Multimodal microwheel swarms for targeting in three-dimensional networks

www.nature.com/scientificreports OPEN Multimodal microwheel swarms for targeting in three‑dimensional networks C. J. Zimmermann1, P. S. Herson2, K. B. Neeves3,4 & D. W. M. Marr1* Microscale bots intended for targeted drug delivery must move through three-dimensional (3D) environments that include bifurcations, inclined surfaces, and curvature. In previous studies, we have shown that magnetically actuated colloidal microwheels (µwheels) reversibly assembled from superparamagnetic beads can translate rapidly and be readily directed. Here we show that, at high concentrations, µwheels assemble into swarms that, depending on applied magnetic field actuation patterns, can be designed to transport cargo, climb steep inclines, spread over large areas, or provide mechanical action. We test the ability of these multimodal swarms to navigate through complex, inclined microenvironments by characterizing the translation and dispersion of individual µwheels and swarms of µwheels on steeply inclined and flat surfaces. Swarms are then studied within branching 3D vascular models with multiple turns where good targeting efficiencies are achieved over centimeter length scales. With this approach, we present a readily reconfigurable swarm platform capable of navigating through 3D microenvironments. Actively manipulated microbots present a promising platform for targeted delivery of therapeutic drugs1,2 by swimming through bulk fluid3–6 or by utilizing nearby surfaces to roll7–10 or walk11. Using applied magnetic fields, individual microbots, proposed for applications including microsurgery12, biofilm eradication13, blood clot removal14, and stem cell t ransplantation15 with structures incorporating h elical16 or flexible c omponents17, 18,19 20 can travel against fluid flow or at speeds up to 600 µm/s in quiescent fluid. Though individual microbot translation can be accurately m odeled21, applications involving therapeutic payloads will require significant microbot numbers and concentrations where swarming behaviors, such as those demonstrated in nature with insects, birds, and fish, have been observed. Such emergent structures include vortices22–24, ribbons25, carpets26, chains27, or d ispersions28 composed of many individual microbots. In addition, swarms can be tuned to change modes to increase hyperthermia29, travel in confined s paces22, or increase translation in various bio-fluids30. While precise microstructures can be f abricated31,32 with good translational c ontrol33, microbots can be difficult to manufacture in bulk in the numbers required for therapeutic applications. Our previous work has focused on wheel-like microstructures (µwheels) that are reversibly and readily assembled in situ from superparamagnetic beads using a weak rotating magnetic field (Fig. 1). Before assembly, these individual particle building blocks are small enough to pass through the smallest capillaries in the body and, when assembled into µwheels, can translate at velocities over 200 µm/s9 on surfaces normal to gravity. For in vivo drug delivery however, µwheels will move as swarms (Fig. 1). Others have shown microbot swarms with multiple modes in 2D22,27, here the contribution is microbot swarm targeting in 3D. During treatment, µwheel swarms may traverse environments such as the circulatory, digestive, or urinary systems that are curved, not normal to gravity, and contain tortuous pathways. An effective platform must therefore be able to navigate highly-branching and inclined systems. To investigate these, we first characterize the behavior of component µwheels in 3D and develop strategies for swarm movement that enable faster translation, better climbing, wider spread, and mechanical action. Then, we investigate the targeting efficiency of µwheel swarms in a model 3D network inspired by the cerebrovasculature. Together, this work presents a complete approach for quickly assembling superparamagnetic beads in situ into concentrated yet highly efficient multimodal µwheel swarms that can adapt to their environment and target across centimeter length scales. With this, we present a microbot-based approach that is not limited to 2D environments and can effectively target within 3D vascular analogues. 1 Department of Chemical and Biological Engineering, Colorado School of Mines, Golden, CO, USA. 2Department of Anesthesiology, University of Colorado Denver, Anschutz Medical Campus, Aurora, CO, USA. 3Department of Bioengineering, University of Colorado Denver, Anschutz Medical Campus, Aurora, CO, USA. 4Department of Pediatrics, University of Colorado Denver, Anschutz Medical Campus, Aurora, CO, USA. *email: Scientific Reports | (2022) 12:5078 | https://doi.org/10.1038/s41598-022-09177-x 1 Vol.:(0123456789) www.nature.com/scientificreports/ Figure 1.  Upon application of a rotating magnetic field (a) individual 4.5 µm beads form into (b) µwheels which subsequently form into (c) swarms. Inset scale = 50 µm. Swarm scale = 300 µm. Figure 2.  µWheel translation on inclines. (a) µWheel angular velocity (ω) as a function of size (R) and incline angle (φ). Dotted line shows the ω ∝ 1/R scaling. (inset) Translating µwheel on an incline. (b) µWheel velocity over incline angles 0–80° with solid lines the variable gap width model (Supplementary Equation 1). All µwheels were propelled with a constant 40 Hz circular rotating field of magnitude 3.7 mT and 30° camber angle (θ). Results µWheel translation. Essential for predicting movement in realistic geometries, we begin by describing the behavior of individual µwheels on inclined surfaces where, upon application of a rotating weak magnetic field (~ 4 mT), µwheels assemble from 4.5 µm Dynabeads® into spinning clusters. While other superparamagnetic beads could be used, these highly-monodisperse particles consist of iron oxide domains within a polystyrene matrix, a relatively biocompatible material available at sizes that can be readily phagocytosed upon µwheel disassembly34,35. In addition, their surfaces can be easily functionalized to create drug delivery vehicles as previously demonstrated14. When oriented with a component normal to the surface, µwheels roll at velocities which depend not only on the µwheel rotation rate, but also on the size (Fig. 2a) and the camber, or tilt, angle θ of the µwheel relative to the surface normal. For this study we hold θ constant, focusing on the size and incline dependScientific Reports | Vol:.(1234567890) (2022) 12:5078 | https://doi.org/10.1038/s41598-022-09177-x 2 www.nature.com/scientificreports/ ence of µwheel velocity. Unlike macroscopic wheels which move by gripping a solid surface, µwheels roll on an intervening layer of fluid and use wet friction to move. Their translational velocity can be predicted by balancing translational fluid drag and wet friction with the surface9. However, for translation up inclined surfaces the normal force, the µwheel distance from the surface, and the resulting frictional fo (...truncated)