The A42R protein retrieved from the Research Collaboratory for Structural Bioinformatics Protein Data Bank (RCSB PDB) had two chains, A and B. Each chain is composed of seven antiparallel β-sheets, three α-helices and a partial helix (Minasov et al., 2022). Herein, the A42R structure was subjected to energy minimization using all-atom optimized potentials for liquid simulations (OPLS/AA) and Chemistry at Harvard Macromolecular Mechanics 36 (CHARMM36) force fields. The A42R protein structure was minimized using the OPLS/AA force field that had a lower potential energy of –3.896 × 10⁵ kJ/mol in 667 steps than that of CHARMM36, which converged in 374 steps with an energy of –3.709 × 10⁵ kJ/mol ( Supplementary Figure S1 ). Thus, the A42R structure which was energy minimized using OPLS was used in this study due to its lower energy which implies a higher stability (Pallio et al., 2023).

Computed Atlas of Surface Topology of proteins (CASTp) 3.0’s (Tian et al., 2018) prediction of binding pockets for A42R resulted in pocket 1 as the largest with an area of 115.519 Ų and a total volume of 31.983 ų ( Table 1 ). Upon visualizing the other pockets in PyMOL, it was observed that the other pockets were relatively small and could not accommodate ligands. Residues lining pocket 1 included Met1, Glu3, Trp4, Lys6, Ile7, Asp10, Ile22, Thr99, Ile104, His124, Ala125, Arg127, Val128, Thr131, and Asn133 ( Table 1 ). Of the twenty predicted binding pockets, pocket 1 stood out as the largest cavity and was the only binding site of A42R assessed in this study.

AutoDock Vina module in PyRx version 0.9.2 successfully screened 26,315 compounds (25,196, 821, and 298 from TCM, AfroDb, and PubChem, respectively) against MPXV A42R profilin-like protein (Trott and Olson, 2010; Dallakyan and Olson, 2015). Docking conformations were visualized for compounds with the lowest docking scores (highest binding affinities). Compounds were checked for binding to pocket 1, the most plausible binding site. TCM compounds ZINC000070455208 and ZINC000085543530, both with the lowest docking score at –9.0 kcal/mol, and ZINC000043552595 at –8.9 kcal/mol were eliminated as they did not bind to pocket 1. Similarly, ZINC000095485942 from AfroDb with a docking score of –8.5 kcal/mol was eliminated. After eliminating compounds not bound to pocket 1, the pose with the most negative docking score was selected as the best for each ligand. A previous docking study screening Plantago lanceolate compounds against A42R resulted in comparable binding energies ranging from –5.3 to –9.9 kcal/mol (Bajrai et al., 2022). The top 1% from TCM was shortlisted comprising 252 compounds with docking scores of –7.7 kcal/mol or less. All compounds passing below the –7.0 kcal/mol threshold were retained from AfroDb and PubChem leaving 44 and 3 respectively. Other molecular docking studies screening for compounds against A42R resulted in binding energies greater than –6.8 kcal/mol (Burkhanova et al., 2022; Preet et al., 2022). The lowest docking score from TCM was observed for ZINC000043552595 with –8.8 kcal/mol, from AfroDb was ZINC000095486204 with –8.3 kcal/mol, and from PubChem CID: 11371962 with –7.2 kcal/mol. These are considered good especially in comparison to tecovirimat, a known MPXV inhibitor whose binding energy was above the –7.0 kcal/mol threshold at –6.7 kcal/mol. It is worth noting that tecovirimat has not been shown to target the A42R protein. Other top compounds included ZINC000000899909, ZINC000001632866, ZINC000015151344, ZINC000013378519, and ZINC000000086470. Interactions such as hydrophobic and hydrogen bonds are important for the stability of the ligand binding to the A42R interface ( Table 2 ). Optimization of both hydrophobic and hydrogen bonds can be used to improve drug selectivity to reduce off-target adverse effects and improve efficacy (Patil et al., 2010).

The protein-ligand interaction maps for the seven potential candidates and tecovirimat are presented in ( Figure 1 ; Supplementary Figure S2 ). Tecovirimat (PubChem CID: 16124688) interacted with A42R residues Glu3, Trp4, Ile7, Arg127, Val128, Thr131, and Asn133 via hydrophobic bonds and formed two interactions with Met1 via hydrogen bonding with lengths 3.19 and 3.33 Å ( Supplementary Figure S2F ). PubChem CID: 11371962 interacted with A42R residues Glu3, Trp4, Ile7, Arg127, Val128, Thr131, and Asn133 via hydrophobic bonds and one 3.08 Å hydrogen bond with Met1 ( Supplementary Figure S2A ). ZINC000000899909 formed hydrophobic bonds with residues Glu3, Trp4, Ile7, Arg127, Val128, Thr131, and Asn133 and a hydrogen bond with Met1 (3.19 Å) ( Figure 1A ). ZINC000001632866 formed hydrophobic bonds with residues Met1, Glu3, Trp4, Ile7, Arg127, Val128, Thr131, and Asn133 ( Supplementary Figure S2B ). ZINC000015151344 formed hydrophobic bonds with residues Glu3, Trp4, Arg127, Val128, and Thr131 and two hydrogen bonds with residues Met1 and Asn133 with bond lengths 3.16 and 3.29 Å, respectively ( Figure 1B ). ZINC000013378519 formed hydrophobic bonds with residues Met1, Glu3, Trp4, Lys6, Ile7, Asp10, Arg127, Val128, Thr131, and Asn133 ( Supplementary Figure S2C ). ZINC000000086470 formed hydrophobic bonds with residues Glu3, Trp4, Ile7, Arg127, Val128, Thr131, and Asn133, as well as two hydrogen bonds with Met1 of bond lengths 3.16 and 2.79 Å ( Supplementary Figure S2D ). ZINC000095486204 formed hydrophobic bonds with residues Glu3, Trp4, Arg127, Val128, Thr131, and Asn133 and a hydrogen bond with Met1 of bond length 3.17 Å ( Supplementary Figure S2E ). In all 8 compounds, the residues Met1, Glu3, Trp4, Arg127, Val128, Thr131, and Asn133 were involved in protein-ligand interactions. Met1 was involved in at least one hydrogen bond for 6 of the 8 compounds and involved in 2 hydrogen bonds for tecovirimat and ZINC000000086470. Ile7 was also a prevalent interaction residue involved in 6 of the 8 protein-ligand interactions. Other studies that looked at interacting residues between A42R and ligands have also mentioned Trp4 and Arg127 involvement (Minasov et al., 2022; Dassanayake et al., 2022).

There were a total of 111 compounds that failed ADME and were eliminated from consideration. These included compounds that violated more than one of Lipinski’s rules or any violations to Veber’s rule. The top compound ZINC000043552595 violated Veber’s rule and was eliminated ( Table 3 ). Twenty other compounds with low docking scores from TCM failed ADME leaving the top compound from TCM to be ZINC000013378519 with a docking score of –8.1 kcal/mol. Top compounds from AfroDb and PubChem remained as ZINC000095486204 at –8.3 kcal/mol, and PubChem CID: 11371962 at –7.2 kcal/mol. A total of 142 out of 252 passed, 43 out of 44 passed, and 3 of 3 passed for TCM, AfroDb, and PubChem, respectively. The shortlisted compounds had a molecular weight between 242.31 g/mol and 482.52 g/mol and TPSA’s ranging up to 88.38 Ų ( Table 3 ).

BBB permeability should be considered in the adaptation of these compounds as potential MPXV inhibitors. There has been a rising concern about neurological complications associated with MPXV (Sepehrinezhad et al., 2023; Pastula et al., 2022; Billioux et al., 2022). Known neurological symptoms of MPXV have commonly included headache, neuropathic pain, depression, and anxiety (Billioux et al., 2022). A rarer symptom associated with MPXV is encephalitis, but it may be linked to a relatively common symptom, conjunctivitis, that occurs in about 30% of unvaccinated patients (Urmi et al., 2023). In the 2022 outbreak, in an examination of two MPXV patients suffering from encephalitis MPXV DNA was detected in the cerebrospinal fluid (Billioux et al., 2022). It has been reported in MPXV and in other viruses that some ocular symptoms like conjunctivitis may play a role in viral infiltration to the brain resulting in encephalitis (Urmi et al., 2023; Koyuncu et al., 2013; Yue et al., 2022). To address the concern of MPXV populating the brain, the permeability of the compounds to cross the blood brain barrier (BBB) was predicted. From the top compounds PubChem CID: 11371962, ZINC000000899909, ZINC000015151344, ZINC000000086470, and ZINC000095486204 were predicted to be permeable to the BBB ( Table 3 ). Excluding the eliminated ZINC000043552595, compounds ZINC000001632866 and ZINC000013378519 were predicted to not cross the BBB, but alternative administration routes could be employed to bypass the BBB (Hersh et al., 2016; Gernert and Feja, 2020; Broni et al., 2023).

Of the twenty-five shortlisted compounds that had high binding affinities to pocket 1 and passed ADME, there was a total of 18 that passed toxicity screening ( Supplementary Table S1 ). There were 21, 19, 14, and 23 compounds predicted to have no toxic effects regarding mutagenicity, tumorigenicity, reproductive effects, or irritancy ( Supplementary Table S1 ) after subjecting them to toxicity risk prediction using DataWarrior 5.5.0. Ten compounds including ZINC000013378519, ZINC000015151344, ZINC000095909830, ZINC000095913878, ZINC000000689683, ZINC000000897930, ZINC000000134782, ZINC000048998695, ZINC000014557836, and ZINC000038658035 had no predicted toxicity in any of the four categories ( Supplementary Table S1 ). Only compound ZINC000095485910 of the top twenty-five had high mutagenic risk. Six compounds had high tumorigenic risk including compounds ZINC000095486204, ZINC000001632866, ZINC000028702248, ZINC000031852149, ZINC000095485910, and ZINC000095486327.

Top compounds ZINC000095486204 and ZINC000001632866 from AfroDb and TCM respectively failed toxicity screening. ZINC000095486204 had both high risk in mutagenicity and low reproductive effect risks while ZINC000001632866 had both low mutagenic and high tumorigenic risks. While this should eliminate them from further use their structures were of interest as they had low docking scores when screened against A42R and good predicted antiviral activity. ZINC000001632866’s binding to A42R was specifically of interest because it had predicted antiviral activity to poxviruses. So, while these drugs should be cautioned against because of their potential toxicity, their structures may be of value in designing new antipoxvirus drugs.

For each of the seven potential lead compounds PubChem CID: 11371962 (N-(3,5-dioxo-4-azatricyclo[5.2.2.02,6]undec-8-en-4-yl)-4-(trifluoromethyl)benzamide), ZINC000000899909 (Sojagol), ZINC000000086470 (Obovatin 5-Methyl Ether), ZINC000001632866 (3-Methylbenzo[c]phenanthrene), ZINC000095486204 ((1S,3R)-7-(4-hydroxy-5-methoxy-7-methylnaphthalen-1-yl)-1,3-dimethyl-3,4-dihydro-2H-isoquinolin-1-ol), ZINC000013378519 (1-[(7-Hydroxy-4-methoxy-9,10-dihydrophenanthren-2-yl)oxy]-4-methoxy-9,10-dihydrophenanthrene-2,7-diol), and ZINC000015151344 (4,18-Dihydroxy-2-oxatricyclo[13.3.1.13,7]icosa-1(18),3,5,7(20),15(19),16-hexaen-10-one) the probability of activity (Pa) obtained for each activity related to viral inhibition, or antiviral activity was greater than the corresponding probability of inactivity (Pi) using the Prediction of Activity Spectra of Substances (PASS) (Lagunin et al., 2000; Parasuraman, 2011). ZINC000000899909, ZINC000000086470, ZINC000001632866, ZINC000095486204, ZINC000013378519, and ZINC000015151344 were all predicted to inhibit viral entry ( Supplementary Table S2 ).

Two compounds that stood out were ZINC000001632866 and ZINC000015151344 which were predicted as antivirals for poxviruses with Pa values of 0.315 and 0.215, and Pi values of 0.052 and 0.136, respectively. These two compounds were also predicted to have antiviral activity to other double stranded DNA viruses in the Adenoviridae, Herpesviridae, and Hepadnaviridae families ( Supplementary Tables S2B, C ). Predicted activity to inhibit adenoviruses had Pa values of 0.387 and 0.381 and Pi values 0.035 and 0.037 for ZINC000001632866 and ZINC000015151344, respectively. In total, each of these two compounds had predicted antiviral activity for ten different viruses, nine of which they had in common including poxviruses, picornavirus, adenovirus, cytomegalovirus (CMV), influenza, herpes, hepatitis C (HCV), rhinovirus, and HIV ( Supplementary Tables S2B, C ). Additionally, ZINC000001632866 was also predicted to have antiviral activity against parainfluenza, and ZINC000015151344 had predicted antiviral activity against hepatitis B ( Supplementary Tables S2B, C ).

MPXV infections can be complicated by other comorbidities. ZINC000000899909, ZINC000001632866, ZINC000015151344, and ZINC000013378519 all had predicted activity to inhibit HIV in several ways by targeting HIV fusion, integration, or reverse transcription and since HIV is a common comorbidity of MPXV these compounds may be useful in further protection from HIV related exacerbations ( Supplementary Tables S2A-D ) (Hoffmann et al., 2022). Human MPXV patients have been reported with inflammation in the spleen and liver though there is a lack of evidence for large amounts of MPXV replication in hepatocytes (Lum et al., 2022). ZINC000000899909, ZINC000013378519, ZINC000000086470, and PubChem CID: 11371962 were all predicted to have antiviral activity against hepatitis either generally or specifically B or C ( Supplementary Tables S2A, D, E, G ). ZINC000000899909, ZINC000001632866, and ZINC000015151344 were predicted to inhibit hepatitis C virus (HCV) internal ribosome entry site important for translation initiation in HCV (Dibrov et al., 2014).

Other antiviral activity predicted included compounds ZINC000000899909 and ZINC000000086470 against rhinoviruses with Pa values of 0.383 and 0.568 and Pi values of 0.111 and 0.009, respectively ( Supplementary Tables S2A, E ). ZINC000000086470 and ZINC000013378519 had predicted antiviral activity against herpes and influenza ( Supplementary Tables S2D, E ). PubChem CID: 11371962 had predicted antiviral activity against human coronavirus with Pa of 0.247 and Pi of 0.080.

There were several other predicted activities that could lead to viral inhibition. All top seven compounds were predicted as RelA expression inhibitors, JAK2 expression inhibitors, and Pin1 inhibitors ( Supplementary Tables S2A-G ). The nuclear factor κB (NF- κB) pathway is important to viruses because this pathway can be activated by detection of viral particles leading to an immune response (Takada et al., 2002). RelA inhibitors block RelA, in NF- κB and Sp1 sufficiently to inhibit HIV-1 replication and decrease HIV-1 transcription (Takada et al., 2002). The highest predicted activity for RelA expression inhibition from the seven compounds was ZINC000000086470 with a Pa of 0.647 and Pi of 0.003, and second was ZINC000000899909 with a Pa of 0.623 and Pi of 0.003.

JAK2 is a substrate for Abl family tyrosine kinases and Abl family tyrosine kinases are a known target for anti-MPXV drugs (Rabaan et al., 2023). MPXV and vaccinia virus (VACV) use a conserved mechanism described previously to move from cell to cell. For both viruses, enveloped virions are important as it has been previously shown that the formation of actin tails necessary for motility require Abl and Src family kinases, though only Abl kinases are needed for release of the enveloped virions (Reeves et al., 2005; Reeves et al., 2011). There has been success with imatinib mesylate in blocking this pathway in mice models successfully inhibiting viral exit of VACV (Reeves et al., 2005). Abl family tyrosine kinase inhibitors have also been reported to inhibit viral replication by interrupting viral DNA synthesis (Rabaan et al., 2023; Reeves et al., 2005). Since JAK2 is known to be activated by poxviruses and it is involved in this pathway important for viral replication and motility, inhibition of JAK2 expression is a good target to inhibit poxviruses (Rabaan et al., 2023; Reeves et al., 2011; Ahmed et al., 2009). JAK2 expression inhibitors are also useful in controlling inflammation caused by response to viral infection. MPXV infection experiments in cynomolgus macaques have reported fatality associated with high numbers of cytokines termed a ‘cytokine storm’ (Lum et al., 2022; Goff et al., 2011). This is not uncommon as similar aberrant immune responses occur in SARS-CoV-2 in human patients and Influenza A in mice studies (Gajjela and Zhou, 2022; Wang et al., 2020). SARS-CoV-2 in human patients when treated with JAK2 inhibitors resulted in preventing severe respiratory side effects resulting from viral infection with minimal impact on the hosts immune system (Gajjela and Zhou, 2022). Influenza A manipulation of JAK2 is vital for viral replication. In a study using JAK2 inhibitor, gingerenone, they were able to limit severe respiratory effects and prolong survival of mice (Wang et al., 2020). There is evidence that poxviruses can activate tyrosine kinases like JAK2 and that inhibition of JAK2 can alter cytokine signals protecting mice subjects from lethal VACV infection (Ahmed et al., 2009). The compound with the highest predicted JAK2 expression inhibition was ZINC000013378519 with a Pa of 0.902 and a Pi of 0.003. Also, with high predicted activity were compounds ZINC000001632866, ZINC000015151344, and ZINC000095486204 with Pa values of 0.860, 0.796, and 0.620 and Pi values of 0.004, 0.008, and 0.029, respectively.

Pin1 is a peptidylprolyl isomerase involved in activating several oncogenes and turning off tumor suppressors which make it a target for viruses including SARS-CoV-2, HIV, and hepatitis B (Kanna et al., 2022; Zhou et al., 2021; Hou et al., 2015). In SARS-CoV-2 and other viruses Pin1 aids in viral growth (Kanna et al., 2022). Pin1 has been shown to promote HIV uncoating, reverse transcription, and viral integration to the host genome, and upon Pin1 inhibition these steps are also inhibited (Hou et al., 2015). Pin1 in hepatitis B viral infection associates with hepatitis B X protein (HBx), a critical protein for viral transcription and replication (Zhou et al., 2021). ZINC000001632866 and ZINC000015151344 had the two highest predicted activities as Pin1 expression inhibitors with Pa values of 0.661 and 0.636 and Pi values of 0.011 and 0.013, respectively.

Compounds ZINC000000899909, ZINC000015151344, and ZINC000095486204 were predicted as APOA1 expression enhancers with Pa values of 0.443, 0.420, and 0.362 and Pi values of 0.047, 0.059, and 0.104, respectively. APOA1 is the gene encoding apolipoprotein A-I, a major component of high-density lipoprotein (HDL) (Singh et al., 1999; Srinivas et al., 1990). HDL has broad antiviral activity inhibiting viral entry into cells (Singh et al., 1999). Apolipoprotein A-I has been shown to limit cell fusion in HIV infected cells, in recombinant vaccinia virus infected CD4+ HeLa cells expressing HIV envelope protein, and herpes simplex virus all of which during viral infection decrease HDL levels (Srinivas et al., 1990; Owens et al., 1990). In herpes simplex virus, Apolipoprotein A-I was able to inhibit cell fusion at 1 µM concentrations (Srinivas et al., 1990).

Other targets for suggested anti-MPXV drugs to block viral replication include DNA or RNA polymerase and topoisomerase inhibitors (Rabaan et al., 2023). All compounds were predicted to have DNA or RNA polymerase inhibition and compounds ZINC000000086470, ZINC000095486204, and PC11371962 were predicted to inhibit topoisomerase I while compound ZINC000013378519 was predicted to inhibit both topoisomerase I and II ( Supplementary Tables S2A-G ). The predicted antiviral activities of the seven potential lead compounds had Pa > Pi and are worthy of further experimental validation in vitro (Jamkhande and Barde, 2014).

PubChem CID: 11371962 was structurally similar to tecovirimat with a score of 0.962. Tecovirimat has shown inhibitory activity against MPXV in vitro with an IC₅₀ of 12.7 nM and in mice models with an EC₅₀ of 0.008 µM against Zaire Central African clade MPXV isolates and an EC₅₀ of 0.006 µM against MPXV isolates from the 2022 Canadian/West African clade (Frenois-Veyrat et al., 2022; Warner et al., 2022). Tecovirimat has limited cases of use in human treatment of MPXV but has shown efficacy in the treatment of MPXV (Mbrenga et al., 2022; Desai et al., 2022). Tecovirimat targets viral p37 and F13L phospholipase needed for enveloping the virus, in this study tecovirimat was used as a reference control, though it does not target A42R specifically (Sherwat et al., 2022; Desai et al., 2022; Mucker et al., 2013). PubChem CID: 11371962 had a higher predicted binding affinity for A42R than tecovirimat ( Table 2 ).

Compound ZINC000000899909 or sojagol, is a natural compound that is extracted from Glycine max or soybeans. Soybean metabolites include different flavonoids, isoflavonoids, and coumarins that have been suggested to play different roles in antimicrobial activities in plants (Silva et al., 2021). ZINC000000899909 was predicted to have structural similarity to (+)-rutamarin alcohol with a score of 0.716. (+)-Rutamarin alcohol’s direct parent is psoralens and is classed as a coumarin (Ulubelen and Öztürk, 2006). (+)-Rutamarin alcohol is reported to target topoisomerase II, that is critical for viral replication (Xu et al., 2014). ZINC000000899909 was reported in PASS to have inhibitory activity of both topoisomerase I and II with Pa’s of 0.272 and 0.147 and Pi’s of 0.018 and 0.035, respectively. (+)-Rutamarin alcohol has effectively inhibited herpesvirus replication in vitro with an IC₅₀ of 1.12 µM and herpes virion production with an EC₅₀ of 1.62 µM (Xu et al., 2014). It also inhibited Epstein Barr virus DNA replication at IC₅₀ 2.38 µM and virion production with an EC₅₀ of 2.94 µM (Hassan et al., 2022). Other coumarins, novobiocin and coumermycin inhibit viral topoisomerase 1B with Ki values of 10-25 µM and 350 µM, respectively (Sekiguchi et al., 1996). Vaccinia topoisomerase 1B has enough differences from human topoisomerase 1B to be selective to the viral version and is a suggested target in poxviruses (Sekiguchi et al., 1996; Sliva and Schnierle, 2007).

Compound ZINC000015151344 or 4,18-Dihydroxy-2-oxatricyclo[13.3.1.13,7]icosa-1(18),3,5,7(20),15(19),16-hexaen-10-one, is a natural compound that can be obtained from the tree Engelhardia roxburghiana. Other compounds extracted from the leaves of Engelhardia roxburghiana include flavonoids with effects as anti-inflammatories, anti-proliferatives, and antioxidants (Xin et al., 2012). ZINC000015151344 was predicted to have structural similarity to zingerone and 5-pentyl-2-phenoxyphenol with scores of 0.813 and 0.802 respectively. Zingerone is a compound from ginger with high antioxidant activity along with other important properties to ease complications associated with viral infection including anti-inflammatory, antimicrobial, anticancer, and antidiarrhoeic effects (Ahmad et al., 2015). Oxidative stress is associated with several viruses including vaccinia virus that causes redox imbalances in its hosts to promote viral replication (Aydemir and Ulusu, 2022). There is support of antioxidant compounds reducing lung inflammation after influenza A and B infection. Terameprocol is an antioxidant with both antiviral and anti-inflammatory effects that in vitro inhibited viral yield in both cowpox and vaccinia virus (Aydemir and Ulusu, 2022). 5-pentyl-2-phenoxyphenol is an antibacterial compound (Shawon et al., 2021). Chemically ZINC000015151344 is classified as a diarylether under the broader diarylheptanoids and 5-pentyl-2-phenoxyphenol’s direct parent is diphenyl ethers. Diphenyl ethers are of interest as new antiviral scaffolds (Kini et al., 2019). Diphenyl ether-based compounds have shown broad antiviral activity including efficacy against vaccinia virus in vitro with an EC₅₀ of 9 µM (Ibrahim et al., 2016).

Compound ZINC000000086470 or Obovatin 5-Methyl Ether, was predicted to have structural similarity to sakuranetin, naringenin, (2S)-7-hydroxyflavanone, 5-deoxyflavanone, hesperetin, and 4’-hydroxyflavanone with scores 0.747, 0.742, 0.74, 0.736, 0.73, and 0.709, respectively. ZINC000000086470 can be found naturally in several species including Tephrosia bracteolata, Lonchocarpus costaricensis, and Pongamia pinnata. Extracts from Tephrosia bracteolata have shown antidiabetic, antioxidant, and antimicrobial properties (Egharevba et al., 2019). Pongamia pinnata is a species of tree with a wide variety of medicinal applications including uses as an antiseptic, and for the treatment of ulcers, malaria, bronchitis, and many more (Al Muqarrabun et al., 2013). More than seven different flavonoids have been extracted from Lonchocarpus costaricensis. Flavonoids are compounds in the flavanone class that are suggested to have broad antiviral activities (Jannat et al., 2021). Flavanoid compounds can inhibit a multitude of viral targets affecting viral binding, entry, and replication (Jannat et al., 2021). Sakuranetin has shown inhibitory activity against Influenza B replication with an IC₅₀ of 7.21 µg/mL (Kwon et al., 2018). Hesperetin has antiviral activity against Sindbis neurovirulent strain with IC₅₀ of 20.5 µg/mL (Paredes et al., 2003).

Compound ZINC000013378519 or 1-[(7-Hydroxy-4-methoxy-9,10-dihydrophenanthren-2-yl)oxy]-4-methoxy-9,10-dihydrophenanthrene-2,7-diol, is a natural product that can be obtained from members of the medicinal orchids, Pholidota chinensis, or Bletilla striata. Both orchid sources have a variety of known medicinal applications. P. chinensis has been used for the treatment of chronic bronchitis, fevers, stomachaches, ulcers, and has shown some anti-inflammatory effects in response to bacterial infections (Martha and Gutierrez, 2010). B. striata has shown activity as an antimicrobial agent, antioxidant, an anticancer agent, an anti-inflammatory, and has been used in wound healing and as a hemostatic agent (Martha and Gutierrez, 2010). The predicted biological activities of the shortlisted compounds corroborate their potential antiviral activity against MPXV. These 7 compounds are attractive antiviral candidates for in vitro experimentation.

To describe the structural conformation changes and atomic motions, 100 ns MD simulations were carried out for the unbound A42R, and A42R complexes with ligands ZINC000000899909, ZINC000001632866, ZINC000015151344, ZINC000013378519, ZINC000000086470, ZINC000095486204, and PC11371962. The top seven compounds used for MD had the highest binding affinities for A42R while binding to pocket 1, passed ADME, and had reasonably good predicted antiviral biological activity. Of the top compounds ZINC000000899909, ZINC000015151344, ZINC000013378519, ZINC000000086470, and PC11371962 also passed toxicity screening, where compounds ZINC000001632866 and ZINC000095486204 failed. ZINC000001632866 was specifically predicted to have antiviral activity against poxviruses. Since both compounds had potential for A42R specific inhibition and had high binding affinities, their ligand-protein interactions were of interest as they could be used as scaffolds or for optimization for future drug design, they were included in the MD simulations.

Free binding energies and other energy contributors namely van der Waals (vdW), electrostatic, polar solvation and solvent accessible surface area (SASA) energies were also calculated using MM/PBSA ( Table 4 ) (Kumari et al., 2014). The vdW energy of the A42R-ligand complexes ranged between –98.944 and –144.534 kJ/mol, where A42R-ZINC000000086470 had the least negative vdW energy and A42R-ZINC000013378519 displayed the most negative vdW energy ( Table 4 ). The second most negative vdW energy was A42R-ZINC000001632866 with –125.513 kJ/mol followed by A42R-ZINC000000899909 with –121.779 kJ/mol. SASA energies ranged from –12.646 to –17.900 kJ/mol where A42R-tecovirimat had the least negative SASA energy and A42R-ZINC000013378519 had the most negative SASA energy ( Table 4 ). The SASA energy has a linear relationship to non-polar solvation energy and differs minimally between structurally similar ligands (Kumari et al., 2014; Kollman et al., 2000; Genheden and Ryde, 2015).

A42R-tecovirimat had the highest binding energy at –68.694 kJ/mol ( Table 4 ). Tecovirimat is a known inhibitor of MPXV, inhibiting p37 and F13L phospholipase and has not been shown to bind A42R (Sherwat et al., 2022; Mbrenga et al., 2022; Desai et al., 2022; Mucker et al., 2013). The other seven ligands had lower binding energies than tecovirimat suggesting that they might have higher affinities for A42R ( Table 4 ). The compound with the lowest binding energy was ZINC000000899909 at –97.140 kJ/mol and close behind was ZINC000001632866 at –94.219 kJ/mol and ZINC000013378519 at –87.652 kJ/mol ( Table 4 ). PC-11371962, was structurally similar to tecovirimat, but had a higher affinity for A42R with a binding energy of –75.443 kJ/mol ( Table 4 ). ZINC000000086470, ZINC000095486204, and ZINC000015151344 all had comparable binding energies with –74.667, –74.196, and –73.252 kJ/mol, respectively ( Table 4 ). These compounds have relatively good binding energies for A42R and provide rationale for in vitro validation of their inhibitory activity against MPXV, as well as other poxviruses. Compounds ZINC000001632866 and ZINC000095486204 did not pass toxicity screening but can be used as scaffolds and for information on A42R binding to aid in drug development targeting A42R, as they had low binding energies.

MPXV protein A42R (PDB ID: 4QWO) experimentally determined by X-ray diffraction with resolution of 1.52 Å, was retrieved from the RCSB PDB (Rose et al., 2017; Burley et al., 2021). The structure from RSCB PDB was in complex with ligands, ions, cofactors, and water molecules that were removed using PyMOL. The retrieved structure had missing residues; thus, chain A was remodeled using EasyModeller, a graphical user interface of Modeller. The complete A42R sequence was retrieved from UniProt with corresponding ID: Q8V4T7 (strain Zaire-96-I-16).

The remodeled structure had a discrete optimized protein energy (DOPE) score of –16024.13965, mol.pdf of 725.07245, and a genetic algorithm 341 (GA341) score of 1.0000. When aligned to the 4QWO structure, an RMSD of 0.092 was observed. GROningen MAchine for Chemical Simulations (GROMACS) v5.1.1 was used to energy minimize the protein structure using two different force fields namely, OPLS/AA and CHARMM36 force fields (Lu et al., 2021; Jo et al., 2008). This was done to compare and select the structure with the least energy after minimization. The A42R energy minimized protein in the GROMACS format (.gro) was then converted to pdb format after removing water and ions.

The binding sites of the A42R were predicted using CASTp 3.0. Usually, relatively large binding pockets correlate to an active site, though there are certainly exceptions (Liang et al., 1998). Predicted sites with relatively small cavity sizes were not considered further.

A screening library was generated to conduct structure-based virtual screening (SBVS) to identify potential binders of A42R. An integrated screening library comprised of 36,366 compounds from Traditional Chinese Medicine (TCM) database obtained from TCM@Taiwan, African Medical Plants (AfroDB), and PubChem (Chen, 2011; Ntie-Kang et al., 2013; Kim et al., 2023) was created. TCM and AfroDb are catalogues from the ZINC15 database (Sterling and Irwin, 2015). There were 35,161 compounds from TCM and 880 compounds from AfroDb. The 35,161 compounds from TCM were pre-filtered for compounds with molecular weights between 150 g/mol and 600 g/mol as done previously, leaving 25,196 compounds used from TCM (Kwofie et al., 2021). There were 325 compounds obtained from PubChem that were structurally similar to smallpox inhibitors tecovirimat, tembexa, and cidofovir. Tecovirimat was included in the library as a control because it has been FDA approved for the treatment of smallpox and has shown inhibition of MPXV (Adler et al., 2022; Frenois-Veyrat et al., 2022; Sherwat et al., 2022; Warner et al., 2022). Ligand structures originating from PubChem were downloaded in 3D spatial data file (.sdf) and merged with the compounds from AfroDb. All compound structures were imported into PyRx, energy minimized using the universal force field (UFF) and conjugate gradient algorithm in 200 steps, and then converted to pdbqt format (Rappe et al., 1992).

AutoDock Vina (embedded in PyRx version 0.9.2), a docking program commonly used to perform protein-ligand docking, was used to screen the library for potential A42R binders and shortlist compounds for further assessment (Trott and Olson, 2010; Dallakyan and Olson, 2015). Docking used an exhaustiveness set to 8 with grid box dimensions of 37.964 × 20.791 × 28.223 ų and A42R centered at x = 30.406 Å, y = 22.08 Å, and z = 27.741 Å. The grid was made by setting the box to the following residues: Met1, Glu3, Trp4, Lys6, Ile7, Asp10, Ile22, Thr99, Ile104, His124, Ala125, Arg127, Val128, Thr131, and Asn133. For each ligand screened, AutoDock Vina generated up to 9 conformers. The poses were visualized using PyMOL to assert that the ligand was accurately bound in pocket 1 (the selected binding site). Due to the size disparity between the TCM database and the AfroDb and PubChem databases they were analyzed separately. The binding energy of -7.0 kcal/mol is a threshold specific to AutoDock Vina that separates putative binders and non-binders (Chang et al., 2007). The –7.0 cutoff has data to support that it filters around 95% of the non-inhibitors, but still passes about 98% of known inhibitors (Chang et al., 2007). The top 1% from the TCM ligands and all ligands below the –7.0 kcal/mol cutoff for AfroDB and PubChem were then shortlisted.

To better characterize the shortlisted compounds’ pharmacokinetic profiles and drug-likeness they were ran through ADME testing via SwissADME (Daina et al., 2017). Shortlisted compounds were selected based on passing both Lipinski’s rule of five and Veber’s rule. The Lipinski’s rule of five requirement is met if the compound has one or less violations of the following rules: ≤ 5 hydrogen bond donors, ≤ 10 hydrogen bond acceptors, a molecular weight < 500 Da, and a lipophilicity or octanol-water partition coefficient (logP) ≤ 5 (Lipinski, 2016; Lipinski, 2004; Lipinski et al., 2001a; Lipinski et al., 2001b; Mullard, 2018). Veber’s rule requires ≤ 10 rotatable bonds and a topological polar surface area (TPSA) ≤ 140 Ų (Veber et al., 2002).

To predict the toxic profiles of the compounds regarding potential mutagenicity, tumorigenicity, irritancy and reproductive effects, OSIRIS DataWarrior version 5.5.0 was used (Sander et al., 2015). DataWarrior predicts potential toxicities of compounds, classifying them as none, low, or high predicted risk for each property in question. Any compounds resulting in low or high toxicity regarding potential mutagenicity, tumorigenicity, or irritancy were removed from further consideration. Eliminating potential carcinogens was of importance because MPXV has been linked to increase tumor immunity and postulated to increase risk of tumor development (Liao et al., 2022). Also, commonly MPXV patients are coinfected with HIV (Hoffmann et al., 2022). HIV has been linked to an increased risk of cancer in what is termed AIDS-defining cancers (Grulich et al., 2007; Hernández-Ramírez et al., 2017). Potential toxicity regarding reproductive effects are reported but did not result in elimination of compounds for consideration.

Prediction of Activity Spectra of Substances (PASS) was used to predict biological activity of the shortlisted compounds. Of most interest was the compounds’ potential antiviral activity. PASS reads the SMILES format of the compound and then compares the structures of the molecules to its dataset comprised of active and inactive structural groups (Lagunin et al., 2000; Parasuraman, 2011; Filimonov et al., 2014). The read out for each compound is then a comparison of the probability of activity (Pa) to the probability of inactivity (Pi) where when Pa is greater than Pi the compound has the potential for that activity. To further corroborate potential activity, a similarity search for the shortlisted compounds was done using DrugBank to identify structural similarities with compounds that have experimentally validated antiviral activity (Wishart et al., 2008; Wishart et al., 2018).

GROMACS v5.1.1 was used for carrying out MD simulations (Abraham et al., 2017; Abraham et al., 2015). GROMACS software accuracy has been assessed in a comparison with experimental data supporting its usage for CADD (Childers and Daggett, 2018). Drug discovery relies on protein-ligand interactions where ligand binding has dynamic properties like flexibility and conformational changes that must be accounted for in drug design (Durrant and McCammon, 2011; De Vivo et al., 2016). MD simulations take into account the conformational changes and the movements associated with receptor-ligand binding interactions (Durrant and McCammon, 2011; De Vivo et al., 2016). These simulations are a computational method used to study the movement of atoms in a system using physics that modulates electric force changes in bonded and non-bonded atoms (Durrant and McCammon, 2011; Cheng and Ivanov, 2012). MD simulations still have limitations including sometimes necessary long simulation run times to accurately describe specific dynamic properties and insufficient mathematical models of forces influencing protein dynamics (Childers and Daggett, 2018). Even with limitations, MD simulations have supported research comparing simulation results with experimental data for their pertinent use in drug discovery (Adelusi et al., 2022; Childers and Daggett, 2018; Durrant and McCammon, 2011; De Vivo et al., 2016; Cheng and Ivanov, 2012).

To prepare ligands for MD, the ligand topologies for OPLS force field were created using LigParGen (Dodda et al., 2017). Solvation of the systems were made using a cubic box with the “TIP4P” water model and sodium or chlorine ions were added to neutralize charges (Lu et al., 2021; Nguyen et al., 2014). A42R-ligand systems prior to MD simulation were subjected to constant number, constant-volume and constant-temperature (NVT) and constant number, constant-pressure and constant-temperature (NPT). To evaluate the structural stability, folding and conformational fluctuations of A42R during MD simulations the RMSD, radius of gyration (Rg) and the root mean square fluctuation (RMSF) were calculated post simulation. At 25 ns intervals, snapshots were generated to ascertain the position of ligands with A42R.

Interaction maps of the top seven compounds and tecovirimat with A42R were generated using LigPlot+. Hydrogen bonds during the MD simulations were monitored using GROMACS “gmx hbond”. Molecular interactions between the ligands and the A42R binding pocket are important to recognize for future studies as potential drug candidates.

MM/PBSA methods have been used successfully to reproduce experimental findings and are becoming more efficient and reliable methods for analyzing protein-ligand interactions (Genheden and Ryde, 2015; Sgobba et al., 2012). MM/PBSA performance evaluations have supported a higher yield of enrichment factors as compared to yields from docking scores alone and give reasonably accurate free energy calculations (Sgobba et al., 2012). MM/PBSA estimates the Gibbs free energy of binding, ΔG_(bind) of ligands to protein (Genheden and Ryde, 2015; Borkotoky et al., 2016). For drug discovery, the most negative ΔG_(bind) can be used to prioritize compounds for experimental trials (Sgobba et al., 2012). For this study the MM/PBSA approach was used to generate binding free energies and to compute the energy contributions per residue for each of the A42R-ligand complexes.