Abstract
Blood clots are a central feature of coronavirus disease-2019 (COVID-19) and can culminate in pulmonary embolism, stroke, and sudden death. However, it is not known how abnormal blood clots form in COVID-19 or why they occur even in asymptomatic and convalescent patients. Here we report that the Spike protein from severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) binds to the blood coagulation factor fibrinogen and induces structurally abnormal blood clots with heightened proinflammatory activity. SARS-CoV-2 Spike virions enhanced fibrin-mediated microglia activation and induced fibrinogen-dependent lung pathology. COVID-19 patients had fibrin autoantibodies that persisted long after acute infection. Monoclonal antibody 5B8, targeting the cryptic inflammatory fibrin epitope, inhibited thromboinflammation. Our results reveal a procoagulant role for the SARS-CoV-2 Spike and propose fibrin-targeting interventions as a treatment for thromboinflammation in COVID-19.
One-Sentence Summary SARS-CoV-2 spike induces structurally abnormal blood clots and thromboinflammation neutralized by a fibrin-targeting antibody.
Persistent life-threatening thrombotic events are a hallmark of COVID-19. Aberrant clots form in multiple organs causing significant morbidity and mortality in COVID-19 patients (1, 2). The high incidence of clotting complications has been attributed to disease severity, inflammation and subsequent hypercoagulable state (3). However, the clinical picture is puzzling because of disproportionate rates of thrombotic events and abnormal clot properties not observed in other inflammatory conditions, such as severe sepsis or different viral respiratory illnesses (4-7). Intriguingly, abnormal clotting is not limited to acutely-ill COVID-19 patients. Pulmonary emboli, stroke and sudden death also occur in young COVID-19 patients with asymptomatic infections or mild respiratory symptoms (8). Persistent clotting pathology is prevalent in post-acute sequelae of SARS-CoV-2 infection (PASC, Long COVID) (2, 7, 9).
The central structural component of blood clots, and a key regulator of inflammation in disease, is insoluble fibrin, which is derived from the blood coagulation factor fibrinogen and is deposited in tissues at sites of vascular damage (10, 11). Hypercoagulability in COVID-19 is associated with inflammation and the formation of fibrin clots resistant to degradation despite adequate anticoagulation (3-5). Extensive fibrin deposits are detected locally in inflamed lung and brain tissues from COVID-19 patients, sometimes without evidence of direct viral infection at autopsy (1, 8, 12-14). The high prevalence of thrombotic events with these unique hypercoagulability features suggests an as yet unknown mechanism of abnormal blood clot formation in COVID-19. We set out to determine how blood clots form in COVID-19 and to identify therapies to combat the deleterious effects of abnormal coagulation occurring in acute and convalescent stages of disease.
Since hypercoagulability in COVID-19 patients has features distinct from those of other inflammatory diseases, we hypothesized that SARS-CoV-2 directly affects the structural and functional properties of blood clots. Incubation of SARS-CoV-2 recombinant trimeric spike protein (Spike) with healthy donor plasma increased fibrin polymerization (Fig. 1A). Spike strikingly altered the fibrin clot structure resulting in thinner fibers with a rough appearance and increased clot density as shown by scanning electron microscopy (SEM) (Fig. 1B, fig. S1), identifying direct effects of SARS-CoV-2 Spike on fibrin clot architecture. Consistent with these structural changes, a solid-phase binding assay revealed binding of Spike to both fibrinogen and fibrin (Kd 5.3 µM and 0.4 µM, respectively) (Fig. 1C). Fibrinogen immunoprecipitated with full-length recombinant trimeric Spike, and studies with deletion mutants identified an interaction with the S2 domain of Spike (Fig. 1D, fig. S2). Fibrinogen is a 340 kDa protein consisting of three pairs of polypeptide chains Aα, Bβ, and γ (10). To identify Spike binding regions in fibrinogen, we generated a custom fibrinogen peptide array consisting of 390 15-mer peptides overlapping by 11 amino acids and spanning the fibrinogen Aα, Bβ, and γ chains (Fig. 1E). Hybridization with His-tagged trimeric Spike identified three binding sites in the Bβ and γ fibrinogen chains, namely Bβ119-129, γ163-181 and γ364-395 fibrinogen peptides (Fig. 1E). The Bβ119-129 peptide contains cleavage sites for the fibrinolytic serine protease plasmin (15). Spike bound to the γ364-395 peptide, which encompasses the γ377-395 cryptic fibrinogen binding site to complement receptor 3 that activates innate immune responses (11, 16). Spike also bound to the γ163-181 peptide, whose function is unknown. Mapping of the Spike binding peptides onto the crystal structure of fibrinogen revealed proximity of the γ163-181 and γ377-395 peptides, suggesting that a 3D conformational epitope in the carboxy-terminal γ-chain of fibrinogen (γC domain) is involved in fibrinogen binding to Spike.
Since Spike binds to fibrinogen sites that regulate plasmin cleavage and binding to complement receptor 3, we tested whether the binding interferes with the degradation and inflammatory properties of fibrin. Incubation of Spike with fibrin delayed plasmin degradation of both the β-chain and the γ-γ dimer (Fig. 1F), suggesting that Spike delays fibrinolysis. This finding is consistent with dense fibrin clots composed of thin fibers we identified and the presence of fibrinolysis-resistant blood clots in COVID-19 patients (5). Dense fibrin clots with thin fibers resistant to lysis are also observed in thromboembolic diseases (17).
Fibrin is deposited locally at sites of vascular damage and is a potent proinflammatory activator and a key inducer of oxidative stress (11, 18). Strikingly, Spike increased fibrin-induced release of reactive oxygen species (ROS) in a concentration-dependent manner in bone marrow-derived macrophages (BMDMs), while Spike alone did not have an effect (Fig. 1G). These results suggest a role for Spike as an enhancer of fibrin-induced inflammation at sites of vascular damage. Overall, these results reveal an unanticipated role for SARS-CoV-2 Spike as a fibrinogen binding protein that alone accelerates the formation of abnormal clots with altered structure and increased inflammatory activity.
In COVID-19 patients, fibrin is deposited in the air spaces and lung parenchyma and is associated with inflammation (8). We developed an experimental platform to study the interplay between fibrin and SARS-CoV-2 Spike in vivo by injecting mice with HIV virions pseudotyped with SARS-CoV-2 trimeric Spike glycoprotein (Spike PVs) (fig. S3), enabling the study of the in vivo effects of Spike independent of active viral replication. Intravenous administration of Spike PVs in wild-type (WT) mice induced extensive fibrin deposition in the lung (Fig. 2A). Double immunofluorescence staining for fibrin and Spike PVs revealed strong overlap of Spike and fibrin deposits (Fig. 2B, movie S1, S2). Fibrin deposition was associated with activated endothelium in the lung, and gene expression analysis revealed increased expression of endothelial and inflammatory markers in Spike PV-injected mice compared to mice injected with control BALD PVs (Fig. 2C, fig. S4, tables S1, S2), consistent with findings of SARS-CoV-2 toxicity to endothelial cells (19). Fibrin activates macrophages and induces oxidative stress through nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (11, 18), which is linked to severe disease and thrombotic events in COVID-19 patients (20). In WT mice, Spike-PVs activated macrophages and increased expression of the gp-91-phox subunit of NADPH oxidase in the lung of WT mice indicating the generation of an oxidative stress response (Fig. 2D). In contrast, control BALD PVs or PVs expressing the Env protein from the HIV-1(HIV-1 PVs) did not induce these effects (Fig. 2D), suggesting that lung pathology was specific for SARS-CoV-2 Spike. Mice genetically-deficient in fibrinogen (Fgα–/– mice), which express all other blood proteins except fibrinogen and are protected from autoimmune and inflammatory conditions (11), did not exhibit lung pathology following Spike PV challenge (Fig. 2E, fig. S5). These results reveal a Spike– fibrinogen-dependent mechanism of clot formation that generates strong inflammatory and oxidative stress responses.
Fibrinogen is causally linked to the activation of macrophages and microglia in autoimmune and inflammatory diseases in the brain and periphery (11, 21). Fibrin is a driver of microglia-induced cognitive dysfunction (22) and is associated with perivascular-activated microglia and macrophages in brains of COVID-19 patients even without signs of infection (12). Stereotactic co-injection of Spike PVs and fibrinogen into the brains of WT mice, a model of fibrinogen-induced encephalomyelitis (21), significantly increased fibrin-induced microglia activation (Fig. 3A), suggesting that Spike enhances the inflammatory function of fibrin in vivo. Furthermore, like recombinant Spike (Figs. 1D, G), Spike PVs co-immunoprecipitated with fibrinogen and increased fibrin-induced oxidative stress in BMDMs (figs. S6, S7). Conversion of fibrinogen to fibrin exposes the cryptic inflammatory γ377–395 epitope in the fibrinogen γ-chain. Genetic or pharmacologic targeting of this epitope has potent therapeutic effects in autoimmune and inflammatory diseases (11, 18, 23, 24). Using alanine scanning mutagenesis, we found that Spike interacts with aa 386-394 in the C-terminus of the γ377–395 epitope (Fig. 3B, fig. S8). Genetic targeting of the fibrin γ377–395 epitope in Fggγ390-396A mice, in which fibrinogen retains normal clotting function but lacks the γ390-396-binding motif, rescued from macrophage activation and oxidative stress in the lung after Spike PV administration (Fig. 3C). Since γ377–395 is a binding site for both Spike (this study) and complement receptor (16, 18, 23, 24), inhibition of this epitope may reduce their interactions with fibrin. Future studies will be required to characterize the biophysical properties of fibrin binding to Spike and to complement receptor 3 and their relative contributions to thromboinflammation. These findings reveal a previously unknown interaction between SARS-CoV-2 Spike protein and fibrin γ377–395 epitope that promotes innate immune activation.
A surge of autoantibody production against diverse immune targets have been detected in COVID-19 patients (25). To determine whether COVID-19 patients develop autoantibodies against abnormal blood clots, we tested autoantibody responses to fibrin. Autoantibodies against fibrin epitopes would be potentially missed by the inherent limitations of phage and yeast library screens to produce post-translationally modified insoluble fibrin polymer. To overcome this challenge, we developed a fibrin autoantibody discovery platform optimized for screening patient samples. We tested longitudinally collected serum samples ranging from acute to convalescent disease stages from 54 COVID-19 asymptomatic, mild, and severe disease patients requiring admission to the intensive care units (table S3). Fibrin autoantibodies were abundant in all three groups of COVID-19 patients and persisted during the convalescent stage, but were scarce in healthy donor controls or in subjects with non-COVID respiratory illnesses (Fig. 4A, B).
Blockade of the thromboinflammation cascade following Spike and fibrinogen/fibrin interaction is an attractive therapeutic target. Based on our genetic rescue results implicating a causal role for the γ377–395 epitope, we tested the effects of 5B8, a monoclonal antibody generated against the fibrin γ377–395 epitope (18). This selective antibody-based approach suppresses fibrin-induced inflammation without altering normal hemostasis (18). 5B8 rescued the enhanced inflammatory effects induced by Spike in fibrin-treated BMDMs (Fig. 4C), suggesting that pharmacologic blockade of the fibrin γ377–395 epitope inhibits the deleterious effects of SARS-CoV-2 Spike as an enhancer of thromboinflammation. Strikingly, the 5B8 antibody reduced macrophage activation and oxidative stress in the lungs of Spike PV-treated WT mice compared to isotype IgG2b-treated controls (Fig. 4D). Collectively, these results identify anti-fibrin autoimmune responses in COVID-19 patients and demonstrate a potent protective effect of fibrin-targeting immunotherapy against thromboinflammation.
In summary, we find that SARS-CoV-2 Spike protein enhances the formation of highly inflammatory clots that are neutralized by a fibrin-targeting monoclonal antibody. Our data shed new light on the enigmatic coagulopathy found in COVID-19 revealing a causal role for fibrinogen in thromboinflammation – even independent of active viral replication. The high incidence of clotting complications in COVID-19 has been attributed to systemic inflammation (3), vascular damage including abnormal levels of circulating coagulation proteins (1, 26), genetic susceptibility to tissue factor and complement genes (27), and prothrombotic autoantibodies (28). Our findings now show that coagulopathy is not merely a consequence of inflammation. Rather, the interaction of SARS-CoV-2 Spike with fibrinogen and fibrin results in abnormal blood clot formation that in turn drives inflammation. Identification of SARS-CoV-2 Spike protein as a fibrinogen binding partner provides a mechanistic basis for the formation of abnormal clots with enhanced inflammatory properties. This mechanism might be in play at sites of local fibrin deposition and microvascular injury perpetuating a hypercoagulable and inflammatory state as reported in COVID-19 patients (14) that could be critical during acute infection, as well as in PASC (2, 5, 7). Fibrin is locally deposited in brain and other organs of COVID-19 patients. Thus, fibrin immunotherapy may represent a novel strategy for reducing thromboinflammation in systemic and neurologic manifestations of COVID-19. Since anti-fibrin antibody 5B8 has protective effects (this study) and protective autoantibodies have been reported in COVID-19 patients (29), some fibrin autoantibodies may be protective against thromboinflammation. Whether the human fibrin autoantibodies in COVID-19 reported here have potential divergent functions, or associate with clinical characteristics, or have biomarker value, warrants further investigation. Furthermore, the fibrin-Spike coagulation and inflammation assays we describe can serve as a discovery platform for identifying new therapeutics for COVID-19. Targeting fibrin may be an efficacious therapy to suppress thromboinflammation in both acute COVID-19 and PASC patients.
Funding
The Roddenberry Foundation (WCG and KA)
National Institutes of Health grants NS120055, R24GM137200 (MHE)
National Institutes of Health grant R35 NS097976 (KA)
National Institutes of Health T32 AI007334 (EGS)
National Science Foundation NSF2014862-UTA20-000890 (MHE)
James B. Pendleton Charitable Trust (WCG)
Edward and Pearl Fein (KA)
Simon Family Trust (KA)
Author contributions
Conceptualization: KA, JKR, WCG
Methodology: JKR, EGS, MM, MHE, TJD, WCG, KA
Software: RS
Validation: EGS, MM, KD, YM, YL, EH, MHE, WCG, KA
Formal analysis: RT, ARP, JKR, EGS, KD, YL, ZY, RS
Investigation: JKR, EGS, KD, MM, YM, YL, EH, TJD, ZY, KLY, RMA, MHE
Resources: CMS, KLY, SSZ, MHE, WCG, KA
Data curation: JKR, EGS, KD, MHE, RS, RT, ARP, WCG, KA
Writing-original draft: KA, EGS, JKR, KD
Writing-Review & Editing: all authors, Supervision: KA, WCG, MHE
Project Administration: KA, WCG
Funding acquisition: KA, WCG, MHE, EGS
Competing interests
KA is a co-founder, scientific advisor, and board director of Therini Bio. KA is the inventor of the 5B8 patent. KA, JKR, MM, and WCG. are inventors on a patent application for 5B8 use in COVID-19. KA, JKR, MM, EGS and WCG are inventors on a patent application related to Spike-induced thromboinflammation model. KA and JKR are inventors on fibrin assay patents. Their interests are managed in accordance with their respective institutions’ conflict of interest policy.
Data and materials availability
Expression plasmid pCAGGS seq nCoV19 of soluble trimeric spike deposited by Florian Krammer is freely available from BEI Resources. Distribution of Spike PVs, mice, and antibodies to non-profit investigators will be under a Material Transfer Agreement (MTA). All data are available in the manuscript or the supplementary materials.
Acknowledgments
We thank Deepak Srivastava and Melanie Ott for critical reading of the manuscript, Florian Krammer for sharing the plasmid for mammalian expression of recombinant Spike, Andrew S. Mendiola, Min-Gyoung Shin, Eunbi L. Ryu and the Gladstone Flow Cytometry Core for technical assistance, and Stephen Ordway for editorial assistance.