Synthesis, Pharmacological Profile and Docking Studies of New Sulfonamides Designed as Phosphodiesterase-4 Inhibitors

Prior investigations showed that increased levels of cyclic AMP down-regulate lung inflammatory changes, stimulating the interest in phosphodiesterase (PDE)4 as therapeutic target. Here, we described the synthesis, pharmacological profile and docking properties of a novel sulfonamide series (5 and 6a-k) designed as PDE4 inhibitors. Compounds were screened for their selectivity against the four isoforms of human PDE4 using an IMAP fluorescence polarized protocol. The effect on allergen- or LPS-induced lung inflammation and airway hyper-reactivity (AHR) was studied in A/J mice, while the xylazine/ketamine-induced anesthesia test was employed as a behavioral correlate of emesis in rodents. As compared to rolipram, the most promising screened compound, 6a (LASSBio-448) presented a better inhibitory index concerning PDE4D/PDE4A or PDE4D/PDE4B. Accordingly, docking analyses of the putative interactions of LASSBio-448 revealed similar poses in the active site of PDE4A and PDE4C, but slight unlike orientations in PDE4B and PDE4D. LASSBio-448 (100 mg/kg, oral), 1 h before provocation, inhibited allergen-induced eosinophil accumulation in BAL fluid and lung tissue samples. Under an interventional approach, LASSBio-448 reversed ongoing lung eosinophilic infiltration, mucus exacerbation, peribronchiolar fibrosis and AHR by allergen provocation, in a mechanism clearly associated with blockade of pro-inflammatory mediators such as IL-4, IL-5, IL-13 and eotaxin-2. LASSBio-448 (2.5 and 10 mg/kg) also prevented inflammation and AHR induced by LPS. Finally, the sulfonamide derivative was shown to be less pro-emetic than rolipram and cilomilast in the assay employed. These findings suggest that LASSBio-448 is a new PDE4 inhibitor with marked potential to prevent and reverse pivotal pathological features of diseases characterized by lung inflammation, such as asthma.


Introduction
positions are given in parts per million (δ ppm), and J values are given in hertz. Signal multiplicities are represented by: s (singlet), d (doublet), t (triplet), q (quadruplet), m (multiplet) and br (broad signal). Melting points were determined with a Quimis 340 apparatus and are uncorrected. The HPLC solvents (methanol, acetonitrile and dimethylsulfoxide) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Water used in the preparations has been previously purified and filtered using a Milli-Q system (Millipore, St Quentin-en-Yvelines, France).

N-(3,4-dimethoxyphenethyl)-N-benzylbenzenesulfonamide
Animals. Male A/J mice (18-20 g) and male guinea pigs (300-400 g) were obtained from the Oswaldo Cruz Foundation breeding facility and kept in the animal-housing facility of the Laboratory of Inflammation at a controlled room temperature (22°C-25°C) and with 12 h light/dark cycle. All the procedures involving care and use of laboratory animals in this study were examined and approved by the Animal Ethics Committee of the Oswaldo Cruz Foundation (CEUA-FIOCRUZ-licence 0213-04).
Tracheal smooth muscle contraction in vitro. Tracheas from guinea pigs were obtained and prepared as described previously [37]. An initial tension of 1 g was applied to the tracheas for 60 min to obtain a constant resting tension. To confirm the viability of the preparation, the response to carbachol (2.5 μM) was recorded. After washout of carbachol and re-establishement of the baseline resting tension, tissues were preincubated with LASSBio-448 (100 μM) or vehicle (DMSO 0.1%) 15 min before re-exposed to cumulative addition of carbachol (10 −8 -10 −4 M). The tracheal rings were pretreated 10 min before application of LASSBio-448 with 100 μM SQ22536 (adenylyl cyclase inhibitor). Stimulation-induced isometric contractile responses were measured with a force-displacement transducer (Ugo Basile, Comerio, Italy) and the readout used to assess contractility was obtained by isolated organ data acquisition software (Proto 5; Letica Scientific Instruments, Barcelona, Spain). Contractile responses were expressed as a percentage of the maximal contraction induced by 2.5 μM carbachol.
Measurement of cAMP intracellular levels in airway smooth muscle cells. Intracellular cAMP concentrations were assayed in primary cultured guinea-pig tracheal smooth muscle cells as reported [37][38]. Briefly, smooth muscle cells obtained from guinea pig tracheas were cultured in DMEM containing 10% fetal bovine serum, 100 units/ml of penicillin, 100 mg/ml of streptomycin and 2 mM of glutamine for 3 to 7 days. After the third cell splitting, 10 6 cells/ well were grown in 24-well plates. At confluence, smooth muscle cells were washed with PBS and incubated with 100 μM LASSBio-448 or forskolin (adenylyl cyclase activator), in the presence or absence of 100 μM SQ22536 (adenylyl cyclase inhibitor) for 30 min. Lysed cells were collected and the intracellular cAMP was evaluated by means of radioimmunoassay (TRK 432-Cyclic AMP [3H] Biotrak assay system-Amersham Pharmacia Biotech, Buckinghamshire, England) following manufacturer's guidelines.
Sensitisation, allergen challenge and treatment protocol. Sensitization was performed by means of a subcutaneous injection of 50 μg ovalbumin (OVA) (grade V; Sigma-Aldrich, St. Louis, MO) adsorbed to 5 mg of Al(OH) 3 in 0.2 mL of sterile 0.9% NaCl (saline) at days 0 and 14. Intranasal OVA provocations (25 μg /25 μL saline) were performed at 19 and 20 days post-sensitization under isofluorane volatile anaesthesia (Cristalia, São Paulo, Brazil). Alternatively, sensitization as described above were done at days 0 and 7, while the OVA provocations were carried out at days 14, 21, 28 and 35 [39]. Negative control groups were represented by sensitized mice in which allergen was replaced by saline as a challenge. Treatments were done orally 1 h before provocations. Test compounds were dissolved in saline containing 0.2% Tween 80. All solutions were freshly prepared immediately before use. Analyses were performed 24 h after the last provocation.
LPS-induced inflammation. A/J mice were anesthetized with isoflurane aerosol and then challenged with LPS (25μg/25 μL) or phosphate buffered solution (PBS) by intranasal instillation as reported [37]. The analyses were performed 24 h after stimulation. Treatment with LASSBio-448 (2.5-10 mg/kg/mice) or cilomilast (1 mg/kg) was performed orally, 1 h before LPS exposure. LASSBio-448 was dissolved in 0.2% Tween 80, while cilomilast was dissolved in 1 M NaOH and further neutralized with 1 N HCl, before adjusting the final volume with 0.9% NaCl solution (saline). The substances were dissolved immediately before use.
Airway hyper-reactivity using non-invasive barometric plethysmography. Using barometric whole body plethysmography (Buxco Research System, Wilmington, NC) as described [38], we measured the enhanced pause responses (Penh) in conscious, spontaneously breathing mice following appropriate provocations. Aerosolized phosphate-buffered saline (PBS) and increasing methacholine concentrations (3, 6, 12 mg/ml) were nebulized through an inlet of the individual chambers for 2.5 min, and Penh readings were recorded for 5 min following each nebulization. Penh averages were obtained at 24 h after the last OVA provocation.
Airway hyper-reactivity using invasive barometric plethysmography. Airflow and transpulmonary pressure were recorded with a Buxco pulmonary mechanics processing system (Buxco Electronics, Sharon, CT), which calculated resistance (cmH 2 O.s/ml) and dynamic lung compliance (mL/cmH 2 O) in each breath cycle. Elastance was calculated as the inverse of compliance values [39,40]. Mice were anesthetized with nembutal (60 mg/kg), and the neuromuscular activity was blocked with bromide pancuronium (1 mg/kg). Animals were allowed to stabilize for 5 min and increasing concentrations of methacholine (3-27 mg/mL) were aerosolized for 5 min each. Baseline pulmonary parameters were assessed with aerosolized phosphate buffer solution (PBS).
Eosinophil peroxidase activity (EPO) in lung tissue. The EPO activity present in lung tissue was determined with a colorimetric assay as described by Strath et al. (1985) [41], with minor modifications. Briefly, the lungs were cannulated and perfused with saline/EDTA (20 mM). Then, lung samples were homogenized in buffer solution containing 0.5% HBSS (Hanḱ s balanced solution, pH 7.4; Sigma Chemical Co.) using a tissue homogenizer (T25 ultra-Tirrax). Samples were centrifuged at 1300 xg for 10 min at 4°C. Red blood cells were removed by hypotonic and hypertonic lysis. The suspended cells were then centrifuged again at 1300 xg for 10 min at 4°C. The pellet was then suspended in a solution of HBSS containing 0.5% of HTAB (hexadecyltrimethylamonium bromide; Sigma Chemical Co.) (pH 7.4). This suspension was subjected to heat shock for lysing the cells in three cycles freeze/thawed. At the end, the supernatants were collected, centrifuged at 12000 rpm for 15 min at 4°C and assayed spectrophotometrically for EPO determination. Fifty μl of the samples were placed, and then 50 μl of the substrate (1.5 mM o-phenylenediamine and 6.6 mM hydrogen peroxide in 0.05 mM Tris-HCl buffer, pH 8.0) was added and, after a 30 min incubation at room temperature, 50 μl of H 2 SO 4 (4M) was added to stop the reaction. The optical density reading was performed on the Spec-traMax M5 spectrophotometer (Molecular Devices) at a wavelength of 492 nm.
Myeloperoxidase (MPO) activity assay. The MPO activity present in the lung tissue was colorimetrically determined as described by Hirano (1996) [42], with minor changes. Briefly, after lung perfusion with 20 mM EDTA in 0.9% saline, lung samples were homogenized in Hank´s balanced solution (HBSS) at pH 7.4. The homogenates were centrifuged for 10 min at 1300 xg and 4°C. Red blood cells were removed by hypotonic and hypertonic lysis and suspended cells were re-centrifuged as described above. The pellet was then suspended in HBSS at pH 7.4 containing 0.5% hexadecyltrimethyl ammonium bromide (HTAB). The homogenates were centrifuged for 30 min at 12500 xg and 4°C. The supernatants were then collected and incubated with 0.167 mg/mL of ortho-dianisidine dihydrochloride for 15 min at 37°C, after which 0.0006% H 2 O 2 in 50 mM phosphate buffer at pH 6 was added. The reaction was stopped 10 min later by adding 50 μL of 1% sodium azide. The measurement was done at 460 nm using the SpectraMax M5 microplate reader (Molecular Devices).
Brochoalveolar lavage (BAL). Twenty-four hours after the last challenge the animals were killed by an over dose of thiopental (200 mg/kg, i.p.) and bronchoalveolar lavage (BAL) was performed three times by intratracheal instillation of 500 μL of PBS and EDTA (10 mM) through a trachea cannula and gently aspirating the fluid. Samples were centrifuged at 300 xg for 10 min, and cell pellets were suspended in 0.25 mL PBS. Total leukocyte numbers were measured in Neubauer chambers using light microcopy after diluting the samples in Türk solution (2% acetic acid). Differential cell counts were performed with cytospin smears using the May-Grünwald Giemsa method. At least 100 cells were counted per slide under light microscopy, and they were differentiated according to standard morphological criteria. The percentages of differential cells were multiplied by total cells number to obtain the number of different cells populations [38][39][40].
Cytokines quantification. Right lungs tissue samples from saline-and OVA-exposed mice were were homogenized in PBS containing 0.05% Triton X-100 and a protease inhibitor cocktail (Complete1, Hoffmann-La Roche Ltd, Basel, Switzerland). After centrifugation at 13000xg for 10 min and 4°C, the levels of IL-13, IL-5, IL-4 and eotaxin-2 were quantified in the supernatant, using commercially available ELISA kits (DuoSet; R&D Systems, Minneapolis, MN, USA). The results were expressed as picograms of cytokine per right lung.
Duration of anesthesia. This assessment was carried out as previously reported [43][44]. Mice of strain A/J were treated orally with rolipram, cilomilast, LASSBio-448 or vehicle and then anesthetized with a combination of xylazine (10 mg/kg) and ketamine (70 mg/kg), administered as a single hindlimb intramuscular injection 30 min post-treatment. The animals were placed in dorsal position and their spontaneous return to the prone position was used as an endpoint to determine anesthesia duration. Both rolipram and LASSBio-448 were freshly dissolved in Tween 80 0.3% immediately before use.

Molecular Docking
Docking studies with GOLD software version 5.2 [32], using goldscore scoring function were carried out to get insights on the detailed interactions of LASSBio-448 with PDE4 isoforms A, B, C and D. Marvin version 6.2.1, 2014, ChemAxon, was used for drawing, displaying and characterizing the ligand's chemical structures (http://www.chemaxon.com).
Since the available crystal structure of unliganded PDE4C2 isoform has a considerable degree of disorder in the region where the conserved Glutamine is located (PDB code 2QYM) [45], we have built a homology model using the FASTA sequence of PDE4C and the structure of PDE4D (best template found) at a high resolution (ID Code 3G4I), chain A [49] in the automated mode in Swiss-Model [50][51].
The set of amino acid residues selected as the binding site to perform docking studies was determined by a distance of 10 Å from the conserved NE2 of Gln369 (PDE4D numbering). All catalytic water molecules have been kept during docking procedures, since it has been shown in the literature that they are important in the binding process [48][49]. Metal ions were included as part of the active site as well. At last, ligand-protein interactions' analysis have been performed with Hermes, available as part of GOLD docking suite and Pymol v. 0.99, the latter of which has also been used to build molecular surfaces.

Statistical analysis
All data are presented as means ± standard error of the mean (SEM) and statistical analysis involving two groups was done, with Student´s t test, whereas ANOVA followed by the Newman-Keuls-Student´s t test were used to compare more than 2 groups. P values of 0.05 or less (two-tail test) were considered significant.

Chemistry
The synthesis of the target sulfonamides 5 and 6a was accomplished in a multistep linear synthesis. The process starting from a regiosselective sulfonation of 5-methylbenzo [d] [1,3] dioxole (7), followed by chlorination of the corresponding salt (8) under the conditions depicted in Fig 3. The key intermediate 9 was subjected to condensation reactions with 2-phenylethylamine and 3,4-dimethoxyphenethylamine to obtain 5 and 6a, respectively. Compounds 6f and 6j were synthetized in high yields by condensation between the commercial chlorosulfonic derivatives (10 and 11) and the 3,4-dimethoxyphenethylamine, in the presence of triethylamine as base and dichloromethane as solvent (Fig 3). The homologous series (6b-i and 6k) were obtained by N-alkylation of compounds 6a, 6f and 6j, using the methodology previously described by Barco and coworkers [12]. All compounds were characterized by NMR 1 H, NMR 13 C, infrared and mass spectrometry. The relative purity was determined by HPLC analysis.
The comparative physico-chemistry properties of rolipram, prototype 4 and its analogues 5 and 6a-k were calculated in silico using the Program ACD/Percepta 14.0. As demonstrated in Table 1 no violations to Lipinski'rule of five were observed, anticipating a probable orally administered drug profile [13].

Biology
In vitro studies. The inhibitory activities of the new sulfonamide derivatives were evaluated for their selectivity against the four isoforms of human PDE4, i.e. PDE4A, PDE4B, PDE4C, and PDE4D, using an IMAP-FP protocol. Compounds (5 and 6a-k) were tested at a screening concentration of 1 μM using rolipram as standard. Those inhibiting 30% or more were selected for further potency investigation. As indicated in Table 1, rolipram showed inhibitory activity against all the four PDE4 isoenzymes, while sulfonamide 6a was able to inhibit PDE4A, PDE4B and PDE4C activities. Compounds 5, 6b, 6d, 6g, 6i and 6j showed only a marginal activity against PDE4D (Table 2).
PDE4A, B and D are the isoenzymes expressed in human leukocytes and they have a central role in inflammatory diseases [14,15]. Recent review shows that PDE4C is not normally found in inflammatory cells and it is not related with inflammatory response [14,15]. Regarding the most important side effect of PDE4 inhibitors, Robichaud and coworkers [16] have demonstrated that emesis is produced as a result of PDE4D inhibitory activity. This isoenzyme is one of the four PDE4 genes products present in the brainstem. Studies carried out with PDE4Dknockout mice confirmed that emesis is strongly linked to the PDE4D inhibition [17]. Therefore, the analysis of the results showed in Table 2 allowed the selection of sulfonamide 6a (LASSBio-448) to further investigation. The comparative potency of rolipram and LASSBio-448 (6a) against PDE4A, B, C and D was established. As demonstrated in Table 3, sulfonamide 6a (LASSBio-448) was able to inhibit recombinant PDE4A, PDE4B, PDE4C and PDE4D with  6b-6e, 6g-6i and 6k)  It is well established that increased intracellular levels of cAMP mediates smooth muscle relaxation [6]. We noted that 6a (LASSBio-448), at 100 μM, inhibited tracheal contraction induced by cumulative addition of increasing concentrations of carbachol (S1A Fig), and upregulated cAMP intracellular levels in cultured tracheal smooth muscle cells, reaching levels comparable to that elicited by the adenylyl cyclase activator forskolin (S1B Fig). Moreover, pretreatment with 100 μM of SQ22,536, a standard adenylyl cyclase inhibitor, clearly abrogated  Further, the comparative in silico ADME profile of LASSBio-448 (6a) and rolipram was established as depicted in Table 4. Both compounds were predicted to be highly permeable (Caco-2), highly absorbed (HIA) and foreseen with great oral bioavailability (F%). The in silico prediction of their ADME profile also anticipate their ability to penetrate in central nervous system (CNS score), and a moderate and strong plasma protein binding (PPB) profile for rolipram and LASSBio-448, respectively, resulting in differences in their calculated volume of distribution (Vd). The metabolic stability of both compounds was predicted as undefined (data not shown).
In vivo studies. In asthmatics, airway hyper-reactivity and other pathological features of the disease are suggested to be associated in a causative manner with lung eosinophilic infiltration [18,19]. Not for nothing, various new anti-asthma therapies based in targeting eosinophils and their products are in development [20]. Using a short-term murine model of asthma, characterized by two allergen provocations at days 19 and 20 post-sensitization [21], we wanted to know, as a first approach, whether 6a (LASSBio-448) might prophylactically inhibit allergeninduced eosinophilic infiltration. We found that LASSBio-448 (100 mg/Kg), administered orally 1 h before challenge, clearly prevented the up-regulation of eosinophil levels in bronchoalveolar lavage fluid, as shown by eosinophil enumeration (Fig 4A) and lung tissue homogenates, attested by EPO colorimetric assay (Fig 4B), in samples obtained 24 h after allergen provocation. PDE4A, PDE4B and PDE4D isoenzymes are expressed in human eosinophils, being quite plausible that increased PDE4 function might account, at least partially, for the  pathogenesis of asthma [22]. In fact, PDE4 inhibitors are very efficacious in inhibiting the activation and release of a range of pro-inflammatory mediators by eosinophils, including cytokines, reactive oxygen species and cationic proteins [23,24]. Additionally, PDE4 inhibitors have been shown to reduce the influx of eosinophils to the lung of allergen-challenged animals, whilst also reducing the bronchoconstriction and elevated bronchial responsiveness [23,25,26]. Therefore, it is not unlikely that the protective effect LASSBio-448 (6a) upon allergeninduced eosinophil recruitment is, in some extent, dependent on the blockade of PDE4 isoenzymes. The biological activity exerted by eosinophils is primarily attributable to the release of their granular content, including EPO, eosinophil cationic protein (ECP) and major basic protein (MBP), as well as pro-inflammatory cytokines and chemokines [20]. In line with this concept, eosinophils and eosinophil-derived products have been found in high amounts in bronchial mucosa, sputum and BAL effluent of asthmatics, and appear to be directly associated with asthma severity and characteristic features including mucus hyper-secretion, extracellular matrix deposition and airway hyper-reactivity [27][28][29][30].
We then investigated what would be the impact of the oral treatment with LASSBio-448 (6a) on ongoing asthma changes. For this purpose, an alternative murine model of asthma was employed, characterized by 4 weekly ovalbumin provocations done at days 14, 21, 28 and 35 post-sensitization. As illustrated in the S2 Fig, an intense peribronchiolar leukocyte infiltration, accompanied by mucus exacerbation and increased production of extracellular matrix could be detected already after the third allergen provocation, compared with sham-challenged mice. Under this condition, a state of airway hyper-reactivity (AHR) to aerolized methacholine was already noted. Values of Penh response (an indirect measure of lung resistance) increased from 0.61 ± 0.04 to 1.18 ± 0.12 at the concentration of 3 mg/ml, from 1.11 ± 0.11 to 1.63 ± 0.09 at the concentration of 6 mg/ml, and from 1.84 ± 0.1 to 2.75 ± 0.28 at the concentration of 12 mg/ ml of methacholine (mean ± SEM) (p<0.05, n = 6). Results were interpreted as indicating that, as performed at the third and fourth weeks of allergen challenges, the treatment would encounter an ongoing asthmatic process. Actually, it is in this context that the exposure to either LASSBio-448 (100 mg/Kg, oral) or rolipram (10 mg/kg, oral) has been shown to impair the progress of allergen-induced eosinophil accumulation in the lung tissue, as pointed out by the colorimetric quantification of eosinophil peroxidase (Fig 5).
Moreover, as revealed by histologic evaluations of lungs from untreated (Fig 6A-6E) and treated mice (Fig 6C and 6F), mucus exacerbation ( Fig 6B) and extracellular matrix deposition ( Fig 6E) were also sensitive to the interventional treatment with LASSBio-448. Quantitative data on the effects of LASSBio-448 (6a) and rolipram are shown in Fig 6H and 6G, concerning mucus and extracellular matrix production, respectively. These treatments were also equally effective in inhibiting airway hyper-reactivity to methacholine, with regard to the functional parameters of airway resistance (Fig 7A) and lung elastance (Fig 7B).
In addition, portions of the lung tissue (right lung), collected at 24 h after the last allergen challenge, were homogenized to evaluate pro-inflammatory cytokine levels. In LASSBio-448 (6a) or rolipram-treated mice, compared to vehicle treated ones, a good correlation between the reduction in the levels of pivotal inflammatory mediators, such as IL-4 (Fig 8A), IL-5 ( Fig  8B), IL-13 ( Fig 8C) and eotaxin-2 (Fig 8D), and down-regulation of crucial asthma features assessed in this model was observed. Taken together, these results suggest that LASSBio-448 (6a), similarly to rolipram, not only can prevent the establishment of inflammatory and adverse remodeling changes as applied prophylactically, but also down-regulates these changes during ongoing allergic provocation as administered therapeutically.
In another setting of experiments, we found that the oral administration of LASSBio-448 (2.5 and 10 mg/kg), 1 h before provocation, dose-dependently prevented LPS-induced airway hyper-reactivity 24 h post-challenge, as attested by quantification of lung elastance values (cmH 2 O/mL) following exposure to aerolized methacholine (3-27 mg/ml) (Fig 9A). This efficacy seems to be accounted for by a marked reduction in the neutrophil infiltration into the lung tissue (Fig 9B). In our conditions, the magnitude of blockade in changes in both functional and inflammatory parameters were similar to that evidenced by cilomilast (1 mg/kg, oral) (Fig  9), suggesting a more comparable effectiveness and potency of LASSBio-448 concerning a reference compound in the current therapy.
While exploring the therapeutic window of LASSBio-448 (6a), we have done investigations into the emetic potential of this compound in comparison to rolipram and cilomilast. For that purpose, we used a well-established pharmacological approach, in which the emetic response associated with PDE4 inhibitors is indirectly attested by the ability of these agents to shorten the duration of α2-adrenoceptor-mediated anesthesia, a behavioral correlate of emesis in mice, rats and ferrets [16,17,31]. In A/J mice, the duration of anesthesia induced by the combined administration of xylazine (10 mg/Kg) and ketamine (70 mg/Kg), as shown by the loss of righting reflex, was 51.7 ± 1.8 min (mean ± SEM; n = 7). Our findings confirmed the anesthesia reversing property of rolipram [31], since as administered orally at 3.5, 11 and 36 μmol/Kg, it yielded sleep time reductions of 30%, 38.5% and 40.8% respectively (10). LASSBio-448 also reduced the duration of anesthesia in 24.1%, 26.6% and 35.5%, following oral doses of 11, 27 and 80 μmol/Kg respectively, suggesting a slight but significant lower pro-emetic activity of the sulfonamide compound (Fig 10). Furthermore, a 28.4% (n = 5) (P<0.01) reduction in the sleep time, induced by xylasine/ ketamine anesthesia, was obtained after treatment with cilomilast (5.8 μmol/kg, oral), which is about the one fifth of the equieffective dose of LASSBio-448. One potential explanation for this difference concerning rolipram and cilomilast is probably related to the lower inhibitory effect of LASSBio-448 upon PDE4D as pointed out in our in vitro experiments. In order to establish the putative binding-model of sulfonamide 6a (LASSBio-448) in the active site of human PDE4 isoenzymes, docking studies were performed using GOLD software version 5.2 (License: G/414/2006) [32]. PDE4D numbering has been used in order to facilitate visual analysis of the docking poses.
Docking analysis of the putative interactions of LASSBio-448 (6a) with PDE4A-D showed similar poses in the active site of PDE4A and PDE4C, but different orientations in PDE4B and D (Fig 11). However, a common feature seen in the four top poses of LASSBio-448 (6a) with the studied isoforms is a π-stacking interaction of the 3,4-dimethoxy phenyl ring against one of the conserved phenylalanine residues, Phe372 (PDE4D numbering), located in the active site, close to the protein surface (Fig 11). This is a common interaction, that has been observed in other crystal structures of PDE4 inhibitors [33,34].
In the top pose with PDE4A (Fig 11A), LASSBio-448 performs van der Waals interactions with amino acid residues Ile336, Met337 and Phe340. It makes a π-stacking interaction with Phe372 and makes hydrogen bonds involving one of the oxygen atoms of the sulfonyl group, the hydroxyl of Tyr159 and one of the hydrogen atoms of a water (HOH) molecule, which is part of a network of interactions with Zn 2+ in the active site of PDE4A.
Analysis of the top pose of LASSBio-448 with PDE4B ( Fig 11B) showed that it also performs van der Waals interactions with amino acid residues Ile336, Met337 and Phe340. It makes a close π-stacking interaction with Phe372 and makes a hydrogen bond involving the oxygen atom of the 4-methoxy phenyl group and the NH-group of the conserved Gln369. This docking pose and observed interactions are consistent with other studies with inhibitors that possess the 3,4-dimethoxy phenyl group [34]. LASSBio-448 (6a) also performs hydrogen bonds to the -SH group of Cys358 via one of the oxygen atoms of the 1,3-benzodioxole ring. Additionally, one of the oxygen atoms of the sulfonyl group is in close contact with the -CH group of His160 and the NE2 nitrogen of His204.
Sulfonamide 6a (LASSBio-448) makes hydrogen bonds involving the oxygen atoms of the sulfonyl group and the hydroxyl of Tyr159 and a water molecule (HOH) from PDE4C ( Fig  11C). It also hydrogen bonds with the conserved Gln369 via the 3-methoxy-phenyl oxygens. It showed a π-stacking interaction involving the 3,4-dimethoxy phenyl ring against Phe372 and Van der Waals contacts involving the 3-methoxy-phenyl group and Met337 and the linking hydrocarbon chain and Ile336 (Fig 11C).
Finally, the top pose of LASSBio-448 and PDE4D (Fig 11D) showed that it is capable of performing very similar interactions compared to the ones with PDE4B. The only exception is that, instead of being in close contact with the NE2 nitrogen of His204 it hydrogen bonds to HOH1006, which is part of the network of interactions involving Mg 2+ in the active site.
With both PDE4A and PDE4C the theoretical binding modes are very similar. The 3,4-dimethoxy phenyl ring of LASSBio-448 (6a) interacts closer to the conserved Gln369 at the solvent accessible surface (Fig 12A) in a folded conformation which does not span the whole active site and in which the 1,3-benzodioxole ring is buried within the pocket. With PDE4B and PDE4D, we have also observed similar conformations (Fig 12B). However, with these isoforms, LASSBio-448 interacts in a more extended conformation in which the 3,4-dimethoxy phenyl and the 1,3-benzodioxole rings are closer to the protein surface and fit the pocket making more contacts. Maybe this is responsible for the similar IC 50 values observed for LASSBio-448 (6a) against these recombinant isoforms ( Table 2). It is worth of note that the active sites of PDE4B and PDE4D are considered comparable [33], while PDE4A has shown displacements of aminoacid residues close to the conserved Glutamine (Gln369). Interestingly, our PDE4C model was based on a PDE4D template but resembles more PDE4A. Overall, different binding conformations and observed affinities may be due to subtle but important differences amongst the active sites of the studied PDE4, especially PDE4A, which has been reported to present conformational divergences compared to PDE4B and PDE4D [34].

Conclusion
In summary, we identified a new PDE4 inhibitor, equipotent to the standard rolipram. This inhibitor (6a) has been shown to be orally active in murine model of asthma. LASSBio-448  (6a) prevented the up-regulation of eosinophil levels in bronchoalveolar lavage fluid, impaired the progress of allergen-induced eosinophil accumulation in the lung tissue, reduced mucus and extracellular matrix production, was effective in the inhibition of airway hyper-reactivity to methacholine and reduced the levels of IL-4, IL-5, IL-13 and eotaxin-2. Moreover, it was at least 7-fold less pro-emetic than rolipram. Its liver microsomal metabolism was recently described [35]. Docking analysis of the putative interactions of LASSBio-448 (6a) with binding site of human PDE4 isoenzymes was performed and will be used to guide the further optimization of this new antiasthmatic lead-candidate.