Berbamine inhibits the infection of SARS-CoV-2 and aviviruses by compromising TPRMLs-mediated endolysosomal tracking of viral receptors

Positive-sense single-stranded ((+)ss) RNA viruses are among the leading causes of human and animal infectious diseases in the world, but so far, no effective antiviral agents are available to treat these infections. Here we found that several bis- benzylisoquinoline alkaloids (e.g. berbamine), potently inhibited the infection of coronaviruses (e.g. SARS-CoV-2 and MERS-CoV), aviviruses (e.g. JEV, ZIKV and DENV), and enteroviruses (e.g. EV-A71) in host cells. Moreover, berbamine protected mice from lethal challenge of JEV. We also found that berbamine inhibited TRPMLs (Ca2+ permeable non-selective cation channels in endosomes and lysosomes), which compromised the endolysosomal tracking of viral receptors, such as ACE2 and DPP4. This led to the increased secretion of these receptors via extracellular vesicles and the concomitant decrease in their levels at the plasma membrane, thereby preventing (+)ss RNA viruses from entering the host cells. In summary, these results indicate that bis- benzylisoquinoline alkaloids such as berbamine, can act as a pan-anti-(+)ss RNA virus drug by inhibiting TPRMLs to prevent viral entry.

The infections of coronavirus cause a number of human and animal diseases, including infections of the gastrointestinal tract and upper respiratory system 9, 10, 11. Although majority of coronavirus infections are asymptomatic or mild, SARS-CoV is the causative for SARS outbreak in 2002 with a total of 8,098 cases and 774 fatalities (i.e. 9.6% of cases) 6, 7, 12, and MERS-CoV is the pathogen responsible for the outbreak of MERS, with fatality rate close to 35% 9, 13. We are currently being affected by coronavirus disease 2019 or COVID-19 due to infection by the SARS-CoV-2. So far, this has been transmitted around the world with more than 3 million cases of infection and a mortality rate close to 7%. COVID-19 has put much of the world into lockdown, and this public health emergency highlights the urgent need for the effective antiviral agents.
With regards to the common aviviruses, DENV infects between 100 and 500 million people each year and causes about 50 million febrile illnesses, including more and more cases of hemorrhagic fever 1, 14.
In contrast, a large proportion of ZIKV infections is asymptomatic and the main features of infected patients who do show symptoms include just mild fever, headache and arthralgia 15,16,17. However, the biggest concern with ZIKV, is the coincidence between the onset and rise of this viral infection and the increase in severe congenital infection leading to microcephaly in new-born babies reported in Brazil 16.
With regard to JEV, over the past few decades, various inactivated and living-attenuated forms of this virus have been successfully used for the production of vaccines in many countries, and the concomitant escalation in immunization programs has markedly decreased the burden of this disease. However, it has been estimated that annually JEV still affects about 68,000 people and results in 10,000-15,000 deaths, with 30%-50% of the survivors showing neurological sequelae 18,19. The widespread global social and economic impact of avivirus infection also urgently requires effective treatment interventions. However, so far, despite commercially available vaccines for yellow fever, Japanese encephalitis, and neonatal encephalitis, there is no effective clinical treatment for any avivirus infection 4, 20.
EV-A71 is one of the most well-studied members of enterovirus family. EV-A71 easily spreads among children and may cause diarrhea, rashes, and hand, foot and mouth disease (HFMD), with the possibility of developing into severe neurological disease. Previously, many EV-A71 epidemics have taken place in Asian countries. For example, in 1998, there was a large EV-A71 outbreak in Taiwan, which led to the death of 78 children out of 405 hospitalized cases 21.

Results
Ca2+ signaling is required for the entry of (+)ss RNA virus into host cells. It has been reported by a number of groups that Ca2+ signaling is essential for virus infection 22, 23, 24, 25, 26, 27, 28, 29, 30. Here, we added to this accumulating evidence by demonstrating that chelation of intracellular Ca2+ with BAPTA-AM markedly suppressed the JEV-induced expression of JEV envelope protein (JEV-E) (Fig. 1A).
Since it has been reported that a Ca2+-dependent signaling pathway regulates the entry of the in uenza A virus 27, 28, we subsequently studied if BAPTA-AM inhibits internalization of JEV. We performed in situ RNA hybridization to detect the positive strand RNA of JEV, the viral genome (vRNA), and showed that JEV was internalized as early as 20 min after infection, and a stronger vRNA signal was detected by 80 min after viral infection. However, when cells were treated with BAPTA-AM, no JEV vRNA was detected inside cells (Fig. 1B). We also examined whether the removal of extracellular Ca2+ might affect ZIKV infection of host cells by performing both in situ RNA hybridization assay to detect the vRNA of ZIKV, and immunocytochemistry to investigate the localization of ZIKV envelope protein (ZIKV-E). We showed that at 90 min after viral infection, in the presence of extracellular Ca2+ ZIKV vRNA was detected inside host cells (middle panel in Fig. 1C), but in the absence of extracellular Ca2+, no ZIKV vRNA was detected inside these cells (right panel in Fig. 1C). Moreover, in the absence of extracellular Ca2+, ZIKV virions, (manifested by ZIKV-E immunostaining), were detected at the plasma membrane of the host cells (right panel in Fig. 1C). These results indicate that Ca2+ signaling is required for the entry of (+)ss RNA viruses into host cells.
Berbamine inhibits JEV, ZIKV, DENV, SARS-CoV-2, and MERS-CoV infection in vitro. Double stranded RNAs (dsRNAs) are an intermediate in viral genome replication. They are generated by viral RNA polymerases following the infection of the (+)ss RNA viruses. These dsRNAs can be detected by performing immuno uorescence staining with an anti-dsRNA antibody 33, 34, 35. Thus, we used a combination of in situ RNA hybridization for the negative strand viral RNA ((-)RNA), and dsRNA immunostaining in the host cells to investigate the infection of the HeLa cells with EV-A71 or JEV. We found that 7h after virus infection, EV-A71 (-)RNA exhibited strong co-localization with dsRNA (Fig. S1A). DsRNA puncta were also detected in cells infected with JEV, and these puncta exhibited weak co-localization with EEA1 (an early endosome marker) or LAMP1 (a late endosome and lysosome marker) (Fig. S1B). This weak colocalization between dsRNA and endosomal or lysosomal markers suggests that for JEV, viral RNA replication does not occur at either the endosomes or lysosomes. Nevertheless, these results con rm that dsRNA immunostaining can be used to detect (+)ss RNA virus infection in host cells.
We, subsequently, developed a high-content image detection platform to detect and quantify the dsRNA immunostaining in virus-infected cells by automated uorescence microscopy ( Fig. 2A). We then applied this high-content image platform to assess the anti-viral activity of various compounds reported in the literature as being modulators of Ca2+ channels or Ca2+ signaling. Thus, A549 cells were seeded in 96well plates, pretreated with different concentrations of various compounds for 1 h, and then infected with JEV, ZIKV or DENV. After 48 h of infection, the cells were xed and subjected to dsRNA immunostaining and DAPI staining, after which images were captured with the CellInsight CX7 high-content screening platform and analyzed in HCS Studio™ 3.0 to quantify the percentage of infected cells ( Fig. 2A).
By applying this high content image platform, we found that berbamine, a bis-benzylisoquinoline alkaloid isolated from the traditional Chinese medicine, Berberis, (which has reported effects on Ca2+ signaling), signi cantly inhibited the infection of JEV (Fig. 2B), ZIKV (Fig. 2C), and DENV (Fig. 2D). In A549 cells, the EC50 of berbamine against JEV and ZIKV was approximately 1.62 mM and 2.17 mM, respectively. Consistently, when we performed a virus titration assay, we could con rm that berbamine signi cantly inhibited the infectious progeny viral particle production of both ZIKV (Fig. 2E) and JEV (Fig.  2F). Taken together, these data not only demonstrate that berbamine is an effective anti-avivirus agent, but also con rm the e cacy of our high-content image screening platform.
In addition to aviviruses, we also assessed the activity of berbamine against coronavirus (e.g. MERS-CoV and SARS-CoV-2) infection in vitro. We treated primary human lung broblasts with berbamine and then infected them with MERS-CoV, after which we performed qRT-PCR to measure the amount of intra-or extracellular viral RNA. We showed that berbamine signi cantly decreased both the intracellular (Fig. 3A) and extracellular (Fig. 3B) level of MERs-CoV RNA. We also assessed the anti-SARS-CoV-2 activity of berbamine in VeroE6 cells, and found that berbamine signi cantly inhibited the viral yield, as quanti ed by qRT-PCR (Fig. 3C). The EC50 of berbamine against SARS-CoV-2 in Vero cells is ~2.3 mM. In summary, these data indicate that in addition to being an effective anti-avivirus agent, berbamine is also an effective anti-SARS-CoV-2 and MERS-CoV drug.
In addition to berbamine and tetrandrine, we also assessed the ability of three additional bisbenzylisoquinoline alkaloids (i.e., isotetrandrine, fangchinoline, and E6 berbamine; We also performed the high-content image assay to assess the effect of berbamine or tetrandrine on EV-A71 infection. As shown in Figure S4B and S4C, treatment with berbamine or tetrandrine signi cantly inhibited EV-A71 infection in RD cells, with EC50 values of ~17. 3 mM or ~7.6 mM, respectively. Taken together, these data indicate that these bis-benzylisoquinoline alkaloids act as pan anti-(+)ss RNA viral agents.
Berbamine can protect mice from a lethal challenge of JEV. We subsequently assessed the cytotoxicity of berbamine ( Fig. 4A) and tetrandrine ( Fig. 4B) in different cell lines, and found that the IC50 for berbamine in different cell lines ranged from ~90 μM to ~126 μM, whereas the IC50 of tetrandrine was from ~20 μM to ~28 μM (Fig. 4C). We also calculated the selectivity index (SI) of berbamine or tetrandrine for JEV and ZIKV infection. The SI helps to determine the window between cytotoxicity and antiviral activity by dividing the EC50 over its IC50 value (i.e., IC50/EC50). As shown in Fig. 4D, the SI values of berbamine were much higher than tetrandrine for these two viruses.
We, thus, assessed the protective effects of berbamine against JEV infection in a mouse model. As shown in Figs. 4E and 4F, berbamine (15 mg/kg, twice per day) indeed protected mice from a lethal challenge of JEV, as demonstrated by the higher survival rate (i.e., 75% in the berbamine-treated group versus 12.5% in the control group) and the better body weight recovery. In summary, these results suggest that in vivo, berbamine is also a potential anti-(+)ss RNA viral drug or at least an anti-aviviral drug.
Berbamine prevents the entry of (+)ss RNA viruses by decreasing the levels of their receptor(s) at the plasma membrane in host cells. Since chelation of intracellular Ca2+ by BAPTA-AM rendered cells resistant to JEV internalization (Fig. 1B), we examined whether berbamine might block the entry of viruses into cells. Thus, A549 cells were treated with berbamine for 1h, and then then incubated with JEV on ice for 1h before putting back to 37 0C for another 80 min before xation. In situ RNA hybridization was subsequently performed to detect the JEV vRNA. As expected, JEV vRNA was only detected in untreated control cells, and not in cells pretreated with berbamine or tetrandrine (Fig. 5A). Therefore, these data demonstrate that berbamine inhibits the entry of JEV into host cells.
We next investigated the mechanism by which berbamine prevents (+)ss RNA viruses from entering host cells, and focused on JEV. The early stages of the avivirus life cycle involve the attachment of viral particles to the surface of the cell membrane. These viral particles bind to membrane receptors and then enter into the cell via receptor-mediated endocytosis, after which the viral RNA is released from the endosomes, due to the fusion of avivirus E proteins with the endosome membrane. To search for potential JEV receptor(s) on host cells, we generated pools of siRNA, and individually knocked down potential virus receptors reported in the literature. In this way, we found that knockdown of the low-density lipoprotein receptor (LDLR) markedly decreased JEV-induced JEV-NS1 protein expression in A549 cells ( Fig. 5B and 5C), which suggests that the LDLR might be involved in JEV infection. The LDLR is a calcium binding protein, which has been reported to function as a receptor for the hepatitis C, rhino-, and vesicular stomatitis viruses 39, 40, 41, 42, 43. We then assessed whether LDLR is involved in JEV entry into cells. Thus, control or LDLR knockdown A549 cells were incubated with JEV on ice for 1h and were then incubated with warm medium at 37 0C for another 80 min followed by in situ RNA hybridization to detect the JEV vRNA. As expected, JEV vRNA was only detected in control cells, not in LDLR knockdown cells (Fig. 5D), indicating that LDLR is required for JEC entry. In addition, we incubated live A549 cells on ice with an anti-LDLR antibody and an Alexa Fluor 488-tagged secondary antibody to label the cell surface LDLR, followed by infection with JEV. Thereafter, cells were incubated at 37˚C, xed at the indicated time points, and subjected to in situ RNA hybridization to detect the vRNA of the JEV (Fig. 5E). We showed that JEV infection induced the internalization of LDLR (green puncta in Fig. 5E), and the internalized LDLR-positive endosomes exhibited strong co-localization with the vRNA particles (red puncta in Fig. 5E) by 30 min after virus infection. The vRNA particles then became dissociated from the LDLR-positive endosomes by 60 min after virus infection (the lower panel in Fig. 5E), and we suspect that this is likely due to the release of viral genome RNA from the endosome. In addition, JEV infection markedly increased the level of LDLR in a time-dependent manner, whereas berbamine treatment abolished this increase (Fig. 5F). These results suggest that LDLR might be one of the receptors used by JEV to gain entry into the host cells, and berbamine decreases the level of LDLR at the plasma membrane thereby rendering cells refractory to JEV infection. We conducted LDLR immunostaining in cells treated with or without berbamine, and con rmed that in the presence of the drug, the level of LDLR at the plasma membrane was signi cantly decreased, but there was little effect on cell morphology (Fig. 5G).
Since SARS-CoV-2 and MERS-CoV are known to target ACE2 44 and DPP4 45, respectively, for their entry into host cells, we suspect that the anti-SARS-CoV-2 and anti-MERS-CoV activity of berbamine might also be due to its effect on these two receptors at the cell surface. By immunolabeling ACE2 or DPP4 in cells treated with or without berbamine, we showed that the drug indeed signi cantly decreased the level of both of these receptor proteins at the plasma membrane ( Fig. 5H and 5I). In summary, these results suggest that berbamine prevents (+)ss RNA viruses from entering host cells by decreasing their respective receptors levels at the plasma membrane. Berbamine inhibits endolysosomal tra cking and induces the secretion of extracellular vesicles to decrease viral receptor levels at the plasma membrane. Since berbamine decreased the levels of LDLR at the plasma membrane (Fig. 5F), we assessed whether berbamine might affect the endolysosmal tra cking of LDLR by performing LDLR and LAMP1 coimmunostaining in cells treated with or without berbamine. In brief, A549 cells were rst incubated with an anti-LDLR antibody on ice for 90 min. The internalization of the LDLR-antibody complex was then initiated when cells were warmed to 37°C 46. In the control cells, after ~30 min to 1 h the internalized LDLR-antibody complex was found in the late endosomes or lysosomes, as shown by the level of colocalization between LDLR and LAMP1, and by ~3 h, most of the internalized LDLR was degraded (top panel in Fig. 6A). However, in berbamine-treated cells, the LDLR-antibody complex was internalized normally but failed to be sent to lysosomes for degradation (bottom panel in Fig. 6A).
Since the endolysosomal degradation of LDLR was compromised in the berbamine-treated cells (Fig. 6A), we reasoned that this should lead to an increase in the total amount of LDLR in these cells. However, we found that berbamine markedly decreased the total amount of LDLR in cells (Fig. 6B). We speculated that the reduced levels of LDLR that occurred in berbamine-treated cells might be due to an increase in the secretion of the endosomes containing these membrane receptors, out of cells. To verify this possibility, we quanti ed the concentration of extracellular vesicles (EVs) in the cell culture medium of control or berbamine-treated cells using a nanoparticle analyzer. As expected, berbamine signi cantly promoted the secretion of EVs (Fig. 6C). We then examined whether these EVs contain elevated levels of membrane receptors in the berbamine-treated group when compared with the control group. Thus, EVs in the culture medium from the control and berbamine-treated cells were collected by ultracentrifugation, and the protein levels of several previously reported cell membrane receptors or membrane binding proteins for (+)ss RNA viruses (e.g., ITGB3 47,48,49,50,51,SCARB1 52,53,LDLR 54,55,56,57 and ANXA2 58,59,60), were analyzed by immunoblot analysis. We showed that the levels of LDLR, ANXA2, ITGB3, and SCARB1, similar to TSG10 (which is an exosome surface protein marker), were all markedly increased in EVs collected from the berbamine-treated cell culture medium, when compared with the control group (Fig. 6D). Similarly, we examined the effect of berbamine on the levels of ACE2 and DPP4 in cells and EVs. We found that when compared with the control cells, berbamine treatment markedly decreased the levels of ACE2 and DPP4 in both A549 cells (Fig. 6E)and Vero cells (Fig. 6F), but markedly increased the level of both in EVs (Fig. 6G). Taken together, these results suggest that berbamine inhibits the endolysosomal tra cking of the plasma membrane receptors of (+)ss RNA viruses. This leads to an increase in the level of secretion of these receptors via EVs and a concomitant decrease in the level of the receptors at the plasma membrane, thereby preventing these viruses from entering host cells.
Berbamine prevents JEV infection by inhibiting lysosomal TRPMLs. It has been previously reported that tetrandrine prevents the entry of Ebola virus into host cells by blocking two-pore channels (TPCs) 25 and TPCs has also been shown to mediate MERS-CoV pseudovirus translation 61. TPCs are Ca2+-permeable non-selective cation channels in the endo-lysosomal system 62, 63. Thus, we rst examined whether berbamine might affect the lysosomal Ca2+ levels by assessing the ability of Gly-Phe β-naphthylamide (GPN) to trigger Ca2+ release from lysosomes in cells treated with or without berbamine. We found that berbamine signi cantly mitigated the GPN-induced cytosolic Ca2+ increase, which suggests that it inhibits lysosomal Ca2+ channels (Fig. 7A). However, Ned-19, an antagonist of nicotinic acid adenine dinucleotide phosphate (NAADP), which can inhibit NAADP-mediated Ca2+ release from lysosomes or endosomes via TPCs 64, failed to inhibit ZIKV infection in A549 cells (Fig. S5A). We also knocked down the expression of TPC1 or TPC2 in 4T1 cells (Fig. S5B), and found that knockdown of either TPC2 or TPC1 had little effect on JEV infection in 4T1 cells (Figs S5C and S5D). In addition, double knockdown of both TPC1 and TPC2 failed to inhibit JEV infection in 4T1 cells (Fig. S5E). These data indicate that berbamine does not target TPCs to inhibit avivirus infection.
In addition to TPCs, several other ion channels in endosomes and lysosomes have been reported to modulate endolysosomal tra cking 65, 66. Among them, TRPMLs in lysosomes and endosomes play critical roles in membrane tra cking, autophagy, and exocytosis 31, 32, 67. Therefore, we assessed whether berbamine might inhibit virus infection via modulating TRPMLs. As shown in Figure 7B, berbamine signi cantly decreased the TRPML-mediated Ca2+ release from lysosomes, which was triggered by ML-SA1, a selective and potent TRPMLs agonist 68. Consistently, knockdown of TRPML2, TRPML3, or both markedly inhibited JEV infection ( Fig. 7D and 7E), although TRPML1 knockdown failed to affect JEV infection (Fig. 7C). In addition, knockdown of both TRPML2 and TRPML3 decreased the levels of LDLR and ACE2 in cells (Fig. 7F). Interestingly, treatment of cells with ML-SA1 reduced the ability of berbamine to decrease the LDLR levels ( Fig. 7G and 7H). In summary, these data suggest that berbamine compromises the endolysosomal tra cking of viral receptors via inhibition of TRPMLs, and this leads to a decrease in the levels of viral receptors for (+)ss RNA viruses, thereby preventing these virus particles from entering the host cells.

Discussion
A number of studies have shown that virus infection changes the cytosolic Ca2+ homeostasis (or the resultant Ca2+ signaling) in the host cells, not only to facilitate the entry, replication, packaging, and release of the virus, but also to inhibit the cellular immune response against virus infection 22, 23, 24, 25, 26, 27, 28, 29, 30. It has also been shown that manipulation of the intracellular Ca2+ levels or Ca2+ signals can inhibit the virus infection 23, 24, 25, 28, 30. However, blocking the general extracellular Ca2+ in ux or the release of Ca2+ from the ER or mitochondria is detrimental to cells since the cytosolic Ca2+ levels or related signals are essential to almost all cellular activities. Thus, it is impractical to use general Ca2+ channel or signaling inhibitors to prevent or treat virus infection. For example, here we showed that BAPTA-AM could completely block infection of JEV and ZIKV ( Fig. 1A and 1B), but it is also cytotoxic. Here, we demonstrated that berbamine inhibited TRPML-mediated Ca2+ release from lysosomes and compromised the endolysosomal tra cking of membrane receptors for many (+)ss RNA viruses, thereby promoting the secretion of these trapped membrane receptors out of cells via EVs (Figs. 5-7). The decreased levels of viral receptors such as ACE2, DPP4 and LDLR, are likely responsible for the widespectrum anti-viral activity of berbamine. Notably, berbamine did not change the basal levels of cytosolic Ca2+ (Fig. 7A and 7B), and this might explain why it was only minimally cytotoxic (Fig. 4A-4C). Indeed, berbamine is actually an over-the-counter medicine that has been widely used to treat leukopenia in China for many years, indicating that it is safe to use in human. We also showed that berbamine protected mice from lethal challenges of JEV ( Fig. 4E and 4F). Thus, considering that the current COVID-19 pandemic has created a global health and economic crisis, berbamine (as a pan-anti-(+)ss RNA viral agent), might have the potential to be developed into an effective therapeutic agent for the prevention or treatment of COVID-19.
The TRPML family comprises three members: TRPML1, TRPML2 and TRPML3 69, 70, and they are permeable to variety of cations, including Ca2+, Na+, Zn2+ and Fe2+. It is known that loss-of-function mutations in TRPML1 lead to mucolipidosis type IV (ML4), a lysosomal storage disease 71, 72, 73. Among the three members, TRPML1 is ubiquitously expressed in all tissues, whereas TRPML2 and TRPML3 appear to have a more restricted pattern of expression, although they can be detected in many tissues and cell lines, including HeLa cells, A549 cells and HEK293 cells. TRPMLs are located in the membrane of early endosomes and recycling endosomes, and they are especially rich in late endosomes and lysosomes. The activation of TRPMLs via PI(3,5)P2 can trigger the release Ca2+ from endosomes and lysosomes, which participates in various endolysosomal tra cking events, including tra cking of endosomal vesicles, fusion events between late endosomes and lysosomes, and lysosome-mediated exocytosis 71, 72, 73. Here, we found that berbamine inhibited the ML-SA1-induced release of Ca2+ from lysosomes (Fig. 7B), which suggests that it is a TRPML inhibitor. Moreover, knockdown of TRPLM2 or TRPML3 markedly inhibited JEV infection (Figs 7D and 7E), and reduced the levels of LDLR and ACE2 in cells (Fig. 7F). In addition, ML-SA1 reversed the reduced level of LDLR expression in berbamine-treated cells ( Fig. 7G and 7H). In summary, our data suggest that berbamine inhibits TRPMLs and thus compromises the endolysomal tra cking of membrane receptors for viruses, including receptors for SarS-CoV-2 and JEV. The accumulation of endosomes leads to an increased level of secretion of these receptors via EVs and a concomitant decrease in the virus receptor levels at the plasma membrane. This decrease in the amount of virus receptors renders the cell resistant to virus infection. Therefore, TRPMLs might be potential therapeutic targets against (+)ss RNA viruses such as SARS-CoV-2.
LDLR was the rst member of the LDLR family to be identi ed; the family is now known to also contain VLDLR, ApoER2, LRP1, LRP2, and LRP6. These family members all share several structural domains, such as LDLR repeats (for ligand binding), an EGF-like domain and a transmembrane anchor motif. The LDLR family members mainly participate in lipoprotein tra cking to maintain cholesterol homeostasis 74. LDLR has also been shown to be one of receptors for the hepatitis C virus, rhinovirus and the vesicular stomatitis virus 39, 40, 41, 42, 43. Here we found that JEV infection triggered the internalization of LDLR, and internalized LDLR exhibited strong co-localization with the viral particles (Fig. 5E). In addition, LDLR knockdown abolished the JEV-induced expression of JEV-NS1 in the host cells (Figs 5D).
These data suggest that LDLR is a potential JEV receptor. However, future work is required to con rm any interaction between LDLR and the JEV-E glycoprotein, which is responsible for the virus binding to the host cells. It is also of interest to assess if other LDLR family members might be involved in JEV infection since they do share a similar ligand binding domain.
Immuno uorescence staining-Cells were xed with 4% paraformaldehyde (PFA) solution, blocked with PBS containing 5% normal donkey serum and 0.3% Triton™ X-100, and then incubated with primary antibody followed by the appropriate uorescent secondary antibody. To label the receptors on the plasma membrane, live cells were incubated with the primary antibody in PBS (+1% BSA) on ice for 90 min, followed by incubation with the uorescent secondary antibody on ice. Images were captured with a Carl Zeiss LSM 880 confocal microscope using a 63×oil objective lens. The primary antibodies used in these experiments are listed in Table S1.
In situ RNA hybridization-In situ RNA hybridization was performed with an RNAscope® Multiplex Fluorescent kit (Advanced Cell Diagnostics, 320851) by following the manufacturer's instructions. In brief, cells attached to coverslips were xed with 4% PFA, and then they were incubated with a speci c RNA probe targeting the JEV (ACD, 435551) or ZIKV (ACD, 463781) viral genome for 2 h at 40°C. Then, up to four signal ampli cation systems were used to detect the target RNA. After RNA hybridization, the cells were subjected to immuno uorescence staining, as described above.
Western blot analysis-The Bradford assay (Bio-RAD) was performed to measure the protein concentration of cell lysates. An equal amount of protein sample was loaded onto 8%-12% SDS-PAGE gels for electrophoresis. The proteins were then transferred to a PVDF membrane (Millipore), blocked with 5% non-fat milk, and blotted with primary and secondary antibodies. The primary antibodies used for immunoblotting are listed in Table S2.
Cytotoxicity assay-BHK-21, Huh7, A549, HeLa, or Vero cells, which had been plated in 96-well plates (Corning, 3603), were treated with different concentrations of berbamine or tetrandrine (Santa Cruz, sc-201492) for 24 h. The cells were then stained with propidium iodide (PI; Invitrogen, P3566) and Hoechst 33258 (Invitrogen, H3570), and images were acquired using a CellInsight CX7 High-Content Screening platform with a 10× objective lens. Quanti cation of the dead (PI positive) cells was performed with HCS Studio™ 3.0 (Thermo Fisher) and the half maximal inhibitory concentration (IC50) was calculated via Graphpad Prism 5.
The anti-virus activity of drugs-A549 cells were pre-treated with the concentrations of chemicals indicated in the Results, for 1 h, and infected with ~1 MOI of JEV or ZIKV. At 48 h post-infection, cells were xed with 4% PFA, stained with an anti-dsRNA antibody, and subjected to the high content screening platform for the acquisition of uorescence images. The percentage of infected cells was determined using HCS Studio™ 3.0., and the concentration for 50% of maximal effect (EC50) was calculated with Graphpad Prism 5. For EV71, 10 MOI of virus was used to infect RD cells and cells were xed at 10 h post-infection.
For anti-SARS-CoV-2 activity of berbamine, VeroE6 cells were pre-treated with berbamine at a titration of different concentrations (0-75 μM) for 6 hours. Then, the cells were washed with PBS and inoculated with SARS-CoV-2 at 0.01 MOI for 2 hours. At 2 hours post infection (hpi), the cells were washed with PBS and treated with berbamine at a titration of different concentrations (0-75 μM). At 24 hpi, 100 μL of viral supernatant was lysed and proceed to total RNA extraction using the QIAamp viral RNA mini kit (Qiagen, Hilden, Germany). The extracted RNA was then used to quantify the replication of SARS-CoV-2 using realtime quantitative RT-PCR (qRT-PCR) as described previously 75.
Puri cation of extracellular vesicles from the culture medium-A549 cells or HeLa cells were grown in 15cm dishes to ~80% con uency. The cells were then rinsed with PBS and incubated in EV-depleted complete medium containing DMSO or berbamine (25 μM) for 48 h. The supernatant was then collected and subjected to sequential centrifugation steps at different centrifugal forces (g) to remove the intact cells, dead cells or cell debris. After each centrifugation, the supernatant was transferred into a new 50 mL tube and the pellet was discarded. Finally, the supernatant was subjected to ultracentrifugation at 120,000 × g for 90 min, and the pellet (now containing the extracellular vesicles) was washed with PBS and subjected to another ultracentrifugation at 120,000 × g for 90 min. Finally, the exosome pellet collected and used for immunoblot analysis. Small interference RNA (siRNA)-Cells were transfected with siRNAs against respective genes (Table S3) using Lipofectamine 3000 according to the manufacturer's instructions. The knockdown e ciency was validated by immunoblot analysis or qRT-PCR.
Anti-JEV activity of berbamine in mice-The anti-JEV activity of berbamine was performed in BALB/c mice as described previously 76. Brie y, 3 to 4-week BALB/c mice were randomly divided into four groups (eight mice per group): an uninfected and PBS-treated group, an uninfected and berbamine-treated group, a JEV-infected and PBS-treated group, and a JEV-infected and berbamine-treated group. Mice were rst injected intraperitoneally with PBS or 15 mg/kg of body weight of berbamine, and 6 h later, mice were infected intraperitoneally with 10 7 TCID50 of JEV (SA14 virus strain). Thereafter, mice were treated with PBS or berbamine (15 mg/kg) twice per day for 14 days. The mice were monitored daily for morbidity and mortality. The mice that showed severe neurological signs of disease were euthanized. All animal studies were performed in B3 level laboratories by strictly following the safety and animal ethical guidelines of university and government.
Statistical analysis-Data are presented as mean ± S.E.M. Statistically signi cant differences were determined by the Student's t-test and P < 0.05 was considered to be statistically signi cant.        cells were transfected with siRNAs against TRPML2 or TRPML3, and were then infected with JEV for 24 h followed by immunoblot analysis against JEV-NS1. TRPML2 or TRPML3 knockdown e ciency was veri ed by qRT-PCR. (E) A549 cells were transfected with siRNAs against TRPML2 and TRPML3, and were then infected with ~50 MOI JEV for 90 min followed by in situ RNA hybridization to detect the RNA genome (red) of JEV. (F) A549 cells were transfected with siRNAs against TRPML2 and TRPML3, followed by immunoblot analysis against ACE2 and LDLR. (G, H) A549 cells in the presence or absence of ML-SA1 were treated with/without berbamine for 3 h, followed by LDLR immunoblot analysis (G). Alternatively, the live cells were immunolabeled with the anti-LDLR antibody, followed by FACS analysis to measure the cell surface LDLR levels (H).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementalinformation.pdf