|Cancer||Cell death||Cell cycle||Cytoskeleton||Exo/endocytosis||Differentiation||Division||Organelles||Signalling||Stem cells||Trafficking|
Cell Biology International (2004) 28, 299310 (Printed in Great Britain)
Shape transformation and cytoskeletal reorganization in activated non-mammalian thrombocytes
Kyeng‑Gea Leeab, Todd Millerab, Ivan Anastassovab and William D. Cohenab*
aDepartment of Biological Sciences, Hunter College, CUNY Graduate Center, New York, NY, USA
bMarine Biological Laboratory, Woods Hole, MA, USA
The nucleated thrombocytes of non-mammalian vertebrates are partially flattened, ovoid cells morphologically distinct from mammalian platelets, and the extent of their functional equivalence is unknown. To test whether they resemble platelets in having similar F-actin-based post-activation stages, rapid fixation/extraction/labeling methods were developed to reveal cytoskeletal organization in dogfish thrombocytes by confocal microscopy. Unactivated cells contained cortical F-actin plus denser F-actin co-localizing with outer marginal band (MB) microtubules. In the post-activation sequence, determined for the first time by continuous observation of individual thrombocytes following thrombin perfusion, cells rounded and blebbed, spread, and eventually flattened extensively. The MB twisted and then became disorganized, with microtubule bundles remaining centrally located and associated with nuclear clefts. In contrast, F-actin occupied blebs and outward-spreading cytoplasm, initially in spiky projections, then predominantly in stress fibers, and inhibitors of F-actin assembly or myosin ATPase blocked shape changes. Thus, the post-activation stages and cytoskeletal events observed in nucleated thrombocytes were found to parallel those of platelets.
Keywords: Marginal band, Nucleated thrombocyte, F-actin, Microtubule, Activation, Apoptosis.
Thrombocytes play a vital role in vertebrate wound healing and innate immunity. Mammalian thrombocytes, i.e. blood platelets, have been the subject of innumerable studies, and their structural and physiological properties are well known (e.g. Allen et al., 1979; Bearer et al., 2002; Kowit et al., 1988; White, 1971). In contrast, although their properties are generally assumed to be similar to those of platelets, non-mammalian vertebrate thrombocytes have received relatively little attention. Morphologically, the two cell types are quite distinct: unactivated platelets are highly flattened, minute (∼3 μm diameter), anucleate discoids, whereas fish, amphibian, reptilian, and avian thrombocytes are thicker nucleated ellipsoids of more typical eucaryotic size (∼10 μm or greater on the long axis; Mainwaring and Rowley, 1985; Shepro et al., 1966).
Despite these differences in size and shape, unactivated non-mammalian thrombocytes, referred to hereafter as nucleated thrombocytes, have two highly distinctive structural features in common with platelets: a “canalicular” membrane system just beneath their surface (Daimon and Uchida, 1985; White and Clawson, 1980), and a prominent marginal band of microtubules (MB) in their plane of flattening (Fawcett and Witebsky, 1964; Lee et al., 2002; Rowley et al., 1988; Schwer et al., 2001). Upon activation of both nucleated thrombocytes and platelets, cellular aggregation, adhesion and morphological alterations are observed. Aggregation of both cell types can be induced by agents such as ADP and thrombin (Aledort, 1971; Haslam, 1964; O'Toole et al., 1994; Pica et al., 1990), and activation of both appears to involve an integrin-like surface protein complex and eicosanoid signaling pathways (Hill et al., 1999; Lloyd-Evans et al., 1994; Passer et al., 1997; Wang and Herman, 1997). For activated platelets, the morphological changes include membrane protrusions, pseudopod formation, and spreading on a surface (Allen et al., 1979; Zucker and Nachmias, 1985), and these activities are correlated primarily with the assembly of F-actin rather than MB reorganization (Debus et al., 1981; Nachmias, 1980).
Although it has been shown that nucleated thrombocytes are similar to platelets with respect to structure and activation signaling, neither their post-activation stages nor their related cytoskeletal changes have been demonstrated. In the work reported here, we tested the hypothesis that nucleated thrombocytes truly resemble platelets functionally by progressing through post-activation stages based on F-actin reorganization. To do so, we examined structural features of unactivated (“resting”) nucleated thrombocytes, the sequence of morphological alteration following activation in individual cells, and the accompanying cytoskeletal reorganization. In addition, inhibitors of F-actin assembly and myosin II ATPase were used to test its function in post-activation events. Dogfish thrombocytes were employed as a model system because they exhibit typical nucleated thrombocyte morphology, are readily obtained (Cohen et al., 1997), and have been the subject of two previous studies (Mainwaring and Rowley, 1985; Shepro et al., 1966).
2 Materials and methods
2.1 Preparation of leucocytes enriched in thrombocytes
Smooth dogfish, Mustelus canis, were provided by the Marine Resources Center of the Marine Biological Laboratory. One volume of blood (usually 5 ml) was drawn into a syringe containing an equal volume of non-pyrogenic 3% saline (Baxter Corp.; isotonic to Elasmobranch blood plasma and seawater). The mixture was divided and centrifuged (IEC clinical, 70×g, 2 min), sedimenting most erythrocytes into a pellet whilst white cells remained suspended. The upper half of the supernatant, rich in thrombocytes, was collected. For experiments requiring higher cell density, the collected supernatant was centrifuged again (70×g, 4 min) in several smaller aliquots (1–2 ml), and the pellets were pooled to half the original blood volume in non-pyrogenic 3% saline.
2.2 Activation by thrombin perfusion
Perfusion chamber slides were devised to observe and capture shape changes of individual selected cells upon exposure to various solutions. A detergent-washed 22×22 mm coverslip was mounted on Fisherbrand “Superfrost Plus” glass slide (improved sample adherence) with just enough water to suspend the coverslip. Two edges of the coverslip parallel to slide length were sealed with nail polish, followed by several washes with water and 3% saline. Prior to mounting the coverslip, rows of scratches were made on the slide, approximately parallel to slide length, where the open coverslip edges would be located. These scratches facilitated flow. In some experiments, surfaces were coated with 0.01% poly-lysine, followed by several washes with non-pyrogenic 3% saline. Use of poly-lysine improved initial cell adhesion, but had no effect on the activation sequence. With careful handling, perfusion was also possible without sealing coverslip edges, allowing for easy removal of the coverslip and facilitating labeling and washing in certain experiments. The same results were obtained whether the edges were sealed with nail polish or not.
A sample of white cell suspension (∼15 μl), enriched for thrombocytes as described above, and pre-treated with F-actin inhibitors in some cases (see below), was loaded into an open edge of the perfusion chamber, and cells were given ∼3 min to initiate adhesion. With cells under continuous observation, activation was triggered by perfusion with 10 μl thrombin solution (Sigma T-4648 bovine thrombin 25 U/ml, in 10 mM CaCl
In some experiments, the thrombocyte-enriched white cell suspensions were incubated in non-pyrogenic 3% saline containing cytochalasin D (5 μg/ml, 2 h) or latrunculin B (1 μM, 1 h) prior to thrombin exposure. Respective solvent controls were 0.5% and 0.2% DMSO. After incubation, samples were loaded in perfusion chambers and exposed to thrombin as described above. For myosin II inhibition experiments, white cell suspensions were incubated in 20 mM 2,3-butanedione monoxime (BDM; Cramer and Mitchison, 1995; Ostap, 2002; Silverman-Gavrila and Forer, 2003) for 1 h, exposed to thrombin for 10 min, and fixed in 2% formaldehyde. For the control, the same white cell suspension was treated as above except for BDM incubation. Although we employed BDM as a myosin-II ATPase inhibitor, it should be noted that its specificity has been questioned because of effects on calcium signaling (Ramachandran et al., 2003).
In some experiments, thrombocytes in whole blood were allowed to activate spontaneously by spreading 100 μl dogfish whole blood on detergent-cleaned coverslips.
2.3 Fixation and permeabilization
Unactivated thrombocytes were either simultaneously fixed and permeabilized in 4% formaldehyde, 0.4% Triton X-100, and 10 mM TAME in PEM [=100 mM PIPES, 5 mM EGTA, and 1 mM MgCl
At selected time points or morphological stages, thrombin-activated cells were simultaneously fixed and extracted with 2% formaldehyde, 0.6% Brij-58, and 10 mM TAME in PEM for a minimum of 10 min.
For spontaneously activated cells, coverslips were immersed in 1% formaldehyde in Elasmobranch Ringer's solution for 10 min. Unattached red cells, and white cells other than activated thrombocytes and granulocytes, were washed away during fixation. Fixation was followed by several washes in PEM, permeabilization for 10 min in PEM containing 0.4% Triton X-100, 10 mM TAME, and protease inhibitor cocktail [Sigma P-8340], a PEM wash, and 30 min post-fixation in PEM containing 4% formaldehyde.
2.4 Fluorescence labeling and microscopy
After a wash in PBS (pH 7.2), fixed and extracted samples from preparations described above were sequentially labeled for F-actin with 260 nM Texas-Red phalloidin (Molecular Probes, T-7471), for tubulin with a 50:50 mix of mouse monoclonal anti α- and β-tubulin antibody (Sigma T-9026 and T-4046) as the primary, followed by FITC-goat anti-mouse IgG (Fab; F-8521) as the secondary, and for DNA with 3 μM DAPI (Sigma D-9542). PBS was used as the wash medium for all treatments. Alternatively, samples were triple-labeled in PBS containing mouse monoclonal anti-α and anti-β tubulins pre-bound with ZenonTM Alexa Fluor 488 Fab (Z-25002, 1:1 by mass; Molecular Probes), 260 nM Texas-Red phalloidin, and 3 μM DAPI for a minimum of 10 min, followed by a wash in PBS.
Labeled samples were examined using either a Zeiss epi-fluorescence microscope with Zeiss Neofluar 100×/1.3 NA oil objective, equipped with a Nikon 950 digital camera, or a Zeiss LSM 510 confocal microscope with Plan-apochromat 63×/1.4 NA DIC oil objective. For the LSM 510, images were recorded as stacks of optical sections and are displayed in the figures either as individual sections or entire stacks as noted in figure legends. Some figures include 3D computer-generated views of stacks. Adobe Photoshop was used to crop and label images, and to contrast-enhance DIC images. During acquisition and processing of confocal images, care was taken not to oversaturate the signals. Although individual microtubules cannot be resolved within the marginal band because of tight bundling, visibility of individual microtubules or minor microtubule bundles external to the band served as a guideline for adequate detector settings and image processing (see Figs. 2 and 3). For studies involving co-localization (i.e., multiple-labeled samples), singly labeled samples were used to check and adjust settings so as to eliminate channel bleeding.
Unactivated dogfish thrombocytes: (A) phase-contrast; (B) DIC. n=nucleus. Bar, 10 μm.
Cytoskeleton of unactivated, double-labeled thrombocyte (confocal): (A) FITC-labeled tubulin distribution; note multiple microtubule bundles in the MB and single microtubules or minor bundles internal to the MB; (B) Texas-Red phalloidin; a band of F-actin colocalizes with the MB, but is thinner than that of microtubules. Network-like F-actin also appears in the interior except for the central dark region where the cell cortex is out of optical plane; (C) Superimposition of (A) and (B). The orange overlap color is pronounced at the MB periphery, whereas streaks of green (microtubules) appear at the MB inner edge; (D) DIC image of the same cytoskeleton, in which bundles of microtubules are distinguishable (arrows), n=nucleus. Bar, 10 μm.
Edge-on view of double-labeled thrombocyte cytoskeleton (confocal): (A) tubulin distribution, near-planar band; (B) F-actin; the MB region displays the highest density of F-actin in a band thinner than that of tubulin in (A). A network of F-actin is also present above and below the MB plane; (C) Superimposition of (A) and (B); (D) DIC. Bar, 10 μm.
3.1 Nucleated thrombocytes prior to activation
Unactivated thrombocytes were ovoid and somewhat flattened in shape and 14∼16 μm in length(Fig. 1). The cell contour appeared smooth and the nucleus, also flattened and ovoid, occupied a large proportion of cell volume. The peri-nuclear region was characterized by granular bodies, visible under both phase-contrast and DIC. As observed by immunofluorescence microscopy, resting dogfish thrombocytesexhibited a prominent circumferential microtubule bundle (Fig. 2A) in the plane of flattening, i.e. a thick marginal band (MB), with few microtubules elsewhere. Confocal image analysis showed that a band of F-actin also existed in the cytoskeleton of the unactivated thrombocytes, and that it co-localized with MB microtubules (Figs. 2 and 3). However, the F-actin band was thinner (∼0.5 μm) than the microtubular MB (∼1.4 μm), and was restricted to the peripheral MB region (Fig. 2C). Edge-on views confirmed the presence of the F-actin band and its co-localization with the outer MB microtubules (Fig. 3). F-actin was also observed elsewhere in the cytoskeleton. Optical sectioning showed that most of it was localized to the cell cortex in mesh-like form (Fig. 4A, C), with occasional dense patches. The region deeper into the interior was largely devoid of F-actin (Fig. 4B, D). Mid-optical sections through cells oriented on their edge revealed that the MB was enclosed by, and abutting, the cortical F-actin (Fig. 4E).
F-actin in the cortex (confocal): (A) optical section through one cortical region of unactivated cell in flat orientation; note the meshwork-like pattern of F-actin with a patchy region of high intensity (arrowhead); (B) section through the mid-region showing F-actin in a band and lack of internal signal; (C) section through the opposite cortical region (relative to A), with similar F-actin distribution as in (A, arrowhead: patch); (D) optical section through mid-Z axis of another cell viewed on edge, exhibiting F-actin in the cortical region; (E) composite of (D) with tubulin distribution, showing location of MB cross-sections (arrows) at opposite extremities of the cortical network. Bar, 5 μm.
3.2 Thrombin-induced activation
Perfusion with thrombin was used to induce activation of individual pre-selected thrombocytes while they were continuously observed under the coverslip (Fig. 5). Overall, cell morphology changed from an elongated, somewhat flattened ovoid, to a refractile spheroid, and then to a highly flattened, relatively circular spreading “pancake-like” form. Thrombocytes in initial stages of activation (Fig. 5A) adhered to the slide and resisted being dislodged by the perfusion flow (Fig. 5B). Shortly after activation, the cell developed a point at one end of its long axis (Fig. 5C) and within ∼10 s became spheroidal (Fig. 5E). Blebs and pointed projections were constantly protruding and contracting from the spheroid and sheet-like extensions emanated in a circular pattern within ∼1 min (Fig. 5F). At ∼3.5 min, cytoplasmic protrusions were evident and the nucleus appeared to be lobed (Fig. 5G). Subsequently, with further cell spreading, the nucleus became lobulated and appeared to be fragmented (Fig. 5H, I). Note: A movie of the described sequence is available. The sequence reported in Fig. 5was verified by more than 50 perfusion experiments and observation of several hundred individual cells.
Stages following thrombocyte activation by thrombin perfusion (phase-contrast): (A–B) initial stage; cell has adhered to the slide; (C–D) pointed stages; (E) spheroidal stage (arrows: blebs; arrowhead: spike); (F) transition to pancake stage (arrow: edge of spreading cell); (G) developing pancake with cytoplasmic protrusions (arrowheads, focus on upper surface): (H–I) late pancake stages exhibiting nuclear lobulation. Bar, 10 μm.
3.3 Cytoskeletal reorganization following activation
Thrombocytes activated spontaneously, though somewhat asynchronously, when whole blood was spread on coverslips. When cells were fixed within 15 min of blood withdrawal, various stages of early activation were observed, corresponding to those induced by thrombin (Fig. 6B vs. Fig. 5A–E). MB shape in these cells corresponded very closely to the cell surface contours in different morphological stages of activation (Fig. 6A vs. B). In the pointed stage, the number and location of points in the MB matched that of points in the cell (Fig. 6, cell 3), whereas, in the spheroidal stage, the entire MB was buckled, conforming to cell shape (Fig. 6, cell 4), which also appeared to be folded relative to the unactivated state (Fig. 6D). Confocal sections (stacks) showed that the buckled MB was enclosing the nucleus (Fig. 7). F-actin, which had co-localized as a band along the entire length of the MB in the unactivated cells, no longer did so in the spheroidal stage cells. However, a significant concentration of F-actin remained in the cortical region and also redistributed into the dynamic surface blebs (Fig. 7; Fig. 9).
Early stages of thrombocyte activation (confocal): (A) tubulin distribution; cell 1: relatively unactivated; cells 2 and 3: pointed stage, corresponding to Fig. 5C and D; cell 4 spheroidal stage; cell g: a non-thrombocyte spheroidal cell lacking MB (a putative granulocyte); (B) DIC, same field; (C) higher magnification view of cell 4 in (A), showing bent MB (tubulin, green) and F-actin distribution (red). F-actin at this stage has already lost colocalization with outer MB microtubules; (D) DIC view of (C), revealing that nucleus is correspondingly folded. Bars, 10 μm (A & B); 5 μm (C & D).
MB geometry at the spheroidal stage (confocal): (A) spheroidal cell with a bleb (arrowhead; DIC); (B) composite of DAPI-stained nucleus (blue) and MB microtubules. The MB is bent and encloses the nucleus; (C) 3D computer rotation of (B) by 22.5°; (D) 90°; (E) F-actin distribution. Note: images projected by 3D computer rotation are commonly blurred with increasing angle of rotation such as in (C) and (D). Bar, 5 μm.
Example of a spheroidal cell that is not a thrombocyte (confocal): (A) DIC view; (B) DAPI-stained nucleus; (C) aster-like microtubules (no MB); (D) F-actin distribution, exhibiting irregular surface corresponding to (A). Note: (B)–(D) are stacks. Bar, 5 μm.
Spheroidal cell with blebs and projections (confocal): (A) DIC, showing blebs; (B) nucleus, in two lobes; (C) altered MB; (D) F-actin distribution in an optical section through apical region, exhibiting several blebs; (E) F-actin distribution in a section through basal region, showing both blebs and a projection. Bars, 5 μm.
To test whether marginal band bending was an effector for generation of spheroidal cells or whetherit was a secondary result of actin-based motility, we utilized the myosin II (non-muscle) ATPase inhibitor, BDM. After pre-incubation of thrombocytes with20 mM BDM for 1 h, thrombin-induced spheroidal cell formation was reduced to 12% from a control level of 96%. Another type of spheroidal cell, with aster-like microtubule distribution instead of a typical MB, was identified as a granulocyte (Fig. 6, cell G; Fig. 8). This cell, also containing blebs, was not easily distinguishable from the spheroidal thrombocyte under phase-contrast or DIC. Thus, presence of the MB served to identify activating thrombocytes from other cells. Subsequently, blebbing spheroidal cells initiated nuclear lobulation, and the nuclear lobes were found to have closely associated microtubules (Fig. 9A-C).
Optical sections through apical regions showed F-actin in several blebs, while basal regions showed F-actin in spike-like projections together with blebs (Fig. 9D–E). At a transition to the pancake stage, the nucleus was divided into lobes (Fig. 10B) with microtubule bundles wrapped around the nuclear furrows. These microtubule bundles were disorganized relative to those of MBs and were frequently arranged in a zigzag pattern (Fig. 10C–D). Computer generated 3D rotation revealed a ring-like structure (Fig. 10E). Spiky F-actin projections became prominent in the basal region, with few remaining blebs (Fig. 10F). With constant extension and retraction, these projections attached to the surface, and regions between them became filled with F-actin sheets in a webfoot-like arrangement.
Transition from spheroidal to pancake stage (confocal): (A) DIC, showing the spreading cell edge (arrows); (B) DAPI-stained nucleus, with lobulation and cleft; natural side view; (C) distribution of MB microtubules, natural side view. MB-derived microtubules have taken on a zigzag pattern; (D) composite of (B) and (C), showing microtubules localized within the cleft; (E) 90° computer 3D rotation of (C), revealing that the nucleus-associated microtubules form a ring; (F) F-actin (stack), exhibiting pointed projections (arrowheads) and sheet like F-actin (arrows). Bar, 5 μm.
Early in pancake formation, the nucleus was multi-lobed without marked lobe separation (Fig. 11A). In most cases, microtubule bundles were associated with each nuclear furrow (Fig. 11B–C). While blebs and spiky projections were no longer evident, F-actin was found distributed throughout the cell in fibrous arrangements (Fig. 11D). In full pancake stages, the nuclei of most cells were extensively lobed and sufficiently separatedto appear fragmented (Fig. 12A–B). The microtubule bundles were disorganized to a greater extent than earlier, with many single microtubules or minor bundles projecting from the major ones (Fig. 12C). When superimposed with nuclear distribution, convoluted MT bundles were found in the gaps between nuclear fragments (Fig. 12D), with major bundles wrapped around the neck of the lobes (Fig. 12E). F-actin was now prominent, occupying the entire spreading cell body in the form of stress fibers (Fig. 12F). Pancakes undergoing nuclear fragmentation exhibited increased stress fiber density in patches near the cell center. However, stress fibers arranged in triangular or quadrilateral geometry were also observed in significant numbers of cells. In these pancakes, there was no apparent nuclear lobulation (Fig. 13). Similar observations were made in pancakes resulting from spontaneous activation on glass surfaces.
Early pancake stage (confocal): (A) Nucleus with 3 lobes; (B) microtubule distribution; (C) composite of (A) and (C), showing microtubule bundles associated with clefts of the lobulating nucleus; (D) F-actin distribution, typical of early cell spreading. Bar, 5 μm.
Advanced nuclear lobulation and fragmentation in a pancake stage cell (confocal): (A) DIC, showing the cell edge and the lobulating nucleus; (B) DAPI-stained nucleus with three major lobes; (C) distribution of microtubules; (D) composite of (B) and (C), revealing loops of microtubule bundles associated with nuclear lobes; (E) close-up of (D), with major loops of microtubule bundles wrapped around the lobes; (F) F-actin distribution. Bar, 10 μm. (E) is a confocal section and all the others are stacks.
Pancakes with and without nuclear lobulation (confocal): (A) DAPI, showing the nucleus of a pancake with no lobes (upper) and of another with four lobes (lower); (B) F-actin, exhibiting stress fibers arranged in geometric (triangular) pattern (upper cell, double headed arrows) and in a less defined pattern (lower cell). Bar, 10 μm.
3.4 Inhibitors of F-actin assembly
Cells exposed to 5 μg/ml cytochalasin D for 2 h or 1 μM latrunculin B for 1 h retained their unactivated ovoid morphology, permitting us to test the effects of these inhibitors on shape changes following activation. Cytochalasin D-treated cells failed to undergo thrombin-induced shape changes (Fig. 14, row A). Cells fixed and permeabilized after 10 min of thrombin exposure exhibited MB microtubules similar to those of unactivated thrombocytes. However, F-actin distribution became slightly altered, with patchy cortical actin and a less-well defined MB-associated F-actin band. Treatment with 1 μM latrunculin B for 1 h produced similar results. Upon thrombin perfusion, cells developed a knob-like extension in one end of the long axis, but progression to the spheroidal stage was never observed. MB microtubules exhibited normal unactivated distribution, while F-actin was more disorganized than in cells treated with cytochalasin D (Fig. 14, row B). Cells exposed to solvent controls, 0.5% DMSO for cytochalasin D and 0.2% DMSO for latrunculin B, progressed to normal thrombin-induced activation and exhibited microtubule and F-actin distribution characteristic of the early pancake stage (Fig. 14C).
Cytochalasin D and latrunculin B inhibit thrombin-induced responses (epifluorescence): (row A) cells pre-incubated in 5 μg/ml cytochalasin D, 2 h; (row B) 1 μM latrunculin B, 1 h; (row C) 0.2% DMSO, 1 h (control); (column 1) phase-contrast, cells prior to attempted activation; (column 2) cells after 10 min of thrombin exposure; (column 3) epifluorescence, tubulin distribution in (column 2); (column 4) F-actin distribution in (column 2). Bar, 10 μm.
4.1 Sequence of events following nucleated thrombocyte activation
As observed in the present study, post-activation events in nucleated thrombocytes do parallel those of mammalian platelets with respect to characteristic morphological states and the corresponding time frame. Approximately2 min after activation, both cell types transform from a planar to a spheroidal state. Cells in this state extend and retract cytoplasmic projections, presumably securing attachment to the surface, or, potentially, to each other. After attachment, at 3–4 min, both develop radially spreading cytoplasm between the projections referred to as hyalomeres in the platelet sequence (Allen et al., 1979), culminating in a fully spread state at ∼10 min. Although the major morphological stages are comparable, one major difference is observed: nucleated thrombocytes develop dynamic blebs concurrent with becoming spheroids (Fig. 7), and these persist for about 30 s prior to extending projections. This has not been reported for mammalian platelets. Obviously, the presence of a nucleus and its subsequent lobulation and fragmentation are additional distinguishing features in comparison with platelets.
4.2 F-actin function prior to activation
We report here that most of the F-actin in the unactivated thrombocyte occurs in a cortical layer throughout the cell, and in a specialized ring of F-actin that is associated with the microtubular MB (Fig. 2; Fig. 3). Association of F-actin with MBs has also been observed in resting mammalian blood platelets (Debus et al., 1981; Takeuchi et al., 1990), and in non-mammalian vertebrate erythrocytes (Kim et al., 1987; Lee and Cohen, 1998; Sanchez and Cohen, 1994). In addition, F-actin has also been reported in platelets as extensions of MB-associated F-actin with a spoke-like arrangement (Takeuchi et al., 1990). Unlike these systems, we found that the F-actin band in unactivated thrombocytes is thinner than that of MB microtubules, as determined within the limits of confocal microscopy. Furthermore, the thrombocyte MB itself appears to be composed of inner and outer microtubule bundles, with only the latter co-localizing with the F-actin band. We suggest that interaction with F-actin may confer increased stability on outer MB microtubules involved in unactivated cell shape maintenance, and that such F-actin-MT interaction is lost when activation triggers rapid cell shape alterations.
In unactivated cells, the MB was enclosed within the cortical meshwork of F-actin (Fig. 4), an arrangement similar to that found in platelets (White, 1971). This supports MB function as a flexible frame that maintains the shape of unactivated nucleated thrombocytes, with the MB exerting pressure against the cortical F-actin layer. The arrangement of cytoskeletal elements in the unactivated thrombocyte is illustrated diagrammatically in Fig. 15. A similar model has been proposed for nucleated erythrocytes, in which the MB is enclosed within the membrane skeleton (Cohen et al., 1998; Joseph-Silverstein and Cohen, 1984). This in turn raises the possibility that MB pressure against the cortical layer of the thrombocyte causes it to be compressed in the plane of contact, accounting for the greater density of F-actin at the MB periphery, i.e. the ring.
Diagrammatic representation of the major cytoskeletal elements in the unactivated thrombocyte. The diagram is a composite of both longitudinal and cross-sectional views. Abbreviations: MB MTs, marginal band microtubules; PM, plasma membrane bilayer; MS, membrane skeleton. The yellow color denotes overlap of MB microtubules with the F-actin ring.
4.3 F-actin function following activation
The hypothesis that post-activation stages in nucleated thrombocytes are primarily due to F-actin reorganization is supported by the results. As early as the spheroidal stage, the F-actin band loses co-localizationwith the buckling MB (Fig. 6C), and cortical F-actin redistributes into the blebs (Fig. 7). The bent and twisted MB shape at the spheroidal stage raised the possibility that the MB might also be an effector of conversion to the spheroid. However, our tests using cytochalasin D and latrunculin B, both of which inhibit F-actin assembly (Spector et al., 1989; Urbanik and Ware, 1989), showed that unactivated thrombocytes failed to progress to spheroids when exposed to thrombin (Fig. 14). In addition, the myosin II ATPase inhibitor BDM blocked spheroid formation. Thus, spheroid formation is dependent on F-actin assembly and myosin II function, indicating that bending of MBs is a passive secondary effect. It has been shown that similarly twisted MBs, observed in erythrocytes of blood clams undergoing shape transformation from flattened ellipsoids to spheroids, are not the effectors of the transformation (Lema-Foley et al., 1999). Thus, shape transformation in the post-activation stages is dependent upon assembly of F-actin.
Cytochalasin D and latrunculin B are known to cause drastic shape changes in cells in which the F-actin is in a dynamic state (Spector et al., 1989). In contrast, we observed stability of unactivated cell morphology in the presence of cytochalasin D and latrunculin B, indicating that the unactivated thrombocyte contains stable F-actin that is not in dynamic equilibrium with G-actin. This is supported by the observation that a substantial amount of F-actin remains in the cortex after incubation with these inhibitors (Fig. 14). Exposure of cells to thrombin would normally convert the stable pre-activation F-actin to a dynamic state, with depolymerization of cortical F-actin maintaining the G-actin pool during reassembly of F-actin elsewhere in the cell. Maintenance of shape in thrombin-exposed cells preincubated in CD and LB fits a model in which the inhibitors block de novo synthesis of F-actin from the G actin pool, preventing pool depletion. The altered G–F-actin equilibrium, in turn, would inhibit further disassembly of cortical F-actin.
In platelets, previous studies have shown that cytochalasin D inhibits F-actin assembly upon thrombin activation (Fox and Phillips, 1981). While cytochalasin D-treated dogfish thrombocytes undergo only a very slight shape change (Fig. 14B2), cytochalasin D-treated platelets do undergo the complete initial cell shape change to spheroids, with a reduction in the number of pseudopods and attenuation of surface convolutions (Casella et al., 1981).
In advanced stages of activation, F-actin distributes outwardly in the form of stress fibers, frequently arranged in geometric patterns, while MTs localize principally in proximity to the nucleus (Fig. 12F; Fig. 13B). Platelet cytoskeletal structures resembling stress-fibers have also been reported in some studies (Morgensternet al., 2001; Tanaka and Itoh, 1998). Thus, in general respects, the spread nucleated thrombocyte cytoskeleton resembles that of activated platelets, in which F-actin spreads outwards and microtubules remain in the interior (Debus et al., 1981; Escolar et al., 1986).
4.4 Microtubule function
While F-actin displays a progressively outward distribution following activation, MB microtubules exhibit the opposite pattern. The MB initially accommodates itself passively into the spheroidal cell shape by buckling or bending (analogous to the seam in a baseball), in effect enclosing the nucleus (Fig. 7). Subsequently, the major bundle of MB microtubules remains intimately associated with the nucleus as such, and becomes increasingly disorganized as the cell spreads. In all observed cases of cells undergoing nuclear lobulation, convoluted bundles of microtubules were found between the lobes (Fig. 12D and Fig. 4H–I), with major bundles wrapped around the neck region between lobes. This explains the presence of microtubules near nuclear lobes in TEM thin sections reported by Shepro et al. (1966).
This intimate association of microtubules with nuclear constrictions suggests that they may provide the force required for nuclear lobulation and fragmentation in this system. However, the nature of this driving force and the mechanism of nuclear fragment migration away from the central region remain to be demonstrated. There is precedence for microtubules exerting a mechanical force to reshape the nucleus. Strongest evidence comes from studies of spermatogenesis in many species, in which manchette microtubules and associated dynein and kinesin motor proteins are believed to be the effectors (Hall et al., 1992; McIntosh and Porter, 1967; Yoshida et al., 1994).
4.5 Activation of nucleated thrombocytes by mammalian thrombin
Activation of mammalian platelets is commonly defined by events such as cell shape change, adhesion, aggregation, and granular exocytosis. Representative agents that induce platelet activation include ADP(Aledort, 1971), thrombin (Haslam, 1964), and the calcium ionophore A23187 (White et al., 1974). Glass surfaces, potentially mimicking damaged vascular endothelia, also induce changes of cell morphology (Baumgartner and Haudenschild, 1972). Our study shows that nucleated thrombocytes undergo morphological changes analogous to mammalian platelets upon contact with a glass surface or, to a much greater extent, exposure to bovine thrombin. Recently, the mechanism of such morphological change has been linked with cell surface receptor-mediated signal transduction. In mammalian platelets, PAR-1 and PAR-4 have been identified as cell surface thrombin-receptors involved in cell aggregation (Kahn et al., 1998). In catfish and zebrafish thrombocytes, an integrin-like complex has been reported as thrombocyte-specific and involved in signaling of aggregation and cell shape alteration (Jagadeeswaran et al., 1999; Passer et al., 1997). Thus, responsiveness of fish thrombocytes to mammalian thrombin indicates the existence of evolutionarily conserved signal transduction pathways.
4.6 Variation within the thrombocyte population
Observations presented here represent the most commonly occurring events. However, variations within the thrombocyte population were also noted. In preparations involving naturally induced activation, in which whole blood is spread on coverslips without anticoagulant, some thrombocytes were found to progress through post-activation stages more readily than others. For example, variations in early response are evident in Fig. 6A, B. It is conceivable that, during blood withdrawal, fast-progressing cells were those exposed to activating signals to a greater extent than slow progressing ones. However, even in thrombin-induced activation, minor percentages of cells did not respond, supporting the notion of population variability. Variations in nuclear lobulation were also seen. Most cells exhibited nuclear lobulation during transition to pancake stages. However, cells in full pancake forms were occasionally found with no apparent nuclear lobulation (e.g. Fig. 13). Such observations further support population variability and, possibly, the existence of thrombocyte subtypes.
4.7 Nucleated thrombocyte activation: a specialized apoptotic pathway?
Several of the major properties that we have observed in activated nucleated thrombocytes are characteristic of apoptotic cells. These include the initial rounding up into a compact form (apoptotic cell “condensation”), early surface blebbing, and nuclear lobulation and fragmentation. Apoptotic surface blebbing involves actomyosin-based cell motility (Mills et al., 1998), and the blebs that we observe are actin-rich (Fig. 7). In addition, in at least some cultured mammalian cells triggered to undergo apoptosis, interaction of microtubules with fragmenting nuclei has been observed(Pittman et al., 1997). Considering that activated thrombocytes are on a pathway leading ultimately to cell death, these similarities raise the possibility that nucleated thrombocyte activation and function may actually be a specialized form of apoptosis.
This work was supported by grants from the City University of New York (PSC-CUNY 63215), the Howard Hughes Medical Institute (HHMI Undergraduate Education Program 71100-534602), and the National Science Foundation (NSF 9808368). We thank Dr Phong Tran and Mara Conrad for helpful comments, and Edward Enos and Andrew Sexton for providing dogfish during the course of the work.
Allen RD, Zacharski, LR, Widirstky, ST, Rosenstein, R, Zaitlin, LM, Burgess, DR. Transformation and motility of human platelets: details of the shape change and release reaction observed by optical and electron microscopy. J Cell Biol 1979:83:126-42
Bearer EL, Prakash, JM, Li, Z. Actin dynamics in platelets. Int Rev Cytol 2002:217:137-82
Casella JF, Flanagan, MD, Lin, S. Cytochalasin D inhibits actin polymerization and induces depolymerization of actin filaments formed during platelet shape change. Nature 1981:293:302-5
Casella JF, Flanagan, MD, Lin, S. . Cohen WD, Sanchez I, Rayos N, Dadacay A. Utility of dogfish erythrocytes for studies of the cytoskeleton and cellular morphogenesis. Marine Models Electronic Record, Biol Bull Publications (online:http://www.mbl.edu/BiologicalBulletin/MMER/COH/CohTit.html) 1997.
Cohen WD, Sorokina, Y, Sanchez, I. Elliptical versus circular erythrocyte marginal bands: isolation, shape conversion, and mechanical properties. Cell Motil Cytoskeleton 1998:40:238-48
Daimon T, Uchida, K. Ultrastructural evidence of the existence of the surface connected canalicular system in the thrombocyte of the shark (Triakis scyllia). J Anat 1985:141:193-200
Debus E, Weber, K, Osborn, M. The cytoskeleton of blood platelets viewed by immunofluorescence microscopy. Eur J Cell Biol 1981:24:45-52
Escolar G, Krumwiede, M, White, JG. Organization of the actin cytoskeleton of resting and activated platelets in suspension. Am J Pathol 1986:123:86-94
Fawcett DW, Witebsky, F. Observations on the ultrastructure of nucleated erythrocytes and thrombocytes with particular reference to the structural basis of their discoidal shape. Z Zellsforsch Mikrosk Anat 1964:62:785-806
Hall ES, Eveleth, J, Jiang, C, Redenbach, DM, Boekelheide, K. Distribution of the microtubule-dependent motors cytoplasmic dynein and kinesin in rat testis. Biol Reprod 1992:46:817-28
Hill DJ, Griffiths, DH, Rowley, AF. Trout thrombocytes contain 12- but not 5-lipoxygenase activity. Biochim Biophys Acta 1999:1437:63-70
Kim S, Magendantz, M, Katz, W, Solomon, F. Development of a differentiated microtubule structure: formation of the chicken erythrocyte marginal band in vivo. J Cell Biol 1987:104:51-9
Lee K-G, Braun, A, Chaikhoutdinov, I, DeNobile, J, Conrad, M, Cohen, W. . Lee K-G, Cohen WD. Co-localization of F-actin and marginal band tubulin in the cytoskeleton of salamander erythrocytes. 38th ASCB annual meeting. Dec. 1998.
Lema-Foley C, Lee, KG, Parris, T, Koroleva, Z, Mohan, N, Noailles, P. Reversible alteration of morphology in an invertebrate erythrocyte: properties of the natural inducer and the cellular response. Biol Bull 1999:197:395-405
Lloyd-Evans P, Barrow, SE, Hill, DJ, Bowden, LA, Rainger, GE, Knight, J. Eicosanoid generation and effects on the aggregation of thrombocytes from the rainbow trout, Oncorhynchus mykiss. Biochim Biophys Acta 1994:1215:291-9
Mainwaring G, Rowley, AF. Separation of leucocytes in the dogfish (Scyliorhinus canicula) using density gradient centrifugation and differential adhesion to glass coverslips. Cell Tissue Res 1985:241:283-90
Morgenstern E, Daub, M, Dierichs, R. A new model for in vitro clot formation that considers the mode of the fibrin(ogen) contacts to platelets and the arrangement of the platelet cytoskeleton. Ann N Y Acad Sci 2001:936:449-55
Pica A, Lodato, A, Grimaldi, MC, Della Corte, F. Morphology, origin and functions of the thrombocytes of Elasmobranchs. Arch Ital Anat Embriol 1990:95:187-207
Pittman S, Geyp, M, Fraser, M, Ellem, K, Peaston, A, Ireland, C. Multiple centrosomal microtubule organising centres and increased microtubule stability are early features of VP-16-induced apoptosis in CCRF-CEM cells. Leuk Res 1997:21:491-9
Schwer HD, Lecine, P, Tiwari, S, Italiano, JE, Hartwig, JH, Shivdasani, RA. A lineage-restricted and divergent beta-tubulin isoform is essential for platelet biogenesis, structure and function. Curr Biol 2001:11:579-86
Shepro D, Belamarich, FA, Branson, R. The fine structure of the thrombocyte in the dogfish (Mustelus canis) with special reference to microtubule orientation. Anat Rec 1966:156:203-14
Spector I, Shochet, NR, Blasberger, D, Kashman, Y. Latrunculins-novel marine macrolides that disrupt microfilament organization and affect cell growth: I. Comparison with cytochalasin D. Cell Motil Cytoskeleton 1989:13:127-44
Wang X, Herman, CA. Eicosanoid synthesis by purified thrombocytes and erythrocytes from warm- and cold-acclimated American bullfrogs (Rana catesbeiana). Gen Comp Endocrinol 1997:106:138-44
White JG, Clawson, CC. The surface-connected canalicular system of blood platelets—a fenestrated membrane system. Am J Pathol 1980:101:353-64
White JG, Rao, GHR, Gerrard, JM. Effects of the ionophore A23187 on blood platelets. I. Influence on aggregation and secretion. Am J Pathol 1974:77:135
Yoshida T, Ioshii, SO, Imanaka-Yoshida, K, Izutsu, K. Association of cytoplasmic dynein with manchette microtubules and spermatid nuclear envelope during spermiogenesis in rats. J Cell Sci 1994:107:625-33
Zucker MB, Nachmias, VT. Platelet activation. Arteriosclerosis 1985:5:2-18
Received 19 September 2003/8 January 2004; accepted 28 January 2004doi:10.1016/j.cellbi.2004.01.008