The Spike Glycoprotein of SARS-CoV-2 Binds to β1 Integrins Expressed on the Surface of Lung Epithelial Cells

The spike glycoprotein attached to the envelope of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) binds to and exploits angiotensin-converting enzyme 2 (ACE2) as an entry receptor to infect pulmonary epithelial cells. A subset of integrins that recognize the arginyl–glycyl–aspartic acid (RGD) sequence in the cognate ligands has been predicted in silico to bind the spike glycoprotein and, thereby, to be exploited for viral infection. Here, we show experimental evidence that the β1 integrins predominantly expressed on human pulmonary epithelial cell lines and primary mouse alveolar epithelial cells bind to this spike protein. The cellular β1 integrins support adhesive interactions with the spike protein independently of ACE2, suggesting the possibility that the β1 integrins may function as an alternative receptor for SARS-CoV-2, which could be targeted for the prevention of viral infections.


Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes a potentially fatal acute pulmonary infection, leading to a cytokine storm that culminates in acute respiratory distress syndrome and other COVID-19 pathologies [1,2]. As of April 2021, the SARS-CoV-2 pandemic is reported to globally comprise more than 133 million infected cases, including more than 2.8 million deaths [3]. SARS-CoV-2 is a single-strand positive-sense RNA virus that on the surface displays a large spike glycoprotein (S-protein), which plays a critical role in the interaction between the virus and host cells at entry sites. Angiotensin-converting enzyme 2 (ACE2), an ecto-enzyme present on the cell surface, has been identified as a major cellular receptor for the viral spike glycoprotein, one that aids viral entry to host alveolar epithelial cells and other epithelial and endothelial cells [4]. ACE2 is regarded as a promising therapeutic target for blocking such viral entry [4,5]. However, multiple cell-surface receptors other than ACE2 may be simultaneously involved in the process of viral entry [6][7][8][9].
Integrins represent the largest family of cell-adhesion receptors. Composed of the α-and β-subunits that bind to ligands on the opposing cells and in the matrix proteins, these integrins mediate adhesive interactions across a wide range of biological processes including organ development, wound healing and angiogenesis, host defense and inflammation, and hemostasis [10]. In both humans and mice, 18 α-subunits and eight β-subunits have now been identified; both participate in the formation of 24 non-covalently associated α/β heterodimer cell-surface receptors [10]. Several integrins, including the fibronectin receptor α5β1 integrin and the vitronectin receptor αVβ3 integrin, specifically recognize a surface-exposed arginyl-glycyl-aspartic acid (RGD) sequence in the cognate ligands in a Mg 2+ ion-dependent manner [11]. Recent investigations have revealed that the S-protein of SARS-CoV-2, but not other coronaviruses such as SARS-CoV and MERS-CoV, contains a surface-exposed RGD sequence [12][13][14]. This component can thereby predict the binding of S-protein to the cellular integrins, which would facilitate the viral entry of SARS-CoV-2. Targeting the integrin-binding RGD sequence in the S-protein could be used for blocking SARS-CoV-2 infections [6,15]. Here, we set forth experimental evidence showing that integrins in human pulmonary epithelial cell lines and mouse primary alveolar epithelial cells bind to the spike protein.

Mice
C57BL/6J mice (10 to 13 weeks old) were obtained from CLEA Japan (Tokyo, Japan). The mice were maintained in the Mie University Experimental Animal Facility under specific pathogen-free conditions with ad libitum access to water and food and with a standard 12-hour light-dark cycle. The experimental animal protocol was approved by the Ethics Review Committee of Mie University (Approval number: #2019-41).

Isolation of Primary Lung Epithelial Cells from Mouse Lungs
Epithelial cells were isolated from mouse lung tissues by using a mechanical dissociation with sterile sieve meshes followed by a discontinuous density-gradient centrifugation using Percoll (GE Healthcare Life Sciences, Chicago, IL, USA) in accordance with previous studies [17,18] with some modifications. Three Percoll solutions (25%, 40%, and 75%) were prepared, the cells from lung tissues were resuspended in 40% solution, and three distinct gradients were carefully placed in the order of 25%-40%-75% from top to bottom in a 15 mL centrifuge tube (Corning, Glendale, AZ, USA). The cells were spun at 780× g for 20 min in a centrifuge (AX-511) (Tomy, Tokyo, Japan) at the setting of minimal acceleration and deceleration. The epithelial cells were centered at an interface between 25% and 40% Percoll solution and were then taken and washed extensively with RPMI1640 (Nacalai) containing 10% FBS (Equitech-Bio) and penicillin/streptomycin (Nacalai).

Reverse Transcription and Quantitative Polymerase Chain Reaction (RT-qPCR)
Total RNA was extracted from the cells by using TRIzol reagent (Thermo Fisher Scientific) and RT was conducted with a Prime Script RT Kit (Takara Bio, Shiga, Japan) according to the manufacturer's instructions. To determine relative ACE2 expression, qPCR was performed by using a PowerUp SYBR Green Master Mix PCR kit (Applied Biosystems, Foster City, CA, USA) and the StepOne Real-Time PCR System (Applied Biosystems) according to the manufacturer's instructions. For endogenous controls, β-actin was used to normalize ACE2 expression. The PCR primers (5 →3 ) used in this study were as follows: human β-actin, CCCTGGACTTCGAGCAAGAG (forward) and ACTC-CATGCCCAGGAAGGAA (reverse); human ACE2, AAACATACTGTGACCCCGCAT (forward) and CCAAGCCTCAGCATATTGAACA (reverse); mouse β-actin, GATCAAGAT-CATTGCTCCTCCTGA (forward) and AAGGGTGTAAAACGCAGCTCA (reverse); and mouse ACE2, TGGGCAAACTCTATGCTG (forward) and TTCATTGGCTCCGTTTCTTA (reverse). Relative expressions were calculated by using the comparative threshold method (2 −∆CT ) normalized to β-actin.

Statistical Analysis
Data are expressed as the mean ± standard error of the mean (SEM) and were analyzed using a two-tailed unpaired t-test for comparisons between two groups and a one-way analysis of variance (ANOVA) for comparisons among three groups. p values less than 0.05 were considered significant. Statistical analysis was done by using the Prism8 software (GraphPad, San Diego, CA, USA).

Results
As pulmonary epithelial cells represent a prime target for SARS-CoV-2, we have investigated the binding of the spike protein to the human pulmonary epithelial cell lines 11-18 and QG-56. We have found that these epithelial cell lines showed good binding to the spike protein ( Figure 1A,B). β1 integrins are the predominant integrins expressed in pulmonary epithelial cells under physiologic conditions, although some levels of αVβ3, α6β4, αVβ5, αVβ6, and αVβ8 integrins are also expressed ( Figure 2) and upregulated under pathologic conditions such as in various cancers and fibrosis [20]. Thus, we have tested the effects of anti-β1 and anti-αV integrin inhibitory antibodies that block all β1 integrins (α1β1, α2β1, α3β1, α4β1, α5β1, α6β1, α7β1, α8β1, α9β1, α10β1, and α11β1) and αV integrins (αVβ1, αVβ3, αVβ5, αVβ6, and αVβ8), respectively. The adhesive interactions of pulmonary epithelial cells with the spike 1 (S1) protein are potently inhibited by pretreatment with anti-β1 or anti-αV integrin inhibitory antibodies ( Figure 1A,B).
As monocytes/macrophages are implicated in the pathogenesis of COVID-19, we have studied human monocyte-like cell line THP-1, thereby showing a good binding to S1 protein ( Figure 1C). THP-1 binding to S1 protein was blocked by not only anti-β1 and anti-αV integrin inhibitory antibodies, but the anti-α4 integrin inhibitory antibody that blocks α4β1, and α4β7 integrins ( Figure 1C). THP-1 cells express α4 integrin, whereas 11-18 and QG-56 cells do not ( Figure 2). Thus, the anti-α4 integrin inhibitory antibody did not show any blocking effects on 11-18 and QG-56 cells ( Figure 1A,B).
To further substantiate our findings regarding the β1 integrin-dependent binding of spike proteins to pulmonary epithelial cell lines, we studied primary pulmonary cells freshly isolated from mouse lungs that were enriched with alveolar epithelial cells. We have shown that mouse primary lung cells robustly bind to S1 proteins, which are inhibited by the antibody to mouse β1 integrins ( Figure 3A).
As we used S1-Fc fusion protein throughout the experiments, we sought to substantiate the specific binding to the S1 protein, thereby comparatively studying the cell adhesion to S1-Fc fusion protein and two types of reference proteins S2-Fc fusion protein and IgG Fc fragment protein ( Figure 4A-D). We have shown that QG-56, 11-18, and THP-1 cells hardly bound to IgG Fc fragment protein, while binding well to S1-Fc protein ( Figure 4B-D). As S2-Fc protein is a scarce reagent for us, we have studied it only with QG-56 cells, which bound significantly better to S1-Fc protein than to S2-Fc protein ( Figure 4A). These results confirmed the specific binding of QG-56, 11-18, and THP-1 cells to S1 protein.
To further validate the specific interaction of β1 integrin with S1 protein, we utilized a β1 integrin knockout (KO) clone of a human breast cancer cell line (MDA-MB-231) that we previously established using CRISPR/Cas9 [16] as a model cell. A flow cytometric examination confirmed the absence of β1 integrin in the KO cells ( Figure 5A). S1 binding of β1-KO cells was completely abolished, compared to that of β1-control cells (i.e., mock transfectants) [16] (Figure 5B), even though β1-KO cells retain αV integrin expression ( Figure 5A). These results have revealed that β1 integrins can support binding to S1 protein independently of αV integrins.  To further substantiate our findings regarding the β1 integrin-dependent binding of spike proteins to pulmonary epithelial cell lines, we studied primary pulmonary cells freshly isolated from mouse lungs that were enriched with alveolar epithelial cells. We have shown that mouse primary lung cells robustly bind to S1 proteins, which are inhibited by the antibody to mouse β1 integrins ( Figure 3A).   To further substantiate our findings regarding the β1 integrin-dependen spike proteins to pulmonary epithelial cell lines, we studied primary pulm freshly isolated from mouse lungs that were enriched with alveolar epitheli have shown that mouse primary lung cells robustly bind to S1 proteins, whic ited by the antibody to mouse β1 integrins ( Figure 3A).  IgG Fc fragment protein ( Figure 4A-D). We have shown that QG-56, 11-18, and TH cells hardly bound to IgG Fc fragment protein, while binding well to S1-Fc protein ( Fig  4B-D). As S2-Fc protein is a scarce reagent for us, we have studied it only with QG cells, which bound significantly better to S1-Fc protein than to S2-Fc protein (Figure 4 These results confirmed the specific binding of QG-56, 11-18, and THP-1 cells to S1 p tein. Figure 4. Binding of human lung epithelial cells and monocytes to S1-Fc. Cellular binding to S1was compared to S2-Fc (A) and to Fc alone (B-D). QG-56 (A,B), 11-18 (C), THP-1 (D) were used this assay. Results are shown as column scatterplots overlayed with the mean ± SEM of at least independent experiments, in which each sample was triplicated. *** p < 0.001; **** p < 0.0001.
To further validate the specific interaction of β1 integrin with S1 protein, we utiliz a β1 integrin knockout (KO) clone of a human breast cancer cell line (MDA-MB-231) t we previously established using CRISPR/Cas9 [16] as a model cell. A flow cytometric amination confirmed the absence of β1 integrin in the KO cells ( Figure 5A). S1 binding β1-KO cells was completely abolished, compared to that of β1-control cells (i.e., mo transfectants) [16] (Figure 5B), even though β1-KO cells retain αV integrin expression (F ure 5A). These results have revealed that β1 integrins can support binding to S1 prot independently of αV integrins. Human pulmonary epithelial cell lines and mouse lung primary cells express moderate to high levels of ACE2, the authentic receptor for the S1 protein ( Figure 6A,B), which may be involved in the observed adhesive interactions of lung cells with the S1 protein. Residual binding to S1 protein in the presence of anti-integrin antibodies could be partly mediated by ACE2. As mouse ACE2 supports S1 protein binding much less efficiently than human ACE2 [21], this could be the case with human pulmonary epithelial cells. To confirm the ability of integrins (e.g., β1 integrin) to bind to the S1 protein independently of ACE2, we utilized ACE2-negative cells, such as human monocytic THP-1 cells and mouse osteoclastic MLO-A5 cells. We have shown that these ACE2-negative cells exhibited good binding to the S1 protein in a β1 integrin-dependent manner (Figures 1C and 3B). Human pulmonary epithelial cell lines and mouse lung primary cells e erate to high levels of ACE2, the authentic receptor for the S1 protein ( Figure 6 may be involved in the observed adhesive interactions of lung cells with th Residual binding to S1 protein in the presence of anti-integrin antibodies cou mediated by ACE2. As mouse ACE2 supports S1 protein binding much le than human ACE2 [21], this could be the case with human pulmonary epithe confirm the ability of integrins (e.g., β1 integrin) to bind to the S1 protein in of ACE2, we utilized ACE2-negative cells, such as human monocytic THP mouse osteoclastic MLO-A5 cells. We have shown that these ACE2-negative ited good binding to the S1 protein in a β1 integrin-dependent manner ( Fig  3B). (B) Cellular binding to S1-Fc was compared between β1-control (Scr.) and β1-KO. Results are shown as column scatterplots overlayed with the mean ± SEM of at least 3 independent experiments, in which each sample was triplicated. **** p < 0.0001. Human pulmonary epithelial cell lines and mouse lung primary cells express moderate to high levels of ACE2, the authentic receptor for the S1 protein ( Figure 6A,B), which may be involved in the observed adhesive interactions of lung cells with the S1 protein.
Residual binding to S1 protein in the presence of anti-integrin antibodies could be partly mediated by ACE2. As mouse ACE2 supports S1 protein binding much less efficiently than human ACE2 [21], this could be the case with human pulmonary epithelial cells. To confirm the ability of integrins (e.g., β1 integrin) to bind to the S1 protein independently of ACE2, we utilized ACE2-negative cells, such as human monocytic THP-1 cells and mouse osteoclastic MLO-A5 cells. We have shown that these ACE2-negative cells exhibited good binding to the S1 protein in a β1 integrin-dependent manner (Figures 1C and  3B). Relative expression of ACE2 to β-actin was determined by using the 2 -ΔCT method. In panel A, statistical significance between two groups (THP-1 vs. [11][12][13][14][15][16][17][18] was analyzed by a two-tailed unpaired t-test, while other comparisons were by one-way ANOVA. Results are shown as column scatterplots overlayed with the mean ± SEM of at least 3 independent experiments, in which each sample was triplicated. ** p < 0.01; and **** p < 0.0001. Figure 6. qRT-PCR measuring the expression of ACE2 in human (A) and mouse (B) cells. Relative expression of ACE2 to β-actin was determined by using the 2 −∆CT method. In panel A, statistical significance between two groups (THP-1 vs. [11][12][13][14][15][16][17][18] was analyzed by a two-tailed unpaired t-test, while other comparisons were by one-way ANOVA. Results are shown as column scatterplots overlayed with the mean ± SEM of at least 3 independent experiments, in which each sample was triplicated. ** p < 0.01; and **** p < 0.0001.

Discussion
In this study, we have demonstrated that native β1 integrins expressed on human and mouse pulmonary epithelial cells bind to the S-protein of SARS-CoV-2, thereby providing experimental evidence to support recent predictions regarding their role in this global disease [12,13]. Our results are consistent with a recent report that showed, in a cell-free ELISA-type assay, that α5β1 integrin protein immobilized on plates supported binding to the SARS-CoV-2 S-protein [15]. Our results demonstrate the critical roles played by β1 integrins in mediating cellular adhesive interactions with the S-protein, although we have shown that αV integrins on pulmonary epithelial cell lines and α4 integrins on monocytic cell lines also support binding to S1 protein. Integrins are conformationally mobile proteins that undergo dynamic structural rearrangements on the cell surface in response to intracellular signaling pathways specific to cell types [22,23]. Thus, demonstrating the ability of the S1 protein to bind integrin expressed on the surface of relevant cells is of great importance, and all of this contributes to emphasizing the significance of the results from the cell-based experiments in our study.
It has been shown in the recent report that a small molecule peptide antagonist of integrin α5β1 (ATN-161, Ac-PHSCN-NH2) inhibited the infection of SARS-CoV-2 in VeroE6 cells in vitro [15]. Due to technical difficulties, we were unable to study the ability of antiintegrin blocking antibodies to interfere with SARS-CoV-2 infection. Whereas further investigations would be necessary to confirm the ability of integrin antagonists to suppress SARS-CoV-2 infection in vitro and in vivo, integrin-targeted approaches may serve as potentially effective anti-viral treatments. A few therapeutic antibodies to integrins have been approved for the treatment of inflammatory disorders such as multiple sclerosis [24] and inflammatory bowel diseases [25]. Although proven to be clinically effective, integrin antagonism in patients can nonetheless pose serious adverse effects. The integrin inhibitor natalizumab, which blocks α4β1 and α4β7 integrins, is reported to sporadically cause progressive multifocal leukoencephalopathy, an often fatal brain disease resulting from reactivation of a latent JC virus infection due to iatrogenic immune suppression [24]. Thus, inhibition of integrins, especially β1 integrins, in clinical settings must be carried out with special precautions. Nevertheless, a multiple sclerosis patient treated with natalizumab has been reported to recover well from SARS-CoV-2 infection [26], which might be relevant to our finding that α4 integrin blockade abolished binding of S1 protein to monocytes.
Integrins have been shown to bind not only to the viral S-protein, as shown in this study, but also to the cellular ACE2, as previously reported [27]. ACE2 contains a conserved RGD sequence that is not surface exposed; however, β1 integrins bind to ACE2 in an RGD-independent manner. A recent and provocative hypothesis proposes not only that integrin binding to ACE2 sterically shields the SARS-CoV-2 binding site within ACE2, but that the cis-interaction of integrins with ACE2 on the same cells naturally prevents the virus from binding to the entry receptor ACE2. In this context, inhibition of integrins may result in the liberation of ACE2 for SARS-CoV-2 binding, potentially enhancing viral entry [14]. Although extremely intriguing, this alternative scenario appears to be unlikely [6]. Simultaneous inhibition of integrin-S-protein interactions and the ACE2-S-protein interactions offers a promising therapeutic approach for robustly suppressing SARS-CoV-2 infections [6]. We have shown that β1 integrin supports the binding of S1 protein to mouse primary lung cells.

Conclusions
To conclude, we have provided important experimental evidence that SARS-CoV-2 spike protein 1 binds to β1 integrins on the surface of pulmonary epithelial cells. β1 integrins bind to the S1 protein independently of ACE2; however, the therapeutic potential of integrin antagonists inhibiting SARS-CoV-2 infections with or without an intervention targeting ACE2 in vivo warrants further investigation.