Visualization of the dynamics of fibrin clot growth 1 molecule at a time by total internal reflection fluorescence microscopy

Much is known about the properties and functions of fibrinogen, including fibrin polymerization. Most studies of fibrin clotting visualized large clot structures either at low resolution by confocal microscopy or at higher resolution by electron microscopy (at certain time points), or used indirect biophysical methods.^1-6  As a result, little is known about the microscopic and molecular structural origins of the clot, especially about the mechanisms in the early stages of polymerization and the molecular kinetics.^1,7  More recently, some aspects of assembly before the gel point were visualized by deconvolution and spinning disk confocal microscopy,^8,9  but the smallest structures observed were protofibrils. Information about clotting, especially in the early stages that determine clot properties and the gel point, has clinical relevance, because the gel point is used diagnostically and disorders of clotting or fibrinolysis accompany or cause many pathologic conditions, including myocardial infarction and stroke.

Here we visualize and characterize individual fibrinogen molecules by total internal reflection fluorescence microscopy (TIRFM) and investigate the dynamics of fibrin assembly and fibrin fiber properties at the single-molecule level. In TIRFM, because of low background, sensitivity is high for events near the coverslip, and single molecules can be detected.^10-12  By analyzing the bleaching of single fibrinogen molecules, we characterize the distribution of labeling and develop a new molecular calibration for the fluorescence intensity of fibers, revealing aspects of the real-time growth of individual fibers at the molecular level.

Approval was obtained from the University of Pennsylvania Institutional Review Board for these studies.

Purified human fibrinogen from Hyphen BioMed was labeled nonspecifically with tetramethyl-rhodamine and Alexa Fluor 488 (supplemental Methods, available on the Blood Web site; see supplemental Materials link at the top of the online article) in parallel samples to assure that none of the observations were the result of artifacts because of dye attachment.

For single-molecule TIRFM observation, 1 mg/mL fibrinogen stocks were diluted 1000× in 20mM HEPES, 150mM NaCl, pH 7.4 (HBS), and ultracentrifuged for 30 minutes at 100 000g to sediment aggregates, leaving only single molecules in the supernatant. The supernatant was diluted to 0.1 μg/mL before introduction into the TIRFM chamber.

To study fibrin fibers in later stages of polymerization, we mixed platelet-poor plasma (PPP; preparation described in supplemental Methods) with fluorescently labeled fibrinogen (1 vol fluorescent fibrinogen to 17 vol PPP) with HBS containing thrombin and CaCl₂ for a final concentration of 0.5 U/mL thrombin and 20mM CaCl₂. The mixture was immediately introduced into the observation chamber.

To study the real-time polymerization of fibrin in plasma, PPP spiked with fluorescent fibrinogen (1 vol fluorescent fibrinogen to 10 vol PPP) was introduced into one side of the observation chamber and HBS containing 5 U/mL thrombin and 200mM CaCl₂ was introduced into the other side to induce polymerization through diffusion at the interface of the 2 liquids.

Data processing was done with an ImageJ plug-in that measures the time evolution of total intensity of single molecules, by recording intensity in regions of interest centered automatically on the bright areas in the first image of a time-series stack and repeating the measurement throughout the stack (ImageJ Version 1.36b software). To measure total intensity of fibrin fibers, similar but larger regions of interest, between branch points of fibers, were selected manually when using the same plug-in. After background subtraction, bleaching correction was performed using the bleaching time constant of each dye (supplemental Methods). A new molecular calibration was developed to determine M, the number of molecules in each fiber (including fluorescent and unlabeled molecules; see supplemental Methods). Using M and considering the fiber structure of parallel rows^1,13,14  we calculated the number of molecules in the cross-section of each fiber, c, and the corresponding diameter, D, of the hydrated fiber (see supplemental Methods).

Individual fluorescently labeled fibrinogen molecules that attached to the glass chamber surface were visualized (Figure 1A). Because of the low fibrinogen concentration, prior removal of aggregates, and low probability of attachment to the surface, most fluorescent signals observed represented single molecules, as previously demonstrated (see supplemental Methods).¹²  Time sequences of fluorescence images showed that most of the signals bleached in 1 step (Figure 1B), some bleached in multiple steps (Figure 1C-E) or presented an exponential decay (Figure 1F). Bleaching in 1 step is the signature of a single molecule labeled with 1 fluorescent probe.¹⁰  Bleaching in several steps indicates the presence of several dyes attached to a fibrinogen molecule. The intensity of the bleaching steps was in the same range for all bleaching events (Figure 1G). An exponential-decay trace comes from the superposition of signals from several dyes that bleach over time, and their number can be calculated from the intensity analysis. With increasing bulk-labeling ratio, the number of 2-, 3-, and 4-step bleaching events increased slightly; however, the most frequent were the 1-step bleaching events, even at higher ratios (Figure 1H). Similar results were obtained with both dyes.

Using this molecular distribution of active fluorophores per fibrinogen molecule (Figure 1H) and the total intensity of a bleaching step, while taking into account the labeling probability, and considering the unlabeled plasma fibrinogen, a single-molecule calibration for the fluorescent fibrin fibers seen in TIRFM was developed (see supplemental Methods) and the total number of molecules in a fiber, including the “not seen” molecules, was determined.

Two different approaches were used to observe aspects of fibrin polymerization. Fluorescent fibrin fibers were observed in TIRFM after induction of polymerization in a vial by adding thrombin into recalcified citrated plasma and transferring the mixture to the chamber by flow (Figure 2A). The flow did not orient the fibers parallel to the flow direction, showing that within seconds, the network was already formed and stable. However, the flow did bend many of the fibers, indicating a deformable network.

To observe polymerization ab initio, fluorescently labeled fibrinogen in plasma was introduced at one side of the microfluidic observation chamber and thrombin was introduced at the other side to induce polymerization at the interface between the 2 liquids (Figure 2B). Because at the μm dimensions of the TIRFM chamber, liquids that make contact do not mix,¹⁵  polymerization was due exclusively to the diffusion of thrombin molecules into the fibrinogen region, and the rate of fiber growth was therefore diffusion dependent. As expected for the low concentration of thrombin reaching the fibrinogen region, the fibers were thick.^16,17  Lateral growth of fibers continued for minutes (Figure 2C) and the single-molecule calibration showed that growth increases to thousands of molecules across (Figure 2D), but some fiber regions reach an early stationary state at lower values (Figure 2D inset). Fiber lateral growth is probably determined by the local kinetics of fibrinopeptide cleavage and oligomerization, as well as physical factors, such as the twisting of fibers.^16,18  The calculated diameters of the hydrated fibers are variable along the length of the fibers and can grow to hundreds of nanometers (Figure 2E) being in the same range as those measured in unbleached fluorescent images.

In conclusion, we demonstrated that the dynamics of fibrin polymerization can be studied at the single-molecule level using TIRFM. Through analysis of bleaching individual fibrinogen molecules, we obtained the molecular distribution of fibrinogen labeling, which was used to obtain a single-molecule intensity calibration. We observed live growth kinetics with molecular accuracy, determined for the first time the number of molecules within individual fibrin fibers, and calculated their diameters, bridging in this way the gap from molecular to assembled structures. This new methodology can be used for further studies of the assembly of fibrin or other biopolymers.

The authors thank Dr Henry Shuman and Dr Yale Goldman for use of their TIRFM setup.

This work was supported by National Institutes of Health grants HL030954 and HL090774.

National Institutes of Health

Contribution: A.H. designed the research, performed experiments, analyzed and discussed results, prepared the figures, and wrote the paper; K.C.G. and D.S. performed the fluorescence labeling of fibrinogen; and J.W.W. discussed results, analyzed data, and wrote the paper.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: John W. Weisel, Dept of Cell and Developmental Biology, University of Pennsylvania Perelman School of Medicine, 421 Curie Blvd, 1054 BRB II/III, Philadelphia, PA 19104; e-mail: weisel@mail.med.upenn.edu.