Novel Mouse Model for Studying Hemostatic Function of Human Platelets

Abstract

Objective:

Platelets are critical to the formation of a hemostatic plug and the pathogenesis of atherothrombosis. Preclinical animal models, especially the mouse, provide an important platform to assess the efficacy and safety of antiplatelet drugs. However, these studies are limited by inherent differences between human and mouse platelets and the species-selectivity of many drugs. To circumvent these limitations, we developed a new protocol for the adoptive transfer of human platelets into thrombocytopenic NOD/SCID mice, i.e. a model where all endogenous platelets are replaced by human platelets in mice accepting xenogeneic tissues.

Approach and Results:

To demonstrate the power of this new model, we visualized and quantified hemostatic plug formation and stability by intravital spinning disk confocal microscopy following laser ablation injury to the saphenous vein. Integrin α_IIbβ₃-dependent hemostatic platelet plug formation was achieved within ~30 seconds after laser ablation injury in humanized platelet mice. Pre-treatment of mice with standard dual antiplatelet therapy (DAPT, Aspirin + Ticagrelor) or PAR1 inhibitor, L-003959712 (an analog of vorapaxar), mildly prolonged the bleeding time (BT) and significantly reduced platelet adhesion to the site of injury. Consistent with findings from clinical trials, inhibition of PAR1 in combination with DAPT markedly prolonged BT in humanized platelet mice.

Conclusion:

We propose that this novel mouse model will provide a robust platform to test and predict the safety and efficacy of experimental antiplatelet drugs and to characterize the hemostatic function of synthetic, stored and patient platelets.

Graphical Abstract

INTRODUCTION:

Platelet-mediated thrombosis contributes to the pathogenesis of ischemic heart disease, ischemic stroke and venous thromboembolism and thus represents a significant global health concern. It is estimated that thromboembolic conditions were responsible for 1 in 4 deaths worldwide in 2010¹. Antiplatelet agents prevent primary and secondary arterial thrombotic events by blocking the activation of platelet agonist receptors, suppressing feedback signaling or preventing α_IIbβ₃ integrin engagement. Although antiplatelet agents represent an important tool for minimizing thrombotic events, they also compromise hemostasis and thus increase the risk of excessive postoperative bleeding, gastrointestinal bleeding and intracranial hemorrhage^2–4. Hence, preventing thrombosis while maintaining hemostasis represents a challenging and persistent dilemma for clinicians and scientists.

Platelet activation at sites of vascular injury is strongly dependent on G protein-coupled receptors, including the thrombin receptors PAR1 and PAR4, the thromboxane (Tx) A2 receptors, and the ADP receptors P2Y₁ and P2Y₁₂⁵. Hemostatic plug/thrombus formation requires platelet-platelet cohesion, mediated by the major platelet integrin receptor, α_IIbβ₃. Dual antiplatelet therapy (DAPT), consisting of aspirin (inhibits cyclooxygenase-mediated TxA2 formation) and a P2Y₁₂ antagonist, is the current standard of care for patients following stenting or for secondary prevention of ischemic stroke ^6–10. Direct inhibitors to α_IIbβ₃ are used in high risk patients only as this treatment strategy comes with a very high risk of bleeding. In 2014, the PAR1 inhibitor, vorapaxar, was FDA approved for clinical use¹¹. As shown in the TRA 2ºP-TIMI 50 clinical trial, co-administration of vorapaxar with DAPT significantly reduced the risk of cardiovascular complications in patients with a history of myocardial infarction, ischemic stroke or peripheral artery disease^12,13. However, this triple anti-platelet therapy also led to a significant increase in moderate to severe bleeding in patients ^12–14. The marked effect of vorapaxar on hemostasis was not anticipated in part because data from preclinical animal models was limited¹⁵.

Mice are a convenient, cost effective and reliable investigational tool commonly used in the early stages of drug development. However, inherent differences between human and mouse platelets, including the absence on mouse platelets of important surface receptors, such as the IgG receptor, FCγRIIA, and the thrombin receptor, PAR1^16,17, limit the applicability of mouse models for the study human platelet biology. These species differences highlight the need for a humanized mouse model capable of directly studying in vivo human platelet function. Previous attempts at humanizing mice have employed NOD/SCID mice, which are capable of grafting xenogeneic tissues by virtue of their immune deficiency^18,19. To date, investigation of in vivo human platelet function has been focused to studies on platelet clearance^19–22, GPVI receptor shedding²³ or transient platelet activation ²⁴. Studies assessing the hemostatic function of human platelets in mice are impossible to interpret as endogenous (mouse) platelets will outcompete transfused (human) platelets for ligand binding (e.g. VWF, collagen, fibrin) at injury sites due to the relatively high ratio of endogenous:transfused platelets²⁵.

Here, we present a powerful and reliable method for the assessment of the hemostatic function of human platelets in mice depleted of endogenous platelets. We performed antibody-mediated depletion of platelets in mice, a widely used method to determine the role of platelets in specific physiological and pathological functions. This method was used to show a role for platelets in angiogenesis²⁶, inflammatory hemostasis²⁷, vascular integrity in cancer²⁸, tumor metastasis²⁹ and liver damage³⁰. Our new protocol is based on our previously published methods for adoptive platelet transfer^31,32 into platelet depleted mice and real-time assessment of bleeding time and platelet/fibrin accumulation at the site of injury³³. To generate humanized platelet mice, NOD/SCID mice were rendered thrombocytopenic by infusion of anti-GPIbα antibodies, and subsequently transfused with human platelets and human VWF (Humate-P). Mice required pre-treatment with a platelet-activating factor (PAF) receptor antagonist to circumvent a systemic shock reaction associated with platelet depletion of NOD/SCID mice. To demonstrate the power of this new model, we quantified bleeding risk, platelet adhesion and hemostatic plug formation in mice treated with an inhibitor to PAR1, a receptor that is expressed on human but not murine platelets. We propose that this model will be of great value to (1) evaluate the efficacy and safety of novel antiplatelet agents, (2) test the in vivo function of synthetic and stored platelets, and (3) identify and/or characterize platelet function defects in patients with inherited/acquired platelet disorders and unexplained bleeding disorders.

MATERIALS AND METHODS:

The data that support the findings of this study are available from the corresponding authors upon reasonable request.

Mice:

NOD.CB17-PRkdc^scid/J mice (stock #: 001303) were obtained from The Jackson Laboratory and housed in the mouse facility of the University of North Carolina at Chapel Hill. Our study is limited to male mice due to the relatively small fat pads overlaying the saphenous vein compared to female mice. Excess fat tissue complicates the surgical preparation needed for our model. All experimental procedures were approved by the Animal Care and Use Committee of the University of North Carolina at Chapel Hill.

Depletion of Endogenous Mouse platelets:

Male NOD/SCID mice (8-12 weeks of age) received 10 mg/kg of platelet-activating factor receptor (PAF-R) antagonist, WEB-2086 (Santa Cruz), via oral gavage 30 minutes prior to depletion of endogenous mouse platelets by intravenous injection of 1 mg/kg of antibodies for platelet depletion (anti-GPIbα, R300, Emfret Analytics) 16-20 hours prior to transfusion of human platelets.

Human Platelet Isolation:

Blood collection from healthy donors was performed in accordance with a protocol approved by the Institutional Review Board at the University of North Carolina at Chapel Hill. Blood was obtained via venipuncture and collected in vacutainers containing 3.8% sodium citrate (BD Biosciences). Blood was centrifuged at 120 x G for 15 minutes at 25°C (without brake) in a centrifuge equipped with a swinging bucket rotor. Platelet rich plasma (PRP) was collected and transferred to a conical tube before addition of ½ volume of Tyrode’s buffer (137 mM NaCl, 12 mM NaHCO₃, 2.0 mM KCl, 0.3 mM Na₂HPO₄, 1 mM MgCl₂, 5mM N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid, 5 mM Glucose, pH 7.3) to the PRP. The final volume of dilute PRP was supplemented with 0.03 units/ml apyrase (Sigma-Aldrich) and centrifuged at 100 x G for 15 minutes at 25°C (with brake) to pellet contaminating red blood cells and white blood cells. PRP was supplemented with 1 μg/ml prostacyclin for 5 minutes (Cayman Chemical) and centrifuged at 650 x G for 15 minutes at 25°C (with brake). The platelet pellet was gently resuspended in Tyrode’s buffer and analyzed by flow cytometry on an Accuri C6 flow cytometer (BD Biosciences). Platelet concentration was adjusted to 3 x 10⁹ platelets/ml by addition of an appropriate volume of Tyrode’s buffer. Washed platelets were allowed to “rest” on the benchtop for 45-60 minutes prior to transfusion.

Human Platelet Transfusion:

Approximately 200-250 μl of washed human platelets, at a concentration of 3 x 10⁹/ml, were treated with 1 μg/ml prostacyclin for 5 minutes prior to incubation with 125 ng/ml CellTrace^™ calcein green-AM (ThermoFisher Scientific) for 5 minutes before being slowly loaded into a sterile 29G insulin syringe with zero dead space (Comfort Point, Excel Int., Korea). Platelets were infused at an approximate rate of 5 μL/second, via timed injections by the investigator, into the lateral tail vein of a NOD/SCID mouse that had been depleted of endogenous platelets 16–20 hours prior, as described above. The mouse was allowed to recover for 30 minutes in a heated cage prior to anesthesia with isoflurane and collection of 50 μl of blood from the retro-orbital plexus. Human platelet concentration and mean calcein green fluorescence intensity was determined from the sampled mouse blood, as was the absence of mouse platelets by staining with an AlexaFluor647-labeled antibody specific for mouse GPIX (Clone: XiaB4, Emfret Analytics). A second preparation of washed and calcein green-labeled human platelets (50-100 μl) were infused into the same recipient mouse to achieve a concentration of 2.0-3.5 x 10⁸ human platelets/ml of mouse blood. The number of circulating human platelets was determined by flow cytometric counting of calcein green-positive events in whole mouse blood collected ~15 minutes after the second transfusion and immediately prior to sacrifice, using the following formula: (calcein green events/μl of dilute mouse blood) x (sample dilution factor) x (1000 μl) = human platelets/ml of mouse blood. Mice were excluded from the study if the circulating human platelet count did not fall within the desired range of 2.0-3.5 x 10⁸ human platelets/ml.

Intravital Saphenous Vein Laser-induced Injury Model:

Mice were prepared for surgery as described recently³³. An AlexaFluor647-labeled antibody against fibrin (2 μg per mouse) (gift from Dr. Rodney Camire) and Humate-P^® (50 u/kg) (CSL Behring) were administered intravenously prior to imaging. The saphenous vein was injured by two maximum-strength 532-nm laser pulses (70 μm; 100 Hz; duration, 7 ns; 10 ms pulse interval) (Ablate! photoablation system; Intelligent Imaging Innovations, Denver, CO, USA). Images were captured on a Zeiss Axio Examiner Z1 microscope equipped with a 20X water immersion objective lens (Zeiss, Jena, Germany) with a numerical aperture of 1 and a working distance of 1.8 mm and a Yokogawa confocal scanning unit, CSU-W (Yokogawa Electric Co.). Fluorescence excitation was achieved with either a multi-color LED light source (Lumencor) for epifluorescence or 488, 561, 640 nm lasers for confocal imaging (Intelligent Imaging Innovations). All images were captured with an Orca Flash 4.0 camera (Hammamatsu). Data was analyzed with SLIDEBOOK 6.0 (Intelligent Imaging Innovations).

Flow Cytometry:

Human platelet activation was determined in heparinized whole blood drawn from the retro-orbital plexus of humanized platelet mice. Briefly, α_IIbβ₃ activation was detected using an AlexaFluor647-conjugated antibody (Clone: PAC1, Biolegend) while an AlexaFluor647-conjugated polyclonal antibody against human p-selectin (a generous gift from Rodger McEver) was used for quantification of α-granule secretion. Humanized platelet mouse blood was stimulated or not with 30 μM adenosine 5′-diphosphate (ADP; Sigma-Aldrich) or 5 μM PAR1-specific activating peptide (PAR1-AP), SFLLRN-NH₂ (Sigma-Aldrich), for 10 minutes in the presence of 2 mM CaCl₂ before analysis by flow cytometry.

Microfluidics Studies:

Ex vivo microfluidics studies were performed as previously described³⁴. Briefly, heparinized whole blood was pre-incubated with human BD FC block (BD Biosciences) prior to labeling of human platelets with FITC-conjugated anti-CD41b (Clone HIP2). Blood was perfused at venous shear rate (400s⁻¹) though a microfluidics chamber coated with a strip of fibrillar type I collagen (200 μg/ml) (Chrono-Log, Haverton, PA) and visualized on a Nikon TE300 microscope (Nikon, Japan) equipped with an Orca Flash 4.0 camera. Data was analyzed with NIS-Elements AR 4.13 software (Nikon) to quantify platelet accumulation.

Statistics:

All data passed the D’Agonstino & Pearson omnibus test for normality, unless otherwise indicated in the figure legends. Statistical significance among multiple groups was assessed by 2-way ANOVA with Bonferroni post-hoc test. Two-group comparisons of parametric data were analyzed using a 1-way ANOVA with Bonferroni post-hoc test or a 2-tailed Student’s t-test, non-parametric data was analyzed using the Kuskal-Wallis test with Dunn’s Multiple Comparison post-hoc test. Statistical analyses were performed using Prism 5 software (GraphPad, La Jolla, CA). A P-value of .05 or less was considered significant. All data are presented as mean ± standard error of the mean (SEM), unless otherwise indicated.

RESULTS:

Antibody-mediated targeting of the GPIbα receptor is a well-established method used by us and others to deplete circulating platelets in mice³¹. Platelet depletion by this method is quick, efficient and occurs without obvious side effects in wild-type (WT) mice. Efficient platelet depletion without side effects following injection of antibodies against the human interleukin-4 (hIL4) receptor was also observed in mice expressing a chimeric GPIb/hIL4R receptor on platelets³². This latter model further allows for the adoptive transfer of WT or other transgenic platelets, which do not express the hIL4 receptor. To establish a similar adoptive transfer method for human platelets, we first induced thrombocytopenia in NOD/SCID mice by injection of antibodies against murine GPIbα. Antibody-treated mice exhibited a >99% decrease in circulating platelet count but surprisingly, also suffered from acute, severe hypothermia (Figure 1A,B) and lethality (Figure 1C). The acute systemic reaction observed in NOD/SCID was reminiscent of that reported for WT mice treated with antibodies directed against the major platelet integrin receptor, α_IIbβ₃³⁵. Mechanistic work identified a critical role for platelet-activating factor (PAF) in the development of anti-α_IIbβ₃ antibody-induced systemic reactions³⁶. Thus, we pretreated NOD/SCID mice with the PAF-receptor (PAF-r) inhibitor, WEB-2086, 30 minutes prior to anti-GPIbα antibody injection. Indeed, PAF-r inhibition protected NOD/SCID mice from severe hypothermia and death while still allowing for clearance of endogenous mouse platelets (Figure 1A,B,C). Based on these findings, subsequent studies were performed in NOD/SCID mice pre-treated with WEB-2086.

Figure 1: PAF receptor inhibition prevents crisis in NOD/SCID mice during antibody-mediated platelet depletion and allows for stable transfusion of human platelets.

(A, B) Platelet count (A, n = 3) and body temperature (B, n = 3) in NOD/SCID mice after injection of platelet-depleting antibodies targeting GPIba in the absence (black bars) or presence (red bars) of PAF receptor inhibitor (WEB-2086). (C) Kaplan-Meier curve of NOD/SCID mortality following antibody-mediated platelet depletion in the absence (black line, n = 3) or presence (red line, n = 20) of WEB-2086. (D) Representative flow cytometry plot of NOD/SCID whole blood following platelet depletion and transfusion of labeled human platelets. (E) Human platelet counts at indicated timepoints following transfusion (T1 = transfusion 1, T2 = transfusion 2) (n = 5-10). (F) Flow cytometric quantification of α_IIbβ₃ integrin activation (Pac1-binding, n = 5-8) and α-granule secretion (P-selectin expression, n = 4-5) in transfused human platelets activated ex vivo with the indicated agonists. (Mean ± SEM; all data are considered to be non-parametric), *** p < 0.001

After establishing the conditions for depletion of endogenous platelets in NOD/SCID mice, we established a protocol for transfusion of human platelets at physiological concentrations. Human platelets were obtained from healthy donors, washed and treated as described in Materials and Methods. Mice were transfused via tail vein injection with calcein green-labeled human platelets and allowed to rest for 45 minutes before a second transfusion of human platelets to achieve the desired concentration of 2.0-3.5 x 10⁸ circulating platelets per ml of blood. Human and mouse platelets in the circulation of NOD/SCID mice were quantified by flow cytometric analysis (Figure 1D). While no mouse platelets were detected, human platelets circulated at a physiological concentration for at least 4 hours after transfusion (Figure 1E). Significant numbers of the transfused human platelets remained in circulation 20 hours later. The ability of transfused human platelets to respond to agonist stimulation was verified by flow cytometry of whole blood collected from humanized NOD/SCID mice. Ex vivo analysis demonstrated robust integrin activation and granule secretion in human platelets transfused into NOD/SCID mice in response to ADP or PAR1-activating peptide stimulation (Figure 1F).

Next, microfluidics chamber experiments were performed to determine whether human platelets show similar adhesive function in mouse blood when compared to human blood. Anticoagulated whole blood from a healthy volunteer and from a NOD/SCID mouse following adoptive human platelet transfer was perfused over a collagen surface at a shear rate of 400s⁻¹. No significant differences in the adhesion kinetics between the two whole blood samples containing human platelets were observed (Supplemental Figure I), suggesting that human platelets interact similarly with murine and human plasma components.

To determine the hemostatic function of transfused human platelets, we next subjected mice to our recently established saphenous vein laser-induced injury model³³. Mice were administered fluorescently labeled anti-fibrin-Alexa647 antibodies (2 μg/mouse) to visualize thrombin-mediated fibrin formation³³. Since human GPIbα interacts better with human VWF than endogenous murine VWF³⁷, mice also received 50 u/kg of human VWF/FVIII (Humate-P^®). We induced injury to the saphenous vein by laser ablation 30 seconds (s) after the start of time-lapse image acquisition (injury 1) and observed subsequent blood loss and accumulation of human platelets and fibrin. The same site was re-injured at 3:30 min. (injury 2) and 6:30 min. (injury 3) to generate lesions of increased severity³³. Time to hemostasis, plug stability, platelet accumulation and fibrin generation were monitored. Accumulation of human platelets after the initial injury was rapid and robust (Video 1, Figure 2C,D), resulting in the cessation of bleeding (28s ± SEM 1.3s) after initial laser ablation (Figure 2A). Bleeding times for injury 2 (39.4s ± SEM 2.2s) and injury 3 (37.3s ± SEM 2.6s) were slightly but significantly prolonged (Figure 2A). Injury size, quantified by measuring fibrin-ring area upon hemostasis, was ~600 μm² following initial laser ablation (injury 1) and ~700 μm² for injuries 2 and 3 (Figure 2B). Following cessation of blood loss platelets continued to accumulate at injury sites, within the lumen of the vessel and downstream of the injury, in the direction of blood flow (Video 1, Figure 2C,D). Three-dimensional growth was most readily apparent after the 2^nd and 3^rd injuries (Video 1, Figure 2C,D, Figure 3D, Video 2), indicative of platelet activation and recruitment induced by second wave mediators (i.e. ADP, Thromboxane A2) released from adherent platelets. Fibrin accumulation increased in a stepwise fashion upon reinjury of the vessel and additional blood loss (Figure 2C,E). Administration of Humate-P^® significantly reduced bleeding time following laser injury (Supplemental Figure IIA). Treatment of mice with the human-specific α_IIbβ₃ integrin antagonist, abciximab, led to a marked impairment of hemostasis at sites of laser injury (Supplemental Figure IIB).

Figure 2: Human platelets form hemostatic plugs in NOD/SCID mice.

(A, B) Bleeding times (A) and lesion sizes (B) at sites of laser-induced injury of the saphenous vein. (C) Representative images of injury sites following laser ablation at t = 00:30 min (Injury 1), t = 3:30 min (Injury 2) and t = 6:30 min (Injury 3) depicting accumulation of platelets (green) and fibrin (red) (yellow lines approximate vessel wall location). (D, E) Quantification of platelet (D) and fibrin (E) accumulation at injury sites (n = 16). (Mean ± SEM), * p < 0.05, ** p < 0.01, *** p < 0.001

Figure 3: Dual anti-platelet therapy impairs adhesive function of transfused human platelets and increases the incidence of uncontrolled bleeding.

(A, B) Bleeding times (A) and lesion sizes (B) at sites of laser-induced injury of the saphenous vein in control (black) and DAPT-treated (green) mice. (# = percent of injuries in DAPT-treated mice that did not achieve hemostasis). (C) Representative images of injury sites in control and DAPT-treated mice at conclusion of observation period; platelets (green), fibrin (red)(yellow lines approximate vessel wall location). (D) Representative 3-dimensional reconstructions from confocal images of hemostatic plugs in control and DAPT-treated mice; platelets (green), fibrin (red). (E, F) Quantification of platelet (E) and fibrin (F) accumulation at injury sites of control (green/red traces, n = 16) and DAPT-treated mice (blue/purple traces, n = 21)(Injury sites with uncontrolled bleeding were excluded from analysis) (Mean ± SEM; data in panel A is non-parametric), * p < 0.05

We next evaluated our model in the context of dual antiplatelet therapy (DAPT), the standard of care for long-term prevention of myocardial infarction. Platelets were isolated from healthy donors and treated with the irreversible COX-1 inhibitor, acetylsalicylic acid, before transfusion. To confirm COX-1 inhibition, washed platelets were activated with arachidonic acid in standard aggregometry. Aggregation was impaired in acetylsalicylic acid-treated but not vehicle-treated human platelets (Supplemental Figure IIIA). Transfused mice also received an intravenous bolus of the P2Y₁₂ inhibitor, ticagrelor (5 mg/kg), 20 minutes prior to intravital microscopy. Ex vivo analysis confirmed impaired α_IIbβ₃ activation (PAC-1 binding) and granule secretion (anti-P-selectin binding) in platelets from ADP-stimulated samples obtained from ticagrelor-treated mice (Supplemental Figure IIIB). Following laser injury, adhesion of DAPT-treated human platelets and the formation of stable hemostatic plugs were delayed when compared on control mice (Video 3, Figure 3A). Mean bleeding time in DAPT-treated mice was prolonged for injury 1 (89.8s ± SEM 21.8s) compared to control mice, in part due to the incidence of uncontrolled bleeding, (5.8% of all injuries). Bleeding time trended towards being longer, but did not reach statistical significance in DAPT-treated mice for Injury 2 (66.4 ± SEM 15.2s) or Injury 3 (51.3 ± SEM 3.2s), compared to controls (Figure 3A). Injury sizes in the DAPT-treated group were not significantly different from controls (Figure 3B). Representative images illustrate the absence of intravascular hemostatic plug growth in DAPT-treated mice (Figure 3C). Spinning disk confocal microscopy yielded 3D computer-generated reconstructions of injury sites from control and DAPT-treated mice, further illustrating differences in hemostatic plug volume (Video 4, Figure 3D). Quantification of platelet and fibrin accumulation is shown in Figures 3E and 3F, respectively. These data are in agreement with published data demonstrating impaired secondary platelet recruitment under flow and increased risk of bleeding in patients receiving DAPT^3,4,6,8.

To test if our novel humanized platelet model can be used to better evaluate the safety of novel antiplatelet drugs, we next evaluated the effect of PAR1 receptor inhibition by a vorapaxar analog, L-003959712 (formerly SCH 602539, Merck & Co., Kenilworth NJ), on platelet adhesion and hemostatic plug formation, both in the presence and absence of DAPT. NOD/SCID mice, depleted of endogenous platelets, were gavaged with L-003959712 (15 mg/kg) 1 hour prior to transfusion of human platelets. Impaired platelet activation in whole blood stimulated with the PAR1-specific activating peptide (PAR1-AP), SFLLRN-NH₂, confirmed successful PAR-1 inhibition in L-003959712-treated mice (Supplemental Figure IIIB). Bleeding times in L-003959712-treated mice were 31.6s (± SEM 2.5s), 59.2s (± SEM 3.6s) and 64.2s (± SEM 4.4s) for injuries 1, 2 and 3, respectively, compared to the control mice (Video 5, Figure 4A). Uncontrolled bleeding was not observed in L-003959712-treated mice. Injury sizes did not differ from control mice (Figure 4B). Platelet accumulation was robust but lacked intravascular growth as shown by representative widefield images (Figure 4C) and 3D reconstructions (Video 6, Figure 4D). Consistently, platelet accumulation reached a plateau following cessation of bleeding (Figure 4E), while continuous platelet accumulation was observed in control mice. Fibrin accumulation was similar to control mice after injury 1, trended higher for injury 2 and was significantly higher after injury 3 (Figure 4F).

Figure 4: PAR1 inhibition on human platelets by L-003959712 minimally prolongs bleeding time without increasing the incidence of uncontrolled bleeding.

(A, B) Bleeding times (A) and lesion sizes (B) at sites of laser-induced injury of the saphenous vein in control (black) and L-003959712-treated mice (purple). (# = percent of injuries in L-003959712-treated mice that did not achieve hemostasis) (C) Representative image of injury site in a L-003959712-treated mouse at conclusion of observation period; platelets (green), fibrin (red) (yellow lines approximate vessel wall location). (D) Representative 3-dimensional reconstruction from confocal images of injury site from L-003959712-treated mouse; platelets (green), fibrin (red). (E, F) Quantification of platelet (E) and fibrin (F) accumulation at injury sites of control (green/red traces) and L-003959712-treated mice (blue/purple traces, n = 19) (Mean ± SEM; data in panel A is non-parametric), * p < 0.05, *** p < 0.001

When given in combination with DAPT, L-003959712 treatment significantly increased the incidence of uncontrolled bleeding after injury 1 (50%) and injury 2 (16.7%) (Video 7, Figure 5A). Mean bleeding times for DAPT+L-003959712 (DAPT+L) treated mice were increased to 337.9s (± SEM 40.1s), 141.5 (± SEM 53.9s) and 65.8s (± SEM 10.1s) for injuries 1, 2 and 3, respectively. Despite the prolonged BT for DAPT+L-treated mice, platelet accumulation following injury 1 was robust when hemostasis was achieved (Figure 5C,E). Time-lapse videos illustrate the delayed accumulation of platelets within the injury site as well as a tendency for large, towering platelet plugs extending beyond the exterior of the vessel before bleeding was halted (Video 7, Video 8, Figure 5D). Fibrin accumulation was increased in DAPT+L-treated mice (Figure 5D) and reached statistical significance following injury 3 (Figure 5F). Platelet and fibrin accumulation for sites with uncontrolled bleeding peaked at ~5-6 minutes after initial injury and remained relatively constant until the end of observation period (Video 9, Figure 5C, Supplemental Figure IVA,B). Hemostatic plug stability was compromised in DAPT+L-treated mice as demonstrated by the occurrence of re-bleeding after primary hemostasis and the accumulation of platelets in the extravascular space (Video 9, Figure 5C,E). Hemostatic plugs became unstable and re-opened approximately 30s after primary hemostasis of injury 1 (Video 7). On average, bleeding was observed for about 30-60s after re-opening of injury 1. The occurrence of re-bleeding contributed to the transient drop in platelet intensity at the 1.6-minute mark of Figure 5E (red arrow). Injury sizes were comparable with control mice for injury 1 (~650 μm²) but significantly smaller for injury 2 (~450 μm²) and trending smaller for Injury 3 (~575 μm²) (Figure 5B), a possible consequence of enhanced fibrin accumulation during prolonged bleeding (Figure 5D,F).

Figure 5: Triple antiplatelet therapy with DAPT and L-003959712 (DAPT+L) significantly impairs hemostasis and increases the incidence of uncontrolled bleeding.

(A, B) Bleeding times (A) and lesion sizes (B) at sites of laser-induced injury of the saphenous vein in control (black) and DAPT+L-treated mice (red). (# = percent of injuries in DAPT+L-treated mice that did not achieve hemostasis) (C) Representative images of DAPT+L-treated injury sites at conclusion of observation period; platelets (green), fibrin (red). (D) Representative 3-dimensional reconstruction from confocal images of hemostatic injury site from DAPT+L-treated mouse; platelets (green), fibrin (red). (E, F) Quantification of platelet (E) and fibrin (F) accumulation at injury sites of control (green/red traces) and DAPT+L-treated mice (blue/purple traces, n = 9) (Sites with uncontrolled bleeding were excluded from analysis, red arrow indicates preponderance of spontaneous re-bleeding) (Mean ± SEM; data in panel A is non-parametric), * p < 0.05, ** p < 0.01

DISCUSSION:

Here we describe a new method for testing the hemostatic activity of human platelets in mice. Our method is based on a unique adoptive transfer approach, where purified human platelets and human VWF are transfused into immunocompromised mice rendered thrombocytopenic by antibody infusion, followed by real-time imaging of hemostatic plug formation and in situ spinning disk confocal microscopy. Using the adoptive transfer approach, we generated mice with circulating human platelet counts, consistent with physiologically relevant levels, which were completely devoid of endogenous murine platelets. Human platelet adhesion and hemostatic plug formation in these mice was quantified following laser ablation injury, both in the presence and absence of clinically used antiplatelet drugs. Consistent with observations from clinical trials^13,14, inhibition of PAR1, a receptor expressed on human but not murine platelets, significantly impaired hemostatic plug formation and increased the risk of bleeding in humanized platelet mice that also received DAPT. We propose that this method will provide an invaluable tool for studies on novel antiplatelet agents and the characterization of in vivo function of synthetic platelets, stored platelets or platelets from patients with a platelet function disorder.

NOD/SCID mice have been used to study the function human platelets in vivo^19,38. Human platelets are either produced by engrafted CD34+ stem cells or transfused megakaryocytes^39–41 or they are transfused acutely, similar to what we describe for our model. Stem cell transplantation is done in lethally irradiated mice and thus leads to mice with circulating human but not murine platelets. However, platelet counts in this model do not seem to reach physiological levels and the time window for performing experiments is quite short. Compared to the stem cell approach, engrafted megakaryocytes produce human platelets over an extended period of time. However, human platelet counts are also low and functional studies are complicated due to the presence of endogenous cells. Transfusing freshly isolated human platelets can solve many limitations of the stem cell/megakaryocyte approaches. However, platelet function studies have been done in the presence of murine platelets, a major limitation for determining the hemostatic function of these cells. Humanizing VWF in immunocompromised mice provides a viable strategy to limit the contribution of endogenous platelets to physiological or pathological thrombus formation, as murine GPIb interacts very poorly with human VWF ⁴². However, even this elegant new model cannot rule out confounding effects by endogenous platelets, as (1) murine platelets may still interact with humanized VWF via α_IIbβ₃ integrin, (2) GPIb can mediate platelet adhesion via interaction with adhesion molecules other than VWF, and (3) the GPIb-VWF interaction is less relevant under low shear conditions ^43,44. Critical to our new model is the replacement of all endogenous murine platelets with freshly isolated human donor platelets. To do so, we modified our recently described adoptive platelet transfer approach, where murine platelets are depleted by antibody targeting of the GPIbα receptor prior to infusion of donor platelets^25,32. Surprisingly, this approach caused systemic shock and death in NOD/SCID mice, unless the mice were pretreated with the PAFr inhibitor, WEB-2086. A similar phenotype was described for WT mice treated with antibodies against the major platelet integrin receptor, α_IIbβ₃^35,36. Critical to platelet antibody-induced shock are mast cells and a cytokine storm, the latter being inhibited in the absence of PAFr signaling⁴⁵. At this point we do not know why anti-GPIbα antibodies induce systemic complications in immunocompromised mice. Mechanistic studies demonstrated that thrombocytopenia induced by antibodies to α_IIbβ₃, but not those to GPIbα, depends on the Fc part of the antibody³⁶. Mice deficient in activating Fc receptors were protected from anti-α_IIbβ₃-induced systemic complications, even though thrombocytopenia still occurred³⁵. It is conceivable that the Fc-independent depletion mechanism of GPIbα antibodies is lacking in NOD/SCID mice, and that these antibodies therefore trigger the Fc receptor response. Alternatively, NOD/SCID mice may lack a cell type which in WT mice prevents anti-GPIbα antibodies from triggering systemic effects.

Irrespective of the exact mechanism, our novel adoptive transfer method provides an elegant method to exchange all endogenous platelets for human donor platelets in mice. A potential limitation arising from the larger size (and maybe different glycoprotein content) of human platelets could be enhanced retention of cells in the microcirculation of the immunocompromised mice, possibly leading to a reduced posttransfusion survival of human platelets. Our protocol specifies the injection of ~1 x 10⁹ human platelets into NOD/SCID mice, yielding a peripheral platelet count of ~2.0-3.5 x 10⁹ platelets/ml blood. This recovery is within the expected range considering a total blood volume of ~2 ml in recipient mice and the expected retention of about a third of all platelets within the platelet reservoir of the spleen⁴⁶. Another potential limitation of our model could be a limited hemostatic activity provided by the remaining endogenous NOD/SCID platelets after treatment with depleting antibody, especially since our studies were done in the presence of murine and human VWF. However, it seems unlikely that these remaining platelets contribute significantly to hemostasis in NOD/SCID mice as (1) hemostatic plug formation following laser injury was not observed in platelet-depleted NOD/SCID mice that were not transfused with human platelets (not shown), and (2) antibody treatment removed more than 99% of circulating platelets, i.e. the ratio of human to murine platelets in humanized platelet mice was > 20/1.To demonstrate the utility of our humanized platelet mouse model, we tested human platelet function ex vivo and in vivo in our recently established saphenous vein laser injury hemostasis model³³. Platelet adhesion to the damaged vein was partially dependent on human VWF and fully dependent on integrin α_IIbβ₃, consistent with the relative contributions of the VWF receptor and α_IIbβ₃ to platelet adhesion under venous shear stress conditions⁴⁷. As shown by us and others, hemostasis at small lesions, such as those generated by laser injury, is only minimally impaired by inhibitors to P2Y₁₂^33,48. Consistent with these results, aspirin and ticagrelor (DAPT) treatment significantly reduced platelet adhesion and had a mild effect on the BT after laser injury in humanized platelet mice. Similar observations were made in humanized platelet mice treated with a PAR1 inhibitor. However, hemostasis was markedly impaired in humanized platelet mice receiving both DAPT and PAR1 inhibitor, a finding that is in agreement with observations in patients receiving similar treatment^12–14. PAR1 inhibition altered hemostatic plug formation in two ways: first, platelet adhesion was altered compared to DAPT-treated mice, consistent with the unique contributions of TxA2, ADP and thrombin to platelet activation at sites of vascular injury. Second, we also observed that hemostatic plugs in triple therapy mice were richer in fibrin, suggesting that platelet aggregates under these conditions are more porous for thrombin (see excellent literature by Stalker and Brass on the diffusion rates of thrombin through the various regions of a thrombus⁴⁸). Thus, humanized platelet mice provide a cheap and faithful model for assessing the hemostatic function of human platelets, consistent with other adoptive transfer models that have been used successfully to study the patho-physiological function of human blood cells in mice^49–52.

Platelets play an important role in many pathophysiological situations, including angiogenesis, development, inflammation, and tumor metastasis/growth^53–55. Our adoptive transfer model will likely not be useful to study human platelets in these settings as it (1) depends on immunocompromised mice as recipients of human platelets, and (2) is cost-prohibitive in situations where platelet exchange must be maintained for extended periods of time. However, considering the availability of a multitude of murine models⁵⁶, we expect this model to be widely used for studies on platelet function in hemostasis and thrombosis. For example, we expect our model to be used for investigating hemostatic defects in platelets from patients with platelet function disorders (PFD). While it is relatively simple to diagnose a platelet function/adhesion defect for patients with severe PFDs, such as Glanzmann Thrombasthenia or Bernard Soulier Syndrome, aggregation tests performed in the clinic are often inconclusive for patients who present with bleeding symptoms typical for a PFD^57,58. It is also important to remember that platelet defects often coincide with alterations in plasma proteins and coagulation activity, i.e. it is sometimes difficult to determine whether and how much impaired platelet function contributes to increased bleeding. , Our model also provides a unique platform for in vivo studies on the hemostatic activity of stored, lyophilized or in vitro generated platelets^38,59–61. Transfusion of fresh and room temperature stored platelets is an effective therapy for the prevention and treatment of bleeding. Platelets derived by improved pharming technologies⁶² and engineered platelet-like particles^62,63 have advanced to the stage of preclinical testing. With our model, these agents can be rigorously tested for their hemostatic potential, alone or in combination with pro-coagulant or anti-fibrinolytic agents, such as Novo7 or tranexamic acid. At the moment, there are no clear guidelines for giving these agents, or for how to combine them with platelets/platelet mimetics.

In summary, we here report a novel method for the adoptive transfer of human platelets into thrombocytopenic mice allowing for the direct study of human platelets in an in vivo environment. Studies on the anti-hemostatic effect of PAR1 inhibition in the absence and presence of DAPT validate our humanized platelet mouse model as a robust platform to test and predict the safety and efficacy of experimental antiplatelet drugs. We propose that this model will also be of great value for the characterization of in vivo function of synthetic platelets, stored platelets or platelets from patients with a platelet function disorder.

HIGHLIGHTS:

1. We establish a novel adoptive transfer approach to replace endogenous mouse platelets with human platelets for in vivo studies in mice.

2. Hemostatic function of human platelets was validated in the absence and presence of antiplatelet drugs, using the saphenous vein laser-induced injury model.

3. In depth analysis of hemostatic plug composition in “humanized platelet mice”was performed using real-time spinning disk confocal microscopy.