Direct inhibitory effect on viral entry of influenza A and SARS‐CoV‐2 viruses by azithromycin

Abstract Objectives Using strategy of drug repurposing, antiviral agents against influenza A virus (IAV) and newly emerging SARS‐coronavirus 2 (SARS‐CoV‐2, also as 2019‐nCoV) could be quickly screened out. Materials and Methods A previously reported engineered replication‐competent PR8 strain carrying luciferase reporter gene (IAV‐luc) and multiple pseudotyped IAV and SARS‐CoV‐2 virus was used. To specifically evaluate the pH change of vesicles containing IAV, we constructed an A549 cell line with endosomal and lysosomal expression of pHluorin2. Results Here, we identified azithromycin (AZ) as an effective inhibitor against multiple IAV and SARS‐CoV‐2 strains. We found that AZ treatment could potently inhibit IAV infection in vitro. Moreover, using pseudotyped virus model, AZ could also markedly block the entry of SARS‐CoV‐2 in HEK293T‐ACE2 and Caco2 cells. Mechanistic studies further revealed that such effect was independent of interferon signalling. AZ treatment neither impaired the binding and internalization of IAV virions, nor the viral replication, but rather inhibited the fusion between viral and vacuolar membranes. Using a NPC1‐pHluorin2 reporter cell line, we confirmed that AZ treatment could alkalize the vesicles containing IAV virions, thereby preventing pH‐dependent membrane fusion. Conclusions Overall, our findings demonstrate that AZ can exert broad‐spectrum antiviral effects against IAV and SARS‐CoV‐2, and could be served as a potential clinical anti‐SARS‐CoV‐2 drug in emergency as well as a promising lead compound for the development of next‐generation anti‐IAV drugs.


| INTRODUC TI ON
Influenza viruses, members of the Orthomyxoviridae family, are persistent and seasonal epidemical health threats to human beings. 1 Influenza viruses consist of influenza A, B, C and D virus (IAV, IBV, ICV, IDV). 2 IAV is the pathogen of most seasonal influenza and influenza pandemics. 1 Classification of IAV is based on the antigenic properties of hemagglutinin (HA) and neuraminidase (NA). The H1N1 and H3N2 subtypes cause seasonal and pandemic infections, while the highly pathogenic avian H5N1 and H7N9 subtypes cause severe mortality. 3 Unlike IAV, coronaviruses rarely cause serious pandemics, but Unfortunately, there are still no regulatory-approved drugs for COVID-19 patients. Although remdesivir and chloroquine have shown potential antiviral ability against SARS-CoV-2 in vitro, 7 the clinical antiviral effects and safety have not been fully verified. 8,9 Interestingly, azithromycin (AZ) showed synergistic anti-SARS-CoV-2 effect with hydroxychloroquine in vitro, but the mechanism is still unclear. 8 The two conventional strategies to fight against influenza pandemics and epidemics are small-molecule antiviral drugs and vaccines. [10][11][12][13] However, influenza vaccines must be reformulated yearly to match with the antigens of circulating viruses. 14,15 Moreover, vaccine production usually lags behind identification of circulating virus for 6 months. 15 Drug repurposing, also known as drug rescue or drug repositioning, refers to the re-examination of existing drugs for new therapeutic purposes. 16 Compared to conventional drug development, drug repurposing possesses significant advantages, such as lower risk of failure and shorter research and development time period due to the known pharmacokinetic properties, tolerance and toxicity of approved drugs.
In the present study, we reported the repurposing of AZ, a macrolide antibiotic, is a potential antiviral drug candidate for high-pathogenic IAV and newly emerging SARS-CoV-2. Macrolide antibiotics have well-established antibacterial, 17,18 anti-inflammatory effects [19][20][21] and certain antiviral effects against rhinovirus. [22][23][24] In addition, it has been recently reported that AZ could reduce Zika viral proliferation and cytopathic effects induced by the virus in glial cell lines and human astrocytes. 25 Furthermore, Li et al also demonstrated that AZ upregulates the expression of IFN-I/III and some of their downstream interferon-stimulated genes (ISGs) in response to Zika virus infection. 26 In spite of the reported antiviral activity of AZ, the mechanism of AZ against viral infection is still not understood. In the current study, we established an A549 cell line with endosome-specific pHluorin2 expression to quantitatively analyse the effect of AZ on acidification of vesicles containing virions and found that AZ could exert antiviral activity through disturbing the acidification of endosomes containing IAV. Moreover, we found that AZ could inhibit the entry of SARS-CoV-2 pseudovirus in HEK293T-ACE2 and Caco2 cells. Taken together, we evaluated the previously unknown antiviral mechanism of AZ against IAV and SARS-CoV-2, and provide a potential candidate for future clinical drug application against IAV and SARS-CoV-2.

| Screening with IAV-luc
Briefly, A549 cells were pre-treated for 8 hours with the indicated drugs, infected with IAV-luc at an MOI of 0.01 for 24 hours. Then, the viral titres were measured by detecting the luciferase activity of supernatant by a microplate reader.

| Establishment of HEK293T-ACE2 cells
As ACE2 has been identified as the host receptor of SARS-CoV and SARS-CoV-2, we constructed the human ACE2 stable expressed HEK293T (HEK293T-ACE2) with lentiviral mediated gene transduction.

| Cytotoxicity assay
The cytotoxicity of the candidate drugs in A549 and HEK293T-ACE2 cells was detected with the cell counting kit-8 (CCK8) after 72 hours treatment.

| Pseudovirus preparation
For IAV, VSV and Ebola entry assays, pseudovirus based on HIV were prepared as previously reported. 27

| Pseudovirus entry assay
For entry assays, pseudovirus entry assay was carried out as previously reported. 27 Briefly, A549, HEK293T-ACE2 or Caco2 cells were pre-treated for 8 hours with the indicated drugs, infected with pseudovirus for 72 hours and lysed for the luciferase assay. and Renilla luciferase expression plasmid as an internal transfection control, as described previously. 28

| Quantitative real-time PCR
Total RNA was obtained using the Ultrapure RNA kit (cwbiotech).

| IAV labelling
To locate or quantitative analyse the IAV particle, WSN stocks were diluted in PBS to 0.1 mg/mL and labelled with Dil (Thermo Fisher Scientific) or R18 (Thermo Fisher Scientific) and SP-DiOC18 (Thermo Fisher Scientific) at RT for 1 hour. The labelled virus particles were filtered through a 0.22 μm-pore filter (Millipore) and stored at 4°C in the dark till used.

| IAV binding, internalization and membrane fusion assay
For binding assay, A549 cells were pre-treated with indicated drugs

| Endosome acidification assay
Total acidification was assessed with Lyso-Tracker Red (Beyotime) as a probe for low-pH organelles. Cells were pre-treated with the indicated drugs at 37°C for 2 hours and then incubated with 50 nmol/L Lyso-Tracker Red for 30 minutes. Cells were analysed by fluorescence microscopy.
The pH calibration curve was generated as described previously. 30,31 The buffers for generating the pH calibration curve

| Statistical analysis
The results were presented as the mean ± SEM. The unpaired twotail Student's t test was used to determine the statistical significance. A P-value < .05 was considered to be statistically significant.
The Prism software program for Windows (GraphPad Software) was used to perform all calculations.

| AZ is identified as a potential inhibitor of IAV in vitro
To rapidly and robustly screen candidate drugs with anti-IAV activity, a previously reported engineered replication-competent PR8 strain carrying luciferase reporter gene (IAV-luc) was used. 33 A549 cells were pre-treated with candidate drugs for 8 hours, and then, they were in-  Figure 1E). To determine whether other macrolide type of antibiotics also possess anti-IAV activity, we detected the anti-IAV activity of multiple macrolide antibiotics, including AZ, erythromycin, roxithromycin, midecamycin, spiramycin, acetylspiramycin, clarithromycin and dirithromycin. While all these macrolide antibiotics significantly inhibited the IAV-luc infection, AZ is one of the most potent macrolide antibiotics with anti-IAV activity ( Figure 1F).

Consistently, AZ also inhibited the infection of IAV-luc in a number
of cell lines of different histological origins (including 293T, Hela and A549 cell lines) and the infection of the wild-type WSN in A549 cell line ( Figure 1G,H). To evaluate the antiviral activity of AZ to various IAV subtypes, we constructed WSN, H5N1 and H7N9 pseudovirus and found that AZ inhibited the infection rates of the pseudovirus of all three IAV subtypes ( Figure 1I). Taken together, these results demonstrate that AZ could inhibit the infection of different IAV subtypes in different cell types.

| AZ has antiviral activity against SARS-CoV-2 pseudovirus in vitro
To screen the candidate anti-SARS-CoV-2 drugs in BSL-2 laboratory, we prepared the HIV-and MLV-based pesudovirus according to a previously reported method 34 with several modifications. As  Our results thus identified AZ as a potential anti-SARS-CoV-2 candidate drug in vitro.

| AZ exerts its antiviral effect independently of the activation of interferon pathway
As the key components of innate antiviral immunity, interferons exert Taken together, these results prove that the major part, if not all, of the antiviral effect of AZ is independent of the interferon signalling.  (Figure 4E,F). Consistently, a lipophilic dye-based fluorescence dequenching assay using R18 (red) and SP-DiOC18 (green, fixable) demonstrated that the fusion of viral and vacuolar membranes was also inhibited by AZ treatment (Figure 4G,H). Moreover, these membrane fusion results are consistent with the pseudovirus assay ( Figure 4I). Therefore, our data demonstrate that the antiviral effect of AZ relies on the alkalinization of acid vesicles to inhibit the IAV entry.

| AZ inhibits the acidification of vesicles containing IAV virions
Measured HA activation pH values across all subtypes and species range from 4.6 to 6.0. 38 To quantify the alkalinization of acidic vesicles by AZ treatment, LysoSensor Yellow/Blue DND-160, a ratiometric pH sensing dye, was used to quantify the pH of acidic vesicles.
Similar to bafilomycin A1 and NH 4 Cl, AZ could dose-dependently increase the pH value of acidic vesicles ( Figure 5A-B). However, this result only showed the average pH of all the acidic vesicles (including early endosome, recycled endosome, late endosome, lysosome, autophagosome, golgi), but not the specific vesicles where IAV locate. To specifically evaluate the effect of AZ on the pH of vesicles containing IAV, we constructed an A549 cell line with endosomal and lysosomal expression of pHluorin2 (an enhanced, ratiometric, pH-sensitive GFP variant). 39 We fused pHluorin2 with the N terminal (signal peptide), transmembrane region 1 (TM1) and C terminal (C-tail) of NPC1, which is expressed specifically in the endosome and lysosome ( Figure 5C-E). Consistent with LysoSensor Yellow/Blue DND-160 assay, pHluorin2 assay indicated that AZ treatment significantly increased the pH of NPC1-labelled endosome and lysosome ( Figure 5F). Further, we specifically detected the pH of vesicles containing Dil-labelled IAV and found that AZ treatment significantly alkalized the vesicles containing Dil-labelled IAV ( Figure 5G). Thus, AZ treatment could alkalize the vesicles containing IAV virions out of the proper pH rang of HA activation.

| D ISCUSS I ON
In this study, we identified that AZ exhibited a powerful antiviral effect against various IAV subtypes, including H1N1, H5N1 and H7N9. Excitingly, AZ also possesses in vitro anti-SARS-CoV-2 and anti-SARS-CoV activities in pseudovirus inhibition assays. As AZ has also reported having antiviral effects against rhinovirus, Ebola virus and Zika virus, [22][23][24][25]40 AZ is therefore a highly promising candidate CoV-2 in vitro and in vivo. 8,[42][43][44] However, other studies reported inconsistent results and adverse effects of HCQ and AZ in COVID-19 patients. [45][46][47] Therefore, more comprehensive clinical and basic research is needed to clarify the anti-SARS-CoV-2 effect of AZ.
As reported in rhinovirus and Zika virus study, AZ treatment also increased the virus-induced interferon mRNA expression, 26,[48][49][50] implying that AZ may exert anti-IAV activity partially through activation of the interferon signalling induced by IAV infection. However, the defection of interferon signalling did not affect the antiviral effect of AZ ( Figure 3B,C), suggesting that the interferon-mediated anti-IAV effect of AZ is dispensable. Moreover, AZ alone did not affect the mRNA level of IFNβ or ISGs, but inhibited the increase of mRNA level stimulated by WSN infection, which is presumably due to the less nucleocapsid released into the cytoplasm ( Figure 3D-F).
Inconsistencies of interferon mRNA expression between these studies are presumably due to the difference of virus classification and the host cells.
As reported in the recently published literature, 51 SARS-CoV-2 pseudovirus entries into the HEK293T-ACE2 mainly through endocytosis, which is regulated by PIKfyve, TPC2 and cathepsin L.
Moreover, it has been shown that SARS-CoV S-pseudotyped virions use the endosomal protease cathepsin L to infect cells. 52,53 Therefore, the activity of cathepsin L (a lysosomal acid cysteine protease) could be weakened by the alkalinization of endo-lysosome, which is consistent with our results that chloroquine and ammonium chloride almost completely suppress SARS-CoV-2 pseudovirus infection in HEK293T-ACE2 ( Figure 2B). Therefore, proper acidic environment maybe a limiting factor for SARS-CoV-2, but this should be verified with live virus and in more cell

models.
Recently, it is reported that AZ inhibits influenza A (H1N1) pdm09 virus infection by interfering with virus internalization process, 54  to the cationic amphiphilic drugs (CAD), which could accumulate in and alkalize acid vesicles. 58 Therefore, AZ may act as a weak base to prevent the acidification of the endosome. However, this hypothesis needs to be carefully verified.
As a critical host factor hijacked by several enveloped viruses, proper acidic environment is a limiting factor for conformation change and/or priming of fusion mediating glycoprotein, suggesting the acidification of endosome and lysosome could be a potential target for antiviral drug development. Unlike to targeting viral proteins, targeting the host factors speculatively possesses a lower likelihood of drug resistance. 59 Therefore, by targeting the acidification of endosome and lysosome, AZ could possess a broad-spectrum antiviral effect and higher barrier to drug resistance.
In summary, we identified AZ as a broad-spectrum antiviral against IAV and SARS-CoV-2, and a potential clinical anti-SARS-CoV-2 drug used for emergency and a promising candidate for the development of clinical anti-IAV drug.

CO N FLI C T S O F I NTE R E S T S
The authors declare no competing financial interests.

AUTH O R CO NTR I B UTI O N
XD, XZ, FM, CH, WO, YH, YG and XZ performed the experiments; XD, XZ, FX and FXQ analysed data and wrote the manuscript; and FX and FXQ was responsible for research design, strategy and supervision.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.