Lepirudin (Refludan^®) was obtained from Celgene (Uxbridge, United Kingdom). GPRP peptide (Pefabloc^® FG or Pefa-6003) was purchased from Pentapharm (Basel, Switzerland). Trisodium citrate was isolated from BD Vacutainer citrate blood sampling tubes (BD, Plymouth, United Kingdom). Heat-inactivated Escherichia coli (E. coli, strain LE392, catalog no. ATCC 33572) was from American Type Culture Collection (ATCC, Manassas, VA). Thrombin-antithrombin (TAT) ELISA kit (catalog no. OWMG15) was purchased from Siemens Healthcare Diagnostics Products GmbH (Marburg, Germany) while β-thromboglobulin (βTG) ELISA kit (catalog no. 00950) was obtained from Diagnostica Stago, Inc. (Parsippany, NJ).

Blood from healthy human donors was obtained by forearm venipuncture using a 21-gauge needle. Informed written consent was obtained from all blood donors. The whole blood (4.5 ml) was collected in polypropylene tubes containing 0.5 ml of either one of the following solutions: i) 0.7% (w/v) citrate, ii) 3.2% (w/v) citrate, iii) the thrombin inhibitor lepirudin (50 μg/ml, final concentration) supplemented with 0.7% (w/v) citrate, or iv) GPRP (8 mg/ml, final concentration) supplemented with 0.7% (w/v) citrate. The volume ratio of blood to anticoagulant was kept at 9:1. An aliquot (100 μl) of whole blood was used for blood cell counting immediately after collection using CELL-DYN Emerald 22 hematology analyzer (Abbott, IL). Experiments with the blood of each donor were performed in triplicates.

The anticoagulated whole blood was centrifugated at 180×g for 15 minutes at 25°C without the brake applied. The liquid phase in upper layer, i.e., the platelet-rich plasma (PRP) layer, was gently transferred to 1.5 ml conical polypropylene tubes without disturbing the buffy layer. Tempered (25°C) modified Dulbecco’s phosphate-buffered saline without Ca²⁺ and Mg²⁺ (PBS, Sigma-Aldrich, MO) was added to the residual blood to compensate for the volume of the PRP removed. An aliquot (100 μl) of the residual blood was taken for cell counting. The blood underwent three washing cycles, each by centrifugation at 180×g for 15 minutes followed by platelet removal, to further reduce the number of residual platelets. After each step of washing, the blood volume was regained by adding PBS. An aliquot (100 μl) of the remaining blood was taken for cell counting after each step.

The separated PRP was centrifugated at 2000×g for 15 minutes at 25°C; the 2/3 top portion of the centrifugated PRP (i.e., platelet-poor plasma) was transferred back to the residual platelet-depleted blood after the last step of washing. The platelet containing portion of the PRP, i.e., the 1/3 bottom, was isolated as the platelet suspension. All cell counts and mean platelet volume (MPV) were measured using CELL-DYN Emerald 22 hematology analyzer (Abbott).

For all steps in the platelet separation from whole blood, the blood was anticoagulated with 3.2% (w/v) citrate, otherwise with either lepirudin or GPRP supplemented with 0.7% (w/v) citrate. The temperature during the preparation was kept at 25°C. Before platelet and whole blood activation, the 0.7% (w/v) citrate was reversed by the addition of 6.25 mM CaCl₂ (Sigma-Aldrich) and 3.2% (w/v) citrate was reversed with 25 mM CaCl₂. The concentration of citrate, which was sufficient to dampen the hemostatic response in GPRP whole blood, was determined by adding citrate of concentrations from 0.2% (w/v) to 3.2% (w/v) to GPRP-whole blood and analysing the expression of CD62P and CD63 on platelets after 15 minutes of incubation. All incubations for functional tests of platelets were carried out at 37°C without or with the addition of E. coli (10⁷/ml). After incubation, 16 μl of a stop solution containing a mixture of a CTAD solution [0.08 M trisodium citrate, 11 M theophylline, 2.6 M adenosine, 0.14 M dipyridamole] and 0.14 M EDTA was added to every 100 μl of blood or plasma.

Platelet activation was quantified by the upregulation of their surface markers, CD63 and CD62P, by flow cytometry. Briefly, after separation from whole blood, the platelet suspension was kept at 25°C for 4 hours, followed by re-calcification and incubation with E. coli for 15 minutes at 37°C. Then, platelets were stained for 30 minutes in the dark at 4°C with an antibody mixture containing platelet CD42a-FITC (catalog no. 348083, BD Biosciences, San Jose, CA), CD63-PE-Cy7 (catalog no. 25-0639-42, Invitrogen, Carlsbad, CA), and CD62P-PE (catalog no. 555524, BD Biosciences). After staining, the solution was centrifuged at 250×g, 4°C for 5 minutes. The supernatant was discarded and the pellet was resuspended and fixed in PBS containing 0.1% paraformaldehyde (PFA) and 0.1% bovine serum albumin (PBSA) and stored at 4°C until analysis with flow cytometer Attune NxT Acoustic Focusing Cytometer (Thermo Fisher Scientific) within 24 hours from sampling. Platelets were gated as CD42a+ population ( Supplementary Figure 1 ) while CD63 and CD62P were used as platelet activation measures. Flow data analysis was performed with FlowJo software version 10 (Ashland, OR).

Thrombin activation was characterized by the quantification of TAT-complexes in plasma, and platelet activation was determined by the level of plasma βTG. Briefly, 200 μl of whole blood or platelet-depleted blood was re-calcified and incubated with or without E. coli (10⁷/ml). PBS was added to the control for volume adjustment. The blood was incubated in a rolling incubator at 37°C for 15 minutes before 32 μl of the stop solution was added. Plasma was collected after centrifugation at 3000×g, 4°C for 20 minutes, and stored at -80°C for further analyses.

Platelets in the platelet suspension were spun down at 250×g at room temperature for 15 minutes. Platelets were allowed to adhere onto Nunc™ Lab-Tek™ II CC2 chamber slide (Thermo Scientific™, catalog no. 154941). The samples then were fixed in 2% glutaraldehyde, dehydrated through an ethanol series and critical point drying (Polaron E3 Critical Point Drier, Polaron Equipment Ltd, United Kingdom). Next, the samples were sputter-coated with a 30 nm-thick layer of platinum in Polaron E5100 sputter coater before being viewed under a scanning electron microscope (SEM) GeminiSEM 300 (Carl Zeiss Microscopy GmbH, Germany).

Platelet-depleted blood (40 μl) was incubated with E. coli (10⁷/ml) in a rolling incubator at 37°C for 15 minutes. After adding 6.4 μl of the stop solution, the leukocytes were stained with antibodies: CD45-Pacific Orange (catalog no. MHCD4530, Invitrogen™), CD14-PerCP (catalog no. 340585, BD Biosciences), CD15-V450 (catalog no. 48-0158-42, Invitrogen™), CD11b-APC/Fire 750 (catalog no. 101262, BioLegend, San Diego, CA), and CD35-Alexa Fluor 647 (catalog no. 1981978, Invitrogen™) for 15 minutes at 4°C. After lysing the red blood cells with fixative-free lysis buffer (Invitrogen™, catalog no. HYL250), the solution was centrifuged at 250×g, 4°C for 5 minutes. After centrifugation, the supernatant was discarded and the pellet was resuspended in PBSA. The expression of the abovementioned markers was analysed with Attune NxT Acoustic Focusing Cytometer (Thermo Fisher Scientific). Granulocytes were gated as CD15+ population ( Supplementary Figure 2 ), while monocytes were gated as CD15- and CD14+ population ( Supplementary Figure 3 ). The activation of granulocytes and monocytes was evaluated by the expression levels of CD11b and CD35 on their surfaces. Flow data analysis was performed with FlowJo version 10.

GPRP-anticoagulated whole blood was immediately supplemented with monensin (GolgiStop^TM, BD Biosciences, San Jose, CA) at 2 µM final concentration and incubated with bacteria for 2 hours. Then, EDTA was added at a 10 mM final concentration. For each test, 25 µl (for monocyte and granulocyte analyses) or 100 µl (for platelet analyses) of whole blood was lysed and fixed utilising the Cytofix/Cytoperm kit containing GolgiStop^TM from BD Biosciences (San Jose, CA). Flow cytometry was performed on a Novocyte Flow Cytometer (ACEA Biosciences Inc., San Diego, CA, USA). Single cells were gated utilizing the forward scatter-area (FSC-A) versus the forward scatter-height (FSC-H) dot plot. Granulocytes were identified in a side scatter (SSC)/CD15 dot plot using anti-CD15 BV605 (BD Biosciences). Monocytes were identified in a SSC/CD14 dot plot from the CD15 negative population using anti-CD14 FITC (BD Biosciences). Platelets were analysed in a separate set-up and gated in a SSC/CD42a dot plot using anti-CD42a FITC (BD Biosciences). Here, monocytes were detected using an anti-CD14 V500 antibody (BD Biosciences). Expression of intracellular IL-8 was detected using anti-IL-8 BV421 (clone G265-8, BD Biosciences and expressed as MFI.

To study the role of platelets in the inflammatory response, whole blood was collected with GPRP supplemented with 0.7% (w/v) citrate. One part was kept as whole blood, and the other part was processed according to the protocol above, i.e., prepared as platelet suspension as one part and blood containing plasma and all cells except platelets as a second part. Half of the platelet-plasma suspension was re-calcified and incubated for 15 minutes at 37°C to activate the platelets. The activated platelet-plasma suspension was centrifuged for 15 minutes at 3000×g, 25°C to remove platelets by hard pellet formation. From the processed blood fractions, the following conditions were prepared: i) whole blood, ii) platelet-depleted blood, iii) platelet-depleted blood with the addition of platelet-plasma suspension, iv) platelet-depleted blood with the addition of plasma after platelet activation, and v) platelet suspension. All components were re-calcified and incubated with E. coli (10⁷/ml) in a rolling incubator at 37°C for 4 hours. Plasma was prepared by centrifugation at 3000×g, 4°C for 20 minutes and stored at -80°C until analysis with multiplex technology by MAGPIX Luminex (Bio-Rad Laboratories, Hercules, CA) according to the manufacturer’s protocols. A panel of ten inflammatory cytokines and chemokines (IL-1β, IL-1Ra, IL-6, IL-8, IFN-γ, IP-10, MCP-1, MIP-1α, MIP-1β, and TNF) were analysed.

Statistical comparisons were calculated using GraphPad Prism 9 (GraphPad Software, CA). Tests employed were paired Kruskal–Wallis one-way analysis of variance. A p-value <0.05 was considered statistically significant. All data are presented as mean ± standard deviation (SD).

This study was designed and performed according to the ethical guidelines from the declaration of Helsinki. Informed written consent was obtained from the blood donors. The study was approved by the ethical committee of the Norwegian Regional Health Authority, ethical permit REK#S-04114, 2010/934.

A low amount of citrate was used to initially stabilize platelet function and its concentration was titrated into GPRP-anticoagulated blood with platelet expression of CD63 and CD62P as the activation read-outs. The expression level of CD63 and CD62P initially decreased in response to the concentration of citrate supplemented ( Supplementary Figure 4 ), and 0.7% (w/v) citrate was the optimal concentration in supplementation in GPRP-anticoagulated blood for the expression lowest of CD63 and CD62P ( Supplementary Figure 4 ). Meanwhile, there was no difference in the intensities of CD63 and CD62P in lepirudin-anticoagulated blood without and with 0.7% (w/v) citrate supplemented ( Supplementary Figures 5A, B ). To re-calcify, 6.25 mM CaCl₂ was required to compensate the Ca²⁺ chelated by 0.7% (w/v) citrate, and 25 mM CaCl₂ for 3.2% (w/v) citrate.

Platelet counts and mean platelet volumes (MPV) in blood anticoagulated with 0.7% (w/v) citrate were similar to those in blood anticoagulated with 3.2% citrate (w/v) and either lepirudin or GPRP supplemented with 0.7% (w/v) citrate ( Supplementary Figures 5C, D ). However, blood anticoagulated with only 0.7% citrate clotted within 10 minutes after blood sampling; hence, data from this blood sample condition could not be further collected. Platelets in 0.7% (w/v) citrated blood transformed their morphologies into early dendritic shapes ( Supplementary Figures 5E, F ), signalling early stage of platelet activation (21). No blood clot was formed with 3.2% (w/v) citrate, a condition used in routine haematological laboratories worldwide and included as control in this study. Neither was any clot formed in 0.7% (w/v) citrate supplemented in either lepirudin or GPRP blood sampling conditions. We found that after one centrifugation, 71.1 ± 2.7% of platelets remained in 3.2% citrated blood after removing the PRP layer ( Figure 1A ). These numbers in lepirudin- and GPRP-anticoagulated blood were 84.9 ± 13.3% and 31.7 ± 29.4%, respectively ( Figure 1A ).

Three more washing steps were applied by repeated centrifugation and the number of platelets remaining in the residual blood reduced after each step ( Figure 1A ). At the end (4^th centrifugation) of the washing process, 6.1 ± 0.6% and 3.7 ± 1.8% of platelets remained in lepirudin- and GPRP-anticoagulated blood, respectively, while this number in 3.2% citrated blood was 7.0 ± 1.3%. Throughout the process, MPV remained at <10 fl throughout the successive washing steps ( Figure 1B ). In addition, the cell counts of other blood cells, i.e., white and red blood cells, remained mainly unchanged ( Supplementary Figure 6 ).

In addition to measuring MPV of platelets as an indicator of platelet activation ( Figure 1B ), we further characterized the expression of platelet activation markers on the platelet surface. Both CD63 and CD62P markers from dense and alpha granules, respectively, were detected at background levels immediately after blood collection and remained almost unchanged after preparing the PRP suspension until 4 hours after blood sampling ( Figures 2A, B ), implying minimal activation of the platelets (22). Equally important, platelets retained their functions in response to E. coli. Upon re-calcification, incubation with E. coli for 15 minutes at 37°C induced a substantial expression of both CD63 and CD62P on platelets in PRP suspension from both citrate 3.2% (w/v) and GPRP ( Figures 2A, B ). In contrast, platelets prepared from lepirudin-anticoagulated blood only expressed background levels of CD63 and CD62P when incubated with E. coli ( Figures 2A, B ).

Upon activation, platelets undergo morphological changes and enhance the expression of surface markers like CD63 and CD62P and release alpha granule-stored βTG (23). In 3.2% (w/v) citrate and in GPRP-anticoagulated blood, βTG was detected at ~1000 UI/ml after re-calcification and incubation at 37°C for 15 minutes ( Figure 2C ). After platelet depletion, the level of βTG in 3.2% (w/v) citrate and GPRP-anticoagulated blood dropped to 10 - 15% of the corresponding platelet-intact whole blood ( Figure 2C ). A virtually identical pattern was observed with TAT, which was detected at levels up to 6 µg/ml before platelet removal and decreased to 0.2 µg/ml after platelet depletion ( Figure 2D ). In contrast, βTG and TAT were detected at background levels in lepirudin-anticoagulated blood both before and after platelet depletion ( Figures 2C, D ). The activation of platelets in blood was confirmed under SEM ( Figures 2E–G ). Platelets isolated from 3.2% (w/v) citrate-, lepirudin-, and GPRP-anticoagulated blood were activated under thrombin stimulation, as evidenced by their wide-spreading ( Figures 2E–G ), indicating that platelets retained their functions during the separation process (21).

Granulocytes and monocytes play an important role in the acute inflammatory response (24, 25). We characterized the activation of these cells in blood before and after platelet separation. The expression level of the activation marker CD11b on both granulocytes and monocytes in all anticoagulants was low immediately after blood sampling and remained unchanged after the blood underwent successive washing steps ( Figures 3A, B ). To confirm that the protocol did not interfere with cell functions, the activation of these cells was evaluated in platelet-depleted blood. E. coli at 10⁷/ml induced a 5-fold increase in CD11b expression, which was comparable to the activation levels in whole blood incubated with E. coli after collection without centrifugation ( Figures 3A, B ).

Complement receptor 1 or CD35, with C3b as the main ligand, is expressed on the surface of both monocytes and granulocytes, and an up-regulation of CD35 is indicative of the activation status of these cells (26). Similar to CD11b, we found that the expression of CD35 was not affected by the platelet separation protocol and responded to E. coli incubation ( Figures 3C, D ).

Having shown that our protocol effectively removed platelets from whole blood ( Figure 1 ), while not activating platelets ( Figure 2 ), granulocytes and monocytes ( Figure 3 ), we further profiled the release of ten inflammatory cytokines and chemokines in platelet-depleted blood in response to E. coli, including IL-1β, IL-1Ra, IL-6, IL-8, IFN-γ, IP-10, MCP-1, MIP-1α, MIP-1β, and TNF. This panel was selected basing on its elevation in response to E. coli in whole blood after a 4-hour incubation. Remarkable reductions were observed in the release of seven cytokines in platelet-depleted blood as compared to whole blood, including IL-8 ( Figure 4A ), IL-1β, IL-1Ra, IL-6, IP-10, MIP-1α, and MIP-1β ( Supplementary Figures 7A–F ). Meanwhile, there was no apparent difference in the release of IFN-γ, MCP-1, and TNF before and after platelet removal ( Supplementary Figures 7G–J ).

IL-8 levels were elevated in GPRP-whole blood in response to E. coli and decreased to background level in platelet-depleted blood ( Figure 4A ). However, IL-8 levels in GPRP-anticoagulated blood were resurgent in platelet-depleted blood upon the addition of either the platelet-plasma suspension or only the supernatant from activated platelets ( Figure 4A ). This result indicated that the increased IL-8, which we previously characterized to be dependent on thrombin, was mediated via the activation of platelets.

We investigated the role of platelets in the upregulation of IL-8 by measuring plasma levels of IL-8 in PRP, in platelet-depleted blood, and in GPRP-whole blood upon E. coli incubation. Interestingly, plasma IL-8 was detected at 4.19 ng/ml in PRP prepared from GPRP anticoagulated blood ( Figure 4B ). Lepirudin was also included here as control and the IL-8-level was detected at 0.22 ng/ml in PRP from lepirudin-anticoagulated blood. Plasma IL-8 was also detected at 5.81 and 5.36 ng/ml in platelet-depleted blood anticoagulated with lepirudin and GPRP, respectively ( Figure 4B ). Notably, IL-8 level increased up to 26.36 ng/ml in whole blood anticoagulated with GPRP ( Figure 4B ).

The activation of platelets was confirmed by a significant increase of platelet-derived growth factor (PDGF) in PRP and whole blood prepared with GPRP ( Figure 4C ) (27). The levels of PDGF were not significantly different between PRP and whole blood prepared with GPRP ( Figure 4C ). In contrast, in platelet-removed blood, PDGF was detected at background levels and there was no difference in PDGF levels between lepirudin- and GPRP-blood ( Figure 4C ).

We further investigated which leukocyte population was the main source of IL-8. Intracellular cytokine staining showed that IL-8 was mainly generated by monocytes ( Figure 4D ). As such, monocytes alone attributed to an average IL-8 intensity of 3485 MFI, which was comparable to that of platelets (3502 MFI) ( Figure 4D ). However, the intensity of IL-8 increased over 17 folds, to 59690 MFI when monocytes were found in conjugation with platelets ( Figure 4D ). Meanwhile, IL-8 in granulocytes was detected at 6417 MFI and increased 1.3 folds to an intensity of 8397 MFI together with platelets ( Figure 4D ).