Development of multi-target QSAR models to screen novel-designed chalcone, 1,3-thiazole, and 1,3,4-thiadiazole derivatives as dual inhibitors of ACE and NEP enzymesThe developed models have significant discriminating power, as demonstrated by the optimum values obtained for statistical parameters including accuracy, precision, sensitivity, specificity, F-measure, and Mathew’s Correlation Coefficient (MCC). All of the models demonstrated similar results for each compound in the test set. Thus, it can be concluded that the developed mt-QSAR models (LDA and RF) are properly capable of determining two endpoints for dual inhibition of ACE and NEP of newly designed compounds.Linear discriminant analysis (LDA) based mt-QSAR model developmentThe QSAR-Co tool’s multi-target modeling leverages the Box-Jenkins method. This method modifies molecular descriptors for each dataset compound, incorporating diverse conditions to enhance the predictive power of the model. The mt-QSAR model was constructed using a dataset consisting of a sub-training set (n = 501) and a test set (n = 214). The best-fit mt-QSAR-LDA model was chosen from all models with the least Wilks (λ)train and highest MCCtrain, as shown in in Table 1 and accompanied by statistical parameters.Table 1 The best-fit mt-QSAR-LDA model (Standard coefficients and fitness scores) for dual inhibition of ACE and NEP enzymes.Table 1 shows standardized coefficient values, indicating selected descriptor’s contribution to inhibitory activity. ‘JGI8_tn’ has maximum (positive) contribution (coefficient value = 137.2343) and ‘VE2sign_Dz(i)_at’ has the maximum (negative) contribution (coefficient value = − 63.265) towards enzymes inhibition. A succinct explanation of the significance, source, and contribution of each descriptor used for the final LDA model, as shown in Table 2.Table 2 Symbols and definitions for the descriptors selected in the mt-QSAR (LDA) model for dual inhibition of ACE and NEP enzymes.The mt-QSAR-LDA model meets the requirements for robustness, quality of fit, and statistical significance. The Wilks λ statistic, with a value of 0.479, indicates the model’s adequate discriminatory power. Table 3 presents the classification results for both the sub-training and test sets, providing an overview of the overall performance of the mt-QSAR-LDA model.Table 3 Overall statistical performance of the final mt-QSAR (LDA and RF) models for dual inhibition of ACE and NEP enzymes.Table 3 demonstrates the model’s strong discrimination ability. Accuracy reached 91.22% and 88.79% for the sub-training and test sets, respectively. Moreover, it accurately classified 94.37% of active samples and 83.56% of inactive ones in the sub-training set, while achieving similar performance in the test set with 93.17% accuracy for active samples and 75.47% for inactive ones. These results support the high degree of efficiency of the mt-QSAR-LDA model to distinguish between active and inactive inhibitors. MCC values (0.7858 for sub-training, 0.6954 for test) further confirm the model’s statistical robustness72.Figure 5 shows the receiver operating characteristic curve (ROC) plot for the training and test set. The area under the ROC curve (AUROC) values of 0.9007 and 0.8138 were obtained, indicating good statistical significance of the mt-QSAR model. The higher AUROC for the training set is expected, as the model is built on this data. However, a value of 0.8138 on the test set suggests good generalizability to unseen data.Figure 5ROC (using tenfold cross-validation) plots for the best LDA model for dual inhibition of ACE and NEP enzymes.The Y-randomization test59 indicates that the mt-QSAR model is not created by chance, as shown in Fig. 6, with Wilk’s lambda values (50 model average λrandom = 0.9201) significantly higher than the original value (λtrain = 0.4791). The QSAR-Co software’s standardization approach60 determined the applicability domain, revealing 6 out of 501 training data points and 5 out of 214 test data points as possible outliers and outside the applicability domain.Figure 6Y-randomization test results for the developed LDA model for dual inhibition of ACE and NEP enzymes.Non-linear mt-QSAR (Random Forest) model development for dual inhibition of ACE and NEP enzymesThe Random Forest (RF) technique was used to construct a non-linear classification-based mt-QSAR model using training and test sets, developed using QSAR-Co software and Weka version 3.9.3 library73. The RF model of QSAR-Co, with its default parameters (tenfold cross-validation procedure), demonstrated superior overall statistical prediction quality compared to the LDA model, as detailed in Table 3. The RF model outperforms the LDA model in predicting inactive and active compounds due to its superior specificity, precision, and accuracy values. Accordingly, employing both models would be always beneficial to perform consensus predictions for queries or newly designed compounds. Further, Fig. 7 shows the plots of the corresponding ROC curves for the RF model, and the AUC values for both the training set (= 0.8453) and the test set (= 0.8225) show that the model has significant discriminatory power.Figure 7ROC (using tenfold cross-validation) plots for the best RF model.Non-linear models with all computed descriptors often produce better predictive models than linear models with a subset of descriptors, but their interpretability is inferior. RF is a method for group categorization that averages predicted outcomes from several different decision trees to produce its predictions. The great precision and superiority of RF have drawn a lot of interest recently74,75,76,77. RF has multiple advantages, one of which is that it is less vulnerable to constructing overfitted models. Consequently, RF can be favoured over several other non-linear machine learning techniques to generate highly accurate mt-QSAR models78,79.Screening of designed derivatives using validated developed mt-QSAR modelsUsing the facility available to screen large designed datasets in QSAR-Co software, the mt-QSAR models were used to screen externally designed derivatives set of four heterocyclic scaffolds viz. chalcones and its derivatives (235 compounds), 1, 3- Thiazole (24 compounds) and 1,3,4-Thiadiazole (107 compounds). Correspondingly, both LDA and RF models are faster in screening large-size databases with an accuracy of equivalent to 90% and an MCC value greater than 0.5. The selection of ligands is based on two primary criteria. Firstly, the ligand must fall within the applicability domain of our LDA models. Secondly, they are required to show a positive score for activity against both ACE and NEP under any given set of experimental conditions, as indicated by the results from our developed LDA and RF models for each of the designed heterocyclic derivatives. Details of this screened dataset and calculated descriptors, as well as the results of the predictions, are provided in Supporting file S6. Out of the 235 designed chalcone derivatives, 85 compounds meet all the criteria. Likewise, out of the 24 designed thiazoles, 8 meet the criteria; out of the 107 designed thiadiazoles, 12 compounds exhibit positive results. Thus, all designed compounds are screened using developed models, and only those molecules that follow the applicability domain and show positive prediction are processed further for molecular docking study (Fig. 8). Altogether, these diverse statistics demonstrate the high internal quality as well as the predictive power of the derived mt-QSAR models.Figure 8Screening of designed compounds using developed mt-QSAR models (LDA and RF). Compounds found active through both models are only selected for molecular docking study.Molecular docking studies of selected screened designed derivativesIn an effort to identify new starting point leads for novel multi-target inhibitors of ACE (C-domain selective) and NEP enzymes, molecular docking simulations were performed for the screened active hit suggested by mt-QSAR models from a combinatorial library of designed derivatives. There are a total of 85 chalcone derivatives, 8 thiazoles, and 12 thiadiazoles derivatives selected for molecular docking which passes the AD criteria of developed mt-QSAR models.Validation of docking protocolThe docking procedure was validated by re-docking the natural ligands (lisinopril and LBQ657 compounds) from crystal structures (PDB ID: 1O86 and 5JMY) into its binding pocket before virtual screening of chosen compounds. The large size and shape of the binding site pose challenges for molecular docking on targets. In cACE and NEP molecular docking calculations, constraints are applied to obtain reliable orientations of ligands, including hydrogen bonds with His353 and/or His513 (cACE) and His711 (NEP) along with metal-chelator interactions43,80. The study found that native ligands in protein structure are maximally superimposed with co-crystallized ligands, confirming the docking protocol’s agreement with previous work and confirming the interaction of native ligands(Fig. 9)17,81. This suggests that the docking methodology is suitable for the virtual screening of dataset compounds.Figure 9Validation of docking protocol by re-docking the native ligands (Lisinopril and LBQ657) at active binding site and interacting amino acid residues (Magenta color original poses, green color redocked pose of lisinopril and cyan color redocked pose of LBQ657).Molecular docking resultsThe binding affinity results of the standard drug molecules, Omapatrilate against the cACE and NEP enzyme are summarized in Table 4. Conventional ACE inhibitors bind to the catalytic region of the active sites of cACE and NEP via chelation with the central Zn2+, while the groups P2′, P1′, P1, and P2, mimicking substrate peptides, are placed inside these subsites. Most of the selected ligands were successfully docked with a plausible pose into the active sites using one of the applied constraint conditions. The resulting docking pose that has metal-acceptor interaction and bonding interactions with His353 and/or His513 (cACE) or bonding with His711 (nACE) is selected for assessment80. The results of molecular docking of selected screened designed derivatives with higher or equivalent docking scores compared to standard drugs with their interacting residues are given in Tables S1 to S3 (Supporting file S8).Table 4 Molecular docking results of selected screened designed derivatives against cACE (PDB ID: 1O86) and NEP (PDB ID: 5JMY) enzyme.Chalcone derivatives exhibited varying docking scores for both cACE and NEP (See Supporting file S7). Notably, C115 demonstrated a docking score of − 5.7009 for cACE and − 7.1057 for NEP, while C148 had a docking score of − 5.6612 for cACE and − 7.1215 for NEP. Compound C105, a member of the chalcone derivatives, displayed a favorable docking score of − 5.6880 for cACE. These results suggest variations in the binding affinities of chalcone derivatives to the target enzymes.Likewise, the 1,3-thiazole derivatives showed a range of docking scores. T1 exhibited a high affinity for the binding of cACE and NEP, with docking scores of − 7.7185 and − 6.5466, respectively. The docking scores for T3 interactions with the enzymes were − 6.9584 for cACE and − 7.1859 for NEP. In comparison, T10 and T20 received a docking score of − 4.9357 and − 4.7410 for NEP and no pose for cACE.Among the 1,3,4-Thiadiazole derivatives, TD7 displayed a docking score of − 5.4932 for cACE and − 5.2422 for NEP. TD104 also exhibited a strong binding affinity with a docking score of − 8.1032 for cACE and − 7.5842 for NEP. TD98 achieved an impressive docking score of − 9.4353 for cACE and − 8.7272 for NEP. TD101 had a minimal docking score of − 0.0854 for cACE and − 7.4954 for NEP. TD64 had a docking score of − 6.1358 for cACE and − 6.1683 for NEP, while D103 had no pose for cACE. TD107 also had no pose for either cACE or NEP.Visualization of docking poses demonstrated the importance of chelation with the central Zn2+, formation of hydrogen (conventional), and hydrophobic (Ï€-π stacking, π-alkyl, and alkyl) interaction required with key residues at S1 and S2’ for the inhibition both cACE and NEP enzymes. From the docking results, the compounds with functional moiety forming the above interactions appeared to be an ideal scaffold to be dual inhibitors. The final docking poses and binding interactions of selected top ligands which have favorable binding energies and poses are illustrated in Figs. 10, 11, 12, 13.Figure 10Schematic representation of 2D (a2 and b2) and 3D (a1 and b1) docking poses of standard drug omapatrilate against cACE and NEP target, binding to the catalytic region of the active sites via a chelation interaction with the zinc atom.Figure 11Schematic representation of 2D (a2 and b2) and 3D (a1 and b1) docking poses of Chalcone derivative (C115) against cACE and NEP target, binding to the catalytic region of the active sites via a chelation interaction with the zinc atom.Figure 12Schematic representation of 2D (a2 and b2) and 3D (a1 and b1) docking poses of 1,3-thiazole derivative (T3) against cACE and NEP target, binding to the catalytic region of the active sites via a chelation interaction with the zinc atom.Figure 13Schematic representation of 2D (a2 and b2) and 3D (a1 and b1) docking poses of 1,3,4-thiadiazole derivative (TD104) against cACE and NEP target, binding to the catalytic region of the active sites via a chelation interaction with the zinc atom.Molecular docking analysisOmapatrilate is a comprehensively experimented dual ACE/NEP inhibitor showing a binding energy of—5.2088 kcal/mol against cACE because of the metallic interactions with Zn2+ and hydrogen bonding interactions with Ala356 and Ser355 through an oxygen atom. Further, the sulfur atom and benzene ring enhanced stability by forming Ï€-sulfur interactions with His513 (subsite S1′) and His353. It also shows alkyl, Ï€-alkyl, and Ï€-Ï€ stacked interactions with residue His387, Val518, Val380, Ala354, and His353 respectively. The interaction between the omapatrilat molecule and NEP is particularly stable (binding energy—5.2360 kcal/mol), as seen in Fig. 10. The P1′ carbonyl group of omapatrilat forms a hydrogen bond interaction with Phe106. Although the seven-membered fused ring only partially reaches into the S2′ pocket, it nevertheless interacts hydrophobically with Val580, His583, and Ala543 residue. Both oxygen atoms of the Omapatrilat P2′ peptide bond form metallic interactions with Zn2+ and conventional hydrogen bonds with His711 and Arg717 residues.In the docking simulation between C115 and cACE, it forms three hydrogen bonds with Gln281 (S2’ subsite), Tyr523, and Glu384 residues Fig. 11. It also forms attractive charge interactions (Ï€-cation, Ï€-anion, and others) with five amino acid residues including His353, Asp377, Glu162, Glu376, and Lys511. In addition, four residues (Ala354, Val380, Val379, and His383) were included in hydrophobic interactions (alkyl, Ï€-alkyl, and Ï€-Ï€ T-shaped interactions) to give stability to docked pose of the ligand (binding energy,—5.7009 kcal/mol). The molecular interaction pattern of compound C115 showed docking interactions with the NEP enzyme accompanied by a docking score of − 7.1057 kcal/mol. It forms two hydrogen bonds with Glu584 and His583 similar to that of omapatrilate. Further, the stability of the protein–ligand complex, on one hand, is supported by the formation of charged interactions (Ï€-cation, Ï€-anion, and others) with Arg110, His711, Arg717, and Asp650. It also forms hydrophobic interaction with Phe106 (Ï€-Ï€ T shaped), and Val580 (Ï€-alkyl) through phenyl ring.Similar to omapatrilate, compound T3, belonging to the 1,3-thiazole derivative formed two hydrogen bonds with Glu384 (Ï€-donor H-bond) and His353 (conventional H-bond) residues included in the binding pocket. Aside from H-bonds, additional hydrophobic (Glu411 and His387: Ï€-anion interactions) and attractive charge interactions (Ala354 residue) were also detected (Fig. 12). On the other hand, hydrophobic cleft formed by Ï€-alkyl, π–π stacked, and π–π T-shaped with His513 and Tyr523 provides additional stability to better fit at the active site of the cACE domain. The utmost affinity of T3 with the NEP enzyme accounted for the formation of hydrogen bonds (conventional and carbon-hydrogen) with Arg717, Phe689, Ala543, and Trp693. In addition to hydrogen bonding interaction, ligands form hydrophobic bonding interactions (alkyl, Ï€-alkyl, Ï€-sulfur, and π–π T-shaped) with Val580, Met579, and Val692.The designed 1,3,4-thiadiazole derivative, compound TD104, showed binding affinity − 8.1032 kcal/mol and − 7.5842 kcal/mol with cACE and NEP enzymes respectively. It interacts with amino acids Gln281, Lys511, Tyr520, Gln369, His353, and His513 by forming both conventional and carbon-hydrogen bonds. It also forms charged interactions with Glu376, Asp377, and Glu162 (Ï€-cation, Ï€-anion, and other attractive charge interactions). The formation of hydrophobic interactions (alkyl, Ï€-alkyl, and Ï€-sulfur) with Phe457, Ala354, Val380, Val518, and Tyr523 amino acid residue respectively provides stability to docked poses. The formation of metal-acceptor interaction at the active site of both enzymes again justifies high binding scores for this compound. Compound TD104 also shows favorable binding affinity and forms conventional hydrogen bonding interaction with Arg717, His711, Asn542, and Ï€-charge interaction with Arg110, Glu584, and Glu646 (Fig. 13).These hydrogen-bonding interactions and hydrophobic interactions between selected designed derivatives (C115, T3, and TD104) and catalytic amino acid [His353 and/or His513 (cACE) and His711 (NEP)] residues of cACE and NEP enzymes make a favorable orientation to interact with zinc ion through metal-chelator interactions. Our docking results indicated that C115, T3, and TD104 could inhibit both targets by inhibiting the active site rather than other secondary sites. Moreover, compound TD104 exhibited greater binding potential for the cACE and NEP compared to C115 and T3.Further, the selected top ligands that form favorable interactions with targets were screened using ADMET studies. Additionally, molecular dynamics simulations of all selected ligand docking poses were run to verify the key residues (catalytic residue) from docking poses.In silico ADME and toxicity studies of selected screened compoundsPrediction of Pharmacokinetic profile (ADME) parameters before experimental studies is among the most vital aspects of the drug design and discovery of drug molecules. These parameters along with toxicity predictions of the compounds are considered important attentive parameters during the transformation of a molecule into a potent drug.Drug-likeness and ADME studies of the selected screened compoundsThe drug-likeness capability of compounds can be prophesied using Lipinski, Ghose, Veber, Egan, and Muegge rules which are predicated on specific physicochemical parameters like logP (for oral in range of 1.35–1.8, sub-lingual > 5) tPSA (should be < 140 Å), no. of donors (< 10), acceptors (> 5), etc.82. The predicted drug-likeness, PAINS, and synthetic accessibility properties of the selected compounds are shown in Table 5. According to the results, the selected compounds in Table 5 showed no violations of the Lipinski and Ghose rules. However, compounds TD75 and TD104 are acceptable with only one violation according to Veber, Egan, and Muegge rules. All of the different compounds in the PAINS investigation were not exhibiting any alerts, except for compounds C93 and C229.Table 5 Predicted drug-likeness, PAINS study, and synthetic accessibility measures of the screened compounds.Table 6 presents the predicted ADME properties of these compounds, including gastrointestinal (GI) absorption, blood–brain barrier permeation (BBB), inhibition of the CYP450 system, and permeability glycoprotein (P-gp) substrate. Furthermore, aqueous solubility and BBB values of the ligands preferably lie in the range of − 6.5–0.5 and − 3.0–1.2 respectively83,84 Also, p-glycoprotein (P-gp) non-substrate causes drug resistance85.Table 6 Predicted pharmacokinetics (ADME) parameters of the screened compounds.According to the results, all the compounds showed high GI absorption, except for compounds TD75 and TD106. BBB permeation potential was predicted for compounds C93 and C105. Compounds C115, C165, and C229 showed potential to be a substrate of P-gp. The potential to inhibit cytochrome P450 (CYP) isoforms was observed for compounds TD104 and TD106 (for 4 isoforms); T3 and T4 (for 3 isoforms); C97 and TD75 (for 2 isoforms); and C93 (for 1 isoform). Chalcone derivatives C105, C115, C146, C165, C191, and C229 were predicted to show no inhibitory activity against any of the CYP isoforms. The computed bioavailability score for all the compounds placed them within the 55–56% probability class, except for compound TD75 (0.11).After evaluating drug-likeness, and ADME characteristics, and studying PAINS alerts, the compounds C93, C229, and TD75 have been excluded from consideration for further investigation. A total of 10 compounds are now undergoing additional toxicological studies, utilizing VEGA QSAR and ProTox-II software.In silico toxicological resultsPredicting the toxicological properties and pharmacokinetic parameters of a compound plays a crucial role in the drug discovery process, as they collectively contribute to 60% of the failures in converting a lead compound into an effective drug86.Toxicological properties prediction using VEGA-QSAR modelsTo evaluate toxicological data, the QSAR modeling method was performed using VEGA-QSAR (https://www.vegahub.eu/portfolio-item/vega-qsar). The software-incorporated algorithm provides the evaluation of reliability prediction as Applicability domain index (ADI) value (Tables 7 and 8). It gives positive results with ADI > 0.5, as indicators of reliability effect; low (0.5 < ADI < 0.6), medium (0.6 < ADI < 0.8), and high (0.8 < ADI < 1).Table 7 Toxicological data of selected designed chalcone derivatives using VEGA-QSAR.Table 8 Toxicological data of selected designed 1,3-thiazole, 1,3,4-thiadiazole derivatives, and standard drug omapatrilate using VEGA-QSAR.All of the selected chalcone derivatives designed in this study were found to exhibit no developmental toxicity (PG model assessment)87, and inactive for estrogen and androgen-mediated effect (IRFMN/CERAPP model)88,89. Furthermore, they were confirmed to be non-reactive for Thyroid hormone receptor α/β (NRMEA model). All compounds have the potential for skin sensitivity, though the reliability of this prediction is low. Nevertheless, among them, only compounds C146, C165, and C115 were predicted to be non-mutagenic as indicated either by CAESAR or SarPy/IRFMN model assessment90 and non-carcinogenic according to the CAESAR or ISS models assessment91 (Table 7).Similar to chalcone derivatives, skin sensitization predictions using the IRFMN/JRC model indicated sensitization potential in both thiazole and thiadiazole compounds, except for compound T3. Both thiazole derivatives (T3 and T4) were predicted to be non-mutagenic by the AMES toxicity (SarPy/IRFMN model), non-carcinogenic (ISS model), non-indicative of developmental/reproductive toxicity (PG model), and inactive for both androgen receptor-mediated effects and thyroid receptor effects (IRFMN/CERAPP and NRMEA models). The two selected thiadiazole derivatives and standard drug Omapatrilate pass all screening parameters when evaluated through the applied QSAR model and their assessment scores. However, skin sensitization predictions have indicated sensitization potential in these compounds. Additionally, hepatotoxicity predictions have raised concerns about their potential toxicity (Table 8).GHS toxicity classification and prediction of LD50 using ProTox-IIThe GHS toxicity categorization places thiazole in Class III and the designed derivatives of chalcone and thiadiazole, in Class V. This indicates that compounds may be harmful if swallowed (2000 < LD50 ≤ 5000)92. The LD50 between 2000–5000 mg/kg indicates a safety range and values showed less potent toxic effects (Table 9).Table 9 GHS toxicity classification of selected designed derivatives.According to the above study, designed compounds C165, C115, T3, T4, TD104, TD106, and Q1934 would make better candidates for further synthesis and development.Molecular dynamics and MM-GBSA resultsMD simulation studies were carried out to understand the stability of protein–ligand interaction. As discussed earlier, the selected designed compounds with favorable screening properties of AD domain values of mt-QSAR models, docking score, and ADMET were selected for MD simulation studies.Omapatrilate was considered the standard multi-target inhibitor of ACE and NEP enzymes. Backbone RMSD analysis was evaluated (RMSD difference ≤ 2.0 Å) suggesting the stability of omapatrilate into the binding pocket of cACE (PDB ID: 1O80) and NEP (PDB ID: 5JMY) proteins. When the RMSD data were compared, each simulation including 20 ns revealed stable conformation. It was observed that Omapatrilate formed H-bond with various amino acid residues of cACE such as His353, Ala356, Glu384, Tyr523 and His410 whereas hydrophobic bond interaction with Trp357, Val380, His383, Phe457, Lys511, Phe512, His513, Val518, Tyr520 and Tyr523. The strong affinity of Omapatrilate with cACE was observed due to the formation of ionic interactions with amino acid residues His383, His387, His353, and Glu411. Additionally, it also forms salt bridge interaction with Glu143, His353, Ala354, Trp357, Glu403, and Pro519 with protein.Similarly, MD simulation of the Omapatrilate-NEP complex reveals the formation of strong ionic interactions with Arg110, His583, Glu584, His587, Glu646 and hydrogen bonding interactions with Asn542, Ala542, Ala543, Glu584, Trp693, His711, Arg717 amino acid residues. The formation of a few hydrophobic interactions with Phe106, Ile558, Val580, and Val710 favored the stability of the complex. Apart from RMSD, the RMSF value of a protein is widely used to access ligand-induced changes in the protein’s internal chains. Figures 14 and 15 show the RMSF plot of the Omapatrilate in complex with cACE and NEP enzymes, respectively.Figure 14MD simulation analysis of Omapatrilate-cACE complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the cACE (d) RMSD of the protein backbone.Figure 15MD simulation analysis of Omapatrilate-NEP enzyme complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the NEP enzyme (d) RMSD of the protein backbone.Likewise, ligand–protein interactions of the selected designed compounds were monitored during the same time trajectory simulation. There are four different categories of contacts viz. as hydrogen bonds, hydrophobic, ionic, and water bridges. The simulation interactions diagram represents more specific subtypes of interactions. Ligands who have maintained contact that occur 10.0% or more of the simulation time are discussed further. Although hydrophobic and hydrogen bonds are weaker compared to ionic bonds, they are too exploited most for the design of new drug candidates93,94.Ligand interactions of compound C115 at different time intervals were analyzed and checked for stability which showed that the proteins got stabilized and the ligand was forming interaction with the protein (RMSD difference ≤ 2.5 Å). In the cACE domain, C115 was forming H-bond interactions with Gln281, Ala354, Asp377, Lys511, His513, and Tyr520 amino acid residues while hydrophobic interactions with His353, Ala354, Val380, and His513 amino acid residues. Moreover, the salt bridge interaction with Ala356 and Tyr523 gives additional stability to the complex. The RMSD and RMSF plots of protein–ligand and the ligand–protein contacts for compound C115-cACE and compound C115-NEP enzyme complexes are shown in Figs. 16 and 17 respectively. On the other hand, compound C115-NEP complex stabilized by hydrogen bond formation with key amino acid residues Asn 542, and Ala543. Amino acid residues Phe106, Ile558, Val580, His583 and Trp 693 in the S1’ subsites contribute to stability by forming important hydrophobic interactions similar to that of the standard omapatrilate.Figure 16MD simulation analysis of compound C115-cACE complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the cACE (d) RMSD of the protein backbone.Figure 17MD simulation analysis of compound C115-NEP enzyme complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the NEP enzyme (d) RMSD of the protein backbone.On the other hand, compound T3 showed stable interactions throughout the simulation period (20.0 ns) which indicates the stability of the ligand in the binding site pocket of the protein (RMSD difference ≤ 2.0 Å). T3 was linked to amino acids like His383, Glu384, His387, and Glu411 by a metal ion coordinated within 3.4 Å of the protein’s and ligand’s atoms. Additionally, the selectivity towards the cACE domain is due to the formation of hydrogen bonding with Glu403, Asp415, Asp453, and Lys454 and hydrophobic interaction with Val380, His383, His513, and Tyr523 amino acid residues. The simulation interactions of the T3-NEP enzyme complex show that, the ligand occupies tightly in the S1 and S2 pocket of the enzyme through ionic interactions with Glu584, His587, and Glu646 amino acid residue. It also forms, an important hydrophobic interaction with catalytic residue His711 amino acid. Unlike, Ompatrilate the stability is contributed by the formation of multiple type contacts with His583, Val541, Ala543, and Tyr545 amino acid residue (Figs. 18 and 19).Figure 18MD simulation analysis of compound T3-cACE complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the cACE (d) RMSD of the protein backbone.Figure 19MD simulation analysis of compound T3-NEP enzyme complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the NEP enzyme (d) RMSD of the protein backbone.MD simulations trajectories revealed that compound TD104 was well stabilized (RMSD difference ≤ 2.0 Å) and made favorable metal ionic contacts and hydrogen bonding with an important catalytic site responsible for inhibition of cACE when compared to compound T3 over the entire simulation trajectory. Along with this Gln281, Ala354, Cys370, Asp377, and His513 amino acid residues play key roles in docked pose stability via H-bonding interaction. Also, hydrophobic contacts with Val380, His383, Lys511, Tyr523, and Phe512 are well maintained during the simulation trajectory. Correspondingly, compound TD104 shows the formation of hydrophobic interaction with Phe106, Trp693, and water bridge interaction with His711, and Asn542 in the catalytic domain suggesting its NEP enzyme inhibition probability. Furthermore, stability is provided by ionic interaction made by His583, His587, Glu584, Asp590, and Glu646 amino acid residues (Figs. 20 and 21).Figure 20MD simulation analysis of compound TD104-cACE complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the cACE (d) RMSD of the protein backbone.Figure 21MD simulation analysis of compound TD104-NEP enzyme complex (a) Simulation interactions diagram (b) Protein–ligand contacts histogram (c) RMSF of the amino acids comprising the NEP enzyme (d) RMSD of the protein backbone.Design compounds, C115, T3, and TD104 were found to form significant key interactions similar to omapatrilate towards the substrate binding pocket of cACE and NEP enzymes, and could probably be the novel lead molecules for multi-target inhibition against the target enzymes.So, the binding potential scores derived from the results of constraint-based docking provide a strong foundation for the MD simulation trajectories. Inferring that the compounds C115, T3, and TD104 have a strong affinity for the cACE and NEP enzymes. Following Lipinski, Ghose, Veber, Egan, and Muegge’s rule, all three compounds exhibited acceptable drug-likeness parameters. Furthermore, selected designed compounds (C115, T3, and TD104) do not have AMES mutagenicity and are Non-carcinogenic., no developmental/reproductive toxicity (PG model), and are found to be inactive for hormone receptors (estrogen, androgen, and thyroid α/β). All these parameters suggest compounds are safe to use. Hence, they could be considered for further synthesis and subsequent screening as promising scaffolds for the development of dual inhibitors targeting cACE and NEP enzymes, potentially contributing to the management of hypertension.The study conducted MM-GBSA calculations to estimate ligand binding energies or affinity (dG Bind), with the results presented in Table 10.Table 10 MM-GBSA values for the selected designed compounds.Based on the MM-GBSA calculations, the ΔG bind values for molecules such as C115, T3, and TD104 were found to be − 16.46, − 12.19, and − 27.76, respectively, for the cACE protein. These binding energies of T3 and C115 closely resemble those of Omapatrilate, indicating a strong binding affinity to the cACE enzyme. Additionally, the ΔG values against NEP were − 8.267, − 5.465, and − 5.574 for the compounds C115, T3, and TD104, respectively, which are in proximity to the ΔG values of Omapatrilate.Based on the two findings, TD104 and C115 exhibited the highest binding affinities for cACE, with C115 demonstrating strong binding with NEP. Hence, the designed compounds C115 and TD104 show promising potential as novel therapeutics and could serve as valuable leads for the synthesis of cardiovascular agents.
