being known for PDE inhibition [19], was tested indirectly for PDE inhibition and elevation in the cAMP levels as PDE inhibitors, by blocking PDE, cause an elevation in tissue cAMP levels. PDE is responsible for converting cAMP to its inactive form (AMP), thus resisting smooth muscle relaxation [32]. Roflumilast, however, completely reversed CCh-mediated contractions with more potency compared to its effect against high K+, thus pointing toward PDE inhibition as an additional gut inhibitory mechanism of roflumilast. It is well known that PDE inhibitors inhibit CCh-mediated spasms of smooth muscles [33]. The potentiation of the inhibitory CRCs of isoprenaline towards lower doses (leftward) in jejunal tissues pre-incubated with roflumilast indirectly confirms its PDE-inhibitory character which was further confirmed when roflumilast pre-incubation causes increase in the levels of cAMP in jejunal tissues, similar to papaverine, a standard PDE inhibitor [34,35].
The PDE inhibition concept was further proved concretely by the in silico molecular docking results. The molecular docking study helped us to gain an insight into the protein-ligand interaction in the active domain of PDE-4. PDE-4 is reported to have four subtypes (A, B, C and D) each with the potential to degrade cAMP, which is the second messenger to react to the inflammatory effect. Roflumilast is a selective and non-specific PDE4 inhibitor that is able inhibit all PDE-4 isoforms. Despite its pharmacological effects, roflumilast can cause side-effects, such as diarrhea, nausea and dizziness, due to a lack of specificity. As PDE-4B and PFDE-4D have been found to be more expressed and extensively studied, hence were selected for the present docking study. The standard drug papaverine exhibited two hydrogen bonding affinities, while roflumilast exhibited better binding affinity with two hydrogen bond and two hydrogen-halogen bonds. Halogen and hydrogen-halogen bonding has been considered more favorable, and this is why halogen-containing compounds are being further explored and believed to be dual agonists [36].

Animals
Rabbits (1-1.5 kg) procured from a local farm in Al-Kharj and Swiss albino mice (20-25 g) from the Animal Care Unit, College of Pharmacy, Prince Sattam bin Abdulaziz University, Saudi Arabia, were kept at optimum temperature (22 ± 1 • C), relative humidity 55% ± 5%, and equal exposure to light/dark cycle. All animals were provided with a standard pellet diet and water ad libitum. Before sacrifice, rabbits fasted for 24 h. Cervical dislocation was performed by a blow on the neck, and death was confirmed by the disappearance of ear reflexes. Due care was taken, and all experiments (in vivo and ex vivo) were performed by following the instructions detailed in NRC [37].

In Vivo Experiments
Twenty mice were randomly allocated into four groups with equal numbers in each group. Following twenty-four-hour fasting, mice in the 1st group were exposed to oral gavage of saline (10 mL/kg) and were labeled as a negative control. After pilot screening for dose selection, the 2nd and 3rd groups (test groups) were exposed to two increasing dosages of 0.5 and 1.5 mg/kg of roflumilast, respectively. The last group (4th) of mice were given loperamide (10 mg/kg) and labeled as a positive control. Separate cages were assigned to each animal, with a blotting sheet on the floor of each cage for observation of the absence or presence of diarrhea by a blind observer. After an hour, all mice were orally exposed to castor oil (10 mL/kg) using a 1 mL syringe. After 4 h, all individual cages' blotting sheets were inspected for typical diarrheal droppings. Protection was noted if no diarrheal drops were observed, as previously reported by Jebunnessa et al. (2009) [38].

Antispasmodic Effect
The method of Shah et al. (2011) [39] was followed to sacrifice the rabbit and to isolate the middle part of the small intestine (jejunum). Following isolation, the required segments of jejunum 2-3 cm in length were cleaned of any adjacent tissues and fecal material, and were mounted in an emkaBath (France) attached with a transducer and IOX software. Fresh Tyrode's solution was filled in 20 mL tissue baths gassed with carbogen, and the temperature was set at 37 • C. The composition of Tyrode's solution was followed as previously described by Khan et al. (2018) [40]. Tension of 1 g was applied by rotating the transducer knob clockwise, and the tissues were left for stabilization for 30 min with multiple exposures of acetylcholine (0.3 µM). After spontaneous contractions of the jejunum were found with a stable band, roflumilast was added to the bath solution in increasing concentrations.
After observing the inhibitory effect of roflumilast on spontaneous contractions, experiments were further extended to determine the mechanism(s) of its antispasmodic effect. The inhibitory effect of roflumilast was tested against sustained spasms mediated by K+ (80 mM) and CCh (1 µM) for multiple pharmacodynamics including voltage-gated Ca++ channel blockade and inhibition of PDE. According to Godfraind et al. (1986) [41], K+ (>30 mM) depolarizes multiple smooth muscles by causing activation of Ca++ channels (L-type), thus causing sustained contractions. Similarly, any drug that reverses CCh-mediated contractions is considered as antimuscarinic and/or a PDE inhibitor [42,43].

Ca++ Channel-Blocking Activity
Following the preliminary inhibitory effect of roflumilast against high K+, confirmation of Ca++ channel blockade was determined by 30 min incubation of jejunal tissues in Ca++-free Tyrode's solution with EDTA (0.1 mM) for Ca++ chelation. To further deplete the intracellular stores of Ca++, the method of Blattner et al. (1978) was followed [44]. Briefly, Ca++-free solution was exchanged with another solution labeled as K+-rich and Ca++-free Tyrode's solution with the following concentrations (mM): potassium chloride, 50; sodium chloride, 91.03; sodium dihydrogen phosphate dehydrate, 0.32; sodium bicarbonate, 11.9; magnesium chloride hexahydrate, 0.50; glucose, 5.05; and calcium chelating agent (EDTA), 0.1. At the end of 1 h of preservation of jejunal tissue in this solution, Ca++ was added exogenously to produce dose-dependent CRCs in the absence and presence of two increasing doses of roflumilast. Parallel experiments were run using verapamil as a positive control.

PDE-Inhibitory Activity
As roflumilast showed efficacy against CCh, which is an indication of PDE inhibition, indirect confirmation was carried out by dose-mediated inhibitory curves of isoprenaline against CCh in the absence and presence of roflumilast. Potentiation of curves to the left is indicative of a blockade of PDE, while papaverine, a standard PDE inhibitor, was used for comparison [45].
Further confirmation of PDE inhibitory activity was done by biochemical method using cAMP enzyme immunoassay kit (ARBOR assays Detect X, direct cAMP enzyme immunoassay kit, Ann Arbor, MI, USA). Jejunal tissues, after contracting with CCh (1 µM), were exposed to relaxant doses of roflumilast and papaverine. Soon after the tissues reached to baseline, the tissues were snapped frozen in liquid nitrogen and were homogenized with 1 mL of sample diluent for every 100 mg of tissue, on ice, which was kept like this for 10 min. After centrifugation at ≥6000 g (4 • C for 15 min), the supernatant was collected and either analyzed immediately. Similar steps were repeated for tissues without administration of any drug and only contracted with CCh, which served as negative control. The cAMP content was expressed as picomole per mg protein.

In Silico Studies
To gain an insight into a PDE protein and its interaction with roflumilast, co-crystallized ligand and papaverine, molecular docking studies were carried out on two isoforms, namely PDE-4B and PDE-4D, using the Autodock vina program from the PyRx virtual screening tool [46,47]. The crystal structure of the PDE protein was downloaded from the protein data bank with PDB ID: 5LAQ and 5WH5 (resolution 2.4A and 1.4A, respectively), and it was further subjected to repairing and extraction of the co-crystallized ligand using Discovery studio visualizer (DSV) 4 [48]. The co-crystallized ligands, were redocked into the active domain of their respective proteins to validate the docking methodology used in the study. The 3D structure of roflumilast was downloaded from PubChem and with the help of open babel, it was minimized and converted to pdb format. DSV was used for visualization of the docked poses.

Statistical Analysis
Data are expressed as the mean ± standard error of the mean (SEM), where n is the number of experiments and EC50 represents the median effective concentration with 95% confidence interval (CI). Protection from diarrhea was statistically evaluated by comparing all groups with the saline control group using Chi-square (x 2 ) test. Statistical significance was considered only when p < 0.05. Graph Pad Prism (version 4) was used for the analysis of concentration-response curves (CRCs).

Conclusions
These findings show that roflumilast possesses antidiarrheal and antispasmodic effects, possibly mediated by blockade of voltage-gated Ca++ channels and PDE inhibition with resultant increase in cAMP levels, although the possibility of β2-adrenergic receptor and/or adenylyl cyclase activation mechanisms cannot be ignored.