Trends Sci. 2026; 23(4): 10899

Biflavonoids from Araucaria Genus as Selective PDE4 Inhibitors: Insights from In Silico and In Vitro Studies

Nafisah¹, Budi Arifin², Setyanto Tri Wahyudi³, Uus Saepuloh⁴,

Kurniawanti², Silmi Mariya⁴ and Purwantiningsih Sugita^2,*

¹Postgraduate Chemistry Student of Postgraduate School, IPB University, Bogor 16680, Indonesia

²Department of Chemistry, Faculty of Mathematics and Sciences, IPB University, Bogor 16680, Indonesia

³Department of Physics, Faculty of Science, IPB University, Bogor 16680, Indonesia

⁴Primate Research Centre, IPB University, Bogor 16151, Indonesia

(^*Corresponding author’s e-mail: [email protected])

Received: 9 June 2025, Revised: 28 July 2025, Accepted: 4 August 2025, Published: 25 December 2025

Abstract

Chronic inflammation is a major contributor to autoimmune diseases, necessitating the discovery of selective phosphodiesterase-4 (PDE4) inhibitors. Biflavonoids, with diverse biological activities, exhibit anti-inflammatory potential. This study employed molecular docking and molecular dynamics (MD) simulations to evaluate the interaction of 25 biflavonoid compounds with PDE4B and PDE4D. The most promising compound was validated using in vitro enzyme inhibition assays. Molecular docking identified 7,7''-di-O-methylamentoflavone as a potent PDE4B inhibitor with strong binding affinity and favourable MM/GBSA binding energy of –49.56 ± 4.12 kcal/mol, compared to its PDE4D binding energy of –39.77 ± 5.21 kcal/mol. Molecular dynamics simulations confirmed the stability of ligand–protein interactions. In vitro assays of six isolated biflavonoids from Araucaria hunsteinii and Araucaria cunninghamii confirmed that 7,7''-di-O-methylamentoflavone as a selective PDE4B inhibitor, with an IC₅₀ value of 13.9 ± 2.38 μM. This study provides new insights into the potential of biflavonoids as selective PDE4B inhibitors. However, further research is required to validate their therapeutic potential, including in vivo evaluation and broader safety profiling.

Keywords: Anti-inflammatory, Biflavonoids, Enzyme inhibition, Molecular docking, Molecular dynamics, PDE4 inhibitors

Introduction

Chronic inflammation is a critical driver of numerous diseases, including chronic obstructive pulmonary disease (COPD), asthma, arthritis and cancer [1]. While a controlled inflammatory response is essential for eradicating pathogens, excessive or prolonged inflammation can damage host tissues and contribute to the progression of degenerative diseases. Consequently, targeting inflammation through effective therapeutic strategies has become a priority in medical research [2]. Phosphodiesterase (PDE) is an intracellular enzyme that plays a role in catalyzing the hydrolysis of cyclic guanosine monophosphate (cGMP) and cyclic adenosine monophosphate (cAMP). These two molecules function as second messengers in signal transduction and regulate various physiological processes, such as visual transduction, cell proliferation and differentiation, cell cycle regulation, gene expression, inflammation, apoptosis and metabolic functions [3]. Phosphodiesterase-4 (PDE4), a key enzyme that regulates inflammatory responses, has emerged as a promising target for the development of anti-inflammatory drugs. PDE4 is predominantly expressed in immune and inflammatory cells, modulating cyclic adenosine monophosphate (cAMP) levels. Inhibiting PDE4 has been shown to suppress a wide spectrum of inflammatory responses both in vitro and in vivo [4]. Although PDE4 inhibitors are effective anti-inflammatory agents, their clinical use is limited by adverse effects such as nausea and emesis, primarily attributed to PDE4D inhibition. Thus, isoform-selective inhibition, particularly toward PDE4B, is highly desirable [5]. As a result, there is a growing interest in developing selective PDE4B inhibitors to achieve therapeutic efficacy with fewer side effects.

Over the past four decades, significant advancements have been made in identifying PDE4 inhibitors as therapeutic agents. This has resulted in a comprehensive collection of preclinical and clinical data that strongly supports drug discovery efforts for new treatments of inflammatory diseases such as COPD, asthma and psoriasis. During this process, many agents with diverse structures have been developed and evaluated pharmacologically, revealing, in some cases, promising therapeutic potential. However, only a few PDE4 inhibitors have been approved as medications. This is primarily due to the adverse effects observed with several PDE4 inhibitors under development, which were initially linked to a lack of specificity in their mechanism of action, hindering their full clinical development. One significant challenge that underscores the complexity of our task is the difficulty in creating isoform-selective PDE4 inhibitors, owing to the high degree of sequence and structural similarity between the various subtypes, especially in conserved regions of the catalytic site [6]. In this context, the identification of new molecules as isoform-selective PDE4 inhibitors continues to be an active field of investigation within drug discovery.

The production of pro-inflammatory cytokines and inflammatory cytokines is regulated via the degradation of cAMP by PDE4 (Figure 1). The inhibition of PDE4 results in the elevation of cAMP level and activation of PKA and exchange protein directly activated by cAMP (Epca 1/2). The activation of PKA leads to the phosphorylation of cAMP-responsive element-binding protein (CREB) and activation of the transcription factor 1 (ATF-1) which results in increasing the production of anti-inflammatory cytokines and decreasing the inflammatory cytokines [7,4]. Rolipram, a first-generation selective PDE4 inhibitor, was investigated for asthma treatment but was limited by gastrointestinal and CNS side effects. Roflumilast, the only PDE4 inhibitor approved by the FDA, is used for COPD but has limited efficacy in acute bronchoconstriction [6,8]. Several biflavonoid compounds such as amentoflavone, podocarpusflavone A, sequoiaflavone, podocarpusflavone B and 7,7″-di-O-methylamentoflavone showed strong inhibitory activity against various PDE isoforms. 7,7″-di-O-methylamentoflavone has been identified as a selective PDE4 inhibitor (IC₅₀ = 1.48 ± 0.21 μM), showing comparable activity to rolipram and making it a promising lead scaffold for drug development [9]. This natural compound appears to be a strong and selective PDE4 inhibitor, making it a promising candidate for further exploration in the development of new treatments for chronic obstructive pulmonary disease and asthma with minimal side effects.

Ligand-based pharmacophore models for diverse classes of PDE4B and PDE4D inhibitors were developed by [10]. Hydrogen bond acceptors were identified to be mainly responsible for PDE4B inhibition, while both hydrogen bond acceptors and hydrophobic groups were found to be responsible for PDE4D inhibition. In another study, six compounds from the result of ligand-based pharmacophore modelling showed higher PDE4B inhibitory activity (2 - 461 nM) than the rolipram [11]. The similarity between the biflavonoid and the reported PDE4 inhibitors (rolipram and roflumilast) (Figure 1) is that they all have an aromatic ring (benzene or heterocyclic systems) and oxygen-containing functional groups (OH, OCH₃, carbonyl), which contribute to their stability and biological activity. Thus, we chose biflavonoids to be the target of this research.

Figure 1 A brief schema of PDE4 in the regulation of inflammatory response.

Biflavonoids are polyphenolic compounds composed of two identical or distinct flavonoid units linked symmetrically or asymmetrically through C-C or C-O-C linkers [12]. Their anti-inflammatory properties have been extensively studied at the molecular level. Specific biflavonoid compounds, such as moreloflavones [13], amentoflavone [14], ginkgetin with an IC₅₀ of 0.75 μM [15] and ochnaflavones [16], have demonstrated the ability to inhibit pro-inflammatory enzymes like phospholipase A2 (PLA2) and cyclooxygenase (COX), although challenges remain regarding their bioavailability [17]. Among these, 7,7''-di-O-methyllamentoflavone has been identified as a selective inhibitor of PDE4 and a promising anti-inflammatory agent, with an IC₅₀ of 1.48 ± 0.21 μM [9]. In addition to their anti-inflammatory effects, biflavonoids exhibit a wide range of biological activities, including antiviral [18-19], anti-osteoporosis [20], antiplasmodium [21], antioxidants [22], antibacterial [23], and anticancer [24] properties. Biflavonoids have been extensively studied over the past 30 years and identified in numerous plant species. Notably, the genus Araucaria, a member of the Araucariaceae family, is known to produce significant secondary metabolites, including biflavonoids and terpenoids [25]. Biflavonoids of the Araucariaceae family significantly modulate key inflammatory pathways, including cytokine suppression and NF-κB inhibition [26]. Several Araucaria species have been reported as sources of biflavonoids, such as A. excelsa and A. cookii from India [27], A. araucana from India [28], Brazil pine (A. angustifolia) [29], pine hoop (A. cunninghamii) [23,24], A. bidwilii from Egypt [32], A. columnaris from India [33], A. hunsteinii K. Schum from Indonesia [13, 18] and A. columnaris from Indonesia [34].

In recent years, in silico approaches have become essential in drug discovery, enabling the efficient identification of new therapeutic compounds. These computational methods play a critical role in the modern pharmaceutical industry by reducing the time and resources required for drug development. This study focuses on identifying biflavonoids with strong anti-inflammatory potential by evaluating 25 biflavonoid derivatives from Genus Araucaria reported by [35]. These biflavonoid derivatives were selected for docking because they fulfil the requirements for an orally active drug, according to Lipinski’s rule of five and of the ADMET prediction study, compared to their basic skeletal biflavonoids such as amentoflavone, cupresuflavone, agathisflavone and robustaflavone [35]. Previous research has explored the anti-inflammatory properties of Araucaria species [26]. For example, studies on the polyphenol-rich fraction of Araucaria bidwillii demonstrated significant anti-inflammatory activity [32], while other work has examined the antioxidant and anti-inflammatory potential of three Araucaria species [36]. Despite these advancements, the specific potential of biflavonoid compounds from the genus Araucaria as anti-inflammatory agents, particularly in terms of inhibiting PDE4 enzymes, remains unexplored. This study aims to address this gap by combining in silico and in vitro approaches to investigate the inhibitory activity of biflavonoid compounds from the genus Araucaria against PDE4B and PDE4D enzymes so that it can be a reference for understanding the molecular mechanism of flavonoids as PDE4 inhibitors and open up new opportunities in the treatment of inflammatory diseases with minimal side effects. This study contributes to the limited number of investigations exploring Indonesian Araucaria-derived biflavonoids as PDE4 inhibitors using a combined in silico and in vitro approach. This study adopts a sequential in silico–in vitro workflow to efficiently identify PDE4 inhibitors from Araucaria-derived biflavonoids. Computational screening guided compound selection for in vitro validation, offering mechanistic insights into PDE4 isoform selectivity and laying the groundwork for future in vivo and preclinical investigations.

To strengthen the efficiency and direction of compound selection, this study employs a sequential in silico-in vitro approach to enhance the efficiency of early PDE4 inhibitor discovery. In silico screening, using docking and molecular dynamics simulations, enabled the identification of potential biflavonoids with high predicted affinity and selectivity for PDE4B. These computational predictions led to the prioritisation of compounds for in vitro validation, thereby reducing the experimental burden and providing mechanistic insights into ligand-enzyme interactions. This comprehensive approach exemplifies a sensible and resource-efficient strategy for discovering selective anti-inflammatory drugs.

Materials and methods

Materials

The materials used for computational research were 25 biflavonoid compounds that had been isolated from the genus Araucaria based on the literature review [26], 3-D crystal structures of target proteins PDE4B (Q07343) and PDE4D (Q08499) obtained through PDB from Uniprot data. Three pure isolates of biflavonoids from A. cunninghamii, namely 4',4''',7,7'''-tetra-O-methylcuprasoflavone, 7,4'',4'''-tri-O-methylrobustaflavone, and 7,7"-di-O-methylamentoflavone in the previous report [37], three compounds from A. hunsteinii, namely 4',7,7''-tri-O-methylcuprasoflavone, 4'',7,7''-tri-O-methylagatisflavone and 7,4'''-di-O-methylcuprasoflavone in the previous report [24] from Bogor Botanical Garden (Indonesia), PDE4B (No. FY-EH23197) and PDE4D (No. FY-EH9346) ELISA kits from Feiyue Biotechnology, Michigan cancer foundation-7 (MCF-7) cells (ATCC HTB 22), phosphate-buffered saline (PBS), trypsin, MCF-7 cell growth media (Roswell park memorial institute (RPMI) 1640 and fetal bovine serum (FBS)), and a mixture of antibiotics (penicillin and streptomycin).

Preparation of receptor target, test ligand co-crystal ligand

The test ligands, which comprised 25 biflavonoid compounds from the genus Araucaria (Table 1), were first created as two-dimensional (2D) structures using the ChemDraw Ultra 12 program. Comparison ligands, including roflumilast, rolipram and cyclic adenosine monophosphate (cAMP), were also prepared. The 2D structures were then converted into three-dimensional (3D) models using Chem3D and saved in *.pdb format. Geometry optimization and energy minimization of molecular structures were carried out using molecular mechanics methods using the Open Babel program with the Steepest Descent and Newton-Raphson methods and the GAFF (General Amber Force Field) force field. The test and comparison ligands were processed using AutoDock Tools 1.5.7 through a command line in Linux. This processing included adding hydrogen atoms, merging non-polar hydrogens, assigning Gasteiger charges and defining the number of active torsions. The modified files were subsequently saved in *.pdbqt format, following the modifications described by [38].

Table 1 Test ligan of biflavonoid compounds from the genus Araucaria.

The target enzymes PDE4B and PDE4D in humans were accessed via UniProt (www.uniprot.org) and ten wild-type structures for each enzyme were selected based on a resolution of less than 2.5 Å and Ramachandran plot analysis. These structures were downloaded in *.pdb format from the RCSB Protein Data Bank (PDB). The receptor files were prepared using AutoDock Tools, employing a command-line workflow that involved removing water molecules, adding Gasteiger charges and associating hydrogen atoms. The processed receptor files were then saved in *.pdbqt format for docking studies [39].

Molecular docking simulation

The self-docking method was validated using AutoDock Vina software executed via the command line. This involved docking the co-crystal ligand with the receptor protein, generating multiple ligand poses. From these, the pose with the smallest root-mean-square deviation (RMSD) value was selected as the optimal reference pose for further analysis. A cross-docking procedure was then performed to evaluate 25 test ligands, along with reference ligands cyclic adenosine monophosphate (CMP), roflumilast (ROF) and rolipram (ROL) against the receptor proteins. AutoDock Vina software was employed for this process, with each docking simulation repeated ten times to ensure consistency. Molecular docking was performed using AutoDock Vina with the following parameters: Grid spacing of 1 Å, grid box dimensions of 20×20×20 ų and exhaustiveness set to 32. A total of 10 output poses were generated per ligand. The random seed was left at its default value, allowing variation between runs. The grid centre were determined based on the optimal results from the self-docking validation (detailed information of the AutoDock Vina command-line input and configuration files in SI). The docking simulations yielded binding energy values (kcal/mol), which were analyzed to assess ligand-receptor interactions. These interactions, including hydrophobic bonds and hydrogen bonds, were visualized using Pymol 3.1.4 and LigPlot+ software. Ligands were ranked based on their binding energy values and their selectivity for PDE4B over PDE4D. The test ligand with the most negative (i.e., most favourable) binding energy for PDE4B was identified as the best candidate. Reference ligands such as ROL, CMP and ROF served as benchmarks to evaluate the affinity and selectivity of the test ligands, providing a comparative framework for the docking results.

Molecular dynamics simulation

Selected protein-ligand complexes were prepared using the AMBER20 program. Water and hydrogen were removed from the complex using Pdb4amber. Next, the pK value of the ionized group on the amino acid residue was calculated using the H⁺⁺ website (http://biophysics.cs.vt.edu/) at pH 7.0. The program H⁺⁺ can predict the amino acid with a non-standard protonation state and adds missing hydrogen atoms according to the specified pH of the environment. The ligand-receptor complex molecules are each given an AMBER ff14sb force field. In the periodic boundary conditions (PBC), the system was dissolved with TIP-3P water molecules with a gap of 1.0 nm and neutralised using Na⁺/Cl⁻. In the heating stage, amino acid residues are limited by 10 kcal/mol with the NVT ensemble (constant number of particles, volume and temperature). The system is heated until it reaches a temperature of 312 K. In the equilibration stage, the heating limitation is released gradually. The production simulation process uses PMEMD.CUDA, to see the free movement of molecules without restrictions for 100 ns at a fixed temperature of 312 K. Evaluation of structural properties, root mean square deviation (RMSD) of the complex and determination of the hydrogen bonds are carried out using CPPTRAJ module from AmberTools. The relative free energy of binding was calculated using the Molecular Mechanics Generalized Born Surface area (MM/GBSA) method [40].

For calculation MM/GBSA of the gas-phase interaction energy (ΔE_MM) and the non-polar part (ΔG_SA) of the solvation energy and the electrostatic solvation energy (ΔG_GB), 50,000 frames evenly extracted from a single MD trajectory of the complex from 0 to 100 ns were used, using the GB model with parameters igb = 2, which generally represents air conditions implicitly, a salt concentration of 0.100 M to represent the physiological environment and using a probe radius of 1.4 Å, which represents the size of air molecules when calculating the solute surface. Binding free energy (ΔGbind) between a ligand (L) and a receptor (R) to form a complex RL is calculated as [41]:

In vitro PDE4B and PDE4D inhibition assay

MCF-7 cells were grown in 24-well microplates with a concentration of 5,000 cells in 100 μL of growth medium (D-MEM, RPMI 1,640 and FBS 5%) and antibiotic mixtures (penicillin 100 U/mL and streptomycin 100 mg/mL). To assess PDE4B and PDE4D enzyme inhibition in MCF-7 cells, we utilised a direct cAMP ELISA-based assay (Feiyue Biotechnology) following cell lysis after compound treatment. Since PDE4 enzymes hydrolyse cAMP, inhibition leads to intracellular cAMP accumulation, which is quantitatively measured as the primary readout. The resulting IC₅₀ values reflect the concentration required for 50% inhibition of PDE4B or PDE4D enzymatic activity. To distinguish PDE4B and PDE4D inhibition, separate ELISA kits with isoform-specific antibodies were used for each enzyme target. Pure isolates of compounds 5, 18 and 22 from A. hunsteinii K. Schum leaves and compounds 13, 20 and 26 from A. cunninghamii leaves were each added after the cells reached 50% confluency (24 h) and a negative control was prepared. All of the samples and controls were carried out with three replicates [42]. Cells are gently washed with cold phosphate-buffered saline (PBS) in moderate amounts and released with trypsin. Cells are centrifuged for 5 min at 1,000×g (suspension cells can be collected by centrifugation directly). The supernatant is discarded and the cells are washed 3 times with cold PBS. Cells are diluted in cold PBS until they amount to 5×106 cells/mL. The freeze-thaw process is repeated several times until the cells are fully lysed, then centrifuged for 10 min at 1,500×g at 2 - 8 °C. The cell lysates are either tested immediately or stored at –20 or –80 °C for later use.

The cell lysates were collected following centrifuging for 5 min at 600×g and then determined by the direct cAMP PDE4B and PDE4D ELISA kit (Feiyue Biotechnology, Wuhan, China) according to the manufacturer’s instruction. Absorbances are read at a wavelength of 450 nm (OD450) using Microplate Reader S/N 11421. The IC₅₀ values were determined by nonlinear regression analysis and a sigmoidal dose-response equation using GraphPad Prism 4 (GraphPad Software Inc., La Jolla, CA, USA) [43].

The MCF-7 cell line was chosen because of its well-characterised cAMP–PKA signalling pathway and modest endogenous production of PDE4 isoforms. Despite being extensively employed in breast cancer research, MCF-7 is appropriate for assessing PDE4 inhibition due to its functional susceptibility to intracellular cAMP modulation. MCF-7 offers a human-derived epithelial model that circumvents the immunological-related variables and variability found in primary immune cells. PDE4B and PDE4D activities were evaluated using their basal expression levels in MCF-7 cells; no inflammatory stimulation was used in this assay.

Results and discussion

Protein structures and validation parameters

Docking simulations involving PDE4B and PDE4D receptors with 25 test ligands of biflavonoid compounds from the genus Araucaria. The 3-D structures of PDE4B and PDE4D proteins obtained from the UniProt and Protein Data Bank sites consist of 40 and 103 protein structures, respectively. Ten selected wildtype structures from each receptor used in this study (Table 2) were based on parameter resolution values ​​ ≤ 2.5 Ǻ, most favoured regions value ​​above 90% and positive G-factors values. A resolution value of ≤ 2.5 Å determines that the docking process runs well [44]. Based on the Ramachandran plot, a good quality model is expected to have more than 90% in the most favoured regions [45]. The G-factor is a value that measures the stereochemistry of a protein model. A low G-factor value indicates the protein model has a low conformational probability. The ideal G-factor value is above –0.5 [46]. The self-docking results in the form of RMSD values ​​are shown in Table 2. A smaller RMSD value (closer to 0) indicates the docked ligand pose’s similarity with the experiment's co-crystal ligand.

Table 2 List of protein structures and RMSD values ​​from self-docking results.

Based on the ensemble docking process, in the virtual screening process, 25 test ligand compounds were docked to 20 structures at once, namely 10 PDE4B structures and 10 PDE4D structures with ten repetitions (Table S1). Ensemble docking is a computational drug discovery technique that accounts for protein flexibility by docking ligands to multiple conformations of a target protein, to identify more accurate binding predictions. Based on Table 3, The ligand with the highest binding affinity value indicated by the most negative value is compound 13 for inhibition of PDE4B and PDE4D, namely –10.90 ± 0.41 and –10.24 ± 0.39 kcal/mol. The lower Gibbs free energy (∆G) indicates that the compound reacts more quickly and spontaneously. This shows the potential and activity to interact and establish strong bonds with its target protein [47]. Compared with comparison compounds, such as rolipram, the binding energy values ​​for PDE4B and PDE4D tend to be the same. At the same time, compound 13 has a greater affinity for PDEB than PDE4D, so it is predicted to have strong potential in selectively inhibiting the PDE4B enzyme.

Table 3 Binding energy (kcal/mol) of test ligands and comparison for PDE4B and PDE4D proteins.

All test compounds have a lower binding energy than the reference compound (rolipram), except for compounds 17, 18, 20 and 21. These results indicate that these compounds can potentially inhibit the PDE4B enzyme, including compounds 5, 13, 22 and 25, which will be further analyzed against MD and in vitro. Compounds 18 and 20 have lower affinity than the reference compounds and will be further analyzed to evaluate the relationship between the docking method, MD and in vitro. Moreover, compounds 5, 13, 18, 20, 22 and 25 have been isolated from A. hunsteinii and A. cunninghamii Indonesia, which can be used to analyse the relationship between data compounds that have low and high binding energy in the results of in silico and in vitro.

Cross-docking results were visualised using Ligplot⁺ (2D) and Pymol 3.1.4 (3D) software to analyse interactions between ligands and receptors (Figure S1). Based on Table 4, the differences in the binding positions of atomic groups and bond distances in compound 13 to PDE4B and PDE4D affect the resulting binding energy. Compound 13 has three hydrogen bonds with a short bond distance to the PDE4B receptor, causing a low binding energy value. The key residues interacting with compound 13 are Asp124, Glu266, Lys354, Glu358, Asn132, His183, Leu152, Phe263, Leu242, Phe295, Met196, Asp241, Tyr382. The bonds involving polar residues such as Asp124 and Leu351 allow hydrogen interactions. The binding site location has more negatively charged residues interacting with the polar groups of compound 13. This suggests that PDE4B may be more affected by electrostatic interactions. The hydrophobic bond with the Asp241 residue is more stable, as evidenced by Al-Nema’s study [5].

Table 4 Binding interactions of selected test compounds to PDE4B and PDE4D proteins.

In the case of PDE4D, the binding energy of compound 13 with the PDE4D receptor is higher, with six hydrogen bonds. The residues interacting with compound 13 are Asp201, Gln343, Gln369, Thr271, Met273, Phe340, Tyr159, Ile336, His160, Phe372, Ser208 and Asn321. In PDE4D, the interactions involve hydrogen bonds with polar residues (Asp201, Gln343) and hydrophobic interactions with aromatic residues (Phe340, Phe372). The differences in interactions of compound 13 with PDE4B and PDE4D may affect the binding affinity and have implications for the selectivity of the compound towards PDE4B or PDE4D.

Visualization of compound 13 interactions with PDE4B and PDE4D proteins can be seen in Figure 2. Hydrogen bonds are the most important specific interactions in the ligand-receptor interaction process. Hydrogen bond interactions regulate the stability of the complexes formed [48]. Compound 13 has a hydrogen bond with the Mg metal ion in the PDE4B protein, whereas, in PDE4D, it does not interact with the Mg metal ion but with the Zn metal ion. Gangwal et al. [11] reported that most selective PDE4B inhibitors showed charge interactions with magnesium metal ions. Residue Asp124 controls inhibitor access to magnesium metal ions. Meanwhile, in PDE4D inhibitor. the selective inhibitor cannot form charge interactions with magnesium ions. Competitive inhibition of the PDE4B enzyme by mimicking the substrate and binding to the enzyme’s active site so that the enzyme experiences reduced or no activity. Hydrogen bond acceptors were identified as mainly responsible for the inhibition of PDE4B. In contrast, hydrogen bond acceptors and hydrophobic groups were found to be responsible for the inhibition of PDE4D [10].

Figure 2 2D and 3D visualization of the interaction between compound 13 on (A) PDE4B and PDE4D (B).