Systems biology meets platelet biology

In this issue of Blood, Purvis and colleagues describe a computational approach, employing 4 interlinked kinetic modules, to model platelet phosphoinositide and calcium regulation in resting platelets and after ADP-mediated P2Y₁ purinergic receptor activation.

The model accurately replicates experimental findings, including the broad frequency distribution and asynchronous calcium spiking behavior in single platelets in response to adenosine diphosphate (ADP).

The P2Y₁ receptor is a G-protein coupled receptor that signals through G_q and mediates ADP-induced platelet shape change and aggregation. Phospholipase Cβ₂ (PLC-β₂) is the major signaling molecule downstream of G_q and is responsible for transient increases in inositol 1,4,5-trisphosphate (IP₃) and calcium concentration as well as the production of diacylglycerol (DAG) and protein kinase C (PKC) activation.^1,2  In resting platelets, the cytoplasmic calcium concentration is actively maintained at approximately 100 nM, primarily through bulk storage of calcium within the dense tubular system (DTS). Calcium homeostasis between the cytoplasm and the DTS is regulated by inositol trisphosphate receptor (IP₃R) channels, which release calcium ions from the DTS in response to IP₃ and by a sarcoplasmic/endoplasmic reticulum Ca²⁺ ATPase (SERCA), which pumps calcium ions from the cytoplasmic compartment into the DTS.³ 

Classically, our understanding of platelet biology has come from a reductionist approach where individual events are studied under a single set of conditions. Here, Purvis et al present the first detailed and integrated “systems biology” approach to platelet signaling, modeling both phosphoinositide and calcium homeostasis in the resting platelet as well as temporal changes post-P2Y₁ receptor activation. The model employs 4 interlinked kinetic modules: a calcium module, where the cytosolic and DTS compartments are separated by the DTS membrane, which contains the IP₃R channels and SERCA; a phosphatidylinositol (PI) module, in which plasma membrane-bound PIs are cleaved by PLC-β₂ to form diffusible inositol phosphates and DAG, which are substrates for resynthesis of PIs; a PKC module, whereby calcium and DAG activate PKC, which migrates to the plasma membrane and phosphorylates PLC-β₂; and a P2Y₁ module, where extracellular ADP activates P2Y₁, accelerating formation of GTP-bound G_q, and subsequent activation of PLC-β₂. Each module was initially considered in isolation for optimization. For example, platelet calcium homeostasis was analyzed by fixing the kinetic properties of the IP₃R channels and SERCA, the resting calcium concentration, the volume of the platelet from experimental measurement, and assessing combinations of values for the number of IP₃R channels per platelet, SERCA pumps per platelet and volume of the dense tubular system, using analog computation. P2Y₁ signaling was con-sidered within the constraint that calcium influx was experimentally excluded, obviating the need to model store-operated calcium entry.⁴ 

This integrated model developed by Purvis and colleagues accurately accounts for known platelet behavior and replicates experimental data at both averaged platelet and single platelet response. Stochastic simulation of the platelet model generated calcium spiking with peak-to-peak interval times favoring 6 to 8 and 11 to13 s gaps, strikingly similar to calcium responses in video-imaged single platelets. In addition, the model allowed several novel predictions: The calcium spiking in single platelets was a consequence of the small platelet volume, the number of SERCA pumps must signifi-cantly outnumber IP₃R channels, and recovery of basal PI levels requires a negative-feedback mechanism in which PKC phosphorylation of PLC-β inhibits its hydrolytic activity. The model further explains that the reason thrombin is a more potent agonist than ADP is primarily due to differences in receptor copy number. The power of the computational approach of Purvis et al is that with continued development and experimental refinement of current variables, the model will not only accurately reflect known platelet behavior, but also predict new experimental findings, allowing a true in silico molecular and kinetic understanding of platelet biology.