Platelet deformation induced by SARS-CoV-2 spike protein


A new study by researchers from the National Heart, Lung, and Blood Institute, U.S. National Institutes of Health, United States using cellular cryo-electron tomography has clearly shown that the spike proteins of the SARS-CoV-2 virus causes deformation of platelets and also causes irreversible stochastic activation leading to coagulopathies.

Detailed cellular cryo-electron tomography revealed dense decorations of S protein on the platelet surface, inducing filopodia formation. Hypothesizing that S protein binds to filopodia-inducing integrin receptors, the study team tested the binding to RGD motif-recognizing platelet integrins and found that the S protein recognizes integrin αvβ3.
The study findings were published in the peer reviewed journal: Nature Communications.

SARS-CoV-2 has shown unique pathological symptoms that can lead to a wide range of coagulopathic events in severe cases. In our study, we probed the direct effect of S protein to the change in morphology of platelets at a molecular level, and we directly visualized the binding of S protein to the platelet surface (summarized in Fig. 6).

We hypothesized that the binding of the SARS-CoV-2 is mediated by integrin receptors based on the following reasons;

(1) the activation of platelets is governed by filopodia formation,

(2) filopodia formation is initiated by integrin receptors,

(3) the major receptors on the platelets are integrin receptors and

(4) SARS-CoV-2 S protein contains a RGD sequence in the RBD, which is recognized by a subtype of integrin, and therefore we tested the interaction of platelet-expressed integrins with S protein.

Our integrin inhibition experiment using cilengitide and in vitro solid-phase binding assays support this hypothesis, particularly with the possibility that S protein recognizes integrin αvβ3. The binding of S protein to integrin was much lower compared to the interaction of integrins with their physiological ligands, and interestingly, we did not detect the binding to the major platelet integrin αIIbβ3.

Previously, an increased binding of the activated integrin αIIbβ3 antibody PAC-1 to platelets was observed in the presence of S protein21. This may be due to an inside-out effect, in which the outside-in signaling is activated by the direct binding of S protein to integrin αvβ3 and in turn, αIIbβ3 would get activated through the intracellular signaling (inside-out).

We surmise that the weak affinity of S protein to platelet integrin receptors and the reversible binding, may reflect the fact that blood clotting defects observed in patients are rare complications and occur in severe cases of COVID-19.

However, here we should also note that there are other receptors on platelets that may also be accountable for the interaction with S protein33,53 and combinatory effects of the binding of S protein to multiple receptors may also occur.

First, S protein binds to receptors on the platelet surface, causing the deformation and priming the activation. Protrusions are forming as a consequence of actin remodeling. This leads to the activation of platelets by the formation of filopodia and the stabilization of the cytoskeleton network. The scheme was created with

SARS-CoV-2 is found in the blood stream of COVID-19 patients 10, and an open question is how it can lead to rare but severe coagulation defects. We showed that the deformation of platelets itself does not always alter their intracellular signaling (Supplementary Fig. 1), or induces activation.

It rather appears that platelets exposed to S protein are primed for the activation upon further stimuli, such as the attachment to an adhesion surface. Based on this observation, we speculate that the combination of the direct binding of S protein to platelets and other identified coagulation factors may induce a synergistic and irreversible activation of platelets, leading to coagulation.

During SARS-CoV-2 infection, several other procoagulant players are active, for example the formation of neutrophil extracellular traps 18, the release of TF23, elevated fibrinogen levels 54 and dysregulated release of cytokines55, creating a hypercoagulative environment in the context of COVID-19.

In our study, we visualized the adaptable attachment of S protein to the platelet plasma membrane with a high degree of flexibility for the engagement to continuously curved membrane surfaces (Fig. 4D, E). Similarly, it has been reported that the stalk domain of S protein proximal to the viral membrane surface contains three hinges, presumably allowing the flexible motion of individual S protein on the viral surface to adapt to curved host cell surfaces 50.

This dual flexibility likely increases the probability for S protein to attach to a host cell receptor, thus, allowing an efficient action of S protein to the membrane surface.


To further analyze the S protein densities on the platelet membrane surface, we manually selected and extracted the densities from 8 tomograms and analyzed them using subtomogram averaging approaches (Fig. 4A). To facilitate a focused alignment of the decorating protein without the influence of the membrane density, the membrane signal was subtracted using PySeg49. We obtained a 3D-averaged density at a resolution of 13.8 Å (Fig. 4A, Fig. S3) showing a characteristic trimeric shape of 15 nm in size. The obtained structure

agreed well with our near-atomic resolution structure of S protein using the same protein batch for single-particle analysis (Fig. S3) as well as previously published structures25,30,46,50, (Fig. 4A, fitted PDB 6vxx), validating the identity of S protein decorations. Furthermore, the analysis of the particles without application of C3 symmetry showed an asymmetric uplift of the tip of the S1 surface that connects to the extra density (Fig 4A, right). This shows that one of the three RBD domains of S protein is lifted up upon its binding to the host cell receptors. The extra density connected to S protein represents the density from the host cell, likely the platelet surface receptor recognized by S protein (Fig. 4I).

To assess how S protein recognizes the platelet surface, the alignment parameters of the individually analyzed S protein densities were applied to the 3D average and plotted back to the original tomograms (Fig. 4B-H). The distribution analysis of neighboring S protein showed that the peak population had a distance of 27.3 nm (median) apart, but some molecules were also more sparsely distributed (Fig. 4B).

This measurement corresponds to a density of one S protein on a surface area of up to 585 nm2 (a radial surface of 585 nm2 is covered by one S protein), although no apparent periodical distribution was detected. Judging from the diameter of S protein (~17 nm), neighboring S protein closely located next to each other. S protein bound to the platelet membrane at a distance of 16 nm between the center of S protein to the membrane surface (Fig. 4F) and interestingly, with a wide range of angular distribution (Fig. 4D-E) with respect to the membrane surface, indicating its flexible attachment to the platelet surface.

In addition, S protein binds to a slightly more curved membrane surface (Fig. 4G and 4H). This may be reflected by the fact that the binding of S protein induces filopodia formation with a membrane protrusion. Taken together, these results indicate that S protein approaches the platelet surface from various geometrical orientations to accommodate and enhance the docking to the membrane surface.

Similarly, a broad angular distribution of S protein has been observed from the viral surface, due to several kinked points in the stalk region46,50. Together with our observation, it suggests the orientational adjustments from both sides of S protein, namely the receptor binding S1 subunit and the stem side at the root on the virus, maximize the efficiency of the attachment of S protein to the host cell receptors.

Consistent with the observed additional density on the lifted RBD domain (Fig. 4A, right), some of the tomograms showed extra densities bridging between plasma membrane and S protein (Fig. 4I, red arrows), presumably those of platelet

receptors recognizing S protein. However, subtomogram analysis only yielded a faint density (Fig. 4A, right) without features, also suggesting a flexible attachment of S protein to its receptor on the membrane surface.

Platelet deformation in the presence of pseudotyped viral particles

After characterizing the effects of S protein on platelets, we hypothesized that locally concentrated S protein on a globular viral surface would be advantageous for increasing the local concentration and evaluated the influence of SARS-CoV-2 pseudo virus-like-particles (VLPs) on platelet deformation. We either generated or obtained SARS-CoV-2 pseudotyped VLPs that are fully intact vesicle-like entities as validated by negative staining EM (Fig. S4A) and by their ability to infect HEK-293T-hACE2 cells (HEK-293T cells constitutively expressing ACE2 receptor, Fig. S4B-S4C). The viral titer determined by flow cytometry was approximately 104-106 particles/ml, comparable to the reported preparation of SARS-CoV-2 pseudo VLPs51, however, it was low compared to VSV-G based lentiviruses.

This low titer did not allow us to readily detect changes in platelet morphologies by live platelet imaging. However, we were able to find an example of a particle located in close proximity to a platelet filopodium (Fig. S5B-S5F). Cryo-electron tomography revealed that the closest distance between this particle and the membrane surface of the filopodium was 20 nm (Fig. S5G and S5H), similar to that measured for S protein alone (Fig. 4F 16 nm, from the center of S protein to membrane).

The cross-section views of the particle showed decorations of proteins on the membrane surface (Fig. S5I, red arrowheads), altogether suggesting that this vesicle may be a bound VLP. In comparison, we also found examples of extracellular vesicles with a similar shape and size (Fig. S5A) that appeared to originate from intracellular vesicles, i.e. exosomes, containing alpha granules (Fig. S5A, left) or budding out from the concave surface of plasma membrane, instead of filopodia and indicative of vesicle release through fusion of lysosome and plasma membrane. These results corroborate our in vitro data of purified S protein inducing morphological changes in platelets.

Integrin receptors recognize SARS-CoV-2 S protein

Several cell receptors were reported to recognize S protein. However, the presence of ACE2, the major S protein receptor, on the platelet surface is still inconclusive16,19,21.Therefore, the

relevance of the ACE2 receptor for platelet malfunction is still an open question. In contrast, integrin receptors are the major class of receptors expressed in platelets. Considering our structural analysis (Fig. 4) and the possibility that ACE2 is not abundantly expressed on platelets, we hypothesized that S protein may directly recognize integrin receptors. Interestingly, the RBD domain of S protein contains a stretch with an “RGD” motif, which is a common motif among integrin ligands35 and a direct interaction of tissue integrin α5β1 and SARS-CoV-2 S protein has been shown 39.

We therefore tested the binding of S protein to known platelet integrin receptors αIIbβ3, αvβ3, and α5β1, enriched in the tissue but also expressed on platelets, all recognizing the RGD ligand motif. We used ELISA-like solid-phase equilibrium binding assays to detect the interaction of S protein with integrins (Fig. 5A). We detected the binding of integrin α5β1 and αvβ3 to S protein, while integrin αIIbβ3 does not have an apparent interaction with it (Fig. 5B).

The extent of binding is most prominent with integrin αvβ3, while integrin α5β1 showed only a weak interaction. However, it should be noted that the observed binding of tested integrins was much weaker (less than 10-fold) compared to those for physiological integrin ligands: vitronectin for αvβ3, fibrinogen for αIIbβ3 and fibronectin for α5β1.

Encouraged by our results, we tested the effect of platelet activation in the presence of cilengitide, a cyclic RGD pentapeptide52 that blocks the binding of integrin to RGD motif-containing extracellular ligands, and indeed the activation was reduced (Fig. 5C). These observations generally agree with a recent discussion of the relevance of integrin recognition by SARS-CoV-2 for vascular dysregulation38.

reference link:


Please enter your comment!
Please enter your name here

Questo sito usa Akismet per ridurre lo spam. Scopri come i tuoi dati vengono elaborati.