Introduction

Platelets are absolutely essential in case of vessel injury. They are not only necessary to initiate hemostasis but also to retract the forming clot thereby pulling together the edges of the ruptured vessel and flattening the clot for improved blood flow. Different approaches to visualize fibrin fibers in whole blood or plasma clots have shown that fibrin fibers represent the main mechanical and structural framework of clots,^1,2 occupying a substantial portion of the clot volume. Previous studies^3–6 have shown that platelets are able to retract an unconstrained plasma clot to a tiny volume. The extent and rate of clot retraction are influenced by several factors, including platelet and red blood cell counts, fibrin and thrombin concentrations, and the geometric context of clot formation.⁷ However, platelets are the primary cellular agents driving such extensive clot compaction, although erythrocytes also contribute.⁸ How do platelets accomplish this tremendous task? Two key studies offer insight: While Fletcher’s lab demonstrated that platelets generate contractile forces comparable to muscle cells,⁹ Weisel’s group provided mechanistic insight, showing that platelets extend filopodia to bind and pull on fibrin fibers, much like tugging on ropes¹⁰ causing localized fiber accumulation.

However, pulling alone may not fully account for the dramatic reduction in clot volume. We hypothesized that efficient clot retraction requires substantial densification of the fibrin fiber network. Using expansion microscopy, we observed that platelets wind-up fibrin fibers similar to balls of wool around individual bulbs. Using a “2D fiber-retraction assay” we found that fibrin patches are initially arranged in a ring-like pattern in close proximity to the circular organization of the actin cytoskeleton in spread platelets. Over time, wound-up fibrin fibers accumulate above the plasma membrane of platelets in a myosin dependent way.

To explore the mechanistical aspects of the platelet mediated fiber compaction we used computational modeling and live-cell video microscopy. Our results suggest that actomyosin-driven swirling within spread platelets on a 2D surface or within individual bulbs of platelets in the 3D clot space may underlie the observed fibrin coiling and compaction.

Results

Platelets become labeled by fluorescent fibrin fibers during clot retraction

We first estimated the capacity of platelets to retract diluted plasma clots (12% plasma) under unconstrained conditions (Fig. 1A). Under these conditions, 4×10⁷ platelets are able to retract the clot volume to approximately 1.5% of its initial volume (400 μl) within 60 min. Even platelets kept overnight at room temperature before the retraction assay were still retracting the clot to the same extent, however after a longer lag phase. We then used live imaging to follow the reorganization of fibrin fibers during unconstrained clot retraction. Platelet rich plasma (PRP) was spiked with fibrinogen-Alexa 488 before thrombin induced clot formation and image acquisition. Very rapidly one can observe the formation of fluorescent fibrin nodes during the retraction process. Since under the assay conditions the clot is composed only of platelets and fibrin fibers, we hypothesized that the nodes may correspond to platelets surrounded by fibrin fibers (Fig. 1B and video 1). To test this hypothesis, we performed a time course analysis of clot retraction in presence of fluorescent fibrinogen under constrained conditions preventing complete retraction. This approach allowed us to stain and visualize individual platelets in a clot even after prolonged retraction times. Hemostatic clots that form in vivo following vessel injury are also likely to experience mechanical constraints, as they remain anchored to the damaged vessel walls while spanning the breach. Retracted clots were fixed and stained for the αIIb-integrin subunit to detect the localization of platelets within the clot. At the position of each platelet, we consistently observe colocalization of a fibrin node at all time points examined (Fig. 1C), confirming our hypothesis that the fibrin nodes observed in figure 1B form at platelet sites. Furthermore, the results show that the build-up of fibrin fibers around individual platelets occurs very rapidly, within 10 min. The fibrin fibers forming the nodes did not decorate the entire platelet surface but surrounded the center of platelets embedded in the clot. Please note, that platelets occupy a larger space after longer retraction times, but the fiber nodes persist even after 80 min of retraction.

Figure 1:

To gain a better insight about the organization of fibrin fibers around platelets within a clot, we used 4x isotropic expansion microscopy. We found that fibrin fibers form a cage-like structure around the platelet center, clearly visible when scrolling through the image stacks (Fig. 2 A-D and animation, Video 2). This pattern was consistently observed across all platelets in the clot and after different clot retraction times (not shown). Most platelets had formed large bulbs, thin filopodia, or both with fibrin fibers typically localized around the base of the bulbs or along filopodia. It remains unclear whether all fibrin segments reside on the external platelet membrane or whether some of them extend into channels of the open canalicular system (OCS). Some of these channels persist in activated platelets.¹¹ Please note that straight fibers radiate from the platelet, indicating a high tension in the platelet clot environment. We want to emphasize that changes in temperature, or plasma/platelet concentrations change the kinetics of retraction but we always observe the cage-like fibrin fiber organization around individual platelets (not shown).

Figure 2:

Platelets, spread on a 2D surface, organize fibrin fibers above them

So far, our results demonstrate that platelets possess a remarkable capacity to precisely organize fibrin fibers around themselves when embedded in a clot. To investigate the mechanism by which platelets arrange the fibers around their center we sought to use a model system with reduced complexity. We, therefore, developed a “2D fiber-retraction assay”, which is a reductionist approach that does not replicate full clot retraction, but can provide insight into how platelets interact with and organize fibrin fibers. To this end, a diluted suspension of PRP (2.5×10⁶ platelets/ml, 4μl plasma/ml PBS) was incubated with fluorescent fibrinogen and thrombin for 15 min to induce platelet activation and fibrin polymerization. Platelets were then allowed to spread on a 2D surface for 30 min before fixation. Under these conditions, we observe four categories of spread platelets and associated fibrin fiber organizations. Platelets without fibrin fiber contact (28.3+2%) and platelets with a small fibrin dot at their center, which could be a fibrin initiation complex (40+3%). Another category of platelets had an accumulation of coiled fibers above them (fiber winding 18.3+0.5%) and a fourth category showed strong fiber densification around them (fiber compaction 13.6+0.9%) (Fig. 3, left image of A, quantification B). We then investigated, whether these fiber accumulations depend on myosin actions as it is the case for clot retraction. Myosin inhibition using blebbistatin strongly diminished the build-up of fibers around the spread platelets (fiber-winding 6.6+0.9% and fiber compaction 0.2+0.2%), while a higher percentage of platelets with fiber initiations or fiber contacts was observed (63.4+2%) (Fig. 3, right image of A and quantification B).

Figure 3:

Then, we used the 2D fiber-retraction assay followed by expansion microscopy to investigate in more detail how the fibers above spread platelets are organized. Platelets were stained either for the αIIb integrin subunit or for myosin. We observed that fibrin fibers are wound-up above platelets (category “fiber winding” in fig. 3), frequently forming ball-like structures wrapped around a bulbous protrusion in the middle of the spread platelet, often referred to as pseudo-nucleus in the literature (Fig. 4A-E and animation, Video 3).

Figure 4:

In addition to these spread platelets with coiled fibrin fibers above them, we also observed platelets entangled by many fibers and therefore unable to spread completely. These platelets had formed filopodia and bulbs similar to platelets within a clot and a dense fiber accumulation is observed around the base of their bulbs (category “fiber compaction” in fig. 3A). These fibers appeared to get coiled into multiple loops, tightly compacted around the base of platelet bulbs similar to individual balls of wool (six small “balls” can be distinguished in Fig. 5A, focal planes at 4 and 5.25 μm and animation, Video 4, see also Fig. 6 and animation, Video 5). Winding-up of fibers by platelets implies that fibers around individual platelets may get twisted and thereby bundled. Local twisting of fibers near platelets is indeed observed in most of the presented images, although the resolution of optical microscopy is often insufficient to resolve regions of densely packed fibers. Particularly clear examples are shown in Figure 6. It should be noted that fiber bundles between two platelets can also be twisted (Fig. 6C-E). The enhanced fibrin labeling observed around these platelets, compared to those in a typical constrained clot (Fig. 2A-D), may be due to multiple winding-up of free, unconstrained fibrin fibers.

Figure 5:

Figure 6:

A ring-like organization of fibrin fiber patches can be observed at the center of spread platelets

We next increased the plasma volume (7μl/ml instead of 4μl/ml) in order to enhance the amount of formed fibrin fibers. This modification led to the formation of a fragile fibrin gel, which was partially lost during the fixation or staining steps. As a result, only very few fibers remained attached to the platelets or to the coverslip.

The residual fibrin fluorescence revealed a ring-like arrangement of fibrin patches at the platelet center. This fibrin “rosette” co-localized with a circular actin organization, forming a gearwheel-like pattern with a compacted fibrin mass at the inner edge of the circle. The actin cytoskeleton was either composed of radial and transverse fibers (Fig. 7 A,B and animation, Video 6) or radial fibers and actin nodules (Fig. 7C and animation, Video 6) depending on the degree of platelet spreading.^12,13 A faint rosette with a compact fibrin mass and an attached fibrin fiber is shown in figure 7D and animation, Video 6).

Figure 7:

We then investigated the localization of myosin II under these conditions and found that it is distributed throughout the platelet. However, a denser ring-like organization of myosin is present in the center of the spread platelet, closely associated with the fibrin rosette (Fig. 8 and animation, Video 7). Some myosin spots are intercalated with compacted fibrin (Fig.8 A, focal plane at z=7.8 µm), while wound-up fibrin fibers can be observed at the periphery and above the myosin ring structure (Fig. 8A-F). The intensity of fibrin rosettes is highly variable between platelets of different donors. Intense rosettes as shown in figure 7 were observed for platelets from two donors (approximately 3-5% of spread platelets with fiber initiation complexes), while faint rosettes (as in Fig. 8AB) were observed for 50% of spread platelets with initiation complexes using blood from three other donors.

Figure 8:

Modeling platelet-mediated winding-up of fibrin fibers

To address the mechanical aspects of the observed platelet-mediated fibrin fiber organizations, we developed a model based on our experimental observations. We first quantified fiber curvature and the number of platelet extensions. Each platelet (in clots or on a glass surface surrounded by fibers and therefore unable to spread) showed 7-10 bulbs/filopodia (Fig. 9A-C). Fiber curvature was determined using transmission electron microscopy (TEM) and fluorescent (FLUO) images. Curvatures were similar for both image types (mean values for κ_TEM=10.09 ± 6.6 μm^-1 and κ_FLUO=6.77 ± 4.7μm¹) with differences likely due to the higher resolution of TEM (Fig. 9D-F).

Figure 9:

Our model incorporates these observations, published parameters (Table 1, methods section) and builds on the framework established by Bershadsky and colleagues.¹⁴ They demonstrated that in fibroblasts, cultured on circular substrates, the actin cytoskeleton is composed of radial actin fibers extending towards the cell center and circular transverse fibers enriched in myosin-IIA. Live-cell imaging in their study revealed that the transverse fibers slide centripetally along the radial fibers, generating a chiral swirling motion directed towards the cell center. Fibronectin-coated beads bound to integrins on the cell surface are dragged along with the intracellular swirling of the cytoskeleton suggesting that extracellularly bound fibrin fibers may experience a similar rotational motion. Chiral rotations of the cytoskeleton are not only observed at the level of entire cells, but also in subcellular regions, such as individual filopodia of the neuronal growth cone.^15,16

Table 1.

Along this line, our simulations suggest that fibrin fibers are anchored to integrins in the plasma membrane of bulbs and filopodia, are pulled towards the platelet center by retraction of filopodia, as demonstrated by Kim et al.¹⁰ Spiral cytoskeletal motions within each bulb drive integrins, and thus the attached fibrin fibers, towards the platelet center, wrapping the fibers around the base of the bulbs (Fig. 10A and animation, Video 8). The efficiency of loop formation depends on the initial fiber position (Fig. 10B-D). The process effectively compacts long fibers into loops with a radius of ∼420-nm near the base of bulbs (Fig. 10D).

Figure 10:

Transmission electron microscopy (TEM) provides evidence of curved and cross-sectioned fibrin fibers at the base of platelet bulbs

To confront the results obtained by expansion microscopy and model simulations with those using TEM, we analyzed TEM images of platelets and fibrin fibers within a clot. We observed numerous instances of cross-sectioned fibrin fibers at the base of bulbs as illustrated in Figure 11. This figure also shows two examples suggesting fibers that curve along bulbs from top to bottom and fibers attached to a filopodium.

Figure 11:

Live imaging of platelet-mediated reorganization and compaction of fibrin fibers

We performed live video microscopy using platelets from three different donors to observe how platelets wind-up fibrin fibers (Fig. 12 and Video 9). In many cases, fibrin fibers were already accumulated above platelets at the start of imaging, with minimal further fiber movements or compactions; likely due to fibers binding to each other, to the dish or to the platelet surface (Fig. 12A). However, dynamic fiber movements, both clockwise and counterclockwise, were observed above two platelets present within the rectangle indicated in figure 12A (see zoom images lower panel). An arrow in Fig. 12A marks the formation of a kink in a fibrin fiber. For the second experiment (blood from a different donor), we reduced the incubation temperature to 21°C in order to slow down the platelet and fiber movements. In the rectangle indicated in figure 12B, a fibrin fiber gets progressively curved and eventually ruptures (at time point 00:48). The resulting free end of the fiber then undergoes a clockwise rotation (see zoom images lower panel). Another platelet producing a fiber densification is marked by an arrow in figure 12B. The video shows a counterclockwise rotation of the central part of the platelet during the fiber densification process. A third experiment (platelets from another donor), again performed at room temperature, shows the rotational (counterclockwise) movement of the pseudo-nucleus and alongside the winding-up of fibers around this bulbous protrusion. This apparently generates strong tension that pulls the pseudo-nucleus towards the edge of the spread platelet, where the movement ceases (Fig. 12C, see zoom images lower panel). These fibrin fiber movements occurred without visible filopodia-fiber interactions, suggesting that the bound fibers are dragged by intracellular movements of the cytoskeleton. Since still images cannot fully illustrate the fiber movements or the rotation of the platelet center, we invite the reader to consult the accompanying time-lapse videos (Video 9). After the time-lapse acquisitions, samples were fixed and stained for the αIIb-integrin subunit to localize the position of platelets on the dish surface. The final fibrin fiber organization around platelets in these live videos of the 2D fiber-retraction assays closely resembles the arrangement of fibers around platelets in a constrained clot, except that some platelets are completely spread on the surface with no or very few fiber contacts. Platelets surrounded by densely packed fibers had formed bulbs and filopodia, but remained attached to the dish surface (Fig. 12D).

Figure 12:

To minimize pre-imaging fiber buildup, we next added preformed fibrin fibers to pre-spread platelets and started imaging immediately. Under these conditions, surface coating is observed and only a few fibers are present, yet a central fibrin accumulation is consistently observed on each platelet, possibly representing a nucleation site. Among 35 platelets analyzed, 15 displayed a central fibrin mass rotating counterclockwise (viewed from below), 4 rotating clockwise, and 16 showed no consistent directional movement (Fig. 12E). In some instances, the rotating complexes stalled and stopped moving. These observations support the idea that cytoskeletal swirling drives fibrin reorganization above spread platelets.

Discussion

Using a 2D fiber-retraction assay, we identified a previously unrecognized capacity of platelets to compact fibrin fibers into a small volume by winding them into coiled structures during platelet induced fiber retractions. To our knowledge the only example of natural fiber compaction is coiling of DNA molecules.¹⁷ In contrast to intracellular, enzyme-mediated DNA compaction, extracellular fibrin fiber compaction requires the cellular activity of platelets. Our investigation began with the observation that platelets become labeled with fluorescent fibrinogen during clot retraction as previously shown by Brzoska et al.¹⁸ They demonstrated that fibrin binding to thrombin activated platelets depends on αIIbβ3 integrins and the actin cytoskeleton. While fibrin fiber retraction by platelets in a clot has been attributed to non-muscle myosin II,^19,20 the mechanism underlying fiber organization and compaction has remained unclear.

One key mechanism, revealed in previous studies, is the extension of platelet filopodia that bind to fibrin fibers and pull them inwards in a hand-over-hand motion.¹⁰ This is clearly the main process by which platelets retract fibrin fibers in a clot. However, this mechanism alone cannot explain the cage-like fiber organization observed around platelets in a constrained clot (Fig. 2). Our findings introduce an additional, distinct mechanism by which platelets can organize fibrin fibers. Platelets wind-up fibrin fibers into dense, ball-like structures, which is an efficient form of compaction (Fig. 3-6). Thus, our model reveals an additional mechanistic aspect of platelet mediated fiber retraction beyond those described by Michael et al.²¹ and Kliuchnikov et al.,²² so far observed only in the 2D fiber-retraction assay. Nevertheless, the platelet morphology and the organization of fibers around them after the retraction process closely resemble those of platelets in a clot and both, 2D and 3D fiber retractions, depend on myosin actions.

The 2D fiber-retraction assay may recapitulate early post-injury events, such as activation of platelets in suspension triggered by agonists released from neighboring platelets, ultimately leading to the spreading of activated platelets onto the subendothelial matrix to anchor the developing clot at the site of vessel injury. This assay revealed two key observations. First, when many fibers are present, some platelets spread with the typical fried-egg phenotype and coil-up fibers above their pseudo-nucleus (Fig. 3, 4). Others form large bulbs in several directions, similar to platelets in a clot, and compacted fibers are observed around the base of each bulb (Fig. 5, 6). Under these unconstrained 2D conditions, fiber accumulation around the bulbs is much higher than in the case of a constrained 3D clot (Fig. 2), where internal tensions limit fiber winding.

Second, when fewer fibers are retained on the culture surface, several spread platelets display a fibrin rosette around their pseudo-nucleus (Fig. 7,8) that colocalizes with actin, myosin and integrins. A dense fibrin mass is seen at the inner periphery of this rosette, sometimes with an attached fiber. This gearwheel-like architecture may represent an early priming/nucleation step of platelet-induced winding of fibrin fibers. The actin cytoskeleton in platelets exhibits a chiral organization²³ which, combined with the action of myosin, could be responsible for the swirling motion observed in spread platelets (Fig.12 E-G and Video 9). Integrin-bound extracellular fibers are pulled along by this vortex inducing their winding-up above the pseudo-nucleus (Fig. 12C). Fibrin rosettes may fade as coiled fibers become more prominent, suggesting a transition from an initiation phase to active coiling and compaction.

Ideas and Speculations

How this 2D “gearwheel” translates to the 3D clot environment remains unclear. One possibility is that the pseudo-nucleus may correspond to a bulb around which fibers are coiled and accumulate. Platelets with bulbs within a clot have been described previously.^24–27

We wondered whether platelet bulbs might represent plasma membrane blebs similar to membrane protrusions formed by megakaryocytes during clot retraction.²⁸ However, the consistent presence of 8–9 bulbs per platelet (Fig. 9A-C) suggests that they reflect a “three-dimensional spreading” morphology, reminiscent of the eight corners of a cube, allowing platelets to sense the clot’s 3D environment along the main directions. Unlike blebs, these bulbs often contain organelles such as mitochondria (Fig. 9D, upper image and Fig. 11B) implying that each bulb may have its own energy supply and actomyosin organization. This could enable swirling within each bulb and explain the observed fibrin fiber accumulation at their base. Thus, at the onset of clot retraction, each emerging bulb may act as a discrete functional unit onto which the gearwheel assembles, enabling fibrin fibers to attach and initiating their subsequent winding and compaction (see scheme, Fig. 13). Indeed, blebs of megakaryocytes retracting a clot may behave similarly as platelet bulbs since fibrin fibers can be seen surrounding the base of blebs during megakaryocyte-mediated clot retraction, as can be observed in the study by Kim et al. (see the supplementary movie S3 of this publication).²⁸

Figure 13:

Our live time-lapse videos of platelet-mediated fiber movements in a 2D environment support the concept of a cytoskeletal vortex that drives the coiling and compaction of extracellular attached fibers. Additionally, the other time-lapse video capturing the rotation of compacted fibrin at the platelet center further suggests the presence of cytoskeletal swirling within spread platelets. This swirling may play a role in scanning the extracellular space for forming fibers that can be anchored to the initiation complex. The process occurs on a notably slow time scale, which could facilitate effective fiber capture by the rotating complex. However, this rate appears insufficient to account for the rapid labeling of platelets during clot formation, suggesting that regulatory mechanisms, possibly involving mechanical feedback, may be at play. Indeed, as mechanosensory cellular particles, platelets adjust their contractile forces and dynamics in response to environmental cues such as fiber tension and clot stiffness.^{9,10,29,30,22}

In conclusion, our study identifies a novel mechanistic aspect of platelet function: the active winding-up and compaction of fibrin fibers. This complements the existing model of platelet-mediated clot retraction¹⁰ and provides a new perspective on how platelets may contribute to clot architecture, mechanical stability, and wound healing.

Materials and methods

Reagents and antibodies

Reagents: Neutral formalin (Sigma-Aldrich, Saint-Quentin Fallavier, France), Pluronic F-127 (Sigma-Aldrich, P2443), thrombin (Sigma-Aldrich, T4648), Jasp-K4-HA (a HAK-actin probe, kind gift of Paul Guichard/Virginie Hamel),³¹ apyrase (Sigma-Aldrich, A6535), heparin (Sigma-Aldrich, H3393), PGI2 (Sigma-Aldrich, P6188)

Antibodies: mouse anti-integrin αIIb (Santa-Cruz, sc-7310), monoclonal rabbit anti-HA (Cell Signaling, C29F4), rabbit anti-non muscle myosin II heavy chain (Covance, PRB-440P), rabbit anti-phospho S19 myosin light chain (Abcam, ab2480), monoclonal rabbit anti-myosin light chain 2 (Abcam, ab92721), Cy3 goat anti-mouse IgG (Jackson, 115-165-166), Cy3 goat anti-rabbit IgG (Jackson, 111-165-003), Alexa 647 goat anti-rabbit IgG (Jackson, 111-605-003)

Preparation of human platelet rich plasma (PRP) and platelet free plasma (PFP)

Whole blood was obtained from the French blood bank (age-range 18-65, mixed genders; pooled blood from multiple donors has not been used). The IRB has authorised our institute’s research projects using human samples under the number DC-2022-4976. PRP was prepared from whole blood drawn into Na-citrate Vacutainers (3 tubes of 2.7 ml per donor). The blood is pooled into a 15 ml falcon tube and centrifuged for 12 min 400g, RT, no brake. The upper phase is the platelet rich plasma (PRP), in which platelets are distributed in a gradient (low concentration on top, high concentration at the bottom). In order to increase the platelet concentration in the final PRP, the upper 3 ml are removed and used to prepare platelet free plasma (PFP, by centrifugation at 12000g, 1 min, RT). The remaining upper phase is collected as the final PRP and the platelet concentration determined.

Platelet washing procedure

PRP was diluted 1/4 with Tyrode/0.35% BSA buffer and platelet activation inhibitors (0.02 U/ml apyrase, 1 μm PGI2, 10 U/ml Heparin, final concentrations) were added. Samples were then centrifuged for 15 min at 900g at RT. The supernatant was aspirated, and platelets were gently resuspended in fresh Tyrode/BSA buffer, 0.02 U/ml apyrase, 1 μm PGI2. The suspension was centrifuged as before and the platelet pellet was resuspended in fresh Tyrode/BSA buffer without inhibitors.

Clot retraction assays

We used three different clot retraction assays illustrated in Supplementary Figure 1.

1. Unconstrained clot retraction (Fig. 1AB): See paragraph “Live video microscopy” below for method details.

2. Constrained clot retraction between two holders (Fig. 1C): In order to perform immunofluorescence stainings of individual platelets within a clot, we used a clot retraction assay around two sterile inoculation loops as previously described.^32,33 Briefly, PRP was adjusted to 1×10⁸ platelets per ml in 50% plasma/50% PBS (to reduce fibrin fiber density), fibrinogen-Alexa 488 final concentration 12.5 μg/ml and 4 μl/ml erythrocytes for color contrast. 400 μl of this suspension was used for clot induction by addition of thrombin (2.5U/ml final) and immediately transferred to 2 ml Eppendorf tubes coated with 2% Pluronic® and containing the inoculation loops. After different retraction times, clots were fixed for 1 h with isotonic formalin (9 volumes formalin / 1 volume 10xPBS). Clots were then washed with PBS and incubated ON at 4°C in 1 ml PBS/15% sucrose. Clots were further incubated in 1 ml PBS/15% sucrose/7.5 % gelatin at 37°C for 4 h and then inclusion blocks were formed in the same solution on ice. Blocks were then trimmed and snap frozen for 1 min in isopentan at −60°C and stored at −80°C until 14 μm thick transversal sections were cut using a cryostat.

3. Constrained clot retraction within an inoculation loop: Small clots of 30 μl platelet suspension (50% plasma/50% PBS; 1×10⁷ platelets/ml; fibrinogen-Alexa 488 final concentration 12.5 μg/ml; 4 μl/ml erythrocytes for color contrast) were induced to form within plastic inoculation loops (4 mm internal diameter, attached to a glass coverslip) by addition of thrombin to a final concentration of 2.5 U/ml as described previously.¹³ Five minutes after clot induction, clots were covered with 500 μl PBS and kept in a cell culture incubator for 10 min. Clots were then fixed with 500 μl isotonic formalin for 15 min, RT, in the dark and then kept in PBS at 4°C until further use. Forming a clot within a small inoculation loop, allowed us to do expansion microscopy without performing cryosections of larger clots, which are difficult to detach from slides during the expansion process.

2D fiber retraction assay

PRP was diluted in PBS (2.5×10⁶ platelets/ml, plasma concentrations were kept at 4 or 7μl/ml depending on the amount of fibrin fibers to be formed; see explanation below) and fibrinogen-Alexa 488 was added to a final concentration of 12.5 μg/ml. To induce fibrin fiber polymerization and platelet activation, thrombin was added (final concentration of 2.5U/ml). The suspension was gently mixed once (to prevent platelet aggregation) and after 15 min of incubation at RT in the dark, 400 μl was transferred into wells of a 24 well plate containing glass coverslips. The plate was centrifuged 3 min, 600 g, RT to allow synchronised contact of all platelets with the glass surface. The plate was then placed in a cell culture incubator for 30 min to allow platelet spreading. Samples were then fixed with isotonic formalin for 15 min at RT and then kept in PBS at 4°C until further use.

To promote fibrin fiber formation, while preventing the development of a clot, the plasma concentration must be kept low (4 μl plasma per ml PBS). This can be achieved by adding a volume of PRP containing 2.5×10⁶ platelets to 1 ml PBS, provided the platelet concentration of PRP is at least 6.25×10⁸/ml (condition used for Fig. 3, 4, 5, 6, 8ABEF). If the platelet concentration in the PRP is lower, platelets must be washed and the appropriate amount of plasma added separately.

At higher plasma concentrations (7 μl plasma per ml PBS), more fibers polymerize leading to the formation of a fragile clot which is aspirated during fixation/washing procedures. Under these conditions very few fibers are left behind. However, a fibrin rosette can be observed in the centre of several spread platelets (condition used for Fig. 7, 8CD).

The 2D fiber-retraction assay followed by immunofluorescence staining and expansion has been repeated 10x with blood from different donors. A total of 18 image stacks stained for fibrin/integrin, 20 stained for fibrin/integrin/actin and 62 stained for fibrin/myosin have been acquired.

Immunofluorescence

Fixed platelets were permeabilized with PBS/0.2% Triton X-100 for 15 min at RT and then incubated with blocking buffer (3% BSA and 10% goat serum in PBS) for 1h at RT. Coverslips were then incubated for 2 h with a primary antibody diluted in blocking buffer, then washed three times and incubated with a secondary antibody diluted in blocking buffer for 2h at RT. After three washing steps, coverslips were either mounted in Mowiol or processed for expansion.

Expansion microscopy

Expansion microscopy was essentially performed according to Gambarotto et al.³⁴ with minor modifications as described previously³³. Briefly, fixed, immunostained platelets or clots on coverslips (12 mm diameter) were incubated in PBS/0.7% formaldehyde/1% acrylamide over night at room temperature. The coverslip was then placed upside down onto an ice cold 35 μl drop of gelation solution (PBS/19% Na-acrylate/10% acrylamide/0.1% bisacrylamide) pipetted onto parafilm on ice immediately after addition of TEMED and ammonium persulfate both to a final concentration of 0.5%. After 5 min of incubation on ice, the samples were transferred to 37°C for 1 h. The coverslip plus gel facing up was then incubated in denaturation buffer (5.7% SDS/0.2M NaCI/50mM Tris, pH9) for 15 min at RT with gentle agitation to detach the gel from the coverslip. The gel was then transferred into an Eppendorf tube with fresh denaturation buffer and boiled for 30 min at 95°C. To expand the gel, it was placed into a large volume of H₂0 and the water was changed twice before incubation ON at RT in fresh H₂0 leading to 4x isometric expansion.

Image acquisition

For image acquisition of some of the expanded samples (Fig. 2, 5CD, 6, 7, 8), a confocal microscope ConfoBright (Nikon A1R+MP) equipped with a home-made adaptive optics module was used with a 40x/1.15 long distance water immersion objective. The geometric optical aberrations were corrected both in excitation and detection light paths in the open loop mode with a deformable mirror (AlpAO DM97-15) inserted between the confocal scanning head and the microscope inverted stand (Nikon Ti2E). Local aberrations were sensed using the metrics of molecular brightness, derived from fluorescence fluctuations, and were iteratively corrected as individual Zernike modes. This configuration allowed to compensate for the refractive index mismatch, the specimen tilt and the eventual light scattering in depth of expanded samples and to bring back the confocal resolution to its diffraction limit in the vicinity of the imaged platelet.

Image acquisition of other expanded samples (Fig. 4, 5AB) was performed with a confocal microscope Dynascope (LSM 710 ConfoCor 3; Zeiss), 63×/1.4 NA Plan Apochromat objective and the acquisition software Zen 2010. The same microscope and software were used for live imaging (Fig. 12) using the Apochromat 40×/1.2 NA water-immersion objective. Axial Z-stacks (step 0.45 µm) were imaged in a time series using the avalanche photodiode detector in photon counting mode at low 488 nm excitation power (0.1% transmission) to limit photobleaching and phototoxicity, the 633 nm laser was set to 0.6% in order to obtain the transmission light images of the platelets. The pinhole was closed to 0.25 Airy units (14 µm) to increase the spatial resolution. The acquisition of transmitted light was concomitant to the fluorescence.

The FIJI software was used for image processing and analysis.³⁵ No treatment that modifies the raw images was applied. FIJI software was used for the Maximum Intensity Projection (MIP) images, 3D reconstructions of image stacks and for the depth-color coded projection image using the rainbow LUT (Fig. 4D) as well as for the depth-color coded images fire LUT (Fig. 6) and the temporal-color coded image spectral LUT (Fig.12E).

Transmission electron microscopy (TEM)

Clots were formed around two holders as described previously.³² Briefly, PRP was adjusted with PBS to 10 % plasma, 1×10⁸ platelets/ml and 4ml erythrocytes per ml were added for color contrast. Clot formation was induced by addition of thrombin (final concentration of 2.5 U/ml). After 1h of retraction at 37°C, clots were fixed using 1,5 ml of 2,5 % glutaraldehyde for 1h at RT. After fixation, a contrast-enhancing step was performed by incubating the samples in a solution containing 1.5% potassium ferrocyanide and 1% osmium tetroxide in 0.1 M sodium cacodylate buffer. Following this, the samples were embedded in EPON resin according to the protocol described by Eckly et al.³⁶ Thin sections (100 nm) were cut and stained with uranyl acetate and lead citrate and examined using a Jeol 1200 Plus transmission electron microscope operating at 120 kV.

Live video microscopy

- Unconstrained clot retraction (Fig. 1B):

Thrombin (final concentration of 2.5 U/ml) was added to 0.4 ml of a platelet suspension (50% plasma/50% PBS; 1×10⁸ platelets/ml; fibrinogen-Alexa 488 final concentration 12.5 μg/ml; 4 μl/ml erythrocytes for color contrast) to induce clot formation. After quick and gentle mixing, the suspension was transferred immediately into a well (8-well Lab-Tek, coated 2% Pluronic® to prevent clot attachment to well walls). Acquisition was started immediately at 100 μm above the bottom of the well using the confocal microscope ConfoBright (for microscope details see paragraph image acquisition).

- 2D fiber-retraction assay (Fig. 12A-D):

Platelets (washed Fig. 12AB or not Fig. 12C) were adjusted to 2.5×10⁶ platelets/ml PBS and 2.5 μl plasma was added and fibrinogen-Alexa488 (final concentration of 12.5 μg/ml). 15 minutes after thrombin induced fibrin fiber formation and platelet activation, 1800 μl of the suspension was transferred into a petri dish (WPI Fluoro Dish, FD35–100). The dish was centrifuged 3 min, 600 g, RT and installed in the microscope incubator at 37°C (Fig. 12A) or room temperature (Fig. 12BCD).

- 2D preformed fiber-retraction assay (Fig. 12E):

Preformed fluorescent fibrin fibers were prepared in the absence of platelets (2.5 μl plasma/ml PBS; fibrinogen-Alexa 488, final concentration 12.5 μg/ml; thrombin, final concentration 2.5U/ml; incubation 15 min RT in the dark). PRP was adjusted to 2.5×10⁶ platelets/ml with PBS and 1800 μl was transferred into a petri dish (WPI Fluoro Dish, FD35–100). The dish was centrifuged 3 min, 600 g at RT to allow spreading of the platelets. After centrifugation the supernatant was replaced by the preformed fibrin fiber suspension and the dish was installed in the microscope incubator at 37°C.

For all time-lapse videos, image acquisition was started immediately after adjusting the focal position and image parameters (<1 min) using a confocal laser-scanning microscope (ConfoBright or Dynascope, for microscope details see paragraph image acquisition).

The model of fibrin winding-up

The platelet bulb was modelled as a straight cylinder (the radius R = 400 nm, the length = 1500 nm). We assumed that the internal cytoskeleton of the bulb underwent the swirling motion with the angular velocity Rate_swirl as it was shown in¹⁴ and simultaneously moved down to the bottom of the bulb with the velocity Rate. Integrins αIIbβ3 on the bulb surface were associated with internal cytoskeletal proteins,³⁷ thus, they underwent the swirling and downstream motion with the same rates.

The fibrin fiber was modelled as it was done previously^21,22 with several simplifications. The fiber composed of N equally spaced nodes connected by elastic springs (see Supplementary Fig. 2A). The spring equilibrium length was L and it was equal to the initial distance between nodes. The fiber was located in the XY-plane and the angle α between fiber and x-axis was variable (Table 1). The motion of the node i in solution was described by the second Newton’s law of motion. The system was assumed to be in an overdamped regime (mä = 0):

where r_i was the position vector of the node i in three-dimensional coordinates (x,y,z); η was the drag coefficient for the fiber node in the blood serum; F_i,i−1 = K_r · (Δr_i,i−1 − L) and F_i,i+1 = K_r · (Δr_i,i+1 − L) were the vector forces which acted on the node i from the node i-1 or i+1 according to the Hook’s law due to the spring elongation or shortening (Δr_i,i−1 and Δr_i,i+1 was the distance between nodes i and i-1 or the nodes i and i+1; Kr was the fiber spring constant); was the Brownian force which acted on the node i due to thermal fluctuations. To solve the equations for the nodes in solution, the system of equations was discretized and the Heun’s method was used. According to this method, the evolution of coordinates in time for the differential equation

was calculated using a two-step algorithm:

where Δt was the time step used in calculation and discretization scheme; tn+1 = tn + Δt; xn was coordinate at the tn moment of time.

Thus, the evolution of the node i coordinates in time was calculated according to a two-step process (in a vector form):

The Brownian force in three dimensions was calculated as described previously.²¹ Briefly, the mean square displacement of a particle over a single time step was , and the iscretized Brownian force acting on the node i was , where εi was sampled from the standard normal distribution.

At the initial time point, the node 0 was bound to the integrin receptor on the bulb surface (top view, Supplementary Fig. 2B). The distance of the binding point from the bulb top was D. The height of the binding point from the bulb surface was equal to the integrin αIIbβ3 molecule length (H = 20 nm) in its elongated conformation.³⁸ The equations for the node 0 swirling and downstream motion were:

When the node i entered the light-magenta zone (Supplementary Fig. 2B), and its distance from the bulb surface became smaller than H, it bound the integrin molecule and its equations of motion became similar to the node 0 equations (7) – (10).

The integrin αIIbβ3 surface density on the bulb surface was relatively high (the distance between the neighbouring molecules was approximately 15 nm). Given such a high density, we assumed that when the node became close enough to the bulb, it immediately encountered the integrin molecule and bound it. Thus, when the distance between the node i and the bulb surface became smaller than H, node i bound to the integrin molecule and the node’s equations of motion became similar to the node 0 equations (7) – (10). The binding was assumed to be irreversible.

The model parameters were listed in the Table 1 below. The model equations were solved using Python 3.11.

Supplementary figures

Supplementary Figure 1:

Supplementary Figure 2:

Data availability

All raw data are stored on the server of our microscope platform and can be accessed upon request.

Acknowledgements

We are grateful to Jacques Mazzega of the microscopic platform and to Arnold Fertin for image analyses. We gratefully acknowledge the MicroCell facility (GIS IBiSA, ISdV, IAB) for its technical assistance and the microscopy system. The MicroCell facility is a member of the national infrastructure France-BioImaging (https://france-bioimaging.org/). We thank Rémy Sadoul and Alexander Bershadsky for critical reading of the manuscript. This work was supported by the Fondation Recherche Médicale (FRM, grant number DEI20151234416) and the University Grenoble Alpes (UGA, grant number AGIR-POLE FRAG15CS08).

Additional information

Authorship contributions

A.G. developed the microscope with adaptive optics corrections and acquired images T.K. and M.P. established the model and simulations. A.-S.R. performed experiments, F.A. helped with image analyses and rendering. A.E. and J.-Y.R. acquired TEM images. L.L. participated in manuscript writing. K.S. initiated and planned the study, performed experiments, interpreted the results and wrote the manuscript. All authors read the manuscript and agreed to its content.

Funding

Fondation pour la Recherche Médicale (FRM) (DEI20151234416)

• Alexei Grichine

University Grenoble Alpes (AGIR-POLE FRAG15CS08)

• Anne-Sophie Ribba

Additional files

Video 1. Depth color-coded time-lapse of an unconstrained clot retraction (1×10⁸ platelets per ml in 50% plasma/50% PBS) in presence of fibrinogen-Alexa 488 (associated with Figure 1B). Fibrin fibers (green) present in z-projections of each time point are also shown upper right corner. Acquisition was started immediately after thrombin addition at 100 μm above the bottom of the well and image stacks were acquired (61 focal planes, step size 0.5 μm) for a time period of 20:27 min (100 frames with a time interval of 12.5 seconds, scale bar 10 μm).

Video 2. Two typical examples of platelets in a constrained clot with attached fibrin fibers (green) are shown (1×10⁷ platelets per ml PBS/50% plasma and fibrinogen-Alexa 488) (associated to Figure 2). Clot retraction was allowed to take place for 15 min before fixation and immunofluorescence staining of the integrin subunit αIIb (magenta). Samples were then processed for expansion (scale bars 10 μm = 2.5 μm after correction for expansion). Left panels, scrolling through the image stack; right panels, 3D reconstruction.

Video 3. Two typical examples of fiber accumulations (green) above spread platelets in the 2D fiber-retraction assay (4 μl plasma per ml PBS, see methods section) are shown (associated to Figure 4). Samples were stained for the αIIb integrin subunit (magenta) and expanded (scale bars 10 μm = 2.5 μm after correction for expansion). Upper panels, scrolling through the image stack; lower panels, 3D reconstruction.

Video 4. Two examples of platelets with bulbs encircled by fibrin fibers (green) in the 2D fiber-retraction assay (4 μl plasma/ml, see methods section) are shown (associated to Figure 5). Samples were stained for the αIIb integrin subunit (A, B; magenta) or the myosin light chain (MLC; C,D; magenta) and processed for expansion (scale bars 10 μm = 2.5 μm after correction for expansion). Upper panels, scrolling through the image stack; lower panels, 3D reconstruction.

Video 5. Two examples of platelets with twisted fibrin fibers (green) are shown in the 2D fiber-retraction assay (4 μl plasma/ml, see methods section) (associated to Figure 6). Samples were stained for myosin II (A; magenta) or phosphorylated myosin light chain (p-MLC; C,D; magenta) and processed for expansion (scale bars 10 μm = 2.5 μm after correction for expansion). Upper panel 3D reconstruction (Fig. 6A). Middle panel, scrolling through the image stack (Fig. 6C); lower panel, 3D reconstruction (Fig. 6D). Please note also the strong fiber accumulation around the platelets similar to platelets shown in Video 4/Figure 5.

Video 6. Four examples of a fibrin rosette associated with spread platelets in the 2D fiber-retraction assay (7 μl plasma per ml PBS, see methods section) are shown (associated with Figure 7). Samples were stained for the αIIb integrin subunit as well as for actin and processed for expansion (scale bars 10 μm = 2.5 μm after correction for expansion). 3D image reconstructions of image stacks. Fibrin fibers (green) and integrin or actin staining (upper and lower panels respectively, magenta).

Video 7. Three examples of fibrin (green) and myosin (magenta) localization for spread platelets in the 2D fiber-retraction assay (4 μl plasma/ml for A, B, E, F or 7 μl plasma/ml C, D; see methods section) (associated with Figure 8). Samples were stained for myosin (A, B, C, D) or phosphorylated myosin light chain (p-MLC; E, F) and expanded (scale bars 10 μm = 2.5 μm after correction for expansion). Upper panels, scrolling through the image stack; lower panels, 3D reconstruction.

Video 8. Simulation of fiber winding-up around a platelet bulb. Angle between the fiber and the x-axis α = π/2.5 (associated with Figure 10).

Video 9. Live imaging of spread platelets (transmission gray) and fluorescent fibrin fibers (green), scale bars 10 μm (associated with Figure 12). Upper three panels: Spread platelets from three different donors are shown, which reorganize and compact fluorescent fibrin fibers. Lower panel: Preformed fluorescent fibrin fibers were added to spread platelets. Surface coating is observed and a dense fibrin mass in the platelet center is rotating in several platelets.