Whole blood clot optical clearing for nondestructive 3D imaging and quantitative analysis

: A technological revolution in both light and electron microscopy imaging now allows unprecedented views of clotting, especially in animal models of hemostasis and thrombosis. However, our understanding of three-dimensional high-resolution clot structure remains incomplete since most of our recent knowledge has come from studies of relatively small clots or thrombi, due to the optical impenetrability of clots beyond a few cell layers in depth. Here, we developed an optimized optical clearing method termed cCLOT that renders large whole blood clots transparent and allows confocal imaging as deep as one millimeter inside the clot. We have tested this method by investigating the 3D structure of clots made from reconstituted


Introduction
Blood clots (thrombi) form in response to vessel injury and prevent excessive bleeding [1]. The formation and mechanical properties of clots are largely determined by complex intermolecular interactions between cells and clotting proteins [2]. The contraction of clots, which requires the same actomyosin machinery involved in cell motility [3,4], is important for hemostasis to restore the flow of blood past obstructive clots [5], and for tissue regeneration [6]. Contraction of whole blood clots leads to a packed core of polyhedrocytes with much of the fibrin and platelets on the exterior [7]. Polyhedrocytes are the result of platelets pulling on fibrin during clot contraction; the erythrocytes are compressed from their normal biconcave shape to polyhedral shapes. The function of this packed structure is believed to stem bleeding in venous hemostasis but may preclude clot dissolution for certain pathologies due to low fibrinolytic enzyme permeability. Changes in the morphology of fibrin-platelet structures have also been associated with multiple pathological conditions such as heart attack, stroke, deep vein thrombosis, hemophilia, and many forms of cancer [8]. Acquiring threedimensional (3D) structures of thick clots or thrombi is of immediate biomedical importance, as it would provide information about internal architecture of these clots or thrombi in various thrombotic and bleeding conditions. Transmission electron microscopy and scanning electron microscopy (SEM) have been used to obtain detailed structural information from the clot surface or from thin sections of the clot interior. Intravital fluorescence microscopy [9,10] has enabled low-resolution studies of in vivo clot formation and light microscopy studies ex vivo [7,11,12] have obtained detailed structural information from near the clot surface. However, studies of whole clot structure taken from mouse models or patients with thrombotic disorders have lagged behind due to the presence of tightly-packed polyhedrocytes in the thrombus core rendering them optically inaccessible [7]. Furthermore, it has been difficult to study the redistribution of fibrin and platelets during clot contraction because of the same imaging limitations. Similar problems are also encountered when studying venous blood clots, which have a higher proportion of erythrocytes than arterial clots. Despite recent advances in understanding blood clot ultrastructure, fundamental aspects of how clot component interactions affect the composition and mechanical stability of clots and thrombi remain unclear.
Several methods for tissue decolorization and 3D imaging have been developed in recent years [13][14][15][16][17][18][19][20][21][22][23][24][25]. These methods have resulted in important advances for the understanding of biological structures and their arrangement within the 3D context of large tissues. For example, CUBIC utilizes aminoalcohols to target the removal of heme from erythrocytes [20]. However, CUBIC was optimized for clearing mouse organs and contains certain reagents that were determined to be counter-productive with a blood clot decolorizing process. Here, we report an optical clearing method termed cCLOT specifically tailored for whole blood clots or ex vivo thrombi. cCLOT quickly and efficiently renders clots transparent and enables fluorescence imaging of relevant 3D structures deep within clots. Importantly, the clot's ultrastructure is preserved with cCLOT, facilitating studies that link 3D clot structure to microstructure-based mechanistic models of blood clot formation and clotting pathologies. cCLOT permits imaging of structural components including fibrin and polyhedrocytes from intact ~1 mm-thick blood clots using two photon and confocal microscopy. Analysis of 3D images reconstructed from confocal z-stacks enables us to assess erythrocyte volumes and evaluate the fibrin network in the presence and absence of blebbistatin, a non-muscle myosin IIA inhibitor [26] that prevents clot contraction. We show that the fibrin network in the presence of blebbistatin is significantly perturbed from those in untreated clots with normal contraction. Interestingly, the volumes of erythrocytes were conserved throughout the clot thickness and were not found to vary significantly with depth, and remained the same in contracted versus non contracted clots suggesting that volume is conserved despite increased cellular density and that contraction does not require regulation of cell volumes. cCLOT was further validated on clots prepared from de-identified human blood, paving the way to detailed structural studies of human thrombi.

Blood collection
Whole blood was obtained from CD-1 outbred mice by cardiac puncture following the Institutional Animal Care and Use Committee (IACUC) approved protocol IACUC #14-04-1740 and collected into acid citrate anticoagulant (9:1). Mouse blood was kept at room temperature and used within 3 hours of collection. Mice were housed at Freimann Life Science Center Animal Facility at the University of Notre Dame. De-identified human blood was purchased from ZenBio Inc. and was obtained by venipuncture from healthy donors and collected in ACD anticoagulant (9:1). This blood was used for proof-of-principle evaluation of cCLOT on clots derived from human blood. The commercial human blood was stored at 4 °C and was used in compliance with University of Notre Dame Institutional Review Board.

Blood reconstitution and fluorescence labeling
Blood samples were reconstituted as previously described [7] to allow specific fluorescent labeling. Briefly, reconstituted blood was made by mixing 60 µl of fresh citrated whole blood, 20 uL of an "erythrocyte fraction" (see below), 20 uL of platelet rich plasma (PRP) and 50 µg/ml final concentration of Alexa 564-labeled human fibrinogen (Invitrogen). To generate the PRP and the erythrocyte fraction, 1 mL of fresh citrated mouse blood was centrifuged at 190 x g for 10 minutes at room temperature. The PRP was saved and the buffy coat discarded. The pelleted erythrocytes were rinsed twice with 2 ml of PBS containing 0.5% BSA (Sigma), by spinning at 190 x g for 10 minutes after each rinse. The final volume of the pelleted erythrocytes was adjusted to 1 ml with PBS and this was kept as the erythrocyte fraction. To facilitate imaging of single cells, a portion (or all) of the erythrocyte fraction was labeled with 1 µM final concentration of Deep Red Cell Tracker (DRCT; Molecular Probes) for 1 hour at 37 °C. To remove excess DRCT, the labeled erythrocytes were rinsed twice with 6 ml of PBS + 0.5% BSA as above and spun at 930 x g for 10 minutes to discard supernatant. Alternatively, erythrocytes were labeled with WGA lectin-Alexa 488 conjugate (Molecular Probes) following the same procedure as described for DRCT labeling. The final lectin concentration used for labeling erythrocytes was 1 μg/ml. Reconstituted blood samples with low (2% v/v) concentration of DRCT or WGA lectin -labeled erythrocytes fraction were prepared by mixing 60 µl of whole blood with 2 µl (2%) labeled erythrocytes fraction with 18 µl non-labeled erythrocyte fraction and the other components as described above. The 2% labeled erythrocyte preparation was used for all the fluorescence microscopy studies (Figs. 3 thru Fig. 7) with the exception of the images in Fig. 4, middle panels (e, f, g, h) in which a high (20% v/v) concentration of DRCT-labeled erythrocytes fraction was used by mixing 60 µl of whole blood with 20 µl (20%) labeled erythrocytes fraction and the other components as described above. Clots derived from human blood (ZenBio) were prepared following the same procedure after blood was equilibrated to room temperature (RT). The purpose of labeling only a fraction of the total erythrocytes (2% or 20%) was to facilitate detailed imaging of individual erythrocyte shapes throughout the clot volume.

Blood clot preparation
Clots were made with reconstituted blood on 1 mm thick 12-well silicon gaskets, pressed onto a microscope glass slide treated with Sigmacote (Sigma). A volume of 25 µl of citrated reconstituted whole blood was added into each well and clotting was initiated by addition of a 1.8 µl mix of bovine thrombin (Sigma) and CaCl2 at a final concentration of 1 U/ml and 20 mM, respectively. The gasket-slide set up was covered immediately by another Sigmacote treated glass slide to promote a flat top surface for the blood clot and to prevent samples from drying. The blood was then allowed to clot for 90 minutes at room temperature in the dark. Each prepared clot was then placed into a well of a 12-well plate and after a brief rinse in PBS the clots were fixed overnight in 10% buffered formalin by gentle rotation at RT.

Optical clearing optimization
A fixed blood clot was placed in each well of a 24-well plate and 1.25 ml of regular CUBIC-1 (25% w/w N,N,N',N'-tetrakis(2-hydroxypropyl)ethylenediamine (TED), 25% w/w urea, 15% w/w Triton X-100, 35% w/w water) [20] or cCLOT (see text for formulation) was added. Following a three hour clearing treatment of the clot samples at room temperature with gentle rotation, the solutions were collected, and the quantity of released heme was determined spectrophotometrically at A600. The clots were imaged and fresh optical clearing solutions added. The procedure was repeated at 6 and 24 hours.

Scanning electron microscopy
PFA-fixed clots were cleared and washed with 0.1 M cacodylate buffer, fixed with 2% glutaraldehyde and 1% osmium tetroxide for one hour, respectively. Clots were then dehydrated in ethanol and critical point dried in liquid CO2 and sputter coated with iridium. Samples were examined in a Magellan 400 field emission scanning electron microscope.

Perturbation of clot contraction with blebbistatin
Reconstituted fresh mouse blood was treated with blebbistatin before clot preparation at a final concentration of 0.05 mM or 0.3 mM. The samples were gently mixed and incubated for 40 minutes at room temperature. The blebbistatin-treated and non-treated mouse blood samples were used to make clots as described above.

Two photon and spinning disc confocal light microscopy
Stained and optically cleared blood clots were equilibrated and mounted in CUBIC-2 (50% w/w sucrose, 25% w/w urea, 10% w/w TED, 15% w/w water). Z-stack images were acquired using the two photon setting on a multiphoton laser scanning inverted microscope (Olympus FV1000) equipped with filter set (460-500, 520-560, 525-625, 650-700 nm) and a modelocked Ti:sapphire laser (Mai Tai DeepSee 690-1040 nm, Spectra-Physics). The two photon setting was used with a 25X water objective (NA 1.0 WD 2 mm) with 4x digital magnification. We used 785 nm wavelength from the IR laser to excite the fluorophores and the emission spectra was collected using the following bandpass filters: Alexa 488 (500-550), Alexa 564 (575-625 nm) and DRCT 640 (650-700 nm). Z-stacks were acquired using the confocal settings of the multiphoton (Olympus FV1000) using a 60X oil objective (NA 1.42 WD 150 µm). Imaging was also performed on a Nikon Eclipse Ti confocal microscope (Nikon Instruments, Melville, NY) with a Yokogawa spinning disc (Andor Technology, South Windsor, CT). Image data were collected on an iXonEM + cooled CCD camera (Andor Technology, South Windsor, CT) using MetaMorph v7.7.9 software (Molecular Devices, Sunnyvale, CA) with a Nikon 60X oil lens (NA 1.49).

Quantitative analysis of fibrin heterogeneity in cleared clots
Confocal z-stacks of cCLOT treated clots were imaged on the Olympus confocal configuration described above at 60X magnification. Two sets of z-stacks were obtained for each clot -one from surface to 40 μm depth and another from 50 to 75 μm depth. Both sets were imaged at 0.1 μm intervals. Fiji was used for all cCLOT quantification [27]. A 25 μmthick z-stack from 50 to 75 µm deep was extracted and a standard deviation projection performed to highlight dense aggregations of fibrin in the samples consistent with platelet aggregation present within the volume in a 2D plane. Prominent fibrin aggregates were segmented and counted in a 179 x 179 μm region using the Otsu auto thresholding algorithm with default parameters and the areas quantified using "Analyze Particles…." Two sets of 4 clots formed in the presence and absence of blebbistatin were analyzed.

Measurement of erythrocyte volumes
The average volume of mouse erythrocytes within optically cleared, reconstituted, whole blood clots at different depths and in the presence of blebbistatin were calculated using segmented z-stacks (n > 150 cells per population). To facilitate segmentation, a 2% (v/v) of DRCT-labeled erythrocyte fraction relative to total reconstituted blood was labeled with DRCT. For the analysis at different depths the sections were selected from a z-stack of 1 µmstep size slices spanning from the surface of the clot to 750 µm deep inside the clot. Images were obtained using an Olympus FV1000 two photon confocal microscope with 25x objective and 4x digital magnification. Likewise, the average volume of mouse erythrocytes in cleared clots of reconstituted whole blood in the presence of blebbistatin was calculated using zstacks of segmented images for two different portions of the clot: surface to 40 µm (40 µm thick stack) and 50 to 75 µm (25 µm thick stack); in this case each z-stack was acquired with 0.1 µm steps using a 60x objective.

Statistics
All reported p-values are calculated from a two-tailed Student's t-test and are summarized in the text and are tabulated by figure in the supplementary spreadsheet (Data File 1).

Developing an optimized optical clearing method for whole blood clots
Initially, we tested whether PFA fixed-whole blood clots could be optically cleared without destroying the clot's ultrastructure. We prepared clots from mouse reconstituted whole blood, fixed with 10% buffered formalin and treated them with CUBIC-1, a clearing solution that has been previously shown to extract heme and decolorize mouse tissues [20]. Treatment with CUBIC-1 for multiple days failed to completely clear the clot (Fig. 1(a)). CUBIC-1 is formulated for lipid extraction of brain tissue and contains a high concentration of Triton X-100, making CUBIC-1 highly viscous. We reasoned that by lowering the concentration of Triton X-100 from 15 wt. % to 0.2 wt. % the combined effect of greatly reduced viscosity and maintained cell membrane permeability would improve the solute exchange rate and more efficiently clear blood clots of the heme chromophore. The visible spectrum of solutions from PFA-fixed erythrocytes treated with either CUBIC-1 or TED aminoalcohol have previously been shown to overlap and correspond to heme with an absorbance peak at 600 nm [20]. Results from spectrophotometric analyses of heme release confirmed that all the modified solutions decolorized blood clots significantly faster than regular CUBIC-1, with 6.25% urea and 0.2% Triton X-100 solutions showing significant improvement over CUBIC-1 over 3 hours (p < 0.001) ( Fig. 1(b)). Next, we investigated whether the concentration of TED affects the rate of clearance, the rationale being that the high viscosity of TED impairs the extraction kinetics of heme. It has been previously shown by Tainaka et al. [20] that a modified CUBIC1 solution containing 25 wt % urea and 15 wt % Triton X-100 in water without the aminoalcohol TED (0% TED), could not decolorize PFA-fixed blood suspensions. We found that best results were achieved at 25-35% TED (Fig. 1(c)), suggesting that any advantage from changes in solution viscosity is outweighed by the higher concentration of hemeextracting TED. For the remainder of the study, TED concentration was fixed at 25% as the reduced viscosity at this concentration gave marginally better clearing over the course of 3 hours than other concentrations tested, while still achieving significantly better heme clearance over a longer, 6 hour, treatment period than 5% TED (p < 0.01). It should be noted that 35% TED did give significantly better heme clearance than 25% TED over the 6 hour treatment period (p < 0.05), and would have also been a viable option for the final clearing solution. The rate of clot heme release was monitored by measuring the absorbance of clearing solutions at 600 nm following treatment at 3 hours and 6 hours with CUBIC-1 or solutions containing 25% TED and various concentrations of urea and Triton X-100. A 6.25% urea and 0.2% Triton X-100 solution significantly decolorized clots as compared to CUBIC-1 over 3 hours (p < 0.001) (c): Efficiency of heme release following treatment with 6.25% urea, 0.2% Triton X-100 and various concentrations of TED. Results are shown as average ± standard deviation (SD) of triplicate samples. TED of 25%-35% achieves significantly better heme clearance over a longer treatment period than 5% TED (6 h, p < 0.01). (d): Relative blood clot area measured at 3, 6 and 24 h. Dotted line represents 100%. A four-fold reduction in urea concentration (6.25%) produced transparent clots that were comparable in size to non-cleared control ( Fig. 1(a), 1(d)) and significantly reduced the 6 h peak expansion (p < 0.05). (e): Excessive expansion of PFA-fixed clots following 24 hours of clearing with 6.25% urea (red, Treatment 1) could be reversed by an additional hour of treatment with PBS (light blue, Treatment 2) with a significantly reduced expansion (p < 0.02), followed by at least 1 hour of equilibration in CUBIC-2 (dark blue, Treatment 3). Clot area (%) for post-clearing solutions normalized to initial clot area. Dotted line represents 100%. Results reported as average ± standard deviation (SD) of triplicate samples.
Urea has tissue clearing properties, but causes increased hydration and expansion of tissues [16]. To test for the effect of urea on clot clearing and morphology, clots were treated with solutions containing various concentrations of urea, while keeping the amount of TED (25%) and Triton X-100 (0.2%) constant. We found that urea contributed greatly to improving clot transparency ( Fig. 1(a)). However the greater clearance at 25% urea was accompanied by significant clot expansion, with peak expansion occurring approximately 6 hours into treatment (p < 0.01). This drastic expansion occasionally caused the fixed clots to fracture ( Fig. 1(a)). At 25% urea, clots were 2-4 times larger compared with non-cleared clots ( Fig.  1(d)). A four-fold reduction in urea concentration produced transparent clots that were comparable in size to non-cleared control ( Fig. 1(a), 1(d)) and significantly reduced the 6 hour peak expansion (p < 0.05). Equilibration of clots in PBS at room temperature for 30 min significantly reduced clearing related expansion (p < 0.02) in instances of substantially elevated clot size after clearing (Fig. 1(e)). Partial loss of clot transparency following PBS treatment was restored by subsequent equilibration with CUBIC-2 [20] for refractive index matching before imaging (Fig. 1(e)). This treatment did not significantly re-expand the clot after equilibration in PBS. Fig. 2. Ultrastructural analysis of cCLOT-cleared blood clots. SEM images of the exterior and interior of blood clots treated with PBS or cCLOT for 24 hours. The structural arrangement with biconcave erythrocytes at the exterior and predominantly compressed polygonal-shaped erythrocytes at the interior of the clot is preserved after optical clearance. Scale bar, 10 μm.
We termed the modified and optimized CUBIC-1 derived solution: "cCLOT", which contains 6.25% urea, 25% TED and 0.2% Triton X-100. The cCLOT protocol recommends a 30 min PBS rinse after clearing to recover the original volume of the clot, as needed. This is followed by equilibration in CUBIC-2 for at least 1 hour before imaging.
We next sought to determine whether prolonged exposure to the optimized cCLOT solution affects the intrinsic ultrastructural properties of a blood clot (Fig. 2). Blood clots were fixed and cleared as previously described and imaged by SEM. We found no visible differences in cellular morphology or overall structural arrangement between cCLOT-cleared and non-cleared blood clots (PBS), which displayed the typical packing of enclosed compressed erythrocytes in the interior and biconcave erythrocytes at the exterior [7]. Clots stained for fibrin showing difference between non-cleared and cCLOT cleared clots. Note the extensive fibrin network at the outer edge and within the clot core (c", d"): XZ orthogonal views of non-cleared vs. cCLOT-cleared clots demonstrate the ability of the cCLOT method to image fibrin networks throughout the clot and at depths that exceeds control clots by ~5-fold. Entire region depicted in a", b", c" and d" represents a depth of 739 μm. Fibrin aggregates can be seen as small, closely packed irregular structures from which fibrin fibers project. Imaged with 25x water objective (1.0 NA, 4x digital magnification) with 1 µm step z-slices. Scale bars are 25 μm.

cCLOT reveals the inner structure of contracted whole blood clots
As a first demonstration of the utility of the cCLOT method, two photon microscopy was performed to image 1 mm-thick optically cleared clots derived from mouse blood made with a fraction of DRCT labeled erythrocytes and Alexa 564-labeled human fibrinogen for fibrin visualization (Fig. 3, Visualization 1). Optical microscopy of non-cleared whole blood clots becomes impossible beyond a thickness of less than 20 μm, failing to reveal important details in fibrin architecture. In contrast, both the erythrocytes and fibrin network of cCLOT-treated clots could be imaged through the entire thickness of the contracted clot (>700 μm) (Fig.  3(a)" vs. Fig. 3(b)" and Fig. 3(c)" vs. Fig. 3(d)"). Average volume of erythrocytes as a function of clot depth. The erythrocyte volume was measured from segmented images of z-stacks obtained at 1 µm steps in clots containing 2% v/v of DRCT-labeled erythrocyte fraction relative to total reconstituted blood (erythrocytes shown in green). Images for (a) and (b) were obtained using two photon microscopy with 25X water lens 4x digital magnification. Erythrocyte volumes in contracted versus non-contracted cCLOT cleared clots (c): Average volume of erythrocytes was measured for control non-treated reconstituted whole blood (WB) clots and clots made with blebbistatin treated reconstituted whole blood. The erythrocyte volume for (c) was measured from segmented images of z-stacks of clots containing 2% v/v of DRCT-labeled erythrocyte fraction relative to total reconstituted blood. Data for figure (c) extracted from images obtained using an Olympus confocal microscope with 60X oil lens. Z-stacks obtained at 0.1 µm steps.
Next, we imaged clots at higher magnification. Confocal microscopy is limited in imaging speed and generates large data sets, which quickly become prohibitive when the goal is to image large samples at high magnification and high resolution. We thus elected to image multiple clot samples at selected intervals through the depth of the clots. Figure 4 shows high resolution imaging of a cCLOT-cleared clot with confocal imaging using a 60x objective. The images demonstrate that cCLOT enables composite views of the fibrin network and erythrocytes in clots from whole blood at greater depths and resolution than previously possible [7]. The composite images were obtained from clots containing different fractions of erythrocytes labeled with DRCT to facilitate imaging (Fig. 4). The middle panels of Fig. 4 show erythrocytes with angular polyhedral shapes within the cleared clot ( Fig. 4(e), 4(f), 4(g) and 4(h)). This result validates and extends previous findings of polyhedral compacted erythrocytes [7] in the clot interior as imaged with light microscopy and SEM. We also observe a higher density of fibrin known to be associated with platelet aggregates towards the surface of the clot (~25 µm depth within the clot's cap) (Fig. 4(a)-4(d)). Localized regions of fibrin aggregates at different depths in the clots are consistent with platelet aggregation (punctae in Fig. 4(j), 4(k)). SEM images cannot show the extent of the 3D fibrin network towards the interior of clots ( Fig. 2 and Fig. 4(i), 4(l)). The extensive fibrin network with platelet aggregates very deep in the clots is consistent with the clots being formed under conditions that do not allow full contraction. Areas of compacted erythrocytes radiating from platelet aggregates as observed in the SEM (exposed core region illustrated in Fig. 4(i) and shown on Fig. 4(l)) resemble zones of the clot core architecture revealed by light microscopy (Fig. 4(j), 4(k), deep in the clot).

Quantitative analysis of erythrocyte volumes
To demonstrate the unique feasibility of the cCLOT method in providing quantitative threedimensional cellular analysis of a clot, we measured the average volume of segmented mouse erythrocytes in a cleared clot at different depths. We found that the volume of mouse erythrocytes is conserved across different depths of the contracted clot (Fig. 5) with an average value of 44 µm 3 , close to the reported mean corpuscular volume of ~40 µm 3 of mouse erythrocytes [28]. The conserved volume of erythrocytes indicates that compression as a result of clot depths does not result in fluid loss from these cells. Image (e) shows the average area of fibrin aggregates ± SD per region for 4 samples. Quantified region at 60x represents an area of 179 x 179 μm. (g, h): Maximum intensity zprojection of a 25 μm slab (50 μm to 75 μm depth) of a contracted clot (g) and uncontracted clot formed in the presence of blebbistatin (h). Scale bars are 10 μm. Fibrin (orange hot) was imaged by incorporating Alexa 564-labeled human fibrinogen in reconstituted blood and contained 2% v/v of DRCT-labeled erythrocyte fraction relative to total reconstituted blood (erythrocytes not shown).

Structural analysis of perturbed clots
In a proof-of-principle study, we tested the ability of cCLOT to reveal abnormalities in blood clot structure that were perturbed pharmacologically. Adding blebbistatin, a cell permeable specific inhibitor of non-muscle myosin II and of platelet contraction, perturbed the formation of contracted clots made with fresh mouse blood. Initial visual observation confirmed complete contraction of un-treated control clots (~30% volume reduction) whereas mouse blood treated with either 0.05 mM or 0.3 mM blebbistatin produced clots that did not contract (no volume change observed) (Fig. 6(a), 0, 0.3 and 0.05 mM blebbistatin). Following clearing with cCLOT, confocal imaging revealed qualitative changes in the appearance of the fibrin network between contracted control clots and non-contracted clots incubated with 0.3 mM blebbistatin ( Fig. 6(b)-6(c)). Blebbistatin-mediated platelet actomyosin inhibition resulted in significantly fewer fibrin aggregates that differed significantly in size. These changes were quantified using Fiji as described in Methods and are shown (Fig. 6(d), 6(e)). Uncontracted clots produced with blebbistatin-treated fresh mouse blood show significantly fewer and larger fibrin aggregates than contracted control clots (Fig. 6(d)). In addition, the fibrin aggregates produced in the contracted clots exhibited more uniformly, but smaller distributed fibrin aggregates that are interconnected by fibrin fibers (Fig. 6(g)). While platelets were not stained separately, these fibrin aggregates are consistent with platelet aggregation. In contrast, uncontracted clots treated with blebbistatin lead to a non-uniform distribution of aggregated platelets corresponding to large fibrin aggregates. The fibrin network shows reduced interconnectivity, which is disrupted due to inhibition of platelet contraction (Fig. 6(h)). Importantly, cCLOT enables analysis of red blood cells throughout the clot. To test whether clot contraction results in volume compression of erythrocytes, we measured erythrocyte volumes with respect to clot depth in non-contracted and contracted clots made with the fresh mouse blood. We found that there were no significant differences between the volumes of erythrocytes in contracted clot vs. perturbed blebbistatin-treated clot (Fig. 5(c)). These results are consistent with increased clot density of contracted clots and indicate that compression as a result of clot contraction does not result in fluid loss from these cells. This analysis serves to demonstrate the quantitative utility of the cCLOT system for fully defining the structure of whole blood clots under both physiological and pathological conditions.

Application of cCLOT method to human blood
To further validate our clot-optical clearing method, we prepared thick clots with commercial de-identified human whole blood for initial evaluation studies (Fig. 7). We show the ability of imaging the complete fibrin network spanning from the clot surface to ~800 µm depth using two photon microscopy ( Fig. 7(a)-7(c), Visualization 2). Additionally we were able to validate the different erythrocyte shapes at the surface of the clot (biconcave) versus deep in the clot (polyhedral) (Fig. 7(f)-7(h)). These results suggest that cCLOT will be an effective method in future studies using freshly drawn human blood or fixed samples of human thrombi. Significant progress has been made in recent years towards developing clearing solutions and protocols to facilitate fluorescence microscopy deeper into tissues [22,23,29,30]. An effort to clear whole mouse body tissues resulted in the development of CUBIC-1, which eluted heme from PFA fixed blood suspensions compatible with fluorescent protein imaging. However, we found that CUBIC-1, which efficiently clears lipid-rich mouse brain, was not able to efficiently clear PFA-fixed whole blood clots. Here, we have established and validated a modified aqueous-based clearing solution, compatible with confocal microscopy, that produces transparent blood clots at physiologically relevant millimeter size scale in less than 24 hours. cCLOT treatment enabled deep 3D fluorescence imaging while maintaining ultrastructural integrity and cellular morphology.
To the best of our knowledge, this is the first report of an optical clearing technique for a whole blood clot with the purpose of detailed visualization of its cellular and fibrous components using fluorescent confocal microscopy. Comparison of different clearing solutions and their ability to preserve synaptic structural integrity using TEM has previously shown that ScaleS [31], which has a low Triton X-100 concentration (<0.2%), is best in preserving cellular membrane structure. By varying the concentrations of the clearing solution components, we found that using a low Triton X-100 concentration (cCLOT contains 0.2%) in addition to a 6.25% urea concentration (as opposed to 25% in CUBIC-1) preserves the morphology of erythrocytes, therefore overcoming problems associated with the implementation of CUBIC-1 for optical clearing of blood clots.
To demonstrate the utility of the cCLOT, we showed that reconstituted contracted whole blood clots can be imaged using two photon fluorescence microscopy to depths on the millimeter scale, whereas previous in vitro contracted blood clot imaging was limited to the clot surface (~20 µm deep) due to the opacity of erythrocytes. Previous findings indicated the presence of surface fibrin as well as formation of polyhedrocytes deeper in contracted clots. However, these studies were done using either SEM of clot fragments or confocal microscopy of the superficial layers ( 20 μm deep) [7]. Our approach provides a new solution for 3D structural visualization allowing assessment of fibrin organization and the spatial packing of erythrocytes throughout contracted clots. Additionally, we are able to visualize the structure of polyhedrocytes within the clot volume. Our results show significant impairment of clot contraction in the presence of blebbistatin, confirming previous findings indicating an important role for actomyosin in platelet-driven clot contraction [32].
Here we additionally validate the ability to optically clear opaque ~5 mm diameter x 1 mm thick clots made from pre-labeled reconstituted human blood. The cCLOT method revealed the complete fibrin network spanning the fibrin cap at the clot surface and along the inner clot. cCLOT also allowed a detailed imaging of the deformed polygonal shapes that erythrocytes under the fibrin cap (~20 um) and deeper in the clot interior as opposed to the disc-shaped erythrocytes observed at the surface.
cCLOT combined with the blood reconstitution and pre-labeling strategy presented here can be a valuable tool in basic and preclinical research for evaluating how changes in precise clot composition made from normal or drug treated blood affect clot formation, morphology and stability. Additionally, the use of microfluidic devices for studying clot formation under flow that mimics in vivo conditions as described in [33] could now be combined with prelabeled blood components followed by fixing and cCLOT optical clearing steps, easily automated, to facilitate high resolution imaging of resulting clot composition. Such cell labeling strategies are frequently used to develop important insights into complex biological processes [34]. Also, microfluidic devices for high-throughput screening could potentially facilitate these and similar studies, and assisting in drug development [35]. Whole blood clots isolated from transgenic mice expressing selected fluorescently labeled hematopoietic markers as previously described [36] would represent an ideal type of model system to study roles of different blood cells in clot formation which could be combined with induced in vivo pathological conditions. Little is known about the structure and composition of opaque venous thrombi and how it relates to the clinical history of the patient. A very exciting application of cCLOT would be to optically clear millimeter scale ex vivo clots aspirated from patients. The specific labeling of cellular components will require improvements in the delivery systems of antibodies in combination with the use of smaller fluorescently labeled molecular probes including nanobodies [37].

Conclusions
Whole blood clots have been a challenge to image with light microscopy due to the opacity of erythrocytes. In this work, we optimized a water based optical clearing solution that renders millimeter-scale whole mouse or human blood clots transparent, while conserving the clot ultrastructure. The clearing ability was demonstrated by imaging fluorescently labeled erythrocytes and fibrin across the whole clot depth using confocal and two photon microscopy. This new method allowed the validation of the erythrocytes deformed shapes (polyhedrocytes) within the clot using confocal microscopy, as well as revealing the entire fibrin network and its interaction with the erythrocytes. This technique allows quantitative analysis of clot composition at the cellular level. We found that erythrocyte volumes do not significantly change upon clot contraction and remain the same at different depths. This supports the idea that clot contraction is achieved by increased packing efficiency of erythrocytes and not through cell volume reduction. Furthermore, we demonstrated the potential use of cCLOT in assessing drug-associated changes in fibrin organization and clot structure by evaluating fibrin aggregates in the presence of a non-muscle myosin II inhibitor drug, blebbistatin.