ACE2-Independent Interaction of SARS-CoV-2 Spike Protein with Human Epithelial Cells Is Inhibited by Unfractionated Heparin

Abstract

Coronaviruses such as SARS-CoV-2, which is responsible for COVID-19, depend on virus spike protein binding to host cell receptors to cause infection. The SARS-CoV-2 spike protein binds primarily to ACE2 on target cells and is then processed by membrane proteases, including TMPRSS2, leading to viral internalisation or fusion with the plasma membrane. It has been suggested, however, that receptors other than ACE2 may be involved in virus binding. We have investigated the interactions of recombinant versions of the spike protein with human epithelial cell lines that express low/very low levels of ACE2 and TMPRSS2 in a proxy assay for interaction with host cells. A tagged form of the spike protein containing the S1 and S2 regions bound in a temperature-dependent manner to all cell lines, whereas the S1 region alone and the receptor-binding domain (RBD) interacted only weakly. Spike protein associated with cells independently of ACE2 and TMPRSS2, while RBD required the presence of high levels of ACE2 for interaction. As the spike protein has previously been shown to bind heparin, a soluble glycosaminoglycan, we tested the effects of various heparins on ACE2-independent spike protein interaction with cells. Unfractionated heparin inhibited spike protein interaction with an IC₅₀ value of <0.05 U/mL, whereas two low-molecular-weight heparins were less effective. A mutant form of the spike protein, lacking the arginine-rich putative furin cleavage site, interacted only weakly with cells and had a lower affinity for unfractionated and low-molecular-weight heparin than the wild-type spike protein. This suggests that the furin cleavage site might also be a heparin-binding site and potentially important for interactions with host cells. The glycosaminoglycans heparan sulphate and dermatan sulphate, but not chondroitin sulphate, also inhibited the binding of spike protein, indicating that it might bind to one or both of these glycosaminoglycans on the surface of target cells.

Keywords: SARS-CoV-2; glycosaminoglycan; heparin; spike protein.

Figure 1

ACE2 and TMPRSS2 mRNA expression in human epithelial cell lines. Total RNA was isolated from the cell lines, converted to cDNA and ACE2 and TMPRSS2 mRNA levels determined by RT-qPCR. The data are shown relative to expression in wild-type (wt) HEK293 cells (=1) and are the means from at least 2 independent experiments.

Figure 2

S1S2 interaction with human cells is temperature dependent. Attachment of S1, RBD and wtS1S2 (black, red and blue lines, respectively) compared to secondary-only control (solid grey), at 4 °C (Top panels) or 37 °C (Lower panels). RT4 cells were incubated with 330 nM S1-Fc, 10 µM RBD or 330 nM S1S2-His6 protein for 60 min at either 4 or 37 °C, before staining with anti-mouse Ig labelled with FITC or anti-His6 secondary antibody labelled with Dylight 488 for 30 min at 21 °C. Cell-associated fluorescence was measured by flow cytometry. Results are representative of at least 3 separate experiments.

Figure 3

ACE2 is not required for wtS1S2 attachment to cells. (A) A549, wtHEK293, ACE2HEK293, Caco2, HaCaT, HCE2 and RT4 cells were incubated with 100 nM wtS1S2 for 60 min, before staining with anti-His6 secondary antibody labelled with Dylight 488. Binding of S1S2 attachment to various human cells. Data are calculated relative to the median fluorescence of secondary antibody alone (=1), means ± SEM from 5 separate experiments performed in duplicate. (B) Attachment of different concentrations of wtS1S2 (squares) or RBD (triangles) to wtHEK cells (open symbols) or ACE2HEK cells (filled symbols). Data are shown as a percentage of 10 µM RBD binding to ACE2HEK cells, means ± SEM from 3 independent experiments performed in duplicate.

Figure 4

RBD can bind to VERO E6 cells that express high levels of endogenous ACE2 but not to RT4 cells. VERO E6/TMPRSS2, A549 and RT4 cells were harvested by either trypsin/EDTA or non-enzymatic cell dissociation solution (CDS) and then incubated with 1000 nM RBD for 60 min, before staining with anti-His6 secondary antibody labelled with Dylight 488. The data are shown as a percentage of the binding of 100 nM wtS1S2 and are the means ± SEM from 5–6 independent experiments performed in duplicate. The significance of the difference from 0 was assessed by a one-sample t test; * p < 0.05; ** p < 0.001.

Figure 5

wtS1S2 from SARS-CoV-1 binds more weakly to cells than SARS-CoV-2 wtS1S2. A549, RT4 and VERO E6/TMPRSS2 cells were harvested by either trypsin/EDTA (TE; filled symbols) or non-enzymatic cell dissociation solution (CDS; open symbols) and then incubated with the stated concentrations of wtS1S2 from SARS-CoV-1 (squares) or SARS-CoV-2 (circles) for 60 min, before staining with anti-His6 secondary antibody labelled with Dylight 488. The data are shown relative to 100 nM SARS-CoV-2 wtS1S2 and are the means ± SEM from 3–4 independent experiments performed in duplicate.

Figure 6

S1S2 interaction may require the furin cleavage site. (A) Dose–response curve for wtS1S2, mS1S2 or RBD to RT4 cells. RT4 cells were incubated with wtS1S2, mS1S2 or RBD at the stated concentrations for 60 min at 37 °C, before staining with anti-His6 secondary antibody labelled with Dylight 488. The data are the means ± SEM of 3 - 6 independent experiments performed in duplicate. (B) Representative histogram of 100 nM wild-type S1S2 (wtS1S2) or mutant S1S2) attachment to RT4 cells.

Figure 7

Concentration-dependent inhibition of S1S2 interaction by unfractionated heparin and low-molecular-weight heparins, dalteparin and enoxaparin but not synthetic pentasaccharide, fondaparinux. RT4 cells were pre-incubated with the stated concentrations of unfractionated heparin, enoxaparin, dalteparin and fondaparinux for 30 min at 37 °C, then with 100 nM wtS1S2 for a further 60 min at 37 °C before fluorescent secondary anti-His6 was added for a further 30 min at 21 °C. Cell-associated fluorescence was measured by flow cytometry and is shown as a percentage of the wtS1S2 attachment to untreated control cells. Data are the means ± SEM of 2–3 experiments performed in duplicate.

Figure 8

Glycosaminoglycan-binding molecules surfen and protamine sulphate can inhibit wtS1S2 binding. RT4 cells were preincubated with the stated concentrations of surfen or protamine sulphate for 30 min at 37 °C, then with 100 nM wtS1S2 for a further 60 min at 37 °C. Bound wtS1S2 was detected using fluorescent secondary anti-His6, measured by flow cytometry. Data are shown as a percentage of the wtS1S2 attachment to untreated control cells. Data are the means ± SEM of 3–4 experiments performed in duplicate.

Figure 9

Spike protein binding to heparin requires a polybasic furin cleavage site. Unfractionated heparin (UFH) and low-molecular-weight heparin, dalteparin (Dalt) were immobilised on 96-well plates and used to detect the binding of wtS1S2, mS1S2 lacking the furin cleavage site (mS1S2) or RBD using a biotinylated anti-His6 antibody and streptavidin-HRP. Data shown are the means ± SD from 2–3 separate experiments performed in duplicate. Tables provide the EC₅₀ values for each protein.

Figure 10

Removal of heparan sulphates only partly inhibits wtS1S2 binding. (A) RT4 cells were pretreated for 3 h with heparinase I/III mixture or control growth medium and then harvested by either trypsin/EDTA (+) or non-enzymatic cell dissociation solution (−) and then incubated with 100 nM wtS1S2 at 37 °C before staining with anti-His6 secondary antibody labelled with Dylight 488. The data are shown relative to the non-trypsinised- and non-heparinase-treated control and are the means ± SEM from 3–4 independent experiments performed in duplicate. Significance of difference from the control was assessed by a one-sample t test. ns = not significant; ** p < 0.001. (B) The effects of heparinase treatment was measured using antibody 3G10 that recognises the cleaved stubs of heparan sulphates. RT4 cells were pre-treated for 3 h with heparinase I/III mixture or control growth medium and then harvested by either trypsin/EDTA or non-enzymatic cell dissociation solution and then incubated with 3G10. Bound antibody was detected by an FITC-labelled secondary and cell-associated fluorescence was measured by flow cytometry. Data are the means ± SD from 2 separate experiments performed in duplicate.

Figure 11

Inhibition of wtS1S2 binding to RT4 cells by heparins and glycosaminoglycans. RT4 cells were incubated with 100 µg/mL of heparin or glycosaminoglycan for 30 min at 37 °C, then with 33 nM wtS1S2 for a further 60 min at 37 °C and bound protein detected using fluorescent secondary anti-His6. Cell-associated fluorescence was measured by flow cytometry and data are shown as a percentage of wtS1S2 attachment to untreated control cells. Data are the means ± SEM of 4–6 separate experiments performed in duplicate. Significance of difference from the control was assessed by a one-sample t test: not significant unless otherwise stated; * p < 0.05; ** p < 0.001; **** p < 0.00001.

Figure 12

Dose–response curves for the inhibition of wtS1S2 and RBD binding to immobilised unfractionated heparin by selected heparins and glycosaminoglycans. Unfractionated heparin (UFH) was immobilised on 96-well plates and used to detect the binding of wtS1S2, mS1S2 and RBD in the presence of increasing concentrations of heparins of increasing chain length (dp10–16, heparan sulphate (HS) or dermatan sulphate (DS)), detected using a biotinylated anti-His6 antibody and streptavidin-HRP. Data are shown as a percentage of binding to untreated control wells and are the means ± SEM of 4 separate experiments performed in duplicate.