Targeting macrophages prevents alloantibody-mediated platelet clearance in a murine model of transfusion refractoriness

Key Points

• •Alloantibodies mediate platelet elimination in a murine model of HLA class I platelet transfusion refractoriness.
• •Liver and spleen macrophages are new targets to prevent HLA class I platelet transfusion refractoriness.

Visual Abstract

Abstract

HLA class I–immunized patients can experience a serious complication known as platelet transfusion refractoriness (PTR). This issue becomes especially relevant in onco-hematology departments where platelet transfusions are at the heart of patient care. Although transfusion failure is evidenced by a rapid elimination of allogeneic platelets from the recipient’s bloodstream, the mechanisms behind it remain poorly characterized. The aim of this study was to better define these mechanisms to improve therapy for PTR. Using a murine model of major histocompatibility complex class I incompatibility to mimic PTR, we first established that antibodies, but not natural killer or CD8 cells, mediated platelet clearance. However, blocking Fcγ receptors with intravenous immunoglobulin or a monoclonal antibody or complement depletion did not correct refractoriness in alloimmune mice. Therefore, we investigated other alternatives beyond antibody-dependent mechanisms. Flow cytometric and microscopic analysis showed that Kupffer cells in the liver and red pulp macrophages in the spleen phagocytose allogeneic platelets during PTR. Moreover, intravital microscopy revealed allogeneic platelets retained in close interaction with macrophages in the red pulp only in alloimmune animals. Splenectomy or Kupffer cell depletion with clodronate in alloimmune mice suggested the existence of compensatory elimination mechanisms in the liver and spleen. Therefore, the simultaneous removal of both macrophage populations was an effective strategy to abrogate PTR. Our study provides an insight into the mechanisms of platelet clearance in alloimmune pathologies and opens up new perspectives for therapeutic targets.

Introduction

Platelet transfusion is a lifesaving procedure for patients with a risk of severe hemorrhage. However, the presence of anti–HLA class I immunoglobulin G (IgG) in the recipient’s bloodstream can compromise the efficacy of this transfusion. In this case, the recognition by antibodies of HLA class I molecules expressed at the surface of the transfused platelets leads to their rapid clearance. This quick elimination results in a therapeutic failure known as platelet transfusion refractoriness (PTR), which occurs in 5% to 15% of transfusions in onco-hematological patients.1, 2, 3

The management of these patients represents a deadlock in transfusion medicine: HLA-compatible platelet units are the first obvious solution, but the obstacles to identify and deliver them in a timely manner rule out this option for patients with an imminent risk of life-threating hemorrhage.^1,4,5 Although the treatment of HLA class I platelet transfusion refractory patients urgently requires effective alternatives, our lack of understanding of the underlying mechanisms of platelet elimination still impedes their development. Two crucial aspects remain poorly understood: the mechanisms leading to the clearance of transfused platelets in PTR and the final elimination sites following alloantibody recognition.

Allogeneic platelets are swiftly recognized by (allo)antibodies before their elimination. This recognition leads to their rapid clearance from the bloodstream, presumably through antibody-dependent mechanisms.⁶ Some studies have already identified platelet elimination mechanisms dependent on antibodies, such as the opsonization of human platelets by autoantibodies in immune thrombocytopenia (ITP)⁷ or activation of complement by certain anti–HLA class I antibodies.⁸ In addition to this humoral response, a cellular response has also been studied, revealing a role for CD8 T cells in the elimination of platelets in an antibody-free mouse model.⁹

The fate of platelets under physiological conditions has long been a subject of debate and controversy.10, 11, 12 It is nevertheless well established that the liver and spleen play a key role in determining the life span of platelets. These 2 organs are in intimate contact with platelets due to their specialized anatomical structure associated with their blood-filtering function.^13,14 They harbor a myriad of cells scavenging for senescent or apoptotic cells, pathogens, and immune complexes to remove from the circulation and degrade or recycle.14, 15, 16 In ITP, an antibody-mediated platelet pathology, macrophages in these organs have already been shown to be capable of eliminating platelets efficiently through Fcγ receptors (FcγR), FcγRI and FcγRIII.^7,17,18 In murine models of ITP, intravenous human IgG (IVIg) treatment and FcγRIV blockade prevent autoantibody-mediated platelet clearance by macrophages.19, 20, 21 These features make the liver and spleen competitive candidates for the capture of transfused platelets recognized by anti–HLA class I alloantibodies.

In a previous study, we set up a murine model of major histocompatibility complex (MHC) I PTR to mimic the human HLA class I refractory state.²² In this work, we first established in this model that the recognition of platelets by alloantibodies is necessary for the manifestation of PTR. We then showed that blocking in vivo antibody-dependent mechanisms such as complement or FcγR was not sufficient to avoid the clearance of allogeneic platelets. Considering aspects other than antibodies, confocal and electron microscopic analyses identified the liver and spleen as end sites for allogeneic platelets in PTR. Using flow cytometry, we measured the percentages of both Kupffer cells (KC) in the liver and red pulp macrophages (RPM) in the spleen positive for allogeneic platelets in naive and refractory mice. Although neither the depletion of KC with clodronate nor splenectomy could prevent PTR, intravital microscopy revealed that RPM eliminated platelets. Finally, we showed that targeting both liver and spleen macrophages was an effective strategy to abrogate transfusion refractoriness.

Materials and methods

Animals

Wild-type C57BL/6J (JAX stock number 000664; H-2^b) and wild-type BALB/c (BALB/cOlaHsd; H-2^d) mice were purchased from Charles River Laboratories (Saint-Germain-Nuelles, France) and Envigo (Horst, The Netherlands), respectively. For some experiments, B6.129(Cg)-Gt(ROSA)26Sor^{tm4(ACTB-tdTomato,-EGFP)Luo}/J (JAX stock number 007676; H-2^b)²³ mice were crossed with C57BL/6-Tg(Pf4-icre)Q3Rsko/J (JAX stock number 008535; H-2^b)²⁴ mice to obtain H-2^b platelets with membrane expression of either tdTomato or enhanced green fluorescent protein (eGFP). Further details can be found in the supplemental Methods. The murine model of alloimmunization has been described previously.²²

Experiments were performed in accordance with European Union Directive 2010/63/EU. This study was approved by the Regional Ethics Committee for Animal Experimentation of Strasbourg, CREMEAS (CEEA 35).

In vivo assays

Details concerning the treatments of mice before allogeneic platelet transfusion can be found in the supplemental Methods.

Platelet transfusion and flow cytometry

Platelet suspensions were prepared either as washed platelets²⁵ or as citrated platelet-rich plasma. The flow cytometric identification of transfused platelets in peripheral blood and immune cells from the organs is described in the supplemental Methods and supplemental Table 1.

Microscopy

Immunohistology and confocal, intravital, and electron microscopy procedures are described in the supplemental Methods and supplemental Table 2.

Statistical analyses

The statistical analyses are detailed in the figure legends and the supplemental Methods.

Results

Anti-H-2^b alloantibodies, but not NK or CD8 cells, mediate elimination of H-2^b platelets

We first performed serum transfer experiments to determine the capacity of alloantibodies to engender PTR. Naive BALB/c (H-2^d) mice were injected with a pool of sera from BALB/c (H-2^d) mice immunized against H-2^b platelets 18 hours before transfusion challenge with fluorescent allogeneic platelets (H-2^b). Flow cytometric analyses (supplemental Figure 1A) showed that only the recipients of alloimmune serum were in a refractory state (P < .0001), confirming that alloantibodies drive the clearance of platelets from circulation in PTR (Figure 1A). We checked that H-2^b platelets were recognized in vivo by the passively administered alloantibodies 5 minutes after transfusion (Figure 1B).

Figure 1.

Anti-H-2^b alloantibodies, but not NK or CD8 cells, mediate H-2^b platelet elimination. (A) Serum transfer from alloimmune to naive H-2^d mice before H-2^b platelet transfusion lead to refractoriness in recipients injected with alloimmune serum but not in those serum (n = 3, nonlinear regression fit comparing 1-phase decay and straight line models; P < .0001). (B) Alloantibodies were bound to H-2^b platelets in vivo 5 minutes after transfusion in naive mice injected with alloimmune serum but not in naive mice receiving normal mouse serum. The histogram overlay shows an example of the shift in mean fluorescence intensity of washed platelets from recipient blood at the 5-minute time point, revealed with a fluorescent goat anti-mouse IgG (G@mIgG) secondary antibody. (C) Antibody depletion of NK and CD8 cells in alloimmune mice did not alter the refractoriness when compared with that in animals injected with an IgG control (n = 3 for anti-NK + anti-CD8, n = 2 for IgG controls, nonlinear regression fit comparing 1-phase decay curves; P = .8023). i.p., intraperitoneal; PLT, platelet.

We then used an active model of alloimmunization to assess the role of the cellular response in PTR. We depleted mice of natural killer (NK) and CD8 cells before transfusion and verified the efficiency of the depletion after the experiment in each individual (supplemental Figure 1B-C). There was no difference in the kinetics of refractoriness between depleted mice and controls receiving IgG (P = .8023; Figure 1C), suggesting that NK and CD8 cells were not responsible for the refractoriness. These results pointed to alloantibodies as the primary mediators of allogeneic platelet elimination in PTR.

Blockade of antibody-dependent mechanisms of platelet clearance is insufficient to correct PTR

To identify a possible therapeutic target to overcome PTR, we blocked different antibody-dependent mechanisms in alloimmune mice before transfusion with allogeneic platelets (Figure 2A). We first targeted FcγR-dependent phagocytosis because these receptors are involved in clearing immune and platelet-antibody complexes, as observed in ITP.^7,26 Alloimmune mice were injected with either IVIg or human albumin 24 hours before transfusion with allogeneic platelets.¹⁹ There was no difference between mice treated with IVIg or albumin (P = .2374; Figure 2B). We measured the alloantibodies just before treatment and just before transfusion. We did not find significant differences between mice receiving IVIg or albumin over time (supplemental Figure 2A). We also isolated granulocytes from spleen 24 hours after transfusion and confirmed that they still bore human IgG bound to their surface, indicating that the blocking was effective throughout the experiment (supplemental Figure 2B).

Figure 2.

Blockade of antibody-dependent mechanisms of platelet elimination is insufficient to overcome PTR in alloimmune mice. (A) Diagram showing the strategy to block antibody-dependent mechanisms in alloimmune mice before transfusion with allogeneic platelets. (B) Transfusion kinetics in alloimmune mice receiving IVIg or albumin 24 hours before transfusion (n = 4, nonlinear regression fit comparing IVIg- and albumin-injected mice; P = .2374). (C) Transfusion kinetics in alloimmune mice treated with anti-FcγRIV (9E9) or an IgG control 4 hours before transfusion (n = 4, nonlinear regression fit comparing 9E9- and IgG control-injected mice; P = .5531). (D) Transfusion kinetics in alloimmune mice treated with CVF or PBS 24 hours before transfusion (n = 3, nonlinear regression fit comparing CVF- and PBS-injected mice; P = .0774). CVF, cobra venom factor; PLT, platelet.

Given that blockade with IVIg has known diverse side effects,²⁷ we decided to block FcγR with a more specific tool. Therefore, we administered a monoclonal antibody (mAb) against FcγRIV, 9E9, which blocks FcγRIII on cells expressing the target FcγRIV.²⁸ FcRγIII has been identified as a foremost receptor for platelet phagocytosis in the context of ITP, especially for splenic macrophages.^7,17,29 Furthermore, FcγRIV blockade has already been proven to be effective in ITP models.³⁰ There was no significant difference in refractoriness between mice receiving anti-FcγRIV treatment or an isotype control (P = .5531; Figure 2C), suggesting that the refractory state was insensitive to this treatment. Injection of the mAb 9E9 4 hours before transfusion was sufficient to block FcγR. Splenocytes were strained with F4/80 as a macrophage marker 9E9 coupled to a fluorochrome 20 hours after transfusion (ie, 24 hours after the injection of 9E9 or control IgG). Coupled 9E9 did not bind to macrophages from mice previously injected with the same mAb (presumably because the binding sites were still occupied), confirming that the treatment was effective over the course of the experiment (supplemental Figure 2C-D).

Finally, we depleted complement with cobra venom factor 24 hours prior to transfusion. It has been reported that although this does not afford 100% depletion of the C3 component, total complement activity is inhibited 24 hours after treatment and for up to 5 days later.³¹ We observed a significant reduction in serum levels of C3 in animals treated with cobra venom factor (supplemental Figure 2E). However, alloimmune mice depleted of complement presented the same kinetics of refractoriness after the transfusion of allogeneic platelets as control receiving vehicle (P = .0774; Figure 2D). In conclusion, blocking FcγR or depleting complement did not prevent PTR in alloimmune mice.

The liver and spleen serve as allogeneic platelet graveyards during PTR

Considering the inhibition of antibody-dependent mechanisms was insufficient to correct PTR, we decided to target posterior elimination sites. First, to identify these sites of elimination, we harvested the liver and spleen 30 minutes after transfusion, when most of the transfused platelets had disappeared from the circulation of refractory mice but before their eventual digestion/degradation.

Liver F4/80⁺ macrophages from refractory mice display an internal accumulation of allogeneic platelets, whereas allogeneic platelets were extremely rarely observed in contact with these cells in naive mice (Figure 3A). Overall, there was a highly significant difference (P < .0001) in the number of platelets in equivalent volumes of the liver tissue between naive and refractory animals (Figure 3B). Remarkably, in the liver of refractory mice, almost all allogeneic platelets were found within F4/80⁺ macrophages (Figure 3C). To further characterize the localization of these platelets and their ultrastructure, we performed transmission electron microscopy experiments. Given that the fragility of the liver tissue precludes immunogold labeling for platelet identification, we loaded allogeneic platelets with fluorescent latex beads during the platelet isolation procedure. Pale, opaque latex beads of uniform diameter were used to distinguish allogeneic from endogenous platelets. We found KC located as expected in the lumen of hepatic blood vessels (Figure 3D), whereas platelets containing dense granules and latex beads were present inside phagosomes, only in refractory mice (Figure 3E).

Figure 3.

Phagocytic cells in the liver and spleen capture allogeneic platelets during PTR. (A) Representative 3D reconstructions of confocal microscopy stacks from the livers 30 minutes after transfusion showed F4/80⁺ macrophages (cyan surface) containing eGFP⁺ H-2^b platelets (red spheres) only in refractory mice. In naive mice, eGFP⁺ H-2^b platelets were rare events within blood vessels (CD31⁺, magenta) and outside F4/80⁺ macrophages (yellow spheres) (n = 4). (B) The density of eGFP⁺ H-2^b platelets was higher in refractory livers than in naive livers (P < .0001, Mann-Whitney test). (C) The median percentage of eGFP⁺ H-2^b platelets inside F4/80⁺ cells in the refractory livers was 92.50% (95% confidence interval [CI], 55.00-97.00]). (D) A representative KC (discontinuous blue contour) from the liver of a naive mouse transfused with eGFP⁺ H-2^b platelets, identifiable through its content of electron-dense phagosomes (white asterisks) and intravascular position. (E) KC in refractory mice contained eGFP⁺ H-2^b platelets, distinguished by dense granules (yellow arrow) and 0.3 μm-diameter latex beads (indicated by “L”). (F) Representative 3D reconstructions of confocal microscopy stacks from the splenic red pulp 30 minutes after transfusion showed F4/80⁺ macrophages (cyan surface), with most of the eGFP⁺ H-2^b platelets inside (red spheres) in refractory mice, whereas naive mice presented with more platelets outside (yellow spheres) F4/80⁺ macrophages (n = 4). (G) Compared with naive mice, refractory mice had a higher density of eGFP⁺ H-2^b platelets in the red pulp (P = .0077, Mann-Whitney test). (H) The percentage of eGFP⁺ H-2^b platelets within F4/80⁺ macrophages was higher in refractory spleens than in naive spleens (P < .0001, t test). (I) Naive mice displayed free-flowing eGFP⁺ H-2^b platelets in the spleen, identifiable through colloidal gold labeling of membrane eGFP by immunoelectron microscopy (IEM) (inset). They were close to other blood elements such as erythrocytes (indicated by “RBC”) and mononuclear cells (indicated by “mono”) but showed no ultrastructural signs of activation, with α-granules (indicated by “α”) in their center and a discoid shape. (J) In contrast, in refractory spleens, eGFP⁺ H-2^b platelets were observed inside phagocytic cells (discontinuous cyan contour) and rich in electron-dense phagosomes (white asterisks) in the red pulp. Despite their aberrant ultrastructure, eGFP labeling (inset) allowed the identification of allogeneic platelets, which presented signs of degranulation. Scales are indicated on each relevant image.

The appearance of the splenic red pulp was different, with allogeneic platelets visible in both refractory and nonrefractory mice (Figure 3F), owing to the fact that this organ constitutes a pooling site for circulating platelets under physiological conditions. However, we observed a higher density of allogeneic platelets (P = .0077) in refractory spleens (Figure 3G). Furthermore, 64.75% of the transfused platelets observed in refractory mice were located within F4/80⁺ macrophages, as compared with only 18.50% in the naive animals (P < .0001) (Figure 3H). To obtain a more detailed view, we used immunoelectron microscopy, which showed allogeneic platelets in the flowing blood, along with erythrocytes and mononuclear cells in naive mice. The central α-granules and discoid shape of these platelets revealed no signs of activation (Figure 3I). In contrast, the allogeneic platelets in refractory spleens, identifiable inside phagocytic cells through eGFP colloidal gold labeling, presented deformed platelet structures, with no clear α-granules and an irregular shape (Figure 3J). All these images revealed allogeneic platelets in association with macrophages in the liver and spleen only in refractory mice.

KC and RPM are responsible for clearing allogeneic platelets from the circulation in PTR

To quantify the proportion of cells positive for allogeneic platelets and characterize their immunophenotype in PTR, we analyzed single-cell suspensions from the liver and spleen of naive and refractory mice by flow cytometry 30 minutes after transfusion. Based on our previous results, we focused our analysis on tissue resident macrophages. Therefore, in our flow cytometric strategy, we first excluded B, T, NK, and remaining red blood cells (supplemental Figure 3). KC in the liver and RPM in the spleen could subsequently be identified through their expression of the pan-macrophage marker F4/80 and low/intermediate expression of the myeloid marker CD11b (Figure 4A,D). We then determined the percentage of cells positive for the fluorescent probe Oregon Green 488, which was used to stain H-2^b platelets during the isolation (Figure 4B,E).

Figure 4.

KC in the liver and RPM in the spleen take up allogeneic platelets in PTR. (A) KC were identified by flow cytometry as lineage (CD3, CD45R, Nkp46, and Ter119) negative, F4/80⁺, CD11b^neg/low cells 30 minutes after transfusion. (B) Representative dot plots to visualize KC positive for H-2^b platelets in naive and refractory mice. (C) Comparison of the percentage of KC positive for H-2^b platelets in naive and refractory mouse livers 30 minutes after transfusion. (D) RPM were identified by flow cytometry as lineage (CD3, CD45R, Nkp46, and Ter119) negative, F4/80⁺, CD11b^neg/low cells 30 minutes after transfusion. (E) Representative dot plots to visualize RPM positive for H-2^b platelets in naive and refractory mice. (F) Comparison of the percentage of RPM positive for H-2^b platelets in naive and refractory mouse spleens 30 minutes after transfusion. In all experiments (n = 4), bars in the graphs denote median and range (Mann-Whitney test). PLT, platelet.

The proportion of KC and RPM positive for H-2^b platelets was significantly higher in the organs of refractory animals than in naive animals 30 minutes after transfusion. Therefore, the median percentage of KC positive for allogeneic platelets was 61.01% in refractory mice, but only 1.70% in naive mice (P = .0143; Figure 4C). The median percentage of RPM positive for allogeneic platelets was 16.44% in refractory mice, but only 2.96% in naive mice (P = .0143; Figure 4F).

Macrophages in the liver and spleen play compensatory roles in the clearance of allogeneic platelets in PTR

To confirm the functional role of these macrophages in PTR, we sought to deplete alloimmune mice of macrophages by injecting them with clodronate liposomes (or control phosphate-buffered saline [PBS]-loaded liposomes) 24 hours before their transfusion with fluorescent H-2^b platelets. The dosage and timing of clodronate treatment to deplete marginal zone macrophages in the spleen,³² while preserving RPM,³³ had been previously described, but these parameters were not available for liver macrophages. We set up a model of KC depletion in naive BALB/c and labeled structures through the in vivo injection of fluorescent-labeled antibodies: anti-F4/80 and anti-CD31. In the spleen, we could still detect F4/80⁺ cells throughout the red pulp (Figure 5A). Complementary to the tissue sections, we quantified the remaining percentage of KC or RPM by flow cytometry (Figure 5B). KC were depleted in mice treated with clodronate (1.22%) when compared with that in control animals (28.13%; P < .0001). RPM were reduced by approximately half (17.60%) in mice injected with clodronate compared with that in those receiving control PBS-liposomes (32.44%; P = .0034). We also sampled blood before and after treatment to detect inflammatory changes in white blood cell counts due to the injection of clodronate-/PBS-liposomes, but we did not find any significant differences in peripheral blood cell counts (supplemental Figure 4A). We observed a rapid removal of allogeneic platelets from the circulation in both clodronate-treated and control alloimmune mice. Nonetheless, after the first 30 minutes, in mice treated with clodronate, approximately one-third of the transfused platelets returned to the circulation for up to 10 hours posttransfusion before being slowly eliminated up to the 24-hour time point (Figure 5C). These findings confirmed that macrophages in the liver and spleen are the end site for allogeneic platelets in PTR, but they also showed that the depletion of KC alone is not sufficient to correct the refractoriness.

Figure 5.

Macrophages in the liver and spleen are compensatory in the clearance of allogeneic platelets in refractory mice. (A) Sections of organs from mice 24 hours after the administration of clodronate- or PBS-liposomes. Macrophages were labeled with F4/80 (cyan) and the endothelium with CD31 (magenta) in vivo (n = 3). Liver images are 3.81-μm-thick maximum intensity projections, and spleen images are 4.16-μm-thick maximum intensity projections. Scale bar, 50 μm. (B) Comparison of the percentage of KC and RPM in mice injected with clodronate-or PBS-liposomes. (n = 4; multiple t tests). Bars in the graphs denote the mean and standard deviation. (C) Transfusion kinetics in alloimmune mice treated with clodronate- or PBS-liposomes (n = 3, nonlinear regression fit comparisons; P = .0019). (D) Transfusion kinetics for 1 hour in splenectomized or sham-operated alloimmune mice (n = 3, nonlinear regression fit; P = .7067). DAPI, 4′,6-diamidino-2-phenylindole; PLT, platelet.

Our results had shown that RPM in the spleen take up allogeneic platelets in PTR. However, when we splenectomized alloimmune mice and waited for 10 days before challenging them with a new allogeneic platelet transfusion, the splenectomized animals displayed no differences in the kinetics of PTR when compared with that in sham-operated mice (P = .7067; Figure 5D). This implied that the spleen is not necessary for the establishment of PTR once alloimmunization has occurred. To investigate this discrepancy and further elucidate the role of splenic macrophages, we performed intravital microscopy experiments in which we injected a fluorescent anti-F4/80 antibody to label RPM, exposed the spleen of naive or alloimmune mice, and transfused them with fluorescent allogeneic platelets. As in our previous microscopy and flow cytometry experiments, observations were conducted during the first 30 minutes after transfusion. The images clearly showed platelets close to the surfaces of macrophages in both naive and refractory animals (Figure 6A), but significantly closer in refractory spleens (Figure 6B). The analysis of the speed of individual platelets also revealed that they were flowing more slowly in the spleens of refractory mice (Figure 6C). Most of them even seemed to remain on the same spot above certain F4/80⁺ areas (supplemental Videos 1 and 2). Platelets closer than 2 μm to the macrophages dwelled for longer periods in refractory than in naive animals (Figure 6D). Considering the platelet speed, their distance from macrophage surfaces, and the dwell time, we deduced that allogeneic platelets were retained extremely close to F4/80⁺ cells in refractory mice. Therefore, we conclude that RPM are not essential for PTR in alloimmune mice, but they eliminate platelets when they are present.

Figure 6.

Splenic macrophages in refractory mice retain allogeneic platelets in vivo. (A) Intravital microscopy images of the spleen of naive and refractory mice transfused with fluorescent allogeneic (H-2^b) platelets. Macrophages (F4/80⁺) are labeled in magenta and H-2^b platelets (Oregon Green 488 CFDA-SE) in green. Platelets were tracked (colored lines) for ∼45 seconds, and platelets that come within <2 μm of a macrophage surface appear in red, whereas those that do not come within this distance from a macrophage surface appear in blue. (B) Quantification of the shortest distance between the center of any given allogeneic platelet and the closest macrophage surface at each time point shows that platelets are closer to macrophages in the refractory spleen. (C) Comparison of the track speed mean of allogeneic platelets reveals that platelets are slowed down in the refractory spleen compared with the naive spleen. The panel on the right focuses on the subset of platelets with the lowest track speed mean in naive and refractory spleens. (D) Comparison of dwell times of platelets closer than 2 μm to a macrophage surface shows that they remain for longer within this radius from macrophages in refractory than in naive spleens. Continuous bars indicate median values (n = 3, Mann-Whitney test). N, naive mice; R, refractory mice.

A combination of splenectomy and liver macrophage depletion abrogates PTR

Given that neither splenectomy nor clodronate treatment was sufficient to prevent transfusion refractoriness, we explored whether other sites or compensatory mechanisms were involved in allogeneic platelet elimination in PTR. Using a double targeting strategy, we administered clodronate to splenectomized mice and compared the kinetics of PTR in these animals and sham-operated mice treated with clodronate (Figure 7). Targeting both pathways corrected PTR, with approximately half of the allogeneic platelets circulating 24 hours after transfusion. For the sham-operated mice, we observed as before (Figure 3B) a progressive recirculation of transfused platelets, but the animals were still in a refractory state. We verified that the latter animals still had alloantibodies binding to H-2^b platelets at the time of transfusion (supplemental Figure 4B). All these results showed that targeting liver and spleen macrophages in parallel is an effective strategy to circumvent PTR in alloimmune mice.

Figure 7.

Absence of both hepatic and splenic macrophages abrogates PTR for 24 hours after transfusion. Transfusion kinetics in alloimmune splenectomized or sham-operated alloimmune mice treated with clodronate (n = 3, nonlinear regression fit; P < .0001). PLT, platelet.

Discussion

This study investigated the fate of allogeneic platelets in the context of PTR. Our results showed alloantibodies to be the mediators of allogeneic platelet elimination. However, blocking the main antibody-dependent mechanisms did not correct the refractoriness. Nevertheless, we identified phagocytic cells in the liver and spleen (ie, KC and RPM, respectively), as being the final destination for allogeneic platelets recognized by anti-MHC class I antibodies in a murine model of PTR. Using an intravital approach, we observed cross talk between transfused platelets and macrophages within the splenic microenvironment. Finally, we showed that removal of both liver and spleen macrophages abrogates PTR.

This work definitively establishes the role of alloantibodies in mediating platelet clearance, thereby ruling out a cellular response in this model. Our findings are in opposition to a previous report identifying CD8 depletion as a means to abrogate refractoriness. This discrepancy may be explained by the different model based on a B-cell deficient mouse strain used in the latter study.⁹ However, even if anti–HLA class I antibodies are the primary mediators in the context of PTR, it would be premature to totally exclude a cellular response in patients. The presence of antibodies does not necessarily correlate with the development of a refractory state,³⁴ and previous assays targeting antibodies did not always lead to a good outcome.³⁵ These arguments support the idea that patients might present a cellular response in association with the humoral response in PTR.

To date, there exists no universal therapeutic strategy to overcome PTR. One study using IVIg found it to be ineffective to manage PTR,³⁶ which was in disagreement with several other reports.^37,38 The specific blockade of FcγR has been explored for ITP, where it demonstrated positive results in 1 report, with short-lived side effects.²⁶ Some authors found that complement-component C1q-binding antibodies were not predictive of refractoriness,³⁹ whereas others highlighted that complement might be an important barrier in PTR.⁴⁰ Agents targeting the platelet desialylation axis have proven to be effective in decreasing desialylation and platelet clearance in ITP.⁴¹ To explore this possibility, we blocked asialoglycoprotein receptors by injecting asialofetuin (or fetuin as the control) into alloimmune mice immediately before transfusion. Asialofetuin consists of desialylated bovine proteins, which act as a competitive inhibitor of these receptors. However, mice injected with asialofetuin displayed no difference in platelet elimination kinetics compared with that in animals receiving fetuin (supplemental Figure 5A). We also checked the efficacy of the inhibition by treating platelets with neuraminidase and transfusing them into an asialofetuin-treated mouse (supplemental Figure 5B-C). Although these results do not allow us to exclude the role of desialylation-dependent platelet clearance, blocking asialoglycoprotein receptors⁴¹ was clearly not effective to prevent PTR in alloimmune mice.

The only way to avoid PTR in our model was to target both liver and splenic macrophages. In the absence of KC, but with half of the RPM still present, we observed a temporary sequestration of all allogeneic platelets, followed by the recirculation of only one-third of them. Our intravital observations may provide a partial explanation for this finding because platelets were observed to be trapped/retained in the spleens of refractory mice. We can hypothesize that KC and RPM collaborate to efficiently remove allogeneic platelets but in the absence of KC, RPM alone are incapable of eliminating all transfused platelets.

The coexistence of multiple mechanisms of platelet clearance has already been described in a mouse model of platelet dysfunction, which showed signs of activation.⁴² In our model, we observed in vitro platelet activation induced by anti-MHC class I alloantibodies (supplemental Figure 6), suggesting that such a process could occur in vivo to facilitate platelet elimination. This potential activation could explain the behavior of transfused platelets once they enter the spleen of alloimmune mice. The spleen displays a complex and unique vascular organization where blood is released into an open area. The complex and unique vascular organization where blood is released into an open area allows the possible binding of molecules on the surface of activated platelets to diverse ligands43, 44, 45 under particular conditions, such as slower blood flow and discontinuous endothelial lining.^46,47 Using intravital microscopy, we observed that the platelets flowed, but not uniformly, because some were retained close to macrophages for different periods, even in naive mice. This finding points to unknown molecular interactions, which enable this type of “touch and go” interplay between platelets and RPM. In any case, the molecular mechanisms of macrophage–platelet interactions remain obscure, but they would seem to depart from those proposed in previous studies, which focused on autonomic agents such as adrenaline,⁴⁸ β1-receptor blockers,⁴⁹ or agonists.⁵⁰ It is also plausible that these transient interactions under basal conditions occur through platelet Glycoprotein Ib (GPIb) and von Willebrand factor on RPM, as has been described for KC.⁵¹

The present results might be useful for the management of HLA class I–immunized patients, indicating that splenectomy should be unlikely to correct PTR. Accordingly, this treatment has afforded variable outcome in patients, being successful in some cases but not in others.^52,53 Our in vivo blocking experiments further suggest that inhibiting antibody-dependent platelet elimination pathways will not suffice to abrogate PTR. Using combined therapy, it was possible to correct the refractoriness only in 1 case.⁵⁴ This emphasizes the need not only to develop predictive tests using patient sera,^55,56 but also to establish assays revealing the impact of an individual patient’s alloantibodies on platelets.

Our murine model oversimplifies the complexity of the HLA system, in addition to the fact that the alloimmune recipients are not thrombocytopenic at the time of platelet transfusion. PTR still has to be modeled in the context of onco-hematological therapy, in which various drugs and underlying pathologies may produce pleiotropic effects that could influence platelet clearance.⁵⁷ Although our means of abrogating refractoriness depended on macrophage depletion, an approach not directly translatable to human patients, we believe that innovative therapies could focus on these cells and prevent them from eliminating platelets. Reprogramming macrophages is emerging as a new tool in cancer immunotherapy,⁵⁸ with potential in transfusion medicine.⁵⁹ It remains to be determined whether platelets binding alloantibodies in vivo are functional and capable of stopping hemorrhage. Several reports have indicated that activated platelets are able to circulate and ensure hemostasis,^60,61 which suggests that targeting macrophages could be a valuable strategy. The answers to these questions could provide a new, ingenious method to avoid PTR.

In summary, this study shows that binding of alloantibodies to allogeneic platelets is necessary for platelet clearance. It also identifies the liver and spleen as graveyards for transfused platelets in PTR, where macrophages are the responsible for eliminating these platelets (supplemental Figure 7). Our findings suggest that compensatory or redundant mechanisms of allogeneic platelet elimination are operating in PTR. Therefore, novel therapies such as the present approach targeting hepatic and splenic macrophages will be required to overcome the refractoriness.

Conflict-of-interest disclosure: The authors declare no competing financial interests.