Dr KikukoTakeuchi Osaka Prefectural College of Health Science, Habikino 3-chome, Habikino-shi, Osaka 583, Japan.
The platelet-sized particle formation in the human megakaryoblastic leukaemia cell line MEG-01 and its subline MEG-01s was examined. MEG-01 and MEG-01s cells spontaneously released platelet-sized particles into the culture medium, in which the cells occasionally extended cytoplasmic processes similar to those of megakaryocyte proplatelets. Scanning electron microscopic images showed cytoplasmic processes elongated from blebs on the MEG-01 and MEG-01s cell surface and were constricted between segments of platelet size. Immunofluorescence staining with anti-tubulin antibody showed that the cytoplasmic processes contained microtubules that were organized into a ring, which is a characteristic of circulating platelets. Some platelet-sized particles, probably released by ruptures at the sites of the process constriction, were metabolically active in an MTT assay (about 50%). Some particles also expressed the platelet-specific glycoproteins GPIIb, IIIa and GMP-140. Rarely, in response to thrombin, particles underwent a shape change from spherical to a shape with irregular membrane protrusions and fine filopodia, and aggregating with one another. The particles also had increased GMP-140 (P-selectin) expression with the addition of thrombin. These results show the usefulness of the MEG-01 and MEG-01s cell lines for the study of thrombopoiesis.
Megakaryocyte precursors originate from pluripotential stem cells and undergo a complex maturation process including the formation of a polyploid nucleus ( Hoffman, 1989). Platelets are shed from mature megakaryocytes by a series of events which are still incompletely understood; the main reason for this is the difficulties in obtaining sufficient numbers of megakaryocytes for studies of megakaryocytopoiesis and thrombopoiesis.
Using several human cell lines exhibiting some morphological and biochemical characteristics of megakaryocytes, it has been revealed by fluorescence and electron microscopy that these cells release platelet-like particles ( Sledge et al, 1986 ; Greenberg et al, 1988 ; Tange et al, 1988b ; Takeuchi et al, 1991 ; Nagano et al, 1992 ). In a few cell lines, however, these particles have not yet been isolated or characterized. The clonal human megakaryoblastic leukaemia cell line MEG-01 and its subline MEG-01s, established by Ogura et al (1985 , 1988), display phenotypic properties that closely resemble those of megakaryoblasts but not other blood cell lineages. These properties demonstrate increased expression after treatment of the cells with phorboldiester, TPA ( Murate et al, 1991 ). The cell lines also release particles identified by a characteristic marginal microtubule and by the localization of platelet-specific glycoprotein (GPIIb/IIIa) in the plasma membrane ( Takeuchi et al, 1991 , 1995). Inhibitors of DNA synthesis, such as aphidicolin, enhance the particle release ( Takeuchi et al, 1995 ). On the basis of these studies, we have examined the mechanism(s) by which particles are formed from MEG-01 and MEG-01s cells and now report that these cells formed long beaded processes which resembled those observed in normal megakaryocytes. Their rupture may yield particles with sizes and functions similar to individual platelets.
MATERIALS AND METHODS
Cell and cell culture
The establishment and properties of the human megakaryoblastic cell lines MEG-01 and MEG-01s have been described in detail previously ( Ogura et al, 1985 , 1988). MEG-01s cells have almost the same characteristics as MEG-01 cells except for three properties: (a) MEG-01s cells proliferate in single suspension, whereas some MEG-01 cells adhere to the culture dish, (b) MEG-01s cells are a little smaller than MEG-01 cells, with diameter of 20–40 μm, and (c) the expression of platelet glycoprotein GPIIb/IIIa in MEG-01s cells is weaker than that in MEG-01 cells. MEG-01 and MEG-01s cells were generally cultivated in RPMI 1640 medium (Gibco, Grand Island, N.Y., U.S.A.) supplemented with 10% fetal bovine serum (FBS) (MAB Co., Rockville, Md., U.S.A.) at 37°C in a humidified atmosphere of 5% CO2 in an incubator. For fractionation of particles, MEG-01 and MEG-01s cells were cultivated in Dulbecco's modified Eagle's medium (DMEM, Nissui Seiyaku, Tokyo, Japan) supplemented with 1% FBS for 2 d ( Takeuchi et al, 1995 ).
Isolation of human blood platelets
Platelets were obtained as described previously ( Takeuchi et al, 1991 ). All procedures were carried out at 37°C.
Fractionation of MEG-01 and MEG-01s particles
The fractionating procedure of particles has been described previously ( Takeuchi et al, 1995 ). Briefly, 2 d after being plated, the cells were treated with 2–5 μg/ml of aphidicolin in culture medium. After 2 d the cells in the medium were stabilized with 0.5–1 μg/ml of taxol (Wako Pure Chemical Industries, Osaka, Japan) at 37°C for 1 h and suspended in a solution of citrate–citric acid–glucose (final concentration: 14.2 m m sodium citrate, 10.8 m m citric acid, 9.3 m m glucose, pH 6.7). The cell suspension was then centrifuged at 40 g for 10 min. The supernatant was concentrated by centrifugation and then washed three to five times with Tyrode-HEPES buffer (136.9 m m NaCl, 2.7 m m KCl, 11.9 m m NaHCO3, 2 m m CaCl2, 1 m m MgCl2, 5.5 m m glucose, 10 m m HEPES, pH 7.4).
Anti-tubulin immunofluorescence was performed as previously described ( Takeuchi et al, 1990 ). Anti-tubulin antibody was from Amersham (Tokyo).
Scanning electron microscopy
MEG-01 and MEG-01s cells and their particles in Tyrode-HEPES buffer (4 × 108/ml) were prefixed in 2% paraformaldehyde dissolved in phosphate-buffered saline (PBS) for 10 min at room temperature and placed on a glass coverslip using a cytospin centrifuge (Shandon Cytospin 3, Runcorn, U.K.) and rinsed twice in PBS. The particles on the coverslip were fixed in either 2% glutaraldehyde for 2 h or in 1% glutaraldehyde for 1 h and then in 2% osmium tetroxide for 1 h, and washed three times in PBS. After fixation, the specimens were dehydrated in a graded ethanol series, subjected to critical-point drying in liquid CO2, coated with platina-palladium, and examined with a JEOL scanning electron microscope (JSM-5400; JEOL, Tokyo, Japan).
Viable particles were measured by a colourimetric assay, based on the tetrazolium salt MTT (3-(4,5-dimethylthiazol-2,5-diphenyltetrazolium bromide; Sigma Chemical Co., St Louis, Mo.). MTT dissolved in PBS at 5 mg/ml (10 μl) was added to the particle or platelet suspension (100 μl) containing the same cell numbers (0.2–4.0 × 107/ml) in a volume ratio of 1:10 and mixed. The samples were incubated for 3–4 h. Acid–isopropanol (100 μl of 0.04 N HCl in isopropanol) was added to the samples, mixed and read on a Beckman Spectrophotometer (DU-65), using a test wavelength of 570 nm and a reference wavelength of 630 nm.
Flow cytometric analysis of GPIIb/IIIa and the P-selectin expression in MEG-01 and MEG-01s particles or human platelets was performed by an indirect immunofluorescent staining technique using primary monoclonal antibodies. Briefly, particles or platelets (2–8 × 108/ml) were fixed in 2% paraformaldehyde (in PBS, pH 7.4) for 10 min at room temperature, and then washed with PBS. The samples were incubated with monoclonal antibodies against either anti-GPIIb/IIIa (Kyowa Medical Co., Tokyo, Japan) or P-selectin (Takara Shuzo Co., Kyoto, Japan) for 60 min at 37°C, washed with PBS and incubated with fluorescein isothiocyanate (FITC)-conjugated goat anti-mouse IgG (Cappel, Durham, N.C.) for 30–60 min at room temperature. After the samples were washed with PBS, immunofluorescence was analysed on a flow cytometer (Epics-XL, Coulter Corp., Hialeah, Fla.).
Thrombin-induced function assay
MEG-01 and MEG-01s particles (2–3 × 108/ml) were incubated with 1–5 U/ml thrombin (Sigma, St Louis, Mo.) in Tyrode-HEPES buffer at 37°C. After 15 min, particle preparations were fixed in 3.7% paraformaldehyde dissolved in PBS for 10 min. After being washed with PBS, the particle preparations were fixed for the analysis of morphology by scanning electron microscopy and expression of GPIIb/IIIa and P-selectin by flow cytometer.
The cytoplasmic processes of MEG-01 or MEG-01s cells
The MEG-01 and MEG-01s cells spontaneously released platelet-sized particles into the culture medium ( Takeuchi et al, 1991 , 1995) in which the cells extending cytoplasmic processes had occasionally been visible as megakaryocytes displaying proplatelets ( Figs 1B and 1C). The thick and thin processes appeared to be beaded due to constrictions at intervals, and to be partly broken into fragments (Fig 1D). The frequency of cells displaying cytoplasmic processes was not more than 0.5%. When platelets were stained using the antitubulin antibody, the fluorescence image showed the characteristic ring structure in the cytoskeleton, which consisted of a circumferential band of microtubules just inside the plasma membrane ( Takeuchi et al, 1991 ). We have previously reported that the particles released from MEG-01 and MEG-01s cells have an anti-tubulin staining pattern similar to that found in platelets ( Takeuchi et al, 1991 , 1995). Figs 1E and 1F show the fluorescence images of the cytoplasmic processes when MEG-01s cells with beaded processes, cultivated on a glass coverslip, were stained with anti-tubulin antibody. Brightly stained microtubule rings were observed in clusters connecting with the cell body by much thinner processes (Fig 1E). The processes detached from the cell body also showed the rings with the thin microtubule tails (Fig 1F). The size of the microtubule rings was approximately 1.5–5 μm in diameter, which was within the same range as those shown in human platelets. The size was also comparable to that measured from the images of the MEG-01s particles shown by Nomarski differential interference contrast microscopy ( Takeuchi et al, 1995 ).
Cytoplasmic processes observed by scanning electron microscopy
The surface features of MEG-01 and MEG-01s cells extending processes were also examined by scanning electron microscopy. Single suspended cells were mostly spherical or ovoid ( Figs 2A–C), whereas the adherent MEG-01s cells had various contours (data not shown). Their surfaces were covered with varying numbers of microvilli ( Figs 2C and 2D) or blebs ( Figs 2A and 2B). Some of the blebs were larger, extending into a tear-drop shape (Fig 2A) which were attached to the MEG-01 cell body by only a thin neck (Fig 2B). In some cases a cell produced a single long process, as long as 30–40 μm and about 0.2 μm in diameter, with irregularly-spaced constrictions (Fig 2C), but occasionally the processes were in clusters from a site on the cell surface. The swelling of the processes separated by developing constriction (Fig 2D, arrow) resembles platelets in size and shape.
Processes related from MEG-01 and MEG-01s cells
MEG-01 and MEG-01s cells were cultivated in medium containing 2 μg/ml aphidicolin, an enhancer of particle release from MEG-01 and MEG-01s cells ( Takeuchi et al, 1995 ), for 2–3 d because the particle release ( Takeuchi et al, 1995 ) and process formation (data not shown) were maximal on days 2–3 of the treatment. The cell suspension was centrifuged at 40 g for 10 min and the resultant supernatant contained smaller process fragments, which were platelet-sized particles thought to have split off from longer processes (Fig 3A). Some particles had one tail or a tail at opposite sites, as though a thin constriction bridge of the beaded process had broken off (Fig 3A, arrow). Longer process fragments, whether single (Fig 3B) or cluster ( Figs 3C and 3D), were in the pellet and resembled a string of beads. As the process fragments in culture medium have several features in common with processes still attached to MEG-01 and MEG-01s cells, this suggests that they fragmented into variable lengths of 1–50 μm and were released into the culture medium.
To ascertain whether the particles in the fragments released from MEG-01 and MEG-01s cells were viable, and similar to those of blood platelets, we used the MTT method, because the particles were too small to measure by the trypan blue exclusion method. When MTT, a pale yellow substance, is incubated with living cells, it is cleaved and produces a dark blue formazan product; it does not do so with dead cells. MTT was added to the MEG-01s fragments or human blood platelets in Tyrode-HEPES buffer. After 4 h incubation at 37°C the formazan product was solubilized by the addition of acid–isopropanol. The relationships between the number of particles (or platelets) and the amount of MTT formazan produced are shown in Fig 4. The absorbances were approximately proportional to the number of particles (or platelets). However, the amounts of MTT formazan produced by the particles were 50% of those produced by platelets in the same numbers. This shows that the particles had only half the metabolic activities of the platelets.
Function of MEG-01 and MEG-01s particles
By a variety of stimulators including thrombin, platelets rapidly undergo the processes of shape change, adhesion, aggregation and secretion. In the present study most of the fractionated MEG-01s particles had not a discoid but a spherical/ellipsoidal appearance in the absence of platelet agonists (Fig 5A). With the addition of thrombin, some of them changed their shape to one with irregular membrane protrusions and several filopods, similar to human platelets (Fig 5B). Using their filopods, they interacted directly and became aggregates (Fig 5C). However, shape change and aggregation were markedly impaired in many MEG-01 and MEG-01s particles in response to thrombin (1–5 U/ml)/ADP (10–100 μm), and thus this aggregation was not detected by measuring the increase in the light transmission in an aggregometer (data not shown).
We have previously shown that the particles released from MEG-01 cells contained the platelet-specific surface glycoprotein GPIIb/IIIa ( Takeuchi et al, 1991 , 1995). To determine the extent of GPIIb/IIIa expression on the particles, fractionated MEG-01 particles were examined by flow cytometric analysis after staining using an anti-GPIIb/IIIa antibody that binds to resting platelets. The proportion of GPIIb/IIIa-positive particles was 28%, whereas that of the positive human platelets, under identical conditions, was 99.5% (Fig 6). Platelet agonists can evoke the release reaction in vitro which involves secretion of the contents of α-granules and expression of P-selectin (GMP-140, CD62) on the α-granular membrane. Fig 7 shows P-selectin expression on MEG-01 particles measured by flow cytometry when 1 U/ml thrombin was added. The percentage of specific CD62+ particles increased from 17.6% to 29.2% with the addition of thrombin, even when allowing for underestimation due to the differences between negative controls, which was 3.1% with and 10.1% without thrombin ( Figs 7A and 7B). The percentage of negative controls with second antibody usually varied with MEG-01 particle specimens, and was slightly higher in the particles than in the human platelets. In contrast, the specific CD62+ platelets increased, under identical conditions, from 63.0% to 80.0% (Fig 7C). Both negative controls were < 0.3% (data not shown). A considerable percentage of the CD62+ population of the MEG-01 particles or human platelets in the absence of thrombin may have been induced by accidental activation occurring during the isolation procedures.
Attempts to develop an in vitro system using cell lines to clarify the mechanisms of platelet production have been made by many investigators, and the release of platelet-like particles has already been reported in several human cell lines of megakaryotic lineage ( Sledge et al, 1986 ; Greenberg et al, 1988 ; Tange et al, 1988b ; Takeuchi et al, 1991 ; Nagano et al, 1992 ). However, little is known of how the release actually occurs. We have previously shown that MEG-01 and MEG-01s cells release platelet-like particles ( Takeuchi et al, 1991 , 1995) and the results of the present study suggest that the release occurs via a mechanism such as the flow model ( Radley & Scurfield, 1980; Stenberg & Levin, 1989). This contention is based on the following evidence. First, the MEG-01 and MEG-01s cells extended processes with constrictions ( Figs 12–3), which closely resemble those found in in vivo megakaryocytes ( Radley & Scurfield, 1980; Scurfield & Radley, 1981) and those supported in culture ( Choi et al, 1995 ; Cramer et al, 1997 ). The process formation first observed in megakaryocytes in bone marrow has been demonstrated in cultures of megakaryocytes of many kinds of animals ( Scurfield & Radley, 1981; Leven & Yee, 1987; Topp et al, 1990 ; Inoue et al, 1993 ; Debili et al, 1995 ; Choi et al, 1995 ; Guerrieróet al, 1995 ). Cytoplasmic processes often develop constrictions between segments of platelet size that give a beaded appearance. A rupture at the site of constriction is thought to release the platelets. Second, the particle fractions prepared from MEG-01 and MEG-01s culture media contained short fragments with tapered ends, often at opposite ends (Fig 3A), and long fragments with several constrictions ( Figs 3B and 3C), but mainly particles without a tail. The fragments with a tail were thought to be shapes detected immediately after rupture of the cytoplasmic processes attached to the MEG-01 cell body. The fragments are comparable to the proplatelet process observed in megakaryocytes. Third, the immunofluorescence imaging using anti-tubulin antibody showed that the cytoplasmic processes were microtubule-containing structures ( Figs 1E and 1F). As shown in our previous reports ( Takeuchi et al, 1991 , 1995), and by the present results, microtubule rings were observed not only in the cytoplasmic processes of MEG-01 cells in culture medium ( Figs 1E and 1F), but also in some platelet-sized particles fractionated from MEG-01 culture medium ( Takeuchi et al, 1995 ). Approximately 20–30% of the fractionated particles had microtubule rings, which were not localized as a mem-brane cytoskeleton in the subplasmalemmal region of the particles, but were located in a coil only in the equatorial region ( Takeuchi et al, 1995 ). Thus, we have used the distinctive morphology of microtubule organization as the first indication that the particles were similar to platelets. The presence of microtubule rings in both cytoplasmic processes and particles from MEG-01 cells indicated that particle formation in MEG-01 cells is similar to platelet formation from megakaryocyte processes. Radley & Haller (1982) observed that spontaneous cytoplasmic process formation was reversed by vincristine. Similarly, Leven & Yee (1987) indicated that both colchicine and vincristine inhibited the process formation stimulated by thrombocytoplasmic plasma. Thus, the formation of microtubule-based cytoplasmic processes may be an important step in platelet formation, although it is still unclear what role the microtubules play in cytoplasmic fragmentation. The flow model suggests that the demarcation membrane is involved in the formation of attenuated processes by megakaryocytes ( Stenberg & Levin, 1989). In our present study we could not examine these relationships because of the difficulty due to the low frequency of the formations of both demarcation membrane and cytoplasmic processes; MEG-01 and MEG-01s cells are megakaryoblasts at a relative early stage in which those structures are seldom found and divide predominantly into daughter cells. Only a few cells differentiate to develop the cytoplasmic processes which break, releasing small fragments, thus forming platelets.
MEG-01 and MEG-01s particles are similar to platelets in their distinctive morphology of microtubule organization and because of the presence of several proteins important for platelet function (GPIb, GPIIb/IIIa ( Takeuchi et al, 1991 , 1995) and P-selectin). In addition, this study showed the functional similarities between MEG-01 and MEG-01s particles and platelets with respect to thrombin-induced shape change (Fig 5B), aggregation (Fig 5C) and expression of GPIIb/IIIa and P-selectin (Fig 7), although the number of shape-changed particles and the amount of aggregated particles were very small. If many dead particles and cell fragments were contained, these results are reasonable. The MTT assay results also support such a possibility (Fig 4). The amount of formazan generated in this assay is directly proportional to the activity of mitochondria. The particle fraction had only about 50% of the metabolic activity of platelet fractions. It seems likely that half of the particles were inactive or dead, but the possibility that all of the particles had low enzyme activity could not be excluded. As MEG-01 and MEG-01s cells are metabolically active cells, such a contamination in the particle fraction may give incorrect results. The microscopic images, in which the formazans produced by MEG-01 and MEG-01s cells were large and thick, and those by MEG-1 and MEG-01s particles and human platelets were very thin (data not shown), clearly showed that the particle fraction did not contain MEG-01 and MEG-01s cells.
Platelet activation requires various membrane glycoproteins specific to platelets, such as GPIIb/IIIa in the surface membrane and P-selectin in the α-granular membrane. The present MEG-01 particles certainly displayed P-selectin expression, increased by thrombin. The low level of the expression probably indicates the presence of particles with inactive P-selectin and/or without P-selectin. The platelets generated from CD34+ cells in vitro ( Choi et al, 1995 ; Cramer et al, 1997 ), mostly lose their discoid-shape and extend spiky and bulky pseudopods as if they were activated. They showed no marginal microtubule bands characteristic of the discoid-shaped platelets. In contrast, shape change and aggregation were so markedly impaired in many MEG-01 particles that they could not be detected by aggregometer. Their ring-like structure of marginal microtubule bands was therefore retained ( Figs 1E and 1F). These differences in features between MEG-01 particles and culture-derived platelets may be related to the levels of expression of glycoproteins, GPIIb/IIIa and P-selectin, which respond to the stimulus and recruit adjacent platelets to form aggregates accompanying the shape change. The culture-derived platelets expressed higher levels of the glycoproteins than MEG-01 particles, and thus were able to induce their aggregation. Although other components in the signal transduction pathway are also required for platelet aggregation, their expression in MEG-01 particles was not examined in this study, except for mito-chondria dehydrogenase activity analysed by MTT assay.
Unfortunately we could not separate the beaded processes connected with the MEG-01 and MEG-01s cell bodies which probably may be functional. The centrifuged culture supernatants contained mainly particles/fragments detached from the cell body. The connected processes were in the pellet together with the MEG-01 and MEG-01s cells. In addition, we could not include any steps to enrich the metabolically active particles in our preparation, as they may decrease sensitivity to stimuli during preparation.
The morphological features observed here in MEG-01 and MEG-01s cells closely approximate those in megakaryocytes, thus allowing the cell lines to be a useful model for the study of megakaryocyte maturation as evidenced by the process formation and cytoplasmic fragmentation into platelets. To obtain a large yield of functional particles, many problems remain to be solved, including candidate factors for maturation, the fragmentation of cytoplasmic processes and particle preparation.
We believe that studies with the MEG-01 and MEG-01s cell lines will contribute to development of new forms of treatment of megakaryoblastic leukaemia based on the use of various components including cytokines to induce differentiation in malignant cells and to clarify the nature of the clinical abnormalities in megakaryopoiesis.
This work was supported in part by a Grant-in-Aid for General Scientific Research from the Ministry of Education, Science, Sports and Culture of Japan.