Preclinical evaluation of GW406108X, a broad-spectrum therapeutic blocking ULK1/2-mediated autophagy to inhibit enterovirus replication

Abstract

Enterovirus D68 (EV-D68) is an emerging respiratory pathogen, with severe cases developing into acute flaccid myelitis, a paralytic condition. Despite growing concern, no approved antivirals currently exist for EV-D68, underscoring the urgent need for therapeutic discovery. Here, we report the development of a high-throughput, phenotypic-based assay to screen 7986 compounds across nine diverse compound libraries for EV-D68 antivirals, on two EV-D68-susceptible cell lines (RD and H1299). The screen identified GW406108X as a potent post-entry inhibitor of EV-D68 infection (EC₅₀: 1.804 μmol/L). Mechanistic studies using luciferase replicon assays, siRNA knockdowns, and drug-resistant mutant generation suggest that GW406108X targets the autophagy pathway through ULK1/2 inhibition. Transmission electron microscopy and fluorescence bioimaging demonstrate a significant reduction in autophagosome formation in treated, infected cells. This disruption likely impairs the virus’s ability to exploit autophagy, which may in turn hinder replication organelle formation and non-lytic virion release, leading to reduced viral replication. Preclinical evaluations showed that GW406108X demonstrates strong antiviral efficacy without detectable cytotoxicity. These findings reveal a previously uncharacterized antiviral mechanism and position GW406108X as a promising candidate for antiviral therapeutic development. Findings from this study expand the current antiviral landscape for enteroviruses and represent a significant step toward clinical intervention for EV-D68 and potentially related viral pathogens.

Graphical abstract

GW406108X inhibits EV-D68 replication by blocking ULK1/2-mediated autophagy, showing strong antiviral activity both in vitro and in vivo.

1. Introduction

Enterovirus D68 (EV-D68) is a common enterovirus from the species Enterovirus D, causing respiratory disease¹. EV-D68 was first identified in 1962 from four children with pneumonia and bronchiolitis and has resulted in cluster outbreaks of respiratory disease in USA, Europe, Southeast Asia, and Africa. In 2014, 2287 cases were confirmed worldwide and reported in 20 countries². Due to a lack of surveillance for EV-D68 infection specifically and its symptoms being similar to that of other respiratory diseases, EV-D68 infection has been difficult to diagnose. With increased awareness of viruses with the potential to cause respiratory disease and their pandemic potential, it is hoped that efforts would be directed to improving the detection of EV-D68.

Unlike many other enteroviruses which transmit through the faecal–oral route, EV-D68 infection mainly transmits through the airways. EV-D68 is acid-labile and similar to rhinovirus in its disease manifestation³. Although EV-D68 infection is more commonly observed in children, healthy adults can also be infected. Most infections present flu-like symptoms including coughing, wheezing, and difficulty breathing. Severe cases of infection develop into pneumonia, respiratory failure, or acute flaccid myelitis, a grave condition of the spinal cord that could result in paralysis⁴. This associative link between EV-D68 and acute flaccid myelitis was made from several cases of EV-D68 being detected in the cerebrospinal fluid of paediatric patients with both respiratory symptoms and acute flaccid myelitis². Patients with severe lower respiratory disease requiring intensive care and acute flaccid myelitis contribute to serious public health concerns, especially if the patient is a young person. The surge in cases worldwide in 2014 is proof of its pandemic-causing potential, which remains a global threat today. This is further exacerbated by the lack of available antivirals and vaccines for disease treatment and prevention. This signals a need for the discovery of new antivirals to target this viral infection, especially since the virus has great potential to cause debilitating effects. This study has successfully identified and characterised GW406108X, a potent antiviral compound inhibiting EV-D68 infection both in vitro and in vivo. Mechanistic studies have shown that GW406108X exerts its effects by inhibiting the autophagy pathway, a cellular process frequently exploited by enteroviruses to facilitate their own replication. Disruption of autophagy by GW406108X impairs efficient EV-D68 replication. Furthermore, mice that were intranasally inoculated with EV-D68 and subsequently treated with GW406108X exhibited potent antiviral therapeutic efficacy. This novel compound holds significant therapeutic potential, offering a promising strategy for antiviral intervention and paving the way for the development of targeted treatments against a broad spectrum of enteroviral infections.

2. Materials and methods

2.1. Cell lines

Mammalian cell lines were cultured at 37 °C and under 5% CO₂ in their respective culture mediums with 10% fetal calf serum (FCS, Cytiva, Malborough, MA, USA) and 0.2% NaHCO₃. The following cell lines were cultured in Dulbecco’s modified Eagle’s media (DMEM, Thermo Fisher Scientific, Waltham, MA, USA) in heat-inactivated FCS: rhabdomyosarcoma (RD) cells, non-small cell lung cancer (H1299) cells, lung carcinoma (A549) cells, and hepatocellular carcinoma (Huh-7) cells. The following cell lines were cultured in Roswell Park Memorial Institute-1640 media (Sigma–Aldrich, St. Louis, MO, USA) in FCS: rhabdomyosarcoma (SJCRH30) cells and baby hamster kidney (BHK21) cells. Human nasal epithelial cells (hNEpC) were cultured in Airway Epithelial Cell Growth Media (PromoCell, Heidelberg, Germany), and adenocarcinoma (HeLa Ohio) cells were cultured in Eagle’s minimum Essential medium (Lonza, Basel, Switzerland) supplemented with heat-inactivated FCS. For majority of the viral titre quantifications with EV-D68 infection, the seeding density of RD cells was 2.6 × 10⁵ cells/well.

2.2. Viruses

The main virus used in this study is EV-D68 strain US/KY/14-18953 (EV-D68, KM851231.1). Other viruses include clinical isolate EV-A71 strain 5865/sin/000009 (EV-A71, AF316321.2), human enterovirus 71 strain H (EV-A71 subgenogroup H, AY053402.1), human enterovirus 71 isolate NUH0083/SIN/08 (EV-A71 subgenogroup B5, FJ461781.1), human enterovirus 71 strain 540V/VNM/05 (EV-A71 subgenogroup C4, JQ965759.1), Coxsackievirus A6 Singapore patient isolate (GenBank accession number not available), Coxsackievirus A16 G-10 (U05876.1), human echovirus 7 strain Wallace (AF465516.1), Coxsackievirus A24 isolate 110390 (KF725085.1), Enterovirus D68 isolate TW-02795-2014 (EV-D68 TW strain, KT711088.1), human rhinovirus A strain HRV-A10_p1038_sR155_2007 (RV-A16, JN541269.1), dengue virus serotype 2 strain New Guinea C (DENV-2, KM204118.1), and chikungunya virus isolate SGEHICHD122508 (CHIKV, FJ445502.2). We would like to thank Prof Shi Shin Ru, Chang Gung University for kindly providing us with Enterovirus D68 isolate TW-02795-2014 (EV-D68 TW strain, KT711088.1). EV-D68 and RV-A16 infection were carried out at 33 °C, whereas all other virus infections performed at 37 °C. Viral plaque assays for all enteroviruses were carried out in RD cells, HeLa Ohio cells for RV-A16, and BHK21 for both CHIKV and DENV-2. For all enteroviruses, the infection duration was 12 h. Infection was DENV-2 and CHIKV were 48 and 24 h respectively.

2.3. Drug libraries

The drug libraries used for the high-throughput screen were Food and Drug Administration (FDA)-approved drug library (Selleckchem, Houston, TX, USA), flavonoid library (TimTec, Newark, DE, USA), natural product library (Enzo Life Sciences, Farmingdale, NY, USA), Angiotensin-converting enzyme 2 inhibitor library (TargetMol, Boston, MA, USA), RNA-targeting compound library (Otava, Vaughan, ON, Canada), Traditional Chinese Medicine compound library (Professor Erwei Hao, Guangxi University of Chinese Medicine, China), glycoside library (MedChemExpress, Monmouth Junction, NJ, USA), peptidomimetic library (MedChemExpress), and kinase inhibitor library (MedChemExpress). All compounds were diluted in dimethyl sulfoxide (DMSO) to a stock concentration of 100 μmol/L and stored at −20 °C.

2.4. High-throughput drug screens

H1299 (or RD) cells were seeded in a clear 96-well plate and left in the incubator at 37 °C and under 5% CO₂ overnight. The seeding densities used were 2.3 × 10⁴ cells/well. The next day, cells were infected with 50 μL of EV-D68 strain US/KY/14-18953 at a multiplicity of infection (MOI) of 10 (5) and incubated in the 33 °C and under 5% CO₂ incubator for 1 h, with rocking every 15 min. After 1 h, the virus was decanted, and 100 μL of drug compounds diluted in DMEM (2% FCS) was added. The final concentration of drug compounds was 10 μmol/L. In each plate, control wells include mock-infected, 0.1% DMSO-treated, and 20 μmol/L fidaxomicin-treated. Media from wells were decanted 24 (12) h post-infection (h.p.i.) and fixed with 50 μL of ice-cold methanol at −20 °C for 10 min. Methanol was then aspirated with the Zoom HT High-throughput Microplate Washer (Berthold Technologies, Bad Wildbad, Germany), and the cells permeabilised with 50 μL of 0.1% Triton X-100 for 20 min. Wells were then washed thrice with 100 μL of 1 × phosphate buffered saline (PBS), and stained with 25 μL of EV-D68 VP1 antibody (GeneTex, Irvine, CA, USA, 1:200) for 1 h at 37 °C in the drying oven. After 1 h, wells were washed thrice with 100 μL of 1 × PBS and stained with 25 μL of anti-rabbit fluorescein isothiocyanate (FITC) antibody (Merck Millipore, Burlington, MA, USA, 1:200) for 1 h at 37 °C in the drying oven. Following three washes with 100 μL of 1 × PBS, 50 μL of 4′,6-diamidino-2-phenylindole (DAPI) (Thermo Scientific, 1:50,000) was added into each well and staining performed at room temperature for 15 min. Finally, wells were washed thrice with 100 μL of 1 × PBS and 100 μL of 1 × PBS dispensed into each well.

High content imaging was performed on the Operetta CLS High-Content Imaging System (Revvity, Waltham, MA, USA). Signals from FITC staining represent EV-D68-infected cells and DAPI staining represent the total number of cells. For each well, two images were taken at different image positions of the well. Images were analysed via CellProfiler 4.1.3 to identify 1) the number of EV-D68-infected cells and 2) the total number of cells (nuclei count). The percentage infection was calculated by taking the ratio of the number of EV-D68-infected cells to the total number of cells. Relative percentages of infection were calculated by normalising to the 0.1% DMSO-treated condition within the same plate.

2.5. Drug compounds

Drug compounds selected for validation were purchased from various sources depending on the drug library from which it was identified. According to the material safety data sheet, compounds were dissolved in DMSO to the highest possible stock concentration. Drug compounds were stored at −20 °C and was diluted in its respective media to the test concentrations before use. For all assays, vehicle control (0.1% DMSO) was included.

2.6. Cytotoxicity assay

Cells were seeded in a clear 96-well plate at a seeding density of 2.3 × 10⁴ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were treated with varying concentrations of the drug compounds and incubated at the respective assay durations at 37 °C (33 °C for EV-D68 and RV-A16) and under 5% CO₂. As controls, 0.1% DMSO-treated wells were included. After which, 10 × alamarBlue Cell Viability Reagent (Thermo Fisher Scientific) was diluted in a 1:10 ratio in the respective maintenance media. Media was decanted from the wells and 100 μL of 1 × alamarBlue reagent added into each well. The cells were then left in the incubator at 37 °C (33 °C for EV-D68 and RV-A16) and under 5% CO₂ for 1–4 h, until wells in the 0.1% DMSO-treated condition turned pink. Fluorescence was then read using the Infinite 200 series microplate reader (Tecan, Männedorf, Switzerland) at an excitation wavelength of 570 nm and emission wavelength of 600 nm. Raw fluorescence values were background-subtracted and normalised to that of the 0.1% DMSO condition to obtain relative cell viabilities.

2.7. Viral plaque assay

Cells were seeded in a 24-well plate at a seeding density of 2.6 × 10⁵ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were infected with 100 μL of virus in serial dilutions up to 10⁻⁶ and incubated in the 37 °C (33 °C for EV-D68 and RV-A16) and under 5% CO₂ incubator for 1 h (1.5 h for CHIKV), with rocking every 15 min. Subsequently, wells were washed twice with 1 mL of 1 × PBS. For the enterovirus plaque assays, wells were then overlaid with 0.5 mL of DMEM (2% FCS) with 0.5% agarose (Vivantis, Subang Jaya, Selangor, Malaysia). For DENV-2 and CHIKV plaque assays, wells were overlaid with 0.5 mL of 1% carboxymethyl cellulose (CMC) in Roswell Park Memorial Institute-1640 media. Plates were incubated at 37 °C (33 °C for EV-D68 and RV-A16) and under 5% CO₂. Plaque assay infection durations are two days for all enteroviruses except EV-D68 and RV-A16, where plates were incubated for four days. DENV-2-infected and CHIKV-infected plates were incubated for six and three days, respectively. Then, cells were fixed and stained with 4% paraformaldehyde (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) and 1% crystal violet solution (Sigma–Aldrich). Plaques were counted manually in the dilution factor with distinct plaque formation to obtain the final value in plaque forming units (PFU)/mL.

2.8. Pre- and post-infection assays

Cells were seeded in a clear 96-well plate at a seeding density of 2.3 × 10⁴ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. For the pre-infection assay, 100 μL of drug compounds diluted in DMEM (2% FCS) was added and the plates incubated in the 37 °C (33 °C for EV-D68) and under 5% CO₂ incubator for 2 h. After 2 h, wells were washed twice with 100 μL of 1 × PBS, infected with 50 μL of virus at an MOI of 1 and incubated in the 37 °C (33 °C for EV-D68) and under 5% CO₂ incubator for 1 h, with rocking every 15 min. For the post-infection assay, virus infection was carried out first and drug compounds added 1 h post-infection. Subsequently, the virus was decanted, and wells washed twice with 100 μL of 1 × PBS. Then, 100 μL of DMEM (2% FCS) was added into each well and incubated for 12 h. Plates were then frozen at −80 °C and two rounds of freeze–thaw cycles performed before viral plaque assay.

2.9. Time-of-addition (TOA) and time-of-removal (TOR) assays

Cells were seeded in a clear 96-well plate at a seeding density of 2.3 × 10⁴ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were infected with 50 μL of virus at an MOI of 1 and incubated in the 33 °C and under 5% CO₂ incubator for 1 h, with rocking every 15 min. Subsequently, the virus was decanted, and wells washed twice with 100 μL of 1 × PBS. For the TOA assay, 100 μL of DMEM (2% FCS) was added into each well. At the timepoints 2, 4, 6, 8, and 10 h.p.i., media was removed from the wells and 100 μL of 10 μmol/L of GW406108X added. For the TOR assay, 100 μL of 10 μmol/L of GW406108X was added into each well. At the same timepoints, drug compound was removed and replaced with 100 μL of DMEM (2% FCS). At 12 h.p.i., the plates were frozen at −80 °C and two rounds of freeze–thaw cycles performed before viral plaque assay.

2.10. Entry bypass assay

Viral RNA was first extracted from 500 μL of viral stock using the QIAmp Viral RNA Mini Kit (Qiagen, Hilden, Germany). Cells were seeded in a clear 96-well plate at a seeding density of 2.3 × 10⁴ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were transfected with the following mixture per well: 100 ng viral RNA, 0.2 μL DharmaFECT 1 (Horizon Discovery, Cambridge, UK), to 20 μL/well with Minimum Essential Medium/Earle’s Balanced Salt Solution (Cytiva, Malborough, MA, USA). Transfection mixtures were incubated at room temperature for 20 min before being added into wells containing 80 μL of DMEM (2% FCS). The transfection duration was 6 h and at 33 °C. After the transfection duration, media in wells were decanted and replaced with 100 μL of drug compounds. Other than the vehicle control, 10 μmol/L pleconaril was included as a negative control. At 12 h post-transfection, the plates were frozen at −80 °C and two rounds of freeze–thaw cycles performed before viral plaque assay.

2.11. Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Cells were seeded in a 24-well plate at a seeding density of 1.8 × 10⁵ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were infected with 100 μL of virus at an MOI of 1 and incubated in the 33 °C and under 5% CO₂ incubator for 1 h, with rocking every 15 min. Subsequently, wells were washed once with 1 mL of 1 × PBS, and 1 mL of drug compounds added. At 3, 6, and 12 h.p.i., the supernatant is removed from the wells, and total RNA extracted using the Total RNeasy Kit (Qiagen, Hilden, Germany) and following the manufacturer’s protocol. From RNA, complementary DNA (cDNA) was obtained using MMLV reverse transcriptase (Promega, Madison, WI, USA). Following cDNA synthesis, qPCR reactions were prepared. For EV-D68, qPCR was performed using the probe-based method. Each 20 μL reaction consisted of 10 μL of 2 × Primer qRT-PCR mix (Integrated DNA Technologies, Coralville, IA, USA), 0.5 μL of FAM5′-NFQ-MGB3′ probe (Integrated DNA Technologies, Coralville, IA, USA), 1 μL of primer, 2 μL of cDNA, and topped up with nuclease-free water.

Reactions were run on the StepOnePlus Real-Time PCR System (Applied Biosystems, Waltham, MA, USA) at the following conditions: 95 °C for 10 min, followed by 40 cycles of denaturation at 95 °C for 15 s, and annealing and extension at 60 °C for 1 min. In the same reaction plate, β-actin was detected as the endogenous control. The double delta cycle threshold (C_t) method was utilised to obtain fold change values for both positive- and negative-sense viral RNA, normalised to the 0.1% DMSO-treated condition. Primer sequences are: EV-D68 sense: 5′-CAAACTCGCACAGTGATAAAYCARCA-3′; EV-D68 anti-sense: 5′-ACTGCYACAGTTGTRAGTATRGTRATYTCAGCA-3′; β-actin forward primer: 5′-TCGGTGAGGATCTTCATGAGGTA-3′; β-actin reverse primer: 5′- TCACCCACACTGTGCCCATCTACGA-3′.

2.12. Sodium dodecyl-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting

Cell lysates were harvested in 1 × sodium dodecyl buffer (SDS) buffer (0.05 mol/L Tris–HCl, 0.1 mol/L dithiothreitol (DTT), 69.25 mmol/L SDS, 1.5 mmol/L bromophenol blue, 1.075 mol/L glycerol) after removal of supernatant and stored at −80 °C. Cell lysates were boiled at 100 °C for 10 min and separated on a 10% SDS-PAGE gel at 100 V. This step was carried out on the Mini-PROTEAN Tetra Vertical Electrophoresis Cell (Bio-Rad, Hercule, CA, USA). Protein bands were transferred using the Trans-Blot Turbo System (Bio-Rad) using a methanol-activated polyvinylidene difluoride (PVDF) membrane and under Mixed Molecular Weight transfer condition. The membrane was first incubated for 1 h at room temperature in blocking buffer which is either 2% bovine serum albumin (BSA, Sigma–Aldrich) or 5% skimmed milk (Nature One, Auckland, New Zealand) dissolved in 1 × Tris-buffered saline with 0.1% Tween 20 (1 × TBST). Primary antibody staining was carried out overnight at 4 °C. The next day, after washing the membrane thrice with 1 × TBST, secondary antibody incubation was carried out for 1 h at room temperature. After washing the membrane thrice with 1 × TBST, the membrane was incubated in 3 mL of Immobilon Western Chemiluminescent Horseradish Peroxidase (HRP) substrate (Merck Millipore). Chemiluminescence was detected via the C-DiGit Chemiluminescence Western Blot Scanner (LI-COR) and ImageStudio Digits Version 4.0 software. For all Western blot assays, a housekeeping gene control was included (β-actin or α-tubulin). Densitometry analysis of protein bands was carried out using ImageJ and protein bands normalised to that of the housekeeping protein band intensity.

The primary antibodies used are rabbit anti-EV-D68 VP1 (GeneTex, 1:2000), rabbit anti-ULK1 (ABclonal, Woburn, MA, USA, 1:2000), rabbit anti-ULK2 (Abcam, Cambridge, UK, 1:2000), rabbit anti-KIF15 (Proteintech, Rosemont, IL, USA, 1:1000), rabbit anti-RAF1 (ABclonal, Woburn, MA, USA, 1:1000), mouse anti-LC3 (Cell SignalLing Technology, Danvers, MA, USA, 1:1000), mouse anti-β-actin (Sigma–Aldrich, 1:10,000), and mouse anti-α-tubulin (Invitrogen, Waltham, MA, USA, 1:10,000). The secondary antibodies used are goat anti-mouse IgG (H + L), HRP (Thermo Fisher Scientific, 1:10,000) and goat anti-rabbit IgG (H + L), HRP (Thermo Fisher Scientific, 1:10,000).

2.13. Construction and establishment of EV-D68 NanoLuc replicon system

For overlap PCR, the fragments were first obtained via PCR from other available plasmids in the laboratory. Following PCR purification, overlap PCR was conducted by mixing the three fragments together in a 1:1:1 M ratio, based on 50 ng of the largest fragment. A PCR reaction was set up with five cycles. Next, primers for the start and end of the combined fragment FHindIII-5′UTR-F-68TH and SwaI-2CGC-R-68TH were added immediately to 1:10, 1:100, and 1:1000 dilutions of the original PCR reaction. A PCR reaction was set up with 30 cycles, and a diagnostic agarose gel electrophoresis run to check for the combined fragment of the correct size. The vector used was pcDNA3.1+. The following primers were used: FHindIII-5′ UTR-F-68TH: 5′-AACTTAAGCTTTTAAAACAGCTTTGGGGTTGTTCCCACTC-3′; 5′UTRGC-Nluc-R-68TH: 5′-GTGAAGACCATGGTGGCTATTAAAATTTTCAAATCAAGGTT-3′; Nluc-5′ UTRGC-F-68TH: 5′-GAAAATTTTAATAGCCACCATGGTCTTCACACTCGAAGATTTCGTTGGG-3′; Nluc-2AGCR-68TH: 5′-TCCGAAGCCTGGACCAGTATTCACTAACGCCAGAATGCGTTCGCA-3′; 2AGC-Nluc-F-68TH: 5′-CGTTAGTGAATGTCCAGGCTTCGGAG-3′; SwaI-2CGC-R-68TH: 5′-ATATCTACATCAAATTTAAATCTGCGTGACAAAGCCTTTG-3′. Deletion of the viral 3Dpol region was performed using the following primers: 3DdelGC-F-68TH: 5′-ATCCCAGAACAACTCTAATTGAAACCCAAGTTATAGTTAC-3′ and 3DdelGC-R-68TH: 5′-GAGTTGTTCTGGGATCTTTAGTCCATCTAATAGATTC-3′.

H1299 cells were seeded in a white 96-well plate at a seeding density of 2.3 × 10⁴ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were transfected with the following mixture per well: 50 ng replicon DNA, 0.1 μL jetPRIME reagent (Polyplus, Illkirch, France), to 10 μL/well with jetPRIME buffer (Polyplus, Illkirch, France). Transfection mixtures were incubated at room temperature for 10 min before being added into wells containing 90 μL of DMEM (2% FCS). Transfection was carried out for 24 h at 33 °C. After the transfection duration, media in wells were decanted and replaced with 100 μL of drug compounds. The control compounds guanidine hydrochloride (Sigma–Aldrich) and cycloheximide (Sigma–Aldrich) were added. At 12 h post-treatment, luminescence was read following the protocol recommended for the Nano-Glo Luciferase Assay System (Promega).

2.14. Drug-resistant mutant generation

Cells were seeded in a 24-well plate at a seeding density of 1.8 × 10⁵ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were infected with 100 μL of virus at an MOI of 1 and incubated in the 33 °C and under 5% CO₂ incubator for 1 h, with rocking every 15 min. Subsequently, wells were washed twice with 1 mL of 1 × PBS, and 1 mL of drug compounds added. At 12 h.p.i., the plates were frozen at −80 °C and two rounds of freeze–thaw cycles performed. Subsequently, the supernatant was harvested and centrifuged at 1321×g for 10 min at 4 °C to remove cell debris. The supernatant following centrifugation was used to infect another plate of seeded cells (passage 1), and the remaining supernatant frozen at −80 °C. This process was repeated for the next four passages. At every five passages, viral plaque assay was performed to check for resistance. Serial passaging was performed 18 times.

2.15. Immunofluorescence assay

Cells were seeded in a 24-well plate at a seeding density of 1.0 × 10⁵ cells/well, with each well containing a coverslip and left in the incubator at 37 °C and under 5% CO₂ overnight. After the desired assays were performed, cells were fixed in either 4% paraformaldehyde at room temperature for 20 min or in ice-cold methanol at −20 °C for 10 min. Immunofluorescence staining was first performed with 50 μL of primary antibody for 1 h at 37 °C. After washing in 1 × PBS, the coverslips were incubated with 50 μL of secondary antibody for 1 h at 37 °C. The coverslips were washed in 1 × PBS and placed on a microscope slide with 2 μL of DAPI with mounting medium (ProSciTech, Kirwan, QLD, Australia). Fluorescence images were taken under the Olympus IX-81 microscope with immersion oil where required.

The primary antibodies used in this assay are rabbit anti-EV-D68 VP1 (GeneTex, 1:200), mouse and rabbit anti-LC3 (Cell Signaling Technology, 1:200), and mouse anti-dsRNA (Absolute Antibody, Oxford, UK, 1:200). The secondary antibodies used in this assay are goat anti-mouse IgG antibody, (H + L) FITC conjugated (Merck Millipore, 1:200), goat anti-rabbit IgG antibody, (H + L) FITC conjugated (Merck Millipore, 1:200), goat anti-mouse IgG (H + L) secondary antibody, TRITC (Thermo Fisher Scientific, 1:200), and goat anti-rabbit IgG (H + L) secondary antibody, TRITC (Thermo Fisher Scientific, 1:200).

2.16. Small interfering RNA (siRNA) knockdown assay

The following siGENOME SMARTpool siRNA were purchased from Dharmacon: human ULK1 (8408), human ULK2 (9706), human KIF15 (56992), human RAF1 (5894), and non-targeting control (NTC). The siRNAs were resuspended in 1 × siRNA buffer (Horizon Discovery, Cambridge, UK) following the siRNA resuspension protocol. H1299 cells were reverse-transfected with either 25 or 50 nmol/L of siRNA with 0.2 μL of DharmaFECT 1 (Dharmacon, Lafeyette, CO, USA), to 20 μL/well with Minimum Essential Medium/Earle’s Balanced Salt Solution. Transfection mixtures were incubated at room temperature for 30 min. Cells were left in the incubator at 37 °C and under 5% CO₂ for 48 h. Subsequently, cells were infected with 50 μL of EV-D68 at an MOI of 0.1 and incubated in the 33 °C and under 5% CO₂ incubator for 1 h, with rocking every 15 min. Subsequently, the virus was decanted, and wells washed twice with 100 μL of 1 × PBS. Next, 100 μL of DMEM (2% FCS) was added into each well. At 12 h.p.i., the plates were either frozen at −80 °C for viral plaque assay after two freeze–thaw cycles or the cell lysates were harvested for Western blotting.

2.17. Transmission electron microscopy (TEM)

Cells were seeded in T-75 flasks at a seeding density of 5.0 × 10⁶ cells/well and left in the incubator at 37 °C and under 5% CO₂ overnight. The next day, cells were infected with 3 mL of virus at an MOI of 20 and incubated in the 33 °C and under 5% CO₂ incubator for 1 h, with rocking every 15 min. Subsequently, media in the flasks were replaced with either 0.1% DMSO or 5 μmol/L of GW406108X in DMEM (2% FCS). Flasks were incubated for 16 h at 33 °C and under 5% CO₂ condition. Flasks were fixed with 4% paraformaldehyde with 2.5% glutaraldehyde for 10 min at room temperature and kept at 4 °C overnight. The next day, flasks were washed twice with 10 mL of 1 × PBS. Post-fixation was performed using 1% osmium tetroxide and 1.5% potassium ferrocyanide in 1 × PBS for 1 h, followed by three 10-min washes in deionized water. Cells were scraped, pelleted by centrifugation, and subjected to a graded ethanol dehydration series (25% for 5 min, 50% for 5 min; 75%, 95%, 100% for 10 min each), followed by two 10-min incubations in 100% acetone. Infiltration was performed using increasing concentrations of Araldite 502 resin in acetone: 1:1 (30 min), 3:1 (30 min), 6:1 (overnight), and three changes of 100% resin (30 min at room temperature, 30 min at 40 °C, and 1 h at 45 °C). Finally, cells were embedded in fresh Araldite 502 resin in BEEM capsules and polymerized at 60 °C for at least 24 h. Prior to embedding, samples were centrifuged for 30 min. Ultrathin sections (90–100 nm) were prepared using a Leica EM UC6 ultramicrotome and stained with 2.5% gadolinium acetate for 10 min prior to imaging with the JEM-1400Flash (JEOL Ltd., Tokyo, Japan).

2.18. In vivo toxicity evaluation

30 μL of DMSO or 10 mg/kg GW406108X was administered to P4 suckling C57BL/6 mice via intraperitoneal (i.p.) injection daily for six days. GW406108X was diluted in sterile-filtered PBS (Invitrogen). Mice were monitored up to Day 17 post-treatment via measurement of body weight and the detailed clinical scores were recorded.

2.19. In vivo efficacy evaluation

30 μL of DMSO, 5 or 10 mg/kg GW406108X was administered to P4 suckling C57BL/6 mice via i.p. injection. After 2 h, mice were intranasally inoculated with 20 μL of mouse-adapted EV-D68 US/MO/14-8949 at 1 × 10⁷ PFU. For the next five days, GW406108X (or DMSO) was administered at the same doses, diluted in sterile-filtered PBS. Mice were monitored up to Day 17 post-infection using the same clinical scoring criteria were recorded.

2.20. Ethical approval

Animal experiments were conducted following the guidelines of an animal biosafety level 2 vivarium with protocol approval by the National University of Singapore Institutional Animal Care and Use Committee (reference number 2021-00091).

2.21. Statistical methods

GraphPad PRISM 9.4.1. (GraphPad PRISM, San Diego, CA, USA) was used for the statistical analyses. Most of the assay results were evaluated with one-way ANOVA and Dunnett’s test for statistical significance, comparing all conditions with the vehicle control treatment condition. Cytotoxic concentration 50% (CC₅₀) values were calculated by fitting the cell viabilities on a non-linear regression curve. The half maximal effective concentration (EC₅₀) values were calculated based on post-infection results fitted to a non-linear regression curve. Resistant mutant results were evaluated with multiple unpaired t-tests for statistical significance, comparing the 0.1% DMSO-treated and drug-treated conditions. Kaplan–Meier survival curves were evaluated using the Mantel–Cox test comparing the vehicle control-treated and drug-treated groups. Statistical significance is shown by the following to compare among group/s: ns signifies P > 0.05, ∗ signifies P ≤ 0.05, ∗∗ signifies P < 0.01, and ∗∗∗ signifies P < 0.001. All data are expressed as mean ± standard deviation (SD).

3. Results

3.1. RD & H1299 are susceptible to EV-D68 infection

Studies of EV-D68 infection have often employed rhabdomyosarcoma (RD) cells, in which the virus replicates efficiently and induces cytopathic effect⁵. We aimed to identify a cell line with greater physiological relevance to EV-D68 infection, as the virus mainly transmits through the respiratory tract. Growth kinetics for five cell lines of various origins were performed to evaluate the susceptibility of these cell lines to EV-D68 replication (Supporting Information Figs. S1 and S2). These cell lines were namely RD, non-small cell lung cancer cell line H1299, human neuroblastoma cell line SK-N-SH, hepatocellular carcinoma cell line Huh-7, and human keratinocyte cell line HaCaT. Plaque assays on cell lysates and supernatants and Western blot assays were used to evaluate the progression of viral replication in these cell lines. As expected, it was observed that cell lysate samples had a consistently higher viral titre as compared to the supernatant. This is because there could still be unreleased virions present in the virus-infected host cell, hence using the cell lysate for plaque assays would be more representative of the viral titres present. Based on the cell lysate viral titre at 0 h post-infection (h.p.i.) and the peak viral titre achieved, a summary of the level of susceptibility of the five cell lines is provided in Table 1. Among the five cell lines tested, HaCaT is the least susceptible to EV-D68 infection, as seen by the lowest peak titre achieved. Overall, it seems that EV-D68 exhibits similar viral growth kinetics in RD, H1299, and Huh-7 cells. Given that EV-D68 infection leads to respiratory illness and is primarily transmitted through the airways, H1299 cells were selected for subsequent experiments alongside RD cells, the commonly used cell line for EV-D68 studies.

Table 1.

Summary of cell lines susceptibility to EV-D68 US/KY/14-18953 infection at MOI 1 by comparing peak viral titres.

Cell line	Tissue origin	Susceptibility to EV-D68 infection
RD	Muscle	+++
H1299	Lung	+++
SK-N-SH	Brain	++
Huh-7	Liver	+++
HaCaT	Skin	+

(+) 1 to 2 log₁₀ value (PFU/mL) increase; (++) 2 to 3 log₁₀ value (PFU/mL) increase; (+++) more than 3 log₁₀ value (PFU/mL) increase.

3.2. Preliminary high throughput phenotypic screen for EV-D68 inhibitors

As mentioned earlier, there has been no available EV-D68 antivirals to date. To identify more hit compounds with potent inhibitory activity against EV-D68 infection, a phenotypic screen was developed on both RD and H1299 cells and used to screen nine different commercially available drug libraries. This was achieved using a commercial antibody for EV-D68 VP1 as a marker for infection. Prior to the primary screen, Z′-factor assessments were performed (Supporting Information Fig. S3). After optimisation of the cell seeding density, multiplicity of infection (MOI), and the timepoint post-infection where the cells were fixed, the calculated Z′-factor values were 0.66 for RD and 0.55 for H1299. This suggests that these assays are well-suited for use in high-throughput settings. RD cells were infected with EV-D68 at MOI 5 and fixed 12 h.p.i., whereas H1299 cells were infected with EV-D68 at MOI 10 and fixed 24 h.p.i. In both assays, immunofluorescence staining was performed after fixation. As a positive control, 20 μmol/L fidaxomicin was used⁶. With this optimised screening workflow, nine drug libraries consisting of 7986 compounds were screened (Fig. 1). Compounds with a normalised percentage infection less than 50% and nuclei count of more than 1000 were considered to be hits. The top 50 hits are shown in Supporting Information Tables S1 and S2. Among the hits, seven compounds were chosen for validation, considering potency, toxicity, and novelty (Table 2).

Figure 1.

Results from primary screen of EV-D68 on H1299 and RD cells. (A) Example images obtained from the primary screens, where immunofluorescence staining was performed to detect EV-D68 VP1 protein (green) and cell nuclei (blue, DAPI). Scale bars represent 200 μmol/L. (B) Scatterplot of results from the primary screen. A total of 7986 compounds were screened; compounds were considered to be hits if the normalised infection rate below 0.5 (black dotted line). Top 50 hits for each screen are shown in Tables S1 and S2.

Table 2.

List of hits selected for validation from EV-D68 immunofluorescence-based phenotypic screen on H1299 cells.

Drug	Chemical structure	Normalised percentage infection	Remarks
Homoharringtonine		30.0%	Echovirus 1 and Chikungunya virus inhibitor (6,7).
Erdafitinib		49.6%	–
Mitoxantrone dihydrochloride		38.4%	EV-A71 inhibitor (5,8).
Dabrafenib		35.4%	–
Baricitinib		33.0%	–
GW406108X		14.5%	–
TC13172		9.46%	–

3.3. GW406108X was identified to be a novel and potent EV-D68 inhibitor

The seven compounds listed in Table 2 were evaluated further through cytotoxicity and post-infection assays (Fig. 2A and B). Cytotoxicity assay was performed to elucidate the CC₅₀ and non-toxic concentrations to be used for post-infection assay. Following the post-infection assay, six out of seven of the chosen compounds inhibited EV-D68 replication. Both homoharringtonine and GW406108X treatment resulted in about a 3log₁₀ PFU/mL reduction in viral titres. For homoharringtonine, the decrease in viral titre was observed from the lowest concentration of 1 μmol/L onwards, whereas in GW406108X a dose-dependent decrease was observed (Fig. 2B). As homoharringtonine has been observed to have antiviral activity against Echovirus 1 and Chikungunya virus, downstream experiments focused on GW406108X instead, which has not been implicated as an antiviral prior to this study^7,8. As GW406108X is a novel compound with CC₅₀ value of >100 μmol/L, EC₅₀ value of 1.804 μmol/L, and a selectivity index of >55.43, its antiviral mechanism of action was investigated in depth.

Figure 2.

Cytotoxicity and post-infection assays for the seven hits chosen for validation. (A) Cytotoxicity assay for seven compounds, where cell viability was normalised to that of the 0.1% DMSO-treated condition, with calculated CC₅₀ values. (B) Post-infection assay for seven hits, carried out at four non-toxic concentrations, with calculated EC₅₀ values. Triplicates were performed with error bars plotted to show the mean and standard deviation. Results were evaluated with one-way ANOVA and Dunnett’s test for statistical significance. ns signifies P ≥ 0.05, ∗P < 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

3.4. GW406108X is a broad-spectrum antiviral compound

To ensure that this antiviral effect is not cell-line specific, GW406108X was evaluated in post-infection assays across various cell lines, including human nasal epithelial cells (hNEpCs), which are primary cells (Supporting Information Figs. S4 and S5). Its spectrum of activity across representative viruses from other species of enterovirus and other positive-sense RNA viruses was also elucidated (Table 3). Altogether, it can be concluded that GW406108X is a pan-enteroviral inhibitor, with similar EC₅₀ values across different cell lines and viruses tested. GW406108X also inhibited orthoflavivirus Dengue serotype 2 infection, but not alphavirus Chikungunya. This could indicate that GW406108X inhibits a shared host factor between enteroviruses and orthoflaviviruses, but not alphaviruses.

Table 3.

Antiviral activity of GW406108X against other viruses.

Cell line	CC50 (μmol/L)	Virus	EC50 (μmol/L)	SI (CC50/EC50)
RD	46.04	Enterovirus A71 subgenogroup B4 (Enterovirus A)	1.120	41.11
		Enterovirus A71 subgenogroup H (Enterovirus A)	2.619	17.58
		Enterovirus A71 subgenogroup C4 (Enterovirus A)	2.143	21.48
		Enterovirus A71 subgenogroup B5 (Enterovirus A)	0.870	52.92
		Coxsackievirus A6 JB141230205 (Enterovirus A)	1.045	44.06
		Coxsackievirus A16 G-10 (Enterovirus A)	1.057	43.56
		Echovirus 7 Wallace (Enterovirus B)	1.210	38.05
		Coxsackievirus A24 110390 (Enterovirus C)	0.329	139.9
H1299	>100	Enterovirus D68 TW-02795-2014 (Enterovirus D)	2.251	>44.42
		Enterovirus D68 US/KY/14-18953 (Enterovirus D)	1.804	>55.43
A549	>100		3.224	>31.02
RD	46.04		3.757	12.25
hNEpC	>100		0.219	456.6
HeLa Ohio	>100	Rhinovirus A16 (Rhinovirus A)	2.785	>35.91
Huh-7	17.73	Dengue virus serotype 2 (DENV-2) 16681	0.936	18.94
SJCRH30	20.54	Chikungunya virus CHIKV-122508	No inhibition	–

3.5. GW406108X acts on a post-entry viral replication stage

For subsequent assays, H1299 was used as the representative cell line. In order to dissect the stages in the viral replication cycle where GW406108X would be involved in, the time-of-addition (TOA) and time-of-removal (TOR) assays were performed (Fig. 3A). From the TOA assay, it can be observed that addition of GW406108X from 4 h.p.i. onwards resulted in increasingly diminishing antiviral effect, evidenced by the steady increase in viral titres at later timepoints post-infection. On the other hand, the antiviral effect of GW406108X persisted despite its removal from 6 h.p.i. and at later timepoints. This indicates that the window of therapeutic effect of GW406108X is likely to be from 4 to 6 h.p.i., coinciding with the viral replication stages viral RNA replication, viral protein translation, virion maturation, and virion release. Results from the entry bypass assay agree with the post-infection timing of GW406108X’s inhibitory effect. This can be seen in Fig. 3B where GW406108X still reduced viral titres even when the entry stages were removed. As expected, the negative control entry inhibitor pleconaril did not inhibit EV-D68 replication in this assay, as viral RNA was directly transfected in H1299 cells⁹.

Figure 3.

Time-of-addition (TOA), time-of-removal (TOR), and entry bypass assays for GW406108X on EV-D68 infection. (A) In TOA, after viral infection, cells were incubated in DMEM (2% FCS) media. At the indicated timepoints, 10 μmol/L of GW406108X was added. In TOR, cells were incubated in 10 μmol/L of GW406108X instead, with wells replaced with fresh media at the same timepoints. All samples were harvested at 12 h.p.i. (B) Entry bypass assay where EV-D68 viral RNA was extracted from live virus and transfected directly into cells (100 ng/well). After 4 h of transfection, the transfection mixture was replaced with GW406108X and 10 μmol/L pleconaril as a negative control. (C) Densitometry analysis with viral protein band intensities normalised to that of β-actin. (D) Representative Western blot on EV-D68-infected cell lysate treated with GW406108X. β-Actin was detected as the housekeeping control. (E) Fold change for strand-specific qPCR calculated via the delta-delta C_t method. Triplicates were performed with error bars plotted to show the mean and standard deviation. Results were evaluated with one-way ANOVA and Dunnett’s test for statistical significance. ns signifies P > 0.05, ∗P ≤ 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

The antiviral effect of GW406108X was also investigated by assessing changes in the newly synthesised viral protein and viral RNA levels. From Fig. 3C and D, GW406108X reduced EV-D68 VP1 protein levels in a dose-dependent manner, with low levels of protein seen from 5 to 20 μmol/L of GW406108X. To detect changes in the newly synthesised viral RNA levels, qRT-PCR with EV-D68-specific primers for positive- and negative-sense viral RNA strands was performed (Fig. 3E). Results show changes in both positive-sense and negative-sense viral RNA levels from as early as 6 h.p.i. At 12 h.p.i., neither positive-sense nor negative-sense viral RNA strands were detected at 10 μmol/L of GW406108X treatment. At 6 h.p.i., there seems to be a greater decrease in positive-sense viral RNA levels than negative-sense viral genomic RNA levels.

3.6. GW406108X inhibits viral genomic RNA replication

In order to differentiate between the drug action on the viral replication processes viral genomic RNA replication and viral protein translation, replicon constructs with NanoLuc luciferase (NanoLuc) that are viral RNA replication-competent and viral RNA replication-defective were utilised, similar to those that have been established in¹⁰. However, instead of transfecting this as viral RNA, it was sought to establish this system as a DNA-launched replicon using the human cytomegalovirus promoter. This is due to the ease of handling plasmid DNA as compared to viral RNA, and the minimisation of steps required to perform the assay. Using the pcDNA3.1+ backbone, the insert was cloned in via fusion PCR (Fig. 4). The insert sequence had the coding region for the viral structural proteins replaced with that of NanoLuc and was joined together via three PCR fragments. To generate the viral RNA replication-defective replicon, site-directed mutagenesis was performed to delete the final 162 nucleotides of the replication-competent replicon, based on previous mutagenesis studies of the viral 3D protein. Following successful construction of both the viral RNA replication-competent replicon and the viral RNA replication-defective replicon, the NanoLuc replicon assay was carried out to assess the inhibitory activity of GW406108X (Fig. 5A and B). Controls for viral RNA replication (guanidine hydrochloride) and viral protein translation (cycloheximide) were used. These plasmid DNA replicons were transfected directly into H1299 cells for 24 h, after which different compounds were added. After 12 h, the substrate for NanoLuc, furimazine was added and luminescence quantified via a plate reader.

Figure 4.

Schematic of the construction of replication-competent and replication-defective viral RNA replicons. The pcDNA3.1+ vector was used as the cloning backbone. Three DNA fragments were generated by extension PCR and assembled using overlap PCR, followed by amplification through purification PCR to produce the full-length insert. This insert was cloned into the pcDNA3.1+ vector using the HindIII and SwaI restriction sites. In the replication-defective replicon, the red region indicates a partial deletion within the 3D coding sequence, corresponding to the removal of the last 162 nucleotides, which renders the RNA replication function inactive. Primer sequences are provided in the Materials and Methods (Section 2.13). Created with BioRender.com.

Figure 5.

GW406108X likely targets a host factor involved in viral genomic RNA replication. H1299 cells were transfected with either the (A) viral RNA replication-competent or (B) viral RNA replication-defective plasmid DNA at 50 ng/well. At 24 h post-transfection, the transfection mixture was replaced with GW406108X, 0.1% DMSO, guanidine chloride (GuHCl), or cycloheximide (CHX). Luminescence levels were read after 12 h. (C) Cells were infected with EV-D68 (P0) and treated with either the vehicle control (0.1% DMSO or 1–2.5 μmol/L of GW406108X. At 12 h.p.i., the virus was harvested (P1) and used to infected RD cells again at MOI 1, with either vehicle control (0.1% DMSO) GW406108X treatment. In this way, the same virus was serially passaged in the presence of increasing concentrations of GW406108X, for 18 passages. Plaque assays were performed to quantify viral titres. (D) Post-infection assay demonstrating GW406108X sensitivity of passage 18 viral pools. Passage 18 viral pools were subjected to post-infection treatment with GW406108X and compared to wildtype P0 virus. Triplicates were performed with error bars plotted to show the mean and standard deviation. Results were evaluated with multiple unpaired t-tests for statistical significance, comparing the 0.1% DMSO and GW406108X conditions within each passage. ns signifies P > 0.05, ∗P ≤ 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. n.d.: not detected.

Results show that the viral RNA replication-competent EV-D68 produced higher luminescence levels than the viral RNA replication-defective EV-D68, implying that viral RNA replication is occurring as expected. Furthermore, viral RNA replication inhibitor guanidine hydrochloride treatment reduced luminescence levels in cells transfected with the viral RNA replication-competent but not the viral RNA replication-defective construct¹¹. Incubation with protein translation inhibitor cycloheximide resulted in a decrease in luminescence levels in cells transfected with either of these constructs, which is an expected finding¹². Overall, as the inhibition profile of GW406108X resembles that of the viral RNA replication inhibitor guanidine hydrochloride, where decreases in luminescence were seen only in the viral RNA replication-competent replicon, it seems that GW406108X inhibits EV-D68 replication through modulating viral genomic RNA replication.

After establishing the stage(s) in which GW406108X likely exerts its therapeutic effect, it was aimed to elucidate the nature of proteins interacting with GW406108X. Serial passaging of EV-D68 in the presence of 1–2.5 μmol/L of GW406108X was performed. Results show a consistent decrease in viral titres in the drug-treated sample as compared to the vehicle control, all the way till 18 passages (Fig. 5C and D). These findings suggest that GW406108X may potentially target a host factor rather than a viral component.

3.7. The viral inhibitory effect of GW406108X is mediated through ULK1 and ULK2, which are key components of the autophagy pathway

To decipher the host protein target of GW406108X involved in its inhibitory effect in EV-D68 replication began with known targets of GW406108X. To date, literature on GW406108X has demonstrated its role as an inhibitor of Unc-51-like Kinase 1 (ULK1), Kinesin Family Member 15 (KIF15), and Raf1 proto-oncogene, serine/threonine kinase (RAF1)13, 14, 15, 16. To investigate the role of these host proteins in EV-D68 infection, siRNA knockdown experiments of these proteins were conducted. As ULK1 has a close homologue ULK2, the effect of ULK2 was also investigated¹⁷. Both the individual and combined effects of ULK1 and ULK2 will be explored. It was first established that knockdown of these host proteins was not toxic to the cells up to 50 nmol/L (Fig. 6A). Western blotting and densitometry analysis showed successful induction of the knockdown (Fig. 6B and C). After reviewing the toxicity and efficiency of the siRNA knockdown, the effect of reduced host protein levels on EV-D68 replication was investigated. Following reverse transfection with the respective siRNAs, H1299 cells were infected with EV-D68. Mirroring the post-infection assay, supernatants were harvested, and viral titres quantified at 12 h.p.i. It can be observed that knockdown of ULK1 and ULK2 both individually and together resulted in a decrease in viral titres (Fig. 6D). This highlights the importance of these two host proteins in viral replication. Interestingly, when viral titres were measured in cell lysates rather than in supernatants, ULK1/2 knockdown did not lead to significant differences in viral titres compared to the NTC. This suggests that ULK1/2 may primarily play a role in viral egress (Supporting Information Fig. S6). Western blotting to probe for intracellular EV-D68 VP1 protein levels also showed a decrease in viral protein levels in cell lysates with knockdown of ULK2 only and in both ULK1 and ULK2 double knockdown conditions (Fig. 6E–H).

Figure 6.

Evaluation of the toxicity and efficiency of siRNA knockdown of host targets of GW406108X. (A) Cytotoxicity assay when reverse transfection of respective siRNAs was performed at 25 and 50 nmol/L for 48 h. (B) Representative Western blots and (C) densitometry analysis of H1299 cell lysate following reverse transfection of siRNA for 48 h and probed with antibodies for the respective host proteins. Replicate Western blots are in Fig. S7. (D) Supernatant of post-infection assay following siRNA knockdown of host targets of GW406108X. (E) Representative Western blots with densitometry analysis of (F) EV-D68 VP1 protein and (G, H) respective host proteins. Replicate Western blots are in Fig. S8. Triplicates were performed with error bars plotted to show the mean and standard deviation. Results were evaluated with one-way ANOVA and Dunnett’s test for statistical significance. ns signifies P > 0.05, ∗P ≤ 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

3.8. GW406108X reduces autophagy induction in EV-D68-infected cells

As ULK1 and ULK2 are canonically known to be involved in the autophagy pathway, it was hypothesised that GW406108X inhibits the activity of both ULK1 and ULK2, which in turn restricts viral usage of the autophagy pathway to promote its replication¹⁸. Studies have shown that poliovirus exploits this pathway for genome replication, assembly, and egress. For example, induction of autophagy results in the production of autophagosomes which are membrane-bound organelles, which are good compartments for virion maturation and assembly¹⁹. The membrane-bound nature of the organelles also enables these virions to evade host cell immune responses²⁰. It has also been hypothesised that enteroviruses utilise secretory autophagy as a non-lytic virus transmission pathway²¹. GW406108X has already been shown to inhibit autophagy by Zachari and colleagues¹⁵, therefore it was hypothesised that GW406108X would exhibit its antiviral effect via the same mechanism, thereby reducing viral titres and viral protein levels. This phenomenon was first verified through Western blotting with an antibody for microtubule-associated protein 1A/1B-light chain 3 (LC3). Conversion of LC3-I to LC3-II via lipidation is a crucial component of the autophagic process and has been used in many instances as markers for induction of autophagy²². GW406108X reduces LC3-I to LC3-II conversion in a dose-dependent manner, confirming its role in inhibiting autophagy under normal cellular conditions (Fig. 7A and B). With its effect on autophagy established, it was further examined if this effect also applied to virus-infected cells. Expectedly, EV-D68 VP1 protein levels decreased with increasing concentrations of GW406108X (Fig. 7C and D). A negative association between the concentration of GW406108X and the LC3-II/LC3-I ratio was also observed, suggesting that GW406108X treatment reduces autophagy levels in virus-infected cells, which in turn reduces viral titres.

Figure 7.

GW406108X inhibits autophagy under both uninfected and EV-D68-infected conditions. Representative Western blots and densitometry analysis of H1299 cell lysate following GW406108X treatment for 12 h in (A, B) uninfected and (C, D) EV-D68-infected conditions. Triplicates were performed with error bars plotted to show the mean and standard deviation. Results were evaluated with one-way ANOVA and Dunnett’s test for statistical significance. ns signifies P > 0.05, ∗P ≤ 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001.

Autophagic induction was also observed through LC3 puncta formation via immunofluorescence staining. This was first verified with H1299 cells treated with 50 μmol/L of hydroxychloroquine (HCQ), a known autophagy inhibitor²³. HCQ inhibits the fusion of autophagosomes and lysosomes, halting the autophagic process and resulting in the accumulation of autophagosomes in the cell (Fig. 8A). Cells were concurrently stained for double-stranded RNA, which is a viral genome replication intermediate in virus-infected cells. The number of LC3 puncta-positive (FITC-positive) cells were counted and compared with mock-infected and EV-D68-infected cells (Fig. 8B). Following EV-D68 infection at 12 h.p.i., the number of LC3 puncta-positive cells increased to about 30% of the total number of cells, marking the induction of autophagy. On the other hand, about 40% of cells treated with 50 μmol/L HCQ had LC3 puncta formation. With the aim of understanding the effect of GW406108X on autophagy induction, immunofluorescence staining was extended to more timepoints post-infection, and with either 5 μmol/L GW406108X or vehicle control treatment (Fig. 8C and D). Across the timepoints from 6 to 12 h.p.i., there is an increase in the percentages of FITC-positive cells with LC3 puncta formation. Treatment with GW406108X not only reduced the number of cells with dsRNA staining, but also reduced the percentage of cells with LC3 puncta to about 10%, which was consistent throughout the timepoints. Therefore, it is likely that GW406108X prevents autophagosome formation, thereby inhibiting viral replication.

Figure 8.

Immunofluorescence staining of H1299 cells for LC3 puncta formation. (A) Representative images where blue stains for cell nucleus and green stains for LC3. (B) Quantification of the percentage of FITC-positive cells with LC3 puncta formation. HCQ: hydroxychloroquine. Cells were fixed with methanol 12 h.p.i., and cells treated with HCQ were fixed 24 h post-treatment. (C) Representative images where blue stains for cell nucleus, green stains for LC3, and red for dsRNA. (D) Quantification of the percentage of FITC-positive cells with LC3 puncta formation. H1299 cells were infected with EV-D68 at MOI 25, and then treated with either 5 μmol/L GW406108X or 0.1% DMSO. At several timepoints post-infection, cells were fixed in methanol and stained with LC3 antibody. Scale bars represent 40 μmol/L.

Beyond immunofluorescence staining, TEM was also carried out to visualise formation of autophagosomes present in EV-D68-infected cells. TEM images in Fig. 9 show the virus-infected cells displayed extensive cytopathic effects, including the presence of numerous double-membrane vesicles (DMVs), which are characteristic of enterovirus replication complexes (red arrow)²⁴. Viral particles were observed clustered within cytoplasmic vesicles and freely dispersed in the cytosol, consistent with the known morphology of enteroviruses. These findings confirm active EV-D68 replication and provide morphological evidence of enterovirus-induced cellular remodelling. In contrast, cells treated with GW406108X had fewer DMVs, and an observed reduced number of viral particles present. These findings suggest that GW406108X treatment prevents the formation of DMVs in EV-D68-infected cells, thereby restricting viral replication.

Figure 9.

Electron micrographs showing GW406108X treatment reverses the effect of EV-D68 infection in H1299 cells. H1299 cells were infected with EV-D68 and treated with either 0.1% DMSO or 5 μmol/L GW406108X and fixed at 16 h.p.i. Representative images of cells are shown for each condition. The top row images were taken at 2000× magnification, and the bottom row images were taken at 5000× magnification. Red arrows indicate double-membrane vesicles (DMVs). N: nucleus; Cyt: cytoplasm; M: mitochondria. Scale bars are indicated in each image.

3.9. GW406108X demonstrates preclinical efficacy in a mouse model for EV-D68 infection

Prior to in vivo efficacy studies conducted in a murine model for EV-D68 infection, the in vivo toxicity of GW406108X was tested in P4 C57BL/6 neonates²⁵. To our knowledge, GW406108X has not been administered in vivo before, hence, 10 mg/kg was first tested (Fig. 10A–C). GW406108X was given via intraperitoneal injection to P4 pups daily for six days. Subsequently, mice were monitored for signs of toxicity for the next 11 days (Supporting Information Table S3). Results show that 10 mg/kg of GW406108X has no observable toxic effects on mice, the mice continued to gain weight at the same rate as those treated with the vehicle control (Fig. 10B) and did not show any clinical symptoms (Fig. 10C). With this favourable result, the in vivo efficacy of GW406108X was investigated in a preliminary study with 5 and 10 mg/kg of GW406108X. A mouse-adapted strain of EV-D68 US/MO/14-8949 was used, where P4 pups were first treated with either drug or the vehicle control (DMSO) via intraperitoneal injection. Two hours later, the pups were intranasally inoculated with the virus. For the next five days, the mice were monitored and dosed daily. From Days 6–17 post-infection, mice were monitored for clinical symptoms for infection. From the Kaplan–Meier survival curve, the vehicle control group had an 83.3% mortality rate (Fig. 10D). Mice in this group showed signs of infection such as rapid breathing and reduced activity beginning from the first day post-infection. Pups in this group also did not gain as much weight as the naïve uninfected control, resulting in high clinical scores reaching the levels for euthanasia (Fig. 10E and F). In contrast, mice in the 5 and 10 mg/kg GW406108X treatment groups did show signs of infection as well, however these symptoms were on average less severe than in the vehicle control group. Lastly, the lung viral titration demonstrated that the GW406108X-treated group had about 80% reduction in viral titres than the vehicle control group (Fig. 10G). Altogether, this preliminary study demonstrates the potential for GW406108X for further in vivo studies.

Figure 10.

GW406108X toxicity and efficacy studies in P4 suckling C57BL/6 mice. For toxicity studies, P4 suckling C57BL/6 mice were treated with a dose of 10 mg/kg of GW406108X once a day for six days, and the mice monitored daily for up to 17 days post-treatment. The following results are shown (A) Kaplan–Meier survival curve, (B) body weight, and (C) physiological scoring based on five different criteria: activity, breathing, movement, body weight loss, and dehydration. For efficacy studies, P4 suckling C57BL/6 mice were pre-treated with a dose of GW406108X and intranasally inoculated with 1 × 10⁷ PFU of EV-D68 2 h post-treatment. 5 or 10 mg/kg GW406108X or the vehicle control DMSO was administered via intraperitoneal injection for the following five days, and the mice monitored daily for up to 17 days post-treatment. The following results are shown (D) Kaplan–Meier survival curve, (E) body weight, and (F) physiological scoring based on the same five criteria mentioned in Table S1. (G) Normalised EV-D68 viral titres in the lung were evaluated with multiple unpaired t-test for statistical significance. ns signifies P > 0.05, ∗P ≤ 0.05, ∗∗P < 0.01, ∗∗∗P < 0.001. n = X which refers to the number of mice in the treatment group.

4. Discussion

Evaluation of multiple cell lines for their susceptibility to EV-D68 infection showed that H1299, a non-small cell lung cancer cell line, supported viral replication in a similar fashion to RD cells, which is the common choice of cell line for EV-D68 studies (Table 1). GW406108X was found to be a potent compound from the phenotypic screen for inhibitors of EV-D68 replication (EC₅₀: 1.804 μmol/L; CC₅₀: >100 μmol/L, Fig. 2C). Testing of GW406108X on other species of enterovirus showed that the mechanism(s) by which GW406108X acts is not specific to EV-D68, but it extends to Enterovirus A–C, Rhinovirus, and even to DENV-2 (Table 3). Its inhibition of EV-D68 infection is also not cell line-specific, as observed by the inhibition seen in H1299, A549, and RD cells. Furthermore, GW406108X impeded EV-D68 replication in hNEpCs, confirming that the inhibitory effect of GW406108X is not restricted to immortalised cell lines but also to primary cells. Of note, GW406108X did not reduce CHIKV replication, hinting at a potentially different mechanism of CHIKV replication compared to the Enteroviruses and Orthoflaviviruses. This is a plausible explanation as CHIKV belongs to the genus Alphavirus, which differs from the other tested viruses. This could be related to differences in a specific viral protein targeted by GW406108X, or different host protein utilisation by different viruses.

As GW406108X was identified from a post-infection phenotypic screen, it was not unexpected that this compound had no effect when it was added prior to virus infection (Supporting Information Fig. S7). It therefore also follows that the possible therapeutic window of GW406108X to be in the later stages of infection as observed from the TOA and TOR assays (4–6 h.p.i.) (Fig. 3A). It is also logical that bypassing the entry stages of viral infection would not abolish the inhibitory effect of GW406108X, as observed in comparison with pleconaril, an entry inhibitor. Nevertheless, further studies evaluating the chemical optimization, pharmacological properties, and head-to-head efficacy against clinically relevant antivirals will be necessary to define the translational potential of GW406108X. The duration of action of GW406108X seems to be 4–6 h.p.i., which would approximately correspond to the viral replication stages of viral genomic RNA replication, viral protein translation, virion maturation, and virion release. The inhibition of viral protein synthesis can be seen from the Western blot assay results. When total RNA was harvested from cell lysate at different timepoint post-infection, qRT-PCR showed decreases in both positive-sense and negative-sense viral RNA levels from as early as 6 h.p.i (Fig. 3E). To decouple the effects of viral genomic RNA replication and viral protein translation, the NanoLuc replicon system was established, with constructs that are viral RNA replication-competent and -defective (Fig. 4). With the necessary controls guanidine hydrochloride (viral genomic RNA replication inhibitor) and cycloheximide (protein translation inhibitor) producing the results as expected, confidence can be given to the results from the NanoLuc replicon assay. The findings from this assay indicate that GW406108X likely inhibits viral genomic RNA replication, as no significant inhibitory effect was observed in the viral RNA replication-defective replicon (Fig. 5A and B). Efforts to seek out the host protein target of GW406108X began with serial passaging of EV-D68 in the presence of this drug, with the same observation made across 18 passages, which meant that the plausible target of GW406108X is a host protein native to the cell, where the possibility of resistance emergence is much lower (Fig. 5C). However, as viral whole-genome sequencing was not performed, the possibility of subtle or non-phenotypic mutations cannot be entirely excluded. Future studies incorporating comprehensive sequencing will be important to fully evaluate the resistance potential of EV-D68 under GW406108X selection pressure.

The search for this host protein began with known literature targets of GW406108X, which are ULK1, KIF15, and RAF1. The role of these three proteins and ULK2 (analogue of ULK1) in EV-D68 infection was investigated through siRNA knockdown studies. Firstly, the knockdown toxicities and efficiencies were tested, where it was found that the knockdown had little effect on cell viabilities (Fig. 6A). The efficiency of the knockdown was verified through Western blotting. ULK1 and ULK2 knockdown efficiencies were low, which could be due to antibody promiscuity for other ULK analogues (Fig. 6B and C)¹⁷. These ULK analogues were excluded from the study due to greater differences with ULK1²⁶.

Cells with knockdown of these targets were subsequently infected with EV-D68, and the cell lysates harvested for viral titre quantification and Western blotting. Results show significant decreases in viral titres in cells with reduced levels of ULK1, ULK2, or both ULK1 and ULK2 (Fig. 6D). It can also be observed that the decrease in viral titre was greater in the ULK2 only condition compared to ULK1, suggesting that the knockdown of ULK2 to be the main contributor to the decrease in viral titre in the ULK1 and ULK2 condition. A similar phenomenon was seen in the Western blotting results for intracellular viral protein levels, with decreases in EV-D68 VP1 protein levels observed in the ULK2 only and the ULK1 and ULK2 conditions (Fig. 6E and F). Dahmane and colleagues¹⁹ reported that ULK1 inhibition in poliovirus-infected cells resulted in the accumulation of intracellular viral arrays, attributed to a defect in virion egress. Consistent with this, knockdown of ULK1/2 in the present study led to significantly reduced viral titres in the supernatant (Fig. 6D), suggesting impaired virion release. Notably, this reduction was not observed following freeze–thaw cycles, which facilitate the release of intracellular virions (Supporting Information Figs. S8 and S9), indicating that viral replication and assembly remained intact. These findings corroborate previous reports implicating ULK1 as a proviral factor that facilitates efficient release of enterovirus particles^27,28.

Although ULK1/2 clearly emerged as the functionally relevant targets in the context of EV-D68 infection, it is important to acknowledge that GW406108X has reported activity against other kinases, including KIF15 and RAF1. While the structural basis of GW406108X binding to ULK1 has been extensively characterized by Zachari et al.¹⁵, additional work will be valuable to comprehensively define its selectivity and target engagement during viral infection. Kinome-wide profiling, cellular thermal-shift assays, and assessment of off-target kinase effects would help clarify the specificity of ULK1/2 inhibition and strengthen the mechanistic link between ULK1 blockade and antiviral activity. Such studies, together with expanded pharmacodynamic and safety evaluations, represent important future directions that will guide the development of GW406108X or next-generation ULK-targeting derivatives as candidate antiviral therapeutics.

Many positive-strand RNA viruses have been shown to hijack the autophagy pathway to promote its own replication, and this has been the case for picornaviruses as well²⁸. Furthermore, the direct interaction between GW406108X and ULK1 has been established by Zachari and colleagues¹⁵, and the role of GW406108X in autophagy inhibition has been studied. Following the identification of ULK1/2 as the targets of GW406108X, the downstream mechanism by which GW406108X inhibits ULK1/2 in the context of enterovirus replication was explored. This was mainly achieved through monitoring LC3-I and LC3-II levels via Western blot and immunofluorescence assays, and TEM imaging. When autophagy is induced, conversion of LC3-I to LC3-II occurs through conjugation of cytosolic LC3-I to phosphatidylethanolamine, leading to its accumulation and association with autophagosomes and autolysosomes. An increase in the LC3-II relative to LC3-I levels is also a signal for autophagy induction. In uninfected cells, GW406108X was shown to reduce the ratios of LC3-II/LC3-I in a dose-dependent manner, confirming its role as an autophagy inhibitor (Fig. 7). In virus-infected cells, the same trend was seen, as well as a decrease in EV-D68 viral protein levels. Hence, an association between GW406108X treatment and reduction in autophagy levels during EV-D68 can be established. Other than quantifying the LC3-II/LC3-I ratios, accumulation of LC3-II in autophagosomes and autolysosomes gives rise to puncta formation, which can be quantified and is commonly used in assessing autophagy levels in cells²⁹. However, LC3 levels alone do not distinguish between increased autophagosome formation and impaired autophagosome turnover. This distinction is particularly important in the context of ULK1 inhibition versus lysosomal blockade. ULK1 inhibition, as expected for GW406108X, would be anticipated to suppress autophagosome initiation and thereby reduce LC3 puncta formation, whereas HCQ allows autophagosomes to form but prevents their maturation, typically resulting in puncta accumulation. From the immunofluorescence staining images, increased puncta formation was detected in cells infected with EV-D68 (as identified by the dsRNA stain in Fig. 8). The percentage of cells with LC3 puncta formation increased across the infection duration. Addition of 5 μmol/L of GW406108X reduced this effect, where a reduced number of cells with LC3 puncta formation was observed. The observation was further verified via TEM imaging, where 5 μmol/L GW406108X treatment reduced the formation of DMVs and viral particles in virus-infected cells (Fig. 9). It therefore appears to be that autophagy is induced during viral infection for the virus to promote its own replication, and GW406108X acts to reverse this effect. A diagram for the proposed mechanism of action of GW406108X in inhibiting enterovirus replication is shown in Fig. 11. During enterovirus infection, the virus hijacks the autophagy pathway to promote its own replication via two ways 1) genome replication organelle formation and 2) non-lytic mature virion release. GW406108X, a ULK1/2 inhibitor, likely acts to reverse this effect by preventing inhibiting formation of the ULK complex, a key component required for the initiation of autophagosome formation. In this way, autophagy is inhibited, thereby preventing virus replication.

Figure 11.

Diagram on the proposed mechanism through which GW406108X inhibits enterovirus replication. During enterovirus infection, the virus hijacks the autophagy pathway to promote its own replication via two ways 1) genome replication organelle formation and 2) non-lytic mature virion release. GW406108X, a ULK1/2 inhibitor, likely acts to reverse this effect by preventing inhibiting formation of the ULK complex, a key component required for the initiation of autophagosome formation. In this way, autophagy is inhibited, thereby preventing virus replication. Created with BioRender.com.

As the inhibitory profile of GW406108X showed a 3log₁₀ PFU/mL reduction in viral titres, it is also reasonable to expect this compound to be multi-targeting, with the combined effect of all targets resulting in this large decrease in viral titres. Follow-up studies deciphering the other possible targets of GW406108X can be performed, with the hope of refining our understanding of the enterovirus replication cycle.

Finally, the in vivo efficacy of GW406108X was also observed in a pilot study conducted with a mouse-adapted strain of EV-D68. Consistent with previous studies, including the original EV-D68 mouse-adapted model, our experiments were conducted in neonatal or immune-compromised mice, as immunologically mature small-animal models that reliably support EV-D68 infection are currently not available. GW406108X has not been tested in any in vivo murine models, hence we took the noteworthy approach to provide initial results for the toxicity of GW406108X in mice. In this study, P4 C57BL/6 neonates were intranasally inoculated with EV-D68 and dosed with 5 or 10 mg/kg of GW406108X every day up to five days post-infection, after a toxicity study showed that doses up to 10 mg/kg are non-toxic (Fig. 10A–C). Results from this preliminary study reveal a 100% survival rate, reduced clinical signs of infection, and reduced lung viral titres in the drug-treated group, demonstrating the potential of GW406108X to be used as an anti-enteroviral compound in the future (Fig. 10D–G). However, as this pilot study was performed in P4 neonatal mice with immature immune systems, future work in adult or more immunologically representative animal models will be essential to better evaluate the translational potential of GW406108X. In addition, while acute toxicity was assessed, comprehensive pharmacokinetic analyses, chronic toxicity studies, and immune safety evaluations will be necessary to fully characterize the safety and therapeutic potential of GW406108X. These findings demonstrate the potential of GW406108X to be used as an anti-enteroviral compound in the future.

5. Conclusions

In conclusion, this study not only expands the repertoire of cell lines suitable for EV-D68 research but also identifies the H1299 cell line as a novel and effective model for enterovirus infection, exhibiting comparable viral growth kinetics to RD cells. Using both cell lines, we conducted a high-throughput phenotypic screen that led to the discovery of GW406108X as a previously uncharacterized antiviral compound. While GW406108X was previously known as a ULK1/2 kinase inhibitor, this work is the first to reveal its antiviral activity against enteroviruses and to demonstrate its preclinical efficacy and safety in an animal model. Mechanistic studies further suggest that GW406108X likely inhibits EV-D68 replication by interfering with virus-mediated hijacking of the autophagy pathway through ULK1/2 inhibition. The data suggest that interference with the autophagy pathway may impair both genome replication organelle formation and non-lytic release, processes known to be essential for efficient enterovirus replication. Together, these findings establish GW406108X as a promising antiviral lead and provide new experimental tools and mechanistic insights that advance the pursuit of therapeutics against enterovirus infections.

Author contributions

Yuhui Deborah Fong: Conceptualization, Methodology, Investigation, Visualisation, Project administration, Supervision, Writing – original draft, Writing – review and editing. Thinesshwary Yogarajah: Conceptualization, Methodology, Funding acquisition, Project administration, Supervision, Writing – review and editing. Justin Jang Hann Chu: Conceptualization, Visualisation, Funding acquisition, Project administration, Supervision, Writing – review and editing. Yi Shan Chan: Investigation. Jasmaadiyah Binte Habib Mohameed: Investigation, Writing – review and editing. Xuan Wei Khoo: Investigation, Visualisation. Angeline Neo: Investigation. Bowen Yi: Investigation.

Conflicts of interest

The authors declare no conflicts of interest.

Appendix A. Supporting information

The following is the Supporting Information to this article: