Red blood cell microparticles show altered inflammatory chemokine binding and release ligand upon interaction with platelets

Abstract

BACKGROUND

Storage of red blood cells (RBCs) under standard blood bank conditions results in reduced structural integrity leading to membrane budding and release of microparticles. Microparticles express the blood group Duffy antigen known to bind multiple inflammatory chemokines, but the functional chemokine binding properties of microparticles are not known.

STUDY DESIGN AND METHODS

We determined whether storage-induced microparticles show inflammatory chemokine binding through the expression of the Duffy antigen, comparing the binding properties to intact RBCs, and assessed microparticle interactions with platelets (PLTs) that release chemokines upon activation.

RESULTS

Intact RBCs retained similar equilibrium dissociation constants for CCL2 (Kd = 7.4 ± 0.9 nmol/L), CXCL8 (Kd = 7.9 ± 1.0 nmol/L), and CXCL1 (Kd = 4.4 ± 1.0 nmol/L) throughout storage. In contrast, microparticles increased in relative counts with storage, showed higher percentages of surface phosphatidylserine, and demonstrated impaired Duffy-dependent chemokine binding affinity with wider variability in dissociation constant for CXCL1(Kd = 362 ± 328 nmol/L; range, 0.6–2000 nmol/L). The altered chemokine binding affinity of RBC microparticles was associated with a propensity to release ligand upon incubation with PLTs. Relative quantification of microparticles, based on criteria of glycophorin A expression and size, underestimated particle numbers with functional chemokine binding, suggesting that glycophorin A–negative particles and nanoparticles contribute to overall chemokine binding capacity.

CONCLUSION

Microparticle burden in transfusates, as determined by functional chemokine binding, is considerable. Altered membrane properties of RBC microparticles enhance PLT interactions to increase inflammatory chemokine bioavailability in vitro.

During storage, red blood cells (RBCs) lose their membrane stability leading to hemolysis and microparticle formation.^1,2 The RBC skeletal framework is essential for membrane stability and consists of a filamentous meshwork of proteins created by a spectrin-spectrin dimer backbone anchored to the lipid bilayer at junctional points through interaction with actin and protein 4.1R as well as interactions with ankyrin.^3,4 Storage reduces the formation of the spectrin-actin-protein 4.1 complex that correlates tightly with reductions in total RBC membrane phospholipid content.³ Storage also shifts the redox potential toward oxidative stress.¹ Indeed, spectrin oxidation correlates strongly with membrane microparticle formation.⁵ It is also notable that microparticles shed during storage generally lack the skeletal protein spectrin. ⁶ This is in contrast to Band 3, a membrane integral protein that provides linkage between the lipid bilayer and the underlying skeletal framework, found in microparticles.⁶ Normally sequestered to the inner leaflet of the membrane, anionic phospholipids such as phosphatidylserine also interact with protein 4.1R and spectrin to strengthen mechanical stability.^7–9 However, microparticles lose phospholipid asymmetry and bear phosphatidylserine on their surface.¹⁰ Thus, the anchorage points where membrane regions attach to the underlying cytoskeleton may represent focal sites that favor budding and microparticle formation.¹⁰

The Duffy antigen is a transmembrane protein and minor blood group antigen that interacts with the protein 4.1R–based macromolecular complex—a complex that includes other transmembrane proteins such as Band 3, Rh, Kell, XK, and glycophorin C.¹¹ Historically known as the RBC receptor for merozoite invasion by malarial parasite Plasmodium vivax,^12–15 Duffy antigen demonstrates high-affinity binding to multiple inflammatory CXC and CC chemokines such as CXCL1, CXCL7, CCL2, and CCL5 but not CCL3 or CCL4 with receptor binding sites estimated at 1000 to 9000/RBC surface.^12,16–20 The role of Duffy antigen in inflammatory states has remained elusive, in part, due to its unique position as a nonsignaling protein on circulating RBCs. However, emerging evidence support the role of RBC Duffy antigen as a modifier and regulator of systemic and local inflammatory chemokine bioavailability.^21–23

We have recently shown that transfusion of stored RBCs amplifies existing lung inflammation and promotes lung injury in a murine model of systemic endotoxemia.²² We also demonstrate that increased duration of RBC storage can promote lung injury and implicate the reduction of RBC chemokine scavenging as one consequence of the storage lesion. Studies have previously shown blood group antigenicity on RBC microparticles,²⁴ in particular the presence of the Duffy antigen.¹⁰ We have previously shown reductions in surface Duffy antigen expression on intact RBCs with storage and this loss from the RBC surface may result from membrane alterations and subsequent microparticle formation. Whether or not Duffy protein expressed on microparticles shows altered functional properties remains unknown.

Exposure of phosphatidylserine on the surface of the microparticle membrane provides a procoagulant surface that facilitates thrombin generation.^25–28 Thrombin, a main agonist of platelet (PLT) activation,²⁹ can induce the release of α-granule contents such as P-selectin and chemokines,³⁰ possibly further amplifying inflammatory processes. However, the interaction between RBC microparticles and PLTs in perpetuating inflammation is largely unknown. The purpose of the study was to determine the expression of Duffy antigen on the surface of microparticles obtained from RBC units under standard blood bank conditions, determine its functional properties in terms of chemokine binding, and assess its interaction with PLTs.

MATERIALS AND METHODS

Human subjects

Adsol-preserved leukoreduced and nonleukoreduced RBC units were obtained from the Institute of Transfusion Medicine, Central Blood Bank of Pittsburgh, Pennsylvania based upon their availability. The majority of the studies were conducted with nonleukoreduced units, but select experiments were performed using leukoreduced units only as indicated. For RBC radioligand binding experiments and assays requiring PLT-poor plasma (PPP) or PLT-rich plasma (PRP), we obtained whole blood from healthy volunteers after obtaining written informed consent. The Institutional Review Board of University of Pittsburgh approved the studies.

Reagents

¹²⁵I-CCL2, ¹²⁵I-CXCL8, and ¹²⁵I-CXCL1/GRO-α were purchased from Perkin-Elmer, Inc. (Waltham,MA). Recombinant human proteins CCL2, CXCL8, CXCL1, CXCL5, and CXCL7 were obtained from PeproTech, Inc. (Rocky Hill, NJ).Mouse immunoglobulin (Ig)G1 anti-humanDuffy Fy6 antibody (clone 2C3), glycophorin A–phycoerythrin (PE), and annexin V–fluorescein isothiocyanate (FITC) were obtained from BD Biosciences (San Jose, CA). Mouse IgG2b, k-PE was purchased from BD PharMingen (San Jose, CA). Mouse IgG1 isotype control antibody was obtained from R&D Systems (Minneapolis, MN).

Isolation and purification of RBCs and microparticles

RBCs were purified from RBC units or from anticoagulated whole blood of healthy volunteers using a method we have previously described.^22,31 The method for isolation and purification of RBC microparticles is a modification of the techniques previously described.^32–34 Briefly, banked RBC units (90 mL) were centrifuged at 2000 × g for 20 minutes at 4°C. The pellet was removed and the supernatant was centrifuged at 3000 × g for 10 minutes at 4°C. The supernatant was immediately filtered through a sterile 0.8-µm pore size syringe-driven polycarbonate filter units (Millipore, Billerica, MA) and examined under the microscope to ensure the absence of RBCs. The supernatant was ultracentrifuged at 37,000 × g at 4°C for 1 hour. The pellet containing microparticles was resuspended in phosphate-buffered saline (PBS) and ultracentrifuged twice under the same conditions, resuspended in 300 µL of final volume of PBS, and maintained at 4°C. The ratio of final resuspension volume to original volume of transfusate was maintained at 1:300, and experiments were conducted within 24 hours of isolation. A small aliquot of the resuspended microparticle pellet was examined under the microscope to confirm the absence of RBCs and visible RBC fragments. Microparticles (5 µL) were labeled with glycophorin A-PE and annexin V-FITC and counted using flow cytometry absolute counting standard microbeads (Bangs Laboratories, Inc., Fishers, IN) as detailed below. Microparticle preps were devoid of PLTs or PLT microparticles, as determined by assaying for the constitutive PLT-specific marker CD42a by flow cytometry (data not shown).

Flow cytometric analysis and relative quantification of RBC microparticles

RBC counts were obtained from each unit bag by manual counting using a hemocytometer. The final supernatant volume examined from each unit bag was normalized to 10⁸ RBCs. The supernatant was labeled with glycophorin A as indicated above. The forward scatter variable was set to log scale to identify microparticle, RBC, and fluorescent absolute count beads. The flow was maintained at slow and stable rate, and events were acquired after samples were running for 2 minutes. To reduce the possibility of carryover, cleaning with 10% bleach and deionized water was performed between samples. Microparticle counting was performed in duplicate or triplicate for each sample. Relative quantification of microparticle counts from each RBC unit bag was obtained on the same flow cytometer (FACSCalibur, BD Biosciences) with identical settings on the same day. Acquired events showed a coefficient of variation (1.34%–18.42%; mean, 10.94%). Microparticles were defined as glycophorin A positive and annexin V events at size approximately 70 to 700 nm.

Immunoelectron microscopy

Isolated microparticles were fixed in suspension with 2% paraformaldehyde in PBS and adsorbed to formvarcoated 100-mesh grids. Grids underwent several washings with PBS and then PBS supplemented with 0.5% bovine serum albumin and 0.15% glycine. Microparticles were blocked in 5% normal goat serum and incubated with a 1:50 dilution of mouse IgG1 anti-human Duffy Fy6 antibody (Clone 2C3) and sheep anti-human hemoglobin (Hb), affinity-purified polyclonal antibody (Bethyl Laboratories, Inc., Montgomery, TX). Grids were incubated with goat anti-sheep biotinylated antibody (1:100, Jackson ImmunoResearch Laboratories, West Grove, PA) followed by goat anti-mouse 5-nm gold-conjugated secondary antibody (1:25, Amersham, Piscataway, NJ) and 10-nm gold-conjugated streptavidin (1:25, Amersham). Grids were fixed in 2.5% glutaraldehyde, poststained in 2% neutral uranyl acetate, and finally embedded in 1.25% methylcellulose. Labeling was observed on a electron microscope (JEM 1011, JEOL, Peabody, MA) at 80 kV. Control grids were incubated with biotinylated anti-sheep, goat anti-mouse 5-nm gold, and streptavidin 10-nm gold reagents in the absence of primary antibody. No gold staining was observed in these samples.

Radioligand competition binding assays

¹²⁵I-Chemokine at 0.25 nmol/L was incubated with either microparticles (approx. 0.5 × 10⁶) or RBCs (1 × 10⁸) in the presence or absence of increasing molar concentrations of unlabeled or “cold” homologous chemokine as competitor. Nonspecific binding was defined as 1 µmol/L cold competitor. The mixture was incubated on ice for 60 minutes. The reaction was terminated by centrifuging the mixture through a sucrose cushion. The supernatant and pellet were measured in a gamma counter. Each data point was performed in duplicate or triplicate. Data were analyzed using a nonlinear regression curve fit program. A comparison of fits was used to define the preferred model for one-site-specific binding (Prism 5 v5.01, GraphPad, La Jolla, CA). For the quick chemokine binding assays, ¹²⁵I-chemokine at 0.25 nmol/L was incubated with either microparticles or RBCs in the presence or absence of 3 × 10⁻⁷ mol/L unlabeled homologous competitor.

PLT assays

Whole blood was drawn into acid-citrate-dextrose at a ratio of 1:6 anticoagulant as previously described.^35–37 The blood was immediately centrifuged at 100 × g for 15 minutes at room temperature and the supernatant was separated and collected as PRP. The remaining blood sample was centrifuged at 1880 × g for 20 minutes and the supernatant was separated and collected as PPP. The absence of RBCs and white blood cells were confirmed by light microscopy, and PLT counts from samples were manually counted using a hemocytometer. PRP and PPP from the same individual served as their own comparison.

Phorbol 12-myristate 13-acetate (PMA; Sigma, St Louis, MO) at 200 ng/mL was added to fixed volumes of the PRP or PPP, incubated for 10 minutes at room temperature. Purified RBC microparticles at a fixed volume were incubated with ¹²⁵I-CXCL1 (0.25 nmol/L) and then added to PRP or PPP. The entire mixture was incubated at 4°C for 60 minutes. The reaction was terminated by centrifuging the mixture through a sucrose cushion. Counts per minute for the supernatant and the pellet fractions were obtained using a gamma counter. Each data point was performed in triplicate.

Statistical analysis

A paired t test was performed to compare two matched groups. A Mann-Whitney nonparametric test was applied to compare the distributions of two unmatched groups. For comparisons involving more than two groups, analysis of variance (ANOVA) was performed and a posttest analysis to compare among two groups. A p value of less than 0.05 was considered significant. Correlation was calculated with Spearman correlation coefficient and p value. Statistics were performed using computer software (Prism 5, GraphPad). Data in figures are presented as mean ± SEM.

RESULTS

Storage does not alter the binding affinity of intact RBCs for chemokines CCL2 and CXCL8

To determine whether in-storage time induces alterations of the chemokine binding pocket and, hence, the binding affinity for ligand, we performed homologous competitive binding assays on fresh and stored RBCs using ¹²⁵I-CCL2 to calculate the equilibrium dissociation constants (Kd) for CCL2. Fresh RBCs showed saturable and specific binding in the presence of increasing molar concentrations of unlabeled or “cold” competitor CCL2 (data not shown). Similar to reports by others,²⁰ the Kd was estimated at 6 nmol/L for a representative sample (Fig. 1A).We also compared the Kd of fresh RBCs to that of stored RBCs ranging from Day 1 to Day 41 and show no significant difference in binding affinities across storage time (Fig. 1C).

Fig. 1.

CCL2 and CXCL8 binding affinity of fresh versus stored RBCs. (A) Equilibrium dissociation constant (Kd) for CCL2 presented in Scatchard plot form, using representative fresh RBCs from a volunteer. (B) Scatchard plot data for CXCL8, using representative fresh RBCs from a volunteer. (C) Fresh versus stored RBCs (Days 1–41) show similar CCL2 binding affinity, a comparison of Kd’s. Fresh RBCs, Kd = 7.3 ± 1.9 nmol/L (n = 4); stored RBCs, Kd = 7.1 ± 0.9 nmol/L (n = 13). (D) Fresh versus stored RBCs (Days 1–37) show similar CXCL8 binding affinity. Fresh RBCs, Kd = 9.0 ± 3.4 nmol/L (n = 5); stored RBCs, Kd = 7.9 ± 1.0 nmol/L (n = 5). Data were obtained from Adsol-preserved leukoreduced and nonleukoreduced RBC units, purified and isolated in the same manner as fresh RBCs, and performed in triplicates. (E, F). Heterologous competitive binding assays show percentage of total ¹²⁵I-CCL2 or ¹²⁵I-CXCL1 binding as a function of increasing molar concentrations of unlabeled CXCL1, CXCL5, CXCL7, and CXCL8. For ¹²⁵I-CCL2 binding, representative Ki’s for fresh RBCs are shown. For ¹²⁵I-CXCL1 binding, representative Ki’s for stored RBCs are shown (n = 2–5 for each ligand tested comparing fresh to older stored RBCs, Days 20–32). The dissociation constant for competitor binding (Ki) is provided for each ligand. Fy^null RBCs do not bind chemokine.

The Duffy antigen is a promiscuous chemokine binding protein that binds both inflammatory CC- and CXC-chemokines. We determined whether storage alters RBC binding affinity for CXCL8, the canonical inflammatory CXC chemokine.Homologous competition binding assays using ¹²⁵I-CXCL8 in the presence of increasing concentrations of unlabeled competitor CXCL8 of fresh RBCs from healthy volunteers also showed saturable and specific binding (data not shown). The estimated Kd for RBCs from a representative healthy volunteer was 15 nmol/L (Fig. 1B). The Kd of fresh RBCs was also compared to that of stored RBCs ranging from Day 1 to Day 37 and showed no significant difference in binding affinities across storage time (Fig. 1D). We also performed heterologous competitive radioligand binding assays and show the ability of various CXC ligands to compete off bound ¹²⁵I-CCL2 and ¹²⁵I-CXCL1 (Fig. 1E and 1F). Storage did not appreciably alter the ability of heterologous ligands to compete off-bound chemokine (data not shown).Notably, PLT chemokines CXCL5 and CXCL7 showed dissociation constants for competitor binding (Ki) between that of CXCL1 and CXCL8, whereas Duffy null (Fya−b−) RBCs did not show specific binding to either CC or CXC ligands (Figs. 1E and 1F). Thus, the findings show that the binding affinity for inflammatory chemokines is not altered on intact RBCs with increasing duration of storage.

Fya−b+ RBCs show reduced surface Duffy antigen expression and chemokine binding sites but binding affinity to CCL2 is not altered and independent of storage duration

We examined whether the twomajor alloantigens of Duffy protein (Fya+b−, Fya−b+) showed differences in RBC inflammatory chemokine binding with storage duration. Fya and Fyb antigens clinically define four phenotypes of the minor blood group Duffy antigen: Fya+b−, Fya−b+, Fya+b+, and Fya−b− RBCs. These phenotypes are the products of codominant alleles comprising genotypes FyA/ FyA, FyB/FyB, FyA/FyB, and FyB*/FyB*, respectively.³⁸ The Fya/Fyb alloantigen system is due to a single polymorphism in Codon 42 that encodes a glycine residue in Fya and aspartic acid in Fyb³⁹ and appears to show no known biologic consequence,³⁸ although the inflammatory chemokine binding properties have not been systematically compared.We observed a reduction in Duffy antigen (Fy) surface expression among Fya−b+ RBCs compared with Fya+b− and Fya+b+ RBCs (Fig. 2A). Fya−b+ RBCs also showed reduced surface ¹²⁵I-CCL2 binding after competition with excess unlabeled competitor CCL2 (Fig. 2B). This reduction in surface radioligand binding can be explained by either reduced numbers of binding sites available or reduced binding affinity. Fya+b− RBCs showed similar CCL2 binding affinity compared to Fya−b+ RBCs (Kd = 6.5 ± 1.1 nmol/L vs. 8.7 ± 1.5 nmol/L, respectively) that was independent of storage duration (Fig. 2C), suggesting that reduced number of binding sites account for the reduced surface ¹²⁵I-CCL2 binding observed in Fya−b+ RBCs (Fig. 2B).

Fig. 2.

Comparison of Fy serotypes according to Fy surface expression, percentage of ¹²⁵I-CCL2 bound, and binding affinity for CCL2. (A) RBCs from Adsol-preserved leukoreduced RBC units matched for storage duration were tested for Duffy surface expression by flow cytometry using monoclonal antibody Fy6. The mean fluorescence intensity (MFI) as a function of Fy serologic status. (B) Percentage of ¹²⁵I-CCL2 RBC surface binding following competition with 200 nmol/L CCL2 (n = 10 units, each point reflecting RBCs from a different RBC unit). (C) CCL2 binding affinity: Fya+b− RBCs, Kd = 6.5 ± 1.1 nmol/L (n = 3, performed in triplicates); Fya−b+ RBCs, Kd = 8. 7 ± 1.5 nmol/L (n = 5, performed in triplicates).Mann-Whitney, p < 0.05.

Based upon the Bmax, the calculated estimates of receptor sites available per cell surface were similar between the two groups (receptor sites 4080 ± 1480/cell surface vs. 4030 ± 1500/cell surface, respectively). This estimate is in agreement with previous published reports for RBCs,¹⁶ but is unlikely to discriminate between small differences that require larger size sampling. Thus, Fya−b+ RBCs possess reduced surface Fy expression compared to Fya+b− RBCs independent of storage duration; however, the binding affinity of the available numbers of surface Fy for CCL2 is not appreciably different between the two phenotypes.

RBC microparticle counts increase with storage duration and express surface Duffy antigen

Microparticles express components of the RBC membrane such as glycophorin A³² and are distinguished by their forward scatter properties (or size) relative to RBCs. We employed 7-µm-sized fluorescent beads and estimated the size of RBC microparticles at approximately 70 to 700 nm by flow cytometry (Fig. 3A). Our findings are similar to those of published reports using scanning electron microscopy.³³ A high percentage of microparticles (57%) expressed surface phosphatidylserine at Day 40 of storage, indicating significant phospholipid membrane redistribution but only a minority of RBCs from the same unit bag (3%) showed this rearrangement (Fig. 3B).

Fig. 3.

Relative quantification of RBC microparticles obtained from RBC units (Adsol-preserved, leukoreduced). (A) RBCs from Day 40 RBC unit were washed, counted, and stained with glycophorin A (right panel) or isotype control antibody (left panel). Fluorescent microbeads (approx. 7 mm in dimension) that were added are the same size as glycophorin A–positive RBCs. Note the forward scatter is in log scale, and microparticles are approximately 70 to 700 nm in size. (B) A significantly higher percentage of glycophorin A–positive microparticles express surface phosphatidylserine, indicated by annexin V binding (approx. 57% of total glycophorin A events) compared with intact Day 40 parent RBCs. (C) Relative quantification of microparticles. Supernatants from Day 13 or Day 29 RBC units were labeled with glycophorin A and annexin V (top left panel). The volume of supernatant assayed from each unit bag was normalized to 10⁸ RBC counts. The numbers of annexin V–positive microparticles per 10⁸ RBCs were higher in Day 29 versus Day 13 RBC units (p < 0.05; top right panel). The percentage of annexin V–positive microparticles were also higher in Day 29 versus Day 13 RBC units (p < 0.05; bottom left panel), with each point representing individual RBC units.

We initially examined leukoreduced RBC units for the amount of microparticle formation as a function of storage time. Microparticles were identified by size and expression of glycophorin A and quantified by assaying the numbers proportional to a known amount of fluorescent microbeads added into each sample. By normalizing to RBC counts in each of the unit bags assayed, we were able to quantify the amount of microparticles in a RBC unit relative to other RBC units assayed. Microparticle counts increased from Day 13 to Day 29 (data not shown), as did the percentage of annexinV–positivemicroparticles with storage duration (Fig. 3C, bottom left panel). Examining the data differently, we showed that total numbers of annexin V–positive microparticles also increased from Storage Day 13 to Storage Day 29 (p < 0.05; Fig. 3C, top right panel). Thus, relative quantification of microparticle counts is possible and, regardless of the method of analysis, microparticles increased with storage duration and a significant proportion of these microparticles showed rearrangement of membrane phospholipids.

Others have previously determined that blood group antigens are expressed on microparticles using indirect agglutination testing²⁴ and by Western blotting.¹⁰ By immunoelectron microscopy, purified Fya+b+ microparticles express surface Duffy antigen but Fya−b− microparticles do not (Fig. 4). Relative to the free Hb observed within the pellet fraction, Duffy expression on any given Fya+b+ microparticle is observed attached to the membrane surface. Based on this observation, we were able to rule out the possibility of significant clustering of Duffy monomeric chains during the process of microparticle formation.

Fig. 4.

RBC microparticles express surface Duffy antigen.Microparticles isolated from Fya+b+ donor and Fya−b− donor (Day 40 Adsol-preserved, leukoreduced unit) were immunolabeled with mouse IgG1 anti-Fy6 and sheep anti-Hb. Five-nanometer gold particle indicates Duffy antigen labeling (large arrowheads). Ten-nanometer gold particle indicates Hb labeling (small arrows). Representative samples of two independent experiments are shown.

RBC microparticles show Duffy-dependent functional chemokine binding that increases during storage

We determined whether RBC microparticles are potential reservoirs for inflammatory chemokine binding by initially examining microparticles from expired RBC units. Microparticles, estimated at approximately 10⁶ counts based on predefined criteria of glycophorin A expression and size by flow cytometry, showed ¹²⁵I-CXCL1 binding that was concentration dependent, as serial dilutions of microparticles resulted in progressive reductions in bound chemokine in the microparticle fraction and increased within the soluble fraction (Fig. 5A).

Fig. 5.

RBC microparticle show inflammatory chemokine binding that is Duffy dependent. Purified microparticle pellets were assayed by flow cytometry using glycophorin A expression and size criteria to quantify the relative numbers. (A) Micro-particles (MP) show concentration-dependent ¹²⁵I-CXCL1 binding. Serial dilutions of microparticles (approx. 10⁶ counts) isolated from expired RBC unit (Adsol-preserved, Fya+b+ serotype) at 1:1, 1:2, 1:4, and 1:8 were prepared and incubated with 0.25 nmol/L ¹²⁵I-CXCL1. Progressive reduction in percentage of ¹²⁵I-CXCL1 binding in microparticle pellet is associated with increases in radiolabeled ligand found in the supernatant (SN). A representative sample of n = 3 individual units examined. (B) ¹²⁵I-CXCL1 binding of microparticles (denoted by Fy serotype) at approximately 1 × 10⁵ relative counts and parent RBCs. Incubation with radioligand alone () or radioligand plus excess unlabeled ligand (□) are indicated. The amount of chemokine binding in counts per minute was expressed as percentage of total ¹²⁵I-CXCL1 added in the absence of microparticles or RBCs. Samples were obtained from leuko-reduced units and performed in triplicate (n = 12 samples examined, four representative samples shown). (C) Relative microparticle counts from RBC units correlate with storage duration. Spearman correlation coefficient r = 0.8375, p < 0.0001. (D) Chemokine binding capacity of microparticles from RBC units correlates with storage duration. Spearman correlation coefficient r = 0.5365, p = 0.04. Each point shown in C and D represent microparticles isolated from a different RBC unit, n = 18.

To confirm Duffy-dependent chemokine binding, we purified microparticles from the four different Fy serotypes between Day 22 and Day 27 of storage, comparing them to the parent RBCs. Chemokine binding was Duffy antigen dependent as bound ¹²⁵I-CXCL1 on Fya+b−, Fya+b+, Fya−b+ microparticles could be competed off with excess cold competitor CXCL1 whereas Fya−b− microparticles showed no appreciable binding (Fig. 5B). Adjusted for approximately 10⁵ relative counts per sample, microparticles exhibited chemokine binding capacity exceeding the capacity of 10⁵ RBCs but lower than that of 10⁸ RBCs (i.e., approximating the binding capacity of 10⁶– 10⁷ RBCs). If RBCs (possessing approx. 10³ binding sites per cell surface) are used as a reference to gauge the number of microparticle binding sites participating in chemokine binding, we estimate roughly 10⁹ to 10¹⁰ sites in these samples.

We also determined chemokine binding capacity normalized to the volume of transfusate (i.e., 1.5 mL). While the relative numbers of purified microparticles isolated from each RBC unit bag increased with storage duration (Fig. 5C), chemokine binding capacity normalized to the same volume of transfusate was variable and appeared to increase by approximately Day 15 (Fig. 5D). Thus, RBC microparticles show a concentration-dependent, Duffy-dependent chemokine binding capacity that exceeds the number of microparticles estimated by flow cytometric analysis and increased by approximately Day 15 of storage.

RBC microparticles show wider variation in CXCL1 binding affinity relative to intact RBCs

We next determined the strength of microparticle binding to chemokine by determining the equilibrium dissociation constant for CXCL1. Although microparticles can show similar CXCL1 binding affinity to intact RBCs (Figs. 6A and 6B, Kd = 4 nmol/L), there was wider distribution in microparticle binding affinity across individual units and, on average, microparticles showed impaired chemokine binding functionality (Kd ≈ 362 nmol/L; range, 0.6–2000 nmol/L). In contrast, RBCs showed very reproducible high-affinity binding to CXCL1 with Kd = 4.4 ± 1.0 nmol/L (Fig. 6B). Neither Duffy antigen serotypes nor storage duration predicted the variations in chemokine binding affinity observed with microparticles (data not shown).

Fig. 6.

Saturation curve, Scatchard plot, and comparisons of CXCL1 binding affinity of microparticles and stored RBCs. (A) Micro-particles show saturable and specific ¹²⁵I-CXCL1 binding. Inset data in Scatchard plot form, Kd = 4 nmol/L. (B) RBCs show saturable and specific ¹²⁵I-CXCL1, Kd = 4 nmol/L. (C) Wide variation in microparticle binding affinity for CXCL1, Kd = 362 ± 328 nmol/L (range, 0.6–2000 nmol/L) compared with RBC binding. Each point represents microparticles or RBCs isolated from individual RBC units, assayed in duplicate or triplicates.

RBC microparticles release chemokine upon interaction with PLTs

RBC microparticles have been hypothesized to show altered interactions with PLTs, although direct evidence of this is lacking. We used PMA to activate PLTs and at the concentration of 200 ng/mL used, we observed approximately 90% of PLTs expressing surface P-selectin expression, a marker of α-granule release (Fig. 7A). When microparticles and ¹²⁵I-CXCL1 were incubated with plasma containing PLTs, there was increase in chemokine released into the soluble fraction and a concomitant reduction in the pellet fraction containing microparticles and PLTs (Fig. 7B). PLTs accounted for approximately 10% of chemokine binding that was nonspecific, as excess “cold” or unlabeled competitor did not compete off “hot” or radiolabeled ligand from PLTs (Fig. 7C). Provided that PLTs do not interact with microparticles, the anticipated outcome would be that chemokine binding in the pellet fraction is not significantly altered, as is the case with autologous RBCs (data not shown), or more likely increased due to the contribution of nonspecific binding fromPLTs.However, the opposite is observed and our findings show that PLTs can alter microparticle-chemokine binding, leading to chemokine release(Figs. 7B and 7D).

Fig. 7.

Microparticle-chemokine binding in the presence of PLTs. (A) PMA induces PLT activation resulting in P-selectin (CD62P) surface expression. PRP was incubated with increasing concentrations of PMA. Y-axis indicates percentage of CD62Ppositive PLTs. Inset shows dual-positive CD62P-positive, CD42a-positive PLTs at a PMA concentration of 200 ng/mL used in subsequent experiments B through D. (B) Chemokine in the RBC microparticle (MP) pellet fraction is released into the supernatant (SN) after incubation with PLTs.MPs were incubated with ¹²⁵I-CXCL1 in the presence or absence of PPP, activated PPP (aPPP), PRP, or activated PRP (aPRP). “Hot” indicates MP in the presence of ¹²⁵I-CXCL1 without plasma or PLTs. *p < 0.05, **p < 0.01 with comparisons between two groups indicated by connecting lines, oneway ANOVA with Tukey’s multiple comparisons test. Representative sample of chemokine release from experiments shown in D. (C) PLTs from three donors do not show specific CXCL1 binding, as a percentage of total ¹²⁵I-CXCL1 placed into the reaction mixture does not change in the presence of excess cold competitor CXCL1. Hot = radioligand alone; hot + xs cold = radioligand plus excess unlabeled ligand. (D) Reductions in percentage of total ¹²⁵I-CXCL1 binding in the microparticle-pellet fractions, n = 9 independent experiments representing microparticles isolated from seven nonleukoreduced units. The mean percentage change in chemokine binding ± SEM is as follows: PPP, −3 ± 6%; aPPP, −2 ± 6%; PRP, −14 ± 7%; aPRP, −22 ± 10%. Percentage change in total ¹²⁵I-CXCL1 binding was calculated by normalizing to MP-pellet fractions incubated with ¹²⁵I-CXCL1 in the absence of plasma or PLTs. *p < 0.05, **p < 0.01 with comparisons between two groups indicated by connecting lines, one-way ANOVA with Tukey’s multiple comparisons test.

DISCUSSION

We present several novel findings of stored RBCs and their membrane-derived microparticles under standard blood bank conditions. We have previously reported surface reduction of Duffy antigen on stored RBCs over time²² and now show that microparticles express the transmembrane Duffy protein displaying functional chemokine binding. Collectively, these microparticles show considerable chemokine binding capacity and may reflect a total number that exceeds the numbers of microparticles measured by flow cytometric analysis, as defined by glycophorin A expression and size criteria. Consistent with prior findings,^10,40 RBC microparticles increase with storage duration, and a higher percentage show membrane phospholipid redistribution over time. Microparticles also show, on average, impaired chemokine binding affinity compared to intact RBCs, allowing for greater propensity to release ligand upon interaction with PLTs.

Stored RBCs undergo oxidative modification such as spectrin oxidation that predisposes toward membrane microparticle formation.⁵ Our findings indicate that intact RBCs do not show functional alterations in CCL2, CXCL8, or CXCL1 binding affinity. Thus, the chemokine binding pocket created by the tertiary structure of the Duffy protein shows relatively preserved binding on intact RBCs during storage with predictable Kd of approximately 4 nmol/L for CXCL1. This is in contrast to the impaired chemokine binding affinity observed for microparticles. We speculate that the phospholipid redistribution of the microparticle membrane may alter the stability of Duffy-ligand binding.

We also observed small differences in surface chemokine binding capacity between Fya+b− and Fya−b+ serotypes, with Fya−b+ RBCs showing both reduced surface Fy expression and reduced ¹²⁵I-CCL2 binding after competition with excess cold competitor. Although there was wider variation in CCL2 binding affinity using Fya−b+ RBCs, the differences were not significant and appeared to be storage independent. We conclude that differences in chemokine binding affinity between the Fy serotypes may be distinct but only uncovered with larger population studies.

Indeed, in a recent genomewide association study involving three independent cohorts (n = 9598), the strongest association regulating circulating CCL2 concentrations was the nonsynonymous polymorphism Asp42Gly in the Duffy antigen gene.²³ The Asp42Gly polymorphism results in the Fya−b+ (A/A genotype, Asp) and Fya+b− (G/G genotype, Gly) serotypes.³⁹ The A/A genotype (corresponding to Fya−b+ serotype) was associated with higher serum CCL2 concentrations but not plasma concentrations, leading the authors to postulate that serum CCL2 concentrations represented the combination of plasma CCL2 and additional CCL2 released from RBC Duffy antigen through the process of clotting. Consistent with our observations, the study also showed a similar reduction in surface Fy expression in individuals expressing the Fya−b+ (Asp, A/A)²³ serotype, although the authors concluded that this finding did not represent a meaningful difference to account for the higher systemic CCL2 concentrations. However, the possibility exists that reduced surface binding sites can result in higher soluble chemokine concentrations in vitro due to the equilibrium dissociation kinetics of ligand to the Duffy protein and that clotting suggests an interaction between PLTs and RBC chemokine binding.

Although relative microparticle counts increased with storage duration, the relationship between chemokine binding capacity and storage was not linear. The increase in microparticle-chemokine binding occurred by Day 15 and remained fairly constant throughout the shelf life of blood (Day 42) although expired blood showed higher chemokine binding. In addition, a limitation of relative quantification was the definition of microparticles. Strictly defining microparticles by glycophorin A expression and size criteria precluded the possibility of including either glycophorin A–negative particles or particles of smaller dimensions with functional chemokine binding. Because we made no attempt to separate out nanoparticles from microparticles during the ultracentrifugation steps, it is reasonable to conclude that our microparticle counts underestimate the amount of particles participating in chemokine binding. Indeed, the chemokine binding capacity attributable to micro-particles is consistent with this hypothesis given the limited numbers of Duffy antigen identified per micro-particle surface by immunoelectron microscopy. If the numbers of predicted chemokine binding sites are an indication of actual particle counts, this suggests that loss of RBC Duffy antigen by microparticle formation point to a region of membrane vulnerability. Based on Duffy’s location within the 4.1R macromolecular complex, ¹¹ this is a region of the membrane where anchorage to the skeletal framework may be compromised during storage.³

RBC microparticles have been shown to facilitate thrombin generation in vitro¹⁰ and this is consistent with surface exposure of phosphatidylserine on microparticles. Our findings show that PLTs attenuate microparticle chemokine binding and activation of PLTs may accentuate this process. One possible explanation for this finding is that PLTs may physically bind microparticles and subsequently reduce the number of binding sites available for chemokines. Physical interaction between PLTs and microparticles may also activate PLTs, resulting in further release of chemokines from alpha granules that can compete off microparticle-bound chemokines. Indeed, PLT chemokines such as CXCL5 and CXCL7 show the ability to compete off both microparticle-bound CCL2 and CXCL1 with Kis between those of CXCL1 or CXCL8. Others have shown that PLT microparticles can provide a mechanism for chemokine deposition to activated endothelium leading to monocyte arrest.⁴¹ We speculate that RBC microparticles show a dynamic relationship with PLTs leading to the binding and release of chemokines that may alter local inflammatory microenvironments in vivo.

In summary, RBC microparticles under standard storage conditions express transmembrane Duffy protein with impaired chemokine binding functionality.While the Duffy protein does not appear to significantly cluster on the microparticle surface, its collective chemokine binding capacity is considerable and suggests that relative estimation of microparticle numbers by flow cytometric analysis using glycophorin A expression and size criteria likely underestimates the true particle count. Moreover, altered membrane property of microparticles can enhance interactions with PLTs to release chemokine bound on the microparticle surface and alter soluble concentrations of ligand.

ABBREVIATIONS

PMA

phorbol 12-myristate 13-acetate

PPP

platelet-poor plasma

PRP

platelet-rich plasma