Favipiravir and Ivermectin Showed in Vitro Synergistic Antiviral Activity against SARS-CoV-2

Despite the urgent need for effective antivirals against SARS-CoV-2 to mitigate the catastrophic impact of the COVID-19 pandemic, there are still no proven effective and widely available antivirals for COVID-19 treatment. Favipiravir and Ivermectin are among common repurposed drugs, which have been provisionally used in some countries. There have been clinical trials with mixed results, and therefore, it is still inconclusive whether they are effective or should be dismissed. It is plausible that the lack of clear-cut clinical bene�ts was due to the �nding of only marginal levels of in vivo antiviral activity. An obvious way to improve the activity of antivirals is to use them in synergistic combinations. Here we show that Favipiravir and Ivermectin had the synergistic effects against SARS-CoV-2 in Vero cells. The combination may provide better e�cacy in COVID-19 treatment. In addition, we found that Favipiravir had an additive effect with Niclosamide, another repurposed anti-parasitic drug with anti-SARS-CoV-2 activity. However, the anti-SARS-CoV-2 activity of Favipiravir was drastically reduced when tested in Calu-3 cells. This suggested that this cell type might not be able to metabolize Favipiravir into its active form, and that this de�ciency in some cell types may affect in vivo e�cacy of this drug.


Introduction
Up to the end of August 2021, the COVID-19 pandemic has resulted in more than 4.5 million deaths. The emergence of variants with antigenic changes causing vaccine escape has obliterated the hope to eradicate the virus, and the disease will likely continue to be a major problem for the foreseeable future.
Effective antivirals are urgently needed to mitigate the disease burden, especially where vaccine supply is insu cient. In addition, infection and illnesses continue even in a vaccinated population, and effective antivirals would ensure the return to normalcy with further-reduced risk of severe disease and death.
While new antivirals such as monoclonal antibodies have been developed, they are costly and not widely available. Furthermore, those monoclonal antibodies can be effectively used only in the early phase of the disease 1 . Repurposed drugs with antiviral activity have been the main target for developing COVID-19 treatments 2 . Remdesivir is the repurposed antiviral with the most supporting clinical data for its effectiveness and is recommended by many treatment guidelines 3,4 . However, the drug cannot be taken orally and therefore, is not used in mild cases or at an early phase. Early antiviral treatment is very likely to be more effective and will prevent not only death but also progression to severe disease 5 . Several repurposed drugs with in vitro antiviral activity have been tested in clinical trials 2 . Some have been shown to be ineffective, some showed mixed results. Marginal or low levels of e cacy could be a reason for the lack of clear-cut bene t for some of these drugs. Drug combination is a common approach to improve antiviral activity and e cacy, and has been successfully used for treatment of many viral diseases 6 . However, synergistic or at least an additive effect is needed to make a combination more effective than single drugs. We, therefore, tested combinations of common repurposed drugs with anti-SARS-CoV-2 activity to guide further clinical development.
Favipiravir was originally developed as an anti-in uenza drug, and has been shown to have broad antiviral activity against viruses in many families 7 . It is metabolized intracellularly to its active form, Favipiravir-ribofuranosyl-5'-triphosphate 8 . It is believed to inhibit viral replication by inhibiting viral RNA polymerase and by inducing lethal hypermutations 9,10 . Favipiravir had been shown to have anti-SARS-CoV-2 activity, and was tested in clinical trials with mixed results 11 . It is being used for COVID -19 treatment in some countries, including Thailand 12 .
Ivermectin is an anti-parasitic drug with broad antiviral and anti-SARS-CoV-2 activity 13,14 . Although many clinical trials have shown its e cacy in COVID-19 treatment and prophylaxis 15 , many others have indicated that it had no clinical bene t 16 . These trials varied in the dosages, time of initiating the treatment, and whether to take the drug with a meal. As earlier treatment is more likely to be effective and taking Ivermectin with a fatty meal was shown to increase absorption and drug plasma levels 17 . It is still possible that these factors might negatively in uence trial outcomes, and the drug may provide some bene t in speci c settings despite the disappointing clinical trial results. Chloroquine and Niclosamide are also repurposed drugs with anti-SARS-CoV-2 activity. While Chloroquine has been dismissed as ineffective by many 18 , Niclosamide is still an interesting drug that is being tested in clinical trials 19 . Regardless of these clinical trial results, these drugs are common repurposed drugs for COVID-19, which have relatively good safety pro les, are inexpensive and widely available. They should therefore be tested in combinations for in vitro synergistic activity, which may help identify combinations with good potential for further clinical testing.

Results
Single drug treatment against SARS-CoV-2 in Vero E6 cells The main features of antiviral activities of Favipiravir, and the repurposed anti-parasitic drugs; Niclosamide, Ivermectin and Chloroquine are shown in Table 1 and Figure 1. Virus production was determined using both plaque assay and qRT-PCR and is expressed as the percent inhibition relative to the DMSO-treated cell control. The IC 50 values determined by both methods are similar. The IC 50 values calculated from the dose-response determined by qRT-PCR for Favipiravir, Ivermectin, Niclosamide and Chloroquine were 40.49, 1.24, 0.048, and 0.89 µM, respectively. And the IC 50 values calculated from the dose-response determined by plaque assay for Favipiravir, Ivermectin, Niclosamide and Chloroquine were 41.81, 1.23, 0.046 and 0.82 µM, respectively. This con rmed that viral RNA quantitation could accurately determine the infectious viral output for these experiments, and for higher throughput, the qRT-PCR was used for screening of two-drug combinations. The selective indexes of each single drug treatment are also shown in Table 1.  (Fig. 2b, Table 2). The doseresponse matrix shows a signi cant inhibitory effect in a large range of combinations (Fig. 2c). The calculated mean Loewe synergy score of 19.245 accounted for the synergistic effect between Favipiravir and Ivermectin, with the peak score of 33.053 (Fig. 2d). Moreover, the synergy score calculation using ZIP, After removing the virus inoculum, the cells were further maintained in the medium containing drugs for two days. Virus production was determined using qRT-PCR. The SynergyFinder was used to calculate the synergy score of two-drug combinations from different 16 pairwise combinations. The dose-response matrix (c) and the synergy map of two-drug combinations treatment (d) are shown. Areas with synergy score less than -10: the interaction between two drugs is likely to be antagonistic; from -10 to 10: the interaction between two drugs is likely to be additive; larger than 10: the interaction between two drugs is likely to be synergistic.  Table 2). The dose-response matrix shows the increasing inhibitory effect due to the increasing concentrations of Favipiravir and Niclosamide (Fig. 3c). The mean Loewe synergy score of 9.61 accounted for the additive effect between Favipiravir and Niclosamide, with the peak score of 18.87 (Fig. 3d). The synergy scores of 12.43, 20.09 and 12.52 were obtained when using ZIP, HSA, and Bliss independence models, respectively which accounted for the synergy effect of Favipiravir and Niclosamide. All 16 pairwise combinations showed no signi cant cytotoxicity (Fig. 3a, b).

Favipiravir-niclosamide Combination
Vero cells were pre-treated for 1 hr. with twofold serial dilutions of Favipiravir in the presence of different xed concentrations of Niclosamide (a) or alternatively, serial dilutions of Niclosamide in the presence of different xed concentrations of Favipiravir (b). The cells were infected with SARS-CoV-2 at m.o.i. 0.01. After removing the virus inoculum, the cells were further maintained in the medium containing drugs for two days. Virus production was determined using qRT-PCR. The SynergyFinder was used to calculate the synergy score of two-drug combinations from different 16 pairwise combinations. The dose-response matrix (c) and the synergy map of two-drug combinations treatment (d) are shown. Areas with synergy score less than -10: the interaction between two drugs is likely to be antagonistic; from -10 to 10: the interaction between two drugs is likely to be additive; larger than 10: the interaction between two drugs is likely to be synergistic.

Favipiravir-chloroquine Combination
The  Table   2). The dose-response matrix shows the increasing inhibitory effect in the pairwise combinations with high concentrations of Favipiravir and Chloroquine (Fig. 4c). The synergy analysis shows the positive synergy score distributed with the high concentrations of both drugs, while the lower concentrations gave zero and negative synergy scores (Fig. 4d). The mean Loewe synergy score of -0.494 accounted for the additive effect between Favipiravir and Chloroquine, with the peak score of 11.189. The additive effect was shown to be driven by only Favipiravir and Chloroquine at high concentrations. The synergy scores of 0.609 and 0.242 were obtained when using ZIP, and Bliss independence models, which accounted for the additive effect of Favipiravir and Chloroquine. While using the HSA model, the synergy score of 10.064 indicated a small level of synergistic effect. All 16 pairwise combinations showed no signi cant cytotoxicity (Fig. 4a, b). After removing the virus inoculum, the cells were further maintained in the medium containing drugs for two days. The virus production was determined using qRT-PCR. The SynergyFinder was used to calculate the synergy score of two-drug combinations from different 16 pairwise combinations. The dose-response matrix (c) and the synergy map of two-drug combinations treatments (d) are shown. Areas with synergy score less than -10: the interaction between two drugs is likely to be antagonistic; from -10 to 10: the interaction between two drugs is likely to be additive; larger than 10: the interaction between two drugs is likely to be synergistic.   (Fig. 6b, Table 3). The dose-response matrix shows the increasing inhibitory effect in relation to the drug concentrations (Fig. 6c). The mean Loewe synergy score of -0.032 accounted for the additive effect between Favipiravir and Ivermectin, with the peak score of 10.614 (Fig. 6d). Moreover, the synergy score calculation using ZIP and Bliss independence models gave values of 3.601, and 3.664, respectively, which indicated the additive effect between Favipiravir and Ivermectin. However, the synergy score of 10.73 was obtained using the HSA model, which indicated a small synergy effect. All 16 pairwise combinations showed no signi cant cytotoxicity (Fig. 6a,  synergy score less than -10: the interaction between two drugs is likely to be antagonistic; from -10 to 10: the interaction between two drugs is likely to be additive; larger than 10: the interaction between two drugs is likely to be synergistic.  20 . Favipiravir (prodrug) receives an intracellular phosphoribosylation to be in an active form (favipiravir ribofuranosyl-5'-triphosphate; Favipiravir-RTP), which is recognized by RdRp and therefore, inhibits the activity of RNA polymerase 8 .
Recently, it was shown that Favipiravir also inhibits SARS-CoV-2 by inducing lethal hypermutations 21 39 . However, further studies are required to assess the Favipiravir-RTP levels, and the activities of cellular enzymes required for phosphoribosylation in Calu-3 cell and lung cells. Moreover, much higher virus input was used in Calu-3 cells to achieve the stable infection level required for drug treatment in this study. Importantly, the plasma concentration in critical patients was much lower, ranging from 28.2 µM to lower than the limit of quanti cation (>1 µg/mL or 6.37 µM) 40,41 , which does not reach the IC 50 concentration thereby raising a concern about its e cacy. Therefore, a drug combination appoarch should be used to improve the antiviral activity of Favipiravir. In this study, the repurposed anti-parasitic drugs with broad-spectrum antiviral activity were selected for the two-drug combination treatment including Ivermectin, Niclosamide and Chloroquine.
Ivermectin is a derivative of avermectin (macrocyclic lactones found in Streptomyces avermectinius) and is being used for anti-parasite medication by blocking the transmission of neuronal signals of the parasites. Ivermectin was previously reported to have broad-spectrum antiviral activity in various types of RNA viruses. The proposed mechanism of Ivermectin is that the drug molecule targets the host nuclear transport importin α/β1 heterodimer, thereby inhibiting nuclear import of various viral proteins. Several studies demonstrated the antiviral activity of Ivermectin in aviviruses, including Dengue virus, West Nile virus, and Zika virus, by inhibiting nuclear import of NS5 protein 13,42,43 . In Human Immunode ciency virus type 1 (HIV-1) and In uenza A viruses, it was shown that Ivermectin inhibited integrase and viral ribonucleoprotein nuclear import, respectively 44,45 . However, the exact mechanism of antiviral activity of Ivermectin has still not been well described.
However, higher concentrations of Ivermectin were detected in various tissues, including fat, skin and nodular tissues 47 , suggesting a potential of Ivermectin in antiviral therapy. Additionally, a combined phase 2/3 clinical study reported that Ivermectin was safe and reduced the level of plasma nonstructural protein 1 in dengue patients 48 .
To improve the antiviral activity of these drugs, a drug combination approach was used. Niclosamide is an anthelmintic drug widely used in humans to treat tapeworm infections. The proposed mechanism is inhibiting oxidative phosphorylation and stimulating adenosine triphosphatase activity in the mitochondria of the tapeworm 51 . From several repurposed drug screenings, Niclosamide was identi ed as a multifunctional drug due to its ability to regulate multiple pathways including mTORC1, showed that Chloroquine treatment in COVID-19 patients was ineffective 66 . In this study, the Favipiravir-Choroquine combination only demonstrated a minimum additive effect. It was previously shown by mathematical modeling that drugs working at the entry step performed poorly in synergism with other drugs 67 . The lack of good synergism with Chloroquine supports with this model. A previous study demonstrated an antagonistic effect of the Hydroxychloroquine and Remdesivir combination against SARS-CoV-2, while Nitazoxanide (that should be similar to Niclosamide) made good synergy with various drugs 68 .
The drug combination approach reported here is used to maximize treatment e cacy. For antiviral treatment, the combination of the antiviral drugs with drugs utilizing different modes of action or acting on cellular targets also helps minimize drug resistance and toxicity 69  SARS-CoV-2 (strain SARS-CoV-2/01/human/Jan2020/Thailand) was previously isolated from nasopharyngeal swabs of a con rmed COVID-19 case in Thailand 72 . The virus was propagated in Vero E6 cells. The supernatants containing virus were harvested by centrifugation to remove cell debris and stored at -80°C. The viral titer was determined by plaque assay in Vero E6 cells or 50% tissue culture infectious dose (TCID 50 ) endpoint dilution assay in Calu-3 cells.

SARS-CoV-2 infection
Vero E6 or Calu-3 cells were seeded in culture plates at a density that allowed 100% and 70% con uence, respectively.The culture media were removed and the cells were incubated with 2%FBS-culture medium One-step qRT-PCR for SARS-CoV-2 The one-step qRT-PCR was used to detect the RNA of SARS-CoV-2 directly from the virus supernatants, without RNA extraction. This assay was validated by comparing the results with viral quantitation using the plaque assay. The procedure was described elsewhere 74 . For the sample preparation, the collected virus supernatants were heat inactivated at 70 o C for 20 min. Then the samples were diluted 1:10 with DNase/RNase free distilled water. One-step qRT-PCR was performed using the Power SYBR one-step kit (Applied Biosystems) and the LightCycler 480 (Roche, LC480). The one-step RT-PCR master mix was prepared following the kit's instructions for a 10 µl reaction volume. and melting curve step to determine the speci city of the PCR product from the melting temperature (T m ) (95 o C for 30s, 60 o C for 30s).
The threshold cycle (Ct) values were calculated from raw uorescence data using Abs Quant/2 nd derivative method. The T m calling analysis was performed to exclude reactions with non-speci c ampli cation by comparing with no template control and the product ampli ed from the positive control. The inhibition of SARS-CoV-2 production in drug treated cells was relative to the cells treated with 0.5% DMSO.

Evaluation of the combination synergy
The SynergyFinder web application was used to analyze and visualize the degree of combination synergy. The synergy scores of two-drug combinations were analyzed by comparing the observed drug combination response (percent inhibition) against the expected response, calculated using reference models 75 . The reference models used in this study, including the Loewe additivity (Loewe), Zero Independence Potency (ZIP), Highest Single Agent (HSA), and Bliss independence models 76 . For the synergy score less than -10: the interaction between two drugs is likely to be antagonistic; from -10 to 10: the interaction between two drugs is likely to be additive; larger than 10: the interaction between two drugs is likely to be synergistic.

Statistical analysis
All independent experiments were performed in triplicate and the data are shown as mean ± SD. The IC 50 and the CC 50 were calculated by non-linear regression analysis using GraphPad Prism 9 (GraphPad Software, Inc., CA). The selective index (SI) was calculated from the ratio between CC 50 and IC 50 .