Integrin activation is an essential component of SARS-CoV-2 infection

SARS-CoV-2 infection depends on binding its spike (S) protein to angiotensin-converting enzyme 2 (ACE2). The S protein expresses an RGD motif, suggesting that integrins may be co-receptors. Here, we UV-inactivated SARS-CoV-2 and fluorescently labeled the envelope membrane with octadecyl rhodamine B (R18) to explore the role of integrin activation in mediating cell entry and productive infection. We used flow cytometry and confocal microscopy to show that SARS-CoV-2R18 particles engage basal-state integrins. Furthermore, we demonstrate that Mn2+, which induces integrin extension, enhances cell entry of SARS-CoV-2R18. We also show that one class of integrin antagonist, which binds to the αI MIDAS site and stabilizes the inactive, closed conformation, selectively inhibits the engagement of SARS-CoV-2R18 with basal state integrins, but is ineffective against Mn2+-activated integrins. RGD-integrin antagonists inhibited SARS-CoV-2R18 binding regardless of integrin activation status. Integrins transmit signals bidirectionally: 'inside-out' signaling primes the ligand-binding function of integrins via a talin-dependent mechanism, and 'outside-in' signaling occurs downstream of integrin binding to macromolecular ligands. Outside-in signaling is mediated by Gα13. Using cell-permeable peptide inhibitors of talin and Gα13 binding to the cytoplasmic tail of an integrin's β subunit, we demonstrate that talin-mediated signaling is essential for productive infection.

. Integrin conformational states antagonist targets and SARS-CoV-2 binding. (A) Integrin States: First, the inactive, bent-closed state (BCS), with a closed headpiece and low affinity for extracellular matrix (ECM) ligands. The bent structure inhibits the receptors from inappropriate signaling due to random binding to extracellular matrix proteins. In the BCS form, binding to large ligands is likely limited. Second, when primed, integrins exhibit an extended-closed state (ECS) with a closed headpiece and higher ligand binding affinity than BCS. Third, active and extended-open state (EOS) with an open headpiece and maximum affinity for ECM ligands. Integrin Affinity Regulation: Mn 2+ binding to the MIDAS site at the αI and βI domain integrin induces integrin extension. α 2 β 1 integrin antagonist BTT 3033 binds to the α-I domain, and stabilizes the BCS. GLP0187 blocks binding to the RGD ligand-binding domain. EOS binding to a macromolecular ligand or ECM generates a force (F) transmitted through the integrin β subunit. (B) Model of Sars-CoV-2 virion structure (https:// www. scien tific ameri can. com/ inter active/ inside-the-coron avirus/). SARS-CoV-2 are spherical or ovoid particles of sizes that span the range of 60-140 nm. The SARS-CoV-2 virion consists of a lipid bilayer envelope membrane covering a large nucleoprotein (N)-encapsidated, positive-sense RNA genome. The lipid envelope is decorated with three transmembrane proteins consisting of trimeric spike proteins (S) that project above the lipid bilayer membrane and relatively small membrane (M) and envelope (E) proteins 78,79 . S proteins bind with highaffinity (1-50 nM) 4 to the angiotensin-converting enzyme 2 (ACE2) for productive infection 80 . (C) Cartoon alignment of the receptor-binding domain (RBD) and RGD sequence on the trimeric spike protein, which favors engagement of activated integrin, adapted from ref. 25 The illustrations were generated using Microsoft ® PowerPoint Version 16 www.nature.com/scientificreports/ integrin signaling during cell entry of SARS-CoV-2 R18 and infectious SARS-CoV-2. Taken together, our results demonstrate that integrins play a significant role in the infectivity of SARS-CoV-2.

Results
Integrin extension promotes SARS-CoV-2 R18 cell entry. To facilitate studies of SARS-CoV-2 hostcell entry outside of BSL-3 containment, we generated UV-inactivated virus particles. Under our experimental conditions, a minimum UV dose of 100 mW s/cm 2 was sufficient to completely inactivate 10 7 virions/ml distributed in 500 µl samples of a twelve well plate ( Fig. 2A,B). UV-inactivated virus samples were fluorescently labeled with a lipophilic lipid probe, octadecyl rhodamine B (R18), intercalating the envelope membrane 44 . Labeled samples were purified and characterized as we have described previously for the Sin Nombre virus 45 .
We hypothesized that activating integrins by Mn 2+24,36 , which induces integrin extension and higher ligand affinity, would provide a favorable spatial orientation of the RGD-binding motifs to facilitate SARS-CoV-2 R18 binding (Fig. 1A). Therefore, we measured initial rates (< 10 min binding time) of binding in activated cells (Mn 2+ /Ca 2+ ) relative to resting (1 mM Ca 2+ only). SNV-CoV-2 R18 bound to the Mn 2+ activated samples at 3 times the rate of untreated cells (not shown). However, at equilibrium (> 20 min incubation), the cell occupancy of SARS-CoV-2 R18 was only ~ 20% higher in Mn 2+ -treated samples compared to untreated samples. As expected, the gap increased when the virus was used as a limiting reagent 46 . We used estimation plots 47 , to assess the precision of the results from multiple experiments of the integrin function assay that measured the relative binding of SARS-CoV-2 R18 to Mn 2+ -activated cells relative to quiescent cells on different days (Fig. 3A-D). The mean difference between the binding to Mn 2+ -activated and resting cells was conserved across different samples when we used at least 5000 SARS-CoV-2 R18 /cell. However, when we used a lower stoichiometric ratio, e.g., 3000 SARS-CoV-2 R18 /cell, the gap between the site occupancy of Mn 2+ activated and resting cells increased to 30%, as discussed below.
To further investigate the role of integrins in SARS-CoV-2 R18 entry into Vero E6 cells, we used high binding affinity integrin antagonists: (1) BTT 3033, a selective antagonist (EC 50 = 130 nM) of integrin α 2 β 1 that binds to Duplicate plaque assays of supernatants of Sars-CoV-2 exposed to increasing doses of 254 nm radiation and then tested for viability. The live virus completely lysed the cells at 1:100 dilution relative to UV exposed virions. (B) Graph shows UV doseresponse, leading to a significant decrease in plaque-forming units at different doses. For our experiments, a 90 s (450 mW s/cm 2 ) UV dose was used to inactivate the virus before removal from the BSL-3 laboratory. (C) Confocal microscopy imaging of cells after incubation with SARS-CoV-2 R18 (magenta) for 15 min, then fixed and labeled for early endosome marker, early endosome antigen 1 or EEA1 (green), an effector protein for Rab5, and nuclei (Hoechst 33,258, blue). SARS-CoV-2 R18 vesicles are trafficked to the perinuclear region, and a subset is co-localized with EEA1. Images are maximum projections and have been brightness and contrast-enhanced. www.nature.com/scientificreports/   49 derived from the synergy region of fibronectin 50 , known to exhibit specific antagonism for α 5 β 1 and α IIb β 3 and also recently shown to inhibit SARS-CoV-2 infectivity 27 , and (3) GLPG0187, a highaffinity, broad-spectrum (EC 50 < 10 nM) integrin receptor antagonist of RGD integrins α 5 β 1 , α v β 3 , α v β 5 , α v β 1 , α v β 6 51 . We used a titrated, fivefold excess of unlabeled SARS-CoV-2 relative to fluorescent SARS-CoV-2 R18 as a control for competitive inhibition of SARS-CoV-2 R18 binding. Paired samples of cell suspensions in Mn 2+ -replete and Mn 2+ -free media were treated with the above integrin antagonists. Total viral binding was normalized to Mn 2+ -treated samples for each experimental condition. The graphs show that Mn 2+ treatment increased SARS-CoV-2 R18 occupancy of cells by ~ 20% compared to Mn 2+ -free conditions (Fig. 3E,F). As noted above, using significantly fewer SARS-CoV-2 R18 than 5000 increases the gap between binding to the Mn 2+ -activated and resting as indicated for the GLPG0187 sample (Fig. 3G). We also show in a subsequent experiment that raising SARS-CoV-2 R18 to a stoichiometric excess limited the site occupancy gap to 20% (Supplemental Figure 1).
The positive control for inhibition (5xCov-2 in data graphs) blocked 80% of SARS-CoV-2 R18 and equally inhibited Mn 2+ -treated and untreated samples. Reasoning that the residual signal of 5× Cov-2 treated samples was due to non-specific binding to the cell membrane, we subtracted the fluorescence of cells blocked with 5xCov-2 and then normalized the data to mock-treated cells. Finally, we compared the relative efficacy of the inhibitors in Mn 2+ -replete and -free conditions of the normalized data ( Fig. 3H-J). The fraction of Mn 2+ -activated integrins (20%) were refractory to BTT 3033 treatment (Fig. 3H). BTT 3033 selectively binds to the BCS integrin structure 48 and does not bind to Mn 2+ activated integrins. In contrast, ATN-161 and GLPG0187 were agnostic to Mn 2+ treated cells, as the same baseline was achieved for either condition (Fig. 3I,J). Overall, GLPG0187 (Fig. 3I) appeared to be a better competitive inhibitor of SARS-CoV-2 R18 compared to ATN-161 (Fig. 3J). The difference for the latter was potentially due to ATN-161's overall specificity for integrin α 5 β 1 . Thus, the expression level of α 5 β 1 , in Vero E6 cells, relative to other integrins with which α 5 β 1 would compete for SARS-CoV-2 R18 engagement governed its apparent efficacy. Also, ATN-161 is known to exhibit U-shaped dose-response characteristics 49 thus, presenting a need to identify an optimally active-dose by titration of ATN-161 27 which is beyond the scope of our present study. The mechanistic specificity of integrin inhibition by these antagonists regarding SARS-CoV-2 uptake strongly supports the idea that (1) integrin RGD engagement is an essential co-factor for cell entry and (2) integrin extension is required for cell entry based on BTT 3033's mechanism of action.

Inhibition of integrin activation or binding to SARS-CoV-2 R18 blocks intracellular trafficking.
We then used live-cell confocal microscopy to visualize Vero E6 cell entry and trafficking of SARS-CoV- www.nature.com/scientificreports/ CoV-2 R18 particles were visible at cell membranes within 3 min, subsequently developed punctate features at the cell periphery, and trafficked to the perinuclear space. The rate of cell entry (time to perinuclear space ~ 10 min) was comparable to infectious virions 52 . For the BTT 3033-treated cells, early peripheral membrane localization of SARS-CoV-2 R18 showed significant diminution of discernable puncta. It did not undergo retrograde traffic towards the perinuclear region within the timeframe of the experiment. The relative amount of virus binding to the surface was also reduced with BTT 3033 treatment (Fig. 4A,B), consistent with reduced binding observed by flow cytometry measurements (Fig. 2).

Blocking of integrin signaling significantly inhibits productive infection of cells by SARS-CoV-2.
Integrin activation is a complex and well-regulated spatiotemporal process involving the synchronized assembly and disassembly of multiple signaling elements at the integrin's β-cytoplasmic tail (β-CT) [53][54][55] . Various groups have described a network of up to 156 interacting components that comprise the integrin adhesome [56][57][58][59] . Some adhesome components relevant to our study are shown in (Fig. 5A). Most β-CTs contain conserved sequences needed for integrin activity, such as the two β chain NPxY/F sequences, which are sites of competitive binding by adaptor proteins that regulate integrin activation and deactivation 41,53 , including sorting signals for clathrin-mediated endocytosis [60][61][62][63] . The phosphorylatable tyrosine (Y) residues of NPxY motifs are key regulatory sites of integrin activation on the β-CT. For example, N 780 PIY 783 and N 792 PKY 795 in β 1 and N 744 PLY 747 and N 756 ITY 759 in β 3 are motifs phosphorylated by Src family kinases (SFK) that may positively or negatively regulate interactions with phosphotyrosine-binding (PTB) domain-containing proteins. During the early stage of integrin activation, inhibitory proteins are displaced from the β-CT in exchange for integrin activators, ending with the recruitment of talin to the integrin tail 38 . However, the early wave of talin-mediated insideout signaling is transiently terminated to allow Gα 13 , the effector of outside-in signaling, to bind to the conserved ExE motif (where x denotes any residue for specific integrins, e.g., EEE for β 3 -CT and EKE for β 1 -CT Fig. 5A), which overlaps the talin binding domain 42 . Integrin binding to macromolecular ligands, such as SARS-CoV-2, facilitates Gα 13 -mediated outside-in signaling. Transmission of the tensile force through the integrin to talin stabilizes high-affinity integrin binding (in the EOS) to the ECM promotes the 'second wave' of inside-out signaling (Fig. 5B). The sequential mechanism of inside-out and outside-in signaling was previously established in part by the use of two myristoylated peptides, mP6 (Myr-FEEERA-OH), derived from the Gα 13 -binding domain and mP13 (Myr-KFEEERARAKWDT-OH) mimicking the β 3 -CT's talin binding domain 42 . It is worth noting that the previous mP6 and mP13 related study by Shen et al. 42 established that the minimal sequence of EEERA does not interact with talin and is a specific inhibitor of Gα 13 association with the β-CT and had no effect on talin-dependent inside-out signaling, or the late phase of outside-in signaling associated with the second wave of talin binding. However, mP13 affects all phases of integrin signaling 42 . To investigate the relationship between the integrin signaling events and SARS-CoV-2 engagement and cell entry, we treated cells with mP6 peptide, which inhibited cell entry of SARS-CoV-2 R18 in flow cytometry and microscopy experiments (Fig. 5C-E). Similarly, mP13 inhibited cell entry in flow cytometry experiments (data not shown). The results for mP6 treated cells suggest that SARS-CoV-2 engagement initiates a Gα 13 -mediated outside-in integrin activation without a known receptor stimulus which is consistent with the idea that SARS-CoV-2 binding induces integrin activation 64 , as we previously demonstrated for the Sin Nombre virus 24 . Figure 5. Inhibition of integrin activation blocks cell entry of SARS-CoV-2 R18 , suggesting integrin-mediated signaling is required for productive infection. (A) Aligned sequences of β1 and β3-integrin cytoplasmic tails (β-CT). The NPxY motif tyrosine residues (shown in brown) and the Ser and Thr residues (shown in purple) are important phosphorylation sites required for exchanging adaptor proteins. Srk family kinase-mediated phosphorylation of the NPxY motifs inhibits the binding of talin while promoting the association of inhibitor proteins such as DOK-1. Interaction zones between β-CT and adaptor proteins are denoted by associated horizontal lines. Functional roles of the proteins are indicated in parenthesis. For a detailed description, see refs. 41,54 . The membrane-permeable peptides mP6 and mP13 were based on the integrin β 3 cytoplasmic tail. (B) Model of outside-inside-out signaling for integrin-mediated cell entry. Hypothetical SARS-CoV-2 binding to integrin β 1 initiates Gα 13 binding to the β 1 cytoplasmic tail, which stimulates outside-in signaling in the absence of a known receptor-stimulated GPCR mediated inside-out signaling. mP6 is a specific inhibitor of Gα 13  www.nature.com/scientificreports/ Because mP6 and mP13 are membrane-permeable peptides, they were suitable for infectivity experiments while obviating the need to expose cells to DMSO for extended periods. We, therefore, tested the efficacy of mP6 and mP13 at inhibiting cell entry and productive infection in Vero E6 cells with a 0.01 multiplicity of infection (MOI) of SARS-CoV-2. For the productive infection assay, infected cells were plated at confluency (500,000 cells/well in a 12 well plate) to minimize cell growth for 48 h post-infection. We used RT-qPCR to measure viral nucleocapsid RNA in the suspended cells or intact cell monolayers at 48 h post-infection, respectively. At 48 h post-infection, inhibition of productive infection by mP13 was significant relative to mock-treated cells, whereas the effect of mP6 was insignificant (Fig. 5F). The failure of mP6 to inhibit productive infection is consistent with the notion that viral replication 52,65 perturbs Ca 2+ homeostasis within the infected cells 66 and thus dispenses with Gα 13 activity in favor of talin-induced outside in-signaling 42

Discussion
This study provides mechanistic evidence for the functionality of extracellular ligand-binding domains of integrin β 1 and cytoplasmic tails of integrins in general 25,28 , which offer possible molecular links between ACE2 and integrins. We show that Mn 2+ , which induces integrin extension and high-affinity ligand binding, enhances the cell entry of SARS-CoV-2 R18 . The increased virus binding and entry is consistent with the notion that integrin affinity and/or extension are essential for cell entry. In support of integrin-dependent endocytosis as a pathway of SARS-CoV-2 R18 internalization, we used broad-spectrum RGD antagonists such as GLPG0187, which inhibited cell entry regardless of integrin activation status. Our study also suggested integrin specificity. BTT 3033, an αI allosteric antagonist that binds to the bent closed conformation of integrin β1 and stabilizes it, supports the possibility of integrin-dependent endocytosis of SARS-CoV-2 R18 upon receptor binding. In a different framework, our data also show that SARS-CoV-2 R18 can bind to low affinity and presumptively bent-conformation integrins 23 , however, in BTT 3033 treated cells, cell entry by SARS-CoV-2 R18 is inhibited because integrin activation post-SARS-CoV-2 R18 engagement is prevented. Thus, our data contextualize integrin extension as the "sine qua non of integrin cell adhesion function," 23 which in turn is an essential condition for integrin-mediated cell entry by SARS-CoV-2. Focal adhesion kinase (FAK) 67 is a well-established component of the adhesome that potentially bridges the signaling gap between integrin signaling turnover 68 and ACE2. FAK is a tyrosine kinase known to direct the recruitment of talin to integrin β 1 -enriched nascent adhesions 60,61 . In ailing heart tissues, ACE2 binds integrin α 5 β 1 in an RGD-independent manner. It is known to regulate FAK mediated cell adhesion and integrin signaling 18 , which terminates with endosomal trafficking 30 (of virion-bearing integrins). The binding of macromolecular RGD ligands to resting integrins elicits ligand-induced integrin activation 64 . We hypothesize that in our in vitro experiments, SARS-CoV-2 binding to inactive integrins triggers a series of spatiotemporally-regulated recruitment of adhesome components, including Gα 13 and talin, to the β-CT. Our data show that talin interaction with integrin β-CTs, which causes integrin extension, is indispensable for productive infection (Fig. 5). Talin binding to the β-CT generates the requisite inside-out signal that increases the affinity of the integrin ectodomain for SARS-CoV-2 binding, which in turn increases viral load. Cell entry of SARS-CoV-2 is clathrin-dependent 69 . Endocytosis of integrins is clathrin-dependent and -independent 70 and involves adaptor proteins such as Dab2 and Numb 71,72 attached to the β-CTs NPxY/NxxY motifs (Fig. 5A). Alternatively, some integrin α-subunits harbor a common endocytosis motif (Yxxϕ) recognized by the clathrin adaptor protein 2 (AP2) 68 .
Mészáros et al. 25 have used bioinformatics to predict the existence of short amino acid sequences (~ 3-10 residues): short linear motifs (SLiMs), such as NPxY/Nxxy, Yxxϕ in the cytoplasmic tails of ACE2 and integrins that mediate endocytosis and autophagy. Some of their theoretical predictions have been validated by experimental studies. First, Kliche et al. 28 confirmed the existence of SLiMs. They extended their findings to establish a potential connection between ACE2 and integrin β 3 cytoplasmic tail interactions with scaffolding and adaptor proteins linked to endocytosis and autophagy. Second, SLiM sequences known to bind and activate the transmembrane glycoprotein neuropilin 1 (NRP1) were identified as potential mediators of SARS-CoV-2 endocytosis 25 . Interestingly, NRP1, which is abundantly expressed in the olfactory epithelium, is now declared as an effector for SARS-CoV-2 infection 34,35 . NRP1 localizes at adhesion sites and promotes fibronectin-bound, activated α 5 β 1 integrin endocytosis, and directs the cargo to the perinuclear cytoplasm [29][30][31][32][33][34] . Studies have shown that the endocytosis of active and inactive integrins to EEA1-containing early endosomes follows distinct mechanisms involving different adaptor proteins. The inactive integrin is promptly recycled back to the plasma membrane via an ARF6and EEA1-positive compartment in a Rab4 -dependent manner 31 . We observed that in BTT 3033-treated cells replete with inactive β 1 integrins, SARS-CoV-2 R18 remained membrane-bound, whereas untreated cells displayed internalization and perinuclear localization of SARS-CoV-2 R18 . This is consistent with the known trafficking of ligand-bearing integrins, including those directed by NRP1, to the perinuclear space 29,30,32 .
Our study has some limitations. Integrin activation is often initiated by other receptors such as G-protein coupled receptors (GPCRs), growth factors, and other integrins 38 . Future studies will explore the effect of receptormediated inside-out signaling, modeled under inflammatory conditions of COVID-19. In addition, the criteria for selecting specific integrins as co-factors of SARS-CoV-2 infectivity are not known and thus worthy of future investigation. Finally, the study is based on the USA-WA1/2020 SARS-CoV-2 strain. Our present study lays the groundwork for examining the activity of the various emergent SARS-CoV-2 variants.
Although several integrins types 12,25-28 are believed to be co-receptors of SARS-CoV-2 infectivity, our study suggests inhibitor specificity for integrin β 1 . This is consistent with known factors: (1) correlated increased expressions of β 1 15 and ACE2 in relevant tissues 16,17 , (2) cytoplasmic tail in cis interactions between ACE2 and integrin β 1 14 , and (3) synergy between ACE2 and integrin β 1 signaling that promotes RGD mediated cell adhesion 18 . To optimize integrin engagement, our cell-binding assays and primary infection assays were carried out in suspension such that ACE2 and integrins were not segregated by cell polarization 73,74 . However, our microscopy studies on adherent cells agreed with the flow cytometry results. Thus, our study represents an initial step toward establishing a mechanistic role for SARS-CoV-2-mediated integrin activation required for cell entry and productive infection.

UV inactivation and fluorescent labeling of the envelope membrane of SARS-CoV-2 with octadecyl rhodamine (R18).
USA-WA1/2020 SARS-CoV-2 strain (from BEI Resources, NIAID, NIH) was cultured in Vero E6 cells in a biosafety level 3 (BSL-3) containment under a protocol approved by the University of New Mexico's Institutional Biosafety Committee or IBC (Public Health Service registration number C20041018-0267). First, live SARS-CoV-2 were harvested at peak titers of 10 7 plaque-forming units/mL (PFU/ml). Next, SARS-CoV-2 was UV inactivated using 254 nm (≈ 5 mW/cm 2 ) U.V. irradiation of a TS-254R Spectroline UV Transilluminator (Spectronics Corp., Westbury, NY) following a similar protocol for inactivating pathogenic orthohantaviruses 45,75 . Briefly, Vero E6 cells were inoculated with SARS-CoV-2 and maintained at 37 °C for 2-4 days. At 70-75% cell death (due to viral cytopathic effect), the supernatant was harvested and subjected to light centrifugation (1000 rpm, 10 min) to remove cellular debris. For UV inactivation, supernatants were added to a 12 well plate at 500 µl aliquot/well. Then UV -irradiated at 3.8 cm above the sample for 0, 10,15, 20, 25, 30, 60, and 90 s and then tested for viability by a 3-day plaque assay as described elsewhere 76,77 . The titration of UV irradiation times was used to establish a minimal UV dose for complete inactivation. After UV treatment, the 500 µl fractions were pooled into 15 mL tubes stored in a − 80 °C freezer pending the results of a plaque assay. Under our experimental conditions, we established that a minimum UV irradiation interval of 25 s was required for the complete inactivation of SARS-CoV-2. A 90 s UV dose was approved by the IBC for removal of inactivated SARS-CoV-2 out of the BSL-3 lab after it was established that the virus particles were capable of specific binding to Vero E6 cells.
Crude UV-inactivated SARS-CoV-2 samples were purified by floating 10 ml of SARS-CoV-2 supernatant on a density gradient comprising 2 ml volumes of 1.2 g/ml and 1.0 g/ml CsCl in PBS media in 14 × 89-mm Beckman polyallomer tubes. The samples were centrifuged for 1.5 h at 4 °C using a Beckman SW41Ti rotor at 30,000 pm. A band was collected at the interface and purified by centrifugation in HHB using 100 kDa cutoff Microcon ® Centrifugal Filter. The purified SARS-CoV-2 samples were stored in 1.0 ml aliquots at − 80 °C. SARS-CoV-2 particles were fluorescently labeled and calibrated according to the same protocol used for the Sin Nombre virus (SNV) 45 . The final volume for each labeled batch preparation was limited to 500 µl. The number of SARS-CoV-2 R18 particles in each sample preparation was estimated from absorption measurements using the following equation: # of moles of R18 (derived from sample absorbance) × Avogadro's number (6.02 × 10 23 molecules mole −1 )/estimated average number of R18 molecules per virion (10,000) 45 . The yield of SARS-CoV-2 R18 particles was typically in the 10 8 /µl range. Batch samples were stored in 20 µl aliquots at − 80 °C.
Flow cytometry binding assays of SARS-CoV-2 R18 to vero E6 cells. For flow cytometry assays, cells were cultured in T25 or T75 flasks to 80% confluence. Cells were then treated with 0.25% trypsin and transferred to minimum essential medium (MEM) media. Cell counts and viability were performed using a Life Technologies Countess II FL Automated cell counter (Thermofisher Scientific). Test suspension cell samples were transferred to microfuge tubes in 40 µl-aliquots (1000 cells/µl). SARS-CoV-2 R18 was added to tubes at 5000 SARS-CoV-2 R18 /cell and incubated using a shaker at 500 rpm for 20 min at 37 °C. For blocking assays, cells were incubated with 5 × unlabeled SARS-CoV-2 or 10 µM integrin inhibitors for 20 min before the addition of SARS-CoV-2 R18 . Samples were centrifuged at 3,000 rpm; the pellet was resuspended in HHB buffer (30 mM HEPES, 110 mM NaCl, 10 mM KCl, 1 mM MgCl 2 ⋅6H 2 O, and 10 mM glucose, pH 7.4) buffer and read on an Accuri flow cytometer. For kinetic assays, Vero E6 suspension cells in 40 µl volumes (1000 cells/µl) were placed in ± Mn 2+ media in duplicate microfuge tubes at 37 °C. Sars-CoV-2 R18 was then added (5000 virions/cell) to the tubes and incubated for 1, 3, 5, 7, 9 min. At each time point, the tubes were quenched in an ice bath, then samples were centrifuged and resuspended in 95 µl HHB buffer and analyzed on a flow cytometer.
Live cell confocal microscopy. Imaging was performed using a Leica TCS SP8 Laser Scanning Confocal Microscope with a 63 × water objective and a Bioptechs objective heater to maintain cells at physiological temperature (~ 36-37 °C). Vero E6 cells were plated in eight-well Lab-Tek (Nunc) chambers at a density of 30,000 cells per well 24 h before imaging. Cells were imaged in Tyrode's buffer (135 mM NaCl, 10 mM KCl, 0.4 mM MgCl 2 , 1 mM CaCl 2 20 mM glucose, 0.1% BSA, 10 mM HEPES, pH 7.2). For integrin inhibition, cells were treated with 10 μM BTT3033-α 2 β 1 or 50 μM MP6 in Tyrode's buffer for 30 min before imaging. ~ 1 × 10 9 SARS-CoV-2 R18 particles were added per well, and z-stacks (300 nm thickness) were acquired every 3 min for 21 min to visualize viral cell entry. R18 was excited using 561 nm light, isolated from the white light source. R18 emission and differential interference contrast (DIC) transmitted light were captured with Leica Hybrid detectors (HyD) in a spectral window of 571-636 nm (for R18 emission). Analysis of the accumulation of SARS-CoV-2 R18 particles in Vero E6 cells was completed using Matlab. Briefly, regions of interest (ROI) were created around the cell membrane, and the mean SARS-CoV-2 R18 intensity was measured within the cell mask at each time point.
Immunofluorescence. Vero E6 cells were plated on 18-mm coverslips overnight in a 6 well plate at a density of 100,000 cells/well. Cells were exposed to ~ 1 × 10 9 SARS-CoV-2 R18 particles/well for 15 min at 37 °C, in the presence or absence of 10 μM BTT 3033. Cells were then washed in phosphate-buffered saline (PBS) and fixed using 4% paraformaldehyde (PFA) in PBS for 15 min at room temperature. Cells were extensively washed with 10 mM Tris (pH 7.4) and PBS and permeabilized with 0.1% Triton. Cells were labeled with anti-EEA1 primary antibody and anti-rabbit Alexa Fluor 647 secondary. Nuclei were stained with Hoechst 33258. Cells were www.nature.com/scientificreports/ mounted on microscope slides using Prolong Diamond Antifade Mountant (Invitrogen, CAT#P33970). Samples were imaged using a Leica TCS SP8 Laser Scanning Confocal Microscope with a 63× oil objective.
Infection inhibition. Vero E6 cells grown at 80% confluency were trypsinized and divided into microfuge tubes aliquots of 1.5 × 10 6 cells in 750 µl media containing 250 µM mP6, 250 µM mP13, 10 µM BTT 3033, DMSO, and media only. Samples were shaken at 500 rpm for 30 min at 37 °C. After transfer to a BSL-3 laboratory, 0.01 MOI of SARS CoV-2 (lot #P3: 1.2 × 10 7 pfu) and then incubated for 60 min while shaking. Tubes were spun down (1000 rpm for 3 min), resuspended in fresh media, and spun down again. The cells were then resuspended in 300 µl of media and transferred to a 12 well plate in 100 µl aliquots (500,000 virions/well) for triplicate measurements. An additional 400 µl were added to each well for a final volume of 500 µl. The plate was transferred to an incubator for 48 h to allow the virus to replicate. The cells were then washed with 1xPBS before RNA was extracted with TRIzol™ (Thermofisher, #15596026) according to the manufacturer's protocol: (https:// assets. therm ofish er. com/ TFS-Assets/ LSG/ manua ls/ trizol_ reage nt. pdf).
The RNA was quantified with a Nanodrop and total cellular cDNA transcribed with the Applied Biosystems™ High-Capacity cDNA Reverse Transcription Kit (Fisher Scientific #43-688-14).
Statistical analysis. SARS-CoV-2 R18 binding was expressed as the mean channel fluorescence (MCF) output of the flow cytometer. For different batch preparations of SARS-CoV-2 R18 , we first tested functional binding of SARS-CoV-2 R18 to cells by comparing the MCF readings of 1 mM Mn 2+ -activated cells and resting cells after correcting for autofluorescence. Then, assuming sampling from a Gaussian distribution, the two groups were compared using estimation plots 47 with unpaired two-tailed t-tests to check for consistency between batch preparations of SARS-CoV-2 R18 . For integrin activation inhibitor tests involving three or more groups, comparisons were performed using Ordinary one-way ANOVA with Tukey's multiple comparison test. Data analysis was done with GraphPad Prism software version 9.2.0. Statistical significance was defined as p < 0.05.