Identifying HSV-1 Inhibitors from Natural Compounds via Virtual Screening Targeting Surface Glycoprotein D

Herpes simplex virus (HSV) infections are a worldwide health problem in need of new effective treatments. Of particular interest is the identification of antiviral agents that act via different mechanisms compared to current drugs, as these could interact synergistically with first-line antiherpetic agents to accelerate the resolution of HSV-1-associated lesions. For this study, we applied a structure-based molecular docking approach targeting the nectin-1 and herpesvirus entry mediator (HVEM) binding interfaces of the viral glycoprotein D (gD). More than 527,000 natural compounds were virtually screened using Autodock Vina and then filtered for favorable ADMET profiles. Eight top hits were evaluated experimentally in African green monkey kidney cell line (VERO) cells, which yielded two compounds with potential antiherpetic activity. One active compound (1-(1-benzofuran-2-yl)-2-[(5Z)-2H,6H,7H,8H-[1,3] dioxolo[4,5-g]isoquinoline-5-ylidene]ethenone) showed weak but significant antiviral activity. Although less potent than antiherpetic agents, such as acyclovir, it acted at the viral inactivation stage in a dose-dependent manner, suggesting a novel mode of action. These results highlight the feasibility of in silico approaches for identifying new antiviral compounds, which may be further optimized by medicinal chemistry approaches.


Introduction
Herpes simplex virus type 1 (HSV-1) is a contagious human pathogen that is estimated to affect 3.7 billion people worldwide [1]. Whereas HSV-1 infections are commonly associated with limited facial-oral lesions, severe disease can occur in neonates or immunocompromised individuals (keratitis, meningitis, encephalitis, and disseminated infections) [2,3]. Viral infections are the most common cause of sporadic life-threatening encephalitis in the United States, and up to 75% of these cases are caused by HSV-1 [3,4].
Primary infection with HSV-1 occurs by virus penetration at mucosal surfaces or through skin abrasions. From here, HSV-1 infects innervating sensory nerves of the trigeminal or dorsal root ganglia, where it establishes a life-long latent infection with a high rate of periodical reactivation [2,[5][6][7]. The current predominant antiherpetic agents are viral deoxyribonucleic acid (DNA) extension inhibitors, such as acyclovir (ACV), penciclovir, valacyclovir, famciclovir, foscarnet, and cidofovir. However, these drugs only act to shorten the duration of initial and recurrent episodes to a limited degree [8][9][10]. Consequently, HSV-1 infection is usually described as an incurable disease [1]. Drug-resistant HSV-1 strains

Virtual Screening in Search of Potential gD Inhibitors
The interaction between gD and its two major cellular receptors, nectin-1 and HVEM, can be described as non-reciprocal competitive binding. Glycoprotein D forms a hairpin structure after interacting with HVEM, which masks the binding site for nectin-1 [16,40]. Similarly, the interaction between gD and nectin-1 blocks the HVEM binding sites [16]. As distinct residues interact with each receptor, we performed targeted molecular docking on each site separately. Docking against the HVEM binding site of gD was conducted first, before a refined set of compounds was docked against the nectin-1 binding site. A schematic diagram of the virtual screening steps and outcomes is presented in Figure 1. AutoDock Vina software was used to estimate the theoretical affinity between the HVEM binding site of gD and a library of 527,209 natural compounds. The generated docking score is based on the change in Gibbs free energy between bound and unbound states; therefore, a more negative score indicates higher putative affinity [41]. A secondary round of screening, which involved performing a greater number of docking runs, was conducted on the top 2% of compounds (n = 10,897). These compounds included 6654 molecules from the Supernatural II database, 3813 from ZINC Natural Products, 210 from Human Metabolome, 206 from Marine Natural Products, and 14 molecules from Phenol Explorer. The docking scores from this screen ranged from −10.6 to −4.3.
To select ACs for downstream in vitro testing, ligands with docking scores ≤−8.2 were filtered for favorable ADMET properties and commercial availability (n = 3898). The selection criteria were based on toxicity ratings and drug-conforming behavior according to the OSIRIS property explorer tool [42]. Consequently, 26 drug-like compounds that could be readily obtained from commercial vendors were selected for further molecular docking analysis against the nectin-1 binding site.
All eight compounds were subjected to in vitro validation.

Investigating the Mechanism of Action of ACs by Time of Addition Assay
Next, the eight ACs were assessed in cell-based assays to test for antiviral activity against HSV-1. Three separate assays were performed to elucidate the mechanism of action ( Figure 4A,C,E). A viral inactivation assay was performed by incubating HSV-1 with ACs at 10 µg/mL or 1 µg/mL (depending on cytotoxicity) before addition to the cells. Viral attachment and entry-inhibition assay involved the simultaneous addition of virus and ACs to cells ( Figure 4C), whereas post-entry effects of ACs were investigated by adding ACs 1 h post-HSV-1 infection ( Figure 4E). The initial screening was performed using a low MOI of 0.01 (Supplementary Materials Figure S1). Three promising ACs, #7, #27, and #28, were selected based on cytotoxicity (Table 3) and preliminary results ( Figure S1) and further tested at MOI of 5, which triggered a cytopathic effect in 90% of cultured cells. The inhibitory effect of each compound was measured by cell survival as determined by using water-soluble tetrazolium 1 (WST-1) reagent ( Figure 4). Two out of eight tested compounds demonstrated promising antiviral activity at different stages of infection. AC#7 exhibited significantly higher activity than the DMSO control at the viral inactivation stage, as shown in Figure 4B. AC#27 was also found to possess antiherpetic activity; however, this effect was seen at the post-entry stage ( Figure 4F), indicating inhibition at later stages of the viral infection. The antiviral activity of AC#7 was found to be most effective when added to the virus prior to addition to the cells. No antiviral activity was observed when cells were preincubated with AC#7 (at 10 µg/mL) prior to HSV-1 infection ( Figure S2). Based on the obtained data, the activity exhibited by AC#7 reflects the hypothesized mechanism of action predicted by the molecular docking study. Therefore, AC#7 was selected for further testing using a dose-response assay.

Estimating the Efficacy of ACs by Plaque Reduction Assay
To visualize and evaluate the activity of AC#7, a plaque reduction assay was performed at concentrations from 400 µM (133.33 µg/mL) to 6.25 µM (2 µg/mL) using two-fold serial dilutions. Cells treated with DMSO only at concentrations ranging from 1.33% (v/v) to 0.02% (v/v) were used as a solvent control, and the results are plotted in Figure 5. AC#7 significantly reduced the number of virus-induced plaques compared to the DMSO control at a concentration of 25 µM ( Figure 5A). This is consistent with the results obtained from the CPE assay, in which the AC was tested at a concentration of 30 µM (10 µg/mL) ( Figure 4B). Notably, AC#7 demonstrated a dose-dependent antiviral response when pre-incubated with the virus compared to the DMSO control ( Figure 5B). The EC 50 value of AC#7 was 183 µM (61 µg/mL), as calculated based on the asymmetric nonlinear regression model (R 2 = 0.84). To determine the therapeutic index (TI), the cytotoxicity of AC#7 was further tested at a concentration of 800 µM. Cell viability was measured 46 h after the 2 h of exposure to AC#7. Negligible toxicity to VERO cells was found, as illustrated in Figure 6. Thus, the TI of AC#7 is estimated to be >4.37, which suggests this molecule could serve as a potential lead compound for future antiviral drug development [43,44].
For comparison, the standard antiherpetic drug ACV was tested in parallel to AC#7. ACV was added post-HSV infection due to its inhibitory mechanism at the viral DNA replication stage ( Figure S3). The EC 50 of ACV was 0.77 µM, determined using the same asymmetric nonlinear regression model (R 2 > 0.99). This is consistent with previously published in vitro results obtained with VERO cells in which the calculated EC 50   The cytotoxicity of AC#7 was tested by incubating Vero cells with AC#7 (800 µM) for 2 h. AC#7 was then removed, and cells were incubated in fresh media for a further 46 h before cell viability was evaluated using water-soluble tetrazolium 1. Cells without any treatment were included as cell-only control and used to calculate cell viability. Cells treated with dimethyl sulfoxide (DMSO) (2.7% v/v), corresponding to the solvent concentration in the tested samples, were used as the solvent control. The wells treated with 1% (v/v) Triton X-100 were used as a negative control. The cytotoxicity assay was performed three times with duplicates. Means of the three experiments are plotted with standard error of the mean. Wilcoxon test was applied to compare the cytotoxicity of DMSO control and AC#7 (ns indicates not significant, p > 0.05).

Discussion
The basis of this study lies in the well-understood interactions between HSV-1 envelope glycoprotein D and its cellular receptors [18,19]. The wild-type strains of HSV-1 exploit either nectin-1 or HVEM as entry receptors to infect host cells, depending on the cell type [47,48]. For key target cells, neurons and human keratinocytes, nectin-1 has been reported as the primary receptor and is therefore the most important. Alternatively, HVEM functions as the main receptor in nectin-1-deficient cell lines, with similar infectivity [47][48][49]. Thus, it may be important for antiviral compounds to inhibit both nectin-1 and HVEM binding sites on gD to be considered potential therapeutic agents. The interaction between HVEM and gD forms an N-terminal hairpin structure that masks the nectin-1 binding site, whereas the binding between nectin-1 and gD blocks the accessibility of the HVEM [16]. Therefore, molecular docking must be performed on multiple regions of the glycoprotein, a requirement that entails significant computational resources to screen the large compound library of over 500,000 molecules used in this study. To overcome this obstacle, an innovative docking strategy was applied, resulting in the selection of eight compounds with predicted affinity to both binding sites in silico.
Among the eight chosen molecules, AC#7, [4,5-g] isoquinolinelin-5-ylidene]ethenone] was identified as a weak HSV-1 inhibitor. The Gibbs free energy values generated by the docking software for AC#7 on both sites were −8.2 kcal/M and −8.6 kcal/M for the HVEM and nectin-1 sites, respectively. However, affinity with gD is not necessarily correlated with the potency of antiviral activity. A number of aptamers that have been reported to demonstrate strong affinity with gD at nanomolar concentrations using surface plasmon resonance were found to possess insignificant antiviral activity in vitro [50]. Therefore, some highly ranked hits, such as AC#28 and #29, may still bind gD, although they failed to show antiherpetic activity in the in vitro assays.
Two weak antiviral ligands were identified in this study; however, only AC#7 demonstrated activity at the early stage of infection, aligning with the in silico results. AC#7 is predicted to interact with nine residues in the HVEM binding interface via hydrophobic contacts and nine residues in the nectin-1 binding interface via hydrophobic contacts and two hydrogen bonds. Notably, the interaction at the nectin-1 interface involves amino acid residue Y38, which has been identified as critical for viral entry via the nectin-1 receptor [51][52][53]. Additionally, there are no overlapping residues between the binding interfaces, suggesting that the compound can bind to both sites simultaneously. However, although AC#7 exhibited low antiviral activity in the early stage of infection, further tests are required to confirm the interaction between this compound and glycoprotein D. Techniques such as surface plasmon resonance could provide deeper insights into the mechanism of action at a molecular level and could guide chemical modifications to improve the antiviral efficacy of this compound.
The compound 9-((2-Hydroxyethoxy) methyl) guanine, commonly known as ACV, is the gold standard in the treatment of HSV-1 infection, with an EC 50 of around 1 µM in vitro [54]. ACV is a typical nucleoside analogue that acts after viral entry, requiring phosphorylation by virally encoded thymidine kinase to prevent viral DNA elongation, resulting in a shorter duration of infection and relief from symptoms [10,55]. Other antiherpetic drugs, including nucleosides, nucleotides, and pyrophosphate analogues, all target viral replication and hence fail to eliminate the lifelong latent infection caused by HSV-1 [10]. Docosanol is the only approved herpesvirus entry inhibitor and directly interferes with the host-cell surface phospholipids [56]. However, drugs targeting host cells may lead to higher toxicity and lower selectivity compared to compounds that interact specifically with viral components [10,57]. Compared to these drugs, AC#7 demonstrated a distinct mechanism, reducing HSV-1 infection at an early stage. This unique activity of AC#7 can be used to exploit this molecule as the basic structure for future drug modification.
Although AC#7 acts at the early stage of infection with an acceptable selectivity (therapeutical index >4) [44], its low antiviral activity remains the main drawback for its consideration as a potential antiviral agent. This may be explained by several reasons. Firstly, it has proven challenging for small molecular compounds to inhibit high-affinity interactions between proteins with contributions from either continuous or discontinuous sites in their respective protein structures [58][59][60]. Another explanation is the potential interaction between these ligands and other viral or cellular components. Given that molecular docking does not account for the complex environment of in vitro conditions, unknown off-target effects are a limitation of virtual screening [61].
Discovery of structures with low activity can lead to the identification of potent compounds via medicinal chemistry approaches. As an example, the HSV-1 ribonucleotide reductase inhibitor BILD 1633 SE demonstrated higher antiviral activity than ACV in vitro and exhibited a potent therapeutic effect against ACV-resistant HSV-1 infections in vivo [62]. This antiherpetic agent was designed via extensive drug modification and optimization of a series of peptides with initial EC 50 values between 30 and 780 µM in vitro [63][64][65]. The drug improvement process of BILD1633 SE demonstrates the feasibility of improving the potency of molecules via medicinal chemistry approaches [62][63][64][65]. More recently, a chemical-modification design strategy targeting the cysteine-3-like protease (3CL pro ) enzyme of severe acute respiratory syndrome coronavirus (SARS-CoV), has given rise to the novel antiviral PF-00835231, the main component of Paxlovid [66]. A previously identified inhibitor of human rhinovirus (rupintrivir) was used as the basis of the study; however, the drug and other modified versions produced weak to undetectable inhibition of SARS-CoV 3CL pro . Consequently, structural binding data from co-crystallography experiments were used to guide chemical modifications, resulting in the production of PF-00835231. Although initially intended for SARS-CoV, this compound demonstrates promising activity against SARS-CoV-2 and is currently being studied in clinical trials [67]. Based on these examples, a similar pathway could be applied to AC#7 to engineer superior antiherpetic compounds.
Natural compounds are an important resource for modern drug development. More than 100 small molecules with diverse antiherpetic activities have been identified from a wide range of organisms [8,43,68,69]. AC#7 was originally derived from glycyrrhiza glabra, which has been found to have versatile bioactivity, including antiviral activity [70][71][72][73]. This molecule contains benzofuran and isoquinoline elements (Figure 2), which have both been found to possess antiviral abilities [74][75][76]. To the best of our knowledge, this is the first report of antiherpetic activity of a molecule with both benzofuran and isoquinoline elements in vitro, suggesting a promising foundation for designing a new generation of antiviral drugs via medicinal chemistry.
The fact that AC#7 reduced HSV-1 at the early stage of infection in a dose-dependent manner shows that this compound likely acts on the viral surface glycoprotein, gD, as predicted by the molecular docking analysis. Identification of a new natural molecule with an antiviral mode of action different from that of ACV provides a fresh basis for subsequent drug development and optimization.

Preparation of Natural Compound Library
A total of 527,209 compounds from five natural compound libraries (SuperNatural II [77], Phenol Explorer [78], Human Metabolome Database [79], Marine Natural Products [80], and ZINC Natural Products [81]) were downloaded in SDF format from the Miguel Hernandez University Molecular Docking site [82]. To prepare ligands for docking, each compound was edited with PyMOL to include polar hydrogens [83]. For the initial round of virtual screening, 3D coordinates were generated by converting files to MOL2 format with Marvin Suite 6.0 from ChemAxon [84]. Ligands were then energy-minimized using the universal force field (UFF) and converted to PDBQT format with Open Babel software v 3.1.1 [85]. For the second round of screening, ligands were energy-minimized and converted from SDF to PDBQT format using the PyRx virtual screening tool [86].

Preparation of Receptor Proteins
Glycoprotein D, the key surface protein of HSV-1 involved in viral attachment to host cells, was selected for molecular docking. The X-ray-derived crystal structures for gD complexed with HVEM (Protein Data Bank (PDB) ID: 1JMA) and in unliganded form (PDB ID: 2C36) were downloaded from the RCSB protein databank [18]. To prepare receptors for screening, files were edited using PyMOL to add polar hydrogens and to remove water molecules, ions, and ligands [83]. PDB files were converted to PDBQT format using the PyRx virtual screening tool [86].

Virtual Screening on the HVEM Binding Site of gD
Virtual screening was performed using AutoDock Vina molecular docking software [41]. A targeted docking approach was employed by defining a search space containing residues involved in the interaction with the native HVEM receptor [18]. The crystal structure of gD bound to HVEM (1JMA) was used as the docking model, and the search space was defined as follows: center: coordinates (x,y,z): −27.284, 50.152, −6.099; dimensions (x,y,z): 15.364, 23.087, 22.147. A flexible docking procedure was adopted to account for changes in the receptor conformation. The gD residues defined as flexible were: C15, R24, V25, E27, A28, and C29.
To identify optimal docking conformations whilst minimizing the number of computational resources required, the protocol was conducted in two stages. The initial round of screening was performed with all 527,209 ligands. For each ligand, the number of docked conformations (modes) was set to 10, and the number of runs (exhaustiveness) was set to 20. A secondary, more thorough screen was then conducted with the top 10,000 ligands, with the number of runs increased to 80. As some ligands produced identical docking scores, the number of compounds screened in this round was 10,897.

ADMET Analysis
To filter for compounds with favorable absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties, ligands ranked within the top 25 scores from the secondary screen were assessed (n = 3893). The OSIRIS property explorer was used to compare the properties of each ligand to currently traded drugs and predict their toxicity based on data from the Registry of Toxic Effects of Chemical Substances (RTECS) database [42]. Ligands were filtered for the absence of toxic fragments (mutagenic, tumorigenic, irritant, or reproductive effects), molecular weight (MW) < 500 Da, hydrophilicity (cLogP) < 5, solubility (logS) >−6, topological surface area (TPSA) < 120Å 2 , drug likeness > 0, and overall drug score >0.5.

Redocking Selected Compounds on the Nectin-1 Binding Site of gD
As nectin-1 is a primary receptor of gD and occupies a binding site distinct from that of HVEM, the ligands ranked with the top 25 scores, filtered for favorable ADMET profiles and commercial availability, were redocked against the nectin-1 binding site of gD (n = 26). For this docking experiment, an unliganded X-ray-derived crystal structure of gD was used due to its high resolution of 2.11Å (PDB ID: 2C36) [19]. The search space was selected based on a region of interacting residues at the nectin-1 binding interface known as surface patch 2 [87]. The following residues were defined as flexible: V37, Y38, H39, Q132, V214, D215, I217, M219, L220, R222, and F223. The search space was defined as follows: center coordinates (x,y,z): 60.637, 42.391, 98.991; dimensions (x,y,z): 20.428, 24.261, 15.139. As with the virtual screen against the HVEM binding site, the number of docked conformations was set to 10, and the number of runs was set to 80.
The docking results were visualized using PyMOL. Residues involved in hydrogenbonding and hydrophobic interactions were plotted using LigPlot + software [88].

Chemicals
A total of 8 synthetic purified ACs with docking scores ≤−8.2 for the HVEM site and ≤−7.6 for the nectin-1 site and favorable ADMET properties were selected for in vitro screening. All ACs were purchased from VITAS-M laboratory, USA, as listed in Table 1. Their molecular structures are illustrated by ChemDraw 18.2 (PerkinElmer). Each compound was dissolved in molecular-grade dimethyl sulfoxide (DMSO) (Sigma, Melbourne, Australia) to a final concentration of 10 mg/mL and kept at −20 • C until use.

Cells and Virus
HSV-1 strain F was used in all antiviral assays in this study. African green monkey epithelial kidney (VERO) cells were cultured in Dulbecco's modified Eagle medium (DMEM) (Lonza, Thermo Fisher Scientific, Walkersville, MD, USA) supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin streptomycin (Gibco, Thermo Fisher Scientific, Scoresby, VIC, Australia) at 37 • C under 5% CO 2 . HSV-1 was propagated in VERO cells and titrated by plaque assay as described in [8]. Viral stocks were aliquoted and kept at −80 • C until use.

Cytotoxicity of ACs
The cytotoxicity of each AC was determined using WST-1 reagent (Sigma, Australia) as described previously [43]. VERO cells were seeded in 96-well plates at a density of 10,000 cells per well for 24 h at 37 • C under 5% CO 2 . Then, the ACs were 10-fold serially diluted in DMEM supplemented with 2% (v/v) FBS and 1% (v/v) penicillin streptomycin to concentrations of 100 µg/mL, 10 µg/mL, 1 µg/mL, and 0.1 µg/mL and incubated with the cells for 24 h at 37 • C. After incubation, cells were washed with DMEM with 2% FBS and 1% (v/v) penicillin streptomycin once and incubated with 10% (v/v) WST-1 for 2 h at 37 • C under 5% CO 2 . To determine cell viability, plates were read by a SpectraMAX iD3 plate reader (Molecular Devices) at 450 nm. The absorbance at 620 nm was used as a reference.
Cells treated with DMSO only were included as solvent controls, and untreated cells were included as cell-only control. Wells containing DMEM without cells were prepared as blanks. The cytotoxicity of each AC was calculated by comparing treated cells with the DMSO control Equation (1). The experiment was performed in quadruplicate.

Time of Addition Assay
A cytopathic effect (CPE) inhibition assay was applied to screen for the antiviral activity of selected ACs at different time points. VERO cells were seeded in 96-well plates at a density of 10,000 cells per well for 24 h before the addition of virus at 37 • C under 5% CO 2 . The initial screening was conducted at a multiplicity of infection (MOI) = 0.01; then, the active compounds were tested with a higher MO1 = 5 to confirm their activity. ACs were tested at 10 µg/mL or 1 µg/mL depending on their cytotoxicity.
To identify compounds with antiviral activity, virus and ACs were added at different time points: viral inactivation (preincubation of the virus with ACs for 1 h prior to addition to cells at 37 • C under 5% CO 2 ), attachment and entry (ACs and virus were added to cells simultaneously), and post-entry stages (ACs were added 1 h after HSV-1 infection of cells and were kept during the entire 48 h incubation at 37 • C under 5% CO 2 ). After infection, the cells were incubated for another 48 h in DMEM with 2% FBS before addition of WST-1 to quantify cell viability. Three controls were included: Solvent controls were prepared by diluting DMSO at concentration of 0.1% (v/v) in DMEM without ACs and added to the cells at different time points corresponding to the samples. Cells cultured with DMEM only were prepared as the cell control, and cells infected with virus without DMSO were included as the virus control. The initial antiviral activity of each AC was calculated using Equation (2) as described in [89], with modifications. The initial screening (with MOI = 0.01) was performed once in quadruplicate, and the secondary experiments (with MOI = 5) were conducted four times in triplicate. The ACs that demonstrated CPE inhibition were further analyzed by plaque reduction assay.

Plaque Reduction Dose-Response Assay
The antiviral activity of selected ACs on VERO cells was quantitatively determined by plaque reduction assay as described in [43], with modifications. VERO cells were seeded in 24-well plates (50,000 cells/well) for 24 h. The virus was diluted to 50 plaque-forming units (PFU) per 0.2 mL and preincubated with the ACs at different concentrations (400, 200, 50, 25, 12.5, and 6.25 µM) or DMSO controls (1.33%, 0.67%, 0.33%, 0.16%, 0.08%, 0.04%, and 0.02%, v/v) for 1 h before being added to the cell monolayer. Then, the cells were incubated with the mixture for 1 h at 37 • C under 5% CO 2 . After removing the inoculum, the cells were rinsed with PBS and overlayed with DMEM containing 1.6% carboxymethyl cellulose (CMC) and 1% FBS. After incubation for 48 h, the cells were fixed with methanol and stained with 0.5% crystal violet (Sigma, Australia). The number of plaques was counted manually using upright microscopy (Olympus, Tokyo Japan). Cells infected with virus without DMSO were included as the virus control. The antiviral activity was calculate using Equation (3) [90], and the EC 50 values were determined by linear regression analysis of the dose-response curve.

Toxicity of AC#7 on VERO Cells at High Concentrations
Since AC#7 did not exhibit cytotoxicity to VERO cells in the plaque reduction assay at a concentration of 400 µM, its toxicity was further tested at 800 µM in 96 well-plates. Due to stock limitation, 800 µM was the highest concentration that could be tested. Briefly, VERO cells were seeded in 96-well plates at a density of 10,000 cells per well for 24 h at 37 • C under 5% CO 2 . Then, the cells were exposed to AC#7 at a concentration of 800 µM or controls for 2 h at 37 • C under 5% CO 2 . AC#7 or controls were then removed and replaced with fresh DMEM with 2% FBS and 1% penicillin streptomycin. Cells were incubated for 46 h at 37 • C under 5% CO 2 . After incubation, cells were washed with DMEM once and incubated with 10% (v/v) WST-1 for 2 h at 37 • C under 5% CO 2 . Cells treated with DMSO at 2.7% (v/v), corresponding to the solvent used to dissolve 800 µM AC#7 in DMEM, were included as solvent control. Cells treated with Triton X-100 at 1% (v/v) (Sigma, Australia) were included as negative control, and untreated cells were included as cell-only controls. Then, the cell viability of AC#7 was calculated using Equation (1). Wells containing DMEM without cells were prepared as blanks. The experiment was performed three times with duplicates.

Statistical Analysis
All statistical analyses were performed using GraphPad Prism 9 software (GraphPad Software, Inc., San Diego, CA, USA). Linear regression curves were obtained using the five-parameter logistic sigmoidal model offered by the same software. Significance testing between multiple samples and controls was performed using Friedman test followed by a post-hoc Dunn's multiple comparisons test. Wilcoxon matched test or t-tests were applied to compare the mean between two samples.

Conclusions
This study represents the first application of a large-scale structure-based virtual screen targeting the binding interfaces of HSV-1 glycoprotein D to identify inhibitors of early viral infection. Two out of eight selected compounds exhibited significant activity against HSV-1 infection in vitro. Notably, compound AC#7 demonstrated a mechanism of action consistent with inhibiting early-stage infection. Although the potency of AC#7 is lower than that of current first-line drugs, these data suggests that it could serve as a structural basis for further drug development. More importantly, this work supports the use of molecular docking as a powerful tool for identifying novel antiviral compounds. Future studies will expand on these computational methods with an aim to screen additional compound libraries, as well as other viral protein targets.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/ph15030361/s1, Figure S1: Preliminary screening of antiviral activity of eight select ACs; Figure S2: Investigating the antiviral activity of AC#7 when added to cells before HSV-1 infection; Figure S3: Inhibition of antiviral activity by ACV in the post-entry stage.