The gold standard measurement of platelet function is the assessment of aggregation in isolated suspensions, platelet-rich plasma or whole blood. Such approaches are limited because in vitro function poorly predicts the platelet response in vivo. This is partly due to the central role of the vascular endothelium in regulating platelet responses [1,2]. A number of valuable animal models of thrombosis are available , although these models do not directly assess platelet responsiveness but rather the multifactorial process of thrombus formation following synthetic vascular injury through mechanisms that remain unclear .
The development of mouse models for assessing platelet aggregation in vivo allowed the evaluation of platelet function directly and in a physiological and genetically malleable setting . There are, however, key differences between human and mouse platelets, meaning that data from mouse models will not translate directly to humans, resulting in a translational gap. Studies of human platelet function are therefore largely limited to in vitro approaches with their inherent limitations. The ability to assess human platelet function in vivo would bridge the gap between in vitro human and in vivo mouse studies and improve the translation of experimental data to the clinic. We have addressed this issue by developing a method for assessing human platelet aggregation in vivo in NOD/SCID (non-obese diabetic modified severe combined immunodeficiency) mice.
To create an environment in which human platelets circulated and aggregated with minimal interaction with mouse platelets, we depleted the endogenous platelet population of NOD/SCID mice by the application of the alkylating agent busulfan, using a dosing regimen previously shown to preferentially target platelets . Busulfan depleted circulating platelet levels by approximately 95% (Fig 1A). Anesthetized, platelet-depleted NOD/SCID mice were subsequently infused with 111Indium radiolabelled human platelets via an exposed femoral vein and the responses to intravenous injection of platelet agonists recorded as changes in platelet-associated counts via external 1-cm scintillation counters positioned over the thoracic region. Additional methodological details are provided as online supplementary information (Data S1). ADP (4–400 μg kg−1) induced dose-dependent platelet aggregation, indicated by a rapid increase in platelet-associated counts, which reached a maximal level before returning to basal levels (Fig. 1B). Response profiles were also obtained with thrombin (16–64 IU kg−1) and collagen (100–200 μg kg−1) (Fig. S1). Dose dependence was demonstrated by measuring the peak responses (data not shown but apparent from Figs 1B and S1) and area under the curve (AUC) for all agonist responses (Fig 1C and S1). Because responses to ADP and thrombin were completely reversible we were able to record dose responses to these agonists within an individual experimental animal (Figs 1D and S2).
We confirmed that increases in radioactive counts corresponded to the entrapment of platelet aggregates in the pulmonary vasculature by observing platelet aggregates in the partially occluded lung vasculature of mice killed at a time-point corresponding to the peak collagen response (1E) and an absence of aggregates in control mice injected with saline (Fig S3). As an additional control, we showed a lack of platelet accumulation in areas outside of the pulmonary region such as the liver (Fig. S4).
When human volunteers took aspirin for a period of 8 days prior to collection of platelets for infusion into NOD/SCID mice, the in vivo response to collagen (100 μg kg−1) was suppressed (Fig. 1F), such that the peak aggregation response and AUC (Fig . S5) were significantly (P < 0.05) reduced compared with experiments conducted when participants had been aspirin free for a minimum period of 21 days.
The genetic malleability of the mouse makes it one of the most valuable experimental tools in cardiovascular research. Although the mouse cardiovascular system replicates many aspects of the human, there are key differences between human and mouse platelets. In 2007 data were published showing circulation of infused human platelets in NOD/SCID mice for up to 2 days , indicating the potential of this mouse strain for the evaluation of human platelet aggregation over a short time-frame in vivo as presented in this study. We confirmed that the recorded increases in radioactive counts in the pulmonary region reflect platelet aggregation through histological analysis. The reduction in platelet aggregation following administration of aspirin to human platelet donors complements published data demonstrating the role of COX-1 in mediating the anti-thrombotic effects of aspirin in mouse models  and evidences the feasibility of the evaluation of drug efficacy for human platelet function using our model. Additionally, assessment of platelets from particular patient groups is possible, as is the evaluation of agents in mouse models of cardiovascular disease.
An increasingly important issue in biomedical research is reduction, refinement and replacement of animal use, as recently reviewed . Our model constitutes a refinement of conventional models of thromboembolic mortality that inflict considerable pain and suffering in conscious experimental animals and are limited scientifically by their evaluation of one extreme end-point of diseases with a broad spectrum . In contrast, our experimental procedures are conducted entirely under general anesthesia and are therefore refined to reduce pain and suffering and model the treatable component of the disease. In addition, the ability to record multiple responses within an individual animal as well as the use of human rather than mouse platelet donors dramatically reduces mouse use.
In summary, we present a model for the evaluation of human platelet aggregation responses in an in vivo experimental setting that will help to bridge the gap between in vitro assessment of human platelet function and animal studies and thus improve the translation of experimental data from bench to bedside.