key: cord-0750927-oi01q4pz
authors: Du, Li; Bouzidi, Mohamed S.; Gala, Akshay; Deiter, Fred; Billaud, Jean-Noël; Yeung, Stephen T.; Dabral, Prerna; Jin, Jing; Simmons, Graham; Dossani, Zain; Niki, Toshiro; Ndhlovu, Lishomwa C.; Greenland, John R.; Pillai, Satish K.
title: Human Galectin-9 Potently Enhances SARS-CoV-2 Replication and Inflammation in Airway Epithelial Cells
date: 2022-05-16
journal: bioRxiv
DOI: 10.1101/2022.03.18.484956
sha: af98ee8e8cf9e647c065029fff7cc36a29ca767a
doc_id: 750927
cord_uid: oi01q4pz

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic has caused a global economic and health crisis. Recently, plasma levels of galectin-9 (Gal-9), a β-galactoside-binding lectin involved in immune regulation and viral immunopathogenesis, were reported to be elevated in the setting of severe COVID-19 disease. However, the impact of Gal-9 on SARS-CoV-2 infection and immunopathology remained to be elucidated. Here, we demonstrate that Gal-9 treatment potently enhances SARS-CoV-2 replication in human airway epithelial cells (AECs), including primary AECs in air-liquid interface (ALI) culture. Gal-9-glycan interactions promote SARS-CoV-2 attachment and entry into AECs in an ACE2-dependent manner, enhancing the binding affinity of the viral spike protein to ACE2. Transcriptomic analysis revealed that Gal-9 and SARS-CoV-2 infection synergistically induce the expression of key pro-inflammatory programs in AECs including the IL-6, IL-8, IL-17, EIF2, and TNFα signaling pathways. Our findings suggest that manipulation of Gal-9 should be explored as a therapeutic strategy for SARS-CoV-2 infection. Importance COVID-19 continues to have a major global health and economic impact. Identifying host molecular determinants that modulate SARS-CoV-2 infectivity and pathology is a key step in discovering novel therapeutic approaches for COVID-19. Several recent studies have revealed that plasma concentrations of the human β-galactoside-binding protein galectin-9 (Gal-9) are highly elevated in COVID-19 patients. In this study, we investigated the impact of Gal-9 on SARS-CoV-2 pathogenesis ex vivo in airway epithelial cells (AECs), the critical initial targets of SARS-CoV-2 infection. Our findings reveal that Gal-9 potently enhances SARS-CoV-2 replication in AECs, interacting with glycans to enhance the binding between viral particles and entry receptors on the target cell surface. Moreover, we determined that Gal-9 accelerates and exacerbates several virus-induced pro-inflammatory programs in AECs that are established signature characteristics of COVID-19 disease and SARS-CoV-2-induced acute respiratory distress syndrome (ARDS). Our findings suggest that Gal-9 is a promising pharmacological target for COVID-19 therapies.

In December 2019, the first known case of coronavirus disease 2019 , caused by SARS-CoV-2 infection, was reported in Wuhan, China. Rapidly, COVID-19 cases were reported worldwide. To date, SARS-CoV-2 has accounted for more than 360 million infections and more than 5.6 million deaths worldwide (1) . The rapid spread of SARS-CoV-2 continues to have a major impact on global health and the economy.

COVID-19 is mainly characterized by pneumonia, including fever, cough, and chest discomfort, and in severe cases dyspnea and lung infiltration (2) .

The major cause of death in COVID-19 cases is acute respiratory distress syndrome (ARDS) accompanied by a cytokine storm (3) . Several reports have identified specific circulating proteins and cytokines in blood plasma that are elevated in the setting of COVID-19 and may constitute clinically useful disease biomarkers (4, 5) . One such specific protein is Gal-9. Studies have recently revealed that plasma Gal-9 levels are elevated in COVID-19 patients and are positively correlated with COVID-19 severity (6) (7) (8) (9) . Furthermore, plasma levels of Gal-9 during COVID-19 were positively correlated with key pro-inflammatory cytokines, including interleukin 6 (IL-6), interferon gamma-induced protein 10 (IP-10), and tumor necrosis factor alpha (TNFα) (6) . However, the mechanism linking Gal-9 to severe COVID-19 disease remains to be elucidated.

Gal-9 belongs to the galectin family, which includes 15 carbohydrate binding proteins sharing a common carbohydrate-recognition domain (CRD) (10) . Conserved CRDs of galectins can homodimerize and have strong binding affinity for poly-N-acetyllactosamine (poly-LacNAc), which is present on the branches of N-and O-linked glycans (11) . Gal-9 is known for its regulation of immune responses and viral pathogenesis through glycan-mediated recognition. It is ubiquitously expressed in different tissues and cells in humans (e.g. endothelial cells, T lymphocytes, dendritic cells (DCs), macrophages, intestinal epithelial cells) and is localized in the extracellular matrix, surface, cytoplasm and nucleus of cells (12) . Circulating levels of Gal-9 serve as sensitive and non-invasive biomarkers in a broad range of conditions including cancer, autoimmunity, and infectious diseases, and the roles of Gal-9 vary with respect to cell type and disease state (13) . For example, Gal-9 suppresses antigen-specific CD8+ T cell effector functions via interaction with its receptor, TIM-3 (14) , and the Gal-9/TIM-3 axis promotes tumor survival through cross-talk with the PD-1 immune checkpoint (15) .

The functions of Gal-9 in the inflammatory response have been studied extensively.

Gal-9 enhances cytokine secretion in the human mast cell line (16) , and potentiates secretion of pro-inflammatory cytokines in inflammatory models of arthritis via TIM-3 interaction (17) . In specific relevance to viral pathogenesis, Gal-9 can bind to glycan structures expressed on the surface of both host cells and microorganisms to modulate antiviral immunity, and to promote or inhibit viral infection and replication (18) . was demonstrated to inhibit human cytomegalovirus (HCMV) infection by its carbohydrate recognition domains (18) . In hepatitis C virus (HCV)-infected individuals, virus infection induces Gal-9 secretion, which in turn induces pro-inflammatory cytokines leading to depletion of CD4+ T cells, apoptosis of HCV-specific cytotoxic T cells (CTLs), and expansion of regulatory T cells (Tregs) (19) . Gal-9 is elevated in human immunodeficiency virus-1 (HIV-1)-infected individuals (20, 21) , mediates HIV-1 6 transcription and reactivation (20, 22) , and potentiates HIV-1 infection by regulating the T cell surface redox environment (23) . These previous reports provide compelling evidence of the diverse roles of Gal-9 in viral infection and virus-associated immunopathology.

To date, a causal role of Gal-9 in SARS-CoV-2 pathology has not been demonstrated.

To address this gap, we investigated the effects of Gal-9 treatment on SARS-CoV-2 replication and pro-inflammatory signaling in immortalized and primary human airway epithelial cells (AECs). Our data show that Gal-9 facilitates SARS-CoV-2 replication and promotes virus-associated immunopathology in the human airway, motivating exploration into Gal-9 manipulation as a therapeutic strategy for COVID-19 disease.

To investigate the impact of Gal-9 on SARS-CoV-2 infection, we first determined the cytotoxicity of recombinant stable-form Gal-9, to optimize dosing in the immortalized Calu-3 AEC line. The 50% cytotoxic concentration (CC 50 ) value detected by MTT assay was 597 nM for Gal-9 in Calu-3 cells (Fig. 1A) . Next, we evaluated the effects of Gal-9 on SARS-CoV-2 replication at 50 nM, 100 nM, and 250 nM concentrations. Calu-3 cells were pre-treated with Gal-9 for six hours before viral infection (MOI=0.01), and Gal-9 was maintained in the media until 24 hours following infection. SARS-CoV-2 infection, as measured by quantitation of viral nucleocapsid (N) gene expression, was increased significantly by treatment with Gal-9 in a dose dependent manner (p<0.0001) (Fig. 1B) , with up to 27-fold induction at the highest concentration of Gal-9. Similarly, release of infectious virus in the supernatant was enhanced significantly by Gal-9 in a dose dependent manner, as measured by median tissue culture infectious dose (TCID 50 ) (p<0.05) (Fig. 1C) . The enhancement of virus production by Gal-9 was confirmed by specific staining of viral nucleocapsid protein using an immunofluorescence assay (IFA) (Fig. 1D ,E). Taken together, these data demonstrate that Gal-9 increases SARS-CoV-2 viral production in susceptible Calu-3 cells.

To investigate the stage of the virus replication cycle impacted by Gal-9, we treated Calu-3 cells with Gal-9 at the concentration of 250 nM, chosen based on our analyses of toxicity and dose response. Calu-3 cells were treated with Gal-9 before and after 8 virus infection. Protocols are illustrated in Extended Data Fig. 1A . Gal-9-mediated enhancement of virus production was significantly higher in pre-treated cells (p<0.05), as compared to cells treated with Gal-9 following viral exposure (Extended Data Fig.   1B ). These results suggested that Gal-9 likely impacts the early stage of the SARS-CoV-2 viral life cycle.

Based on the results of our time-course experiments, we sought to determine whether Gal-9 affects SARS-CoV-2 viral entry. Firstly, we examined the role of Gal-9 in SARS-CoV-2 cell-surface attachment. Cells were incubated with SARS-CoV-2 at 4ºC for 2 hours and attached SARS-CoV-2 viral particles were detected after washing the cells three times. Gal-9 induced a substantial, highly significant (p<0.0001), and dosedependent increase in cell-surface SARS-CoV-2 attachment ( Fig. 2A) . We next determined the capacity of Gal-9 to affect the entry of VSV-SARS-CoV-2 spike-ΔGluciferase reporter pseudovirus (SARS-2-S) into Calu-3 cells. Positive serum (P Serum), which was predetermined to possess SARS-CoV-2 neutralizing activity, potently reduced SARS-2-S infection (p<0.01) but did not suppress VSV-G infection (Fig. 2B ).

Gal-9 markedly enhanced SARS-2-S infection in a dose-dependent manner in Calu-3 cells which endogenously express ACE2 and TMPRSS2 (p<0.01) (Fig. 2B) , indicating that Gal-9 can potentiate SARS-CoV-2 attachment and entry. Unexpectedly, entry of VSV-G was also significantly enhanced by Gal-9 treatment (p<0.0001), suggesting that this pro-viral activity may be generalizable to other viral taxa.

ACE2 has been identified as the critical receptor for SARS-CoV-2 binding and entry (24) . To explore whether Gal-9-promoted virus entry depends on ACE2 binding, we treated cells with an anti-ACE2 antibody that competitively binds to the receptor. As expected, anti-ACE2 antibody blocked SARS-2-S but not VSV-G entry in a dose dependent manner (p<0.05) (Fig. 2C ). Anti-ACE2 antibody also blocked Gal-9enhanced virus infection (p<0.01) (Fig. 2D) , indicating that Gal-9 facilitates SARS-CoV-2 entry in an ACE2-dependent manner.

We next investigated the potential mechanisms underlying the Gal-9-mediated enhancement of SARS-CoV-2 viral entry. SARS-CoV-2 can use the endosomal cysteine proteases cathepsin B and L (CatB/L) and the serine protease TMPRSS2 to prime entry (25) . Only TMPRSS2 activity is essential for viral spread and pathogenesis in the infected host, whereas CatB/L activity is dispensable. In Calu-3 cells, which express ACE2 and TMPRSS2 (Fig. 3A) , SARS-CoV-2 entry was demonstrated to be primed by TMPRSS2 (26) . To determine if Gal-9 modulates ACE2 and TMPRSS2 surface expression to promote virus entry, we measured Calu-3 cell surface ACE2 and TMPRSS2 expression by flow cytometry. Gal-9 exhibited no effects on ACE2 and TMPRSS2 expression (Fig. 3B) , suggesting that Gal-9 facilitates virus entry independently of ACE2 and TMPRSS2 induction. We next investigated the direct impact of Gal-9 on the interaction between the SARS-CoV-2 spike protein and the ACE2 entry receptor, using purified spike and ACE2 proteins in an established sandwich ELISA protocol (27) . Gal-9 significantly enhanced binding between ACE2 and spike (p<0.05) ( Fig. 3C) , indicating Gal-9 facilitates virus entry, and viral replication at large, by strengthening ACE2 and spike interaction.

To explore the impact of Gal-9-glycan interactions on SARS-CoV-2 infection, we first determined if Gal-9-enhanced SARS-CoV-2 entry was dependent on CRD activity, relying on lactose, a competitive inhibitor of galectin carbohydrate-binding activity (28) .

Our data demonstrated that lactose treatment significantly abrogated Gal-9-mediated SARS-CoV-2 entry in a dose-dependent manner (p<0.001) (Fig. 4A ). We then deglycosylated Calu-3 target cells using two complementary approaches, kifunensine treatment and PNGase F exposure (29) . Kifunensine inhibits mannosidase I enzymatic activity within the cell, preventing hybrid and complex N-linked glycosylation of synthesized proteins, and PNGase F chemically removes N glycans from the cell surface. Loss of host complex N-glycans achieved using both of these strategies led to significant inhibition of Gal-9-enhanced SARS-CoV-2 entry (p<0.01) (Fig. 4B,C) . In the absence of Gal-9 treatment, PNGase F administration inhibited SARS-CoV-2 entry (p<0.05) (Fig. 4C ). Collectively, these data demonstrate that Gal-9 promotes SARS-CoV-2 attachment and entry into host cells in a glycan-dependent manner.

We next evaluated the temporal characteristics of Gal-9-mediated enhancement of SARS-CoV-2 replication. Our growth curves are compatible with the concept that Gal-9 facilitates SARS-CoV-2 entry, as the expression of the viral N gene was significantly increased within one to three hours post-infection (hpi) in Calu-3 cells (p<0.0001), as compared to untreated controls (Fig. 5A) . Maximum viral yields were detected at 36 hpi with Gal-9 treatment, and 48 hpi without Gal-9 treatment. In accordance with the viral growth kinetic data, microscopy images showed that virus-mediated cytopathic effects (CPE) were much more pronounced in infected cultures treated with Gal-9 at both 36 and 72 hpi, as compared to infected, Gal-9 untreated cultures (Fig. 5B ). The We next leveraged the RNA-seq data to examine the impact of Gal-9 treatment on SARS-CoV-2 expression, achieved by aligning sequencing reads against the USA-WA1/2020 reference genome. The number of reads mapping to each region of the viral genome was calculated and interpreted to infer viral expression patterns (Fig. 6E ). The transcription of SARS-CoV-2 exhibited an uneven pattern of expression along the genome, typically with a minimum depth in the first coding regions with ORFs 1a and 1b, and the maximum towards the 3' end. This skewing likely reflects the relative abundance of these sequences due to the nested transcription of SARS-CoV-2 subgenomic RNAs; all viral transcript variants include the terminal 3' genome segment (31) . Importantly, in accordance with our quantitative PCR data, Gal-9 treatment increased the expression of SARS-CoV-2, resulting in a ~10-fold increase in the number of sequencing reads mapping to the USA-WA1/2020 reference (Fig. 6E ,F), maintaining and even amplifying the observed 3' skewing of viral transcripts. and induction of TNFα was observed but failed to achieve statistical significance (p=0.055) (Fig. 7B ). Taken together, our findings in primary AECs validate and extend our previous observations, confirming that Gal-9 promotes SARS-CoV-2 replication and associated pro-inflammatory signaling in the airway epithelium.

The elevation of plasma Gal-9 levels in COVID-19 cases and severe COVID-19 disease has been confirmed in multiple reports (6, 33) . Here, we leveraged a recombinant stable-form of Gal-9 as a proxy for endogenously produced Gal-9, and investigated its impact on SARS-CoV-2 replication and host immune signaling. Our data reveal that Gal-9 enhances SARS-CoV-2 replication in AECs including human primary ALI-cultured AECs, facilitating cellular entry in a galectin-glycan interaction-dependent manner. Our transcriptomic data show that Gal-9 accelerates and exacerbates several virus-induced pro-inflammatory programs in AECs. These observations are highly relevant to the clinical manifestations and management of COVID-19, suggesting that circulating Gal-9 has a direct impact on viral infectivity and the cytokine milieu in the airway epithelium, which constitutes the critical initial site of SARS-CoV-2 attachment and infection (34) .

Collectively, our findings complement previous reports highlighting plasma Gal-9 level as a biomarker of severe COVID-19 disease, providing a novel molecular and immunologic framework linking Gal-9 activity to disease pathology (Extended Data Fig.   7 ). Importantly, our data build on a robust literature featuring Gal-9 as a key host factor regulating viral immunopathogenesis.

We observed potent Gal-9-mediated enhancement of SARS-CoV-2 cellular entry, using both pseudoviral constructs and wildtype replication-competent viruses. Previous studies have demonstrated that Gal-9 promotes HIV entry through retaining protein disulfide isomerase (PDI) on the CD4+ T cell surface (23) . Therefore, we initially hypothesized that Gal-9 promoted SARS-CoV-2 entry by retaining or increasing viral receptor expression on the cell surface. However, our flow cytometry data revealed no impact of Gal-9 exposure on ACE2 or TMPRSS2 surface expression. These data were further validated by our transcriptomic analysis which failed to show any modulation of ACE2 or TMPRSS2 mRNA transcripts by Gal-9. Studies have found roles for Gal-9 in bridging pathogen glycans to host cell surface glycans to promote target cell attachment (35) . We therefore conducted experiments to determine if this phenomenon underlies Gal-9 can promote cell-surface binding, internalization and cell invasion of many sexually transmitted pathogens, including bacteria, parasites, and viruses (37) . This enhancement is mediated by bridging the pathogen surface and receptors on host tissues in a glycosylation-dependent manner (37) . Our data further revealed that the cellular entry of VSV-G, like SARS-CoV-2, is enhanced by Gal-9 exposure, suggesting that entry enhancement of multiple viral taxa could be mediated by galectin-glycan lattice interactions.

Accumulating evidence suggests that fatal COVID-19 is characterized by a profound cytokine storm (38) . The overproduction of pro-inflammatory cytokines, such as IL-6, IL-8, TNFα, and IL-1β, leads to an increased risk of vascular hyperpermeability, acute lung injury, multiorgan failure, and eventually death when the high cytokine concentrations are unabated over time (39) . In direct relation to this phenomenon, our transcriptomic analysis revealed that Gal-9 and SARS-CoV-2 infection synergistically induced the expression of key pro-inflammatory programs in AECs including the IL-8, IL-17, EIF2

and TNFα signaling pathways. These data were validated in part at the protein level using a bead-based immunoassay to characterize the AEC secretome. Bozorgmehr and colleagues demonstrated highly significant positive correlations of plasma Gal-9 levels with a wide range of pro-inflammatory biomarkers in COVID-19 patients. They further demonstrated that Gal-9 treatment of monocytes in vitro enhanced expression and production of key pro-inflammatory molecules associated with severe COVID-19 disease (6). Gal-9 induces secretion of inflammatory cytokines in several immune cell lineages including monocyte-derived macrophages and neutrophils (40, 41) . Our findings validate and extend these observations to SARS-CoV-2 infection of the airway epithelium.

Importantly, the combined effects of SARS-CoV-2 infection and Gal-9 exposure on proinflammatory signaling are much stronger than either Gal-9 or SARS-CoV-2 alone, reflecting a synergistic interaction. We observed little transcriptional perturbation by SARS-CoV-2 alone after 24 hours of infection, which is consistent with other reports that profound transcriptional changes are only evident after 48 hours (42, 43) .

Interestingly, the IL-17 signaling pathway was significantly activated after Gal-9 treatment in the presence or absence of SARS-CoV-2. Our data are in accordance with a previous report demonstrating induction of IL-17 signaling by Gal-9 treatment in vivo, in the setting of sepsis (44) . For MERS-CoV, SARS-CoV and now SARS-CoV-2, disease severity has been shown to positively correlate with levels of IL-17 and other T helper 17 (Th17) cell-related pro-inflammatory cytokines (45) (46) (47) . IL-17 inhibition has been adopted as a common and successful strategy to reduce the injury associated with inflammatory autoimmune diseases (48) . Thus, inhibition or neutralization of Gal-9

would not only decrease SARS-CoV-2 replication but also attenuate IL-17 signaling and other damaging pro-inflammatory cascades.

Our study has limitations that must be considered. Firstly, we focused exclusively on the airway epithelium, and it is now established that SARS-CoV-2 is capable of infecting other cell lineages including monocytes, monocyte-derived macrophages, and microglia (49) (50) (51) . It is possible that Gal-9 does not exert similar effects on viral replication or immune signaling in other target cell types. Secondly, our studies were all performed in vitro or in transplant tissue-derived primary epithelial cells ex vivo. As it is wellestablished that Gal-9 exerts conditional, pleiotropic immunomodulatory effects, the net effect of Gal-9 signaling on SARS-CoV-2 pathogenesis cannot be fully appreciated outside of an animal model with a functional immune system. Validation and extension of our observations in murine, hamster or nonhuman primate models of SARS-CoV-2 infection will help in evaluating the clinical relevance of our reported findings. In relevance to the implementation of animal models, our study did not elucidate the principal cell or tissue sources responsible for secreting Gal-9 in the setting of SARS-CoV-2 infection. Leveraging an animal model to identify these source compartments will be critical in developing interventions to manipulate Gal-9 signaling as a therapeutic approach.

To our knowledge, the data presented here are the first to show that Gal-9 is directly 

The Calu-3 human lung adenocarcinoma epithelial cell line was obtained from ATCC (ATCCHTB-55) and cultured in Eagle's Minimum Essential Medium (EMEM). Vero E6 cells were purchased from ATCC (CRL-1586) and cultured in Dulbecco's Modified Eagle's medium (DMEM). Vero E6 cells stably expressing TMPRSS2 (Vero E6-TMPRSS2) were established and cultured in DMEM in the presence of puromycin (1µg/ml). All media were supplemented with 10% FBS and 1% penicillin/streptomycin.

All cells had been previously tested for mycoplasma contamination and incubated at 37°C in a humidified atmosphere with 5% CO2.

Human bronchus was harvested from five explanted healthy lungs. The tissue was submerged and agitated for one minute in PBS with antibiotics and 5 mM dithiothreitol to wash and remove mucus. After three washes, the tissue was placed in DMEM with 0.1% protease and antibiotics overnight at 4ºC. The next day, the solution was agitated and the remaining tissue removed. Cells were centrifuged at 300g/4ºC for five minutes, then the cell pellet was resuspended in 0.05% trypsin-EDTA and incubated for five minutes at 37ºC. The trypsinization reaction was neutralized with 10% FBS in DMEM, then cells were filtered through a cell strainer and centrifuged at 300g/4ºC for five minutes. The cell pellet was resuspended in 10% FBS in DMEM and 10 μ l was stained with trypan-blue and counted on a hemocytometer. 75,000 cells were plated onto each 6mm/0.4mm Transwell ALI insert after treatment with FNC coating mixture. 10% FBS in DMEM and ALI media were added in equal volumes to each basal compartment and cultures were incubated at 37ºC with 5% CO 2 . The next day, the media was removed and both compartments were washed with PBS and antibiotics. ALI media was then added to each basal compartment and changed every three days until cells were ready for use at day 28.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), isolate USA- 

The cytotoxic effect of Gal-9 on Calu-3 cells was determined using an MTT assay kit (abcam, ab211091) according to manufacturer's guidelines. In brief, Calu-3 cells cultured in 96-well cell culture plates were incubated with different concentrations of Gal-9 (0-5,000 nM). After 48 h, the media was removed and 100 μ l MTT reagent (1:1 dilution in DMEM medium (serum free)) was added to each well and incubated for 3 h at 37ºC. Then the medium was removed, and 150 μ l MTT solvent was added into each well. Quantification was performed by reading absorbance at OD=590 nm. The data from three independent experiments was used to calculate the CC 50 by nonlinear regression using GraphPad Prism 8.0 software.

Stable-form recombinant Gal-9 was used in all experiments (20, 52) . Calu-3 cells were seeded at 0.5 x 10 6 cells per well in 0.5 ml volumes using a 24-well plate, or were to enable virus entry. Subsequently, the cells were washed and fresh ALI medium (500 μ l) was added into the basal compartment. Cells were incubated at 37ºC (5% CO 2 ) and harvested for analysis at 36 hours post-infection.

Viral production by infected cells was measured by quantifying TCID 50 . Vero E6 cells were plated in 96-well plates at 5 X 10 4 cells per well. The next day, supernatants collected from Calu-3 cells were subjected to 10-fold serial dilutions (10 1 to 10 11 ) and inoculated onto Vero E6 cells. The cells were incubated at 37ºC with 5% CO 2 . Three to five days post infection, each inoculated well was evaluated for presence or absence of viral CPE. TCID 50 was calculated based on the method of Reed and Muench (56) .

Cells were washed with 1 X PBS, then were fixed and permeabilized with cold methanol : acetone (1:1) for 10 min at 4ºC. Next, cells were washed with 1 X PBS and incubated in a blocking buffer (5% goat serum (Seracare Life Sciences Inc, 55600007)) at room temperature for 30 min. Cells were then incubated with a primary antibody (monoclonal rabbit anti-SARS-CoV-2 nucleocapsid antibody (GeneTex, GTX135357)) in 1 X PBS (1:1,000) overnight at 4ºC. The following day, cells were washed three times with 1 X PBS and incubated with a secondary antibody (Goat anti-Rabbit IgG (H+L) secondary antibody, FITC (Thermo Fisher, 65-6111)) in 1 X PBS (1:200) for 1 h at 37ºC. Then cells were washed three times with 1 X PBS and incubated with DAPI (300 nM) (Thermo Fisher Scientific, D1306) for 5 min at room temperature. Images were acquired using a fluorescence microscope.

Total RNA was extracted using the chloroform-isopropanol-ethanol method. 500 ng of RNA was reversed transcribed into cDNA in a 20 

RNA concentration and quality was measured using High Sensitivity RNA ScreenTape Analysis (Agilent, 5067-1500). cDNA libraries were prepared using the Illumina TruSeq Stranded total RNA library prep kit (Illumina, 20020597) and sequencing was performed on the Illumina Nextseq 550 platform generating 75 bp paired-end reads. The quality of raw sequencing reads was assessed using FastQC. Differentially-expressed genes were identified by GSA or ANOVA in Partek® Flow® imported into the QIAGEN Ingenuity Pathway Analysis (IPA) software application. IPA was used to identify gene ontologies, pathways and regulatory networks to which differentially-expressed genes belonged to, as well as upstream regulators (57) . Reads were also aligned to the SARS-CoV-2 isolate WA-1 and analyzed using the QIAGEN CLC Genomics Workbench.

VSVΔG-luciferase-based viruses, in which the glycoprotein (G) gene has been replaced with luciferase, were produced by transient transfection of viral glycoprotein expression plasmids (pCG SARS-CoV-2 Spike and pCAGGS VSV-G) or no glycoprotein control into HEK293T cells by TransIT-2020 as previously described (58) . In brief, cells were seeded into 15-cm culture dishes and allowed to attach for 12 hours before transfection 

Calu-3 cells were plated into 96-well plates. The following day, cells were pre-treated with indicated concentrations of Gal-9 for six hours. In order to block ACE2 on the cell surface, cells were pretreated with indicated concentrations of an anti-ACE2 antibody for one hour. An unrelated anti-goat IgG antibody was used as a control. Pseudovirus harboring either SARS-CoV-2 spike or VSV-G glycoprotein were diluted in EMEM containing indicated concentrations of Gal-9, and then were added to Calu-3 cells.

Controls included wells with serum predetermined to possess neutralizing activity. Cells were incubated for 24 h at 37ºC with 5% CO 2 . Supernatants were then removed, cells were lysed, and luciferase activity was read using a commercial substrate (Promega, E1500).

Cells were detached with 10% EDTA containing Zombie NIR (1:300) (BioLegend, 423105) for 10 min at 37ºC. Then cells were washed three times with 1 X PBS and incubated with human ACE2 Alexa Fluor 488-conjugated antibody (R&D Systems, FAB9332G) or human TMPRSS2 Alexa-Fluor 594-conjugated antibody (R&D Systems, FAB107231T) for 30 min at room temperature. Cells were washed three times with 1 X PBS again. Analytical flow cytometry was performed with BD LSRII flow cytometer.

Data was analyzed using FlowJo.

Assessment of Gal-9 effects on the binding of the SARS-CoV-2 spike to human ACE2 was performed using a commercially available spike-ACE2 binding assay kit (CoviDrop SARS-CoV-2 Spike-ACE2 Binding Activity/Inhibition Assay Kit, EPIGENTEK, D-1005-48) following the protocol provided by the manufacturer. Gal-9 or positive inhibitor control was mixed with indicated amounts of recombinant human ACE2 protein, then added to an ELISA plate coated with recombinant SARS-CoV-2 spike protein and incubated at 37ºC for 60 min. Unbound ACE2 was removed. The amount of captured ACE2, which is proportional to ACE2 binding activity, is then recognized by an ACE2 detection antibody and measured by reading the absorbance at a wavelength of 450 nm.

Cytokines in the cell culture supernatants were measured using the human cytokine 

To measure the frequency of infected cells, randomly-selected areas were imaged.

Each treatment had three replicates. The percentage of GFP-positive cells was determined by dividing the number of GFP+ cells by the number of DAPI+ cells. All samples were analyzed at the same threshold values. For quantification of GFP+ cells, CellProfiler-3 software was used to determine the fraction of GFP+ cells. Briefly, we used the software pipeline CorrectIlluminationCalculate to calculate an illumination correction function for each of the two channels (DAPI (blue) and GFP (green)). We then used another pipeline, CorrectIlluminationApply, to load each image and correct its illumination using the pre-calculated functions. Next, we ran ColorToGray to change all slides to gray and ran IdentifyPrimaryObjects to identify the number of each channel.

Finally, data were exported by using the ExportToSpreadsheet pipeline. The same threshold value was applied to the images of each area.

The combinatorial effects of Gal-9 treatment and SARS-CoV-2 infection on proinflammatory cytokine expression were analyzed using the SynergyFinder web application, implementing the Bliss Independence model. The Bliss model generates synergy scores from a response matrix (59) .

Statistical analysis was performed using GraphPad Prism version 8 software. Data were presented as means ± SEM or median. Data were analyzed for statistical significance using an unpaired or paired Student's t test to compare two groups, or using a paired 

Cellular toxicity was examined in Calu-3 cells using an MTT assay and was expressed as relative cell viability as compared to Gal-9-untreated control (set at 100%). The are shown by the 2D synergy landscape. When the synergy score is less than -10, the interaction between two treatments is likely to be antagonistic; from -10 to 10, the interaction between two treatments is likely to be additive; larger than 10, the interaction between two treatments is likely to be synergistic. 

An interactive web-based dashboard to track COVID-19 in real time

Characteristics of SARS-CoV-2 and COVID-19

How COVID-19 induces cytokine storm with high mortality

An inflammatory cytokine signature predicts COVID-19 severity and survival

Profiling serum cytokines in COVID-19 patients reveals IL-6 and IL-10 are disease severity predictors

Galectin-9, a Player in Cytokine Release Syndrome and a Surrogate Diagnostic Biomarker in SARS-CoV-2 Infection

Interleukin-18 Is a Potential Biomarker to Discriminate Active Adult-Onset Still's Disease From COVID-19

Proteomic blood profiling in mild, severe and critical COVID-19 patients

High Levels of the Cleaved Form of Galectin-9 and Osteopontin in the Plasma Are Associated with Inflammatory Markers That Reflect the Severity of COVID-19 Pneumonia

Carbohydrate-recognition domains of galectin-9 are involved in intermolecular interaction with galectin-9 itself and other members of the galectin family

Glycosylation-dependent galectin-receptor interactions promote infection

Galectin-9 and VISTA Expression Define Terminally Exhausted T Cells in HIV-1 Infection

Galectin-9 as a biomarker of disease severity

Galectin-9/TIM-3 interaction regulates virus-specific primary and memory CD8 T cell response

Galectin-9 interacts with PD-1 and TIM-3 to regulate T cell death and is a target for cancer immunotherapy

Galectin-9 enhances cytokine secretion, but suppresses survival and degranulation, in human mast cell line

Promotion of tissue inflammation by the immune receptor Tim-3 expressed on innate immune cells

Restriction of Human Cytomegalovirus Infection by Galectin-9

Interaction between Galectin-9/TIM-3 pathway and follicular helper CD4 T cells contributes to viral persistence in chronic hepatitis C

Human Galectin-9 Is a Potent Mediator of HIV Transcription and Reactivation

Elevated cerebrospinal fluid Galectin-9 is associated with central nervous system immune activation and poor cognitive performance in older HIV-infected individuals

Effect of the use of Galectin-9 and blockade of TIM-3 receptor in the latent cellular reservoir of HIV-1

Galectin-9 binding to cell surface protein disulfide isomerase regulates the redox environment to enhance T-cell migration and HIV entry

ACE2: Evidence of role as entry receptor for SARS-CoV-2 and implications in comorbidities

Mechanisms of SARS-CoV-2 entry into cells

Cell Entry Depends on ACE2 and TMPRSS2 and Is Blocked by a Clinically Proven Protease Inhibitor

Circulating plasma concentrations of angiotensin-converting enzyme 2 in men and women with heart failure and effects of renin-angiotensin-aldosterone inhibitors

suppresses B cell receptor signaling and is regulated by I-branching of N-glycans

Kifunensine, a potent inhibitor of the glycoprotein processing mannosidase I

Pathological inflammation in patients with COVID-19: a key role for monocytes and macrophages

The architecture of the SARS-CoV-2 RNA genome inside virion

SARS-CoV-2 infects an upper airway model derived from induced pluripotent stem cells

Marked T cell activation, senescence, exhaustion and skewing towards TH17 in patients with COVID-19 pneumonia

SARS-CoV-2 infection of primary human lung epithelium for COVID-19 modeling and drug discovery

Microbe-Host Interactions are Positively and Negatively Regulated by Galectin-Glycan Interactions

Potential Role of Galectin-glycan Lattices in SARS-CoV-2 Infection and Pathogenesis: A Hypothesis

Galectins as potential therapeutic targets in STIs in the female genital tract

SARS-CoV-2 infection: The role of cytokines in COVID-19 disease

van Spriel AB. 2019. Intracellular Galectin-9 Controls Dendritic Cell Function by Maintaining Plasma Membrane Rigidity. iScience

Galectin-9 expands unique macrophages exhibiting plasmacytoid dendritic cell-like phenotypes that activate NK cells in tumorbearing mice

Imbalanced Host Response to SARS-CoV-2 Drives Development of COVID-19

Characterization of the SARS-CoV-2 Host Response in Primary Human Airway Epithelial Cells from Aged Individuals

Galectin-9 prolongs the survival of septic mice by expanding Tim-3-expressing natural killer T cells and PDCA-1+ CD11c+ macrophages

MERS-CoV infection in humans is associated with a pro-inflammatory Th1 and Th17 cytokine profile

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

SARS-CoV-2 infection initiates interleukin-17-enriched transcriptional response in different cells from multiple organs

COVID-19: a case for inhibiting IL-17?

SARS-CoV-2 cell tropism and multiorgan infection

Monocytes and Macrophages, Targets of Severe Acute Respiratory Syndrome Coronavirus 2: The Clue for Coronavirus Disease

SARS-CoV-2 Infection of Microglia Elicits Pro-inflammatory Activation and Apoptotic Cell Death

Development of highly stable galectins: truncation of the linker peptide confers protease-resistance on tandem-repeat type galectins

482 Phase1/2 study of an anti-galectin-9 antibody, LYT-200, in patients with metastatic solid tumors

Targeting the C-terminus of galectin-9 induces mesothelioma apoptosis and M2 macrophage depletion

Galectin antagonist use in mild cases of SARS-CoV-2; pilot feasibility randomised, open label

A Simple Method for Estimating Fifty Per Cent Endpoints

Causal analysis approaches in Ingenuity Pathway Analysis

SARS-CoV-2 seroprevalence and neutralizing activity in donor and patient blood

SynergyFinder 2.0: visual analytics of multi-drug combination synergies

In the pre-infection treatment experiments, Calu-3 cells were pre-treated with 250 nM Gal-9 for six hours. Cells were washed and incubated with SARS-CoV-2 (MOI=0.01) for one hour. Then cells were washed again and were supplemented with fresh media. 24 hpi, cells were harvested for RNA isolation and RT-qPCR targeting the N gene. In the post-infection treatment experiments, cells were infected with SARS-CoV-2 (MOI=0.01) for one hour and washed with PBS

Extended Data Fig. 2 Gal-9 treatment and SARS-CoV-2 infection induce secretion of select pro-inflammatory cytokines

This study was supported by the Program for Breakthrough Biomedical Research, which is partially funded by the Sandler Foundation. Additional support was provided by National Institutes of Health grants R01MH112457 (SKP) and the University of California San Francisco-Gladstone Institute of Virology & Immunology Center for AIDS Research (P30 AI027763).

The authors declare no competing interests.