Magnetically powered microwheel thrombolysis of occlusive thrombi in zebrafish

We previously demonstrated that only µwheels functionalized with tPA can achieve lysis of fibrin clots in vitro (18). To test this in vivo, we first infused fluorescently tagged microparticles into circulation at 5 days post fertilization (5 dpf) through the retro-orbital plexus (Fig. 1A). After 1 h in circulation, no adverse effects nor any microparticle accumulation in any particular location was observed. We have previously shown the ability to produce occlusive thrombi using laser-mediated endothelial injury in 3 to 5 dpf zebrafish larvae (21), as well as the ability to lyse spontaneous thrombi using human tPA (23). We performed injury prior to microparticle infusion and found accumulation at the upstream edge of the occlusive thrombus (Fig. 1 B and C and Movie S1). These studies demonstrated that zebrafish larvae tolerate circulation of the beads and that they can be easily visualized.

While, 1 ng of tPA infused into individual larvae is approximately equal to the dosing in patients (0.9 mg/kg) we tested a range of doses, from 1 to 40 ng/larvae, for recanalization. Only high concentrations of tPA, greater than 10 ng/larvae, were able to achieve recanalization within 4 h and without significant toxicity (Fig. 2). Greater concentrations resulted in toxicity evidenced by gross morphologic changes. These studies suggest that, to achieve efficient lysis with minimal toxicity, a high local concentration of tPA is needed.

To test the ability of µwheels to recanalize the vessel, we used microparticles functionalized with recombinant human tPA or Atto-488 fluorescent dye. We used a tPA-specific fluorometric assay to measure the concentration of active tPA bound to the microparticles and calculated that there was approximately 8.4 pg infused per larva. Following endothelial injury, a magnetic field was applied to produce µwheels (Fig. 1 D–F). All of the larvae with tPA-functionalized µwheels recanalized within 30 min, while control µwheels did not (Fig. 3 and Movies S2–S5). We infused tPA-functionalized microparticles without application of a magnetic field and found no evidence of significant recanalization within 30 min. We also infused non-functionalized particles followed by application of a magnetic field using corkscrew motion. We were only able to apply the magnetic field for 30 min given the heat generated by the coils, but there was no lysis in the absence of tPA. Taken together, these data show that only tPA-functionalized µwheels are capable of thrombolysis.

The corkscrew motion has been previously shown to be the most effective at lysis in vitro (16). To examine whether this is also the case in vivo, we performed a comparison to additional modes: switchback, flipping, and rolling movements. We compared all of these using tPA-functionalized µwheels and found that corkscrew motion was most effective at recanalizing occlusive thrombi, in a median time of under 10 min. The other modes were also able to recanalize occlusive thrombi within 30 min, but with a median of ~15 min or higher (Fig. 4).

tPA is well known to mediate its effects by activating plasminogen to plasmin. Although the plasminogen (plg) gene is single copy and conserved in the zebrafish genome, it has not been previously shown whether it has the same function. After endothelial injury, tPA-functionalized microparticles were infused into larvae generated from plg^+/− incrosses, followed by application of the magnetic field. µWheels were able to recanalize occlusive thrombi in the majority of wild-type and heterozygous mutants within 30 min, while nearly all of the homozygous mutants were unable to do so (Fig. 5A). There was a small significant decrease in the percentage of heterozygous mutants that were able to recanalize compared to wild-type. These data show that fibrinolysis is plasminogen-limited, at least at high local tPA concentrations (19). Finally, when human plasminogen was infused into plg mutants, the ability to recanalize was restored (Fig. 5B).

Zebrafish were maintained according to protocols approved by the University of Michigan Animal Care and Use Committee. Wild-type fish were a hybrid line generated from crosses of AB and TL zebrafish acquired from the Zebrafish International Resource Center. Tricaine (Western Chemical) was used for anesthesia and rapid chilling in an ice-water bath for euthanasia in accordance with American Veterinary Medical Association (AVMA) guidelines.

Recombinant tPA was biotinylated using N-hydroxysuccinimide (NHS) activated biotin by incubation on ice for 3 h. Excess biotin was removed using a desalting column, and biotinylated tPA was stored in 20 µL aliquots at −80 °C. 5 µL of 1 µm streptavidin-conjugated iron oxide beads (ThermoFisher Dynabeads MyOne Streptavidin T1) were mixed with biotinylated tPA, incubated at 4 °C overnight, washed three times to remove excess tPA (resulting in negligible free tPA), and resuspended in 15 µL 2% bovine serum albumin (BSA) in saline. Atto 488-biotin was similarly conjugated to Dynabeads without tPA functionalization and used as a control. tPA activity on functionalized beads was measured using the tPA-specific fluorogenic peptide substrate Phe-Gly-Arg-AMC (Centerchem Inc. Norwalk CT) and compared to a standard curve of human tPA. These data indicated that the active tPA concentration of bead slurry was 39.7 nM, which is equivalent to 2.75 pg/nL. Thus, each 3 nL infusion had ~8.4 pg of tPA on ~150 microparticles (~0.056 pg tPA/bead).

Laser-mediated endothelial injury was performed on 3 dpf zebrafish larvae (21). Larvae were anesthetized in tricaine and mounted in 0.8% low melting point agarose on glass coverslips. The agarose around the head of the larvae was removed and replaced with water. A laser (MicroPoint Pulsed Laser System, Andor Technology, Belfast, Northern Ireland) was used to injure the endothelium of the posterior cardinal vein (PCV) 5 somites caudal to the anal pore using 99 pulses (36).

After the endothelial injury, 1 to 40 ng tPA or 3 nL (~8.4 pg total tPA) microparticles were infused via retro-orbital vessels using pulled capillary micropipettes (21) (Fig. 1 A and B). Standard tPA dosage was 0.9 ng/mg [6 dpf larvae have been shown to weigh ~1 mg (37), and 3 dpf is estimated to be 0.5 to 1 mg], corresponding to clinical dosages of 0.9 mg/kg (4, 38). Plasminogen [2 nL of 79 µg/mL solution (Prolytix, Essex Junction, VT)] was infused into plg mutants prior to laser injury and infusion of microparticles.

After introduction of the microparticles into the zebrafish larvae, µwheels were assembled using a rotating magnetic field generated with a custom apparatus (39). This apparatus consists of five coils arranged with two opposing pairs in the x and y plane and a fifth coil in the z-direction above the zebrafish sample (Fig. 1D). A set of sinusoidal currents are produced by an analog output card (NI-9263, National Instruments, Austin, TX), run through an amplifier (EP2000, Behringer, Willich, Germany), and sent to each pair of coils producing a rotating magnetic field. The circular, rotating magnetic field is positioned normal to the rolling surface, and the direction of the µwheels is manipulated by varying the ratio of current supplied to the individual coils (Fig. 1 E and F). The magnetic field strength and frequency are controlled with custom software, MuControl (40, 41). Previously, four different magnetic field modes were investigated for controlling µwheel swarms, including switchback, flipping, corkscrew, and rolling (17). Here, we used these modes to investigate their effect on the time to recanalization. Switchback mode results from switching the heading angle of the wheel back and forth which provides better movement over inclined surfaces (Movie S6). Flipping results from changing the camber angle back and forth and provides a way for wheels to prevent assembly into larger wheels (Movie S7). Corkscrew uses a synchronized change in heading angle and camber together that results in enhanced penetration (Movie S8). Rolling is our standard mode in which the heading is fixed manually to control directed movement through complex geometries (Movie S9) (17, 39).

Time to recanalization was observed optically for up to 30 min and defined by the resumption of passage of blood cells beyond the occlusive thrombus and into the distal PCV. Longer times were not feasible due to coil heat generation. All observations were collected by an observer blinded to condition or genotype.

We used a CRISPR/Cas9-mediated genome editing system to produce a 23 base pair deletion in the plasminogen (plg) gene, as previously described (42). Briefly, single guide RNAs (sgRNAs) were identified using ZiFiT Targeter (43), and sequence GGGAGTACTGCAATATTGAG in exon 12 was selected. DNA templates were produced using pDR274 (44), and sgRNAs transcribed. sgRNAs and Cas9 mRNA were co-injected into one-cell stage zebrafish embryos, at concentrations of 12.5 and 300 ng/µL, respectively. F0 offspring were raised to adulthood and crossed with wild-type zebrafish to verify F1 germline transmission (36). Heterozygotes were incrossed to produce homozygous mutants.

Genomic DNA of zebrafish larvae was isolated after incubation in lysis buffer (10 mM Tris-Cl, pH 8.0; 2 mM EDTA, 2% Triton X-100, and 100 µg/mL proteinase K) at 55 °C for 2 h followed by proteinase K inactivation at 95 °C for 5 min. Mutations were detected by PCR using primers 5′-AGATCTAAGGAGAAACCTGT-3′ and 5′-CTTTTTCTGAGGGAGCAGAT-3′.

Statistical analysis was performed with Mantel-Cox log-rank testing. Significance testing was performed using Prism (GraphPad software, CA).

M.H.H.P., C.-J.K., M.J.O., D.D., Y.L., M.W., D.A.L., D.W.M.M., K.B.N., and J.A.S. designed research; M.H.H.P., C.-J.K., Y.L., and M.W. performed research; M.H.H.P., C.-J.K., M.J.O., D.D., and M.W. analyzed data; and M.H.H.P., D.A.L., D.W.M.M., K.B.N., and J.A.S. wrote the paper.

J.A.S. has been a consultant for Sanofi, Takeda, Genentech, CSL Behring, and HEMA Biologics. D.W.M.M. and K.B.N. are co-inventors on patent US 10,722,250 B2.