Polymerase-inhibitor drug synergy and mutational signatures in different epithelial cell models of RSVA and hPIV3 infection

Abstract

Despite the huge global health burden presented by respiratory viruses, effective broad-spectrum antiviral therapeutic options remain limited. Here we evaluated the antiviral activity of four RNA-dependent RNA polymerase (RdRp) inhibitors, remdesivir, ribavirin, favipiravir, and molnupiravir, as monotherapy or dual-drug combinations against respiratory syncytial virus (subtype A, RSVA) and human parainfluenza (serotype 3, hPIV3) using epithelial cell lines and primary human airway culture models. Remdesivir showed the greatest potency across both viruses, while ribavirin and favipiravir also demonstrated inhibition. Molnupiravir was active against RSVA but not hPIV3. Several dual-drug combinations, including remdesivir–favipiravir, remdesivir–molnupiravir and favipiravir–molnupiravir, produced marked synergy against RSVA, and more limited synergy for hPIV3. Antiviral efficacy was validated in primary airway epithelial cultures, where effective concentrations preserved epithelial integrity and attenuated viral disruption of ciliary function. Across both viruses, increasing antiviral exposure was associated with dose-dependent signature mutagenesis. Antivirals induced significantly higher RSVA mutation burden in the primary airway model. These findings highlight the therapeutic potential of RdRp inhibitor combinations for RSVA and hPIV3, provide mechanistic insight through antiviral-related mutational signatures, and demonstrate advantages of the primary human airway culture model for development of effective multi-drug regimens and broad-spectrum antiviral preparedness.

Introduction

Respiratory viral infections, especially RNA-viruses, are a major cause of global morbidity and mortality, underscoring the need for effective antiviral strategies. Respiratory syncytial virus (RSV) and human parainfluenza (hPIV) are leading causes of lower respiratory tract infection and viral pneumonia^1–3, particularly in infants, the elderly, and immunocompromised patients^4,5. RSV alone infects approximately 33.8 million children below the age of five annually, while hPIV contributes substantially to paediatric hospital admissions⁶. More broadly, respiratory RNA-viruses have driven major outbreaks and pandemics.

Current post-exposure treatments rely on symptom management, and there remains an unmet need for quick and effective therapies, especially in severe cases such as respiratory viral infections in immunocompromised patients. One promising approach is targeting the RNA-dependent RNA polymerase (RdRp) protein. The RdRp is a highly conserved enzyme across RNA virus species that is essential for viral replication⁷. Interest in RdRp inhibitors increased during the COVID-19 pandemic with efforts to repurpose existing nucleoside analogues^8,9. Ribavirin, originally developed for hepatitis C virus, is licensed for RSV and shows in vitro activity against hPIV, but clinical efficacy is inconsistent and limited by toxicity, cost and safety concerns^10–17. Other RdRp inhibitors such as remdesivir, favipiravir and molnupiravir (active metabolite EIDD-1931) were initially developed for influenza and severe acute respiratory syndrome (SARS)^7,18. All three have subsequently been investigated for potential efficacy against other RNA viruses, most notably being repurposed for SARS-CoV-2 during the COVID-19 pandemic^18–20.

Alternatively, drug combination may enhance antiviral efficacy and tolerability through additive or ideally synergistic inhibitory effects²¹ and has been highly successful in other viral diseases, most notably HIV²². Within the field of respiratory RNA viruses, combination therapy has been suggested as an important approach for improving influenza treatments, particularly for mitigating the emergence of resistance²³. In recent years, the use of multiple antivirals has been investigated for SARS-CoV-2, with many combinations demonstrating synergy in vitro and in vivo^9,24,25. Broad-spectrum combinations with differing viral susceptibilities may be particularly advantageous for vulnerable populations such as immunocompromised patients²⁶. Robust evaluation of such strategies, however, requires experimental systems that reliably capture both antiviral activity and host epithelial responses across multiple respiratory viruses.

Most antiviral studies rely on immortalised epithelial cell lines as an infection model, which offer practical advantages such as ease of culture, reproducibility and suitability for higher-throughput screening. However, these models incompletely represent the complexity of the human airway epithelium, and different respiratory viruses frequently require distinct cell lines to achieve efficient replication or quantifiable cytopathology, limiting standardisation and complicating direct comparisons of antiviral efficacy across viral species²⁷.

Primary human airway culture models provide a more physiologically relevant approach. When differentiated at an air–liquid interface, these cultures recapitulate the pseudostratified architecture of the human airway, incorporating ciliated and secretory cell populations, functional tight junctions and mucus secretion. Importantly, this model supports infection by multiple respiratory viruses within a unified experimental model²⁸, enabling direct evaluation of antiviral efficacy, epithelial toxicity and virus-induced functional disruption, such as impaired ciliary beating.

Here, we address these gaps by systematically evaluating the in vitro efficacy of selected RdRp inhibitors, both as monotherapies and in combination, against RSV and hPIV. Using conventional epithelial cell lines alongside primary human airway epithelial cultures, we aim to define antiviral potency, synergistic interactions and epithelial outcomes in models that better reflect human respiratory infection, thereby informing the development of effective and broadly applicable antiviral strategies.

Results

In vitro efficacy of remdesivir, favipiravir, molnupiravir or ribavirin as monotherapies against RSVA and hPIV3

We first optimised the infection assays for RSVA and hPIV3 to achieve sufficient infection progression to investigate antiviral efficacy. Infection of both viruses was assessed for three epithelial cell lines (VeroE6, Calu3 and LLC-MK2) using two readouts: endpoint cytopathic effects (CPE) by crystal violet staining of surviving cells, and fluorescence intensity generated by GFP-tagged viral replication.

For RSVA, only VeroE6 cells supported measurable infection by both CPE and GFP signals. For hPIV3, LLC-MK2 cells were permissive to infection, whereas Calu3 and VeroE6 cells demonstrated little to no detectable infection, even at higher multiplicity of infection (MOI). An MOI of 0.02 and a 7-day incubation were selected as optimal for the subsequent antiviral assays, using VeroE6 for RSVA and LLC-MK2 cells for hPIV3 assays [Fig. 1A]. The results for other timepoints and MOIs for all cell line and virus combinations are summarised in Supplementary Fig. 1.

Fig. 1. Assay schematic and monotherapy antiviral efficacy.

A Comparison of cell toxicity versus viral-GFP intensity (relative fluorescence units) at 7 days post-infection with RSVA or hPIV3 (both at MOI 0.02) with the epithelial cell lines VeroE6, Calu-3 or LLC-MK2 (n = 3). B, C. Representative images of the antiviral infection assay in 96-well format demonstrating dynamic range, here showing RSVA infection of VeroE6 cells with and without single or dual-drug antiviral treatments. Cytopathology is visualised by endpoint crystal violet staining of adherent cells (B), whilst the fluorescence micrograph of the same representative assay plate (C) demonstrates visualisation of GFP-tagged RSVA infection. D Comparison of viral infection quantification using cell viability (quantified by crystal violet absorbance) and viral GFP fluorescence, as normalised against virus-only and mock controls. Representative dose response curves are modelled on data for remdesivir against RSVA. E–L Dose response curves for monotherapy concentrations of remdesivir (E&I), favipiravir (F&J), EIDD-1931 (G&K), and ribavirin (H&L) tested against RSVA (E-H) or hPIV3 (I-L) infection. The viral inhibition % model is shown as the blue line, while drug-only endpoint cytotoxicity (as a % reduction of adherent cells) is indicated by the red line. The shaded box indicates drug concentrations with cytotoxicity exceeding 20%. (n = 5 for RSVA, n = 3-4 for hPIV3).

To determine the monotherapy efficacy of these antivirals, infection assays were performed using serially diluted drug concentrations selected to bracket clinically achievable values: remdesivir²⁹ (0.06-32 mg/L; 0.1–53.1 μM), favipiravir³⁰ (0.5–256 mg/L; 3.2–1629 μM), EIDD-1931³¹ (0.02-16 mg/L; 0.08–61.7 μM), and ribavirin³² (0.06-256 mg/L; 0.2–1048 μM). Endpoint CPE quantification at day 7 [Fig. 1B] was used to calculate relative viral inhibition versus virus-only and mock controls, from which data a dose-response curve was modelled to determine EC₅₀/EC₉₀ values for each drug-virus interaction [Table 1]. Wells were also scanned for GFP fluorescence intensity, with drug-induced viral inhibition reflected in reduced fluorescence intensity [Fig. 1C&D].

Table 1.

Summary of monotherapy antiviral efficacy

Virus	Drug	EC50 [95% CI]		EC90 [95% CI]
		mg/L	μM	mg/L	μM
RSVA	Remdesivir	3.6 [2.8-4.4]	6.0 [4.6-7.3]	8.5 [6.7-10.3]	14.1 [11.1-17.1]
	Favipiravir	15.2 [9.9-20.6]	96.8 [63.0-131.1]	136.9 [88.2-176.5]	871.4 [561.4-1123.5]
	EIDD-1931	0.8 [0.6-0.9]	3.1 [2.3-3.5]	2.1 [1.7-2.6]	8.1 [6.6-10.0]
	Ribavirin	6.1 [3.3-8.9]	25.0 [13.5-36.4]	54.7 [30.2-76.1]	224.0 [123.7-311.6]
hPIV3	Remdesivir	1.1 [-3.3-5.5]	1.8 [-5.5-9.1]	1.3 [1.3-1.4]	2.2 [2.2-2.3]
	Favipiravir	80.2 [77.4-83.1]	510.5 [492.7-528.9]	159.2 [153.7-164.7]	1013.3 [978.3-1048.34]
	EIDD-1931	NA*	NA*	NA*	NA*
	Ribavirin	5.6 [4.9-6.3]	22.9 [20.1-25.8]	8.1 [7.7-8.5]	33.2 [31.5-34.8]

*EIDD-1931 cytotoxicity occurred in LLC-MK2 cells above 2 mg/L (8 μM), before 50% viral inhibition reached.

All four RdRp inhibitors achieved 100% viral inhibition against RSVA within the concentration range tested [Fig. 1E–H]. For hPIV3, remdesivir, favipiravir and ribavirin all demonstrated effective in vitro viral inhibition [Fig. 1I–L]. Remdesivir demonstrated the overall highest potency against both viruses, with EC₅₀ observed at 1.1 mg/L (1.8 μM) and 3.6 mg/L (6.0 μM) against hPIV3 and RSVA, respectively. Additionally, no cytotoxicity was observed throughout the clinically relevant concentrations of remdesivir tested on either cell line [Fig. 1E, I]. Ribavirin also demonstrated consistent viral inhibition against RSVA and hPIV3 with EC₅₀ values of 6.1 mg/L (25.0 μM) and 5.6 mg/L (22.9 μM), respectively. Ribavirin cytotoxicity was only detected at higher doses, from 128 mg/L (524 μM) for VeroE6 and 32 mg/L (131 μM) for LLC-MK2 [Fig. 1H, L].

Favipiravir required substantially higher concentrations to achieve antiviral effects, with EC₅₀ of 15.2 mg/L (96.8 μM) and 80.2 mg/L (510 μM) for RSVA and hPIV3, respectively [Fig. 1F, J]. For both viruses, significant endpoint cytotoxicity was observed at or below the predicted EC₉₀ doses of favipiravir [Fig. 1 & Table 1]. EIDD-1931 demonstrated the lowest EC₅₀ against RSVA, 0.8 mg/L (3.1 μM) [Fig. 1G], but an EC₅₀ could not be determined for hPIV3 at the concentrations tested. In LLC-MK2 cells, EIDD-1931 caused increasing endpoint cytotoxicity from 2 mg/L (7.7 μM) to the maximum tested dose of 8 mg/L (31 μM), with no inhibition of hPIV3 infection evident below this concentration [Fig. 1K].

Whilst Fig. 1E–L shows endpoint cytotoxicity for the highest tested doses of EIDD-1931, favipiravir, and ribavirin after 7 days, further cell viability and lactate dehydrogenase (LDH) release assays did not demonstrate any early drug-induced toxicity for either VeroE6 or LLC-MK2 cells after 24 h exposure [Supplementary Fig. 2A, B]. Endpoint cytotoxicity in the cell lines did interfere with the use of viral-GFP fluorescence intensity for effective dose modelling. Using hPIV3 data as an example, relative viral inhibition as calculated by reduction in GFP resulted in a EC₅₀ of 1.9 mg/L (3.2 μM) for remdesivir and 11.3 mg/L (46.3 μM) for ribavirin. However, where cytotoxicity occurred at concentrations below complete viral inhibition for favipiravir and molnupiravir, any reduction in viral-GFP intensity due to drug inhibition cannot be separated from the confounding effect of drug-induced host cell death [Supplementary Fig. 3].

Synergistic dual-drug combinations of remdesivir, favipiravir, molnupiravir and ribavirin against RSVA and hPIV3

We next used our cell-line antiviral assay to evaluate dual-drug combinations of remdesivir, favipiravir, molnupiravir and ribavirin. For each pairwise combination, 4-6 concentrations per drug were selected based on the monotherapy dose-response ranges [Fig. 2, 3]. Concentrations were selected to span the full dynamic range of each single-drug dose response range whilst remaining below the prior cytotoxicity thresholds in these cells.

Fig. 2. RSVA antiviral combinations.

Dual-drug combinations of antivirals were assayed for RSVA inhibitory dose response and cytotoxicity (without virus) using VeroE6 cells. For each panel (A–F), the first row shows a dose response grid representing relative viral inhibition % and the second row shows a drug-only cytotoxicity response grid indicating cell viability %. Also shown are 3D contour plots of synergy interaction scores as calculated by the HSA model (row 3) and cytotoxicity (row 4) at the corresponding antiviral concentrations. (n = 3 per combination) Favi = favipiravir, Rem = remdesivir, EIDD = EIDD-1931 (molnupiravir), Riba = ribavirin.

Fig. 3. hPIV3 antiviral combinations.

Dual-drug combinations of antivirals were assayed for hPIV3 inhibitory dose response and cytotoxicity (without virus) using LLC-MK2 cells. For each panel (A–F), the first row shows a dose response grid representing relative viral inhibition % and the second row shows a drug-only cytotoxicity response grid indicating cell viability %. Also shown are 3D contour plots of synergy interaction scores as calculated by the HSA model (row 3) and cytotoxicity (row 4) at the corresponding antiviral concentrations. (n = 3 per combination) Favi = favipiravir, Rem = remdesivir, EIDD = EIDD-1931 (molnupiravir), Riba = ribavirin.

Remdesivir exhibited synergy with each of the other RdRp inhibitors against RSVA [Fig 2A, C, D] and with all but molnupiravir (which demonstrated no efficacy in monotherapy) against hPIV3 [Fig. 3A, C, D]. Synergistic effects were observed at concentrations as low as 0.5 mg/L (0.8 μM) for remdesivir against both viruses. Such combinations were dose sparing, with remdesivir alone having an EC₉₀ of 8.5 mg/L (14.1 μM) against RSVA yet reached a viral inhibition of 97.6% at 0.5 mg/L (0.8 μM) when combined with 0.5 mg/L (1.9 μM) molnupiravir (also below the latter drug’s EC₅₀ of 0.8 mg/L (3.1 μM)). Importantly, these synergistic concentrations were not associated with cytotoxicity.

Favipiravir also demonstrated synergy at substantially reduced concentrations when combined with remdesivir or molnupiravir against RSVA [Fig 2A, B]. While its monotherapy EC₅₀ was 15.2 mg/L (96.8 μM), interaction scores reached +62.1 and +60.8 (HSA model) when 8 mg/L (50.9 μM) favipiravir was combined with 0.06 mg/L (0.2 μM) and 0.5 mg/L (1.9 μM) molnupiravir, respectively.

Overall, synergy was less pronounced against hPIV3, with dose responses primarily driven by the higher potency of remdesivir and ribavirin for this virus. The most productive combinations were 0.5 mg/L (0.8 μM) remdesivir with either 64 mg/L (407 μM) favipiravir (HSA model score +41.7) or 4 mg/L (16.4 μM) ribavirin (HSA + 38.3) [Fig. 3A, D].

Across these dual-drug combinations for both RSVA and hPIV3 there was minimal evidence of antagonism. Instances of negative HSA scores were mostly attributable to cytotoxicity at high concentrations, particularly the upper molnupiravir concentrations used in combinations (8 mg/L (30.9 μM) for VeroE6 and 2 mg/L (7.7 μM) for LLC-MK2 cells) [Fig. 2, 3]. A summary of the most synergistic antiviral combinations is provided in Table 2, with interaction scores calculated using the HSA model, as well as alternate Bliss, Loewe and ZIP models.

Table 2.

Selected combinations of antivirals with greatest synergy in epithelial cell infection assays

Virus	Drug A			Drug B			Viral Inhibition	Interaction Scores
	Drug	mg/L	μM	Drug	mg/L	μM	(%)	HSA	Bliss	Loewe	ZIP
RSVA	Favipiravir	8	50.9	EIDD-1931	0.06	0.2	82.3	+62.1	+61.8	+58.6	+55.8
RSVA	Favipiravir	8	50.9	EIDD-1931	0.5	1.9	92.7	+60.8	+46.3	+34.2	+29.2
RSVA	Remdesivir	0.5	0.8	EIDD-1931	0.5	1.9	97.6	+60.4	+46.0	+37.4	+28.0
RSVA	Remdesivir	0.5	0.8	EIDD-1931	0.06	0.2	52.0	+34.1	+33.7	+32.5	+34.7
RSVA	Remdesivir	0.5	0.8	EIDD-1931	1	3.9	93.0	+17.9	+11.5	+16.8	+4.6
RSVA	Remdesivir	0.5	0.8	Favipiravir	8	50.9	78.5	+42.0	+27.9	+29.1	+26.2
RSVA	Remdesivir	4	6.6	Favipiravir	1	6.4	93.7	+19.3	+15.0	+17.3	+11.7
RSVA	Remdesivir	0.5	0.8	Favipiravir	1	6.4	43.6	+21.2	+9.3	+20.8	+16.8
hPIV3	Remdesivir	0.5	0.8	Favipiravir	64	407	76.3	+41.7	+41.0	+33.3	+41.0
hPIV3	Remdesivir	0.5	0.8	Ribavirin	4	16.4	48.5	+38.3	+37.4	-27.5	+37.4

Benchmarking RSVA and hPIV3 infection in a physiologically relevant human airway epithelial model

To benchmark antiviral responses in a more physiologically relevant system, we next performed infection assays using differentiated human airway epithelial cells cultured at air-liquid interface (ALI) [Fig. 4A]. Cultures were maintained for at least four weeks to achieve full mucociliary differentiation, confirmed by transepithelial electrical resistance before experiments (TEER; >300 Ω·cm² threshold), and experimental endpoint immunofluorescent staining showing intact tight junctions and the presence of mucous-secreting and ciliated epithelial cell types [Fig. 4B, C].

Fig. 4. Benchmarking RSVA and hPIV3 infection of differentiated human airway epithelial cell cultures.

A Graphical representation of methodology for culture of primary airway epithelial cell model at air-liquid interface (ALI) and infection with RSVA or hPIV3 (Created in BioRender. Woodall, M. (2026) https://BioRender.com/0d9xwg6). B, C Representative confocal micrographs of primary airway epithelial cells in ALI culture. Cultures were infected with RSVA, hPIV3, or a mock control and fixed for immunofluorescent staining and imaging at 7 days post infection (dpi). Channels represent signal for DAPI staining of cell nuclei (blue), anti-GFP for the GFP-tagged virus (green), phalloidin for tight junctions (white), alpha-tubulin for ciliated epithelial cells (violet), and MUC5AC to indicate mucus-secretory cells (yellow). Orthogonal views (C) are also shown for representative cultures, with the same immunofluorescent staining. D–G Cilia-beat frequency (CBF) was analysed using high speed video microscopy at 7 dpi with RSVA (D&F) or hPIV3 (E&G) versus mock. Distribution of measured CBF is shown by histogram and box plot (D, E) for each virus. Representative regions of interest (ROI) from video analysis indicate the relative abundance of detectable cilia movement in mocks versus RSVA (F) and hPIV3 (G) infected cultures. H Representative well scans showing brightfield and GFP signal at 7 dpi for mock and infection conditions with each virus using cell lines (VeroE6 or LLC-MK2 in 96-well format) compared to primary ALI cultures (24-well format). I The Ct result from qRT-PCR performed on supernatant (for cell lines) or apical wash (for ALI) samples used for viral genomic sequencing, indicating respective viral load at 7 dpi. Sequencing results for total mutations (J), and the proportion of mutations which were observed in both models or unique (K). Representative images and functional analyses are shown from triplicate ALI cultures per experimental condition. (***P < 0.001).

Both RSVA and hPIV3 established productive infection in the ALI cultures, as determined by GFP fluorescence and microscopy [Fig. 4B, C]. Despite using the same MOI and time course of infection, RSVA spread across larger, contiguous regions of the epithelium, whereas hPIV3 remained largely confined to individual infected cells [Fig. 4B, C, H]. This observed difference was quantified by image analysis of GFP-positive well area [Supplementary Fig.4A] and aligned with relative levels of viral RNA as measured in apical surface washes by qRT-PCR [Fig. 4I]. Whole well-scanning fluorescence imaging further illustrated the greater intensity of RSVA replication in ALI cultures compared with mammalian cell lines [Fig. 4H]. The brightfield scans confirmed preservation of epithelial integrity at 7 days post-infection (dpi), a timepoint at which pronounced cytopathology and cell loss are already evident in cell line models [Fig. 4H].

High speed video microscopy was used to evaluate ciliary function and the impact of infection [Fig. 4D–G]. Mock-infected cultures displayed abundant, rapidly beating cilia, consistent with mature mucociliary differentiation. In contrast, both viruses induced substantial ciliary dysfunction, reflected in altered ciliary beat frequency (CBF) distributions [Fig. 4D, E], and representative regions of interest from the video analysis [Fig. 4F, G]. RSVA infection reduced mean CBF from 8.7 to 5.8 Hz [Fig. 4D, F]. Consistent with its more limited replication at this timepoint, hPIV3 caused a smaller reduction in the proportion of actively beating cilia. Interestingly, hPIV3 infection also produced localised regions exhibiting abnormally high CBF relative to mock [Fig. 4E, G].

Comparative analysis from viral genomic sequencing for cell line and ALI culture samples demonstrated robust concordance of mutagenic patterns [Fig. 4J, K]. This is consistent with mutation accumulation during culture as expected due to the viral RdRp lacking proofreading exonuclease activity³³. For both RSVA and hPIV3, 85.4% and 81.7% of mutations were shared between the two experimental systems, respectively, at the 7-dpi endpoint. The mutation analysis revealed contributions from both intra-host single nucleotide variants (iSNVs, representing minority variants with allele frequency <50%) and single nucleotide polymorphisms (SNPs, representing consensus-level variants with allele frequency >50%), with the iSNV representing only 1% of the total mutations. The predominance of shared mutations across experimental models provides a robust baseline for the following analysis of the additional drug-induced mutagenic effects.

RdRp inhibitors reduce viral replication and preserve functional cilia activity during infection of the primary airway epithelial model

Concentrations of the RdRp inhibitor antivirals were selected for testing in the primary airway model based on the dual-drug combinations that showed the strongest synergy in cell line assays, along with the corresponding single drug doses. Antivirals were delivered via the basolateral media component of the ALI cultures. Inhibition of viral replication was determined from viral-GFP data [Supplementary Fig. 4B, C], used to calculate relative viral inhibition in live cultures and subsequently validated by post-fixation immunofluorescent staining, for both RSVA [Fig. 5] and hPIV3 [Fig. 6].

Fig. 5. Efficacy of antiviral combinations on RSVA infection of primary airway epithelial cell cultures.

A Representative micrographs of primary airway epithelial cells cultured at air-liquid interface for infection assays with RSVA (A), with whole-well brightfield (BF) and viral GFP fluorescence shown at 7 days post-infection (dpi). B GFP intensity was quantified by image analysis to calculate relative viral inhibition for all condition (One-way ANOVA versus virus only control). C High speed video microscopy was used to analyse cilia beat frequency (CBF) for all conditions, with the red line indicating the mean CBF in Hz. D, E Histogram and box plot representation of CBF distribution for RSVA only and selected antiviral drug conditions (Wilcoxon versus virus only). F Representative confocal micrographs of RSVA-challenged ALI culture with 4 mg/L remdesivir and 8 mg/L favipiravir combination, fixed for immunofluorescent staining and imaging at 7 dpi. Channels represent signal for DAPI staining of cell nuclei (blue), anti-GFP for the GFP-tagged virus (green), phalloidin for tight junctions (white), alpha-tubulin for ciliated epithelial cells (violet), and MUC5AC to indicate mucus-secretory cells (yellow). Orthogonal view also shown for representative culture, with the same immunofluorescent staining. G Ct result from RSVA qRT-PCR performed on apical wash samples collected for RSVA viral genomic sequencing, indicating respective viral load at 7 dpi. H Observed mutation rate with selected antiviral conditions, as change from virus-only baseline, normalised against Ct. I The Transition/Transversion (titv) ratio of mutation analysis with selected antiviral conditions. Representative images and functional analyses are from triplicate ALI cultures per experimental condition. Favi = favipiravir, Rem = remdesivir, Mol = EIDD-1931 (molnupiravir). (**P < 0.01, ***P < 0.001, ****P < 0.0001).

Fig. 6. Efficacy of antiviral combinations on hPIV3 infection of primary airway epithelial cell cultures.

A Representative micrographs of primary airway epithelial cells cultured at air-liquid interface for infection assays with hPIV3 (A), with whole-well brightfield (BF) and viral GFP fluorescence shown at 7 days post-infection (dpi). B GFP intensity was quantified by image analysis to calculate relative viral inhibition for all condition (One-way ANOVA versus virus only control). C High speed video microscopy was used to analyse cilia beat frequency (CBF) for all conditions, with the red line indicating the mean CBF in Hz. D, E Histogram and box plot representation of CBF distribution for hPIV3 only and selected antiviral drug conditions (Wilcoxon versus virus only). F Representative confocal micrographs of hPIV3-challenged ALI culture with 0.5 mg/L remdesivir and 64 mg/L favipiravir combination, fixed for immunofluorescent staining and imaging at 7 dpi. Channels represent signal for DAPI staining of cell nuclei (blue), anti-GFP for the GFP-tagged virus (green), phalloidin for tight junctions (white), alpha-tubulin for ciliated epithelial cells (violet), and MUC5AC to indicate mucus-secretory cells (yellow). Orthogonal view also shown for representative culture, with the same immunofluorescent staining. G Ct result from hPIV3 qRT-PCR performed on apical wash samples collected for hPIV3 viral genomic sequencing, indicating respective viral load at 7 dpi. H Observed mutation rate with selected antiviral conditions, as change from virus-only baseline, normalised against Ct. I. The Transition/Transversion (titv) ratio of mutation analysis with selected antiviral conditions. Representative images and functional analyses are from triplicate ALI cultures per experimental condition. Favi = favipiravir, Rem = remdesivir, Riba = ribavirin. (*P < 0.05, **P < 0.01, ****P < 0.0001).

We observed a range of viral inhibition using these selected antiviral concentrations in the primary airway model. For RSVA, 4 mg/L (6.6 μM) of remdesivir, alone or combined with 8 mg/L (50.9 μM) favipiravir, demonstrated near complete prevention of viral-GFP signal [Fig. 5A, B]. Against hPIV3, 0.5 mg/L (0.8 μM) remdesivir in combination with 64 mg/L (407 μM) favipiravir reduced relative infection by 94.8% in the ALI model [Fig. 6A, B]. RSVA infection caused marked epithelial disruption, including loss of epithelial tight junction integrity identified by phalloidin staining, whereas these structural changes were prevented by effective antiviral concentrations such as 4 mg/L (6.6 μM) remdesivir and 8 mg/L (50.9 μM) favipiravir [Fig. 5F]. For both viruses, antiviral concentrations which caused strong inhibition of replication as determined by viral GFP also resulted in reduced viral load in apical washes, and were associated with accumulation of drug-induced viral genomic mutations [Figs. 5G–I, 6G–I]. The inhibition of RSVA and hPIV3 infection by selected antiviral doses in this 24-well ALI culture model was replicated in supplementary screening with high-throughput 96-well ALI cultures generated from six further paediatric donors [Supplementary Fig. 4D].

High speed video microscopy was further used to analyse whether antiviral treatment could prevent virus-induced ciliary disruption [Fig. 5C–E & 6C–E]. RSVA infection caused both loss of motile cilia and a significant reduction in mean CBF, but this was partly attenuated by certain antiviral drug combinations that produced high levels of viral inhibition, such as 4 mg/L (6.6 μM) remdesivir with 8 mg/L (50.9 μM) favipiravir [Fig. 5D, E]. As hPIV3 alone induced less disruption to CBF than RSVA, the effects of antivirals on cilia activity relative to viral inhibition were less obvious. However, combinations of 0.5 mg/L (0.8 μM) remdesivir with 4 mg/L (16.3 μM) ribavirin or 64 mg/L (407 μM) favipiravir both reduced the proportion of abnormally fast beating cilia previously observed during hPIV3 infection [Fig. 6D, E]. Neither hPIV3 infection alone, nor with any of the investigated single or dual-antiviral concentrations, caused any reduction in barrier integrity as measured by TEER at 5 or 7dpi [Supplementary Fig. 2C]. Fully differentiated ALI culture wells exposed to single or dual-antiviral concentrations for 24 hours in the absence of virus did not demonstrate any direct drug cytotoxicity by LDH release [Supplementary Fig. 2D].

RdRp inhibitors induce signature mutations associated with viral inhibition across infection models

Endpoint samples, consisting of culture supernatant for cell lines or apical surface wash for ALI culture, were collected for qRT-PCR, viral genomic sequencing and mutation analysis. For both RSVA and hPIV3 in both the cell line and primary airway culture models, increasing drug concentrations were associated with a reduction in viral load and inhibition of viral replication [Fig. 7A]. Consistent with prior reports for molnupiravir and favipiravir in SARS-CoV-2 infection, the total number of viral mutations, corrected for viral load and normalised against mutations occurring over the same time-period in untreated cultures [Fig. 7B], increased in a dose-dependent manner³⁴.

Fig. 7. Comparative genomic analysis of antiviral efficacy between cell lines and ALI culture.

A Mean viral load by qRT-PCR CT values across antiviral concentrations for RSVA (left) and hPIV3 (right) in endpoint samples from cell line (orange circles) versus ALI model (orange triangles) infection assays. Lower CT values indicate higher viral replication. B Total number of mutations compared to zero drug normalized by CT values across antiviral treatments for RSVA (left) and hPIV3 (right) in cell lines (blue circles) and ALI samples (blue triangles). C Transition/transversion (Ts/Tv) mutation ratios across antiviral treatments for RSVA (left) and hPIV3 (right) comparing cell lines (green circles) and ALI samples (green triangles) values are normalised for CT values and zero drug. Boxes indicate selected corresponding single and dual-combination concentrations for remdesivir-molnupiravir (red boxes) or remdesivir-favipiravir (black boxes). D Heatmap showing the proportion of consensus mutations across viral genes (y-axis) for different antiviral concentrations (x-axis) in cell line experiments. Colour intensity represents the percentage of consensus mutations (0–100%). E Comparison of total viral mutations (adjusted for viral load) under matching antiviral concentrations between the respective cell line and ALI models for RSVA and hPIV3 (Wilcoxon matched-pairs test, **p < 0.01). Favi = favipiravir, Rem = remdesivir, Mol = EIDD-1931 (molnupiravir), Riba = ribavirin.

Unexpectedly, remdesivir was also mutagenic for both RSVA and hPIV3 across the cell lines and ALI models, contrasting with observations in SARS-CoV-2 infection, where remdesivir is generally considered a non-mutagenic chain terminator^35,36. As expected for RdRp-targeting nucleoside analogues, transition point mutations (A ↔ G; C ↔ T) increased with antiviral treatment, whereas transversion substitutions (purine↔pyrimidine) did not. Accordingly, the transition/transversion (Ts(G-A;C-T) /Tv) ratios increased for all antiviral treatments relative to untreated controls for both RSVA and hPIV3 [Fig. 7C].

Notably, the increased viral inhibition seen with dual-drug combination therapies, such as favipiravir (8 mg/L; 50.9 μM) combined with molnupiravir or remdesivir against RSVA, or favipiravir (16 mg/L; 101 μM) combined with ribavirin or remdesivir against hPIV3, was not associated with further increased mutagenesis [boxed points, Fig. 7A-C]. Across both viruses, mutations were dispersed randomly across the genome, with no clustering, enrichment, or evidence of selection or concentration at known drug-binding or active sites [Fig. 7D].

Despite the described differences for individual doses, comparing the two culture systems at a population level for viral inhibition with matching antiviral concentrations suggested no significant difference between cell line and ALI models for RSVA (p = 0.70) or hPIV3 (p = 0.12) [Supplementary Fig. 4E]. However, for RSVA, antiviral-induced viral mutagenesis was significantly higher in ALI cultures compared to VeroE6 cells (p = 0.0078) [Fig. 7E]. While hPIV3 showed a minor trend towards higher mutation rates in ALI cultures for most tested antiviral doses, this did not reach statistical significance overall (p = 0.22) [Fig. 7E].

Discussion

In this study, we demonstrate that four clinically relevant RdRp inhibitors, remdesivir, ribavirin, favipiravir and molnupiravir, exhibit distinct but complementary antiviral activities against RSVA and hPIV3 across both traditional epithelial cell lines and a physiologically relevant primary airway epithelial ALI model. We show that remdesivir is the most potent monotherapy across viruses and models, that multiple dual-drug combinations act synergistically, and that effective antiviral exposure is associated with dose-dependent increases in transition mutagenesis. We further show that antiviral effects and mutagenesis profiles translate into the ALI culture system, where structural and functional preservation of the epithelium including maintenance of ciliary beating provides additional biological validation of therapeutic benefit^15,37.

We first characterised the monotherapy activity of each RdRp inhibitor using optimised epithelial cell-line infection assays. The selection of cell lines was driven empirically by the requirement for robust infection and quantifiable CPE and GFP readouts [Supplementary Fig. 1]. LLC-MK2 cells, a monkey kidney epithelial cell line, were the most permissive for hPIV3 infection, consistent with their extensive use in wider PIV studies³⁸,while VeroE6 cells provided reliable RSVA infection and are well-established in other RSV and antiviral research^39,40.

Across both viruses, remdesivir emerged as the most effective antiviral tested. We calculated EC₅₀ values for hPIV3 (1.1 mg/L; 1.8 μM) and RSVA (3.6 mg/L; 6.0 μM), which are consistent with or slightly higher than previous reported values using similar strains and assay formats, noting that earlier infection timepoints (72 hours post-infection) used in other studies may yield lower EC₅₀ estimates due to reduced cytopathic contribution¹⁷. The efficacy of remdesivir against different RSV strains in alternative cell lines (HEp-2, A549) with EC₅₀ < 1 mg/L ( < 1.7 μM) has also been documented, supporting its broad antiviral potential^41–43.

Ribavirin also demonstrated strong antiviral activity against both viruses. Our EC₅₀ result for hPIV3 (5.6 mg/L; 22.9 μM) aligns with EC₅₀ values of 2–13 mg/L (8.2-53.2 μM) previously published for both laboratory-adapted and clinical hPIV3 strains^16,17. Against RSVA, we observed an EC₅₀ of 6.1 mg/L (25.0 μM), comparable to values of 3.9 mg/L (16.0 μM) and 2.7 mg/L (11.1 μM) in HeLa/M and HEp-2 cells^44,45.

Favipiravir inhibited both RSVA and hPIV3 but required higher concentrations, and cytotoxicity emerged near the EC₉₀ in both VeroE6 and LLC-MK2 cells. Nonetheless, an EC₅₀ of 15.2 mg/L (96.8 μM) for RSVA is within the 10.7–41.0 mg/L (68.1–261 μM) range published in studies using HEp-2 infection models^46,47. In vitro favipiravir efficacy against hPIV infection has been reported at 10.8 mg/L (68.7 μM) and 21.7 mg/L (138 μM) elsewhere, although in different host cells such as Vero-118 and PLC/PRF/5, whilst our EC₅₀ value of 80.2 mg/L (510 μM) in LLC-MK2 cells is likely beyond the clinically achievable range^16,46.

Molnupiravir (using the active metabolite EIDD-1931) produced the lowest-dose antiviral effect against RSVA, with an EC₅₀ of 0.8 mg/L (3.1 μM), which is consistent to results between 0.1–1.0 mg/L (0.4–3.9 μM) obtained with clinical RSV strains in HEp-2 cells⁴⁸. In contrast, we observed no inhibition of hPIV3 due to pronounced cytotoxicity above 2 mg/L (7.7 μM) in LLC-MK2 cells, which would prevent any accurate determination of potential hPIV3 inhibition above this concentration. To our knowledge, hPIV sensitivity to molnupiravir has not been reported previously. It is possible that sensitivity of hPIV3 to molnupiravir may be measurable by screening in ALI cultures, if less sensitive to cytotoxicity than LLC-MK2 cells, but that was not tested in this study.

We next evaluated dual-drug combinations, which can provide several advantages over monotherapy including additive or ideally synergistic inhibition of viral replication²¹. Recently the use of combined antivirals has been investigated in greater detail in the context of SARS-CoV-2 therapy, including demonstrated synergism in cell culture and animal models^9,24,25. Our group and others have demonstrated RdRp inhibitors such as remdesivir and molnupiravir have comparable effective dose ranges against SARS-CoV-2 laboratory and clinical strains in vitro, highlighting the potential for broader-spectrum use of these drug combinations across common respiratory RNA viruses^9,49. In a directly relevant clinical setting, a case series at our partner hospital, Great Ormond Street Hospital for Children, has been recently published outlining potential benefits from combination favipiravir, ribavirin, and nitazoxanide to treat respiratory viral infections in severe T-cell deficient children⁵⁰.

Consistent with findings in SARS-CoV-2 models, we observed synergy for several RdRp inhibitor pairs against RSVA in vitro, including favipiravir-molnupiravir, remdesivir-molnupiravir, and remdesivir-favipiravir⁹. Against hPIV3 infection, synergy was more limited but evident for remdesivir–favipiravir and remdesivir-ribavirin. These combinations were also dose-sparing: for example, remdesivir alone required 8.5 mg/L (14.1 μM) to achieve an EC₉₀ against RSVA, yet 0.5 mg/L (0.8 μM) in combination with 0.5 mg/L (1.9 μM) molnupiravir achieved 97.6% inhibition. Importantly, synergistic doses showed minimal cytotoxicity. Synergistic combinations of RdRp inhibitors have previously been shown for SARS-CoV-2, including using the in vivo Syrian hamster model^25,51,52. Radoshitzky et al. have also shown a broad-spectrum potential of remdesivir against many viral classes with no antagonism with a favipiravir combination⁵³.

Our mutation analysis parallels earlier work showing that transition mutagenesis (A ↔ G; C ↔ T) increases with nucleoside analogue exposure and can serve as a biomarker for antiviral activity against norovirus, influenza B, SARS-CoV-2 and RSVA^34,54–56. As expected, favipiravir and molnupiravir induced increasing transition mutations, after correcting for viral load and normalising to untreated controls. Unexpectedly, remdesivir, which is not reported as directly mutagenic as with favipiravir or molnupiravir in SARS-CoV-2, also induced significant increases in total and transition mutations in both RSVA and hPIV3 in both infection models^34–36. Our data, which are the first to examine potential mutagenic signatures in this way for RSVA and hPIV3, raise the possibility of previously unrecognised virus-specific properties of remdesivir’s activity in paramyxoviruses. A secondary mechanism of action has previously been suggested against coronaviruses via template-dependent inhibition by embedded remdesivir, which could induce copy-error mutation accumulation⁵⁷.

It has been previously observed in SARS-CoV-2 infected hamsters that combination therapy modestly increased total mutation burden compared with each drug alone³⁴. While the current study was not designed to measure that effect, we did observe a modest increase in total mutation burden for favipiravir plus remdesivir for hPIV3 in LLC-MK2 cells, although this was less obvious in RSVA infected ALI cultures [Fig. 7B, C]. A targeted study would be required to determine whether mutagenesis contributes mechanistically to synergy.

The primary airway ALI system provided several advantages over standard monolayer cell lines in this study. First, unlike VeroE6 and LLC-MK2, which supported RSVA or hPIV3 but not both, ALI cultures supported infection by both viruses, allowing direct cross-virus comparison under identical culture conditions²⁷. Second, the mammalian cell lines have limited biological relevance to human infection, and do not display important respiratory epithelial characteristics such as transport proteins, mucus secretion and active motile cilia²⁸.

Studies on respiratory viruses have begun to utilise these advantages for diverse research aims, such as investigating the impact of tissue donor age on RSV infection kinetics, the mutation accumulation rates when passaging hPIV, and the cell tropism and epithelial responses to SARS-CoV-2 infection^58–60. Respiratory ALI models are also emerging as a viable tool for improved antiviral testing. ALI culture of human nasal epithelial cells has been applied for screening antiviral nucleosides for SARS-CoV-2 inhibition^25,61. CBF is also a powerful functional readout in this model. Our data showed RSVA disrupts ciliary beating, as has been reported previously, but this effect was attenuated by some of the tested antiviral doses⁴⁵. Change in CBF is useful in the context of drug screening as a sensitive indicator of cytotoxicity and has been applied elsewhere in studies of pharmaceuticals or bacterial toxins^62–64. Furthermore, we have previously demonstrated the potential application of a 96-Transwell ALI model for high-throughput screening with RSVA⁶⁵.

Overall, most of the effective or synergistic doses determined from the cell line assays for the RdRp inhibitors used in our study translated into viral inhibition in the ALI infection model. Remdesivir even showed enhanced efficacy in the ALI model for both viruses. Whilst the cell line EC₅₀ doses were 3.6 mg/L (6.0 μM) for RSVA and 1.1 mg/L (1.8 μM) for hPIV3, a lower dose of 0.5 mg/L (0.8 μM) was already sufficient for 59.7% and 45.8% viral inhibition in ALI cultures, respectively. In contrast, favipiravir efficacy against RSVA was reduced in the ALI model, with 8 mg/L (50.9 μM) ineffective alone and not enhancing the efficacy of remdesivir or molnupiravir at the included dual-combination concentrations. Limited published studies have compared relative antiviral efficacy between such models. Mirabelli et al. previously determined EC₅₀ for ribavirin and other replication or fusion inhibitors against RSV using HEp-2 cells, and then demonstrated their efficacy with fully differentiated human airway epithelium cells at ALI, although this study differed in delivering higher doses (10-100x EC₅₀) at later timepoints (1-5 dpi) to mimic therapeutic regimens⁴⁵. Our data further aligns with the findings of Lin et al. for hPIV infection models. They also reported synergy for specific combinations of remdesivir and ribavirin in LLC-MK2 cells, as well as investigating ribavirin alone and in combination with GS-441524, the parent nucleoside of remdesivir, in infections of human nasal airway epithelial ALI cultures¹⁷.

While viral inhibition was comparable between the models, we did observe higher drug-associated mutation rates for RSVA using primary airway cultures. While there is some evidence that ALI cultures are associated with a lower viral mutation rate during routine passaging⁵⁹, it is also established that these primary airway cells reproduce complex innate immunity and antimicrobial responses and secretions which could influence the mutation pressure during viral replication^60,66. While distinct epithelial immune responses to different viruses are described for primary airway cultures, the specific mechanisms behind an increased antiviral drug-induced mutagenesis for RSVA in this model requires further study⁶⁷. Yet this finding does suggest these antivirals have greater potential mutagenic activity in the more physiologically-relevant differentiated airway model, with implications for their in vivo applications.

This study acknowledges several limitations. Firstly, comparing antiviral activity across the different epithelial models of immortalised cell lines and the ALI cultures introduces variability driven by differences in viral entry receptors, innate signalling, and cellular responses, all of which can influence drug potency. Comparison to previous literature is complicated by varied use of different cell lines elsewhere for RSV or hPIV (such as HEp-2, A549, Hela/M, Vero-118). Endpoint cytotoxicity further constrained concentration ranges for some antivirals (most notably for molnupiravir in LLC-MK2 cells), limiting full dose-response characterisation for hPIV3. These factors highlight the need for physiologically relevant models such as ALI cultures, while also emphasizing the importance of broader standardisation across laboratories to improve reproducibility. Secondly, in vitro findings may not fully predict in vivo therapeutic efficacy, pharmacokinetics, or host immune responses. Future studies using small animal models or clinical data will be essential to confirm the translational relevance of the synergistic combinations identified here. This study also evaluated antiviral drug effects on viral replication rather than post-exposure viral infectivity or viability, which is necessary in future work to assess any impact on transmissibility with these therapies. The ALI assays in this study were aimed at validating selected single and combined antiviral concentrations on viral inhibition specifically, but future work should further explore full synergy modelling with the ALI system as well as address any functional effects of the polymerase-inhibitor drugs alone on the epithelium, such as cilia beating. While a promising approach to improve efficacy, a potential disadvantage for the translation to clinical approval of novel combination therapies even with existing drugs is the need for specific new clinical trials. This study used representative strains of RSVA and hPIV3, subtypes associated with higher severity in infants and the immunocompromised³⁸, but efficacy across other subtypes should be explored in the future. Finally, while mutagenesis analysis provided important mechanistic insights, sequencing was performed at single endpoint timepoints. Time-resolved analysis would better characterise mutation accumulation, distinguish lethal mutagenesis from other inhibitory processes, and clarify whether mutagenic synergy contributes directly to antiviral synergy.

Overall, our results show that combinations of RdRp inhibitors suppress RSVA and hPIV3 replication in 2D epithelial cell models, with remdesivir displaying the greatest potency and several combinations exhibiting strong synergy, especially remdesivir-favipiravir, remdesivir-molnupiravir, and favipiravir-molnupiravir. The inhibition of viral replication translated into primary airway epithelial cultures, where effective antiviral concentrations preserved epithelial integrity and ciliary function after viral challenge. RdRp inhibitor–associated mutagenesis increased dose-dependently across models, including unexpected mutagenic signatures for remdesivir in RSVA and hPIV3, revealing potential virus-specific mechanisms. These findings strengthen the rationale for combination RdRp therapy for respiratory RNA infections, particularly in vulnerable patient groups, and as a broad-spectrum option for future viral threat preparedness.

Methods

Viral strains and culture

A laboratory strain of GFP-tagged RSVA was purchased from ViraTree (product code R121) and was titrated for use for RSVA antiviral assay infections. All hPIV experiments used a GFP-tagged hPIV3 strain purchased from ViraTree (product code P323). Titrated viral stocks were aliquoted, snap frozen and stored at -150 ^oC until experimental use.

RSVA and hPIV3 viral stock titration was performed by TCID₅₀ assays and syncytial plaque forming unit (SPFU) counting by GFP fluorescence, both using serial dilution inoculation of either VeroE6 cells for RSVA or LLC-MK2 cells for hPIV3, on 96-well culture plates. TCID₅₀ CPE functional titre was calculated by the Spearman & Kärber algorithm as previously described⁶⁸. TCID₅₀ plates were further imaged on a Nikon Ti-E microscope, using GFP-signal to count syncytial plaques per well to validate SPFU/ml titre. The calculated titres were used for all MOI calculations for subsequent experiments in this study.

Epithelial Cell Lines

African green monkey kidney cell line VeroE6 (ATCC: C1008-CRL-1586) was provided and authenticated by The Francis Crick Institute, London, UK, for use in this study. VeroE6 cells were routinely maintained in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal calf serum (FCS, Thermo Fisher), 100 units/mL Penicillin and 100 µg/mL Streptomycin (Sigma-Aldrich). Human lung adenocarcinoma cell line Calu-3 (HTB-55, ATCC) and rhesus monkey kidney cell line LLC-MK2 (CCL-7, Caltag-Medsystems Ltd) were maintained in DMEM supplemented with 10% fetal bovine serum (FBS, Thermo Fisher), 100 units/mL Penicillin and 100 µg/mL Streptomycin. All cells were incubated at 37 °C and 5% CO₂ in a humidified incubator, with weekly passaging and media replacement 1–3 times a week, as needed.

Primary airway epithelial culture

The air-liquid interface (ALI) culture model for human primary airway epithelium was performed as previously described, from a nasal epithelial brush biopsy collected from a healthy paediatric donor (3-year-old female)⁶⁰. Written informed consent was obtained from the donor’s parent or legal guardian, consistent with the Declaration of Helsinki. Ethics approval was granted through the Living Airway Biobank, administered through the UCL Great Ormond Street Institute of Child Health (REC reference 19/NW/0171, IRAS project ID: 261511, Northwest Liverpool East Research Ethics Committee).

In brief, basal epithelial cells were first expanded using co-culture with mitomycin-inactivated 3T3-J2 fibroblasts, then separated from feeder cells using differential trypsinization. Epithelial cells were seeded onto collagen I-coated, 24-well plate semi-permeable membrane supports (Transwell, 0.4 µm pore size, Corning). To induce differentiation, apical fluid was removed to expose cells to air (5% CO₂), and basolateral media was replaced with complete PneumaCult ALI differentiation media (STEMCELL Technologies), exchanged three times weekly. The apical cell surface was washed approximately once per week with PBS to remove excess mucus. The ALI cultures were maintained for a minimum of 4 weeks for complete differentiation before downstream experiments. Transepithelial electrical resistance (TEER) was regularly measured using an EVOM2 resistance meter (World Precision Instruments) to indicate differentiation status and monitor epithelial barrier function.

Supplementary ALI assays also used cells from healthy paediatric donors cultured at ALI on HTS Transwell-96 Permeable Supports (Corning), using the same general methodology as above, and as we have previously described⁶⁵.

Antiviral Drugs

The following antiviral drug compounds were acquired for this study: remdesivir (Bio-Techne), favipiravir (Fisher Scientific), EIDD-1931 (active metabolite of molnupiravir; Sigma-Aldrich), ribavirin (Cambridge Bioscience). These were purchased as lyophilised product and reconstituted in DMSO at a standard stock concentration of 10 mg/mL. The drug stocks were stored as single-use aliquots at –20 °C or 4 °C, per manufacturer recommendations.

Cell line model for assessing antiviral efficacy and cytotoxicity

Cell line assays in 96-well format were adapted from our previous SARS-CoV-2 antiviral study⁹, with the following modifications for optimisation with these viruses. For RSVA assays, VeroE6 cells were seeded the day before infection at a density of 1-2 × 10⁴ cells per well in a black walled, clear bottomed 96 well culture plate (Corning Costar). Six plates were seeded per experiment, for triplicate plates measuring antiviral infection inhibition (with virus) and antiviral drug cytotoxicity (without virus).

For infection assays, culture media was removed and viral inoculum added at MOI of 0.02 (or media control for cytotoxicity plates) in 50 µL of serum-free DMEM and incubated at 37 °C and 5% CO₂ for 1-2 h. During this incubation, the antiviral drugs were prepared at 4x desired experimental dose in a separate 96 well plate, also in serum-free DMEM. After the inoculation, 50 µL of the prepared 4x antiviral media solution was added to corresponding assay wells (or DMEM only for mock and virus-only control wells). Each well was then supplemented with 100 µL of culture media (10% serum), for a final well volume of 200 µL at 1x experimental antiviral drug concentration and 5% FCS.

To assess antiviral drug cytotoxicity in absence of virus, in parallel with the infection assay the remaining three plates of VeroE6 cells were mock-infected with 50 µL of serum-free DMEM for 1-2 h. As above, 50 µL of the prepared 4x antiviral media solutions was added to the corresponding assay wells (or DMEM only for no-drug mock control), followed by 100 µL of culture media (10% serum), for a final well volume of 200 µL at 1x experimental antiviral drug concentration and 5% FCS. Supplementary drug toxicity and viability assays used the CyQUANT^TM LDH Cytotoxicity Assay Kit (Invitrogen) or alamarBlue (Bio-Rad), per manufacturer assay protocols.

Assay plates were incubated at 37 °C and 5% CO₂, without media replacement, until the experimental endpoint (typically 7 days post-inoculation unless otherwise specified). The plates were regularly monitored for CPE by inverted light microscope, and at chosen experimental timepoints all wells were recorded for GFP fluorescence on a Nikon Ti-E microscope using an automated NIS-Elements JOBS (Nikon) module. At experiment endpoint, media was collected for downstream applications including PCR and sequencing, and remaining cells were fixed and stained using 4% paraformaldehyde and 0.1% v/v crystal violet (Sigma). Excess stain was washed off with water and plates air-dried, before crystal violet staining was quantified by absorbance at 595 nm for each well using Spiral Averaging on a FLUOstar Omega plate reader (BMG Labtech).

The exact same methodology was applied for hPIV3 assays, other than the use of the corresponding cell line LLC-MK2 and the infection media containing 5% FBS.

Statistical analysis of in vitro antiviral efficacy

Analysis of the antiviral efficacy was performed as previously described⁹. In brief, the dependent variable for the statistical analysis of antiviral effect was the percentage inhibition read from the optical density at 595 nm from crystal violet staining of remaining cells. Statistical analyses were performed in R (version 4.4.2) including calculation of the variables effective dose for 50 and 90% inhibition (EC₅₀/EC₉₀) for monotherapies, and the SynergyFinder version 3.14.0 package for R was utilised for drug combination analysis. The R code is available on GitHub at https://github.com/ucl-pharmacometrics/RSV-PIV-in-vitro-analysis. Some supplementary figures and statistical analyses used GraphPad Prism (version 10.1).

Primary air-liquid interface (ALI) cell culture model for assessing antiviral efficacy

Fully differentiated primary epithelial cultures were grown at air-liquid interface in 24-well Transwell culture plates as described above. Prior to infection, the apical cell surface was washed with PBS for 10 min to remove excess mucus, and any pre-infection TEER measurements or microscopy performed. Viral inoculum (or media-only mock control) was added to the apical Transwell at an MOI of 0.05 in 200 μL of PneumaCult media and incubated at 37 °C and 5% CO₂ for 1 h. The inoculum was removed, and the basolateral media replaced with 600 μL of PneumaCult media containing chosen antiviral drug concentrations (or media only for controls), in triplicate replicate wells per experimental condition. The cultures were incubated at 37°C and 5% CO₂ without replacement of the antiviral-containing media until an experimental endpoint at 6-7 days post-infection (dpi). Mock-infected cultures were used to confirm tolerance of this longer window without basolateral media exchange, by TEER and phalloidin staining for tight junctions, immunofluorescent staining for cell types, and analysis of cilia beat activity. All wells were assessed for GFP fluorescence and cilia beat frequency on a Nikon Ti-E microscope using an automated NIS-Elements JOBS (Nikon) module. At endpoint, apical washes were collected from all wells by adding 200 μL PBS for 10 min, and stored for downstream PCR and sequencing analysis. All cultures were fixed in 4% PFA for 30 min and stored submerged in PBS + 0.02% Sodium Azide for downstream immunofluorescent staining and confocal microscopy.

Cilia beat frequency (CBF) analysis

Cilia beat measurements from the primary airway model were performed using an environmental chamber (5% CO₂, 37 °C) connected to an inverted microscope system (Nikon Ti-E). Image acquisition was automated using NIS-Elements JOBS (Nikon) module for fast time-lapse at an interval of 100fps. CBF was determined from timelapse files by first extracting the average pixel intensities using ImageJ (NIH) and, secondly by performing a fast Fourier transformation (FFT) on this data using R; 6400 regions of interest (ROI), each with an area of 16.8 µm² were analysed per file (pixel resolution = 0.32 µm). CBF was computed using ciliR software as described previously⁶⁵.

qRT-PCR

Culture supernatant samples collected for downstream analysis were stored at -70 °C, with dilution at 1:1 in DNA/RNA shield (Zymo Research) for longer-term storage. RNA extraction was performed using the Qiagen Viral RNA mini kit (Qiagen) according to the manufacturer’s instructions. For qRT-PCR, 4 µL of extract was combined with 16 µL of master mix (Takara One Step PrimeScript III and primer-probes for PIV 1/2/3 and RSV A/B) and manually loaded onto a MicroAmp™ EnduraPlate™ Optical 384-Well Plate. The qRT-PCR was performed using a QuantStudio 5 real-time PCR system, and negative template control and positive plasmid controls for RSV A/B and PIV types 1/2/3 were included in each run to control for contamination and amplification, respectively.

Library preparation and sequencing

Viral whole genome sequencing was carried out by bait capture with biotinylated RNA oligonucleotides used in the Agilent SureSelect^XT (SSXT) protocols. A design targeting PIV 1-3, RSV A and B and Metapneumovirus was designed in-house and synthesised by Agilent Technologies.

11 µl RNA extract was used in first-strand cDNA synthesis with random primers and SuperScript IV (Thermo Fisher Scientific) according to the manufacturer’s instructions. Second-strand cDNA was synthesised using the NEBNext Ultra II Non-Directional RNA Second Strand Synthesis Module (New England BioLabs) according to the manufacturer’s instructions. Double-stranded cDNA was purified using AMPure XP Beads (Beckman Coulter) with a 40 µl elution volume and was quantified using a high-sensitivity dsDNA Qubit kit (Fisher Scientific).

Double stranded cDNA (bulked with male human gDNA (Promega) if required) was sheared using a Covaris E220 focused ultra-sonication system (42 s, PIP 75, duty factor 10, cycles per burst 1000). End-repair, non-templated addition of 3’ poly A, adaptor ligation, pre-capture PCR, hybridisation, post-capture PCR and all post-reaction clean-up steps were performed using the SureSelect^XT Low Input kit (Agilent Technologies) with minor modifications to the manufacturer’s protocol to account for variable pathogen loads. Quality control steps were performed using the 4200 TapeStation (Agilent Technologies). Samples were sequenced using the NextSeq 500 platform (Illumina). Run data were demultiplexed and converted to fastq files using Illumina’s BCL Convert Software v3.7.5.

Sequencing analysis

The raw fastq reads were adapted, trimmed and low-quality reads removed using the fastp algorithm. The reads were aligned against the RSVA reference genome (KX765960.1) and the hPIV3 reference genome (NC_001796.2); consensus sequences were then called at a minimum of 20X coverage. The entire processing of raw reads to consensus was carried out using nf-core/viralrecon pipeline⁶⁹. Only samples producing genomes with at least 90% genome coverage at 10X sequencing depth were kept for further analysis. Variant calling was done using the iVAR algorithm using the metagenomic protocol⁷⁰. Minority variants were filtered if the total coverage at that position was less than 200×, with at least 20 reads in total and at least 5 independent reads with no strand-bias supporting each of the main two alleles, and minimum allele frequency threshold was 5%. Antiviral induced mutations were identified by comparing mutations (nucleotide) from virus only positive control samples with sequences from the antiviral drug challenge conditions. General data processing was carried out in R 4.2.0, figures were made using the ggplot2 package⁷¹.

Immunofluorescent staining and confocal microscopy of primary airway model

For immunofluorescence confocal imaging, ALI cultures were fixed using 4% (v/v) paraformaldehyde for 30 min, permeabilized with 0.2% Triton X-100 (Sigma) for 15 min and blocked using 5% goat serum (Sigma) in PBS for 1 h. The cultures were then incubated overnight at 4 °C with primary antibody dilutions prepared in 5% goat serum in PBS with 0.1% Triton X-100. The primary antibodies used included mouse anti-MUC5AC (Sigma-Aldrich, MAB2011, diluted 1:500), Goat Recombinant Anti-alpha Tubulin (acetyl K40) antibody (Abcam, EPR16772, diluted 1:100), Rabbit Anti-GFP antibody (Abcam, EPR14104, diluted 1:100).

Following primary antibody incubation, cultures were washed and then incubated with secondary antibodies diluted in 1.25% donkey serum in PBS with 0.1% Triton X-100 at room temperature for 1 h the following day. The secondary antibodies included Donkey Anti-Mouse Alexa Fluor® 594 preadsorbed (Abcam, ab150112, 1:500), Donkey anti-rabbit Alexa Fluor® 488 preadsorbed (Abcam, ab96919, 1:500), Donkey anti-goat rabbit Alexa Fluor® 647 preadsorbed (Abcam, ab15013, 1:500).

After the secondary antibody incubation, cultures were stained with Alexa Fluor 555 phalloidin (ThermoFisher, A34055, 4 μg/ml) for 1 hour and DAPI (Sigma, 2 μg/ml) for 15 min at room temperature to visualize F-actin and nuclei, respectively. Samples were washed three times with PBS containing 0.1% Tween 20 after each incubation step.

Imaging was carried out using an LSM710 Zeiss confocal microscope, and the resulting images were analysed using Fiji/ImageJ v.2.1.0/153c54.

Supplementary information

Supplemental Figures ^{(1MB, pdf)}

Author contributions

S.E., A.I.J. and J.B. designed the study. S.E. and A.I.J. conducted experiments, analysed data, and co-wrote the manuscript. C.M.S., S.Z., J.B. and L.B. performed data analysis and contributed to manuscript and figures. M.W. performed microscopy experiments and analysis. J.A., M.P., E.K.H., H.M., A.S. and M.T. supported viral and cell culture and in vitro experiments. HT and RW carried out and oversaw viral genome sequencing. C.M. and A.K. supported qRT-PCR experiments and sample processing. C.O.C. provided support through ethics and donor recruitment for primary airway cultures. J.B. conceived the study and obtained funding. J.B., C.M.S. and J.F.S. oversaw data analysis and interpretation, and the write-up of the manuscript.

Data availability

Data supporting the findings of this study are available within the paper and its supplementary information, or available upon request. The R code for statistical analysis of antiviral efficacy is available on GitHub at https://github.com/ucl-pharmacometrics/RSV-PIV-in-vitro-analysis. The viral sequencing dataset is deposited as an NCBI BioProject under accession number PRJNA1465822.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s44259-026-00222-7.