Megakaryopoiesis and thrombopoiesis
Although the existence of a humoral factor that stimulates megakaryocyte production was postulated by Kelemen  near the end of the Descriptive Period based on rodent studies of thrombocytopenia, attempts to purify thrombopoietin (Tpo) during the next 30 years achieved only limited progress [43,44], in part because they relied on complex in vivo, radiolabeled methionine incorporation assays . In vitro culture methods on semisolid media developed in the 1970s made it possible to define the colony- and burst-forming megakaryocyte units (CFU-MK and BFU-MK), along with their precursors, as well as variably-specific transcription factors (e.g. SCL, GATA1, GATA2, NF-E2, and ets family members) and cytokine receptors involved in megakaryopoiesis . The breakthrough, however, came when several groups recognised that the cellular proto-oncogene of the murine myeloproliferative leukaemia virus (c-Mpl) was a member of the cytokine receptor family, and that because it was expressed in megakaryocyte precursors, it was a strong candidate to be the receptor for Tpo. In 1994, studies by multiple groups using c-MPL for affinity purification or direct biochemical purification techniques resulted in the purification and cloning of Tpo [46–49]. Tpo was shown to increase platelet counts in animals and humans by enhancing both megakaryocyte proliferation and survival. Loss of either Tpo or its receptor was shown to result in severe thrombocytopenia, demonstrating Tpo’s important physiologic role. In fact, a number of patients with inherited amegakaryocytic thrombocytopenia were found to have mutations in c-MPL .
Studies of Tpo also provided insights into the regulation of the platelet count, which is normally maintained at a level that is ten-fold higher than that required to prevent spontaneous bleeding. Much can be explained by a model in which Tpo catabolism is controlled by the total mass of platelets and megakaryocytes via their binding and degradation of Tpo , with potential additional roles for regulated Tpo expression and/or cytokine (e.g. IL-6)-induced upregulation of Tpo expression .
Studies of megakaryopoiesis have also provided insights into the endomitotic process that allows megakaryocytes to increase their ploidy and mass without undergoing division [52–55]. They also shed light on how organelles and granules move along microtubule tracks to the tips of proplatelets [56,57] before they extend into the vascular sinuses in the bone marrow, bend and fragment, giving rise to thousands of platelets [58–61]. Studies of the myeloproliferative and megakaryocytic disorders associated with Down’s syndrome trisomy 21 have provided valuable insights into transcription factors that control megakaryopoiesis [62,63].
The early phase of the Mechanistic Period was marked by bringing the new tools of biochemistry to bear on the platelet phenomena that had already been described and by developing more quantitative techniques to evaluate and dissect the phenomena. The platelet’s protein and lipid composition, inorganic constituents, enzymatic activities, and energy metabolism were defined and quantified [10,64]. Hellem et al.  developed a method for measuring platelet ‘adhesiveness’ using columns of glass beads, and found an association between ‘adhesiveness’ and the blood hematocrit. Based on that finding, they postulated that red cells must have a powerful platelet activating agent and this led them to discover the platelet aggregating activity of ADP . Modifications of the platelet ‘adhesiveness’ assay (later termed the platelet retention test since it reflects both adhesion to glass and platelet aggregation) helped to define the abnormalities in platelet function in uraemia, afibrinogenaemia, Glanzmann thrombasthenia, Bernard–Soulier syndrome, and von Willebrand disease, including the accentuation of the platelet function defect in von Willebrand disease under high shear conditions [67–69]. It also furnished the first assay to support the purification of von Willebrand factor (VWF) . Biochemical analysis of platelet granule contents demonstrated the presence of adhesive glycoproteins such as fibrinogen, VWF, and thrombospondin-1 in α granules, along with cytokines and platelet-specific proteins such as platelet factor 4 and the β-thromboglobulin family of proteins [71–73]. The presence of a ‘storage pool’ of ADP, ATP, and serotonin in the calcium-rich dense bodies was also discovered and characterised .
The aggregometer and flow chamber studies
The need for more quantitative and robust methods of assessing platelet function led both Born and O’Brien to develop turbidometric platelet aggregometry in 1962 [75,76]. This allowed for the characterisation of the unique platelet effects of the different agonists that had been, and were being, discovered, including ADP, epinephrine, thrombin, collagen, vasopressin, and serotonin. The discovery that with citrated platelet-rich plasma ADP and epinephrine produce aggregometer tracings that could be divided into two separate phases, or waves, led to the definition of the role of the platelet release reaction, including the release of ADP from dense bodies and the release of arachidonic acid and its subsequent conversion into thromboxane A2, in amplifying the platelet aggregation response [77–80]. The subsequent development of the lumi-aggregometer allowed for simultaneous measurement of both platelet aggregate formation and release of ATP from dense bodies . Aggregometer studies were also crucial in more precisely defining the roles of phospholipids, phosphoinositol hydrolysis, diacylglycerol, and Ca2+ mobilisation in platelet activation [82–86], as well as the functional defects in many of the previously described inherited platelet disorders [87–90], and the newly described Storage Pool disorders , the Gray platelet syndrome , and disorders of platelet secretion [93,94]. The roles of fibrinogen and VWF in mediating platelet–platelet interactions via binding to αIIbβ3 [95–98] and GPIb , respectively were also facilitated by aggregometer studies. Similarly, studies using newly designed flow chambers defined platelet-ligand and platelet–platelet interactions as a function of shear rate, introducing and partially simulating an important aspect of the in vivo milieu [100–103].
Platelet coagulant activity
Platelets were shown to greatly accelerate blood coagulation and this property was initially termed platelet factor 3 (PF-3) [10,104]. This factor was only ‘made available’ when platelets were activated under certain conditions, or when they were subjected to physical trauma, as with freezing and thawing. Extensive studies of the platelet lipids and proteins responsible for the nearly 100 000-fold increase in the rate of thrombin generation identified exteriorisation of the negatively charged phospholipid phosphatidylserine as a primary contributor to recruitment of coagulation factors and enhancement of catalytic efficiency [105,106]. A series of platelet protein receptors for coagulation factors were also identified and their contributions continue to be the subject of study [107–109]. The role of fibrin in enhancing platelet-mediated thrombin generation was also identified . Platelet microparticles, small fragments of platelets that circulate in health and in disease, were found to have coagulant activity and to be biologically active [111,112]. Studies of a single patient, Mrs Scott, first described by Weiss et al. in 1979  provided important insights. Her bleeding symptoms differed from patients with qualitative or quantitative defects in platelets in that she did not have easy bruising; she did, however, have menorrhagia and post-partum haemorrhage. Her platelets failed to expose negatively-charged phospholipids upon activation, apparently as part of generalised defect in cell vesiculation [114–116]. Information from advanced biophysical tools and structural biology insights have been combined to provide a detailed image of the underlying mechanisms by which platelets contribute to thrombin generation [111,112,117–120].
The endothelial lining, inhibitors of platelet function, and platelet–leukocyte interactions
The endothelial lining of blood vessels prevents platelet interactions with the vessel wall by both synthesising platelet inhibitors and ‘hiding’ the subendothelial platelet-adhesive proteins from the platelet’s view. Studies by the groups led by Moncada and Weksler demonstrated that endothelial cells synthesise prostaglandin I2 (PGI2, prostacyclin) from arachidonic acid [121,122]. Subsequently, endothelial cells were found to synthesise the platelet inhibitor nitric oxide (NO), which acts synergistically with prostacyclin [123–125]. Finally, Marcus et al. [126,127] also identified an endothelial ecto-ADPase [CD39; ecto-nucleotide pyrophosphatase/phosphodiesterase (E-NPP)] capable of rapidly metabolising the platelet activator ADP (as well as ATP) to the inactive compound AMP. The subsequent conversion of AMP to adenosine by the ecto-5′-nucleotidase CD73 on the endothelial cells of some vessels and on some leukocytes yields adenosine, an inhibitor of platelet activation [128–130]. Marcus [131,132] also identified the extraordinary high content of arachidonic acid in platelets and the complex transcellular eicosanoid metabolism that occurs between platelets, endothelial cells, and leukocytes.
Platelets can also interact with, and roll on, intact endothelium via multiple receptors when the endothelium is activated to secrete large VWF multimers and expose P-selectin . Complex mechanisms involving similar interactions underly platelet–leukocyte interactions [134–136].
When the endothelial cell lining is disrupted, platelets adhere to the denuded surface via receptors for the adhesive proteins that are exposed and/or deposited from plasma, in particular, the interactions of platelet GPIbα with VWF [137–139] and both GPVI and α2β1 with collagen [140,141]. GPIbα has a carbohydrate-rich region that allows it to extend far from the platelet surface and so it attaches first. Engagement of these receptors initiates platelet activation, resulting in activation of αIIbβ3 and recruitment of additional platelets.
Platelet receptors and ligands
The concept of cell receptors controlling cell physiology also emerged in the early part of the Mechanistic Period. Initially receptors were characterised functionally and pharmacologically based on biochemical responses and later they were characterised by their ability to release intracellular Ca++, monitored by cation-sensitive dyes [142,143]. The introduction of radionucleotides in biomedical research made it possible to perform direct ligand binding studies in select cases [95,96]. The introduction of sodium-dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) allowed for detailed biochemical assessment and was a great aid to receptor purification and characterisation [99,144]. It also provided the initial glycoprotein (GP) nomenclature for many of the receptors . The sensitivity and specificity of SDS-PAGE was greatly enhanced by combining it with platelet surface labelling techniques and immunoblotting [145,146]. Other immunologic techniques, including cross-immunoelectrophoresis allowed for assessing protein–protein interactions under non-denaturing conditions, helping to define ligand–receptor interactions [99,144,145,147]. Monoclonal antibody development was an additional major advance, providing essentially unlimited amounts of standardised reagents that could unequivocally identify specific receptors by immunoblotting and other techniques [147,148]. In addition, the development of monoclonal antibodies that could inhibit platelet function provided tools that could dissect the contribution of each receptor [147,149–152]. This was particularly important in understanding the roles of receptors that could bind more than one ligand, such as αIIbβ3 and αVβ3, and of ligands like VWF and fibrinogen that could bind to more than one receptor. Monoclonal antibodies that reacted selectively with activated αIIbβ3, most notably PAC1 , or with P-selectin [154,155], which is confined to α-granule membranes, made it possible to monitor the activation state of αIIbβ3 and the release reaction, respectively. Similarly, monoclonal antibodies were developed that reported on the conformational changes in receptors after ligand binding . Many of the initial monoclonal antibody studies were performed with radiolabeled antibodies, which required special care and handling. The development of flow cytometry made it possible to substitute more generally available fluorescently-labelled antibodies and permitted analysis of the binding of antibodies to individual platelets [157–159] and even to small platelet microparticles, which are biologically active and support blood coagulation [160,161].
The introduction of molecular biologic techniques allowed for the detailed analysis of the subunit structures and homology of different receptors, defining the major families of receptors as integrins (αIIbβ3, αVβ3, α2β1, α5β1, and α6β1), leucine-rich repeat receptors (GPIbα, GPIbβ, GPV, GPIX), immunoglobulin repeat receptors (PECAM-1, GPVI, FcγRIIA, CLEC-2), selectins (P-selectin), tetraspanins (CD9, CD151, TSSC6, CD63, and Tspan9) [162,163], and G protein-coupled 7 transmembrane receptors (including, among others, PAR-1, PAR-4, P2Y1, P2Y12, vasopressin V1a, serotonin 5-HT2A, and adenosine receptors A2a and A2b) [128,164,165]. Of particular importance in integrin receptor biology was the extrapolation from studies of α5β1  that both αIIbβ3 and αVβ3 could be inhibited by peptides containing the Arg-Gly-Asp (RGD) sequence . Structural biology studies using nuclear magnetic resonance, x-ray crystallography, electron microscopy, and molecular dynamics simulations now permit atomic level resolution of the receptors and receptor–ligand interactions [168–173]. Transgenic mice with targeted deletions of receptors or mutant receptors have confirmed and extended our understanding of the roles of the different receptors [174,175], although extrapolation from mouse to human platelets needs to be tempered because of the unique aspects of human vs. mouse platelets, such as the repertoire of functional thrombin receptors , and the differences in physiology, most notably the mouse’s approximately 600 beats per minute heart rate!
Insights into the signal transduction pathways triggered by ligand binding also derived from studies of the receptors. Our understanding of collagen signalling through GPVI had its origin in the 1989 description of a patient with an isolated defect in collagen-induced platelet aggregation whose platelets also had a defect in GPVI [177,178]. Subsequent functional studies and the cloning of the GPVI led to the recognition that it is coexpressed with the FcR γ-chain and requires a dimeric structure to achieve high affinity for collagen . GPVI cross-linked by polymerised collagen (or collagen-related peptides or the snake venom convulxin) was elegantly shown to initiate Srk kinase-dependent tyrosine phosphorylation of the FcR γ-chain ITAM, leading to recruitment and activation of Syk. Downstream from Syk is a cascade of intermediates similar to those in immune cell signalling, including LAT, SLP-76, PI3-kinase and PLCγ2, ultimately leading to the conversion of PIP2 into 1,2-diacylglycerol and inositol 1,4,5-trisphosphate IP3, resulting in an increase in cytoplasmic calcium and granule exocytosis . Signalling through the GPIb/X/V complex leading to inside-out signalling could be inferred from the contribution of αIIbβ3 to ristocetin-induced platelet aggregation and from studies identifying the roles of VWF and fibrinogen in shear-induced platelet aggregation . The signal transduction details, including the roles of tyrosine kinases, were then worked out in several laboratories [182–185]. Signalling through αIIbβ3 (‘outside-in’) was shown to include roles for Src, Syk, Gα13 and the phosphorylation of two tyrosine residues in the β3 integrin subunit cytoplasmic tail [186–190]. ‘Inside-out’ signalling to activate αIIbβ3 via changes in its conformation so that it adopts a high affinity ligand binding state was elucidated by elegant functional studies of the roles of talin and kindlin in binding to the β3 cytoplasmic tail and the separation of the transmembrane domains [191–199]. The physiologic role of kindlin-3 was supported by the identification in 2009 of patients with a profound defect in platelet function as part of leukocyte adhesion deficiency-III (LAD-III) resulting in a serious bleeding diathesis in association with mutations in kindlin-3 [197,198,200–202]. CLEC-2, a C-type lectin receptor, has multiple ligands including the snake venom rhodocytin, tumor-associated podoplanin and HIV-1, and has been implicated in haemostasis, thrombosis, tumor metastases, and lymphangiogenesis; it also signals through tyrosine phosphorylation of its cytoplasmic tail followed by the binding of Syk [203–208]. Signalling through the G protein-coupled 7 transmembrane domain receptors was investigated in great detail including the pairings with specific Gα subunits, the downstream effects on second messengers, and the feedback control mechanisms [164,165]. Balancing the platelet receptors involved in initiating signalling are inhibitory receptors containing the immune receptor tyrosine-based inhibitory motif (ITIM), most notably PECAM-1 and G6b-B [209–214]. A third platelet ITIM containing receptor, TREM-like Transcript-1, however, may not serve an inhibitory function; in fact, it may enhance platelet function and contribution to platelet–leukocyte interactions .
Clinical diagnosis, carrier detection, and prenatal diagnosis
Biochemical and molecular biological advances, most notably the application of the reverse polymerase chain reaction (PCR) to study the small amount of platelet mRNA , revolutionised the diagnosis and characterisation of patients with disorders affecting platelet receptors. In particular, Glanzmann thrombasthenia was defined as a defect in either αIIb or β3 [217,218], Bernard–Soulier syndrome was defined as due to an abnormality in GPIbα, GPIbβ, or GPIX , platelet-type von Willebrand disease was defined as a defect in GPIb , and other defects were identified in P2Y12, GPVI, and many other receptors . Monoclonal antibodies were rapidly applied to the diagnosis of the disorders and in select cases, carrier detection; they also made possible prenatal diagnosis based on fetal blood sampling at approximately 18–20 weeks of gestation [222,223]. Since megakaryocytes also synthesise VWF, platelet mRNA-based PCR was also used to define the molecular biologic defects in VWF and correlate them with the current subtyping of the disorder [224,225]. Practical benefits to patients accrued rapidly from the advances in molecular biology, including the ability to perform unequivocal carrier detection and prenatal diagnosis using chorionic villus sampling at just approximately 11 weeks of gestation .
The molecular biologic tools also revolutionised our understanding of the alloimmune thrombocytopenias, including neonatal alloimmune thrombocytopenia and post-transfusion purpura, which had been characterised serologically in the 1960s [227–229]. Thus, the amino acid substitutions in the glycoproteins responsible for antibody production were defined, tools were developed to test for potential incompatibilities, and effective therapies introduced [230–235].
von Willebrand disease
Building on the observation that the clinical platelet function abnormality in von Willebrand disease could be dramatically improved by a plasma factor [32,33,236], Ted Zimmerman in Oscar Ratnoff’s  laboratory established that most patients with von Willebrand disease have reduced levels of a protein present in normal plasma and the plasma of patients with haemophilia A. A series of biochemical studies ultimately showed that VWF and factor VIII are separate molecules, but as was suspected from the transfusion studies, they circulate as a complex in plasma [22,238]. The unfortunate choice of the term ‘factor VIII-related antigen’ for VWF created considerable confusion, some of which remains for those new to the field to this day. Studies of VWF immunology, biochemistry, and molecular biology defined its subunit structure, its remarkable multimeric structure, its association with factor VIII, its ability to bind to collagen, GPIbα, and αIIbβ3, and its cleavage by ADAMTS13 [139,239–242]. Studies of VWF carbohydrate defined functional variants  and the role of blood group carbohydrates in its in vivo survival, leading to close correlations between blood type and VWF plasma levels that also translate into haemorrhagic and thrombotic risk [244–247]. The important roles of endothelial cells and megakaryocytes in VWF synthesis were also defined, and the importance of platelet vs. plasma VWF debated [248,249]. The remarkable discovery in 1971 by Howard and Firkin, who systematically studied drugs that produced thrombocytopenia in humans, that the antibiotic ristocetin could agglutinate/aggregate platelets only in the presence of VWF provided more robust functional assays than the platelet retention test for diagnosing von Willebrand disease, purifying VWF, and demonstrating that the higher molecular weight multimers are functionally more effective in agglutinating/aggregating platelets [22,250–252]. Subsequent studies showed that the snake venom protein botrocetin could also agglutinate platelets in the presence of VWF, providing another tool for assaying VWF . The important role of ADAMTS13 in controlling the size of VWF multimers through proteolysis was later defined, spurred by research on thrombotic thrombocytopenic purpura . Collectively, these advances allowed for the classification of von Willebrand disease into a large number of different subtypes, including the rare platelet-type associated with abnormalities in GPIbα, with both diagnostic and therapeutic importance .
Studies of fibrinogen defined the importance of the γ chain C-terminal peptides and other regions in interacting with αIIbβ3 and the role of other regions in binding to αVβ3 and other receptors such as αMβ2 and αXβ2 [254–259]. The differences between platelet interactions with fibrinogen vs. fibrin were also characterised [260–265].
Thrombotic thrombocytopenic purpura (TTP)
For more than a half-century after Moschcowitz’s description of TTP, haematologists were powerless to prevent the usually rapid progression to death of the uncontrolled thrombotic microangiopathic process manifested by haemolytic anaemia, thrombocytopenia, and ischaemic organ damage, especially to the brain and kidney . Thus, the 1976–1977 reports by several groups of striking benefit from exchange transfusion, plasma or cryoprecipitate-poor plasma infusion, and/or plasmapheresis were met with enormous interest . Plasma exchange rapidly became the treatment of choice and dramatically improved the prognosis . Despite employing sophisticated machinery, however, plasma exchange is intellectually very unsophisticated since it is based on the premise of removing whole plasma to presumably remove a single toxic substance and replace whole plasma to presumably replace another single substance. Moake and colleagues provided a vital clue in 1982 when they observed unusually large VWF multimers in the plasma of patients with TTP and postulated that these contributed to the development of platelet thrombi and resulted from a defect in multimer processing [19,266]. They were prescient on both accounts as subsequent groups in 1987–1988 identified the VWF processing activity as a metalloproteinase, later identified as ADAMTS13, and confirmed a defect in VWF multimer cleaving activity in patients with TTP, identifying the cause as autoantibody formation [267–269]. This led to the introduction of rituximab treatment to diminish autoantibody production, another therapeutic advance . Shortly thereafter, investigators identified long strings of VWF extending out from the surface of endothelial cells to which platelets could attach in the absence of ADAMTS13, providing an explanation for the source of the unusually large VWF multimers [19,133]. Cytokines or hormonal changes presumably trigger TTP in the absence of ADAMTS13 activity by enhancing endothelial cell production of VWF. Subsequently, patients with inherited forms of the disease (Upshaw–Schulman syndrome) were shown to have mutations in the ADAMTS13 gene [271–275]. The success of plasma exchange could then be explained by its ability to both remove the autoantibodies to ADAMTS13 and replace the missing ADAMTS13. Most recently, preliminary data with an oligonucleotide aptamer that inhibits the binding of VWF to GPIb (ARC1779) appear promising in treating the acute thrombotic phase of the disorder . Looking to the future it is possible that recombinant ADAMTS13 therapy could reverse the acute thrombotic process in TTP and perhaps serve as an antiplatelet therapy more broadly. Since complement activation has also been implicated in the pathophysiology of the disorder in some patients, it may provide an additional target for intervention .
Drug-induced thrombocytopenia was recognised to be associated with a broad array of agents, some of which produced direct or indirect effects on megakaryocytes, others of which operated through immune clearance, and others, such as quinine and quinidine, that could actually cause platelet lysis . Heparin-induced thrombocytopenia (HIT), and its association with the thrombotic ‘white clot’ syndrome, was identified as being of major clinical importance, with a broad spectrum of manifestations including both arterial and venous thrombosis . The important role of PF4 in HIT was elucidated and a variety of diagnostic assays were developed, but none of these are ideal for real-time clinical decision making .
Thrombopoietin and related drugs
Soon after Tpo was cloned, two companies developed recombinant forms of Tpo for clinical use and demonstrated their ability to increase platelet counts in healthy humans. Platelet increments were also induced in patients receiving some forms of chemotherapy, but not the most myelosuppressive forms [296–299]. Clinical development of these agents was stopped, however, when healthy individuals who received one of the preparations developed thrombocytopenia in association with developing antibodies to Tpo . After a multiyear lag, two agents that activate the thrombopoietin receptor demonstrated safety and efficacy in treating patients with immune thrombocytopenia. One agent (romiplostim), a success of phage library screening and ‘peptibody’ engineering, consists of a human immunoglobulin IgG1 Fc fragment framework onto which was grafted two dimeric copies of a Tpo receptor stimulating peptide distinct from the Tpo amino acid sequence [301–303]. The other (eltrombopag), a success of organic molecule library screening, is an orally active small molecule that binds to the receptor at a site near the transmembrane region [302,304]. Both drugs have proved efficacious in treating refractory immune thrombocytopenia, albeit perhaps with a minor risk of thrombosis in those at high risk [301,305]. Benefit has also been reported in treating patients with hepatitis C and thrombocytopenia , but thrombotic complications of the therapy have led to its reappraisal for this indication [307,308]. Future uses of these agents may include some of the inherited platelet abnormalities associated with abnormalities of nonmuscle myosin heavy chain IIA (MYH9; May–Hegglin anomaly, Epstein syndrome, Fechtner syndrome, and Sebastian platelet syndrome) and neonatal thrombocytopenia [309,310].