Thrombin-dependent nanoscale architecture and fibrinolytic remodeling of hydrated fibrin clots revealed by expansion microscopy

Abstract

Background:

Fibrin network architecture governs clot stability, permeability, and susceptibility to fibrinolysis, yet nanoscale characterization of hydrated fibrin structures remains challenging.

Objective:

To apply expansion microscopy (ExM) for quantitative subdiffraction-resolution imaging of fixed, hydrated fibrin architecture and determine how fibrin structure varies with thrombin, regulates clot permeability, and remodels during fibrinolysis.

Methods:

Fibrin structure in human platelet-free plasma clots was quantified across thrombin concentrations (0.3–1.5 U/mL) and during tissue-type plasminogen activator-induced internal fibrinolysis using ExM, with validation against confocal, stimulated emission depletion (STED), and scanning electron microscopy (SEM). Fiber diameter, pore size, porosity, and clot permeability were measured to relate nanoscale structure to macroscopic transport.

Results:

Increasing thrombin produced progressively thinner fibers, without altering network porosity, indicating redistribution of fibrin mass at constant void fraction. In contrast, pore area decreased monotonically and exhibited a strong linear relationship with clot permeability, suggesting that pore geometry, rather than bulk porosity, is closely associated with hydraulic transport under constant fibrinogen concentration. Fiber diameters measured by ExM closely matched those by STED and were systematically larger than SEM values, consistent with preservation of hydrated ultrastructure. During fibrinolysis, ExM resolved spatiotemporal remodeling of fibrin networks, including fiber thinning during active lysis followed by increased heterogeneity, apparent fiber bundling, and marked pore enlargement in late lysis.

Conclusions:

These findings define a nanoscale structural sequence linking thrombin-regulated polymerization and fibrinolytic degradation to macroscopic clot function and establish ExM as a robust platform for quantitative analysis of fibrin architecture in hydrated plasma clots.

Introduction

Fibrin forms the primary structural backbone of blood clots, assembling into a three-dimensional polymeric network that governs clot stability, permeability, and lysis^1–3. The architecture of the network, defined by the fiber diameter, branching density, pore size, porosity, and crosslinking, determines the mechanical properties of clots, resistance to enzymatic degradation, and transport of proteins and fluid through the fibrin matrix^4–7. These structural properties play an essential role in the balance between hemostasis and thrombosis and alteration in fibrin organization has been associated with a range of thrombotic and bleeding disorders, including arterial and deep vein thrombosis, venous thromboembolism, and hemophilia^8–13.

Fibrin polymerization is initiated by the thrombin-mediated cleavage of fibrinopeptides from fibrinogen, generating fibrin monomers. These monomers spontaneously assemble into half-staggered, double-stranded protofibrils, which then undergo lateral aggregation to form thicker fibers. The fibers branch and interconnect to form a three-dimensional fibrin network^3,14. Fibrinolysis occurs concurrently with and following clot formation and involves plasmin-mediated proteolysis of fibrin fibers, a process necessary to maintain vascular patency and prevent excessive fibrin accumulation^15–17. The balance between fibrin polymerization and degradation is crucial for normal hemostasis.

Prior studies have demonstrated that fiber diameter modulates fiber stiffness¹⁸, that network conformation influences clot viscoelasticity⁴, and that denser networks composed of thinner fibers are more resistant to fibrinolysis⁵. However, several key quantitative structure–function relationships remain incompletely defined. First, although thrombin-dependent changes in fibrin structure have been reported, the nanoscale organization of fibrin fibers and pores in fully hydrated plasma clots has not been systematically quantified. Second, it remains unclear how pore size and bulk network porosity each contribute to hydraulic permeability in hydrated fibrin clots. Third, while fibrinolysis is known to remodel fibrin architecture, the temporal evolution of fiber diameter distributions and pore geometry during internal fibrinolysis in intact hydrated networks has not been directly resolved.

Addressing these questions have been limited by methodological constraints. Although fibrin structure has been extensively examined using light microscopy, electron microscopy and atomic force microscopy^5,19–21, nanoscale visualization and accurate characterization of the three-dimensional fibrin network architecture under physiologically relevant (hydrated) conditions remain technically challenging.

Fluorescence confocal microscopy and scanning electron microscopy (SEM) are among the most commonly used methods to visualize fibrin architecture^{5,19,20,22,23}. Confocal microscopy enables three-dimensional visualization of fluorescently labeled fibrin but is constrained by the diffraction limit of light (~200–300 nm)²⁴, which impedes accurate measurement of individual fiber diameters that often fall within the 50–200 nm range^18,25,26. SEM provides superior nanometer-scale resolution of fibrin and blood clots²², but the sample dehydration and metal coating can introduce substantial sample shrinkage and structural distortions that preclude accurate quantitative assessment of the native morphology of biological objects^27–29. Dehydration may reduce the size of biological samples by up to 20%²⁸, which can potentially lead to overestimation of fiber density and underestimation of fibrin network porosity and pore size. Furthermore, network collapse during drying can obscure the native, straight, tensioned fiber architecture of hydrated fibrin, as well as the pore size. Moreover, sputter coating of metal artificially increases the apparent diameter of fibers. These limitations underscore the need for imaging strategies that preserve the native, hydrated state of fibrin while enabling accurate ultrastructural analysis.

Expansion microscopy (ExM) is an emerging super-resolution imaging technique that circumvents optical limits by physically enlarging biological specimens embedded in a swellable hydrogel^30,31. The method involves fluorescent labeling of the sample, followed by anchoring the fluorophores to a polymer matrix formed in situ. Subsequent enzymatic digestion of structural proteins homogenizes the mechanical properties of the sample, ensuring isotropic expansion. This physical enlargement increases the distance between labeled features, enabling nanoscale visualization using standard diffraction-limited microscopes. ExM has been successfully applied to diverse biological systems, including cultured cells, tissue sections, intact organs, early-stage embryos, and various microorganisms^32–38. These studies demonstrated the versatility of ExM across different scales, from single cells to millimeter-scale tissues such as the brain^39,40, and its ability to enhance effective resolution from ~200–300 nm to ~50–70 nm⁴¹. Despite its broad applicability, ExM has not yet been used for structural characterization of fibrin networks or blood clots.

In this study, we adapt and apply expansion microscopy to quantify nanoscale features of fixed, hydrated fibrin networks within human plasma clots and to relate structural organization to macroscopic clot transport properties and fibrinolytic remodeling. We analyze clots polymerized across a range of thrombin concentrations (0.3–1.5 U/mL) and directly compare ExM measurements with those obtained using conventional confocal microscopy, super-resolution stimulated emission depletion (STED) microscopy, and scanning electron microscopy (SEM). We quantify fibrin fiber diameter distributions, pore size, and network porosity, and measure clot permeability to link structural parameters to hydraulic transport. We further apply ExM to characterize architectural changes during tissue-type plasminogen activator (tPA)–induced internal fibrinolysis. Together, our results establish a hydration-preserving nanoscale view of fibrin architecture and reveal how thrombin-dependent polymerization and fibrinolytic remodeling shape the structure–function relationships governing plasma clot behavior relevant to (patho)physiological conditions.

Material and Methods

Plasma preparation and clot formation for microscopy

Frozen pooled human plasma collected by plasmapheresis (see Supplementary Methods) was used for all clot experiments to reduce donor-to-donor variability and enable controlled comparisons across experimental conditions. Plasma was thawed at 37 °C and centrifuged at 8000 × g for 15 min to remove residual platelets and debris. The supernatant was used as platelet-free plasma (PFP).

To visualize fibrin using confocal microscopy and Stimulated Emission Depletion (STED) super-resolution microscopy, Alexa-Fluor 647-labeled fibrinogen (Molecular Probes, 40μg/ml final concentration) was added to PFP, constituting 3% of the total volume. Clots were prepared using 30 μL plasma for STED and scanning electron microscopy (SEM), and 5 μL plasma for ExM-confocal imaging. Clotting was induced by adding CaCl₂ (17mM, final concentration) and thrombin (0.3–1.5 U/mL) to PFP samples. Samples were immediately transferred to glass-bottom dishes (P35G-1.5–14-C, MatTek Corporation) and incubated in a humidity chamber at 37°C for 40 min to allow complete polymerization. For internal fibrinolysis experiments, human recombinant tissue-type plasminogen activator (tPA; 1.75 ng/μL) was added to PFP prior to recalcification and thrombin addition (1 U/mL).

Fibrin clot immunostaining

Fully polymerized plasma clots were fixed with 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) for 30 min at room temperature and washed three times with PBS (5 min each) under gentle agitation. Samples were blocked with 1% BSA in PBS, containing 22.52 mg/mL glycine and 0.1% Tween-20, for 30 min and incubated with sheep anti-human fibrinogen polyclonal antibody (PA1–85429, Thermo Fisher Scientific; 1:200) at 4 °C for 48 h. For internal fibrinolysis experiments, the primary antibody incubation time was reduced to 24 h. After washing with PBS, samples were incubated with Alexa Fluor 568-conjugated donkey anti-sheep secondary antibody (A-21099, Thermo Fisher Scientific; 1:200) overnight at 4 °C, followed by an additional 1 h incubation at room temperature in the dark. Samples were then washed three times with PBS prior to imaging.

Expansion microscopy of plasma clots

Expansion microscopy (ExM) was performed using an adapted protocol comprising anchoring, gelation, digestion, and expansion steps. Immunostained fibrin clots were treated with AcX (6-((acryloyl)amino)hexanoic acid, succinimidyl ester) to covalently anchor proteins to the hydrogel matrix, followed by polymerization of an acrylamide–sodium acrylate gel. The samples were subsequently digested with proteinase K to homogenize the network and expanded in deionized water to achieve 4 to 5-fold linear expansion. Detailed ExM preparation procedures and validation of global and local hydrogel expansion are provided in the Supplementary Methods.

Imaging of PFP clots using different microscopy modalities

Confocal microscopy:

Imaging of expanded PFP samples was performed using Zeiss LSM 800 inverted confocal laser scanning microscope with a Plan-Apochromat M27 63x/1.40 oil immersion objective lens. Samples were imaged inside 35 mm poly-L-lysine precoated glass-bottom dishes (No. 1.5, MatTek). Each expanded gel was cut with a razor blade to ensure it fit the dish. Subsequently, a sample was applied on the glass-bottom coverslip of the dish, and excess water was removed with a micropipette. The gel was adjusted with a paintbrush to ensure its tight adherence to the coverslip with no bubbles. For fibrin fiber structural analysis, z-stack images of fibrin in expanded hydrogels were acquired at a resolution of 2048 × 2048 pixels (0.0495 μm per pixel) with 0.43 μm spacing between slices (10 slices per sample). Non-expanded fibrin networks were acquired under the identical settings, except with a 0.25 μm step size between slices. During ExM confocal imaging, the digital scan zoom was set to 0.5× because the expanded gel exceeded the field of view at the original zoom setting. All acquisition parameters, i.e., laser power, gain, pixel size, and the distance from the coverslip to z-stack planes, were kept consistent for quantitative comparison across clots formed at varying thrombin concentrations.

Stimulated emission depletion microscopy (STED):

STED was performed using a Nikon AXR inverted point scanning confocal microscope with a super-resolution Abberior STEDYCON module, equipped with a pulsed excitation laser (640 nm) and a 775 nm depletion laser. Imaging was performed with a Plan-Apochromat Lambda D 100x/1.45 NA oil-immersion objective. Z-stacks of 2.5 μm total depth were collected using a 0.25 μm z-step. The acquisition resolution was 667×667 pixels for fiber thickness quantification and 1333×1333 pixels for porosity evaluation, with a pixel size of 0.03 μm.

Scanning electron microscopy (SEM):

Plasma clots were rinsed with PBS and fixed with 2.5% glutaraldehyde in 50 mM cacodylate buffer (pH 7.4, 150 mM NaCl) for 2 h at room temperature. Samples were washed, dehydrated through graded ethanol, dried with hexamethyldisilazane, mounted on aluminum stubs, and sputter-coated with ~15 nm gold–platinum, as described⁴².

Imaging was performed using an FEI Quanta 250 FEG microscope at 10 kV with magnifications of 5,000× and 10,000×. At least nine fields were analyzed for each condition.

Structural characterization of fibrin

For each condition, fibrin network structure was assessed from z-stack images in Fiji (ImageJ v1.54p, NIH), measuring fiber diameter, fiber length, network pore size and porosity, and fiber junction density. To correct for potential sample drift during ExM z-stack acquisition, images were post-processed using the ImageJ plugin StackReg (translation transform).

Fiber diameter:

For ExM-confocal, conventional confocal, and STED images, diameters were measured by placing a 10-pixel-wide line Region of Interest (ROI) perpendicular to the local fiber axis in a well-focused segment. The fluorescence intensity profile measured using ImageJ Plot Profile tool was averaged across the ROI width, and the diameter was defined as the full width at half maximum (FWHM) of this averaged profile (Supplemental Fig.2). N>200 fibers per sample were analyzed and pixel distances were converted to μm using the image calibration (for ExM data, expansion-corrected diameters were obtained by dividing the physical distance by the gel linear expansion factor). For SEM images, where fiber edges are sharply delineated, diameters were measured directly by drawing a line perpendicular to the fiber axis with endpoints at the outer boundaries of each fiber and recording its length in ImageJ.

Fiber segment length and junctions:

Fiber segment length of fibers in confocal images was measured as the geodesic distance along the fiber centerline between adjacent network nodes (junctions). Junctions were detected and labeled from standard confocal z-stacks and categorized by node degree: 3 (Y-junction/branch), 4 (X-junction/crossing), or ≥5 (high-degree). Junction density was reported as the count of junctions per unit volume (μm⁻³).

Fibrin network porosity:

To determine fibrin network porosity, first a Gaussian blur (sigma = 4.0) was applied to z-stack images to enhance the signal-to-noise ratio. Next, the grey-scale fluorescent microscopy images were transformed to binary images using Otsu’s automatic thresholding method⁴³. Then, the “calc_porosity” function in Quanfima (Version 0.1a3) was applied to calculate the ratio of dark (void) voxels to total volume, defining the porosity value⁴⁴. Pore area was estimated using the fast-marching method with the Python extension module scikit-fmm. An optional multi-scale Sato filter was adopted to enhance fibrillar structures⁴⁵. The pore center candidates were determined by finding the local maxima of Euclidean Distance Transform inside each pore. Next, a wavefront was propagated outward from the pore centers, constrained by the surrounding fibers, to generate a watershed segmentation map that accurately delineates pore boundaries (Supplemental Fig.3). To validate the robustness of this measurement, an alternative approach of the linear intercept method was also adapted (Supplemental Materials, Supplemental Fig.4)⁴⁶.

Plasma clot permeability assay

Plasma clot permeability was assessed using a fluid permeation assay as previously described²⁶ with minor modifications. Clots were formed in cylindrical silicone tubes (MasterFlex L/S14, 96410–14; inner diameter 1.6 mm; length 20 mm) using PFP recalcified with CaCl₂ (17 mM) and thrombin (0.3–1.5 U/mL). Tubes were pre-coated with fibrinogen by incubation with PFP for 1 h at 37 °C, rinsed with deionized water, and air dried. 10μL of plasma were pipetted to one end of each tube. Following clot formation (incubation at 37 °C for 45 minutes) the clots were fixed using 4% PFA for 20 minutes and washed with PBS. Next, a column of sterile filtered double-distilled water (ddH₂O) was placed above the clot, and the flow of buffer through the clot was measured under constant hydrostatic pressure (typically 30–32 cm H₂O). The effluent was collected over a defined period (e.g., 20–60 minutes). To determine the flow rate, the mass of 21 sequential drops was measured using an analytical balance (Torbal AGCN220). As the flow rate required time to stabilize, the flow rate was calculated based on the average mass and time of drops 11 through 16. The Darcy permeability coefficient for each plasma clot was calculated using the Darcy formula: $K=\frac{QL\mu }{\text{S}\Delta P}$, where Q is the volume flow rate, L – is the length of the clot, μ - is the viscosity of the permeating fluid, S – is the cross section area of the clot, and ΔP – is the pressure drop across the clot.

Turbidity measurements of fibrinolysis

50μL of PFP were premixed with tissue-type plasminogen activator (tPA; 1.75ng/μL (~25nM), final concentration), recalcified with CaCl₂ (17mM, final concentration), and clotted with thrombin (1 U/mL) in a 96-well plate, thereby entrapping tPA within the fibrin network to assess internal fibrinolysis. The final concentration of tPA was selected to impose an accelerated fibrinolytic challenge and to permit observation of fibrin formation and degradation within the experimental time window. The samples were immediately transferred to the Molecular Devices SpectraMax^® iD3 where the absorbance measurements at wavelength of 405nm of the samples were recorded every 15s for 1 hour at RT. Control wells included PFP containing clotting reagents without tPA. Absorbance values were background-corrected and normalized to control wells. Turbidity curves were used to identify clot formation and fibrinolysis phases for time-point selection in parallel ExM experiments performed at the same tPA concentration.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism 10.5.0 (GraphPad, San Diego, CA, USA) and MATLAB R2023b (MathWorks, Natick, MA, USA) with specific tests described in the figure legends.

Results

Nanoscale quantification of thrombin-dependent fibrin fiber thickness in hydrated plasma clots

We first quantified how fibrin fiber diameter varies with thrombin concentration in hydrated clots. Human platelet-free plasma (PFP) clots were polymerized across thrombin concentrations ranging from 0.3 to 1.5 U/mL, and fibrin fiber diameters were measured using expansion microscopy (ExM), a hydration-preserving super-resolution approach. To benchmark these measurements, fiber diameters were also quantified using conventional confocal microscopy, STED super-resolution microscopy, and scanning electron microscopy (SEM), enabling direct comparison across imaging modalities (Fig.1, Supplemental Fig.5).

Figure 1. Representative comparative imaging of fibrin networks formed from human PFP using different microscopy techniques.

(A) Confocal microscopy of an expanded sample (ExM-confocal), (B) conventional confocal microscopy, (C) stimulated emission depletion (STED) microscopy, and (D) scanning electron microscopy (SEM) of clots polymerized with 0.5 U/mL thrombin. Insets (bottom left in each frame) show higher-magnification views of selected fibers; scale bars are indicated in each image. For ExM images in (A), the scale bar was corrected for the linear expansion factor and is shown in pre-expansion-equivalent dimensions. ExM-confocal imaging reveals improved resolution of fibrin fibers compared with conventional confocal microscopy, apparent resolution comparable to STED microscopy, and thicker-appearing fibers in hydrated clots than in SEM images.

All imaging modalities revealed consistent inverse relationship between fiber thickness and thrombin concentration. However, absolute diameter measurements differed markedly between methods. Across all thrombin concentrations (Fig.2), ExM-confocal imaging yielded significantly smaller fiber diameters than conventional confocal microscopy (mean range: 128–207 nm vs. 340–370 nm), reflecting the improved effective resolution afforded by physical sample expansion. Notably, ExM-derived diameters closely matched those obtained by STED microscopy (mean range: 145–215 nm), revealing no substantial difference between ExM and STED. These findings indicate that ExM enables super-resolution measurements of fibrin fibers without specialized optics. In contrast, SEM yielded about 1.1–1.5-fold smaller diameter values (mean range: 88–193 nm) compared to ExM, likely reflecting fiber shrinkage and collapse during dehydration, fixation, and metal coating steps.

Figure 2. Comparison of fibrin fiber diameters in PFP clots measured by different imaging modalities across thrombin concentrations.

Box plots show fibrin fiber diameter measurements obtained using expansion microscopy-confocal (ExM-confocal), standard confocal, STED, and scanning electron microscopy (SEM) in human PFP clots polymerized at different thrombin concentrations. Box plots show medians and interquartile ranges (IQR); dashed lines indicate nonlinear one-phase exponential decay fits for the clot-level median of each modality (R² = 0.99, 0.78, 0.92, and 0.99 for each corresponding modality). Sample sizes: 213–2,156 fibers from n=3–6 clots per condition. Statistical comparisons across imaging modalities at each thrombin concentration were performed using clot level summary values (see Supplemental Table 1).

To further compare ExM and SEM measurements, we analyzed the full distribution of fibrin fiber thicknesses across thrombin concentrations (0.3–1.5 U/mL) (Fig.3). Both methods revealed a consistent right-skewed (or log-normal) diameter distributions; however, ExM consistently yielded larger mean fiber diameters than SEM. The broader distribution observed in ExM suggests better retention of the hydrated fibrin structure, whereas SEM preparation may induce partial fiber collapse. Across all samples, ExM captured a larger fraction of fibers exceeding 200 nm in diameter that were underrepresented in SEM measurements.

Figure 3. Distributions of fibrin fiber diameters revealed by expansion microscopy (A-E) and scanning electron microscopy (F-J) at different thrombin concentrations.

Frequency histograms show measured fiber diameters with overlaid log-normal fits (black curves). The corresponding coefficients of determination (R²) and thrombin concentrations are indicated in each panel. Sample size: N=992–2156 fibers from n=6 clots per condition for ExM and N = 910–1113 fibers from n=3 clots per condition for SEM.

Together, these results provide nanoscale quantification of the inverse relationship between fiber diameter and thrombin concentration in hydrated plasma clots and demonstrate that ExM measurements are consistent with those obtained by STED microscopy while avoiding dehydration artifacts inherent to SEM.

Thrombin-dependent polymerization alters fibrin network pore architecture without changing porosity

To determine how thrombin concentration affects fibrin network organization beyond fiber thickness, we quantified clot porosity and pore size in PFP clots polymerized with 0.3–1.5 U/mL thrombin using ExM, conventional confocal microscopy, STED microscopy, and SEM.

Analysis of plasma clot ExM images showed that the mean clot porosity remained essentially unchanged across thrombin concentrations, varying by less than 1.5% (Fig.4). This indicates that the overall void fraction of the fibrin network was largely independent of thrombin concentration and consistent with the constant total fibrinogen concentration across all conditions. In contrast, porosity assessed by standard confocal microscopy showed a nonlinear decrease as thrombin concentration increased, likely reflecting diffraction-limited resolution that overestimates fiber volume fraction in dense networks. STED microscopy measurements corroborated the ExM results and likewise showed minimal changes in the mean porosity with increasing thrombin concentration, albeit with greater variability (coefficient of variance of 2.3%). In contrast, SEM measurements underestimated porosity by approximately 1.6-fold compared to ExM, likely due to network collapse during sample dehydration and preparation.

Figure 4. Quantitative assessment of fibrin network porosity across thrombin concentration using different imaging modalities.

Box plots show porosity of fibrin networks in human PFP clots formed in the presence of 0.3–1.5 U/mL of thrombin, measured using ExM confocal microscopy, conventional confocal microscopy, STED microscopy, and SEM. Sample sizes: N = 10–350 images from n=3–8 clots per condition. Statistical comparisons were performed using clot-level summary values (see Supplemental Table 2).

Next, to establish the pore-scale organization of fibrin networks, we quantified the absolute area of individual pores in the ExM-confocal images of PFP clots using the fast-marching method (see Methods). Analysis revealed a 1.9-fold decrease in the median pore area as thrombin concentration increased from 0.3 to 1.0 U/mL (Fig.5A), reflecting formation of a denser fibrin network. As thrombin increased further from 1.0 and 1.5 U/mL, pore area showed only a modest additional reduction (~1.2-fold), suggesting that the network approached a lower structural limit at high thrombin levels. The variation in the pore area (or the width of pore area distribution) (Supplemental Fig.6) decreased 1.7-fold at 1.5 U/mL of thrombin, indicating more uniform pore sizes at higher thrombin concentrations.

Figure 5. Thrombin-dependent changes in fibrin network pore area in plasma clots.

(A) Box plot show the absolute pore area measured from ExM-confocal images using the fast-marching method. Boxes indicate the interquartile range with median values shown; whiskers denote the 10th-90th percentiles. Sample sizes ranged from N = 2709 to 7165 pores from n=8 clots per condition. Statistical significance was determined using clot-level summary values with the Kruskal-Wallis test followed by post hoc Dunn’s multiple comparison test (****P < 0.0001). (B) Relative changes of fibrin network pore area with increasing thrombin concentration measured by ExM-confocal, conventional confocal, STED, and scanning electron microcopy. Data shown as Mean ± SD (n = 3–8 clots per condition). Pore area values were normalized by their median value at 0.3 U/mL thrombin for each imaging modality.

We next compared relative pore area changes across all four microscopy modalities using the fast-marching method (Fig.5B). All modalities showed an overall decrease in relative pore area with increasing thrombin concentration, although the magnitude of this effect varied by technique. ExM-confocal imaging exhibited a monotonic ~55% reduction in the average relative pore area as thrombin concentration increased from 0.3 to 1.5 U/mL thrombin. Conventional confocal microscopy followed a similar trend but showed a larger apparent decrease (~67% at 1.5 U/mL), consistent with diffraction-limited blurring that overestimated fiber volume fraction in dense networks. STED microscopy revealed minimal change between 0.3 and 0.5 U/mL thrombin, followed by a pronounced decrease at higher concentrations. SEM measurements were broadly consistent with ExM, but exhibited an approximately 15% larger relative pore area decrease at 0.75 U/mL thrombin, likely reflecting fiber collapse during dehydration. Pore area estimates obtained using the fast-marching approach were consistent with independent measurements derived from a linear-intercept method (Supplemental Fig.7). In addition to changes in pore area, higher thrombin concentrations were associated with reduced fiber segment length and increased junction density (Supplemental Fig.8), consistent with formation of a more highly branched and compact fibrin network.

Together, these analyses demonstrate that ExM provides stable, hydration-preserving, and quantitatively reliable pore-size measurements across polymerization conditions, while confocal microscopy, STED, and SEM each exhibit modality-specific biases that become more pronounced as fibrin density increases.

Plasma clot permeability assessment and relation to fibrin network pore size

To determine how fibrin network architecture influences fluid transport, we directly measured the hydraulic conductivity (permeability) of plasma clots polymerized across thrombin concentrations ranging from 0.3 to 1.5 U/mL. Permeability decreased approximately threefold over this range (Fig. 6A), with clots formed at 0.3 U/mL exhibiting the highest Darcy constant (~1.0–1.2 × 10⁻¹³ m²) and those formed at 1.5 U/mL showing the lowest values (~4–5 × 10⁻¹⁴ m²). This monotonic decline in clot permeability is consistent with the formation of thinner fibers and denser fibrin networks as thrombin concentration increases.

Figure 6. Permeability of plasma clots and relation to pore area.

(A) The measured Darcy’s constant of human PFP clots formed at various thrombin concentrations. Boxes indicate the interquartile range with median values shown; whiskers denote the minimum and the maximum measurements. Sample size ranged from N = 8 to 10 clots for different thrombin concentration groups. Statistical significance was determined using Kruskal-Wallis one-way test, followed by post hoc Tukey’s multiple comparison test (*P<0.05, ***P<0.001, ****P<0.0001). (B) Correlation between the mean fibrin network permeability and pore area, demonstrating a positive relationship between pore size and hydraulic permeability. The solid line represents a linear fit to the data.

To determine how the macroscopic clot transport properties are related to clot structure, we examined the relation between clot Darcy constant and fibrin network pore area. Linear regression analysis revealed a strong positive correlation between permeability and pore area (Fig.6B). Clots with larger pores (~4.2–5.2 μm²) were markedly more permeable than those with smaller pores (~1.8–3.0 μm²), which was in agreement with the Kozeny-Carman relation⁴⁷. These findings indicate that pore size is closely associated with hydraulic transport in fibrin clots and that ExM-derived structural measurements can help relate fibrin network structure to macroscopic transport properties.

Structural remodeling of the fibrin network during internal fibrinolysis

To assess the dynamic structural changes of plasma clots during internal fibrinolysis induced with tPA, ExM imaging was performed at defined time points determined from spectrophotometric turbidity measurements of clot formation and lysis kinetics (Fig.7A). In control clots without tPA, turbidity increased rapidly over ~12 minutes at a linear rate of 8.0×10⁻⁴AU/s and then plateaued with a slight further increase at 7.0×10⁻⁵AU/s, indicating a stable fibrin network. In contrast, clots formed in the presence of tPA, exhibited a lower, earlier peak turbidity, followed by a rapid decline between ~20 and 35 minutes at −6.0×10⁻⁴AU/s, and a slower decline thereafter at −2.0×10⁻⁴AU/s, eventually approaching a low baseline consistent with near-complete fibrin degradation. These three kinetic phases (pre-lysis, active lysis, late lysis) defined the timepoints for subsequent ExM-based structural analysis.

Figure 7. Quantitative structural assessment of fibrin fibers during internal fibrinolysis.

(A) Normalized turbidity curves of fibrin content during clot formation and tPA-induced internal fibrinolysis. The blue and gray curves represent clots formed in the presence and absence of tPA, respectively (Mean±SD, n=5 clots). Distinct lysis phases (pre-lysis, active lysis, and late lysis) were defined based on changes in OD slope and used for subsequent structural analyses. (B) Representative ExM-confocal images illustrating fibrin remodeling due to tPA-induced lysis of PFP clots at distinct time points (5 min, 20 min, and 35 min) of different kinetic phases (pre-, active, and late lysis). (C) Histograms of the distribution of the fiber diameter for each lysis phase. (D) Bar plots of the cumulative relative frequency for 4 diameter ranges: 0–0.1μm, 0.1–0.2 μm, 0.2–0.3 μm and 0.3–1.0 μm. N=100 fibers from n=5 clots. (E) Fiber diameters assessed with ExM for each lysis phase corresponding to (C, D). Results are presented as median, IQR between the 25^th and 75^th percentiles, and 10^th and 90^th percentiles, (F) Fibrin network pore size assessed using fast-marching method for each lysis phase. Results are presented as Mean±SD, (N=100 pores, n=5 clots). Statistical significance was determined using the Wilcoxon test (*P<0.05, **P < 0.01, ***P < 0.001).

ExM imaging across these phases revealed substantial remodeling of the fibrin architecture, characterized by dynamic changes in both fiber diameter and network organization (Fig.7B).

Fiber-level analysis showed distinct diameter distributions at each phase (Fig.7C). The mean fiber diameter was ~150 nm prior to lysis, decreased to ~100 nm during active fibrinolysis, and subsequently increased to ~200 nm in the late phase (Fig.7D). This pattern indicates initial fiber thinning due to plasmin-mediated degradation followed by an apparent increase in measured diameter, likely reflecting potential fiber collapse, bundling, or aggregation as the network loses structural integrity. Fiber diameter heterogeneity also increased markedly as fibrinolysis progressed, reflected by rising coefficients of variation (CV): 37.7% (pre-lysis), 62.6% (active lysis), and 70.5% (late lysis) (Fig.7C). Cumulative probability distributions further highlighted these phase-dependent shifts (Fig.7D). Specifically, during the active phase, the fraction of thin fibers (<100 nm) increased sharply, accompanied by a reduction in intermediate-diameter fibers (100–200 nm). In contrast, the late lysis phase showed a larger proportion of thicker structures (>300 nm), consistent with putative fiber collapse or residual bundles rather than intact fibrin filaments.

Consistent with these structural changes, ExM-derived pore measurements revealed an approximately six-fold increase in pore area during the late lysis phase (Fig.7F), indicating pronounced loosening of the fibrin network as fibers retract and degrade. Pore size remained relatively stable between pre-lysis and active lysis but expanded dramatically in the late phase, marking the transition from localized proteolysis to large-scale structural breakdown of the clot.

Discussion

Understanding how fibrin architecture governs the mechanical and transport properties of blood clots is central to elucidating the biophysical mechanisms of regulation of hemostasis and thrombosis. However, nanoscale visualization of fibrin fibers within hydrated, three-dimensional networks has remained a longstanding challenge due to the intrinsic limitations of optical resolution and the structural distortion induced by electron microscopy preparation. Here, we demonstrate that Expansion Microscopy (ExM) enables quantitative sub-diffraction imaging of fibrin networks using standard confocal microscopes and provides improved preservation of fixed hydrated fibrin architecture relative to SEM. By combining ExM with complementary imaging modalities and permeability measurements, we establish quantitative relationships between fibrin fiber structure, pore geometry, and macroscopic transport properties in plasma clots.

Our results establish several advantages of ExM for fibrin imaging. First, fiber diameter measurements obtained with ExM-confocal imaging confirmed that physical expansion enhances effective resolution sufficiently to resolve 50–200 nm fibrin fibers using conventional optics, yielding measurements comparable to those by STED. Second, ExM preserved fixed, hydrated fiber morphology, capturing thicker fibers and broader diameter distributions than SEM, which consistently underestimated fiber size due to dehydration-induced shrinkage. Third, porosity and pore-size measurements derived from ExM were consistent across polymerization conditions, unlike conventional confocal microscopy, which exhibited significant resolution-induced artifacts, particularly at higher fibrin fiber densities.

The ExM-derived structural metrics provided a mechanistic link between nanoscale fibrin architecture and macroscopic transport. Specifically, clot permeability decreased monotonically with increasing thrombin concentration in parallel with reductions in fibrin pore size. This trend is consistent with the Kozeny–Carman relation, which relates hydraulic permeability in porous materials to pore size and network geometry, underscoring pore geometry as an important structural correlate of hydraulic conductivity in fibrin clots⁴⁷. These results highlight the utility of ExM for quantitatively linking structural and functional transport properties of clots.

The observed inverse relationship between fibrin fiber diameter and thrombin in plasma clots is consistent with previous studies by Belcher, et al.⁴⁸ and Domingues et al.⁴⁹, who used different imaging techniques including SEM, confocal microscopy, and STORM⁴⁸ as well as transmission electron microscopy and atomic force microscopy⁴⁹ to evaluate fiber thickness. Fibrin fiber diameters have also been estimated from the wavelength dependence of light scattering using turbidity measurements, which relate scattering intensity to the mass–length ratio of fibrin fibers⁵⁰. However, such approaches provide ensemble-averaged diameter estimates and do not resolve spatial heterogeneity or full fiber diameter distributions within three-dimensional networks. Furthermore, our study directly addresses a critical issue highlighted by Belcher, et al., namely, that reported fibrin fiber diameters vary widely across studies because different imaging modalities introduce systematic measurement biases. Our results support this perspective: SEM consistently yielded the smallest diameters, whereas hydrated ExM and STED measurements converged closely. Recent efforts to standardize SEM diameter quantification, such as the protocol proposed by Cai et al.⁵¹, improve the reliability of SEM-based measurements, but inherently cannot eliminate dehydration-related distortions. ExM therefore offers a powerful complementary approach, providing a hydration-preserving, high-resolution benchmark for evaluating fibrin fiber diameters across conditions and modalities.

Importantly, ExM also enabled us to examine internal fibrinolysis in hydrated plasma clots with sub-diffraction resolution and to quantify progressive changes in fiber and pore structure. Earlier studies^5,52 used dynamic confocal imaging and turbidimetry to relate fibrin network conformation to fibrinolysis speed, while other works inferred relative changes in apparent fiber thickness from averaged cross-sectional intensity profiles⁵³. These approaches established that denser fibrin networks composed of thinner fibers lyse more slowly than less dense, coarse networks, and suggested that changes in fiber diameter and network packing accompany proteolysis. However, such approaches operate at diffraction-limited resolution, rely on indirect intensity-based diameter estimates, and do not capture full diameter distributions in three-dimensional hydrated clots. In contrast, our ExM-based analysis directly quantifies the distribution of fibrin fiber diameters during distinct phases of internal fibrinolysis induced by tPA. We identify an active lysis phase characterized by relatively uniform fiber thinning, followed by a late lysis phase marked by pronounced heterogeneity and the emergence of apparent thickened residual structures. This late-phase apparent “thickening” likely reflects putative collapse, buckling, or bundling of partially degraded fibers as network connectivity is lost, rather than new fibrin polymerization. These diameter-distribution changes occurred in parallel with an approximately six-fold increase in pore area. The idea that fibrinolysis involves both fiber thinning and progressive pore enlargement has been proposed based on prior imaging and modeling studies^5,53–55, but the detailed sequence and nanoscale distributional changes have not been fully defined. Our results support this conceptual framework, while providing a direct, quantitative, nanoscale-resolved view of how fiber diameter and pore geometry remodel over time in intact, hydrated plasma clots.

The structural fidelity afforded by ExM has potential clinical and translational relevance. Alterations in fibrin architecture, including changes in fiber diameter, network density, and pore structure, have long been associated with thrombotic disease. Early ultrastructural studies demonstrated that molecular abnormalities in fibrinogen can lead to substantial changes in fibrin fiber diameter and clot mechanical properties, as illustrated by electron microscopy analysis of fibrinogen Dusart clots that exhibited altered fiber structure and viscoelastic behavior⁵⁶. Subsequent clinical work further showed that abnormal fibrin architecture is associated with impaired fibrinolysis and increased risk of premature coronary artery atherothrombosis⁵⁷. More recently, using SEM, Pretorius et al.¹⁹ reported marked differences in fibrin fiber diameters and network organization between healthy individuals and patients with thromboembolic ischemic stroke, with patient clots exhibiting thicker and more heterogeneous fibers. Interpretation of such disease-associated differences is complicated by the preparation artifacts inherent to SEM. ExM offers an attractive alternative for assessing patient-specific clot architecture under hydration-preserving conditions, enabling robust comparison of fiber diameter distributions, pore structure, and potentially derived mechanical and transport characteristics between clinical groups. In combination with prior work showing that fiber diameter strongly modulates fiber stiffness¹⁸, that network conformation underlies rheological behavior⁴, and that fibrin architecture regulates fibrinolysis rates⁵, ExM-based measurements may help clarify how pathological changes in fibrin structure contribute to abnormal clot stability, impaired lysis, and thrombotic risk.

In summary, our results provide a quantitative nanoscale view of fibrin architecture in hydrated plasma clots and reveal how thrombin-dependent polymerization and fibrinolytic remodeling determine fibrin network structure and transport properties. By enabling quantitative nanoscale analysis of fibrin fiber diameter distributions and pore geometry in hydrated networks, ExM bridges the longstanding resolution gap between conventional optical microscopy and electron microscopy. These measurements reconcile discrepancies in reported fibrin fiber diameters, refine the structural consequences of thrombin-dependent polymerization, help link pore geometry to clot permeability, and provide a quantitative view of dynamic remodeling during internal fibrinolysis. Together, these capabilities position ExM as a valuable platform for re-examining fundamental questions in clot biophysics and for investigating disease- or therapy-induced changes in fibrin architecture relevant to hemostasis, thrombosis, and biomaterial design.

Data Sharing statement:

Quantitative data supporting the findings are provided in the article and Supplemental Material. Representative raw microscopy datasets supporting this study are deposited in Figshare and publicly available. Additional data are available from the corresponding author upon reasonable request.