Drug-likeness and Toxicological Profiling of Designed Quinoline Analogues as Anti-Tubercular Agents: An In-Silico Approach

In drug discovery, medicinal chemists reduced the frequency of clinical drug trials by using a computational approach to improve the success rate in identifying potential and safer drug candidates. This approach makes the drug discovery process less expensive, less time-consuming, and even reduces animal deaths due to in vivo animal studies. Predicting the ADMET (absorption, distribution, metabolism, excretion, and toxicity) profiling of designed drug pharmacophores before their synthesis is safer than in vivo studies. Therefore, pharmacokinetic and toxicological profiling of the designed ligands is predicted to determine their biological potential. The quinoline is a pharmacologically important fused heterocyclic scaffold bearing two six-membered rings, i.e., a pyridine and a benzene ring. This naturally occurring heterocyclic scaffold is found in cinchona alkaloids. Furthermore, their semi-synthetic and synthetic candidates have a wide range of reported pharmacological effects, like anti-malarial, anti-tubercular, antimicrobial, and antitumour [1-2] .  In this research study, in silico methods are employed to compare drug-like activity and toxicity studies of substituted quinolines to predict future drugs for safe and effective tuberculosis treatment. Herein, we designed new quinolines (Figure 1) by incorporating variants of the piperazine moiety and benzylamine into the 4^th position of the quinoline scaffold to obtain novel hybrids inspired by the reported compounds of two research teams, i.e., Pissinate et al. 2016 [3], and Giacobbo et al. 2017 [4]. Both research groups demonstrated the synthesized compounds for TB potential.

Figure 1. Newly designed 4-substituted quinolines

Thus, five newly designed 4-substituted quinoline analogues were subjected to an in silico computational approach; their drug-likeness parameters and other physicochemical and toxicological characteristics were determined using pkCSM, admetSAR 3.0, ProTox 3.0, Swiss ADME, and ChemDraw software.

MATERIALS AND METHODS:

Computer System and Software

Computer system (HP) with a 64-bit operating system having the following specifications: CPU @ 2.40GHz, Intel® 11th Gen Intel Core i5, 8.00 GB installed RAM was used during this research study. ChemBioDraw Ultra software V.14.0 for drawing structures. Apart from this, ProTox 3.0 [5], pkCSM [6], admetSAR 3.0 [7] and Swiss ADME [8] online web server tools were used for the in-silico research findings.

In silico predictions of drug-likeness and toxicity studies

In silico predictions for the designed analogues were performed using freely available pkCSM, admetSAR 3.0 and SwissADME, free web-server tools that permit worldwide medicinal researchers and academicians to determine the drug-likeness and toxicological profiling of small molecules. Herein, pkCSM and admetSAR 3.0 were used to determine the physicochemical parameters, while SwissADME was used for the boiled-egg plot [9] and bioavailability radars [8-9]. Additionally, the toxicological profiling was obtained using ProTox 3.0 [5].

RESULTS AND DISCUSSION:

Drug-likeness studies

An in-silico drug-likeness analysis was performed on designed 4-substituted quinolines using pkCSM, admetSAR 3.0, and Swiss ADME web-based tools. These tools were used to assess various parameters related to drug-likeness, lipophilicity, aqueous solubility, and toxicity studies. In the drug discovery process, Drug likeness is a qualitative method used in the initial stages for selecting a lead molecule. Its theoretical advancement is presumed to be Lipinski’s characteristics, known as the ‘rule of five’. According to pharmacokinetic characteristics, only those drug candidates are considered therapeutically effective and safer which can cross the biological membrane and bind to the desired target site in the human body at optimal concentration. Initially, Lipinski’s characteristics were determined for the designed derivatives S01-S05 along with three reference drugs, rifampicin (RIF), isoniazid (INH) and bedaquiline (BDQ). The in-silico predictions for the designed compounds S01-S05 were within the acceptable or optimal range, without any violations (Rule-of-five). This indicates the probability of a designed scaffold being AV when taken orally (i.e., drug-likeness). The results indicated that the designed derivatives S01-S05 are likely to have more drug-like properties compared to reference drugs, as shown in Table 1.

Table 1. Predicted Lipinski’s Drug-likeness properties of S01-S05 for drug-likeness

^**Standard drugs: RIF = Rifampicin; INH = Isoniazid; BDQ = Bedaquiline; 1#all the physicochemical properties determined by pkCSM, admetSAR 3.0 and ProTox 3.0; Mol.Wt. = Molecular weight; LV = Lipinski’s violations, LogP = Octanol-water partition coefficient; H-A = Hydrogen bond acceptor; H-D = Hydrogen bond donor; Molr. Ref. = Molar refractivity.  In SwissADME, the boiled egg plot is a highly reliable method for predicting gastrointestinal tract (GIT) absorption passively and blood-brain-barrier (BBB) permeation of designed quinoline analogues. In the boiled egg plot, egg-white provides a visual representation of GIT absorption, while the egg yolk represents BBB penetration of the compounds. The boiled-egg graphs (Figure 2) and bioavailability radars (Figure 3) are the two graphical outputs of the Swiss ADME for the five designed quinolines derivatives. Boiled-egg graphs generated by Swiss ADME used two prime and classical physicochemical descriptors. Among them, one is WLOGP (partition coefficient between n-octanal and water) for lipophilicity, and another is TPSA (topological polar surface area) for apparent polarity. The predictions of all concerned descriptors are reported in Table 2.  

Table 2. In silico predicted physicochemical parameters of the designed compounds S01-S05

^a boiled-egg graphs generated by these parameters, ^b bioavailability radars generated by these parameters.

In addition to this, substrate of P-gp (P-glycoprotein) determination is another benefit of boiled-egg plots and is directly linked to the AV efflux of compounds through both biological barriers (GI wall and BBB). Blue dot predicts the compound as a P-gp substrate (PGP+), which is AVly effluxed from both barriers. Red dot predicts compound as a P-gp non-substrate (PGP−), so it is passively absorbed/effluxed through the GI wall and has passive BBB access. Therefore, two compounds, S02 and S04, were effluxed from the brain because these are identified as P-gp substrates (blue dot in boiled-egg graph), but can passively cross the BBB (egg yolk region), while three compounds, S01, S03, and S05, were identified as P-gp non-substrates, therefore, brain-penetrants (egg yolk region) but cannot AVly efflux (red dot in boiled-egg graph).

Figure 2. Boiled-egg graphs determined by Swiss ADME for S01-S05

The bioavailability radars also predict the drug-like behaviour of the designed quinoline derivatives. In these graphical radars, the coloured area represents the acceptable range of each of lipophilicity (LIPO), flexibility (FLEX), size, insaturation (INSATU), solubility (INSOLU), and polarity (POLAR). These predictors are essential for oral bioavailability. To consider the designed analogues drug-like, the bioavailability radars of the predicted compounds must be within the pink coloured (Figure 3). Herein, the bioavailability radars of the four compounds S01, S02, S03, and S04 fall entirely within the pink region, indicating their drug-like nature. These are orally bioavailable compounds. Among these compounds, S04 meets all six properties to a greater extent. The radar for the compound S05 falls somewhat out of the pink region due to the Csp³ parameter (0.16).

Figure 3. Bioavailability radars for S01-S05

Toxicological Profiling of S01-S05 via ProTox 3.0

The toxicological profiling of five newly designed compounds was investigated using a web-server tool named ProTox 3.0. This tool predicts the oral toxicity studies (Table 3) and organ-specific toxicities (Table 4). The oral toxicity predictions include the predicted LD₅₀, toxicity class, average similarity, and accuracy percentage. ProTox 3.0 classified drugs into six different classes based on predicted LD₅₀. Class I drugs are fatal if taken orally (LD₅₀ must be less than 5); Class II drugs also fatal if taken orally (LD₅₀ must be in between 5-50); Class III drugs are toxic if taken orally (LD₅₀ must be in between 50-300); Class IV drugs are harmful if taken orally (LD₅₀ must be in between 300-2000); Class V drugs may be harmful if taken orally (LD₅₀ must be in between 2000-5000); and Class VI drugs are non-toxic (LD₅₀ must be greater than 5000). The predicted results indicated that the designed derivatives S01-S05 are safer compared to standard drugs. Three compounds, S01, S04 and S05, lie in IV toxicity class, and two compounds, S02 and S03, lie in V toxicity class. On the other hand, INH and BDQ, predicted as Class III drugs. None of the designed compounds were predicted as class III drugs (Table 3). 

Table 3. Toxicity predictions of S01-S05 when taken orally

^**Standard drugs: RIF = Rifampicin; INH = Isoniazid; BDQ = Bedaquiline.

ProTox 3.0 predicts organ-specific toxicities related to liver (hepatotoxicity), brain (neurotoxicity), kidney (nephrotoxicity), lungs (respiratory toxicity), and Heart (cardiotoxicity). The predicted organ-specific toxicological profiling of the newly designed derivatives is shown in Table 4. These results indicated lower toxicity of the designed derivatives compared to standard drugs.

Table 4. Predicted organ-specific toxicological profiling of the newly designed derivatives

^**Standard drugs: RIF = Rifampicin; INH = Isoniazid; BDQ = Bedaquiline; IA=Inactive, AV=Active