Extracellular Vesicle Size Reveals Cargo Specific to Coagulation and Inflammation in Pediatric and Adult Sickle Cell Disease

Abstract

Aberrant coagulation in sickle cell disease (SCD) is linked to extracellular vesicle (EV) exposure. However, there is no consensus on the contributions of small EVs (SEVs) and large EVs (LEVs) toward underlying coagulopathy or on their molecular cargo. The present observational study compared the thrombin potential of SEVs and LEVs isolated from the plasma of stable pediatric and adult SCD patients. Further, EV lipid and protein contents were analyzed to define markers consistent with activation of thrombin and markers of underlying coagulopathy. Results suggested that LEVs—but not SEVs—from pediatrics and adults similarly enhanced phosphatidylserine (PS)-dependent thrombin generation, and cell membrane procoagulant PS (18:0;20:4 and 18:0;18:1) were the most abundant lipids found in LEVs. Further, LEVs showed activated coagulation in protein pathway analyses, while SEVs demonstrated high levels of cholesterol esters and a protein pathway analysis that identified complement factors and inflammation. We suggest that thrombin potential of EVs from both stable pediatric and adult SCD patients is similarly dependent on size and show lipid and protein contents that identify underlying markers of coagulation and inflammation.

Introduction

Homozygous sickle cell disease (SCD) occurs because of a single point mutation in both HBB genes leading to the production of sickle hemoglobin (HbS). HbS polymerizes under progressive deoxygenation causing both reversible and irreversible sickling of red blood cells (RBCs). This process leads to both erythrophagocytosis in multiple tissue compartments and RBC lysis within the circulatory compartment. ¹ Intravascular hemolysis causes exposure to cell-free hemoglobin (Hb) and arginase, leading to an increase in nitric oxide (NO) scavenging and reduction in NO synthesis, respectively. ² The combination of hemolysis, cell-free Hb, and NO depletion is a key contributor to vasculopathy and coagulopathy in SCD patients. ³ In addition, heme partitioning from Hb to pattern recognition receptors (e.g., TLR4) together with persistent endothelial activation and expression of NF-κB, ICAM-1, VCAM-1, E-selectin, and VWF also establish heme as a downstream Hb degradation activator of inflammation and vascular stasis. ⁴ Aberrant coagulation is a disease characteristic that favors hemostasis in both steady-state and crisis patients. ⁵ To date, experimental data from murine and human studies support a concept that microvascular and macrovascular thrombosis is consistent with induced thrombin generation (TG),^6,7 immune thrombosis (neutrophil extracellular traps), and platelet activation.^5,8,9

Extracellular vesicles (EVs) are heterogeneous membrane structures that contain cargo to facilitate normal physiological homeostasis, but also participate in processes that exacerbate pathophysiology. ¹⁰ EVs are classified into 3 main groups by size (diameter): exosomes (40-150 nm), microvesicles/microparticles (100-1000 nm), and apoptotic bodies (1000-3000 nm). ¹¹ In SCD, procoagulant EVs are shed from activated platelets and expose phosphatidylserine (PS).^12,13 PS is a key lipid exposed on EVs shed from sickled RBCs, ¹⁴ vascular endothelium, ¹⁵ and immune cells. ¹⁵ Further, tissue factor (TF) is an upregulated transmembrane glycoprotein observed across multiple cells and on the surface of EVs in murine ¹⁶ and human SCD. ¹⁷ Both PS and TF on EVs can activate thrombin, facilitating its binding to platelet receptors (PAR-1, 3, and 4) and cleavage of fibrinogen to fibrin on activated platelets. ¹⁸

Coagulopathy is observed in both pediatric and adult steady-state SCD patients, increasing their risk for thrombosis^19,20; however, coagulation testing in children remains a challenge and plasma coagulation factors are known to differ based on the age of healthy children compared to adults. ²¹ Moreover, the current literature does not establish a definitive coagulation signature in the pediatric or adult SCD patient populations. Several studies do suggest that TG, TATs (Thrombin-antithrombin complexes) , and d-dimers may be a marker of general coagulation risk, ⁷ cerebral vascular disease,^22,23 and risk of vaso-occlusive episodes ²⁴ in adults and children. Cell-derived EVs within the plasma compartment of SCD patients are easily isolated and potentially contain important pathophysiological markers that identify similarities and differences between pediatrics and adults.

We hypothesized that differences in pediatric and adult EV populations can demonstrate cargo-specific hemostatic activity which can be revealed by TG assay parameters. Moreover, EV populations isolated from pediatric and adult plasmas would demonstrate unique multiomics signatures. To test these hypotheses, we obtained blood from pediatric and adult steady-state SCD patients treated within the same academic medical center. Platelet-poor plasma was processed to separate and characterize small EVs (SEVs, 40-150 nm) and large EVs (LEVs, 100-1000 nm) also known as exosomes and microparticles, respectively. ²⁵ RBCs were collected to evaluate the distribution of Hb variants. We then tested the TG of SEVs and LEVs alone and after blocking vesicular PS. The lipidomic and proteomic signatures of SEVs and LEVs from patient plasmas were evaluated to determine the differences in vesicular composition by defining the top lipids and proteins and to develop pathway analyses that support observations from our TG assay. Further, lipid and protein cargo signatures defined here allow for advancing an understanding of EV markers specific to pediatric and adult patients.

Material and Methods

Patients

This observational study was approved by the University of Colorado Anschutz Medical Center Institutional Review Board (Inflammation and cellular function in sickle cell disease, protocol number: 20-0505). All blood samples for this study were obtained from consenting SCD patients (pediatrics n = 18; adults n = 12) at the time of routine clinical visits to the Colorado Sickle Cell Treatment and Research Center over a 1-year period. The study size was dependent on patients who consented to blood sampling for research-based analysis with a 1-year period. For this observational study, all patient data was obtained from chart-based review.

Sample Collection and Handling

All blood samples (4 mL) were collected from SCD patients into sodium citrate (0.109 m, 3.2%) Vacutainer blood collection tubes (Becton Dickenson, Franklin Lakes, NJ, USA). Samples were processed as described previously ²⁶ and detailed information is provided in the Supplemental Material and Methods.

Small and Large Extracellular Vesicle Isolation

SEVs and LEVs were isolated as described previously^18,26 and detailed information is provided in the Supplemental Material and Methods.

Dynamic Light Scattering

The hydrodynamic size of isolated SEVs and LEVs was determined by the dynamic light scattering (DLS) principle using a Zetasizer (Nano ZS, Malvern Instruments, Malvern, UK).

Transmission Electron Microscopy

Isolated SEVs and LEVs from SCD patient samples were evaluated by transmission electron microscopy (TEM) to visualize their size and morphology as described previously. ²⁶

Flow Cytometric Characterization of SEVs and LEVs

Flow cytometry analysis was performed to classify the cellular origin of purified SEVs and LEVs derived from platelets, RBCs, or endothelial cells, based on their surface markers. Detailed information is provided in the supplementary material. The list of antibodies used in the flow cytometry analysis and TG assays is included in the Supplemental Material and Methods.

Thrombin Generation

TG was performed based on the assay originally developed by Hemker et al ²⁷ with modifications. ²⁸ Detailed information is provided in the Supplemental Material and Methods.

Tissue Factor Activity

TF activity was measured using a human tissue factor activity assay kit (ab108906, Abcam, Cambridge, UK) with slight modifications to the manufacturer protocol as described previously. ²⁶

Lipidomics and Proteomics

High-throughput lipidomic and proteomic analysis was performed on SEVs, and LEVs isolated from SCD patient plasma. Detailed information about the sample processing is provided in the Supplemental Material and Methods.

Statistical Analysis

All statistical analyses and graphing of data were performed using GraphPad Prism software (version 9.2.0). Patient clinical and demographic data are represented as the group median and intraquartile range (IQR). For SEV and LEV concentration, the Mann–Whitney analysis for comparison within groups was used. For TG and comparisons of RBC Hb composition data, comparisons across groups were analyzed with either a Student t test or a 1-way ANOVA and Tukey's multiple comparison test, as appropriate. For all comparisons, the familywise α = .05. Data for lipidomic and proteomic data analysis was performed using Maven (1.4.20-dev-772), and quality controls were maintained as described. ²⁹ LipidSearch (4.2.27) and Compound Discoverer (3.1.0.305) in tandem performed untargeted data analysis. Heat maps and correlation data were generated by MetaboAnalyst (5.0). ³⁰ Graphs were produced using GraphPad Prism (9.2.0).

Results

General and Clinical Characteristics of Study Subjects

We acquired plasma and RBCs from 30 patients with SCD. Subjects were divided by age into adult (n = 12, median age 35) and pediatric (n = 18, median age 13) groups. In this study, any patients from birth to 20 years of age were considered pediatrics, while patients greater than 21 were considered adults. The demographics, comorbidities, and SCD-associated medications of these groups are described in Table 1. Each patient may have been on more than one drug therapy at the time of blood collection. Further, blood samples were collected prior to transfusion (regular exchange or simple). Clinical characteristics, including patient hematology and chemistry labs, are presented in Table 2, alongside adult and pediatric reference ranges.

Table 1.

Baseline Patient Characteristics.

	Adult (n = 12)	Pediatric (n = 18)
Demographics
Age (years)	35 (9)	13 (7)
Sex
Male	3 (25%)	11 (51%)
Female	9 (75%)	7 (39%)
Ethnicity a
Hispanic	0 (0%)	2 (11%)
Non-Hispanic	11 (100%)	16 (89%)
Comorbidities
Asthma	4 (33%)	8 (44%)
Pulmonary hypertension	3 (25%)	0 (0%)
Chronic kidney disease	3 (25%)	2 (11%)
Avascular necrosis	3 (25%)	0 (0%)
Stroke	4 (33%)	2 (11%)
Hypertension	1 (8%)	0 (0%)
Ulcers	1 (8%)	0 (0%)
Diabetes	0 (0%)	0 (0%)
Medications
Hydroxyurea	5 (42%)	12 (67%)
Voxelotor	0 (0%)	0 (0%)
Crizanlizumab	1 (8%)	0 (0%)
l-Glutamine	0 (0%)	1 (6%)
Iron chelator	0 (0%)	5 (28%)
Aspirin	1 (8%)	3 (17%)
Erythropoietin	1 (8%)	0 (0%)

Data are median (IQR) or n (%).

^a Adult n = 11.

Table 2.

Clinical Characteristics.

Patient labs	Adult (n = 12)	Adult reference range	Pediatric (n = 18)	Pediatric reference range
Hematology
RBCs (106/µL)	3.1 (1.1)	M: 4.0-5.7 F: 3.9-5.1	3.1 (0.6)	2-6 y.o.: 3.9-5.3 6-12 y.o.: 4.0-5.2 M (12-18 y.o.): 4.5-5.3 F (12-18 y.o.): 4.1-5.1
WBCs (103/µL)	7.9 (7.6)	4.5-11.0	9.5 (6.2)	2-4 y.o.: 6.0-17.0 4-6 y.o.: 5.5-15.5 6-8 y.o.: 5.0-14.5 8-14 y.o.: 4.5-13.5 14-18 y.o.: 4.5-13.0
Absolute neutrophil count (103/µL)	4.8 (5.5)	1.7-7.3	5.2 (2.6)	1-6 y.o.: 1.5-8.5 6-10 y.o.: 1.5-8.0 10-18 y.o.: 1.8-7.3
Monocytes (103/µL)	0.8 (1.1)	0.1-1.7	0.9 (1.3)	2-4 y.o.: 0.1-1.0 >4 y.o.: 0.0-0.7
Absolute lymphocyte count (103/µL)	2.0 (1.4)	1.3-3.5	3.3 (2.0)	2-4 y.o.: 3.0-9.5 4-6 y.o.: 2.0-8.0 6-8 y.o.: 1.5-7.0 8-10 y.o.: 1.5-6.8 10-12 y.o.: 1.5-6.5 12-14 y.o.: 1.2-6.0 >14 y.o.: 1.3-1.5
Platelets (103/µL)	423 (235)	153-367	386 (282)	153-367
Hemoglobin (g/dL)	9.1 (1.9)	M: 12.6-17.4 F: 11.9-15.7	9.5 (1.8)	2-6 y.o.: 11.5-13.5 6-12 y.o.: 11.5-15.5 M (12-18 y.o.): 13.0-16.0 F (12-18 y.o.): 12.0-16.0
Hematocrit (%)	27.1 (6.2)	M: 37-50 F: 35-45	27.3 (5.6)	2-6 y.o.: 34-40 6-12 y.o.: 35-45 M (12-18 y.o.): 37-49 F (12-18 y.o.): 36-46
Mean corpuscular volume (fL)	88.3 (14.5)	80-96	89.5 (13.3)	2-6 y.o.: 73-87 6-12 y.o.: 77-95 M (12-18 y.o.): 78-98 F (12-18 y.o.): 78-102
Chemistry
Ferritin (ng/mL)	185 (756)	M: 18-464 F (0-50 y.o.): 6.2-137 F (>50 y.o.): 11.1-264	226 (993)	M: 18-464 F: 6.2-137
Iron (µg/dL) a	94 (29)	M: 49-181 F: 37-170	122 (103)	M: 49-181 F: 37-170
Total iron-binding capacity (µg/dL) b	334 (160)	M: 257-470 F: 274-546	335 (107)	M: 257-470 F: 274-546
Transferrin saturation (%)	25 (20)	20-50%	36 (23)	20-50%
Creatinine (mg/dL)	0.58 (0.33)	M: 0.66-1.25 F: 0.52-1.04	0.33 (0.18)	1-12 y.o.: 0.22-0.63 12-18 y.o.: 0.42-0.92

All data are median (IQR).

Abbreviation: F, Female; M, Male; Y.O, Years old.

^a Adult n = 10; pediatric n = 15.

^b Adult n = 10; pediatric n = 14.

Hemoglobin HPLC

We determined the composition of Hb in SCD patients using HPLC (Figure S1A and B). Patients showed a range of phenotypes, with differential proportions of fetal, adult, and sickled Hbs (HbF, HbA, and HbS, respectively). Patients treated with hydroxyurea had a significantly higher proportion of HbF than those who did not (P = 0.0012, Figure S1C). Those who were transfused had an HbA fraction ranging from 7.1% to 77% of total Hb (Figure S1D). Patients who received exchange transfusions had a significantly higher HbA fraction than those who received simple transfusions (P = 0.0307) (Figure S1E).

Characterization of SEVs and LEVs

We isolated SEVs and LEVs from the plasma of pediatric and adult SCD patient plasma and characterized them simultaneously. We determined the particle sizes of SEVs and LEVs from both the groups. The average pediatric SEV and LEV sizes were 149.8 ± 33.5 nm and 256.8 ± 55.2 nm, respectively. The average adult SEV and LEV sizes were 174.2 ± 44.7 nm and 403.0 ± 118.8 nm, respectively (Figure 1A). The particle concentrations of SEVs and LEVs from both the groups were determined (Figure 1B and C). The morphologic differences of SEVs and LEVs from both pediatrics and adults were analyzed by TEM imaging (Figure 1D and E) and the result was consistent with the size range obtained from particle sizing experiments.

Figure 1.

Characterization of EVs. (A) Size comparison between pediatric and adult sickle cell patient SEVs (n = 7) and LEVs (n = 8). (B and C) Graphical representation of SEV and LEV particle concentration in pediatric (SEV/LEV) and adult samples (SEV/LEV). (D) TEM images of pediatric patient SEVs and LEVs (representative image of 1 sample). (E) TEM images of adult patient SEVs and LEVs (representative image of 1 sample). n = 5 each of pediatric and adult SEVs and LEVs was used for TEM analysis. (F and G) Flow cytometric analysis of EV-specific tetraspanin markers (CD63 and CD81) of pediatric and adult SEVs/LEVs. (A) Data points indicate individual measurements, and P values are from the 1-way ANOVA analysis with Tukey's multiple comparison between groups. (B and C) Data points indicate individual measurements, and P values are from the Mann–Whitney analysis for comparison within groups. ns, P > .05; ***P ≤ .001. EDP, EV-depleted plasma; LEV, Large extracellular vesicles; PS, Phosphatidylserine; SEV, Small extracellular vesicles; TG, Thrombin generation.

Flow Cytometric Analysis of SEV and LEV Surface Antigen Expression

We next characterized pediatric and adult SEVs and LEVs using flow cytometric analysis for EV-specific markers to confirm measurements made using particle sizing techniques. We used a flow cytometry submicron particle size reference kit (0.1, 0.2, and 0.5 µm) to create the flow cytometry gate for SEV and LEV particles, which was then applied to all samples as shown in Figure S2A. Purity of EV-depleted plasma (EDP) was determined sing the same gate as compared with pooled normal plasma (PNP) (Figure S2B). Isotype controls along with unstained and single staining of EVs for CD63 and CD 81 are shown in Figure S3.

SEVs and LEVs from both pediatric and adults were identified by the exosome tetraspanin surface markers CD63 and CD81, and their distributions were normalized to their mode (Figure 1F and G). These markers were differentially expressed on SEVs and LEVs from both pediatric and adult patients. Higher fluorescence intensity of LEVs as compared to the corresponding SEVs was observed as a clear positive shift of the CD63 histogram, but not the CD81 histogram as shown in Figure 1F and G.

Further, the histograms of cell-specific markers CD41, CD235a, and CD31 demonstrated a visible positive shift of LEVs from SEVs isolated from both pediatrics and adults (Figure S4A-C and E-G). Mean fluorescence intensity (MFI) of CD63 was significantly higher in pediatric and adult LEVs compared to respective SEVs (pediatric: P < 0.0001; adult: P < 0.0001) (Figure S4I). The MFI of CD81 was significantly higher in adult LEVs compared to SEVs (P = 0.0027) but not in pediatric LEVs (P = 0.0559) (Figure S4J). In all the cases, the MFI of CD41 (pediatric: P = 0.0015; adult: P = 0.0155), CD235a (pediatric: P = 0.0098; adult: P = 0.0123), and CD31 (pediatric: P = 0.0045; adult: P = 0.0012) indicated significantly higher expression of LEVs than the corresponding SEVs from both pediatric and adult patients (Figure S4K-M).

At the same time, there was no noticeable difference in the fluorescence intensity (EV markers and cell-specific markers) between pediatric and adult plasma isolated SEVs or LEVs (Figure S4I-M).

Flow cytometric analysis of annexin V suggests that PS is enriched in LEVs compared to SEVs from both pediatric and adult patients (Figure S4D and H). The MFI of annexin V demonstrated that pediatric LEVs (P = 0.0002) and adult LEVs (P < 0.0001) exposed significantly higher PS compared to SEVs (Figure S4N).

Pediatric and Adult LEVs Demonstrate Increased Thrombin Generation

Studies have shown that EV concentrations between 40 and 400 µg/mL induce TG.^18,26 SEVs, and LEVs from SCD pediatric and adult patients were resuspended in EDP at 2 concentrations (50 and 200 µg/mL). SEV- and LEV-spiked EDP was then evaluated in our TG assay (Figure 2A-D and E-H). The mean value TG curves at 50 and 200 µg/mL are shown in Figure 2A and E, respectively. The thrombin curves of SEVs and LEVs from pediatric and adult patients did not visually differ at either of the concentrations tested (Figure 2A and E), but there were observed differences between SEVs and corresponding LEVs within the adult and pediatric groups (Figure 2A and E). The peak height, peak velocity, and endogenous thrombin potential (area under the curve [AUC] of the TG curve) were calculated as the main parameter of the TG curves as previously reported.^31,32 There were no significant differences between pediatric and adult SEVs or pediatric and adult LEVs in peak height, peak velocity, or AUC values at either concentration (Figure 2B, C, F, and G). However, comparison of thrombin peak heights, peak velocities, and AUC values between SEVs and LEV—irrespective patient age—were significantly different (P < 0.0001) at both 50 and 200 µg/mL concentrations (Figure 2A-C and E-G). AUC values also showed a significant difference between SEV and LEV (P < 0.0001) in both groups and at both the concentrations (Figure 2D and H).

Figure 2.

Thrombin generation induced by pediatric and adult SEVs and LEVs: (A) 50 µg/mL and (E) 200 µg/mL—mean TG curves for SEVs and LEVs. (B-D) Peak height, peak velocity, and AUC values of pediatric and adult SCD patient SEVs and LEVs at 50 µg/mL. (F-H) Peak height, peak velocity, and AUC values of pediatric and adult SCD patient SEVs and LEVs at 200 µg/mL. 50 and 200 µg/mL: pediatric SEVs (n = 16) and LEVs (n = 17), and adult SEVs (n = 12) and LEVs (n = 12). Data points indicate individual measurements, and P values are from the one-way ANOVA analysis with Tukey's multiple comparison between groups. ns, P > .05; ****P ≤ .0001.

LEV but Not SEV Phosphatidylserine Drives Thrombin Generation

To assess the contribution of PS to TG, we performed blocking experiments with annexin V. The TG curves for EDP, pediatric and adult SEVs, and LEVs at specified annexin V concentrations (0.1, 0.5, and 2.5 µg/mL) are shown in Figure 3A, B, E, and F. Preincubation of SEVs and LEVs with annexin V inhibited TG in a concentration-dependent manner, with highest concentration (2.5 µg/mL) abolishing TG altogether (Figure 3C, D, G, and H). The concentration-dependent effect of annexin V on TG was significantly higher in pediatric and adult LEVs, compared to SEVs. Interestingly, this effect was greater in adult LEVs, when compared to EDP (peak height: 0.1 and 0.5 µg/mL annexin V: P < 0.0001 and P = 0.0020; peak velocity: 0.1 µg/mL annexin V: P = 0.0006) (Figure 3G and H). AUC comparisons between EDP and pediatric/adult LEVs were significant (P < 0.0001 for both; Figure 3A, B, E, and F). We tested the activity of TF in both SEVs and LEVs; however, we did not observe differences (Figure S5).

Figure 3.

Effect of annexin V on SEV/LEV-induced thrombin generation: mean TG curves for EDP, pediatric (A) and adult (B) SEVs, and pediatric (E) and adult (F) LEVs with different annexin V concentrations (0.1, 0.5, and 2.5 µg/mL). Peak height and peak velocity values of EDP, pediatric and adult SEVs (C and D) and LEVs (G and H) with different annexin V concentrations. EDP (n = 4), SEVs (n = 5), and LEVs (n = 5). ns, P > .05; *P ≤ .05; **P ≤ .01; ***P ≤ 0.001; ****P ≤ 0.0001. EDP, EV-depleted plasma; LEV, Large extracellular vesicles; PS, Phosphatidylserine; SEV, Small extracellular vesicles; TG, Thrombin generation.

Multiomics Analysis of SEVs and LEVs Reveals a Size Dependent Procoagulant and Proinflammatory Signature

The top 25 lipids in SEVs and LEVs isolated from SCD pediatric and adult patient plasmas are grouped based on their classification (Figure 4A and B). The primary lipids observed in EVs were diacylglycerols, phosphatidylcholines, cholesteryl esters (ChE), phosphatidylethanolamines, lysophosphatidylcholines, triacylglycerols, and PS. Here, ChE was the most significantly enriched lipid class in pediatric and adult SEVs compared to LEVs (P < 0.05 and P < 0.001, respectively) (Figure 4C and D). Conversely, both pediatric and adult LEVs were highly enriched for PS compared to SEVs (P < 0.001 and P < 0.05, respectively) (Figure 4E and F). A comparison of PS isomers is shown for isolated SEVs and LEVs from pediatric and adult plasmas (Figure 4E and F). Both PS (18:0;20.4) and PS (18:0;18:1) were the most highly accumulated isomers in LEVs irrespective of patient age, in keeping with the overrepresentation of stearoyl, oleyl, and arachidonyl-conjugated lipids in the total lipidome. Collectively, these data support results from our TG assay before and after annexin V blocking, suggesting similarity in the procoagulant potential of EVs of stable pediatric and adult patient plasmas.

Figure 4.

Lipidomics from SEV and LEV isolated from pediatric or adult populations: heat map analysis of metabolites by MetaboAnalyst of SEVs and LEVs isolated from pediatric (A) or adult (B) populations. Lipids compared by specific lipid type in SEVs and LEVs isolated from pediatric (C) or adult (D) populations. The top phosphatidylserines in SEVs and LEVs isolated from pediatric (E) and adult (F) patient plasmas. *P ≤ .05; **P ≤ .01; ****P ≤ .0001. EDP, EV-depleted plasma; LEV, Large extracellular vesicles; PS, Phosphatidylserine; SEV, Small extracellular vesicles; TG, Thrombin generation.

The proteomic data comparing SEVs to LEVs in both pediatrics and adults highlight data sets for size differentiation and markers of inflammation and coagulation (Figure 5A-C). Data are shown as heat maps for the top 25 proteins in SEVs and LEVs isolated from the plasmas of pediatric and adult SCD patients (Figure 5A). SEVs from both patient groups demonstrated enrichment of proteins consistent with an increase in the innate immune response, especially within the complement pathway (eg, C1QB, C1QA, C3, and C5) also shown in the pathway analysis for SEVs (Figure 5B). The pathway analysis for LEVs from both patient groups suggests an interconnection between complement activation and the coagulation system (eg, C1s, RBP4, and ORM1), microvascular coagulation (eg, RAB35), and thrombin-mediated platelet activation (eg, RAB6A) as shown in Figure 5C. Consistent with our lipidomic analysis, LEV protein pathways support a concept of coagulation risk that is similar in both steady-state pediatric and adult patients.

Figure 5.

Proteomics from SEVs and LEVs isolated from pediatric or adult populations. (A) Heat map analysis of metabolites by MetaboAnalyst of SEVs and LEVs isolated from pediatric or adult populations. (B) Pathway analysis by –omics.net of SEVs isolated from pediatric and adult populations. The legend emphasizes the significant pathways of response to wound healing (P = 2.86 × 10⁻¹³), innate inflammatory response (P = 7.2 × 10⁻¹²), and acute inflammatory responses (P = 1.3 × 10⁻¹¹). (C) Pathway analysis by –omics.net of LEVs isolated from pediatric and adult populations. The legend emphasizes the significant pathways of response to blood coagulation (P = 3.56 × 10⁻¹³), coagulation (P = 7.2 × 10⁻¹¹), and wound healing (P = 9.8 × 10⁻¹⁰). ORM1, Orosomucoid 1; GC, GC vitamin D-binding protein; A1BG, alpha-1-B glycoprotein; ITIH4, inter-alpha-trypsin inhibitor heavy chain 4; RBP4, retinol binding protein 4; C2, complement 2; HNF4A, hepatocyte nuclear factor 4-alpha; KNG1, Kininogen 1; ORM2, Orosomucoid 2; F10, coagulation factor F; APOA2, apolipoprotein A2; APOD, apolipoprotein D; AZGP1, alpha-2-glycoprotein, zinc-binding; UBC, ubiquitin C; PIGR, polymeric immunoglobulin receptor; PITX3, pituitary homeobox 3; C1QB, complement C1q beta chain; SERPINA3, serpin family A member 3; C1QA, complement C1q beta chain; EFEMP1, endothelial growth factor containing fibulin extracellular matrix protein 1; LASP1, LIM and SH3 domain protein 1; NRAS, N-Ras; C1s, complement C1s; ELMO1, engulfment and cell motility 1; NSF, N-ethylmaleimide-sensitive factor; CUL3, cullin-3; RAB35, RAS-related protein 35; PDCD6IP, programed cell death 6 interacting protein; GRB2, growth factor receptor-bound protein 2; VCAM1, vascular cell adhesion molecule 1; FN1, fibronectin 1; SERPINF1, serpin family F member 1; ITGA4, integrin subunit alpha 4; AP2B1, adaptor related protein complex 2 subunit beta; EIF2S1, eukaryotic translation initiation factor 2 subunit alpha; EDP, EV-depleted plasma; LEV, Large extracellular vesicles; PS, Phosphatidylserine; SEV, Small extracellular vesicles; TG, Thrombin generation.

Discussion

Pediatric ³³ and adult ³⁴ SCD patient EVs have been studied independently, but not directly compared for TG and multiomics analysis. Here, we compared stable pediatric and adult patients within the same study using these techniques. All patients were being actively managed for their disease with indicated drug therapies such as hydroxyurea, as well as simple and exchange transfusions; therefore, the totality of the disease and treatment process was captured in this study. Characterization of EVs revealed that pediatric and adult patient plasma SEVs were of similar size and distribution. Conversely, LEVs from adults demonstrated significantly larger sizes than those isolated from pediatric patient plasmas, suggesting potential for functional differences. Our TG experiments showed that LEVs isolated patient groups similarly increased thrombin peak heights, velocities, and AUCs compared to their respective SEVs. No differences were observed in TG parameters between pediatric and adult SEVs or between pediatric and adult LEVs. Because PS surface expression is a critical component of injured RBCs and activated platelets, we further pursued the role of PS in our study.

Anionic phospholipids, especially PS, are an important component of EV membranes that trigger TG. ³⁵ PS exposure on platelet, erythrocyte, and endothelium-derived microparticles are well described.^36,37 Here, annexin V blocking showed a concentration-dependent reduction in thrombin peak heights, velocities, and AUCs specific to both pediatric and adult plasma LEV isolates. Flow cytometric analysis also supported this data, but more concrete evidence came from metabolomic analysis of EV isolates. Both pediatric and adult EVs revealed that the top 25 metabolites were lipids that grouped into cholesterol esters, phosphatidylcholines, diglycerides, and triglycerides. Interestingly, cholesterol esters were highest in both pediatric and adult SEVs and this observation was consistent with exosome biogenesis. ³⁸ Conversely, PS was enriched in both pediatric and adult LEVs and we observed significantly increased PS (18:0;20:4) and PS (18:0;18:1). During the process of LEV production, PS (18:0;20:4) and PS (18:0;18:1) translocate from the inner to the outer leaflet of the plasma membrane, predominantly through the action of scramblases, ³⁹ resulting in a circulating pool of PS-enriched vesicles. LEVs with high PS content that originate from RBCs and platelets are reported to increase TG without the need for TF, ⁴⁰ which may suggest a rationale for our TF activity data showing similar results when applied to SEVs and LEVs.

As with the lipidomic data, proteomics revealed that the top 25 proteins were strikingly similar for pediatric and adult EVs. Further protein pathway analysis showed a specific signature for SEVs and LEVs that favored inflammation and coagulation, respectively. SEVs from both pediatric and adult patient groups demonstrated enrichment of proteins consistent with an increase in the innate immune response, especially within the complement pathway (eg, C1QB, C1QA, C3, and C5). Complement is activated in SCD patients ⁴¹ and murine studies demonstrate the importance of complement markers C4d and C5a as mediators of endothelial activation and occlusive crisis.^42,43 Pulmonary hypertension (PH) is one of the most significant vasculopathies of SCD and is diagnosed in 6% to 10% of patients, ⁴⁴ and 4 (13%) of patients included in this study were diagnosed with progressive PH. It is notable that data are emerging on SCD lung injury and proinflammatory circulating LEVs. ⁴⁵ Further, studies in a calf model of PH also show that C3-rich SEVs are released by pulmonary vascular fibroblasts and polarize macrophages to a proinflammatory phenotype. This finding offers a novel mechanism that drives vascular remodeling in PH. ²⁹ SCD patients with genetic anemias often receive regular or intermittent transfusion to avert end organ injury ⁴⁶ ; however, EVs are also known to accumulate during blood storage and activate complement post transfusion. ⁴⁷ In our study, a significant number of patients received regular exchange or simple transfusions to prevent end organ injury. As a result, sequelae of SCD and common treatments to achieve steady-state disease are likely to impact EV cargo and their accumulation in the circulatory compartment; however, it is unknown how exogenous sources of EV exposure impact SCD disease.

Although we identify PS as the likely cause for TG by LEVs, the protein pathway analysis for LEVs from both patient groups suggested an interconnection between complement and the coagulation system. Identification of C1s, RBP4, and ORM1 in this analysis reinforced this signature in LEVs. ⁴⁸ We identified RAB35 in LEVs, which is a critical regulator of endothelial activation, exocytosis of Weibel–Palade bodies, and VWF-initiated microvascular coagulation. ⁴⁹ Further, the presence of RAB6 suggests a process of thrombin activation of platelets consistent with modulation of vesicular trafficking. ⁵⁰

The present study is limited by the number of patients available per group and as such we were not able to classify patients by drug treatment (eg, hydroxyurea) or transfusion status; however, all drug therapies, transfusions, and RBC Hb variants were identified and are presented. Most importantly, this study is not able to define the SEV and LEV populations or their cellular origin as specific to the underlying process of SCD. However, this study does describe SEVs and LEVs based on the totality of disease and treatment. Another limitation of this study is the protocol used for LEV purification; due to volume constraint, we are unable to purify LEVs using other techniques. LEVs isolated by centrifugation from healthy donor plasma showed increased thrombogenicity ¹⁸ and that impurities coisolated with LEVs do not alter the thrombogenicity in a significant manner. ⁵¹ Different techniques are available for EV purification^11,52 but either ultracentrifugation or in combination with other techniques is still the most used EV separation method. ⁵³

In conclusion, our data identified similar function and cargo compositions of pediatric and adult SEVs that were distinctly different than LEVs from both patient groups. LEVs isolated from stable SCD patient plasmas increased thrombin in a similar manner for both pediatric and adult patients. Further, PS was identified as a primary contributor to LEV-driven TG in both patient groups. This observation was supported by both lipidomic analysis and protein pathway analysis. Unique to SEVs in our study was a clear increase in complement components that suggest a qualitative interconnection between inflammation and coagulation identifiers. Further studies may reveal that SEVs and LEVs from stable pediatric and adult SCD patients serve as a useful marker of inflammation and aberrant coagulation.