Platelet Toll-like Receptor 4-related Innate Immunity Potentially Participates in Transfusion Reactions Independent of ABO Compatibility: An ex Vivo Study


 Purpose: The role of platelet TLR4 in transfusion reactions remains unclear. This study analyzed platelet TLR4, certain DAMPs, and the effect of ABO compatibility on TLR4 expression after a simulated transfusion ex vivo.Methods: Donor red blood cells were harvested from a blood bank. Recipient blood from patients undergoing cardiac surgery was processed to generate a washed platelet suspension. Donor blood was added to the washed platelets at 1%, 5%, or 10% (v/v). Blood mixing experiments were performed using four groups: 0.9% saline control group (n = 31); M, matched blood type mixing (n = 20); S, uncross-matched ABO type-specific mixing (n = 20); and I, ABO incompatible blood mixing (n = 20). Platelet TLR4 expression was determined after blood mixing. Levels of TLR4-binding DAMPs (HMGB1, S100A8, S100A9, and SAA) and that of LPS-binding protein and endpoint proteins (TNF-α, IL-1β, and IL-6) in the TLR4 signaling pathway were evaluated.Results: The 1%, 5%, and 10% blood mixtures significantly increased TLR4 expression in three groups (M, S, and I; all P < 0.001) in a concentration-dependent manner. TLR4 expression did not significantly differ among the three groups (P = 0.148). HMGB1, S100A8, and S100A9 showed elevated levels in response to blood mixing; SAA, LPS-binding protein, TNF-α, IL-1β, and IL-6 did not. Conclusion: Blood mixing may elicit innate immune responses by upregulating platelet TLR4 and DAMPs unassociated with ABO compatibility, suggesting that innate immunity through TLR4-mediated signaling may induce transfusion reactions. The trial was retrospectively registered at Chinese Clinical Trial Registry (ChiCTR2100045606) with date of registration on 19 April 2021.


Introduction
Blood transfusions are critical interventions, particularly in patients undergoing hemorrhagic shock.
However, despite signi cant advancements in the safety of blood transfusions, these procedures are still associated with important risks. The mechanisms of immune reactions during blood transfusion are unclear, speci cally those of innate immune reactions. Transfusion reactions can result from transfusion with cross-matched blood, uncross-matched type-speci c blood, and ABO incompatible blood [1].
Immune reactions to blood transfusion involve complex interactions among various soluble factors and immune cells, including platelets [1].

Blood sampling
This study was approved by the institutional review board of Tri-Service General hospital (TSGHIRB 1-102-05-014, TSGHIRB 1-107-05-015), and written informed consent was obtained from the participants prior to enrollment. Rh-negative patients were excluded from this study. Red blood cells (unwashed) with a hematocrit of 55-60% were acquired from the blood bank of our hospital and stored in a cold room at 2-4 °C. Recipient blood samples were obtained from patients scheduled for cardiac surgery. Blood samples were obtained from an arterial catheter before anesthesia induction (Fig. 2). All samples were treated with a 1:9 volume of 3.8% sodium citrate solution as an anticoagulant. Blood transfusion was simulated ex vivo by performing blood mixing. Blood mixing reactions were segregated into four groups: 0.9% saline control group; group M, matched blood type mixing group; group S, uncross-matched ABO type-speci c mixing; group I, ABO-incompatible mixing group.
The primary outcome measure was the effect of blood mixing on platelet TLR4 expression in each of the three groups (M, S, and I) and the differences in TLR4 expression among the three groups (M, S, and I).
The secondary outcome measure was the effect of blood mixing on the levels of LPS-binding protein and DAMPs including HMGB1, SAA, S100A8, and S100A9. Additionally, the levels of molecules downstream of TLR4 signaling, such as TNF-α, IL-1β, and IL-6, were evaluated in each of the three groups (M, S, and I) Flow cytometry analysis of TLR4 expression Recipient whole blood was centrifuged at 37°C for 10 min at 200 × g. The upper phase (plasma) was carefully collected, whereas the lower phase (red cells) and interphase (buffy coat containing mainly leukocytes and a few platelets) were discarded. The upper phase was then centrifuged at 2,000 × g for 10 min and the remaining pellet was gently resuspended in platelet wash buffer. The suspension was centrifuged at 2,000 × g for 10 min to prepare washed platelets, which were resuspended in HEPESbuffered Tyrode's solution; the suspension was adjusted to a nal platelet count of 150,000-450,000 platelets/μL. Thereafter, recipient washed platelets without or with 0.9% saline added at 1%, 5%, or 10% (v/v) were used as controls, respectively (Fig. 3). Consequently, for blood mixing reactions, donor red blood cells were immediately mixed with recipient washed platelets at 1%, 5%, or 10% (v/v) and incubated at 37°C for 5 min (Fig. 4).

Quanti cation of TLR4 expression in washed platelets
We investigated platelet TLR4 expression in mixed samples in the presence of thrombin ( nal concentration: 0.2 U/mL) by incubating the samples at an ambient temperature of 23-26 °C for 5 min.
Thrombin is generated during tissue injury, such as cardiac surgery [9], and is a key component of the blood coagulation cascade and a potent stimulator of platelets. Thrombin is commonly used in sample preparation protocols for platelet analysis. To quantify TLR4 expression, the samples were stained with a saturating concentration of the anti-CD41a-FITC and anti-TLR4-PE monoclonal antibodies and incubated at 22-26 °C in the dark for 20 min. The samples were then xed with 1% paraformaldehyde at 4 °C for 30 min and analyzed using ow cytometry. FITC-labeled IgG1κ and PE-labeled IgG1κ served as background controls. Individual platelets were identi ed through side scatter (granularity characteristics) and anti-CD41a-FITC immuno uorescence in a logarithmic-scaled dot plot. The results are expressed as the mean uorescence intensity (MFI) of TLR4-PE expression, and reads from 10,000 platelets were collected for each sample ( Fig. 3; Fig. 4).

ELISA analysis
We then assessed the levels of LPS-binding protein and of certain DAMPs that may interact with TLR4 after blood mixing. Donor red blood cells were immediately added to recipient whole blood at 1%, 5%, or 10% (v/v) and incubated at 37°C for 30 min. The mixed blood was then centrifuged at 37°C for 10 min at 100 × g. The upper phase (plasma) was carefully collected, whereas the lower phase (red cells) and interphase (buffy coat containing mainly leukocytes and a few platelets) were discarded. The plasma was analyzed to determine the concentrations of HMGB1 (Aviva Systems Biology, San Diego, CA, USA), S100A8 (Circulex, MBL, Nagano, Japan), S100A9 (Circulex, MBL), SAA (Abnova Co., Taipei, Taiwan), and LPS-binding protein (Aviva Systems Biology) using ELISA kits according to the manufacturer's protocols.

Statistical analysis
One-way analysis of variance (ANOVA) was performed to compare demographic variables of the four groups (control, M, S, and I). Furthermore, ANOVA was performed to compare the levels of TLR4; LPSbinding protein; DAMPs including HMGB1, S100A8, S100A9, and SAA; and TLR4-regulated cytokines including TNF-α, IL-1β, and IL-6 in each of the three groups (M, S, and I) followed by Scheffé post-hoc test. Differences among the three groups (M, S, and I) over various concentrations were analyzed by two-way ANOVA. All tests were two-sided, with P < 0.05 considered as statistically signi cant. SPSS software (Version 20; SPSS, Inc., Chicago, IL, USA) was used for all analyses.

Results
No signi cant differences in demographic characteristics were observed among the four groups (Table  1). Figure 1 shows the results of ow cytometric analysis of platelet TLR4 expression. Figure 1 depicts uorescence dot plots representing TLR4 expression in isotype controls (Fig. 1A), unstimulated platelets (Fig. 1B), thrombin-stimulated platelets (Fig. 1C), and thrombin-stimulated platelets mixed with 10% donor blood (Fig. 1D); Fig. 1E shows an overlay of the histograms of unstimulated platelets, thrombinstimulated platelets, and thrombin-stimulated platelets mixed with 10% donor blood (Fig. 1E). Table 1 Demographic characteristics of recipients in 0.9% saline control and groups M, S, and I. Effect of TLR4 expression after the addition of 0.9% saline The 0.9% saline control groups were not exposed to red blood cells but were treated with 0.9% saline (Fig.  3). We examined the control groups to determine whether TLR4 expression levels could be determined in blood mixing experiments using a recipient washed platelet mixture containing 0.9% saline [1%, 5%, and The TLR4 expression levels did not signi cantly differ following stimulation by thrombin or the 0.9% saline control [1%, 5%, and 10% (v/v)] (P = 0.892, Fig. 3).
Effect of DAMPs and LPS-binding protein in the plasma HMGB1, S100A8, and S100A9 levels were signi cantly increased in response to blood mixing (Fig. 5).
Effect of pro-in ammatory cytokines downstream of the TLR4 signaling pathway in the plasma The levels of total LPS-binding protein involved in TLR4 signaling (P = 0.526) and endpoint proteins including total TNF-α, IL-1β, and IL-6 were not elevated after blood mixing (P = 0.998, P < 0.806, P < 0.87, respectively; Fig. 6).

Discussion
This study showed that after ex vivo blood mixing, TLR4 expression levels were upregulated in platelets in the matched (M), uncross-matched ABO type-speci c mixing (S), and ABO incompatible (I) groups. In addition to its crucial role in hemostasis and thrombosis, platelet TLR4 signi cantly contributes to amplifying in ammatory and immune responses [2]. This may also be true in the platelet TLR4-related transfusion immune and in ammatory responses [2]. Platelet TLR4 expression in uences innate immunity [5,7,8,10,11], leading to an adaptive immune response [3][4][5][6][7] and signi cant in ammation [12] as observed in transfusion reactions. Thus, platelet TLR4 may serve as a pathophysiological link between innate immunity and transfusion reactions.
TLR4 recognizes DAMPs [13], which are associated with host cell components that are released upon cell damage [5,14,15]. We found that certain DAMPs, including HMGB1, S100A8, and S100A9, were elevated after blood mixing, which is corroborated by previous reports [16-20] on blood transfusions (Fig. 5). For example, rst, red blood cell transfusion increases the susceptibility to lung in ammation through the release of HMGB1 and induces necroptosis in lung endothelial cells [19]. Second, stored human red blood cells contain soluble HMGB1, the levels of which are elevated during storage [16]. Third, salvaged blood analyses revealed sustained high levels of certain DAMPs, including S100A8 and S100A9. Above all, transfusion reactions may result from increased levels of TLR4 and certain DAMPs including HMGB1, S100A8, and S100A9 that bind with TLR4 (Fig. 5). Upregulation of TLR4 and certain DAMPs suggests that innate immunity through TLR4-mediated signaling can induce the transfusion reaction.
TLR4 expression upregulation and the elevated levels of some DAMPs (HMGB1, S100A8, and S100A9) were observed in response to blood mixing. However, this did not involve SAA and LPS-binding protein and did not lead to the release of pro-in ammatory cytokines downstream [8], such as TNF-α, IL-1β, and IL-6, after blood mixing (Fig. 6). These nding may be because of the following: rst, LPS-binding proteins is a soluble protein that is synthesized by hepatocytes and found in the blood [ . Third, these results may have occurred because of the short duration of blood mixing (30 min) during which time pro-in ammatory cytokines were not expressed; thus, the interaction between platelet TLR4 and certain DAMPs does not trigger this signaling pathway of pro-in ammatory cytokine release. Fourth, ex vivo blood mixing was not able to provide an adequate context in which to observe the complete range of platelet response to transfusion in vivo. Above all, although our results suggest that platelet TLR4 serves as a pathophysiological link between innate immunity and transfusion, the levels of pro-in ammatory cytokines downstream of TLR4 signaling, including TNF-α, IL-1β, and IL-6, were not increased remain unanswered.
Among the three groups (M, S, and I), we did not observe signi cant differences in TLR4 levels. The major variances among the three groups (M, S, and I) were the responses of antibodies recognizing transfused (foreign) antigens [22]. Human red blood cell membranes containing ABO system components are important in most blood transfusions. Individuals often produce antibodies (alloantibodies) against the alleles they lack within each system. Such antibodies are responsible for the most serious reactions to transfusions. However, ABO systems are not present on platelets [22]. In addition, surface antigens on red blood cells include Duffy, Kell, Kidd, MNS, and P systems, which are not present on platelets [22]. We found no differences in platelet TLR4 expression among the three groups (M, S, and I), which implies that reactions between antibodies and antigens did not play a role in inducing TLR4 expression.
TLR4 ligands recognize not only DAMPs but also pathogen-associated molecular patterns (PAMPs) [5,14,15,23], such as circulating LPS (endotoxin), which are associated with microbial pathogens [24]. PAMP signaling was less likely to participate in this response for the following two reasons: rst, during blood mixing, PAMPs such as circulating LPS from microbial pathogens were disregarded, as the patients did not have a pathogenic infection; second, after blood mixing, the levels of LPS-binding protein, which binds with PAMPs but not DAMPs, did not increase (Fig. 6) [14,15]. Therefore, blood mixing may initiate platelet TLR4 expression, which may trigger the innate immune system, likely independent of PAMPs.
The rationale for selecting patients undergoing cardiac surgery in this study was as follows: rst, blood transfusion is routinely required during cardiac surgery, and the preparation of surgery routinely requires cross-match testing of blood at least one day before surgery. Therefore, cross-matched red blood cells were available routinely and sent to operation room from the blood bank before anesthesia for us to obtain donor red blood cells. The red blood cells also used for heart-lung machine priming or stored until transfusion later; second, patients undergoing cardiac surgery require arterial catheterization for aggressive hemodynamic monitoring. The arterial catheterization also used for recipient blood sampling before anesthesia induction and operation of skin incision prevents contamination due to anesthetics and tissue damage, which may be a confounding factor affecting platelet function and activation (Fig. 2). Red blood cells from groups S and I were also obtained from the blood bank for other major surgeries in this hospital.
Our blood mixing procedure is clinically relevant. In this study, we performed ex vivo mixing of donor and recipient blood, which is similar to the cross-matching procedure. Cross-matching is performed prior to blood transfusion to determine whether donor blood is compatible (or incompatible) with recipient blood.
The mean body weight of subjects in this study was 66.8 kg; thus, the calculated total blood volume was approximately 4620 mL (estimated 7% of body weight). We used mixtures of 1-10%, which were equivalent to approximately 46.8 to 467.6 mL, respectively; these volumes are commonly used in clinical transfusions.
This study had two limitations. First, the baseline levels of HMGB1, S100A8, S100A9, SAA, LPS-binding protein, TNF-α, IL-1β, and IL-6 were not detected in the donor red blood cells. However, donor red blood cells were added to recipient whole blood at 1%, 5%, or 10% (v/v). The donor red blood cells have a hematocrit level of 55-60% and were standardized by our blood bank center. Thus, they would have contained less plasma than recipient whole blood. Therefore, the plasma volume of our donor red blood cells was far smaller than that of recipient whole blood. Second, we primarily used an ex vivo model because we included group I (ABO incompatible blood mixing), which could be harmful to patients if conducted in vivo. Therefore, the in vivo host innate immune responses were not evaluated. Platelets are not only directly involved in immune defense, but also assist and regulate several functions of innate immune cells. Platelets have also been shown to participate directly in the modulation of immune cell function by physically tethering to them or releasing platelet-derived microvesicles, lipid mediators, nucleosides, mitochondrial DNA, growth factors, cytokines, and chemokines. Platelets and their releasates have broader effects on differentiation, migration, phagocytosis, microbicidal activity, formation of extracellular traps, pathogen clearance, and cytokine response of innate immune cells [24].

Conclusions
Platelet TLR4 functions at the crossroads of thrombosis and the innate immune response. We found that allogeneic blood mixing may modulate the innate immune response by upregulating platelet TLR4 and DAMPs, including HMGB1, S100A8, and S100A9, which may bind to TLR4, thus suggest that platelet TLR4 links transfusion and innate immunity in blood mixing reactions. Considering the large number of circulating platelets, the potential interaction between platelet TLR4, and DAMPs and the induction of innate immune responses leading to transfusion reaction is possible. However, because TLR4 downstream pro-in ammatory cytokines (TNF-α, IL-1β, and IL-6) were not detected, it remains unclear whether TLR4 signaling leads to a transfusion reaction under in vivo conditions. This upregulation of TLR4 expression was not associated with ABO compatibility. These ndings suggest that TLR4 contributes to transfusion reactions that are unrelated to antibodies against red blood cell antigens.
Whether platelet TLR4 can be considered a novel prophylactic and therapeutic target in transfusion reactions or a new target to modulate innate immunity remains further studies. Figure 1 Flow cytometric analysis of platelet Toll-like receptor 4 (TLR4) expression. Individual platelet events are identi ed by their characteristic side-scatter properties (granularity; x-axis) and positive labeling using platelet-speci c monoclonal antibodies (CD41a-FITC; y-axis). Dot plot of the uorescence of isotype control (A), unstimulated platelets (B), thrombin-stimulated platelets (C), and thrombin-stimulated platelets with 10% donor blood (v/v) (D). Histogram of the uorescence of unstimulated (control) and thrombin-stimulated platelets (TLR4-PE; x-axis) (E). MFI, mean uorescence intensity.

Figure 2
Key steps of general anesthesia and cardiac surgery performed in this study. First, essential hemodynamic monitoring was conducted. Second, cross-matched red blood cells were sent to the operation room. Then, arterial catheterization was performed to obtain donor red blood cells and recipient blood samples. Anesthetics were administered, and skin incision was performed. Blood sampling before anesthesia induction and skin incision can theoretically eliminate confounding factors caused by anesthetics and tissue damage-related factors in the blood circulatory system.

Figure 3
Effect of Toll-like receptor 4 (TLR4) expression after 0.9% saline was mixed with washed platelets (n = 31; n = 10, group M; n = 10, group S; n = 11, group I). MFI, mean uorescence intensity; vol, volume; WP, washed platelets. Data are presented as the mean ± SD.  Effect of DAMPs (HMGB1, S100A8, S100A9, and SAA) in the plasma prepared from mixing donor red blood cells and recipient whole blood on total blood mixing, matched blood type mixing (Group M), uncross-matched ABO type-speci c mixing (Group S), and ABO incompatible blood mixing (Group I).
HMGB1, high mobility group box-1; SAA, serum amyloid A; RBCs, red blood cells; vol, volume; WP, washed platelets. Data are expressed as the mean ± SD. *P < 0.05. Effect of LPS-binding protein and pro-in ammatory cytokines in the plasma, prepared from mixing donor red blood cells and recipient whole blood, on matched blood type mixing (Group M), uncross-matched ABO type-speci c mixing (Group S), and ABO incompatible blood mixing (Group I). RBCs, red blood cells; vol, volume; WP, washed platelets. Data are expressed as the mean ± SD. *P < 0.05.