Real‐time tracking of fibrinolysis under constant wall shear and various pulsatile flows in an in‐vitro thrombolysis model

Abstract A great need exists for the development of a more representative in‐vitro model to efficiently screen novel thrombolytic therapies. We herein report the design, validation, and characterization of a highly reproducible, physiological scale, flowing clot lysis platform with real‐time fibrinolysis monitoring to screen thrombolytic drugs utilizing a fluorescein isothiocyanate (FITC)‐labeled clot analog. Using this Real‐Time Fluorometric Flowing Fibrinolysis assay (RT‐FluFF assay), a tPa‐dependent degree of thrombolysis was observed both via clot mass loss as well as fluorometrically monitored release of FITC‐labeled fibrin degradation products. Percent clot mass loss ranged from 33.6% to 85.9% with fluorescence release rates of 0.53 to 1.17 RFU/min in 40 and 1000 ng/mL tPa conditions, respectively. The platform is easily adapted to produce pulsatile flows. Hemodynamics of human main pulmonary artery were mimicked through matching dimensionless flow parameters calculated using clinical data. Increasing pressure amplitude range (4–40 mmHg) results in a 20% increase of fibrinolysis at 1000 ng/mL tPA. Increasing shear flow rate (205–913 s−1) significantly increases fibrinolysis and mechanical digestion. These findings suggest pulsatile level affects thrombolytic drug activities and the proposed in‐vitro clot model offers a versatile testing platform for thrombolytic drug screening.


| INTRODUCTION
Thrombosis is the pathologic formation of a blood clot that obstructs flow through vasculature. Complications of thrombosis can be lifethreatening, such as ischemic stroke, myocardial infarction, and pulmonary embolism (PE). The mainstay of thrombosis treatment is either through mechanical thrombectomy, which requires a specialized device and an experienced clot retriever-operator, or by intravenous infusion of thrombolytic agents. 1,2 Both of the aforementioned approaches maintain inherent risks, bleeding being the primary. 3 Numerous in-vitro drug testing methods have been developed to assist in the screening and development of novel thrombolytic agents. A static in-vitro clot lysis assay using a statically formed clot is the most common protocol found in the literature. 4,5 Tests are performed by adding drugs to preformed clots and tracking changes in clot weight to depict drug efficacy. 5 This method can address scenarios where diffusion is the dominant biophysics of thrombolysis, for example, in patient with deep vein thrombosis (DVT) but is limited to address patients with PE or stroke not to mention arterial thrombosis. Nevertheless, most DVT patients are not recommended for thrombolytic therapy, according to Ziqian Zeng and Alexei Christodoulides contributed equally to this study. the guideline by the American Society of Hematology. 6 Static thrombolysis inherently ignores human hemodynamics where parameters like trans-thrombus pressure drop, and turbulent flow can dramatically affect clot permeation, resulting in distinctive drug efficacy profiles. [7][8][9] Properties of statically formed clots are different from those of native thrombi due to the lack of shear effects during clotting. Flowing blood even at low shear conditions (<500 s À1 ) promotes the formation of distinct clot motifs rather than complete stasis. [9][10][11] Thus, devices have been engineered in attempts to achieve more relevant physiological digestion conditions. 7 Microfluidic-related assays are useful in studying thrombosis progression, given their highly ordered flow patterns and ease of imaging. Clots formed in these devices tend to be dissimilar to native thrombi due to the nature of device lumen size and difference in flow dynamics. 12,13 Utilizing a Chandler loop device to form clot analogs and study clot digestion has gained popularity in thrombosis societies in recent years ( Figure 1a). 14,15 Clot analogs formed in the Chandler loop have revealed a good resemblance to native venous thrombi, arterial thrombi, and pulmonary emboli. 10 The Chandler loop also allows for clot digestion under shear; however, its over-simplified nature makes it a less representative model since the setup lacks important physiological circulatory components such as a reservoir, pressurized flow conditions. Importantly, while studying steady-state shear or continuous pressure effects on clot digestion is useful, modeling flow pulsatility generated by the human heart also demands research attention as it more closely represents in-vivo hemodynamics. Although pulsatile flow and cyclic wall stress have long been considered to impact invivo drug or enzyme activities, only a few studies have explored this relation in the past and the information on how pulsatile pressure amplitude (the pulsatile pressure difference between systolic and diastolic pressure over a cardiac cycle) affects the thrombolytic efficacy is missing. [16][17][18] Given the drawbacks and benefits of existing methods, developing a drug screening platform that combines a physiologically relevant clot and biomimetic flow is essential to more meaningful drug screening. This model may not eliminate the need for preclinical animal testing but can exclude inefficient agents earlier in the drug development pipeline to expedite the drug evaluation process. This study reports the design, validation, and characterization of an in-vitro thrombolysis model for thrombolytic drug screening. This model is a tubing-based system that incorporates a peristaltic pump, a flow dampener, a temperature-controlled reservoir, pressure sensors, a fluorometer, and a Chandler loop-formed clot analog (Figure 1b). The system can be used to develop a laminar flow at a constant shear or offer more human-relevant pulsatile flows through engineering scaling. According to the Buckingham-Pi theorem, the kinematic similarity of a flow can potentially be achieved by having identical dimensionless factors.
These include the ratio of length scales to ensure a dimensional similarity, Reynolds number to have a similar flow pattern, and Fanning friction factor to have a comparable fluid shear effects on lumen surfaces. When the pulsatile flow is relevant, Womersley number should also be matched to that of the target in-vivo condition to keep dynamic similarity. 19,20 In addition, the clot analog utilized is fluorescently tagged by using a previously reported low-impact labeling strategy to facilitate real-time tracking of fibrinolysis. 21

| Generating a physiologically relevant clot analog
A common means of introducing fluorescence into clot analogs for clot lysis study is using fluorescein isothiocyanate-labeled-fibrinogen (FITC-Fg). 21,24,25 Concentrations of FITC-Fg could range anywhere from 0.0075 to 0.6 mg/mL using 1-15 FITC per fibrinogen with no justifications as to why a particular concentration or labeling density is preferred. 9,[26][27][28] To create a truly representative thrombolysis system relying on FITC-Fg labeling, we had previously demonstrated endogenous Fg to FITC-Fg = 50:1 present minimal impact and 10:1 present small deviation on cell-free fibrin clot properties. 21 Herein, we expanded to optimize the ratio of endogenous Fg to FITC-Fg present in whole blood samples as the introduction of modified Fg could largely affect clot structure. We aimed for the highest level of fluorescence labeling while maintaining clot integrity and architecture. Clotting mixtures were prepared by adding FITC-Fg (14-FITC per fibrinogen) to whole blood at ratios of native Fg to FITC-Fg at 1:0 (control), 50:1, 10:1, and 5:1. Thromboelastography (TEG) was elected to compare clotting parameters across groups. Increasing the amount of FITC-Fg added to whole blood led to decreased maximum amplitude (MA), increased time to maximum amplitude (TMA), and decreased angle ( Figure S1). At native Fg to FITC-Fg = 10:1, samples showed minimally reduced MA (5.6%, p = 0.0208), similar TMA and angle values compared to the control group ( Figure S1). The proposed thrombolysis system utilizes clot analogs formed under pulmonary shear (464 s À1 ). Thus, a Chandler loop device was utilized to further characterize labeled clots at the mentioned FITC-Fg ratios. The final clot masses did not deviate significantly across groups; the 5:1 group (103.6 mg) and the 10:1 group (102.6 mg) had nearly identical masses to controls (Figure 2b; 103.2 mg). Subsequent digestion of these clots was further performed under shear in the Chandler loop with Alteplase (tPa). Clot mass loss after digestion was not significantly different ( Figure S2). Grossly, all clots showed similar F I G U R E 2 Whole blood clot analog generation and characterization. (a) Gross images of clots formed in the Chandler loop using varying FhF ratios under room light (top row) versus UV light (second row). Scale bars represent 20 mm. H&E (third row) and epifluorescence images (fourth row) acquired of Chandler loop clots at the same view. Scale bars represent 2 mm. For H&E staining, RBCs, fibrins, and WBCs were stained in dark red, pink, and blue. Black arrow indicates packed fibrin motifs. Note the loss of packed fibrin motifs in the 5:1 group, indicating a significant perturbation of clot architecture. (b) Masses and (c) RBC percentage of clots formed in the Chandler loop utilizing respective ratios of FITC-Fg. No significant differences ( p > 0.05) were found for clot mass and RBC percentage across groups. H&E, hematoxylin and eosin; RBC, red blood cell; WBC, white blood cell.

| RT-FluFF assay device characterization
The RT-FluFF assay utilizes a system that incorporates a peristaltic pump with an adjustable pulsatile-flow dampener that can produce physiologic levels of shear or flow dynamics in a versatile system allowing for great interchangeability of tubing lumen size and geometry ( Figure 5). Pressure sensors were placed before and after the clot Flow parameters used for factor calculations were collected from normal patient MPA data (Table S1). Three dampeners (60, 20, and 6 cc) were utilized to create three cyclic pressure amplitudes at 913 s À1 corresponding to 4, 20, and 40 mmHg, respectively (Figure 3d).
Final system characterization revolved around understanding how a clot would behave under various conditions of shear before the addition of thrombolytic drugs. Chandler loop-formed clots were formed with either no FITC-Fg, 5:1, or 50:1 ratio of FITC-Fg, that is, the upper and lower limits previously described. Analysis of the percent change in length between the groups at various rates of shear, 0-900 s À1 , showed that clots underwent highly reproducible degrees of elongation with respect to a given shear rate ( Figure S6). In other words, clot behavior was very consistent between varying levels of shear, ensuring that clots would be exposed to similar mechanical forces if thrombolytic drugs were introduced. study were higher than that of the human heart. To directly compare human-relevant pulsatility on thrombolysis, an additional experiment was conducted by adding a human heartbeat pulsation rate (1 Hz) to the highly dampened (60 cc dampener) 523 s À1 averaged wall shear condition. The pump speed of the heartbeat setup is doubled to give an equivalent volumetric output as the highly dampened 523 s À1 and noted as 523 s À1 and 1 Hz on/off. tPA-induced clot digestion rates were 63% higher ( p = 0.0165) in the heartbeat setup than in the highly dampened 523 s À1 , while an inversed result (p = 0.641) was seen when comparing no exogenous tPA control groups. These results further confirmed that the addition of pulsatility contributes to an improved drug-induced fibrinolysis. impacts the physical properties of the clot, particularly its mechanical strength, as we experienced during the clot handling process. 31 Interestingly, studies done on the effects of fibrin carbamylation, the reaction that FITC undergoes with fibrinogen, have shown nearly identical results to ours in terms of perturbations in clotting parameters. Binder et al. showed that carbamylation of fibrin led to decreased rates of polymerization, decreased fibrin cross-linking, and even increased propensity to clot lysis. 32,33 Our proposition for the ideal ratio of FITC-Fg to native fibrinogen should not be viewed as the ultimate standard to be utilized across all assay designs considering that various assays might demand varying levels of reporters. Rather, this data should be used as a guide to others if their aim is to generate physiologically relevant clot analogs for subsequent analysis of fibrinolysis. Additionally, one could easily propose using alternative reporters, labeling strategies, or increasing the utilized FITC-Fg concentration at the expense of decreasing the number of FITC molecules per fibrinogen, a possibility we did not explore in this paper specifically to maintain a higher degree of native fibrinogen in the assay.

| Thrombolysis and fluorometric monitoring of whole blood clot fibrinolysis
Thrombolysis, in particular fibrinolysis, was accomplished utilizing tPa introduced into human pooled plasma. Thus, leading to activation of both plasminogen within the plasma itself, in addition to plasminogen incorporated within the clot analogs through the process of clot formation. Importantly, as platelet contraction affects clot permeability that can cause a significant reduction in fibrinolysis, whole blood clot analogs utilized in this study were formed in the Chandler loop for 1 h to allow for a complete platelet contraction to ensure consistent clot permeability despite clots being visually stable within 15 min after clot initiation. 34 We aimed to understand fibrinolysis of whole blood clots via two primary means commonly presented in Although tPa concentrations were conserved between all digestion modalities, mechanical dissolution of the clots stemming from shear forces was not identical. It is these mechanical forces that drive clot mass loss as RBCs get released from the clot upon fibrinolysis. 37  Furthermore, increased RBC release is also suggested by the fact that the RT-FluFF assay does seem to maintain a slightly higher degree of percent clot mass lost at comparable tPa concentrations than those in the Chandler loop despite the latter setup having a slightly higher wall shear rate. This rise stems both from direct increased fibrinolysis and increased RBC release due to mechanical shear forces from pressuredriven flow acting on a degrading fibrin network. 38

| Shear and pulsatile impact on plasma clot fibrinolysis
Higher shear rate could contribute to drug replenishment, drug permeation, and increased mechanical clot tearing. Studies have shown increased drug-induced fibrin clot dissolution in parallel plate setups at higher shear rate. 42,43 Results from the RT-FluFF assay are in accordance with these findings where significantly faster tPA-induced plasma clot dissolution is observed in most groups at high shear (913 s À1 ) compared to those at lower shear (205 s À1 ).  46 We used fluorescently labeled plasma clots instead of whole blood clots for these tests to better understand the fundamental effect of pulsatility on fibrinolysis. Future works could implement whole blood clot tests at these pulsatile conditions to study the impact of pulsatility on thrombolysis.  50 The portable fluorometer we designed is capable of tracking FITC fluorescence signals in plasma, as plasma has low background fluorescence values with the wavelengths used. However, fluorescence signals can be significantly quenched by RBC release even at levels as low as 0.1% v/v within the system. This is in part due to a 550 nm sensitive photo sensor used for signal acquisition, a wavelength overlapping with RBC absorption at 538 nm. 36 Although experiments were not largely affected, since our digestion medium was plasma, different fluorophores or different photosensors would be recommended for studies that aim to utilize whole blood as a medium. Additionally, reliance on FITC-fibrinogen presents the issue of fluorescence quenching due to FITC residue burial within the protein complex. 51 To ensure sta-

| Ethics
Informed consent was obtained from all subjects, and/or their legal guardians, before blood donation. All experimental protocols were approved by the IRB ethics committee at Indiana University School of Medicine, USA.

| Whole blood and plasma processing
Venous whole blood was collected from healthy volunteers (n = 5) by a phlebotomist in accordance with guidelines/methods outlined in our institutionally approved IRB protocol (1610652271). All collection and handling of human specimens have been priorly approved by the IRB at our institution. Blood was collected into 3.2% Sodium-Citrate tubes and immediately pooled into 15 mL tubes for use (BD Vacutainer).
Sample hematocrits were calculated utilizing microhematocrit tubes centrifuged at 2750g for 3 min (Thermo Fisher Scientific). Whole blood was stored at 4 C and brought to room temperature (RT) for over 30 min before use. Human plasma units were donated on behalf of the Eskenazi Blood Bank for research use and were aliquoted before storage at À20 C.

| FITC-fibrinogen synthesis
Fibrinogen was labeled with FITC as previously described. 21 In short,

| Whole blood clot formation under shear
Whole blood clots were formed utilizing a Chandler loop to mimic exposure to physiologic levels of shear. In brief, FITC-Fg was added to citrated whole blood at a final ratio of 10:1 (0.88 μM) before recalcification. After gentle inversion of the mixture, the whole blood was recalcified utilizing CaCl 2 to achieve a final concentration of 11.8 mM.
Recalcified whole blood was loaded, using a syringe, into 5/32 in. tubing (Tygon 100-65 medical tubing) such that half of the volume of the end-joined loop would remain empty-final volume of whole blood utilized $2 mL (U.S. Plastics). The tubing loops were immediately placed on a rotating semi-submerged drum of 5.5 cm radius to begin clot formation. The drum was set to rotate at 40 RPM (calculated shear: 464 s À1 ) for 60 min. Partial submersion of the Chandler Loop drum in a water bath ensured clot formation was conducted at a constant 37 C. Upon completion, individual clots were weighed and imaged under both white light and UV light. Clots were subsequently re-submerged in their residual serum to ensure they remained hydrated during waiting periods.

| H&E staining and epifluorescence
A single representative clot formed in the Chandler loop was collected from each subject and stored in 10% neutral buffered formalin for 24 h before being placed in 70% EtOH. Once preserved, clots were submitted to the Indiana University Histology Core for sectioning and staining with H&E. Analysis of H&E slides was performed utilizing threshold analysis in ImageJ to isolate contributions of RBCs, white blood cells, and fibrin as part of clot composition. Non-stained sections from the same clots were utilized for epifluorescence imaging utilizing an LSM 800 confocal microscope (Zeiss).

| Static digestion protocol
A subset of the FITC-Fg-labeled whole blood clots previously described was subjected to fibrinolysis under static conditions. Frozen plasma was thawed at 37 C and aliquoted to subsequently be loaded with Alteplase at concentrations of 0 (i.e., no Alteplase added), 40, 200, or 1000 ng/mL (Genentech). Clots were gently loaded into 1.5 mL centrifuge tubes and filled to the 1.5 mL mark with plasma.
Subsequently, tubes were submerged in a 37 C water bath for the duration of the 60-min digestion. Triplicate samples of the plasma were taken at 0-, 30-, and 60-min marks of digestion for fluorescence quantification. A spectrophotometer (Molecular Devices, SpectraMax M5) was utilized for all readings. Settings for the spectrophotometer were as follows: 495 nm excitation, 519 nm emission, and 515 nm auto-filter. Clot weights were also recorded at the end of the digestion period to calculate the percent clot mass lost.

| MPA mimetic pulsatile flow setup
The system was adapted to create an in-vitro MPA flow condition to explore the impact of various pulsatility levels (4, 20, and 40 mmHg or ±2, 10, and 20 mmHg) on thrombolytic drug-induced clot digestion. 22 Given that average blood output in MPA is about 5.2 ± 1.0 L/min where average Re = 1570 ± 404 with a maximum volumetric flow of 21 L/min, the in-vitro flow model should be down-scaled to a level that maintains the system fidelity and saves experimental material. 53 To mimic the kinematic dynamics in MPA, parameters in the RT-FluFF model were adjusted to match an averaged Re where Q is the averaged volumetric flow rate of a pulsatile cycle, D is the lumen diameter, and ν is the kinematic viscosity using the following equation:  Table S1. 54 The reservoir is elevated to match average

| RT-FluFF assay digestion protocol
FITC-Fg-labeled whole blood clots were employed in the RT-FluFF assay for digestion in the constant shear setup. FITC-Fg-labeled platelet-free plasma clots were employed for all pulsatile flow tests.
Each clot was fixed in the tubing using a 31-gauge syringe needle at one-tenth the distance from the clot head. To mimic human pulmonary flow conditions, the reservoir was heated at 37 C and lifted to 8 cm above the clot level to give an average flow pressure of 12 mmHg, and the pump rate was either adjusted to generate a 398 s À1 wall shear flow for the constant shear setup or multiple other wall shear flows at different pressure amplitudes. For each experiment, the model was perfused with newly thawed pooled plasma with premixed Alteplase. Clot digestion was monitored utilizing the in-line fluorometer. Flow pressure, clot appearance, and clot break-off times were recorded.

| Shear-stretch analysis
Chandler loop-formed whole blood clots were fixed in the RT-FluFF assay tubing using a 31-gauge syringe needle at one-tenth the distance from the clot head. PBS flowed through the system at various rates of shear including 0, 300, 600, and 900 s À1 . Clots were allowed to stretch and equilibrate at each shear for 2 min before video capture. Quantification of the video frames was conducted utilizing Ima-geJ in which clot length could be accurately measured from clot head to clot tail. Percent change in length could then be calculated based on respective initial lengths.

| Data analysis
All data were collected/processed using Microsoft Excel. ANOVA analysis was used to ascertain statistical differences among groups with three or more conditions, followed by a Tukey test for individual subset comparisons. Student t tests were utilized to compare two categorical variables with each other. Statistical significance was deemed to be a p value < 0.05.