SARS-CoV-2 binds platelet ACE2 to enhance thrombosis in COVID-19

For this study, we recruited COVID-19 patients admitted to the First Affiliated Hospital of Zhengzhou University, Henan Province, China and three centers of the Affiliated Hospital of Anhui Medical University, Anhui Province, China between January 10th and March 18th 2020. The detailed study design, including the inclusion and exclusion criteria as well as laboratory data collection, is described in Additional file 1: Expanded Materials and Methods. This study was approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (2020-KY-121) and the Ethics Committee of Anhui Medical University (2020-AH-114) and complied with the Declaration of Helsinki and good clinical practice guidelines. All participants provided written informed consent.

Detailed descriptions of the reagents, commercial ELISA kits, and viral RNA detection in the blood are provided in Additional file 1: Expanded Materials and Methods. The primary antibodies and dilutions used in this study are listed in Additional file 1: Online Table 1.

Platelets and peripheral blood mononuclear cells (PBMCs) were prepared [32] as detailed in Additional file 1: Expanded Materials and Methods. All experiments using human subjects were performed in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Zhengzhou University.

Washed platelet aggregation and secretion in response to thrombin (0.025 U/mL) or collagen (0.6 μg/mL) as well as washed platelet aggregation in response to ADP (5 μM with 10 μg/mL fibrinogen) were measured as previously described [33]. Spreading and clot retraction were measured as previously described [34, 35] and detailed in Additional file 1: Expanded Materials and Methods. To explore the effects of SARS-CoV-2 and its Spike protein on platelet function, SARS-CoV-2 (1×10⁵ PFU [36]) or Spike protein (2 μg/mL) was added to platelets for the indicated times before agonist-induced stimulation [37‐39].

The human colon cell line Caco-2, the human lung cell line Calu-3, the human prostate cell line PC-3, and the human cervical carcinoma cell line HeLa were all cultured in the appropriate medium detailed in Additional file 1: Expanded Materials and Methods. All cells were collected and processed for RNA and protein extraction.

SARS-CoV-2 virus was isolated from a COVID-19 patient in Shanghai (GenBank accession No. MT121215) and propagated in Vero E6 cells [23]. The supernatant from the SARS-CoV-2 infected cells was aliquoted and stored at − 80 °C until use. The supernatant from mock-infected cells was used as a control. For the platelet functional studies, around 300 μL of platelets (2 × 10⁸/mL) were incubated with 1 × 10⁵ PFU SARS-CoV-2 in Tyrode’s buffer. The infectious titers have been used in other cells [36]. All experiments involving live SARS-CoV-2 virus were performed in a biosafety level-3 (BLS-3) laboratory.

Total RNA was isolated from different cells and 1 μg of RNA was reverse transcribed to cDNA using an RNA isolation kit and polymerase chain reaction (RT-PCR) kit (TaKaRa, Japan), respectively [35]. PCR reactions were performed using specific primers (Additional file 1: Online Table 2).

Flow cytometry was used to evaluate platelet activity in COVID-19 patients, non-COVID-19 patients and healthy volunteers, and to evaluate platelet activity and ACE2 and TMPRSS2 expression in SARS-CoV-2 virus or Spike protein treated healthy platelets and leukocyte-platelet aggregates. All of these assays were completed using the previously described methods [23, 40, 41], and detailed in Additional file 1: Expanded Materials and Methods. The primary antibodies and dilutions used in these assays are listed in Additional file 1: Online Table 1.

Washed platelets subjected to different treatments were lysed with equal volumes of chilled 2× NP-40 lysis buffer (100 mM Tris-HCl pH 7.4, 300 mM NaCl, 2 mM NaF, 2% NP-40, 2 mM EDTA, and 2× protease and phosphatase inhibitor solution) on ice for 30 min. The supernatants were then precleared using protein A/G agarose beads for 3 h at 4 °C and then centrifuged. Immunoprecipitation was carried out using an anti-phospho-Ser/Thr antibody and then incubated with protein A/G-agarose beads overnight on a rocker at 4 °C. The beads were then harvested and rinsed 3 times with 1× NP-40 lysis buffer. Bead-captured ACE2 was detected by immunoblot and the primary antibodies and dilutions used in these assays are listed in Additional file 1: Online Table 1.

Platelets were attached to poly-l-Lysine-coated coverslips and fixed with precooled methanol and then blocked with BSA in PBS. After incubation with the primary antibodies and appropriate secondary antibodies, confocal images were captured using a laser-scanning confocal microscope as previously described [44] and detailed in Additional file 1: Expanded Materials and Methods. The primary antibodies and dilutions used in these assays are listed in Additional file 1: Online Table 1.

PF4, TNF-α, IL-8, IL-1β Factor V, and Factor XIII concentrations were determined using commercial ELISA kits according to the manufacturer’s instructions. More information is provided in Additional file 1: Expanded Materials and Methods.

Wild-type C57BL/6 mice and hACE2 transgenic mice (Cat No. T037657) were purchased from Jiangsu Gempharmatech, China. All animal procedures were carried out in accordance with the ethical approval granted by the Ethics Committee of Zhengzhou University. Mouse platelets were prepared as described previously [35]. Adoptive platelet transfusions were performed based on previous reports [45]. More information is provided in Additional file 1: Expanded Materials and Methods.

Intravital microscopy of FeCl₃-injured thrombus formation in mouse mesenteric arterioles was performed as previously described with minor modifications [34, 46, 47]. SARS-CoV-2 Spike protein (200 μg/kg) was injected intravenously into wild-type mice according to a previously reported method with minor modifications [30]. More information is provided in Additional file 1: Expanded Materials and Methods.

The flow chamber assay was prepared as described previously with minor modification [44, 48], and detailed in Additional file 1: Expanded Materials and Methods.

Given the inherent differences between patients, a series of propensity score analyses were performed for the following variables: age, sex, history of smoking, hypertension, diabetes mellitus, hypercholesterolemia, stroke, and COPD. When matched with the healthy group, variables included age, sex, and history of smoking; when matched with other groups, variables included age, sex, history of smoking, hypertension, diabetes mellitus, hypercholesterolemia, stroke, and COPD. The propensity scores were estimated using a logit model. Matching was conducted using 1:1 or 1:2 nearest neighbor methods with a caliper width of 0.25*SDs in the logit propensity score, which yielded 28 severe and critically severe COVID-19 patients matched with 56 healthy subjects, 37 severe and critically severe COVID-19 patients matched with 37 non-COVID-19 patients, 29 severe and critically severe COVID-19 patients matched with 58 mild and moderate COVID-19 patients, 64 mild and moderate COVID-19 patients matched with 64 healthy subjects, and 59 non-COVID-19 subjects matched with 115 mild and moderate subjects. All matching analyses were performed using R software (version 3.6.0).

Continuous variables are expressed as the mean ± standard deviation of the mean (SD) or median (interquartile range was defined as the difference between twenty-fifth and seventy-fifth centiles) depending on data distribution, and compared using an unpaired Student’s t test or Mann-Whitney U test, as appropriate. Categorical variables were expressed as numbers and percentages and compared using the Chi-square test or Fisher’s exact test. All data were evaluated for normality (Kolmogorov-Smirnov) and subjected to the Bartlett’s test for homogeneity of group variances prior to statistical analysis. Group comparisons were made using one-way ANOVA, followed by Tukey’s post hoc analysis or Kruskal-Wallis test with Bonferroni correction. Pearson’s correlation analysis was used to investigate the relationships between two variables. A P value of less than 0.05 was considered statistically significant. Statistical analyses were performed using SPSS (version 21.0) and GraphPad Prism (7.0).

Our study population comprised 201 healthy volunteers and 589 patients suspected of having COVID-19. Of these 589 patients, 422 were identified as SARS-CoV-2 infected, while the remaining 167 patients were identified as non-SARS-CoV-2 infected. We excluded 35 healthy volunteers, 107 non-COVID-19 patients, and 181 COVID-19 patients leaving a total of 166 healthy volunteers, 60 non-COVID-19 patients, and 241 COVID-19 patients including 184 mild and moderate patients, and 57 severe and critically severe patients in the study (see Flowchart in Additional file 1: Online Figure 1). The characteristics of the healthy group, non-COVID-19 patients, mild and moderate COVID-19 patients, and severe and critically severe COVID-19 patients are provided in Additional file 1: Online Table 3.

Severe and critically severe COVID-19 patients presented with abnormal platelet parameters, including decreased platelet counts and plateletcrit (PCT), increased mean platelet volume (MPV), and platelet distribution width (PDW) as well as abnormal coagulation parameters including increased prothrombin time (PT), international normalized ratio (INR), activated partial thromboplastin time (APTT), d-dimer and fibrinogen degradation products (FDPs), and decreased prothrombin time activity (PTA) when compared with healthy donors, non-COVID-19 patients, and mild and moderate COVID-19 patients (Additional file 1: Online Table 3). Similar results were observed after propensity score matching of severe and critically severe COVID-19 patients to the other groups (Additional file 1: Online Table 4). In addition, mild and moderate COVID-19 patients presented with decreased platelet counts and increased MPV, compared with healthy donors and non-COVID-19 patients. Similar results were observed after propensity score matching (Additional file 1: Online Table 3 and 5). MPV was shown to correlate with platelet activity and is considered a marker of platelet activity [49, 50]; therefore, increased MPV in COVID-19 suggests that these platelets may present as hyperactive.

The relationship between the platelet count and coagulation parameters in COVID-19 patients is described in Fig. 1a–f. For those with a normal platelet count (> 125 × 10⁹/L), the platelet count did not significantly impact the outcomes of the PT, PTA, INR, APTT, d-dimer, and FDPs tests. However, in patients with thrombocytopenia (< 125 × 10⁹/L), when the platelet count decreases, PT, INR, APTT, d-dimer, and FDPs increase exponentially, while PTA decreases exponentially. We further examined platelet count over time in severe and critically severe COVID-19 patients (n = 22). Our results suggest that platelet count decreases gradually after hospital admission (Fig. 1g).

Consistent with the increased MPV, we found that integrin αIIbβ3 activation (PAC-1 binding) and P-selectin (CD62P) expression were increased in platelets of COVID-19 patients (Fig. 1h, Additional file 1: Online Figure 2). The severe and critically severe COVID-19 patients presented the highest integrin αIIbβ3 activation and P-selectin expression on platelets. Platelet activation leads to platelet consumption which causes thrombocytopenia [51]; notably, PAC-1 binding and CD62P expression were both moderately correlated with decreases in platelet count (Pearson │r│ for PAC-1 = 0.50, P < 0.01; Pearson │r│ for CD62 = 0.64, P < 0.01; Fig. 1i) in COVID-19 patients.

Of the 241 COVID-19 patients, a total of 15 (6.22%) patients, including 12 severe and critically severe and 3 mild and moderate cases of COVID-19, were positive for blood SARS-CoV-2 RNA. In severe and critically severe patients, patients with detectable blood viral RNA presented with higher integrin αIIbβ3 activation and P-selectin expression on platelets, compared with those with undetectable blood viral RNA (Fig. 1j). To test the effect of SARS-COV-2 RNA positive blood on platelet aggregation, we incubated healthy platelets with SARS-COV-2 RNA-positive platelet-poor plasma (PPP) and SARS-COV-2 RNA-negative PPP from severe and critically severe patients, and used healthy PPP as control. We found that SARS-COV-2 RNA-positive PPP enhanced platelet aggregation, compared with SARS-COV-2 RNA-negative PPP and healthy PPP (Additional file 1: Online Figure 3). These results suggest that the presence of SARS-CoV-2 viral RNA in blood is an indicator of platelet hyperactivity.

Since SARS-CoV-2 infects host cells via ACE2, we explored whether platelets express ACE2. We found that human platelets exhibit robust expression of ACE2 at both the RNA and protein levels as detected by RT-PCR (Fig. 2(A1)) and Western blot (Fig. 2(B1)). These levels were similar to the human colon cell line Caco-2 and the human lung cell line Calu-3, which are used as SARS-CoV-2-infected host cells [52, 53]. The HeLa cell line was used as a negative control for ACE2 expression [54]. The detected ACE2 in platelets is not from the contaminated PBMCs, because the PBMCs-specific marker CD14 was not detectable in our platelet samples when evaluated by RT-PCR (Fig. 2(A2)) or Western blot (Fig. 2(B2)). ACE2 level in platelets is comparable with that in PBMCs (Fig. 2(B2)). Mouse platelets exhibited abundant expression of ACE2, similar to lung and heart tissues at the RNA (Fig. 2(C)) and protein levels (Fig. 2(D)).

A recent study reported that cellular serine protease TMPRSS2 (transmembrane protease serine 2) primes SARS-CoV-2 for cell entry [53]. Interestingly, we detected robust expression of TMPRSS2 in platelets comparable with that in human Caco-2 cells, human Calu-3 cells, and mouse lung tissues at the RNA (Fig. 2(E)) and protein levels (Fig. 2(F)). Furthermore, abundant ACE2 and TMPRSS2 protein expression was confirmed in platelet progenitor megakaryocyte cells (Meg-01 cell line, Fig. 2(G)). Finally, ACE2 and TMPRSS2 protein expression was further confirmed in human and mouse platelets using flow cytometry (Fig. 2(H)) and in human platelets using confocal immunofluorescence (Fig. 2(I)).

After providing evidence that platelets express ACE2 and TMPRSS2, we went on to evaluate whether SARS-CoV-2 virus could directly activate platelets. We found that incubation with SARS-CoV-2 (1 × 10⁵ PFU [36]) for 30 min did not induce platelet aggregation in human washed platelets from healthy donors (Additional file 1: Online Fig. 4), but in the range of 0.1 to 1 × 10⁵ PFU, SARS-CoV-2 dose-dependently potentiated platelet aggregation in response to collagen (0.6 μg/mL), thrombin (0.025 U/mL), and ADP (5 μmol/L). Consistently, SARS-CoV-2 increased platelet dense granule secretion (ATP release) in response to collagen and thrombin in a dose-dependent manner (Fig. 3a). SARS-CoV-2 induced integrin αIIbβ3 activation and P-selectin expression under basal conditions and enhanced both integrin αIIbβ3 activation and P-selectin expression following agonist activation (Fig. 3b).

We then went on to assess platelet spreading on immobilized fibrinogen and clot retraction and found that pre-incubation with SARS-CoV-2 markedly enhanced platelet spreading (Fig. 3c) and clot retraction (Fig. 3d). In addition, we observed SARS-CoV-2 particles attached to platelet membrane using SEM (Fig. 3f), indicating that SARS-CoV-2 can bind directly to platelets. Moreover, fluorescent confocal microscopy and TEM revealed that SARS-CoV-2 particles were present inside the platelets (Fig. 3e, g), suggesting that SARS-CoV-2 can infect these cells.

ACE2 degradation plays an important role in the pathogenesis of COVID-19. SARS-CoV2 has been reported to promote ACE2 internalization and subsequent degradation [55]. Similarly, we found that SARS-CoV-2 induced a time-dependent decrease in ACE2 levels in platelets (Additional file 1: Online Figure 5a), indicating the degradation of ACE2 in platelets upon ACE2 activation. Consistently, we found that ACE2 levels were decreased in platelets from COVID-19 patients, compared with healthy donors. Severe and critically severe COVID-19 patients presented the lowest level of ACE2 expression (Additional file 1: Online Figure 5b).

Since SARS-CoV-2 binds to host cells via interactions between the Spike protein and ACE2, we sought to explore whether Spike protein could regulate platelet function. Incubation with Spike protein for 9 min did not induce platelet aggregation in human washed platelets from healthy donors (Additional file 1: Online Figure 4). However, similar to the results from the SARS-CoV-2 virus experiments, we were able to demonstrate that the Spike protein dose-dependently enhanced platelet aggregation and ATP release (Additional file 1: Online Figure 6). We further found that SARS-CoV-2 Spike protein (2 μg/mL, 5 min), Spike protein subunit 1 (S1, 2 μg/mL, 5 min), but not Spike protein subunit 2 (S2, 2 μg/mL, 5 min), potentiated platelet aggregation and dense granule secretion in response to different agonists (Fig. 4a). Using flow cytometry, we found that Spike protein induced integrin αIIbβ3 activation and P-selectin expression in the absence of agonist. Furthermore, Spike protein and S1, but not S2, enhanced both integrin αIIbβ3 activation and P-selectin expression in the presence of agonist (Fig. 4b). These data indicate that S1, but not S2, binds ACE2 to regulate platelet function, which corroborates the finding that the receptor-binding domain (RBD) of the Spike protein is found in the S1 subunit [56]. After incubation with Spike or S1 protein, platelets also displayed markedly accelerated spreading (Fig. 4c) and clot retraction (Fig. 4d).

Mitogen-activated protein kinase (MAPK) has been well documented in platelet activation and thrombosis. ACE2/MAPK pathway activation has been reported to mediate SARS-CoV-2-induced cytokine modulation in lung cells [27, 57]. Thus, we examined whether the ACE2/MAPK pathway is involved in SARS-CoV-2-mediated platelet activation. SARS-CoV-2 and Spike protein stimulated platelet ACE2 phosphorylation at 15 and 3 min, respectively (Fig. 5(A) and Additional file 1: Online Figure 7a). MAPK (Erk, p38, and JNK) kinase was phosphorylated apparently later than ACE2 phosphorylation (Fig. 5(A) and Additional file 1: Online Figure 7b). These results agree with a previous study conducted on lung cells [27] and supports the hypothesis that the MAPK pathway is activated downstream of ACE2 in platelets.

We investigated whether MAPK signaling mediates the potentiating effects of SARS-CoV-2 on platelet activation. As shown in Fig. 5(B and, C), the inhibition of Erk, p38, and JNK with PD98059 (10 μM), SB203580 (10 μM), and SP600125 (10 μM) abolished the potentiating effects of SARS-CoV-2 (Fig. 5(B)) or Spike protein (Fig. 5(C)) on agonist-induced platelet aggregation. In addition, we observed that the accelerated clot retraction elicited by SARS-CoV-2 or Spike protein was also prevented by Erk, p38, or JNK inhibitors (Fig. 5(D)).

We further detected MAPK phosphorylation in platelets from COVID-19 patients. Our results demonstrated that COVID-19 patients presented increased phosphorylation of Erk, p38, and JNK in platelets, compared with healthy donors (Fig. 5).

Together, these data suggest that MAPK is phosphorylated downstream of ACE2 and may mediate the potentiating effects of SARS-CoV-2 on platelet activation.

We then tested whether SARS-COV-2 Spike protein activates platelets and thereby enhances thrombus formation in vivo. To eliminate non-platelet hemostatic or thrombotic factor variations, we injected 10⁹ platelets from transgenic mice bearing human ACE2 (hACE2) or wild-type mice into thrombocytopenic wild-type mice. We simultaneously intravenously injected 200 μg/kg SARS-CoV-2 Spike protein into mice [30], and examined its effects on FeCl₃-induced thrombus formation in the mesenteric arterioles. Representative images of the thrombus development at various time points following FeCl₃ injury are shown in Fig. 6a. Notably, Spike protein had no effect on thrombus formation following mesenteric arteriole injury in mice transfused with platelets from wild-type mice, which agreed with the results that SAR-CoV-2 Spike protein cannot bind mouse ACE2 protein [53]. However, Spike protein did potentiate thrombus formation in mice transfused with platelets from hACE2 transgenic mice.

We then assessed the role of the Spike protein on mice platelets for thrombus formation under arterial flow conditions using a microfluidic whole-blood perfusion assay. Throughout the perfusion period, thrombus formation was significantly increased when pre-treated with Spike protein (2 μg/mL, 5 min), in whole blood from hACE2 transgenic mice, but not in whole blood from wild-type mice (Fig. 6b). In addition, the Spike protein potentiated platelet aggregation and ATP release in response to agonists in vitro and enhanced thrombosis formation in vivo on hACE2 transgenic mice, while it had no effect on wild-type mice (Fig. 6c and Additional file 1: Online Figure 8).

Platelets are an important source of coagulation factors V and XIII in α granules, and serve as adhesion sites for coagulation factors activation via their surface exposure of phosphatidylserine (PS) [58]. We demonstrated that SARS-CoV-2 stimulated platelets to release both Factor V (Additional file 1: Online Figure 9A) and Factor XIII (Additional file 1: Online Figure 9B), at levels that were comparable to thrombin stimulation. However, SARS-CoV-2 had no effect on platelet PS exposure (Additional file 1: Online Figure 9C). Similar results were also observed in the Spike protein-treated platelets (Additional file 1: Online Figure 9D, 9E, and 9F).

Platelets carry a large variety of inflammatory cytokines in α granules, which can be quickly secreted upon activation to participate in the immune response [59]. We found that SARS-CoV-2 and its Spike protein stimulated PF4 (Additional file 1: Online Figure 10A), TNF-α (Additional file 1: Online Figure 10B), IL-8 (Additional file 1: Online Figure 10C), and IL-1β (Additional file 1: Online Figure 10D) secretion from platelets. Consistently, we found that plasma PF4 levels were increased in COVID-19 patients, compared with healthy donors. The severe and critically severe COVID-19 patients presented the highest plasma PF4 levels (Additional file 1: Online Figure 10E).

P-selectin is a key adhesion molecule that mediates the interaction between platelets and leukocytes via P-selectin glycoprotein ligand 1 (PSGL-1). Consistent with our finding that SARS-CoV-2 increased P-selectin expression, we found that SARS-CoV-2 and Spike protein increased the proportion of leukocyte–platelet aggregates (LPAs) (CD45⁺CD41⁺ aggregates, Additional file 1: Online Figure 11). Specifically, both monocyte–platelet aggregates (MPAs, CD14⁺CD41⁺) and neutrophil-platelet aggregates (NPAs, CD65⁺CD41⁺) were significantly increased after SARS-CoV-2 or Spike protein treatment.

Recombinant human ACE2 protein has been reported to inhibit SARS-CoV-2 infection in engineered human tissues [60]. SARS-CoV-2 neutralizing antibodies produced by the host immune response are a critical part of the toolbox of therapies for COVID-19 [61]. We pretreated platelets from healthy donors with ACE2 protein or an anti-Spike antibody. Both the ACE2 protein and the anti-Spike antibody suppressed SARS-CoV-2 (Fig. 7(A)) or Spike protein (Fig. 7(B))-potentiated platelet aggregation and reversed SARS-CoV-2 or Spike protein-induced PAC-1 binding and CD62P expression (Fig. 7(C and D)). The accelerated platelet spreading (Fig. 7(E)) and clot retraction (Fig. 7(F)) elicited by exposure to SARS-CoV-2 or Spike protein were also prevented by pretreatment with the ACE2 protein and anti-Spike antibody. Of note, treatment with the ACE2 protein and anti-Spike antibody also suppressed the thrombus formation induced by the Spike protein following mesenteric arteriole injury (Fig. 7(G)). These data clearly show that SARS-CoV-2 stimulates platelet activation via the interaction of its Spike protein and ACE2, which can be suppressed by exogenous addition of ACE2 or an anti-Spike antibody.

Our study has several limitations. Firstly, animal experiments of virus invasion are lacked due to the limitations of our experimental conditions. Whether our in vivo thrombus formation study using Spike protein can be also reproduced using living SARS-CoV-2 virus still needs to be further investigated. Secondly, platelets in infected lungs may meet different concentrations of viruses in COVID-19; the in vivo effect of SARS-CoV-2 on platelet activation needs to be further investigated in COVID-19 patients. Thirdly, whether our study in washed platelets can be also reproduced using platelet-rich plasma still needs to be further investigated. Fourthly, the relationship between ACE2 degradation and disease severity, platelet activity, or blood SARS-COV2 RNA existence still needs further investigation.