(278b) Pairwise Agonist Scanning (PAS) of Human Platelets Reveals a Global 6 Dimensional Calcium Response to Combinatorial Agonists

During thrombosis, platelets respond simultaneously to collagen activation of GPVI and α2β1, ADP activation of P2Y1, P2Y12, P2X1, thromboxane activation of TP, and thrombin activation of PAR1 and PAR4, while NO and PGI2 dampen responsiveness. Using intracellular calcium mobilization to quantify platelet activation, we developed a high throughput assay to rapidly phenotype an individual's platelets. Dose responses (EC50) of convulxin (GPVI activator), ADP, U46619 (thromboxane analog), SFLLRN (PAR1 agonist), AYPGKF (PAR4 agonist) and the IP agonist PGE2 were first determined. Then, platelets were challenged with all 154 pairwise combinations (2 replicates) of these 6 agonists at concentrations of 0, 0.1, 1 and 10x EC50 to yield the level of mutual synergy or antagonism. A neural network (NN) was trained on pairwise interaction data (34,000 data points) to predict the experimentally observed pairwise synergies/antagonisms. Once trained, the NN predicted the measured 64 ternary combinations of the agonists ADP, SFLLRN and convulxin, as well as various sequential tests of these agonists where striking desensitization and cross down-regulation were noted. NN predictions of 45 conditions (ranging from extreme synergy to extreme antagonism) out of 4077 possibilities in the 6 dimensional agonist space were experimentally verified (r = 0.88). ?Super-synergy? conditions at high U46619/PGE2 ratio in the 6-agonist space were consistent with the known cardiovascular risk of COX-2 therapy. Additionally, a novel thromboxane-PAR1 synergy was detected in all donors. Finally, Pairwise Agonist Scanning (PAS) for 4 healthy individuals demonstrated unique inter-donor traits, thus allowing a patient-specific Systems Biology prediction of global platelet-dependent thrombotic response.