Target identification, screening and in vivo evaluation of pyrrolone-fused benzosuberene compounds against human epilepsy using Zebrafish model of pentylenetetrazol-induced seizures

Pyrrolone-fused benzosuberene (PBS) compounds were semi-synthesized from α,β,γ-Himachalenes extracted from the essential oil of Cedrus deodara following amino-vinyl-bromide substituted benzosuberenes as intermediates. These PBSs compounds classified as an attractive source of therapeutics. The α-isoform of PI3K which is a pivotal modulator of PI3K/AKT/mTOR signaling pathway, responsible for neurological disorders like epilepsy, found as a potential target molecule against these 17 semi-synthesized PBS compounds using in silico ligand-based pharmacophore mapping and target screening. The compounds screened using binding affinities, ADMET properties, and toxicity that were accessed by in silico docking simulations and pharmacokinetics profiling. Ultimately two compounds viz., PBS-8 and PBS-9 were selected for further in vivo evaluation using a zebrafish (Danio rerio) model of pentylenetetrazol (PTZ)-induced clonic convulsions. Additionally, gene expression studies performed for the genes of the PI3K/AKT/mTOR pathway which further validated our results. In conclusion, these findings suggested that PBS-8 is a promising candidate that could bedeveloped as a potential antiepileptic.

Identification of a target molecule. Further, to identify the target molecule against 17 PBS compounds, we used a ligand-based virtual screening approach 21 with the help of Accelrys Discovery studio package. The 3D www.nature.com/scientificreports www.nature.com/scientificreports/ pharmacophore model against these PBS ligands were mapped using the interaction pattern of cations, anions, aromatic, aliphatic, hydrophobic and hydrogen bond donors/acceptors 5 . The pharmacophore model thus generated was then used to search the pre-existing structured databases to identify the molecular structure that best matches with the pattern of that pharmacophore map. This similarity search unearths PI3K (α-isoform) as the biological target against our PBS compounds.

Analyses of binding energies and binding interactions. For enumeration of specific inhibitors against
α isoform of PI3K lipid kinase, we docked our 17 naturally originated compounds with this isoform. We calculated the energy of interaction between PI3K-α and 17 PBS ligands. Docking with Autodock 4.2.6 22 Table 1. The atomic interactions were further explored by LigPlot + v.1.4 software 23 . This software could plot 2D views of in-depth ligand

Analyses of drug-likeliness.
To determine the drug-likeliness of our PBSs compounds, we calculated their ADMET properties. The screened results of ADMET were summarized in Table 2, revealing six descriptors such as absorption, solubility, cytochrome P 450 2D6 (CYP2D6), plasma protein binding (PPB), hepatotoxicity, and  www.nature.com/scientificreports www.nature.com/scientificreports/ AlogP98. Easy absorption of the drug through blood brain barrier (BBB) measured by its AlogP98 value which must be less than 5. The obtained absorption levels determine drug absorption and absorption decreases inversely with the level, i.e., level 0 denotes proper absorption, level 1 denotes moderate absorption and so on. To determine the inhibitory effect and toxicity of the drug CYP2D6 and hepatotoxicity descriptors gave two predicted classes, such as class 0 (non-inhibitor and non-toxic) and class 1 (inhibitor and toxic). Another ADMET descriptor PPB has given two results true or false that symbolizes the binding or non-binding of the drug. Moreover, solubility descriptor predicted the molar solubility of the drug, which gives good molar solubility within a range of −2.0 and −4.0. If it is below that range, solubility decreases gradually and becomes extremely low below −8.0, and above that range, it increases gradually to become too soluble at 0.0.
ADMET model also developed using two descriptors such as 2D polar surface area (2D-PSA) and AlogP98 (Fig. 5). This model includes the eclipses of 99% and 95% confidence limits which were used to define the regions with compounds having proper intestinal absorption and penetration across the BBB. The results of the obtained  Table 2. ADMET descriptors. AlogP98 must be less than 5 for good absorption through BBB. Absorption of drug was determined by obtained levels: 0(good), 1(moderate), 2(low), 3(very low). Hepatoxicity determines toxicity of drug by predicted classes: 0(non-toxic) and 1(toxic). CYP2D6 descriptor determines inhibitory effect by predicted classes: 0(non-inhibitor) and 1(inhibitor). PPB (plasma protein binding) determines binding of drug, true symbolizes binding and false symbolizes non-binding. ADMET solubility descriptor predicts molar solubility of drugs within the ranges: −6.0 to −4.0 (low solubility), −4.0 to −2.0 (good solubility), and −2.0 to 0.0 (optimal solubilty). www.nature.com/scientificreports www.nature.com/scientificreports/ ADMET model showed that all our PBS compounds fell inside all the eclipses expect two of them fell outside the eclipse of 95% confidence limits for intestinal absorption.
Another method used to determine the therapeutic compatibility of the drug is toxicity prediction by komputer assisted technology (TOPKAT), summarized in Table 3. TOPKAT is a useful tool for in-silico prediction of toxicity quantitatively, and it is employed in quantitative structure-activity relationship (QSTR) models. Moreover, with the help of these QSTR models, it calculates probability values and evaluates toxicity through them. It follows the criterion of checking the components in the optimal predictive space (OPS), and when they lie outside then the results were considered as unreliable, i.e., false positives. Obtained Ames probability values determine the level of toxicity, such as, 0.0 to 0.30 (non-toxic), 0.30 to 0.70 (inter vocal), and 0.70 to 1.0 (toxic). Other additional descriptors provided was Ames mutagenicity, Ames enrichment and Ames score to determine the reliability of the predictions. It also provides values of the carcinogenic potency of TD50 mouse, rat oral LD50, rat inhalational LC50, and Daphina EC50. Increase in TD50, LD50, LC50, and EC50 values predicts the decrease in toxicity and increase in safety index of the drug which makes it more potent.
Further analyses of drug-likeliness were performed by Lipinski's rule-of-five to determine the ability of the drug to diffuse passively through the BBB. Lipinski's rule-of-five follows the criteria of number of violations listed in Table 4 such as, molecular weight of compound should be less than 500, AlogP value should be less than 5, hydrogen bond donors should be less than 5, hydrogen bond acceptors should be less than 10, and number of rotatable bonds should be less than 5, respectively.
Effect on PTZ-induced clonic seizures. The exposure with 8 mM PTZ showed the appearance of clonic seizure in vehicle control larvae with a latency of 4.42 ± 0.15 min. Pre-incubation with 1 µM concentration of PBS-9 (P = 0.002) and PBS-8 (P < 0.001) resulted in a marked increase in latency to clonic-like seizures in comparison to vehicle control. However insignificant difference in latency to first clonic-like seizure was observed at 1 µM concentration of PBS-9 and PBS-8 (P = 0.670). The group pre-incubated with sodium valproate showed a significant (P < 0.001) increase in latency to clonic-like seizures. The effect of both PBS-8 (P = 1.00) and PBS-9 (P = 0.639) at 1 µM concentration was found to be equipotent when compared to sodium valproate. Both test compounds were found to be ineffective at 0.25 µM and 0.5 µM concentrations as compared to vehicle control (Fig. 6).
Effect on mRNA levels. Larvae exposure to PTZ resulted in a significant increase in mRNA levels of  www.nature.com/scientificreports www.nature.com/scientificreports/ decreased in PBS-8 (P = 0.004), and PBS-9 (P = 0.005) exposed larvae in contrast to vehicle control. Furthermore, pre-incubation with a 1 µM concentration of PBS-8 and PBS-9 showed a significant (P < 0.001) decrease in mRNA levels of AKT, PIK3CA, PIK3R1, mTOR, and Rps6kb1 as compared to the vehicle control larvae. A marked decrease in mRNA level of Rps6 was also observed in the larvae exposed to PBS-8 (P = 0.049) and PBS-9 (P = 0.005) and as that of vehicle control (Fig. 7).

Discussion
For identification of the biological target against ligand 1-17, they were undergone through in silico studies. The procedures of virtual screening were performed to identify the biological target using Discovery studio package. Virtual screening is an efficient approach that is widely used for the discovery of novel compounds 24 . In the absence of target molecule information ligand-based virtual screening approach had been successfully applied, such as pharmacophore mapping and similarity searching 5 .
PI3K-α was a resulted target of ligand-based virtual screening against these 17 PBSs compounds. PI3K as reported plays an essential role in the activation of mTOR signaling pathway. The role of the PI3K/AKT/mTOR pathway has been widely deciphered in epilepsy 14,25 . Studies suggested that injury due to seizures led to activation of the pathway which further propagates seizure progression and related pathogenic changes 26 . Furthermore, the constitutive stimulation of the pathway has been well explored in various in vitro 27 and in vivo models 28 . Studies conducted on PI3K suggested its phosphatidylinositol group to be responsible for the emergence of a large number of mitotic factors, thus resulting in the proliferation of cells 29 through phosphorylating AKT and activation of its downstream genes such as mTOR, Rps6, and Rps6kb1. It has been found that all mammalian cells when activated by receptor tyrosine kinases, expresses at least one of the isoforms of PI3K. Moreover, the PI3K (class IA) is functional when the catalytic subunit p110α binds with its regulatory adapter protein p85α to form a dimer 30 . Thus this target is beneficial for the development of new treatment avenues in epilepsy.   www.nature.com/scientificreports www.nature.com/scientificreports/ In this study, in silico docking and pharmacokinetic profiling were performed for screening of more drug-likely PBSs ligands against the PI3K-α protein. Molecular docking is a computational approach widely used in the drug discovery process 31,32 . These PBSs compounds were then screened by obtaining binding energies using docking simulations. Smaller the binding energy, more potential it is. Based on binding energies nine compounds were screened viz., PBS-2, PBS-3, PBS-5, PBS-6, PBS-8, PBS-9, PBS-10, PBS-11 and PBS-12, and based on the torsional free energy they were screened down to eight viz., PBS-3, PBS-5, PBS-6, PBS-8, PBS-9, PBS-10, PBS-11, and PBS-12. Consequently, the screening of these remaining eight PBS compounds were done using their ADMET properties. PBS-12 was screened out based on ADMET adsorption level descriptor as well as PBS-5 was screened out based on its inhibition effect and toxicity showed by CYP2D6 and hepatotoxicity descriptors. After ADMET screening six compounds were selected viz., PBS-3, PBS-6, PBS-8, PBS-9, PBS-10, and PBS-11. Further screening of PBSs compounds were done based on TOPKAT results. PBS-3 and PBS-11 compounds were screened out based on their low potency values obtained by TOPKAT carcinogenic potency of TD50 mouse descriptor. Moreover, PBS-6 and PBS-10 were screened out based on the applicability of rat oral LD50, rat inhalational LC50, and Daphina EC50. Finally,in silico docking studies and pharmacokinetic profiling suggested that PBS-8 and PBS-9 compounds were found suitable inhibitor for PI3K-α. Finally, they were evaluated through in vivo studies using zebrafish as a model organism for human epilepsy.

PBS compounds
Since the past few decades, zebrafish has gained popularity as a developed disease model. The foremost criterion for using zebrafish in epilepsy research is to ascertain seizure like clonic convulsions as depicted by the Racine scale in rodent model 33 . Accordingly, in the present study, seizure-like behavior in 7 dpf (days post fertilization) zebrafish larvae were established using PTZ, and latency to first clonic seizure was recorded. The study found that, exposure to the test compounds (PBS-8 and PBS-9) depicted a considerable increase in PTZ-mediated clonic seizure latency at 1 µM concentration. Epileptic studies conducted on rodent models have revealed that seizure generation leads to immediate early genes expression, particularly c-fos upregulation in the brain 34 , a similar observation has also been made earlier in the zebrafish larvae model 33 . In line with this observation, the present study showed increased c-fos expression in PTZ exposed vehicle control larvae, and subsequent decrease upon pre-incubation with PBS-8 and PBS-9, supporting its anti seizure effect. Our findings were further supported by previous work showing c-fos downregulation in larvae treated with an antiseizure compound 35 .
Our study revealed that the genes encoding these units, i.e., p110α (PIK3CA gene) and p85α (PIK3R1 gene) were upregulated in the larvae exposed to PTZ. Pre-incubation with PBS-8 and PBS-9 showed a reduction in their expression. The fact that the intricate mechanism of PI3K/AKT activation leading to mTOR hyperactivation in rodent models of epilepsy 36 can be corroborated with our findings, which showed that the expression of all the www.nature.com/scientificreports www.nature.com/scientificreports/ downstream genes, i.e.,AKT, mTOR, Rps6, and Rps6kb1 were reduced dramatically following drug treatment on PTZ treated larvae.
The study identified few potential lead compounds against epilepsy. Our adopted approaches addresse the complexity in searching enormous natural bioactive space. Moreover cellular targets of a natural lead is crucial for the process of drug discovery. We strongly recommand computational exploration in target identification and screening of lead before going for an in-vivo analysis. It helps to reduce the effort and time of a researcher.

Materials and Methods
Synthesis of chemical compounds. All the 1-17 pyrrolone-fused benzosuberene (PBS) compounds used under this study were synthesized following our earlier developed protocols 20 . These compounds (ligand 1-17) were synthesized with different functional groups at the specific position denoted by R, as shown in Fig. 2. Further, these molecules were investigated for therapeutic applications.
Ligand-based Virtual screening to natural analogs. Accelrys Discovery studio package (Dassault Systèmes BIOVIA, 2017R2, San Diego) was used for deriving pharmacophore mapping 5 , that is a type of ligand-based virtual screening. Naturally extracted seventeen ligands were fed in the Accelrys Discovery studio package to generate pharmacophore model, with default parameters. The pharmacophore model was selected and then used to search the 3D structure database to identify the appropriate receptor structure.
Protein dataset. The crystallographic structure of PI3K (α-isoform) class I lipid kinase was achieved from Brookhaven PDB (Protein Data Bank; www.rcsb.org) server 37 . We selected a catalytic subunit of α isoform (PDB ID: 4JPS) of Homo sapiens organism, solved by X-ray diffraction method at a resolution of 2.2 Å 13 .
Pharmacokinetic properties. Drug-likeliness of 17 PBS compounds were analyzed by assessing Lipinski's rules, Absorption, distribution, metabolism, excretion, and toxicity (ADMET) descriptors, and TOPKAT descriptors using Accelrys Discovery studio package. ADMET analyses were performed using sixdescriptors, such as absorption, solubility, CYP2D6, Plasma Protein Binding (PPB), hepatotoxicity, and AlogP98. Also, Toxicity Prediction by Komputer Assisted Technology (TOPKAT) analyses were performed using carcinogenic potency of LD50 mouse, Ames mutagenicity, Ames probability, Ames enrichment, Ames score, rat oral LD50, rat inhalational LC50, and Daphina EC50. Zebrafish maintenance. Adult wild-type zebrafish of 4-5 months were housed in a ZebTEC Stand-Alone system (Tecniplast, Buguggiate, Varese, Italy) set at temperature 26-28 °C, pH 7.0-7.5 and conductivity 400 -600 µS. The room photo period cycle was maintained at 14:10 h light: dark and fish were fed twice a day with freshly hatched live Artemia (Inve Aquaculture, Inc., Salt Lake City, USA). The experimental protocol was duly approved by the Institutional Animal Ethics Committee of CSIR-IHBT and was performed in accordance with the approved guidelines.
Egg collection. The eggs were obtained from the induced spawning of healthy adults in separate breeding tanks. Briefly, at the end of the light cycle of the day prior to collection of eggs, two males and four females (1:2 ratio) were transferred to a breeding tank (Tecniplast, Buguggiate, Varese, Italy) separated by a transparent divider, containing system water maintained at 28.5 °C temperature. The breeding setup comprised of an internal grid bottom tank (Model: ZB10BTI) with a sliding transparent divider (Model: ZB10BTD), fitted into the external solid bottom tank (Model: ZB10BTE) covered with a transparent lid (Model: ZB10BTL). On the day of egg collection, the lid was removed after 30 min of the start of the light cycle. Healthy fertilized eggs were collected in sterile petri dishes using a pipette and were cleaned with 2-3 rounds of washing with system water. Around 50 eggs per plate were kept in a BOD incubator (Relitech, Ambala, India) at 28.5 °C. Regular water changes (twice a day) were done till 7 dpf. PTZ-induced epileptic seizures. The larvae at 7 dpf in different groups (n = 6) were pre-incubation with  Table 5. Primer sequence of target genes.