Fibrin clot characteristics and anticoagulant response in a SARS‐CoV‐2‐infected endothelial model

Abstract Coronavirus disease 2019 (COVID‐19) patients have increased thrombosis risk. With increasing age, there is an increase in COVID‐19 severity. Additionally, adults with a history of vasculopathy have the highest thrombotic risk in COVID‐19. The mechanisms of these clinical differences in risk remain unclear. Human umbilical vein endothelial cells (HUVECs) were infected with SARS‐CoV‐2, influenza A/Singapore/6/86 (H1N1) or mock‐infected prior to incubation with plasma from healthy children, healthy adults or vasculopathic adults. Fibrin on surface of cells was observed using scanning electron microscopy, and fibrin characteristics were quantified. This experiment was repeated in the presence of bivalirudin, defibrotide, low‐molecular‐weight‐heparin (LMWH) and unfractionated heparin (UFH). Fibrin formed on SARS‐CoV‐2 infected HUVECs was densely packed and contained more fibrin compared to mock‐infected cells. Fibrin generated from child plasma was the thicker than fibrin generated in vasculopathic adult plasma (p = 0.0165). Clot formation was inhibited by LMWH (0.5 U/ml) and UFH (0.1–0.7 U/ml). We show that in the context of the SARS‐CoV‐2 infection on an endothelial culture, plasma from vasculopathic adults produces fibrin clots with thinner fibrin, indicating that the plasma coagulation system may play a role in determining the thrombotic outcome of SARS‐CoV‐2 infection. Heparinoid anticoagulants were most effective at preventing clot formation.

low-molecular-weight-heparin (LMWH) and unfractionated heparin (UFH). Fibrin formed on SARS-CoV-2 infected HUVECs was densely packed and contained more fibrin compared to mock-infected cells. Fibrin generated from child plasma was the thicker than fibrin generated in vasculopathic adult plasma (p = 0.0165). Clot formation was inhibited by LMWH (0.5 U/ml) and UFH (0.1-0.7 U/ml). We show that in the context of the SARS-CoV-2 infection on an endothelial culture, plasma from vasculopathic adults produces fibrin clots with thinner fibrin, indicating that the plasma coagulation system may play a role in determining the thrombotic outcome of SARS-CoV-2 infection. Heparinoid anticoagulants were most effective at preventing clot formation.

K E Y W O R D S
anticoagulation, blood coagulation, cell culture, electron microscopy, viruses

INTRODUCTION
The coronavirus disease 2019 (COVID-19) pandemic, caused by SARS-CoV-2, is linked to haemostatic dysregulation with both severe and mild cases of COVID-19 associated with coagulopathy [1]. Between 10% and 20% of COVID-19 patients are admitted to the intensive care unit, and 31% of those patients experience a thromboembolic event [2,3]. Hospitalised COVID-19 patients have increased fibrin monomer and d-dimer levels, with severe and critical COVID-19 patients having prolonged prothrombin time and thrombocytopenia [4][5][6][7]. There is also substantial evidence that COVID-19 may trigger vasculitis and vasculopathy, which may be linked to the systemic effects of severe disease [8]. COVID-19-associated coagulopathy contributes significantly to the overall morbidity and mortality from COVID-19, with post-mortem evidence of thrombosis detected in the lung vasculature [9]. COVID-19-associated coagulopathy does not impact all people in the same way. COVID-19-infected children younger than 12 years of age have low rates of thrombosis, while adults with vasculopathy and COVID-19 have significantly higher thrombotic rates and suffer a rapid and more severe decline in health [10,11].
While there is an abundance of research into COVID-19 and the response to SARS-CoV-2, a comprehensive understanding of underlying mechanisms of COVID-19-induced coagulopathy is lacking. An overwhelming inflammatory response coupled with increased platelet activation, increased blood viscosity and endothelial injury create a prothrombotic vascular environment [12][13][14]. The close interaction of inflammation, immunity and coagulation has been well established, and their co-functioning has been termed thromboinflammation [15].
Inflammatory molecules, endothelial activation or the pathogen itself may interrupt or stimulate coagulation at any stage of blood-clotting [16]. Certain viral infections, such as Ebola, promote a bleeding phenotype, while other viral infections, such as Influenza A, promote thrombosis [17,18]. Avian influenza A/H5N1 infection exemplifies the complex interplay in thromboinflammation, as this virus stimulates both a bleeding and thrombotic phenotype [19].
There are multiple factors that can affect fibrin characteristics in a blood clot, including changes in pH, temperature and surrounding ion concentration [20]. However, the impact of a viral infection on the fibrin clot structure is currently unknown. When considering potential interventions, it is crucial to understand the structure of the SARS-CoV-2induced blood-clot, particularly as factors such as fibrin fibre thickness and density can predict clot stiffness, resistance to lysis and therefore disease-related morbidity [21][22][23].
Anticoagulation is widely used in the treatment of severe COVID-19; however there is some debate as to the optimal drug, dose or administration schedule [24]. Unfractionated heparin (UFH) and lowmolecular-weight heparin (LMWH), bivalirudin and defibrotide are among the drugs that have been used and/or studied in the setting of COVID-19 [25][26][27]. The heparinoids (UFH and LMWH) both act upon antithrombin to inactivate thrombin, FXa and FIXa. UFH is large enough to act by binding antithrombin and thrombin, while the smaller LMWH binds antithrombin and FXa directly [28]. Bivalirudin, a direct thrombin inhibitor, binds to thrombin to produce a transient inhibitory effect of coagulation [29]. The mechanism of anticoagulation of defibrotide is not fully understood, although some of its pathways of action include directly antagonising thrombin, inhibiting the tissue factor pathway, and up-regulating tissue plasminogen activator [30,31].
There is little clarity on the effectiveness of prophylactic anticoagulation, despite some evidence that prophylactic heparinoids may reduce mortality in hospitalised COVID-19 patients [32]. Reliable evidencebased interventions are required [33].
We hypothesised that SARS-CoV-2 infection of endothelial cells promotes thrombosis and that this could be simulated in vitro. Based on the evidence of vasculitis in the pathogenesis of COVID-19, we further hypothesised that individuals with a pre-existing vasculopathy may be at a higher risk. We aimed to show the impacts of SARS-CoV-2 infection on the fibrin characteristics in healthy children, adults and adults with vasculopathy and to study the impacts of anticoagulants on this system. To facilitate this, we also aimed to develop an in vitro endothelial cell-culture model to examine the contribution of the endothelium to haemostasis.

Cell culture
Human umbilical vein endothelial cells (HUVECs) were cultured in this experiment. To produce HUVECs, flasks were coated with Geltrex (Thermo Scientific A1413302) in Dulbecco's Modified Eagle Medium (DMEM) (Thermo Scientific, 111965118) in a ratio of 1:100, 1 h prior to seeding. HUVECs were cultured to 95% confluency and split on a 24 well Geltrex coated plate (0.05*10 6 ) 2 days prior to the experiment.
All cells were cultured in endothelial growth media-2 (EGM-2) (Lonza, CC3202). Blood from healthy children was collected as a part of the HAPPI Kids study [34]. Specifically, healthy children attending the hospital for minor, elective day-surgery were enrolled into the study. Blood was collected via the cannula inserted as a part of standard clinical care.

Plasma sample collection
Blood from healthy adults was collected by venepuncture. Blood samples were collected in tubes containing 1-part sodium citrate to 9-parts blood. Plasma was collected from blood samples by centrifugation at 2500 g for 10 min and stored at −80 • C until experimentation. Vasculopathic plasma samples were collected from adults who had previously suffered from their first ischaemic stroke episode and were enrolled in the START study [35]. Blood was collected from vasculopathic adults by venepuncture of the antecubital region tubes containing 1-part sodium citrate to 9-parts blood. Plasma was extracted by centrifugation at 1100-1300 g for 10 min and stored at −80 • C until experimentation.
Plasma samples from children, adults and vasculopathic adults were all collected as part of studies undertaken prior to the COVID-19 pandemic. Hence all samples were from SARS-CoV-2 negative subjects.
Plasma was thawed and formed into healthy child, healthy adults and vasculopathic adult plasma pools on the day of each experiment. After formation, plasma pools were stored at 4 • C and used within 4 h of thawing.

Viral infection
The SARS-CoV-2 virus (Vic/01) was isolated in Vero cells at the Peter Doherty Institute as previously described [36]. Influenza and was used as a non-coronavirus control. Cell culture infections using SARS-CoV-2 were performed at Biosafety level 3 laboratory conditions. Viruses were diluted in phosphate-buffered saline (PBS) to achieve a viral inoculum of 200 µl/well at 0.1-1 multiplicity of infection (MOI). Cells were gently washed in PBS before addition of virus.
PBS was used as a control in mock-infected cells. Inoculated cells were incubated at 37 • C and 5% CO 2 for 1 h, after which the inoculum was removed by pipetting. Cells were then washed with PBS and incubated in the appropriate culture medium at 37 • C and 5% CO 2 .
This viral infection procedure was repeated using Vero cells, with the supernatant used as a positive control for the viral replication detection experiment.

Clot formation experiment
HUVEC monolayers were cultured in 24-well plates. Cells were inoc-

Anticoagulation experiments
HUVECs were infected with SARS-CoV-2 as above. Prior to addition to the infected HUVECs, plasma pools were incubated with one of the following anticoagulation drugs: 0.02 µg/µl defibrotide, 0.005 µg/µl bivalirudin, 1 U/ml or 0.5 U/ml LMWH and 0.7 U/ml or 0.4 U/ml UFH.
The doses of LMWH and UFH, representative of a higher therapeutic dose and lower prophylactic dose, were 0.25 U/ml and 0.125 U/ml of UFH, and 0.2 U/ml and 0.1 U/ml of LMWH.

Viral detection in HUVEC cultures
HUVECs were cultured on Nunc Thermanox coverslips (Thermo Fisher In each HUVEC clot formation experimental replicate, 400 µl of supernatant was sampled from two SARS-CoV-2-infected wells and one mock-infected well that did not receive human plasma at 48 h postinoculation. Sample aliquots were frozen at −80 • C until testing by qRT-PCR as described above.

Image quantification
Fibrin strands in SEM images were manually quantified using FIJI ImageJ software (National Institutes of Health, NIH). Using the line and measure tool, two blinded independent reviewers selected 20 strands, from three randomly selected regions of the image (60 total strands total per image).

Statistical analysis
Analysis

Viral detection in HUVEC cultures
In the replication kinetics experiment, amplification of the SARS-CoV-2 E gene was detected in HUVEC culture supernatant by qRT-PCR following inoculation with Vic/01 at 0.1 MOI. Figure 1 shows the viral

Clot formation experiment
Infection of HUVEC monolayers using influenza A/Singapore/6/86 (H1N1) has been previously reported to induce procoagulant activity in vitro [37]. HUVECs remained adherent to the wells at 48 h following inoculation and allowed for stable clot formation suitable for further imaging. HUVECs were therefore used in subsequent experiments using SARS-CoV-2.
The clot formation experiment was repeated three times, with three different plasma pools. In the SARS-CoV-2 experiments, fibrin was visible in all wells that were incubated with plasma. None of the wells that were incubated with EGM-2 alone showed any fibrin, and cells were present on wells after SARS-CoV-2 or influenza infection (Figure 2). Fibrin was not detected in wells that only contained media, and there were no significant differences between fibrin in mock-infected wells with different plasma. Fibrin density appeared to be greater in SARS-CoV-2-infected cells compared to influenza and healthy controls, although there was no apparent difference in density of fibrin formed from plasma collected from healthy children, healthy adults or vasculopathic adults. Statistically significant differences in fibrin thickness across groups were confirmed by ANOVA (p = 0.21), and fibrin was significantly thinner in vasculopath clots compared to paediatric clots (p = 0.0165) (Figure 3).

Drug response in SARS-CoV-2-infected cells
At standard concentrations, bivalirudin and defibrotide had no effect on preventing fibrin clot formation (Figure 4)

DISCUSSION
We showed significant differences in fibrin characteristics in an in vitro model of SARS-CoV-2 infection of endothelial cells using plasma samples from healthy children, healthy adults and vasculopathic adults.

Fibrin clot characteristics
The characteristics of a fibrin clot can be used as an indicator of clot strength, severity and resistance to lysis [20]. The size of the pores in the fibrin scaffold represents the clots strength, permeability and susceptibility to breakdown by lytic enzymes [39]. Increased thickness of fibrin fibres has been correlated with a less densely packed clot structure, that is, less resistant to lysis [40]. In the context of our results, this knowledge suggests that the thinner fibres produced in vasculopath plasma may represent a clot that has a greater resistance to natural fibrinolysis and is more likely to become patholog-ical. Recent research has indicated that fibrinolysis is inhibited in patients with severe COVID-19 [41]. This is particularly concerning in the context of our findings that formation of vasculopathic clots was also resistant to anticoagulation, particularly low doses of UFH. We expected that plasma from healthy children would produce thicker fibrin strands than healthy adults, and vasculopathic adults, as COVID-19-associated coagulopathy is extremely rare in children [42]. Based on our findings, we propose that individuals with a history of vasculopathy (such as stroke) are more vulnerable to serious COVID-19-associated coagulopathy.

Drug responses
Despite knowing that coagulopathy is a major complication of SARS-CoV-2 infection in humans since the virus pandemic emerged in early 2020, optimal treatment for COVID-19-induced coagulation has not been identified [24]. Our findings that LMWH and UFH were effective in preventing fibrin clot formation are particularly interesting. LMWH is often seen as a more attractive alternative to UFH due to more predictable pharmacodynamics, longer half-life and less requirement for laboratory monitoring [43]. Our results suggest that UFH may be more effective in the context of COVID-19. LMWH has a predominant anti-Xa effect, while UFH has both anti-Xa and anti-IIa effect, all mediated via antithrombin. While this appears to suggest that the anti-IIa behaviour of UFH is indeed the explanation for its increased effectiveness, we showed no anticoagulant effect of the direct Factor IIa inhibitor bivalirudin at therapeutic dose. However, the broader activity of UFH may be important. The viral replication data demonstrated that the difference was not due to a viricidal effect of UFH, which was unlikely in any case following the short duration of UFH incubation with the cell surfaces. However, we cannot exclude that UFH had an impact on endothelial cell drivers of plasma coagulation. While defibrotide interacts with a number of systems, it has been identified as an anticoagulant and profibrinolytic [31]. However, we failed to detect an effect of defibrotide on prevention of COVID-19-associated coagulation at therapeutic dose. While it may be possible that higher doses of bivalirudin or defibrotide may be effective, we did not go increase concentration beyond therapeutic dose.
It has been previously suggested, based on the structural differences, that paediatric clots were more susceptible to lysis [21]. We found no difference in the effectiveness of anticoagulants in children and healthy adults. This may suggest that the structural differences detected previously do not directly translate to clot behaviour, as previously hypothesised.

Endothelial haemostasis assay
Previous experiments have examined fibrin clot structure using SEM and implemented automated analysis software to characterise fibrin characteristics [21]. Previous studies formed clots on artificial surfaces such as filter paper, which may have been prothrombotic and did not account for the substantial impact of endothelial cells to coagulation [44]. Importantly, while we re-calcified the citrated plasma samples using a small amount of calcium to reverse the effects of sodium citrate we did not add any specific coagulation activator-rather, the endothelial surface provided the stimulus for thrombosis.

Limitations and next steps
HUVECs are derived from discarded umbilical cords of healthy babies and are typically used to model the behaviour of endothelial cells in vitro [45]. HUVECs do not necessarily represent all endothelial cell types in the body, and endothelial cells from differing sources may respond differently in vitro, particularly to infection [46]. Future studies may choose to focus on lung-specific endothelial cells to mimic the site of infection more accurately. In this study, we studied the impact of anticoagulants; however future experiments that incorporate fibrinolytic drugs, such as tissue plasminogen activator or urokinase, after clot-formation may be valuable to characterise fibrin-clot resistance to fibrinolysis.
These experiments were conducted without blood cells, including red blood cells, white blood cells and platelets. This is of relevance, as UFH also acts on macrophages and platelets , potentially masking some of the effects of the drug in our model [43]. Further, drugs were added 48 h following viral infection, and endothelial cells may have been activated significantly prior to addition of the drug. This may have reduced the effect of the anticoagulants, particularly defibrotide, which acts directly on the endothelium [30].

CONCLUSION
In conclusion, we demonstrated that fibrin characteristics caused by SARS-CoV-2 infection in endothelial cells exposed to plasma are impacted by age and prior vascular disease in the plasma source.
Plasma from stroke patients produces fibrin clots in the presence of low doses of LMWH, while UFH was effective in all groups at extremely low doses, suggesting it may be valuable for COVID-19-associated coagulopathy.