In Vitro Combinations of Baloxavir Acid and Other Inhibitors against Seasonal Influenza A Viruses

Two antiviral classes, the neuraminidase inhibitors (NAIs) and polymerase inhibitors (baloxavir marboxil and favipiravir) can be used to prevent and treat influenza infections during seasonal epidemics and pandemics. However, prolonged treatment may lead to the emergence of drug resistance. Therapeutic combinations constitute an alternative to prevent resistance and reduce antiviral doses. Therefore, we evaluated in vitro combinations of baloxavir acid (BXA) and other approved drugs against influenza A(H1N1)pdm09 and A(H3N2) subtypes. The determination of an effective concentration inhibiting virus cytopathic effects by 50% (EC50) for each drug and combination indexes (CIs) were based on cell viability. CompuSyn software was used to determine synergism, additivity or antagonism between drugs. Combinations of BXA and NAIs or favipiravir had synergistic effects on cell viability against the two influenza A subtypes. Those effects were confirmed using a physiological and predictive ex vivo reconstructed human airway epithelium model. On the other hand, the combination of BXA and ribavirin showed mixed results. Overall, BXA stands as a good candidate for combination with several existing drugs, notably oseltamivir and favipiravir, to improve in vitro antiviral activity. These results should be considered for further animal and clinical evaluations.


Introduction
Since 1977, two subtypes of influenza A viruses, H1N1 and H3N2, along with influenza B viruses, have co-circulated in the human population. These viruses are responsible for seasonal epidemics and may cause global pandemics [1]. Influenza epidemics can lead to millions of cases of severe respiratory illnesses and to half a million deaths annually worldwide [2] representing a major public health challenge. Vaccination still remains the first line of prevention against seasonal influenza infections [3][4][5] but the protection conferred by actual vaccines varies from year to year due to the evolution of the circulating

Evaluation of Antiviral Combinations in HAE
The apical poles of HAE were gently washed twice with warm OptiMEM medium (Gibco, ThermoFisher Scientific, Gaithersburg, MD, USA)) and then infected with a 150 µL dilution of A/California/7/2009 (H1N1)pdm09 or A/Texas/50/2012 (H3N2) in OptiMEM, at a multiplicity of infection (MOI) of 0.1 or 0.01, respectively. For mock infection, the same procedure was followed using OptiMEM as inoculum. After 1 h incubation at 37 • C, the viral inoculum was removed. At 5, 24 and 48 hpi, treatments with specific dilutions of antiviral molecules alone or in combination in MucilAir ® culture medium (Epithelix SAS, Saint Julien-en-Genevois, France) were applied through basolateral poles. At 24, 48 (before treatment renewal) and 72 hpi, apical washes with 150 µL OptiMEM were performed and stored in two aliquots at −80 • C for TCID50 viral titration. At 24, 48 and 72 hpi, transepithelial electrical resistance (TEER) between the apical and basal poles of the HAE, considered as a surrogate of epithelium integrity, was measured using a dedicated volt-ohm meter (EVOM2) and compared to baselines values measured before infection (t = 0).

Statistical Analysis
All experimental assays were performed in triplicate at a minimum, and representative results are shown. HAE viral titers in the drug combination groups were compared against those of both the untreated (*) and most performant single-treated (#) groups using mixed model two-way analysis of variance (ANOVA) with Bonferroni post hoc test. The testing level (α) was 0.05. Statistical analyses were performed on all available data, using GraphPad, Prism 7.

Antiviral Activity of Single Drugs against Two Influenza A Strains
As shown in Table 1, effective dose responses for influenza A(H1N1)pdm09 and A(H3N2) replication were determined for each drug. The EC50 values of the wild-type A(H1N1)pdm09 strain were 3.87 ± 0.36 and 4.05 ± 0.88 µM for ribavirin and favipiravir, respectively. Oseltamivir (0.10 ± 0.05 µM) and zanamivir (0.13 ± 0.07 µM) had the same potency. Peramivir and BXA had the strongest inhibitory activity against A/California/7/2009 virus with EC50 values of 15.00 ± 5.77 and 0.48 ± 0.22 nM, respectively. The EC50 value of oseltamivir and favipiravir were 0.42 ± 0.29 and 10.32 ± 1.89 µM against the A(H3N2) virus, respectively. Similar activities were observed for ribavirin and zanamivir with EC50 values of 2.22 ± 1.55 and 2.48 ± 0.96 µM, respectively. Peramivir (48.43 ± 21.83 nM) and BXA (19.55 ± 5.66 nM) also had the most potent activity against the A(H3N2) strain. Data are means EC50 (50% effective concentration) ± standard deviation (SD) from at least three independent experiments.

In Vitro Two-Drug Combination Activity against Two Influenza A Strains
Combination drug experiments against the A(H1N1) strain (Table 2A) showed that BXA exhibits synergistic effect when combined with each NAI. The combination index weights (CIwt) were 0.40, 0.48 and 0.48 for zanamivir, oseltamivir and peramivir, respectively. Additionally, BXA and favipiravir had synergistic effects against influenza A(H1N1) with a CIwt of 0.54. However, BXA in combination with ribavirin induced an antagonist effect with a CIwt of 1.91.
In the A(H3N2) background (Table 2B), BXA also demonstrated synergistic effect when combined with each NAI. The CIwt were 0.47, 0.49 and 0.42 for zanamivir, oseltamivir and peramivir, respectively. Furthermore, the combination of BXA and favipiravir also had a highly synergistic activity with a CIwt value of 0.16. Again, the combination of BXA and ribavirin led to an antagonist effect with a CIwt of 1.23.

Two-Drug Combination Activity in Influenza A-Infected HAE
Based on the results obtained in ST6-GalI-MDCK cells, we further evaluated the efficacy of the selected two-drug combination ratios determined in Table 2A in the MucilAirTM reconstituted HAE infection model [32]. Note that different MOIs, 0.1 and 0.01, for A/California/7/2009 (H1N1) and A/Texas/50/2012 (H3N2) viruses, respectively, were based on previous characterization experiments and were used to reflect a comparable replication kinetics curve for both subtypes. As shown in Figure 1, A(H1N1) apical viral titers peaked at 48 hpi in the untreated group, reaching mean values of 4.4 (±3.6) × 108 TCID50/mL. As expected, treatment with 10 nM BXA was strongly effective by inducing a by TEER values, which remained stable all throughout the experiment, contrary to what was observed for the mock-treated controls. Of note, the extreme disparity of the EC50 ratio obtained for the combination of BXA and ribavirin (1:10,000) hampered its evaluation under experimental conditions in which the ribavirin concentration was sufficiently low not to induce a very significant antiviral effect. In the case of A(H3N2), apical viral titers peaked at 72 hpi, reaching mean values of 7.5 (±6.4) × 109 TCID50/mL, in the untreated group ( Figure 2). Interestingly, treatment with 10 nM BXA almost completely inhibited viral replication (), for which this dosage was too high to be compatible with the evaluation of drug combination in the A(H3N2) HAE model. We therefore used 5 nM BXA as the reference treatment, which induced a 1.5 log10 reduction in mean peak viral titers, reaching 4.9 (±1.78) × 108 TCID50/mL at 72 hpi. Treatment with zanamivir 625 nM (ratio 1:125) and peramivir 12.35 nM (ratio 1:2.47) induced significant reductions in viral production, as evidenced by 1.7 log10 (Figure 2A) and 1.3 log10 ( Figure 2C) lower mean peak viral titers compared to those of the mock-treated controls, respectively. On the other side, single-drug treatment with oseltamivir 105 nM (1:21), favipiravir 2.63 µM (ratio 1:526) and ribavirin 555 nM (ratio 1:111) ( Figure 2B,D,E) demonstrated relatively mild antiviral effects, inducing less than 0.5 log10 reductions in mean peak viral titers. Similar to what we observed for A(H1N1)pdm09, the antiviral effects of the different two-drug combinations tested against A(H3N2) were higher than those of single-drug treatments. Of note, the observed differences were statistically significant. The best synergy was obtained with the BXA + oseltamivir combo, which showed a 3 log10 reduction at 48 hpi and a ≥2 log10 reduction at 72 hpi in mean viral titers compared to the mock-treated but also the single-drug treatment conditions. Moreover, BXA + zanamivir and BXA + peramivir improved single-drug treatment at 72 hpi by almost 2 log10 and 1 log10, respectively. TEER measurements further supported these observations, as they remained stable for BXA + oseltamivir, BXA + zanamivir and BXA + peramivir throughout the experiments, which was not the case for single-drug treatments other than BXA. Interestingly, despite the almost negligible antiviral effect of favipiravir and ribavirin single-drug treatments, both BXA + favipiravir and BXA + ribavirin showed 2.8 log10 and 2.2 log10 reductions in viral titers at 48 hpi, respectively. Nevertheless, such an effect was not observed at 72 hpi, which is in line with the reduction in TEER values observed at this time point.

Discussion
Antivirals play a crucial role in the treatment and control of influenza epidemics and pandemics [33]. The M2 ion-channel blockers (adamantanes including amantadine and rimantadine) [34], NAIs (oseltamivir, zanamivir, peramivir or laninamivir) [35] and the polymerase complex inhibitors (BXM or favipiravir) have demonstrated clinical benefits, but the emergence of viral resistance may limit the clinical efficacy of monotherapies [36,37]. Antiviral combinations have been suggested to prevent or delay the emergence of resistance by targeting various steps in the virus replicative process [38]. Multiple therapies may allow for dosage reduction [33], in addition to improving therapeutic effect [38]. Drug combinations already exist to treat cancer [39] and infectious diseases caused by HIV and hepatitis C [40][41][42][43]. In a clinical study, the triple combination of oseltamivir, amantadine and ribavirin has been reported to decrease influenza viral shedding compared to monotherapy [24].
In this study, we aimed at investigating the potential benefits of combining BXA to most approved antivirals against seasonal influenza A(H1N1) and A(H3N2) strains using both a classic cell line and a physiologically and predictively relevant HAE model.
Firstly, we observed that BXA had the lowest EC50 (Table 1) against A(H1N1) and A(H3N2) viruses. These effective concentrations of BXA were followed by those of NAIs and other polymerase inhibitors. Of note, EC50 values were higher for A(H3N2) than for A(H1N1). Depending on the antiviral, effective concentrations could differ by 2.5-to 39-fold between strains. Additionally, BXA susceptibilities of influenza viruses have been shown to vary depending on viral strain or subtype [20,44]. Therefore, determinations of EC50 value for each drug were the first step in the systematic evaluation of drug combinations by using an equipotent ratio between BXA and other antivirals.
Secondly, our results show that combinations of BXA with NAIs (oseltamivir, peramivir or zanamivir) have higher potency compared to individual drugs, against A(H1N1)pdm09 and A(H3N2) viruses. Such synergistic effects were confirmed in the HAE model, for which the physiological relevance and predictive value have been previously described by our group and others [33]. These findings are in agreement with those reported by Fukao et al., where in vitro combinations of BXA and approved NAIs also resulted in a synergistic effect against the A/PR/8/34 (H1N1) virus [26]. In that study, the authors further showed that the combination of suboptimal doses of BXM and oseltamivir enhanced therapeutic effects in mice, including reduced lung viral titers.
Combinations of other polymerase inhibitors with NAIs, as well as combinations of two polymerase inhibitors, have also been studied. Ormond and colleagues demonstrated that the combination of oseltamivir with favipiravir negatively impacted viral replication after several passages [45]. These results are in accordance with different animal studies where peramivir-favipiravir [46] or oseltamivir-favipiravir [13] combinations provided various benefits, including improving survival, controlling drug-resistant A(H1N1) strains or delaying mortality in immunocompromised animals.
In our study, we also assessed two polymerase inhibitors: favipiravir and ribavirin (a transcriptase inhibitor of Polymerase Basic 1-PB1) [47,48] in combination with BXA. We obtained synergistic effects for the BXA-favipiravir combination against the two influenza A strains in cell culture, which were further observed in HAE mainly in the case of A(H3N2). In contrast, the BXA-ribavirin combination showed mixed results compared to single-drug treatment, depending on the infection model and viral subtype used. In fact, considering that BXA and ribavirin compounds inhibit two polymerase subunits (PA and PB1, respectively), we suggest that these two agents targeting similar but non-identical functions can interfere or compete in their mechanism of action, resulting in antagonism. Indeed, the PA and PB1 subunits, which interact extensively [49], could be in close proximity during the transcription/replication process, which may limit their accessibility to the two inhibitors simultaneously. In a modeling drug action experiment, Yin et al. [50] demonstrated that, when used in combination, drugs interact in many unexpected ways and show a plethora of different outcomes, leading to drug synergy or antagonism. Biological functions involve many molecules that interact in a network manner. Thus, the final effects of drug combinations also depend on the interactions of their targets in a network manner [50], especially in physiological models, such as HAE, which is constituted by primary differentiated human cells mimicking in vivo epithelia. Indeed, a household study showed that the combination of oseltamivir and zanamivir, administrated within 24 h of onset of symptoms, was more effective in reducing the transmission of influenza compared to monotherapy [51], whereas we reported that the same combination was not more protective than zanamivir monotherapy against A(H1N1)pdm09 and A(H3N2) infections in mice [52]. This difference could be explained by the severity of influenza infections (uncomplicated versus severe infections). Moreover, the optimal dose of each component greatly affects drug interactions [53]. Our combination experiments were based on a fixed ratio calculation. We hypothesize that this method could not apply to the combination of BXA and ribavirin and that a different ratio may lead to another type of drug response. Indeed, different dose combinations may lead to different effects [54]. Interestingly, in some cases, drugs with antagonist effects could also be beneficial against resistant viral strains [50,55].
Additional in vitro studies using resistant strains (I38T or H275Y variants) should be done to better understand drug activities and to evaluate the impact of combination therapy on mutant viral replication. Additionally, further animal studies should be conducted to evaluate the effectiveness of our synergistic combinations in a complete system.
In summary, our results suggest that BXM stands as a potential good candidate for combination therapy with several existing drugs to improve antiviral activity and eventually delay drug resistance of influenza type A viruses. Combination therapies with BXM should be further considered for animal studies and clinical investigations.