Denovo Designing, Virtual Screening and Lead Optimization of Potential Drug Candidate for Herpes Disease

Herpes simplex viruses (HSV) 1 and 2, the members of herpes virus family Herpesviridae are ubiquitous, highly contigious agents which infect humans [1,2]. They cause cold sores, herpetic keratitis and genital herpes [3-5]. An estimated 400 million people worldwide are currently infected by HSV-2 and infection of HSV-1 is spreading to developed regions like USA, Western Europe, Australia and New Zealand [WHO Fact Sheet, 2015]. HSV-1 persists in the body in dormant state by escaping the immune system and results in sporadic viral reactivation [6].


Introduction
Herpes simplex viruses (HSV) 1 and 2, the members of herpes virus family Herpesviridae are ubiquitous, highly contigious agents which infect humans [1,2]. They cause cold sores, herpetic keratitis and genital herpes [3][4][5]. An estimated 400 million people worldwide are currently infected by HSV-2 and infection of HSV-1 is spreading to developed regions like USA, Western Europe, Australia and New Zealand [WHO Fact Sheet, 2015]. HSV-1 persists in the body in dormant state by escaping the immune system and results in sporadic viral reactivation [6].
HSV evades the immune system by interfering with MHC Class I antigen presentation on the cell surface. Infected cell protein (ICP) 47 caused major histocompatibility complex (MHC) class I proteins retaintion in the endoplasmic reticulum (ER). MHC presents the antigen to CD + 8 T cells that are inhibited after expression of ICP 47 in the cell [7]. ICP 47 blocks peptide (viral epitope) transport across the ER membrane by transporter associated with antigen processing (TAP) [8,9] so that MHC class I proteins remain in ER without peptides and cytotoxic T-lymphocytes activation can be stopped. This allows virus to reside for a protracted period in the host.
Due to serious threat to human population a proper treatment and cure of HSV is required. For last two decades HSV glycoprotein D, has been the predominant HSV vaccine candidate [10,11], but the outcome of clinical trials of the vaccine have been really disappointing [12]. Antiviral drugs (valacyclovir, acyclovir, famciclovir, etc.) used against HSV infection, inhibit DNA polymerases and slows down infection by reducing viral replication but does not cure the infection completely [13]. Inappropriate prescribing and wide spread uncontrolled use of antiviral drugs led to emergence of resistance in the viruses. Recently, it has been observed that natural serine protease inhibitor (serpins) like serpinantithrombin III (AT III) inhibit HSV infection during an entry event. Antithrombin III demonstrated a promising result at experimental levels but yet it is far from any therapeutic use [13,14].
Due to inaccessibility of any effective drug and vaccine there is therefore, an urgent need for an effective HSV vaccine or drug that provides protection against infection and also thwart the virus entering a latent state [15]. Computer aided drug discovery design (CADD) has emerged as a way to significantly decrease compounds screening while retaining same level of lead compound discovery at the same time [16]. In the present manuscript we have designed lead compound by denovo designing strategy and molecule having highest score was further optimized by virtual screening to increase its affinity to target and reduce toxicity by toxicological analysis by online ADMET profiling tools of FAF DRUG 20 online tool and ORISIS data warrior.

Sequence retrival
DNA sequence of ICP-47 of HSV-2 was retrieved from NCBI and it was blasted to human genomic and transcriptomic database. The structure of target protein ICP-47 was retrieved from rcsb.org in PDB (text) format (PDB I.D. 1QLO).

Denovo drug designing
After exhaustive literature survey and database searching, no ligand was found that bind to this target, so automated de-novo drug design strategy was adopted to get a hit molecule. eLEA3D program was used for denovo generation of ligand, which create new molecules by using a library of molecular fragments and by determining best combinations of molecular fragments that fit user-defined physicochemical properties

Abstract
Herpes simplex virus (HSV1, HSV 2) is a neurotropic and neuroinvasive virus which becomes latent and causes a lifelong infection. HSV-1 and 2 produce infected cell protein (ICP)-47 against MHC class I antigen presentation pathway by inhibiting the transporter associated with antigen processing (TAP). ICP 47 is also responsible for evasive nature of HSV in human immune system. Currently available antiviral drugs and vaccines only slowdowns the infection but it does not cure the infection. In present study, we have in silico designed a potential drug candidate against HSV ICP-47 target through de-novo pathway using eLEA3D. The derived ligand docked with the natural viral receptor ICP-47 showed the binding affinity of -4.07, but it was found toxic in FAF DRUG online ADMET tool, due to presence of high risk imine group. Further manual optimization led to generation of many bioisosteres and final lead structure showed no toxicity and a high binding affinity of -7.53. Our designed lead can act as a potential therapeutic compound against HSV. used for the de novo generation of the ligand was based on genetic algorithm that evolves the molecular structures generation after generation until the appearance of fitted molecules [17]. Each molecule of each generation is evaluated by fitness function (constraints) which is either molecular properties or an affinity prediction by a docking program. Submitting the query generated 10 structures of possible lead molecule along with their scores. The best structure had a maximum scoring percentage of 71.46% ( Figure 1). Rest structures had the scoring percentage of 50 and hence only first result was chosen as a lead molecule for further steps.
De-Novo designed lead molecule was further docked with ICP-47 and analyzed with FAF DRUG 2.0 ADMET tool. Lead molecule showed a binding affinity of -4.07 and a high risk of toxicity due to presence of imine group.
To increase binding affinity and reduce toxicity, lead compound was optimized manually by combinatorial optimization method. Due to small size of lead molecule it became possible to mark all replaceable functional groups and substitute with similar functional groups to create bioisosteres [20]. Each functional group replacement was docked with ICP-47. Out of total 57 replacements of functional groups, (also called constraint function) [17]. For De-novo ligand design, PDB structure of the target protein is uploaded in the server. Binding site for the ligand is set around 17th residue (Valine) with the binding site radius of 10 angstrom. Weight in the final score and conformational search for the given ligand was set to one, also ionization of carboxylates, phosphates, and guanidiniums was allowed.

Autodocking and toxicity profiling
The resulting ligand was docked with natural viral receptor ICP-47 in Autodock4 [18] and toxicity profiling was done with FAF DRUG 2.0 online ADMET tool [19]. To increase its binding affinity and reduce toxicity we manually optimize the ligand by virtual combinatorial ligand optimization. The members of each class of substituent were numerated and combinatorially combined to create bioisosteres. Each functional group replacement was docked against the ICP-47 receptor taking other functional groups constant to check whether they are positively affecting the binding affinity and ADMET prolife is also studied simultaneously. Threshold binding energy value is set to -4.11. All functional group possessing binding energy less than -4.11 were selected for further optimization by OVAT (One Variable at a time) method and progressive method.

Results
The blast search of the DNA sequence was performed to see whether ICP-47 of HSV showed any similarity with the human or not and it did not show significant similarity to any of the DNA/ protein sequence. Firstly, the available chemical databases were screening for their binding to the HSV ICP-47 (PDB I.D.-1QLO). No ligand was observed which had the binding affinity to target protein. Hence, automated denovo drug design strategy was adopted to get a hit molecule. eLEA3D (Ligand by Evolutionary Algorithm) programme was  10 functional group were selected having better value than threshold of -4.11 (Table 1). NH 2 functional group has only one better replacement (-CH 3 ), so we replaced it in all further structural analysis, for the other 9 passed groups exhibiting binding energy >-4.11 (Figure 2), the OVAT method was utilized and docking with ICP-47 and ADMET profile was simultaneously studied (Table 2). Total 24 iterations were performed which showed promising results with top binding energy of -7.16, -6.6, -6.57 but toxicity still persisted (Table 3).
On analysis it is found that toxicity was being caused by the nitrogen ring replacement, so progressive method was used to generate structure excluding the ring replacements. This gave the best result of -5.96 but the toxicity was still there (Figure 2). To remove toxicity, N atoms of the ligand with the best binding energy (-7.16) were replaced by carbon one at a time and ADMET profiling was done by FAF DRUG 2.0 until any non-toxic ligand was generated. So it would be much close to natural biological compounds and toxicity of imine group and azo group can be removed.
It was further docked in Autodock 4 and finally a non-toxic lead molecule having binding energy of -7.53 (Figures 3 and 4) was obtained. The toxicity of the lead molecule is checked with FAF Drug 2.0 which categorized it non-toxic ( Figure 5). Toxicity was also crosschecked with help of ORISIS data warrior which categorized final lead molecule as an irritant. The lead molecule is also checked for binding with any receptor 1.

Discussion
In recent years computer-aided drug design (CADD) approaches have revolutionized the field pharmaceutical research and many powerful standalone tools for CADD were developed, e.g., e-LEA3D, i Drug, iSMART etc. [17,22,23]. In these web servers many techniques were combined (pharmacophore mapping, similarity calculation, scoring, and target identification) to create a web interface which was more consistent and user friendly. Hartenfeller and Schneider have also emphasized that novel biologically active drug like lead molecules can be generated computationally on the basis of rule based fragment assembly [24]. Our initial hit molcule was also generated by fragment assembly method of e-LEA3D. Villoutreix et al.. also generated a non enzymatic tyrosine kinase inhibitors lead by virtual screening and its ADMET profiling was done with FAF Drug 2.0. These lead molecules can be used to discover potential Anti-Allergic drugs [25]. Similarly rational drug designing was used to generate many drug like molecules e.g. cimetidine (prototypical H 2 -receptor antagonist) [26], enfuvirtide [27], nonbenzodiazepines [28] and raltegravir [29].

Conclusion
The lead molecule generated by the procedure can be checked for it chemical synthesizability and can checked for its in vivo ADMET profiling in clinical trials. Its biding with other receptors can be further analysed to access its pharmaceutical potential in other diseases. Denovo drug design and virtual lead optimization and in vitro screening can further lead to decrease the timeline of any drug development and hence holds true potential in modern pharmacology and rational drug design.