Extracellular vesicles in plasma and cerebrospinal fluid in patients with COVID‐19 and neurological symptoms

Increased levels of extracellular vesicles (EVs) are associated with haemostatic disturbances in various clinical settings. However, their role in COVID‐19 patients is still not fully clear. In the present study we investigated EVs in plasma from patients with COVID‐19 and neurological symptoms in relation to the activation of coagulation.

Coronavirus disease 2019 (COVID-19) patients exhibit neurological manifestations that commonly include fatigue, headache and myalgias 1 whereas age and severity of disease are, among other things, significant factors contributing to a higher incidence of neurological complications. 1,2Furthermore, a considerable number of patients develop acute respiratory distress syndrome (ARDS) with a high incidence of coagulopathy, including microvascular thrombosis. 3,4Coagulopathy in COVID-19 presents with prominent elevation of D-dimer and fibrin/fibrinogen degradation products and has been proposed to be connected to impaired fibrinolysis. 5There is also a correlation between COVID-19 severity, neurological complications and central nervous system (CNS) injury markers in the CSF. 6mplement activation, part of the innate immune system against pathogens, plays a major role in ARDS as well as CNS injury. 7,8Key proteins of the complement system have been found both in CSF and brain tissue of humans and animals with brain injury. 8,9The final products of the complement cascade collectively influence different aspects of coagulation, forming and enhancing crosstalk between these two closely connected systems, as reviewed by Keragala et al. 10 Simultaneous activation of the complement and coagulation pathways in patients with severe COVID-19 manifests as an inflammatory state and hypercoagulability. 11tracellular vesicles (EVs), previously called membrane microparticles, are considered to be important mediators of intercellular communication involved in normal physiological processes as well as in various pathological conditions. 12Categories of EVs, such as plateletderived EVs, have shown excellent inflammation-targeting ability and potential immune-stimulatory activities, while they may promote procoagulant mechanisms in the pathogenesis of venous thrombotic disease. 13In COVID-19, EVs have been proposed to transfer the virus receptor ACE2 to recipient cells, thus making these cells susceptible to virus docking. 14,15 this study we aimed to highlight the interplay between activation of complement anaphylatoxins C3a and C5a as well as the terminal complement complex (TCC) and coagulopathy in COVID-19 patients suffering neurological manifestations during the course of the disease.Therefore, we sought to identify and measure the levels of EVs of different cell origins in plasma and cerebrospinal fluid from these patients in relation to coagulation disturbances assessed by global haemostatic assays.

| Patients
This was a prospective single-centre study at Uppsala University Hospital (UUH).Patients with positive PCR results for SARS-CoV-2 in nasopharyngeal swabs (Abbott, Abbott Park, IL, USA) and at least one new-onset neurological symptom were included from April until July 2020.Pathological neurological findings were documented as follows: cranial nerve affection, central or peripheral paralysis, extrapyramidal, sensory symptoms and altered mental status including confusion, encephalopathy and reduced level of consciousness graded by means of the Glasgow Coma Scale (GCS).The most severe neurological symptoms, including GCS data during the disease were documented, as well as GCS results within 24 h of lumbar puncture analysis.A total of nineteen patients were included, 37% female, with a median age of 64 (39-85) years.

| Ethics
The study was approved by the Swedish Ethics Review Authority (2020-01883).Informed consent was obtained from each patient, or next-of-kin if a patient was unable give consent.The collection of plasma samples from twenty-three healthy subjects was part of an ongoing study already approved by the Regional Ethics Committee in Stockholm (#2014/987-31).The Declaration of Helsinki and its subsequent revisions were followed.

| Plasma and CSF sampling
Routine blood and CSF work-up was carried out in conjunction with or after the period of most severe neurological symptoms.In detail, for both patients and healthy controls, blood was drawn by venipuncture directly into BD Vacutainer Citrate Tubes containing 0.109 M sodium citrate.Within one hour of collection, the tube was subsequently centrifuged at 2400 Â g for 20 min at room temperature for preparation of platelet-poor plasma.Aliquots were immediately frozen and stored at À70 C until further analysis.
CSF lumbar puncture (LP) was performed in patients with neurological symptoms and no contraindications (fourteen out of our nineteen COVID-19 patients).CSF was collected in sterile tubes and stored at À70 C for flow cytometric analysis of EVs.
All analyses with plasma and CSF samples were performed after a single freeze-thaw cycle.

| Flow cytometric analysis of EVs
Plasma and CSF samples were tested for the presence of EVs of size 0.1-1 μm, as described by Zong et al., 16 using flow-cytometry.All samples were thawed in a water bath for 5 min at 37 C and immediately processed for immunolabelling.20 μL of plasma or CSF were incubated with antibodies for 30 min at RT in the dark.The following antibodies were purchased from BD Biosciences unless stated otherwise: anti-myeloperoxidase (MPO)-PE, anti-CD61-APC, anti-CD45-APC, anti-CD51/61-PE and anti-CD142-PE for identification of MPO+ EVs, platelet-derived EVs (PEVs), leucocyte-derived EVs (leucoEVs), endothelial-derived EVs (EEVs) and tissue factor (TF+) EVs, respectively.Anti-TCC-FITC, anti-C3a-FITC and anti-C4d-FITC were purchased from USBiological, Novus Biologicalis and LSBio respectively.Lactadherin-FITC was purchased from Haematologic Tech for identification of phosphatidylserine (PS+) EVs.A complete list of the antibodies can be found in Supplementary Table  The mixture was subsequently supplemented with 500 μL PBS and tested using a FACs Canto flow cytometer (BD Biosciences).Fluorescent compensation parameters are shown in Supplementary Table 2.For acquisition, the EV gate was determined using Megamix-Plus SSC beads (BioCytex) of sizes 0.16, 0.20, 0.24 and 0.5 μm on side scatter.This SSC-standardized gate allows standardized acquisition of EVs of between $0.17 μm-eq.and 0.5 μm-eq. in (similar to the range of 0.3-1 μm-eq.using FSC beads), which corresponds to the size of EVs.To discriminate between EVs and negative events and to set the gates for positive EVs, a background control using Triton X-100-lysed EVs was performed as previously described by Rousseau et al 17 The gating strategies for identification of different EV subpopulations in plasma and CSF samples are presented in Supplementary Figure 1 and Supplementary Figure 2, respectively.In this study, the results are expressed as EV events/μL, using FACS Canto volume measurement over 90 sec, as described by Campello et al 18 Files were exported and data were evaluated by using FlowJo software (BD Biosciences).

| Overall haemostatic potential (OHP) and clot turbidity assays
We determined the overall haemostasis potential in citrated plasma by constructing fibrin aggregation curves, as described previously. 19iefly, citrated plasma was recalcified and supplemented with thrombin (Sigma-Aldrich) at 0.04 U/mL, with or without tissue plasminogen activator (t-PA) (Boehringer Ingelheim) at 300 ng/mL.The real-time absorbance change in the sample was monitored at λ = 405 nm, every 12 s for one hour.The values for the overall coagulation potential and the overall haemostasis potential were calculated on the basis of fibrin formation and fibrinolysis curves, respectively, and the difference in these curves gives the overall fibrinolytic potential (OFP).
Fibrin clot density was characterized by clot turbidity assay, this being a modification of the OHP assay.The turbidimetric curve for determination of OCP was used to assess the following parameters, as described previously 20 : "lag-time", the time-point when exponential growth of the curve begins as a measure of the clotting time; "max absorbance" (Max Abs), the average value of three consecutive points where the curve reached a plateau as a measure of clot density; "max-Abs time", the time needed for the curve to reach the plateau and the "slope", measuring the polymerization rate of fibrin.

| Scanning electron microscopy (SEM) of fibrin clots
After performing the OHP assay, the clots were washed, fixed in 2.5% glutaraldehyde and stored at 4 C.The specimens were analysed in an Ultra 55 field emission scanning electron microscope (Carl Zeiss) and individual fibre thickness was measured as previously described. 21

| Routine laboratory parameters
White blood cell and platelet counts, together with the analysis of C-reactive protein (CRP), D-dimer and interleukin-6 were performed according to routine laboratory analytical methods at Uppsala University Hospital, Sweden.

| Statistics
Statistical analysis was performed using Graph Pad Prism (version 9.1 for Windows) and SPSS Statistics (26 IBM) software.All parameters were evaluated for normality by using the Shapiro-Wilk test and are presented as mean ± standard deviation of the mean (for parametric data), or median and interquartile range (for nonparametric data).Differences in estimated variables between groups were assessed by using Student's t-test (parametric) or the Mann-Whitney (nonparametric) unpaired t-test.Correlations between parameters were evaluated by using Spearman's (nonparametric) rank correlation test.
Significant differences were statistically defined by p values of ≤0.05.

| RESULTS
Clinical characteristics and laboratory parameters of the patients are summarized in Supplementary Table 3.The most common symptoms at admission were anosmia and headache and the most common symptom at the time of lumbar puncture was altered mental status.All patients had at least one neurological symptom.The most frequent comorbidities were hypertension and diabetes mellitus, while increased BMI (>29.9 kg/ m 2 ), indicating overweight, was present in the whole patient group.At the time of blood sampling thirteen (72%) patients were being treated with low-molecular-weight heparin (LMWH) and four (22%) were on low-dose aspirin (LDA).A total of 23 healthy controls were included, 70% female, with a median age of 58 (53.5-63).
Healthy subjects were not treated with anticoagulant drugs and had no known active infection or cancer.

| Haemostatic parameters and fibrin clot properties
COVID-19 patients exhibited higher OCP ( p < 0.01) and OHP (p < 0.001) levels, while OFP values were significantly lower ( p < 0.01) in comparison with controls (Table 1).Significantly prolonged Lag time and time to Max Abs, and also higher Max Abs values and slower clotting rates were found in COVID-19 patients compared with healthy controls.Image analysis by SEM revealed fibrin networks of regular structure that consisted of thin fibres, both in clots originating from COVID-19 patients (Figure 1A) and from healthy controls (Figure 1B).Measurement of fibre diameter showed thinner fibres in the clots from COVID-19 patients compared with controls (Figure 1C).

| Identification of EVs in plasma and expression of different complement biomarkers
The concentrations of all EVs exposing phosphatidylserine (PS+) and terminal complement complex (TCC+) were significantly higher in COVID-19 patients than in healthy controls, while there were no differences in the levels of all EVs expressing complement factors C4d and C3a between patients and controls.EVs derived from platelets (PEVs) and neutrophils (MPO+ EVs) and the subpopulations of these EVs expressing complement factors TCC and C3a, respectively, were predominant in patients compared with controls.No difference in the concentrations of TF+ EVs was found between patients and controls.
The presence of EVs derived from leucocytes (LEVs) and endothelial cells (EEVs) was also explored, resulting in only a few events registered in flow-cytometry and were regarded as nonspecific results (data not shown).The numbers of events are presented in Table 2 and illustrated in Supplementary Figure 3. Representative flow cytometry plots of different EVs are presented in Supplementary Figure 1.

| Correlation of plasma EVs with coagulation
Significant correlations between different circulating EV populations and coagulation parameters were found (Table 3).OCP levels were significantly correlated with MPO+ EVs, PS+ EVs and TCC+ PEVs, whereas OHP levels were significantly correlated with PS+ EVs, PEVs, TCC+ EVs and TCC+ PEVs.OFP levels showed a significant negative correlation with PEVs, TCC+ EVs and TCC+ PEVs.

| Identification of EVs in CSF
We examined the CSF of fourteen out of our nineteen COVID-19 patients and identified the presence of different EV populations (Table 4).The most prominent population was that of PEVs, with the  highest number of positive events detected within the gate, followed by the populations of MPO+ and leucocyte-derived EVs.

| DISCUSSION
In the present study, we investigated the role of EVs in plasma from patients with COVID-19 in relation to the activation of coagulation, using global haemostatic assays, flow-cytometry and imaging techniques.We confirmed the procoagulant status of the patients and found significantly increased levels of several plasma populations of EVs, together with higher expression of complement factors on EVs compared with healthy controls.Moreover, we identified the presence of EVs in the CSF of patients suffering neurological symptoms during COVID-19 infection.
SARS-CoV-2 interactions with the blood-brain barrier may contribute to neurological dysfunction. 22Neuroradiological findings include microvascular pathology, especially concerning the corpus callosum and juxtacortical areas.Ischaemic and haemorrhagic manifestations are commonly observed, as well as leucoencephalopathy and olfactory bulb signal abnormalities.Also, prominent optic nerve subarachnoid spaces and enhancement of the parenchyma, leptomeninges, and cranial and spinal nerves have been noted. 23Perfusion abnormalities in MRI as well as cerebral ischaemic stroke have also been described in neuro-Covid patients. 24However, the underlying mechanism is not known.The rate of fibrinolysis is also directly influenced by the fibrin structure itself, with looser fibre networks being lysed faster than tightly packed fibrin fibres. 28Indeed, we noted that fibrin clots formed from the plasma of COVID-19 patients had thinner fibres forming a dense network with smaller pores compared with healthy controls, which indicates a potential role of altered fibrin architecture in CNS coagulopathy during COVID-19, as previously found in thromboembolic ischaemic stroke patients. 29One can assume that activated coagulation during viral infections can be beneficial, creating resistance to fibrinolysis and helping to counteract viral infection by trapping viruses in the fibrin network.However, in excessive inflammatory and coagulative responses during SARS-Cov-2 infection, massive fibrin production and deposition occurs, causing severe organ damage.
In the course of COVID-19 infection and local tissue injury, activation of the complement system is necessary for endothelial activation, with consequent loss of thrombomodulin and glycocalix and exposure of negatively charged phospholipid surfaces and TF.Therefore, it is not surprising that circulating phosphatidylserineexposing EVs (PS+ EVs) predominate in samples from patients with severe COVID-19.Exposure of PS on the surface of EVs in COVID-19 patients, alone or in combination with TF, facilitates the assembly of prothrombinase complexes, resulting in increased thrombin generation and subsequent fibrin production. 30The significant correlations between PS+ EV concentrations and OCP and OHP values may suggest an association with increased fibrin generation and impaired fibrinolysis, as previously reported in hypercoagulable patients. 31wever, we have not yet found significantly increased concentrations of TF+ EVs in samples from COVID-19 patients compared with healthy controls.We suggest that in COVID-19 patients with neurological manifestations, the procoagulant effect of PS+ EVs is due to the exposure of phosphatidylserine, independently of TF.
In our study we observed circulating PEVs at higher levels in COVID-19 patients compared with control subjects, similar to the situation previously described and partially attributed to platelet hyperactivation and aggregation. 32Together with correlation analysis data concerning the concentrations of PEVs with OHP and OFP values, our findings support the idea that PEVs may be associated with a higher rate of fibrin formation and a lower rate of fibrinolysis in COVID-19 patients with neurological symptoms.The procoagulant activity of PEVs may contribute to denser fibrin structures consisting of thinner fibres, as shown in our SEM analysis.We can hypothesize that this alteration could be a result of direct contact between fibrin fibres and PEVs in a COVID-19 setting.Previous findings have indicated that PEVs may participate in alterations of the fibrin network by attaching to particular positions on fibrin fibres in an in-vitro plasma model. 16However, to answer this question, further studies must be implemented.
Neutrophils have been found to be upregulated in the bronchoalveolar lavage fluid of COVID-19 patients. 33Circulating activated neutrophils change their cytoskeletal structure, with retention in the pulmonary capillary bed and pass through the lung interstitial region and epithelial cells into the alveoli.Therefore, we sought to detect EVs expressing MPO, originating mostly from neutrophils but also from monocytes, previously shown to be the hallmark of an inflammatory response in patients with ANCA-associated vasculitis. 34We observed that MPO+ EVs are significantly increased in the circulation of COVID-19 patients with neurological symptoms, and they are correlated with higher OCP values, indicating an association with increased coagulation potential.We hypothesize that during neutrophil migration into the alveolar cavity, MPO+ EVs may play a role in epithelial and endothelial injury, thus potentially contributing to coagulation activation, fibrin deposition, activation of macrophages and tissue damage, as previously demonstrated. 33tivation of the terminal complement pathway has been previously described in brain injury. 7,8In this regard, we are the first to report the elevated expression of TCC, the final product of the complement cascade, on the membrane of EVs in COVID-19 patients with neurological complications.This is in agreement with previous findings that increased levels of TCC complex in plasma precede the development of ARDS and are associated with increased mortality rate and thromboembolic complications. 35The correlation between concentrations of TCC+ PEVs and values of OHP, OCP and OFP suggest that PEVs exposing TCC may be involved in hypercoagulability and impaired fibrinolysis.
We further examined the presence of anaphylatoxin C3a, the cleaved product of C3, on the surface of EVs.We made the novel observation that a higher number of EVs express C3a with and without co-expression of the neutrophil marker MPO in COVID-19 patients.We might speculate that these EVs are mediators of activation and polarization of lymphocytes through the vascular endothelium.The elevated levels of neutrophil-derived EVs expressing C3a may also contribute to the deregulated action of neutrophils involved in lung injury, as described previously. 36Similar to our results, increased levels of C4d, C3a, C3d as well as C5b-9 have been previously found in the plasma of critically ill COVID-19 patients and this was associated with organ damage, illness severity scores, and survival. 37s have been previously found in the cerebrospinal fluid (CSF) and they have been involved in the transfer of toxic proteins between cells, with direct implications as regards neurodegenerative disorders such as Alzheimer's disease. 38To the best of our knowledge, we are the first to provide direct evidence of the presence of EVs in the CSF of COVID-19 patients suffering neurological symptoms, with plateletderived EVs being the most prominent population.This suggests that the generation of EVs is an event associated with a procoagulative profile and neurological symptoms.We have previously reported increased CNS injury markers in CSF as well as SARS-Cov-2 antigens S/S1 in this cohort. 6,39It is possible that EVs found in CSF could contribute to complement factors driving an intrathecal detrimental inflammatory response, blood-brain barrier disruption and finally CNS injury.The terminal complement complex C5b9 has previously been detected in the CSF and brain tissue of patients with traumatic brain injury as well as in experimental animal models. 8,9r study has several limitations.Firstly, the number of patients included was relatively small, and it was not possible to synchronize the blood sampling with respect to disease onset.Secondly, limited availability of plasma samples did not allow the employment of all our assays in every recruited patient, resulting in limited statistical power in the correlation analysis, and making it impossible to detect potential differences in patient subgroups.Nevertheless, we were able to apply multiple assays with moderate throughput but high-level clinical relevance.Thirdly, at the time of blood sampling most patients were receiving anticoagulant therapy according to guideline recommendations.Thus, it was difficult to examine changes in procoagulant and fibrinolytic potential specifically induced by COVID-19.However, we provide data on a persisting procoagulant condition in these patients despite the ongoing anticoagulant treatment.Further studies should include more patients and plasma samples so that potential differences in patient subgroups can be detected, and other potential variables and parallel mechanisms should be further explored.
In conclusion, our study shows that COVID-19 patients with neurological symptoms are in a procoagulant state and that EVs might be the drivers.We observed, for the first time, the expression of TCC/C5b9 on circulating EVs in plasma and CSF.This is also the first study reporting on EVs in CSF that might contribute to neurological manifestations.

F I G U R E 1
Representative SEM images of fibrin structures from plasma of (A) a COVID-19 patient and (B) a healthy control.(C) Fibre thickness (mean ± standard deviation) in fibrin clots in patients versus controls, n = 10 fibrin clots.Data are expressed as mean ± SD, *** p < 0.001.
time in clots from COVID-19 patients which clearly suggests enhanced fibrin formation.Diminished OFP values were also presented in COVID-19 patients which directly implies impairment of fibrinolysis in COVID-19 patients, similarly to the situation among critically ill patients with other diseases where hypofibrinolysis is a common feature.27

1
Parameters of global haemostatic assays and fibrin clot turbidity parameters in plasma.
Correlation of extracellular vesicles with other biomarkers in COVID-19 patients and controls.
Thus, we employed the OHP assay to assess fibrin formation and fibrinolytic kinetics.The summed absorbances over 60 min in parallel reactions of clotting (no tPA) and lysis curves (with tPA) revealed higher OHP and OCP values in COVID-19 patients, despite ongoing anticoagulant treatment in the majority of patients.The clot turbidity assay showed higher Max absorbance together with prolonged slope T A B L E 2 Concentrations of extracellular vesicles (EVs) in plasma from COVID-19 patients and controls.T A B L E 3 T A B L E 4 Concentrations of extracellular vesicles (EVs) in CSF of patients with COVID-19.