key: cord-0937885-csyykpyx
authors: Yeung, Man Lung; Lee Teng, Jade Lee; Jia, Lilong; Zhang, Chaoyu; Huang, Chengxi; Cai, Jian-Piao; Zhou, Runhong; Chan, Kwok-Hung; Zhao, Hanjun; Zhu, Lin; Siu, Kam-Leung; Fung, Sin-Yee; Yung, Susan; Chan, Tak Mao; Kai-Wang To, Kelvin; Fuk-Woo Chan, Jasper; Cai, Zongwei; Pui Lau, Susanna Kar; Chen, Zhiwei; Jin, Dong-Yan; Yat Woo, Patrick Chiu; Yuen, Kwok-Yung
title: Soluble ACE2-mediated cell entry of SARS-CoV-2 via interaction with proteins related to the renin-angiotensin system
date: 2021-03-02
journal: Cell
DOI: 10.1016/j.cell.2021.02.053
sha: 17b4300606d7ebda1f416600e99bdab597c7479d
doc_id: 937885
cord_uid: csyykpyx

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can cause acute respiratory disease and multiorgan failure. Finding human host factors that are essential for SARS-CoV-2 infection could facilitate the formulation of treatment strategies. Using a human kidney cell line—HK-2—that is highly susceptible to SARS-CoV-2, we performed a genome-wide RNAi screen and identified virus dependency factors (VDFs), which play regulatory roles in biological pathways linked to clinical manifestations of SARS-CoV-2 infection. We found a role for a secretory form of SARS-CoV-2 receptor, soluble angiotensin converting enzyme 2 (sACE2), in SARS-CoV-2 infection. Further investigation revealed that SARS-CoV-2 exploits receptor-mediated endocytosis through interaction between its spike with sACE2 or sACE2-vasopressin via AT1 or AVPR1B, respectively. Our identification of VDFs and the regulatory effect of sACE2 on SARS-CoV-2 infection shed insight into pathogenesis and cell entry mechanism of SARS-CoV-2 as well as potential treatment strategies for COVID-19.

Coronavirus disease 2019 (COVID-19) caused by a novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has become a pandemic with >109 million confirmed cases and approximately 2.4 million fatalities (https://covid19.who.int/). SARS-CoV-2 can cause substantial pulmonary diseases. Besides, numerous extrapulmonary manifestations such as cardiac and renal complications have been reported to be associated with an increased risk of death in patients with COVID-19 (Gupta et al., 2020; Li et al., 2020) .

Host factors required for SARS-CoV-2 infection are largely unknown due to the lack of highly susceptible human cell lines for studying SARS-CoV-2 infection. As of now, a few studies have attempted to identify host factors required for SARS-CoV-2 infection either using non-human cells or modified human cells-that may not truly reflect authentic the infection process. These include studies using the CRISPR screening approach to identify host factors in SARS-CoV-2-infected Vero-E6 cells, which is of non-human origin (Wei et al., 2020) , or in modified human cell lines-A549 and Huh7.5.1-transduced with SARS-CoV-2 entry factors to become susceptible to SARS-CoV-2 infection (Daniloski et al., 2020; Heaton et al., 2020; Schneider et al., 2021; Wang et al., 2021) . In addition, another study reported the use of a focused CRISPR screening to target a small number of host factors, which were recently identified to be involved in SARS-CoV-2 infection (Hoffmann et al., cell surface, where it is cleaved by host proteases such as disintegrin and metalloproteinase 17 (ADAM17), to release an enzymatically active soluble form of ACE2 (sACE2) into the plasma (Lambert et al., 2005) . The sACE2 retains an intact SARS-CoV-2 interaction site, suggesting its ability to bind to SARS-CoV-2. Interestingly, a recent study has observed that ACE2 shedding may be induced by regulatory pathways influenced by and that the concentrations of sACE2 may correlate with the level of systemic inflammation that occurs (Kornilov et al., 2020) . However, most of the current research focus on studying the function of cACE2 in SARS-CoV-2 pathogenesis; the impact of circulating sACE2 on viral entry is largely unknown.

Tissue tropism of SARS-CoV-2 cannot be fully explained only by the expression pattern of cACE2. While a majority of cell susceptibility studies are based solely on the mRNA level of cACE2 (Sungnak et al., 2020) , studies on its protein expression pattern is limited (Bertram et al., 2012; Hamming et al., 2004) . Recently, Wang et al. comprehensively investigated the tropism of SARS-CoV-2 in human tissues and discovered discordance between mRNA and protein expression levels of ACE2 in many tissues . Indeed, studies of cACE2 mRNA and protein expression levels showed that only a small population of the lung cells has a detectable expression level of cACE2-that is in contrast to the common knowledge that the lungs are the primary site of infection Zou et al., 2020) . Potentially, circulating sACE2, which retains the interaction site for binding to SARS-CoV-2, could interact with SARS-CoV-2 in the extracellular compartment forming complexes, which may affect the infectivity.

In this study, we report the identification of a human renal cell line-HK-2-which is highly susceptible to SARS-CoV-2. Using HK-2 cells, we performed a genome-wide RNAi screening and successfully identified virus dependency factors (VDFs), the expression of which is required for SARS-CoV-2 infection. Detailed analysis of these VDFs J o u r n a l P r e -p r o o f 6 suggested that perturbation of their functions could be linked to clinical symptoms and complications of COVID-19. Finally, we elucidated the important but overlooked role of sACE2 in SARS-CoV-2 infection using multiple methods, thus advancing our understanding of pathogenesis and treatment strategies for COVID-19.

To find an optimal human cell line that is highly susceptible to SARS-CoV-2 infection, we examined eleven human cell lines of different organs, including HK-2, Caco-2, A549, Calu3, Huh7, HepG2, PLC/PRF/5, RD, HeLa, NT2, and 293T, inoculating the virus at a multiplicity of infection (MOI) of 1. At 72 hours post-infection, median tissue culture infective dose (TCID 50 ) results showed that Vero-E6 cells strongly supported SARS-CoV-2 replication with high viral titer of 5.7 × 10 8 TCID 50 /mL ( Figure 1A) .

Much lower viral titers, ranging from 4.2 × 10 3 to 7.9 × 10 4 TCID 50 /mL, were detected in most human cell lines ( Figure 1A) . Intriguingly, SARS-CoV-2 can replicate efficiently in human renal proximal tubule cells (HK-2) with high viral titers of up to 3.8 × 10 8 TCID 50 /mL ( Figure 1A ). Among these cell lines, Vero-E6 and Caco-2, representing unmodified cell line of non-human and human origins respectively, which are known to be susceptible to SARS-CoV-2 (Bojkova et al., 2020) . Therefore, effective replication of SARS-CoV-2 in HK-2 and Vero-E6 cells was further confirmed by western blotting (WB) and immunofluorescence assay (IFA), which detected strong viral protein expression in the infected cells ( Figure 1B and C). In comparison, WB analysis was unable to detect NP expression in Caco-2 cells ( Figure 1B) , which was consistent with the IFA results showing a few positive-stained cells ( Figure 1C) . Consistent with viral load and the protein expression data (Figure 1A -C) , strong and moderate CPEs were observed in HK-2 and Vero-E6 cells, respectively, but not in Caco-2 cells upon SARS-CoV-2 infection ( Figure 1D ). SARS-CoV-2 exhibited similar replication kinetics in HK-2 and Vero-E6, whereas in Caco-2 cells, it replicated less efficiently ( Figure 1E ). Taken together, HK-2 strongly supports the lytic infection of SARS-CoV-2, representing a suitable cell line model of human origin for studying the molecular interactions between SARS-CoV-2 and human host.

J o u r n a l P r e -p r o o f

To identify the VDFs essential for SARS-CoV-2 replication, we used a puromycinmarked feline immunodeficiency virus (FIV)-based lenti-vector-shRNA library to target human transcripts in HK-2 cells followed by the challenge of SARS-CoV-2 (Figure 2A and S1). We anticipated that HK-2 cells that were knocked down for mRNA-which were dispensable for SARS-CoV-2 replication-would succumb rapidly upon infection. In contrast, a shRNA clone silenced an mRNA essential for lytic SARS-CoV-2 replication would survive. The identity of shRNA in the survival cells was determined by highthroughput sequencing (Figure 2A ). Principal Component Analysis (PCA) plot and heatmap based on the abundance of all identified shRNA sequences mapped against the human reference genome revealed good spatial-temporal separation between SARS-CoV-2-infected and mock-infected samples (Figure 2B -C) . Pathway analysis of VDFs with ≥3-fold enrichment in SARS-CoV-2-infected HK-2 cells revealed 16 significantly affected biological pathways (p <0.05) related to cardiac (pink), pulmonary (blue), liver (green), and renal diseases (brown) ( Figure 2D) . Notably, 11 of these 16 biological pathways were related to cardiac diseases (pink; Figure 2D ). In addition, PANTHER pathway analysis of the identified VDFs was performed using a cut-off p-value of 0.001 and 8 major affected molecular pathways were discovered ( Figure 2E ). Among these, the top-affected molecular pathways "Vasopressin-like receptors (R-HSA-388479)", and "Defective AVP does not bind AVPR1A,B and causes neurohypophyseal diabetes insipidus (NDI) (R-HSA-5619099)" were also involved in the regulation of the cardiovascular system (pink; Figure 2E ). Knockdown of individual VDFs involved in the vasopressin-related pathway using corresponding siRNAs resulted in significant inhibition of SARS-CoV-2 infection (p <0.01; 45.6% -90.7% of J o u r n a l P r e -p r o o f inhibition; Figure 2F ). The strong inhibitory effects warranted further study of the VDFs that are functionally related to the vasopressin pathway on SARS-CoV-2 infection.

Vasopressin can enter cells via receptor-mediated endocytosis through membrane remodeling and vesicle trafficking (Innamorati et al., 2001) . Indeed, our Gene Ontology (GO) enrichment analysis revealed a close relationship of the VDFs to plasma membrane (79%), membrane raft (12%), as well as intracellular trafficking and secretion, and vesicle (9%), which are known to be important in regulating cytokinesis and vesicle trafficking ( Figure   3A ; left). Similarly, we also performed GO enrichment analysis using gene candidates that were previously reported to be important for the replication of other coronaviruses Bairoch et al., 2005; McKusick, 2007; Oughtred et al., 2019; Pruitt et al., 2005) . Consistently, gene candidates related to plasma membrane (20%) as well as intracellular trafficking and secretion, and vesicle (18%) also contributed substantially to the biogenesis of coronaviruses (Figure 3A ; right).

Next, we performed a volcano plot analysis to identify VDFs that could be involved in SARS-CoV-2 replication ( Figure 3B ). Manual gene annotation revealed that certain enriched VDFs have functional roles related to the regulation of the cardiovascular system, the cytokinesis and vesicle trafficking pathway, and the renal-related disease ( Figure   3B ), which was consistent with the results generated from our pathway and GO enrichment analyses ( Figure 2D -E and 3A) . The differential gene enrichment of the VDFs with a false positive rate (FDR) of <5% is displayed in the heatmap ( Figure 3C ). Of these, twenty-two VDFs were selected according to their biological roles for further validation. Consistent with our RNAi screening results, knockdown of individual VDFs using corresponding siRNAs J o u r n a l P r e -p r o o f resulted in significant inhibition of SARS-CoV-2 infection (p <0.01; 46.5% -95.4% of inhibition; Figure 3D ). Particularly, potent inhibition (>80%) was observed in some VDFs with biological functions related to the control of cytokinesis and vesicle trafficking pathway (GPR176, CAPNS1, IL18RAP, ARL4D, and CXCR1; red boxed; Figure 3D ) and the endosomal/lysosomal system (ANXA8 and ANXA1; green boxed; Figure 3D ). Together, our results evidenced that cytokinesis, vesicle trafficking, and the endosomal/lysosomal system are important biological pathways for the replication of SARS-CoV-2.

In view of the importance of vasopressin-related VDFs in SARS-CoV-2 infection ( Figure 2F ), we further tested if SARS-CoV-2 infection depends on vasopressin. Compared to the mock-treated control, HK-2 cells pre-treated with increasing doses of vasopressin showed a significant increase in the SARS-CoV-2 infectivity as determined by the TCID 50 assays ( Figure 4A ). It is known that circulating sACE2 is an important regulator of the RAS, while vasopressin production could also be regulated by the RAS (Reid et al., 1983) . Given that sACE2 retains an intact interaction site for binding to SARS-CoV-2, we hypothesized that there may be interaction between vasopressin and sACE2 which could modulate the SARS-CoV-2 infectivity. To examine their interactions, we spiked recombinant S of SARS-CoV-2 into culture medium of 293T cells, which were doubly transfected with FLAG-tagged vasopressin and V5-tagged ACE2. Secretions of sACE2 and vasopressin were confirmed by WB analyses of the conditioned medium of transfected cells ( Figure 4B ; lane 1). As expected, pulldown of sACE2 could effectively co-immunoprecipitate the S ( Figure 4B ; lane 2). Intriguingly, vasopressin was detected in the same co-immunoprecipitation suggesting that sACE2 could form a complex with S and vasopressin ( Figure 4B ; lane 2). To further examine if there was any interaction between S and vasopressin, we removed vasopressin J o u r n a l P r e -p r o o f from the co-immunoprecipitation experiment. Pulldown of sACE2 was able to coimmunoprecipitate the S ( Figure 4B; lane 3) , but the amount was lower compared to the reaction containing vasopressin ( Figure 4B ; lane 1). Remarkably, when S was removed from the co-immunoprecipitation reaction, vasopressin was no longer able to be coimmunoprecipitated by sACE2, suggesting that the S was required for the formation of sACE2-S-vasopressin complex ( Figure 4B; lane 4) . The interaction between vasopressin and S was supported by successful co-immunoprecipitation of S by pulling down of vasopressin in the absence of ACE2 ( Figure 4B ; lane 5). The interactions between vasopressin, sACE2, and S of SARS-CoV-2 were further confirmed by reciprocal co-immunoprecipitation using HK-2 cells. It is expected that the HK-2 cells, which is highly susceptible to SARS-CoV-2, Vasopressin functions by binding to its membrane receptors, which could trigger a process called receptor-mediated endocytosis. Consistent with our interaction assay results, which showed that vasopressin could form a complex with sACE2 and S ( Figure 4B ), knockdown of a vasopressin receptor-AVPR1B-potently inhibited SARS-CoV-2 infection (89.66% ± 0.14 inhibition; Figure 2F and S2). Therefore, we hypothesized that SARS-CoV-2 may enter cells via the receptor-mediated endocytosis through AVPR1B. To test this hypothesis, non-permissive 293T cells were first transfected with a plasmid overexpressing J o u r n a l P r e -p r o o f AVPR1B, followed by inoculating the conditioned supernatants, which contained the sACE2-S-vasopressin complex prepared in Figure 4B . IFA showed that the AVPR1B was primarily localized on the plasma membrane of the transfected 293T cells (Figure 4C ). At 6 hours post-inoculation, the AVPR1B-transfected cells showed decreased immunofluorescent intensity on the plasma membrane but increased puncta-staining pattern in the cytosol ( Figure 4C ). IFA revealed a strong co-localization signal between S and AVPR1B at the puncta ( Figure 4C ), suggesting internalization of sACE2-S-vasopressin complex via AVPR1B.

Interaction between ACE2 and cellular receptor AT1 has been reported (Deshotels et al., 2014) . To test if the sACE2-S-vasopressin complex-like AVPR1B-could also be internalized into cells via AT1, we performed overexpression experiments using the same experimental conditions as described above. Over the infection time course, IFA results detecting the YFP-tagged AT1 and recombinant S showed that 293T cells transfected with the AT1 also displayed increasing puncta staining pattern in the cytosol after inoculation with conditioned supernatant containing sACE2-S-vasopressin complex ( Figure 4D ). As our interaction assay showed that the sACE2 could interact with S in the absence of vasopressin ( Figure 4B ), we tested if the sACE2-S complex could also enter cells by utilizing the same cell entry mechanism. Under the same experimental conditions, similar puncta staining pattern could be observed in the YFP-tagged AT1-transfected 293T cells following inoculation of the sACE2-S complex ( Figure 4E ). Together, our results suggested that the sACE2-S complex and sACE2-S-vasopressin complex could enter cells via receptormediated endocytosis through AVPR1B and/or AT1 receptors.

Receptor-mediated endocytosis is known to be regulated by cytokinesis, vesicle trafficking, and the endosomal/lysosomal system (Simonetti and Cullen, 2019) . To further examine the role of cytokinesis and vesicle trafficking in SARS-CoV-2 infection, we treated the HK-2 cells with two cytoskeletal drugs-Jasplakinolide and Cytochalasin D (CytD)prior to SARS-CoV-2 infection. Compared to the mock-treated control cells, HK-2 cells pretreated with Jasplakinolide and CytD showed 2.15-log (p <0.001) and 1.33-log (p <0.001) reductions in viral load, respectively ( Figure 4F ). WB and IFA results confirmed undetectable and minimal SARS-CoV-2 NP expression in pretreated cells (Figure 4G and H). Besides a substantial reduction in the number of positively stained cells in the CytDtreated HK-2 cells, we observed an altered staining pattern with the viral antigen expression primarily restricted to the plasma membrane forming a "ring-like" structure ( Figure 4H ).

These results suggested that perturbation of the cytoskeletal organization may hamper the entry of SARS-CoV-2 into the cytosol.

Besides, CytD is known to be able to modulate the function of the endosomallysosomal system. Indeed, our RNAi screening also identified several VDFs that are related to this system, of which four (ANXA8, ANXA1, ANXA8L1, and ALS2) have been validated ( Figure 3D ). Given that endosomal acidification plays important roles in receptor-mediated cell entry in some viruses (Glebov, 2020) , we examined if the change of endosomal pH would affect SARS-CoV-2 replication by impairing the function of the proton pumpvacuolar H + -ATPase (V-ATPase)-which is responsible for controlling the endosomal pH via a specific inhibitor, Bafilomycin A1 (BafA1) (Cotter et al., 2015) . Pretreatment of HK-2 cells using the BafA1 at a concentration of 25 ηM showed a significant inhibition (>4-log) of SARS-CoV-2 infection when compared with the mock-treated control cells (p <0.001; Figure 4F ). Pretreatment with a 10-fold lower dose (2.5 ηM) of BafA1 also achieved a similar level of inhibition (>4-log) (p <0.001; Figure 4F ). Consistently, viral protein J o u r n a l P r e -p r o o f expression was not detected by WB and IFA in the BafA1-treated cells ( Figure 4G and H).

Thus far, our results supported a potential cell entry mechanism by which SARS-CoV-2 exploits the endocytosis pathway ( Figure 4C -H 

Dyn2-dependent endocytosis involves a vesicular transport event to facilitate the internalization and recycling of receptors (Cullen and Steinberg, 2018) . In addition to being a receptor for SARS-CoV-2, ACE2 controls blood pressure homeostasis, which requires complex regulatory events involving protein trafficking as well as post-translational modifications. First, surface trafficking and sorting are required for plasma membrane expression of the cACE2. On the surface, cACE2 requires proteolytic cleavage by proteases such as disintegrin and ADAM17, to release circulating sACE2 to the extracellular space. It is conceivable that the shedding of cACE2 could modulate SARS-CoV-2 infectivity. To test this possibility, prior to SARS-CoV-2 infection, we pretreated HK-2 cells with GW280264X, which inhibits the enzymatic activity of ADAM17. Following the treatment, successful inhibition of sACE2 but not cACE2 expression was confirmed by WB ( Figure 5A) . Notably, pretreatment of GW280264X potently inhibited SARS-CoV-2 infection in a dose-dependent manner as shown by WB, TCID 50 , and IFA ( Figure 5A -C).

In addition, we independently performed knockdown experiments to confirm the regulatory effect of ADAM17 on SARS-CoV-2 infectivity. Consistent with the results of the GW280264X treatment experiment, a reduction of sACE2 was also observed in the siADAM17-transfected cells ( Figure 5D ). Dose-dependent reductions in viral loads (p <0.001) and NP expression were observed in siADAM17-transfected cells (Figure 5D -F).

Overall, our results suggested that targeting ACE2 shedding could modulate SARS-CoV-2 infectivity.

Our results showed that sACE2 could facilitate virus cell entry through receptormediated endocytosis, suggesting its potentially important role on SARS-CoV-2 infection. To further substantiate its role, we tested the effect of sACE2 on SARS-CoV-2 infectivity in HK-2 cells at a low MOI of 0.01. Under this infection condition, ~10% of cells could be infected ( Figure 6A ). With increasing doses of recombinant ACE2 (rACE2) on HK-2 cells, SARS-CoV-2 infectivity increased to 30% and 50%, respectively, as evidenced by stronger CPE and higher expression of SARS-CoV-2 NP observed in the infected cells ( Figure 6A ).

The results suggested that the addition of rACE2, mimicking the sACE2, could enhance SARS-CoV-2 infectivity. This effect, however, was not observed when performing the experiment using MERS-CoV under the same conditions ( Figure S3 ). To further confirm the role of sACE2 in SARS-CoV-2 infection, we repeated the same experiment using two human lung cell lines-Calu3 and A549. Consistently, WB and TCID 50 results showed a dosedependent augmentation of SARS-CoV-2 infectivity in both cell lines treated with an increasing dose of rACE2 ( Figure 6B ). Intriguingly, cell lines-including RD, NT2, HepG2, J o u r n a l P r e -p r o o f and Huh7-derived from organs other than the lung also showed increase in SARS-CoV-2 susceptibility when treated with an increasing dose of rACE2 ( Figure 6B ).

To investigate the role of sACE2 under physiologically relevant conditions, we selected a panel of human cell lines from different organs and examined their relationship between the susceptibility to SARS-CoV-2 and sACE2 and cACE2 expressions. Consistent with previous studies on the rare expression of ACE2 in many cell types (Lukassen et al., 2020; Wang et al., 2020) , WB results demonstrated that the protein expression of ACE2 could only be detected in a few cell lines ( Figure 6C) Figure S4A, lane 2) , suggesting that most ACE2 (i.e., GFP-ACE2∆TM) was exported out into the extracellular space and not retained inside the cells. These results were consistent with the IFA results where the GFP signal could be detected in cells transfected with the wild-type ACE2 but not the GFP-ACE2∆TM ( Figure S4B) . We then performed confocal microscopy analysis of the GFP-ACE2∆TM-transfected cells upon SARS-CoV-2 infection. Remarkably, orthogonal projections of confocal sections revealed strong co-localization of SARS-CoV-2 NP and GFP-ACE2∆TM ( Figure 6D ). Three-dimension (3D) reconstruction of the confocal images is available in Video S1. Profiles of fluorescent intensity measured at different focal planes also revealed a high degree of overlapping signals between the SARS-CoV-2 NP and GFP-ACE2∆TM ( Figure 6E and Video S1). Further optical sectioning of the infected cells revealed that similar to the formation of vesicles during the endocytosis, a punctate pattern of viral antigen predominantly localized on the cell surface ( Figure 6F) . Together, these results support that sACE2 could interact with SARS-CoV-2 during the infection process, and that sACE2 has a potentially determining role in SARS-CoV-2 infection. Increasing evidence support the association between COVID-19 and cardiovascular and renal-related diseases (Camm and Camm, 2020; Yang et al., 2020) . These observations were also in line with the RNAi screening results, in which the functional annotation of some enriched VDFs and their biological pathways were also linked to the regulation of cardiac and renal functions. In fact, pathway analysis of VDFs showed ~70% of significantly affected biological pathways are related to cardiac diseases (pink; Figure 2D ). The most affected biological pathway is related to cardiac arrhythmias, which is often observed in SARS-CoV-2-infected patients, especially in critically ill cases ( Figure 2D ) (Babapoor-Farrokhran et al., 2020; Karamchandani et al., 2020; Rav-Acha et al., 2020) . Besides, we also identified three significantly affected renal disease-related pathways (brown; Figure 2D ), which is in line with the high incidence of acute kidney injury reported in SARS-CoV-2-infected cases . A previous study also reported the presence of coronavirus-like particles in >75% kidney biopsies of deceased SARS-CoV-2 patients (Su et al., 2020) . Furthermore, PANTHER pathway analysis of the enriched VDFs identified vasopressin-related pathways as the top-affected molecular pathways ( Figure 2E) . The important roles of these vasopressin-related VDFs in SARS-CoV-2 infection were further validated by performing knockdown experiment ( Figure 2F) . Previous studies showed that vasopressin can induce differentiation of stem cells into cardiomyocytes and promote heart muscle homeostasis (Yasin et al., 1994) . It can also control tonicity of body fluids by converting to arginine vasopressin (AVP) that can further regulate the arterial blood pressure (Demiselle et al., 2020) . Consistently, due to their regulatory roles in maintaining homeostasis of kidney and J o u r n a l P r e -p r o o f heart, dysregulated expressions of vasopressin and AVP were also directly linked to the development of renal failure and cardiovascular system diseases (Czarzasta et al., 2018) .

Therefore, further investigation of the identified VDFs involved in cardiac and renal disease development may provide mechanistic insights into SARS-CoV-2 induced extrapulmonary diseases.

sACE2 and vasopressin play important roles in SARS-CoV-2 infection. ACE2, being a receptor for SARS-CoV-2 , is also known to serve as one of the key regulators controlling the release of vasopressin into the plasma for the maintenance of blood pressure homeostasis via the RAS (Matsukawa and Miyamoto, 2011; Reid et al., 1983) . To serve its regulatory role in the RAS, the tissue-bound form of ACE2 (i.e., cACE2) is shed by proteases to produce the sACE2, which can then enter the circulatory system. As sACE2 preserves the binding site for SARS-CoV-2, it is possible that sequestration of SARS-CoV-2 by sACE2 may enable cell entry of tissues where cACE2 is poorly expressed. Indeed, our in vitro data showed that endogenous sACE2 could interact with the S of SARS-CoV-2 in the extracellular compartment ( Figure 4B) . The resulting sACE2-S complex could then enter cells through receptor-mediated endocytosis via the AT1 surface receptor (Figure 4D and E) .

Additionally, we found that the S of SARS-CoV-2 could interact with vasopressin forming an sACE2-S-vasopressin complex, which facilitated cell entry via another vasopressin receptor, AVPR1B (Figure 4B and C) . This new cell entry mechanism may explain our data showing that cells from various organs could be sensitized to SARS-CoV-2 upon administration of rACE2 (Figure 6A and B) .

sACE2 expression contributes to the cell line susceptibility to SARS-CoV-2. Little or low infectivity of SARS-CoV-2 was detected in all tested human cell lines, except for the HK-2 cells ( Figure 6C) . In contrast to HK-2 cells, SARS-CoV-2 is unable to replicate efficiently in 293T, although both cell lines were derived from human kidney. We speculate J o u r n a l P r e -p r o o f that the differential susceptibility may be linked to their differences in sACE2 level. We also noted that while highly susceptible HK-2 cells exhibited very strong expressions of both cACE2 and sACE2 ( Figure 6C; lane 11) , expression of cACE2 alone does not render the cells susceptible to SARS-CoV-2 as exemplified in HepG2 and 293T cells where cACE2, but not sACE2, was detected ( Figure 6C; lane 6 and 9) . In contrast, expression of sACE2 alone Figure 6A and B) .

Together, our in vitro infection data using human cell lines that originated from different organs support the important role of sACE2 in SARS-CoV-2 infection.

We discovered the dual role of sACE2 in SARS-CoV-2 infection. Modulating the SARS-CoV-2 infectivity using recombinant sACE2 has been previously suggested as a treatment strategy for COVID-19. Attempts have been made to utilize recombinant soluble human ACE2 to inhibit SARS-CoV-2 infection using in vitro model (Cocozza et al., 2020; Monteil et al., 2020) . In these studies, very high concentrations of rACE2 [~10 -200 µg/mL of ACE2, concentrations are much higher than its physiological range in plasma i.e., ηg/mL (Ridwan et al., 2019; Sama et al., 2020) ] were required to achieve inhibitory effects. Indeed, our results were also in line with their findings, where 25 and 100 µg/mL of rACE2 could inhibit SARS-CoV-2 infection ( Figure S3 ). We speculate that the addition of excessive amounts (i.e., µg/mL level) of recombinant ACE2 may saturate endocytic recycling of the ACE2 receptor, competing with the SARS-CoV-2-ACE2 complex for cell entry, therefore J o u r n a l P r e -p r o o f resulting in the reduction of SARS-CoV-2 infectivity. In contrast, rACE2 concentrations close to physiological range (i.e., ηg/mL level) could enhance SARS-CoV-2 infection ( Figure 6A , B, and Figure S3 ). Interestingly, a similar phenomenon was recently reported in a study which provided an important clue on the in vivo effect of the rACE2 in SARS-CoV-2 infected patient (Zoufaly et al., 2020) . The study showed that substantial and steady increases in viral loads were detected in tracheal aspirates (from ~10 3 copies/mL at day 0 to ~10 5 copies/mL at day 2) and nasopharyngeal swabs (from ~10 4 copies/mL at day 0 to ~10 5 copies/mL at day 5), respectively, after the administration of rACE2. Although the patient eventually recovered after the appearance of neutralizing antibodies, this in vivo data and our findings suggested that treatment that may alter sACE2 level in patients with COVID-19

should be carefully considered.

The discovery of a highly permissive human renal cell line-HK-2-to SARS-CoV-2 is useful for studying the biogenesis of SARS-CoV-2 and the possible therapeutic options for COVID-19. The physiological and pathological relevance of using this HK-2 cell line in studying SARS-CoV-2 biogenesis is supported by recent ultrastructural studies which detected SARS-CoV-2 in renal autopsy of patients with COVID-19 (Farkash et al., 2020) . The versatility of HK-2 is further supported by our RNAi screening results, in which many successfully identified VDFs were not only linked to renal-related diseases but also other clinical manifestations and complications of COVID-19 (Figure 2D and E) . Furthermore, most current studies on SARS-CoV-2 infection were based on the use of pseudovirus-expressing viral surface spike protein or non-human cell lines, which may not truly reflect an authentic infection process (Heaton et al., 2020; Shang et al., 2020) . Our findings using authentic virus to infect a physiologically relevant HK-2 cells revealed that SARS-CoV-2 entered the cells mainly through Dyn2-dependent endocytosis (Figure 4F -I) . The involvement of cytokinesis, vesicle trafficking, and the J o u r n a l P r e -p r o o f endosomal/lysosomal system in SARS-CoV-2 cell entry was further confirmed through the use of inhibitors (Jasplakinolide, CytD and BafA1) and a dominant-negative Dyn2 construct ( Figure 4F -I) . Of note, our results showed that a potent inhibition (>4-log) of SARS-CoV-2 replication could be achieved even when cells were pretreated with a low dose of BafA1 (i.e., 2.5 ηM) (Figure 4F -H) , which could inhibit receptor-mediated endocytosis (Harada et al., 1996) . Further investigation of the BafA1 and its analogues as potential therapeutic agents for COVID-19 is warranted.

Modulating the shedding of sACE2 may provide insights into treatment strategies for COVID-19. A previous study suggested that the ADAM17 activity could be correlated with infectivity of SARS-CoV, which also utilized ACE2 as a cellular receptor (Haga et al., 2008) .

The authors reported that the S of SARS-CoV induced the ADAM17 activity causing an enhanced ACE2 shedding which was positively correlated with SARS-CoV infectivity. It is possible that cells could be sensitized to both SARS-CoV-2 and SARS-CoV by enhancing the production of sACE2 through the induction of ADAM17 activity. On the flip side, controlling the activity/expression of ADAM17 could be a potential treatment strategy for COVID-19. Indeed, we demonstrated that inhibition of ADAM17 by a sheddase inhibitor, GW280264X, or by a specific siRNA potently suppressed SARS-CoV-2 infection ( Figure   5A -F). The therapeutic potential of targeting the ADAM17 has also been widely investigated in cancer research with minimal side effects (Bandsma et al., 2015; Blaydon et al., 2011) , suggesting that it may represent a safe target in controlling SARS-CoV-2 infection.

Overall, our identification of inhibitors, including the sheddase inhibitor, the cytoskeletal drugs and/or siRNAs/drugs targeting the VDFs, could provide insights into future development of drug combination therapy, which can be considered as an effective strategy in preventing the generation of drug-resistant mutants, particularly in the case of RNA viruses (Presloid and Novella, 2015) , for the treatment of COVID-19.

This study has several limitations as the findings are based on HK-2 cells which are of renal origin. As the primary infection site of SARS-CoV-2 is lung, further examinations will be needed to study the involvement of these factors during pulmonary infection. Shek Chi Wai Foundation. We also thank the support from the Centre for PanorOmic Sciences, Genomics and Bioinformatics Core, at the University of Hong Kong. The funding sources had no role in the study design, data collection, analysis, interpretation or writing of the paper. 

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Man Lung Yeung (pmlyeung@hku.hk).

All requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact author. Materials will be made publicly available either through publicly available repositories or via the authors upon execution of a Material Transfer Agreement. Table S2 (see also Figure S5 ). Viruses SARS-CoV-2 (HKU-001a) was isolated from a nasopharyngeal aspirate of a laboratoryconfirmed COVID-19 patient in Hong Kong . MERS-CoV (EMC/2012 strain) and was a gift from Ron Fouchier (Erasmus Medical Center) (Zaki et al., 2012) . The isolates were propagated through Vero-E6 cells (ATCC) in DMEM (Gibco) supplemented with 10% FCS (Gibco) and 100 units/mL penicillin plus 100 µg/mL streptomycin (1% PS). All experiments entailing live SARS-CoV-2 followed the approved standard operating procedures of our Biosafety Level 3 (BSL-3) facility (Yeung et al., 2016) .

FIV-based shRNA library was produced as described previously (Yeung et al., 2009; Yeung et al., 2018) . Briefly, 2 µg of the shRNA library in lentiviral constructs (SBI) were cotransfected with 10 µg of the pPACK packaging plasmid mix into HEK293T cells using

Lipofectamine and Plus reagents according to the manufacturer's protocol (ThermoFisher).

At 48 hours after transfection, conditioned culture media were collected to harvest the Step 2: 94°C for 30 seconds and then 68°C for 1 minute;

Step 3: Repeat

Step 2 for 20 cycles;

Step 4: 68°C for 3 minutes. Nested PCR reactions using 1 µL of the 1 st round PCR products were performed using 100 µM of H1 Forward primer (5′-GTTCTGTATGAGACCACTTGGATCC-3′), 100 µM of RevPrimer (5′-AAAGAATGCTTATGGACGCTAGAA-3′), and iProof (Bio-Rad). PCR amplification of the shRNA target region was performed under the following conditions:

Step 1:94°C for 2 minutes, 50°C for 2 minutes and then 68°C for 1 minutes;

Step 2: 94°C for 30 seconds and then 68°C for 30 seconds;

Step 3: Repeat

Step 2 for 18 cycles;

Step 4: 68°C for 3 minutes.

The PCR products were then gel purified using QIAquick PCR purification kit (QIAGEN) as J o u r n a l P r e -p r o o f described in the manufacturer's protocol. The purified PCR products were then submitted to Centre for PanorOmic Sciences at HKU for high-throughput sequencing.

For independent validation, siRNAs (Human ON-TARGETplus) were purchased from ThermoFisher. These siRNAs were transfected into the HK-2 cells using Lipofectamine 2000 (ThermoFisher) according to the manufacturer's protocol. Forty-eight hours after transfection, the cells were challenged with SARS-CoV-2 (MOI = 1). At day 3 post-infection, total RNA of the shRNA-transduced HK-2 cells were harvested as described above.

All samples for high-throughput sequencing were processed by the Centre for PanorOmic Table S2 (see also Figure S5 ).

J o u r n a l P r e -p r o o f

The median tissue culture infective dose (TCID 50 ) per mL was determined for SARS-CoV-2

in Vero-E6 cells as previously described (Yeung et al., 2018) . Briefly, cells were seeded in 96-well plates at a density of 5 ×10 4 cells/well in 150 µL of DMEM. The virus was serially diluted by half-log from 10 3 to 10 14 in DMEM. One hundred microliters of each dilution were added per well; and the plates were observed daily for CPE for five consecutive days.

Vero-E6, HK-2, 293T, and Caco-2 cells were fixed in 4% paraformaldehyde solution in 1×PBS containing 0.1% triton X-100. After 1 hour blocking with 3% bovine serum albumin (BSA) at room temperature, the cells were stained with anti-SARS-CoV-2 antibodies (developed by our group), anti-FLAG (Sigma) and/or with anti-ACE2 antibodies (R&D systems) for 1 hour at room temperature as we previously described (Yeung et al., 2008) .

Unbound antibodies were washed away six times with 1×PBS. Positively stained cells were detected by secondary IgG (H+L) antibodies conjugated either with Alexa Fluor 488 or Alexa

Fluor 594 (Life Technology) for 30 minutes at room temperature. Following six 1×PBS

washes, the stained cells were mounted onto glass slides with VECTASHIELD mounting medium with 4′,6′-diamidino-2-phenylindole (DAPI) (Vector Lab) and examined with a Leica TCS-NT microscope (Leica Microsystem) or a LSM700 confocal microscope (Zeiss).

Total RNAs and viral RNAs were isolated using TRIzol (ThermoFisher) and Viral RNA Mini kit (QIAGEN), respectively, as described in our previous literature (Yeung et al., 2016) .

Following RNA quantification, one microgram of RNA was reverse transcribed into cDNA using random hexamers. Detection of SARS-CoV-2 and MERS-CoV was performed using primers 5′-ACAGGTACGTTAATAGTTAATAGCGT-3′ and 5′- 

Cell lysates or tissue extracts were resolved by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE). Proteins were transferred to polyvinylidene difluoride (PVDF) membrane by electroblotting (Hoefer) at constant current of 150mA overnight. Detection of antigens was performed using anti-SARS-CoV-2 antibodies (developed by our group) or anti-hACE2 antibodies (R&D systems) as previously described (Yeung et al., 2016) . As loading controls, the membranes were stripped with Restore western blot stripping buffer (Pierce) before reprobing with anti-γ-tubulin (Sigma).

Recombinant SARS-CoV-2 S proteins (Sino Biological) were incubated with protein lysates expressing V5-tagged or untagged ACE2 and FLAG-tagged vasopressin under previously described conditions (Yeung et al., 2018) with the following modifications: both recombinant S and the protein lysates were mixed in the presence of 0.1% Tween 20 at 4°C overnight with

shaking. The next day, the anti-FLAG ® M2 affinity agarose gel (Sigma) or anti-V5 agarose affinity gel (Sigma) was added into the mixture and allowed to incubate for an additional 2.5 J o u r n a l P r e -p r o o f hours. Following six washes with a buffer containing 50 mM of NaH 2 PO 4 , 300 mM of NaCl, and 20 mM of imidazole pH 8.0, agarose-bound proteins were fractionated by SDS-PAGE and detected by WB analyses using anti-S (Abcam), anti-FLAG (Sigma), and anti-ACE2

(Abcam) antibodies.

The CPEs of unfixed, unstained, infected cells were measured using an optical microscope, with the condenser down and the iris diaphragm partly closed as described in "Cytopathic Effects of Viruses Protocols," by Sushman E and Blair C (ASM Microbe Library. American Society for Microbiology. Archived from the original on June 2, 2012. Retrieved 20

November 2014). We consider total detachment of the monolayer cell as 100%. To determine the percentage of CPE, Vero-E6, HK-2, and Caco-2 cells were inoculated with SARS-CoV-2 at a virus titer of MOI = 1. The virus induced CPEs were monitored daily.

High-throughput sequencing raw reads were processed with Cutadapt (Martin, 2011) removing constant lentivirus vector and hair-pin loop nucleotide sequences. Resulting sequences were then filtered to contain only the anti-sense shRNA arms ranging from 15 to 40 bp, which were representative of the shRNA sequences. The shRNA sequences were then aligned to human reference genome (GENCODE GRCh38) using Bowtie2 (Langmead and Salzberg, 2012) . Subsequently, the number of reads mapping to shRNA probes and shRNA target genes were counted using HTSeq . The shRNA probes and shRNA target genes were generated by retrieving unique and properly paired alignments to nonoverlapping genomic locations with a mapping quality over 15, following by target gene annotation (GENCODE GRCh38 annotation v24). Finally, shRNA counts were median-J o u r n a l P r e -p r o o f normalized and analyzed with DESeq2 to infer changes between samples and to evaluate statistical significance. Next, we adjusted the gene counts of raw reads within each treatment group by median normalization. For each treatment group, the gene counts with missing values from individual replicates were removed. The resulting datasets were subjected to clustering analyses, including principal components analysis (PCA), volcano plot, and heat map hierarchical clustering analysis, using the stat function in R, ggplot2 function in R, and Complex Heatmap, respectively .

The average of the adjusted gene counts from SARS-CoV-2-infected samples was compared with that of the mock-infected samples. Fold changes of gene counts between the SARS-CoV-2-infected samples and the mock-infected samples were generated with counts from replicates higher than 100. GO enrichment analysis was performed on 3-fold enriched genes using DAVID GO Cellular Component using default parameters with p <0.05 . Similarly, molecular pathway analysis was conducted by analyzing the enriched VDFs into PANTHER (Protein Analysis Through Evolutionary Relationships) using the FISHER test type .

To generate the coronavirus life cycle map, relevant genes reported from different database:

UniProt , Gene Ontology , NCBI Reference

Sequence , NCBI OMIM database (McKusick, 2007) and BioGRID were obtained. Text mining of the genes properties, including biological processes, molecular functions, and cellular components, were analyzed using PubMed query via iHop (Hoffmann and Valencia, 2004) . The information was summarized J o u r n a l P r e -p r o o f 43 and integrated into a database framework to form a schematized coronavirus viral lifecycle.

Newly discovered VDFs were added manually into this framework according to the cellular functions inferred from literature curation and database review. All SARS-CoV-2 VDFs identified in this study are available in Table S1 .

Statistical significance was determined as p <0.05 using GraphPad Prism 8 unless otherwise

indicated. Experiments were analyzed by unpaired two-tailed t tests. J o u r n a l P r e -p r o o f

HTSeq--a Python framework to work with highthroughput sequencing data

Gene ontology: tool for the unification of biology. The Gene Ontology Consortium

Arrhythmia in COVID-19

The Universal Protein Resource (UniProt)

Loss of ADAM17 is associated with severe multiorgan dysfunction

Influenza and SARS-coronavirus activating proteases TMPRSS2 and HAT are expressed at multiple sites in human respiratory and gastrointestinal tracts

Inflammatory skin and bowel disease linked to ADAM17 deletion

Proteomics of SARS-CoV-2-infected host cells reveals therapy targets

The cardiac effects of SARS-CoV2: COVID-19 special issue

Engineering human ACE2 to optimize binding to the spike protein of SARS coronavirus 2

Extracellular vesicles containing ACE2 efficiently prevent infection by SARS-CoV-2 Spike protein-containing virus

Recent Insights into the Structure, Regulation, and Function of the V-ATPases

To degrade or not to degrade: mechanisms and significance of endocytic recycling

Effect of Chronic Kidney Disease on Changes in Vasopressin System Expression in the Kidney Cortex in Rats with Nephrectomy

Identification of Required Host Factors for SARS-CoV-2 Infection in Human Cells

Vasopressin and its analogues in shock states: a review

Angiotensin II mediates angiotensin converting enzyme type 2 internalization and degradation through an angiotensin II type I receptor-dependent mechanism

Ultrastructural Evidence for Direct Renal Infection with SARS-CoV-2

Angiotensin-Converting Enzyme 2: SARS-CoV-2 Receptor and Regulator of the Renin-Angiotensin System: Celebrating the 20th Anniversary of the Discovery of ACE2

Understanding SARS-CoV-2 endocytosis for COVID-19 drug repurposing

Complex heatmaps reveal patterns and correlations in multidimensional genomic data

Extrapulmonary manifestations of COVID-19

Modulation of TNF-alpha-converting enzyme by the spike protein of SARS-CoV and ACE2 induces TNF-alpha production and facilitates viral entry

Tissue distribution of ACE2 protein, the functional receptor for SARS coronavirus. A first step in understanding SARS pathogenesis

Bafilomycin A1, a specific inhibitor of vacuolar-type H(+)-ATPases, inhibits the receptor-mediated endocytosis of asialoglycoproteins in isolated rat hepatocytes

SRSF protein kinases 1 and 2 are essential host factors for human coronaviruses including SARS-CoV-2. bioRxiv

Functional interrogation of a SARS-CoV-2 host protein interactome identifies unique and shared coronavirus host factors

A gene network for navigating the literature

DAVID Bioinformatics Resources: expanded annotation database and novel algorithms to better extract biology from large gene lists

The long and the short cycle. Alternative intracellular routes for trafficking of G-protein-coupled receptors

Attenuation of ligand-induced activation of angiotensin II type 1 receptor signaling by the type 2 receptor via protein kinase C. Scientific reports 6, 21613

Cardiac Arrhythmias in Critically Ill Patients With COVID-19: A Brief Review

Plasma levels of soluble ACE2are associated with sex, Metabolic Syndrome, and its biomarkers in a large cohort, pointing to a possible mechanism for increased severity in COVID-19

Tumor necrosis factor-alpha convertase (ADAM17) mediates regulated ectodomain shedding of the severe-acute respiratory syndrome-coronavirus (SARS-CoV) receptor, angiotensin-converting enzyme-2 (ACE2)

Fast gapped-read alignment with Bowtie 2

Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia

Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

SARS-CoV-2 receptor ACE2 and TMPRSS2 are primarily expressed in bronchial transient secretory cells

Cutadapt removes adapter sequences from high-throughput sequencing reads

Angiotensin II-stimulated secretion of arginine vasopressin is inhibited by atrial natriuretic peptide in humans

Mendelian Inheritance in Man and its online version

PANTHER version 14: more genomes, a new PANTHER GO-slim and improvements in enrichment analysis tools

Inhibition of SARS-CoV-2 Infections in Engineered Human Tissues Using Clinical-Grade Soluble Human ACE2

The BioGRID interaction database: 2019 update

RNA Viruses and RNAi: Quasispecies Implications for Viral Escape

The sequence of human ACE2 is suboptimal for binding the S spike protein of SARS coronavirus 2

NCBI Reference Sequence (RefSeq): a curated non-redundant sequence database of genomes, transcripts and proteins

Cardiac arrhythmias among hospitalized Coronavirus 2019 (COVID-19) patients: prevalence, characterization, and clinical algorithm to classify arrhythmic risk

Interactions between vasopressin and the renin--angiotensin system

Decreased Renal Function Induced by High-Fat Diet in Wistar Rat: The Role of Plasma Angiotensin Converting Enzyme 2 (ACE2)

Circulating plasma concentrations of angiotensinconverting enzyme 2 in men and women with heart failure and effects of renin-angiotensinaldosterone inhibitors

Genome-Scale Identification of SARS-CoV-2 and Pan-coronavirus Host Factor Networks

Cell entry mechanisms of SARS-CoV-2

Actin-dependent endosomal receptor recycling

Renal histopathological analysis of 26 postmortem findings of patients with COVID-19 in China

SARS-CoV-2 entry factors are highly expressed in nasal epithelial cells together with innate immune genes

Consistent Detection of 2019 Novel Coronavirus in Saliva

Receptor Recognition by the Novel Coronavirus from Wuhan: an Analysis Based on Decade-Long Structural Studies of SARS Coronavirus

Genetic Screens Identify Host Factors for SARS-CoV-2 and Common Cold Coronaviruses

A comprehensive investigation of the mRNA and protein level of ACE2, the putative receptor of SARS-CoV-2, in human tissues and blood cells

Genome-wide CRISPR Screens Reveal Host Factors Critical for SARS-CoV-2 Infection

Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a singlecentered, retrospective, observational study

Endothelin-1 stimulates the in vitro release of neurohypophyseal hormones, but not corticotropin-releasing hormone, via ETA receptors

MERS coronavirus induces apoptosis in kidney and lung by upregulating Smad7 and FGF2

A genome-wide short hairpin RNA screening of jurkat T-cells for human proteins contributing to productive HIV-1 replication

Human tryptophanyl-tRNA synthetase is an IFN-gamma-inducible entry factor for Enterovirus

Roles for microRNAs, miR-93 and miR-130b, and tumor protein 53-induced nuclear protein 1 tumor suppressor in cell growth dysregulation by human T-cell lymphotrophic virus 1

Isolation of a novel coronavirus from a man with pneumonia in Saudi Arabia

Single-cell RNA-seq data analysis on the receptor ACE2 expression reveals the potential risk of different human organs vulnerable to 2019-nCoV infection

Human recombinant soluble ACE2 in severe COVID-19

Hoffmann and Valencia, 2004) infection-permissive human kidney cell line, Yeung et al. show that a soluble form of ACE2 that is cleaved and liberated from the host cell surface mediates SARS-CoV-2 binding and uptake by receptors