Joseph E. Italiano, Translational Medicine Division, Karp 6, Rm 215, 1 Blackfan Circle, Brigham and Women’s Hospital, Harvard Medical School, Boston, MA, USA. Tel.: +1 617-355-9005; fax: +1 617-355-9016; e-mail: firstname.lastname@example.org
Summary. The cellular and molecular basis of the intricate process by which megakaryocytes (MKs) form and release platelets remains poorly understood. Work has shown that proplatelets, long cytoplasmic extensions made by mature MKs, are essential intermediates in platelet biogenesis. Microtubules are the main structural component of proplatelets and it is microtubule sliding, driven by dynein motors within cortical bundles, which elongates and thins proplatelets. Kinesin motors carry their cargo of platelet-specific granules and organelles into the proplatelets using the microtubule bundles as tracks. Extension of proplatelets is associated with repeated actin-dependent bending and bifurcation, which results in considerable amplification of free proplatelet ends. Large proplatelets, dissociated from the residual MK cell body, have the capacity to mature platelets. Only the ends of proplatelets form marginal microtubule coils similar to that observed in mature platelets, demonstrating that platelet formation completes primarily at proplatelet ends. Understanding the molecular basis of platelet formation requires detailed knowledge of how the MK microtubule machinery interacts to generate proplatelets and release platelets.
Megakaryocytes (MK) are specialized cells living in bone marrow that enlarge and become polyploid  through repeated cycles of DNA replication without cell division (endomitosis) [2,3]. This DNA amplification is ultimately used to direct the protein and lipid synthesis that increases the size of the cell and fills it with platelet-specific granules , cytoskeletal proteins, and an extensive internal membranous labyrinth called the demarcation membrane system  that supplies membrane during proplatelet formation . MK maturation culminates with the assembly of long cytoplasmic protrusions called proplatelets that release platelets into circulation. Proplatelets are tubular elongations extended by MKs that have periodic platelet-sized swellings along their length. Although detailed characterization of proplatelets remains incomplete, they have been recognized both in vitro and in vivo [6–8] and as only proplatelet-producing MKs yield platelets in culture, they are an essential intermediate in platelet production [9,10]. Mice lacking distinct hematopoietic transcription factors have severe thrombocytopenia and fail to produce proplatelets in culture, underscoring the correlation of proplatelets with platelet biogenesis in vivo [11–14]. Mature MKs move near to the bone marrow venous sinusoids  and proplatelet-like structures have been observed extending from MKs in the bone marrow through junctions in the lining of blood sinuses [9,16–19].
The stages of proplatelet formation
The discovery of thrombopoietin and development of MK cultures that reconstitute platelet formation in vitro have provided systems to study MKs in the act of forming platelets. Mouse MKs in culture follow a maturation program that ends in proplatelet and platelet formation (Fig. 1). The purpose of this process is to concentrate specialized platelet materials in the proplatelets. Proplatelets are uniform, long tubular structures that have the appearance of beads linked by thin cytoplasmic bridges . The process of forming and elongating proplatelets unfolds over 4–10 h, initiating with erosion of one pole of the MK to generate large pseudopodial-like structures that elongate, thin, and branch to yield slender tubular tree-like structures with shafts with uniform diameters of 2–4 μm. As proplatelets elongate outward at a steady rate of ∼1 μm min−1, proplatelet lengths of 0.5–1 mm are commonly obtained in the 4–8 h required to complete this task in culture. Proplatelet formation is a dynamic process that is complicated by a repetitive dynamic bending and branching that bifurcates the shaft multiple times and thereby increases the number of free proplatelet ends. In the proplatelet tips, a single microtubule, derived from the microtubule bundles of the proplatelet shaft, rolls up into a circumferential coil that defines the territory of an individual platelet. Once the coil has been established, the nascent platelet fills with its content of granules and organelles. The filling of proplatelets with platelet-specific granules begins early during the proplatelet elaboration process. Once delivered into the proplatelets, granules track back and forth slowly over microtubules in the shafts . Granules that reach the ends of proplatelets enter nascent platelet buds and become trapped.
A considerable portion of the proplatelet mass releases from the MK cell body before a significant conversion into platelets occurs. However, even dissociated from MKs, proplatelets retain their capacity to mature into platelets (Fig. 2) and platelet formation occurs from both ends of the dissociated proplatelet. Microtubule movements, in the fractured shaft of the proplatelet, rapidly convert it into a bulbous structure by forming a microtubule loop in it identical to that found at the proplatelet tips. The presence of proplatelet-like structures in blood has been long recognized, and thus, it is likely that proplatelets routinely fragment from the MK body, enter the blood, and mature into platelets there.
Microtubules provide the force to elongate proplatelets
The first insights into the cytoskeletal mechanics of platelet formation date from the work of Tablin and Leven , who found that microtubule poisons, such as colchicine, prevent proplatelet formation. The observation that mice lacking the hematopoietic-specific β1-tubulin show profound thrombocytopenia and fail to produce proplatelets in vitro, adds molecular support for the essential role for microtubules in platelet assembly [21,22]. Antitubulin immunofluorescence and high-resolution electron microscopic studies have now delineated the changes in microtubule organization that lead to the elaboration of proplatelets (Fig. 1). Proplatelets are filled with 50-200 microtubules that derive from microtubules formed and sent to the cortex of mature MKs. Once in the cortex, these microtubules organize into interdigitating arrays that orient parallel to the plasma membrane. Sliding movements between the microtubules in these arrays first generate initial broad pseudopodia. Next, the pseudopodia narrow and elongate to become proplatelets. As this occurs, the cortical microtubule bundles twist and collapse upon themselves, forming loops at the tips. As the proplatelet elongates, the microtubule bundles taper as they approach the bulbous proplatelet ends, but do not terminate. Instead, they make a U-turn and re-enter the proplatelet shaft. This teardrop-shaped structure forms the template for the microtubule coil that forms in the nascent platelet.
While microtubules are used to propel proplatelet elongation, an actin-dependent reaction is used to bifurcate the proplatelet shaft, thereby increasing the number of proplatelet tips available to participate in platelet formation [8,10]. Branching begins when a portion of the proplatelet shaft bends to form a U-shape, from which a new daughter process protrudes and elongates. The branching mechanism is mediated by actin-based forces and becomes quiescent after treatment of MKs with drugs that disrupt the actin cytoskeleton, such as cytochalasin B and D.
Cytoplasmic dynein powers microtubule sliding in proplatelets
Cytoplasmic dynein and kinesin are motor proteins that move on microtubules and attach to cargo that is moved over microtubules. Cytoplasmic dynein is primarily responsible for sliding proplatelet microtubules relative to one another in proplatelets , while kinesin associates with granules and organelles within the proplatelets . Antibodies to kinesin heavy chain label membrane-bound organelles in both the cell body and proplatelets (Fig. 3) but do not label proplatelet microtubules, while cytoplasmic dynein or dynactin antibodies strongly stain the microtubule composing the bundles in proplatelets. Only cytoplasmic dynein remains associated with detergent-treated proplatelets that can be reactivated by adenosine triphosphate (ATP) to elongate via microtubule sliding.
The localization and association of cytoplasmic dynein with proplatelet microtubules suggests that it is the source of the forces that slide neighboring microtubules. To directly test if cytoplasmic dynein contributes to proplatelet elongation, we have expressed GFP-dynamitin in MKs. Overexpression of dynamitin (p50) dissociates the p150 subunit from the dynactin complex and disrupts dynein function in cells [24–28], including during axonal outgrowth in cultured neurons . MKs expressing dynamitin p50-GFP exhibit a striking phenotype not found in control transfected or untransfected cells that is characterized by short pseudopodia extending from the cell body.
The association of kinesin with granules/organelles in proplatelets suggests that this family of motor proteins provides the motile forces that move cargo over microtubules into the proplatelets. Organelles move slowly (∼0.1 μm min−1) along the microtubule bundles of proplatelets, and frequently stop and then change direction (Fig. 3). As most kinesin proteins move toward the plus-end of microtubules, reversal of direction by organelles is best explained by the mixed microtubule orientation in the bundles.
Toward a mechanistic model for proplatelet elongation
We have focused on understanding how microtubule mechanics are used to make proplatelets and platelets. As discussed, microtubules collect into bundles in the shaft of proplatelets that begin in the MK cell body and extend to the tip of the proplatelet, where they make U-turns and run back toward the cell body. The proplatelet shaft, therefore, is composed of at least two microtubule bundles with opposite polarities . Cross-sections of the proplatelet shaft reveal that the shafts of proplatelets have, in fact, four or more distinct bundles, each composed of 10 or more microtubules (Fig. 1B).
Individual microtubules composing bundles are dynamic and are continuously in the process of assembly–disassembly. We have marked, and visualized in real time, the assembling ends of microtubules using end-binding protein-3 tagged with green fluorescent protein (EB3-GFP), a protein that promotes microtubule growth in cells . Because the localization and movement of EB3-GFP defines growing plus-ends of microtubules, analysis of their number, direction and velocities of movement provides essential information on the microtubule motor of proplatelets. Firstly, it informs us that bundles have mixed microtubule polarity (i.e. EB3-GFP moves in both directions in a bundle). Secondly, it reveals that there is on average 0.5 microtubule ends per μm of proplatelet length. Hence, by comparing the number of EB3-growth sites to estimates of the number of microtubules in a proplatelet determined in the electron microscope1, we can conclude that most microtubules assemble dynamically (e.g. proplatelets are not cored by stable microtubules). Analysis of the rates of EB3 movements in the microtubule bundles (microtubule assembly rates) also provides direct evidence that microtubules slide relative to each other in the shafts of proplatelets. Microtubule growth rates are biphasic in proplatelets: one population grows at rates of ∼5 μm min–1 while the other elongates much faster at rates of ∼14 μm min–1. In contrast, the plus-end growth rates of microtubules undergoing centrosomal nucleation in immature MKs, which lack proplatelet projections, are monophasic (∼10 ± 1.5 μm min−1). As this differential rate behavior is specific to MKs with proplatelets, the simplest explanation for it is that it derives from a superimposition of microtubule sliding (at rates of 4–5 μm min−1) onto the 10 μm min−1 polymerization rate. Given that bundles have mixed polarity, individual microtubules sliding opposite to their assembly end would appear to grow more slowly, while microtubules sliding in their growth direction would appear to grow faster. As microtubules assemble tenfold faster than the proplatelets elongate and four to five times faster than the microtubule sliding rate, neither process can be directly coupled to proplatelet elongation.
Although formal measurements of microtubule disassembly rates remain to be reported, it is likely that assembly is linked to disassembly, as the microtubule mass and percentage polymer do not dramatically increase in MKs when proplatelets are extended. On the other hand, one can ask: is microtubule assembly a prerequisite for proplatelet elongation? Experiments addressing this issue have temporally arrested microtubule assembly with drugs and found that these drug-treated MKs continue to elongate their proplatelets and do so at rates close to normal rates .
These experiments support the concept that proplatelet elongation depends primarily on microtubule sliding. To test this notion, proplatelets were permeabilized using 0.5% Triton X-100 in a microtubule-stabilizing buffer, a condition that retains cytoplasmic dynein. Motor proteins were then reactivated with ATP, causing the permeabilized proplatelets to elongate, and do so at average rates of 0.65 ± 0.13 μm min−1. In some preparations, individual microtubules are extruded from the ends of the permeabilized proplatelets by the sliding process, demonstrating that microtubules can indeed slide past each other within proplatelets .
Gaps in our understanding of platelet release
To understand fully even the final stages of platelet production, several questions remain to be addressed. First, what is the basic process by which individual platelets release from the ends of proplatelets? Surprisingly, the actual mechanical events involved in the release of individual platelets have never been observed. Second, during this process, how does the marginal microtubule coil form inside the nascent platelet? Third, how and when is the membrane skeleton of the platelet assembled? Fourth, does the microtubule motor power the final cytokinesis of platelets from proplatelets? Fifth, are the microtubule movements coupled throughout the length of proplatelets? And last, how dynamic are the microtubules in proplatelets, that is, how fast do they turn over?
We estimate that proplatelet shafts contain, on average, 200 MTs, each ∼100 μm in length. This estimate is based on the number of MTs in cross sections of proplatelet shafts, on the length of proplatelets (0.5 mm) and on the length of the major MT in a resting platelet.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interests.