Swarming magnetic photonic-crystal microrobots with on-the-fly visual pH detection and self-regulated drug delivery

Abstract

Swarming magnetic micro/nanorobots hold great promise for biomedical applications, but at present suffer from inferior capabilities to perceive and respond to chemical signals in local microenvironments. Here we demonstrate swarming magnetic photonic-crystal microrobots (PC-bots) capable of spontaneously performing on-the-fly visual pH detection and self-regulated drug delivery by perceiving local pH changes. The magnetic PC-bots consist of pH-responsive hydrogel microspheres with encapsulated one-dimensional periodic assemblies of Fe₃O₄ nanoparticles. By programming external rotating magnetic fields, they can self-organize into large swarms with much-enhanced collective velocity to actively find targets while shining bright “blinking” structural colors. When approaching the target with abnormal pH conditions (e.g., an ulcerated superficial tumor lesion), the PC-bots can visualize local pH changes on the fly via pH-responsive structural colors, and realize self-regulated release of the loaded drugs by recognizing local pH. This work facilitates the development of intelligent micro/nanorobots for active “motile-targeting” tumor diagnosis and treatment.

INTRODUCTION

After millions of years of evolution, many living organisms can change their colors, shapes, and other behaviors in response to environmental stimuli for camouflage, mating, hunting, and intimidating predators.^1–5 For example, chameleons, neon tetra fishes, and Morpho sulkowskyi butterflies display iridescent structural colors utilizing their surface natural periodic micro/nanostructures, known as photonic crystals (PCs), and some of them can further change their colors with environmental cues by tunning the spacing (lattice constant) of their natural PCs.^1,6–9 Inspired by this astonishing phenomenon, considerable effort has been devoted to creating synthetic responsive PCs by periodically embedding colloidal particles in a responsive polymer matrix or adopting responsive colloidal particles as building blocks.^10–12 In this way, a variety of PC-based chromic materials are available that show the structural colors in response to surrounding physicochemical conditions, such as temperature, humidity, chemicals, and so on; and they have important potential applications in camouflage, anti-counterfeiting, and sensing.^12–19 Nevertheless, the so-far developed responsive PCs usually do not involve free locomotion and on-the-fly sensing.

Motile micro/nanorobots (MNRs) are capable of propelling and navigating in various liquid media, and have been envisioned to provide revolutionary technological advances for drug delivery, microsurgeries, and micro/nanoengineering.^20–29 In terms of energy sources, MNRs can be categorized as biohybrid MNRs,^30–33 chemically powered MNRs,^34–36 and those powered by external fields, such as light, ultrasound, electric field, and magnetic field.^37–41 Among them, magnetically powered MNRs hold great promise for practical biomedical applications, such as single-cell surgery and drug (cell) delivery, due to their wireless fuel-free actuation, strong propulsion, precise motion control, and high biocompatibility.^42–47 Similar to swarming organisms in nature, such as flocking birds and shoaling fish, magnetic MNRs can also self-organize into coherent groups through local magnetic and hydrodynamic interactions, and exhibit intriguing emergent collective behaviors, such as enhanced driving forces, adaptive reconfigurations, and high imaging contrast.^48,49 Consequently, swarming magnetic MNRs can execute complex assignments that single ones cannot do.^50–53 For instance, the microswarms of magnetic iron oxide micro/nanoparticles show dynamic group reconfigurations (e.g., from chains, vortex to ribbons) upon adjusting the applied alternating magnetic field, and cooperatively transport heavy cargoes exceeding the capacity of single robots.^54,55 However, the present swarming magnetic MNRs lack the capabilities to perceive and respond to chemical signals in local microenvironments and are extremely challenging to perform autonomous theranostic tasks.

In this work, we demonstrate that swarming soft magnetic photonic-crystal microrobots (PC-bots) can spontaneously perform on-the-fly visual detection of local pH value and self-regulated drug delivery according to local microenvironmental pH. They consist of pH-responsive poly(acrylic acid-co-acrylamide) (poly(AA-co-AM)) hydrogel microspheres with encapsulated aligned one-dimensional (1D) periodic assemblies of Fe₃O₄ nanoparticles (NPs), and are fabricated by an emulsion droplet-templated-polymerization method assisted by a magnetic field. The magnetic PC-bots under a rotating magnetic field (B(t)) can move near a substrate in a rolling-while-slipping mode with controlled velocity and directions, and, further, self-organize into large swarms with bright optically traceable “blinking” structural colors and much-enhanced velocity. With pH responsiveness, the swarming PC-bots under magnetic actuation and navigation can perform both on-the-fly visual detection of local pH via structural color changes and self-regulated release of the loaded anticancer drugs after reaching targeted tumor sites. The PC-bots are promising for active “motile-targeting” diagnosis and treatment of ulcerated superficial tumor lesions, and the proof-in-concept of swarming PC-bots with environmental sensitivity can evoke a design strategy for intelligent MNRs.

RESULTS Conceptual design and functions

Tumor cells can stimulate significant molecular changes within their host tissues to support tumor growth and progression.⁵⁶ For example, during tumor development, insufficient oxygen supply, and building-up metabolic wastes often result in an acidic tumor microenvironment (pH 5.8–7.2).⁵⁷ Thus, the pH value has become an important indicator for tumor diagnosis and can be exploited to accomplish site-targeted anticancer-drug release.^58–60 Here, we propose to design swarming magnetic microrobots capable of perceiving and responding to local pH conditions. The individual magnetic PC-bot was designed with a pH-responsive poly(AA-co-AM) hydrogel microsphere with encapsulated aligned 1D periodic assemblies of Fe₃O₄ NPs. They are expected to carry integrated functions of magnetic propulsion coming from its magnetic property,^49,61 pH-responsive structural colors owing to its adjustable lattice constant (interparticle distance, d),¹⁷ and pH-triggered drug loading/releasing due to the pH-modulated interactions between the poly(AA-co-AM) scaffold and antitumor drugs (e.g., doxorubicin, DOX).⁶² With the integrated functions, a group of PC-bots under a rotating B(t) can collectively move toward a targeted tumor lesion (e.g., superficial lung, esophageal, gastric, or bladder tumor lesion^63–65) via hydrodynamic interactions, and, more importantly, spontaneously perform on-the-fly visual pH detection by employing their pH-responsive structural colors and self-regulated drug delivery utilizing their pH-dependent drug release (Figure 1 and Video S1). Thus, they exhibit promising applications in active “motile-targeting” tumor diagnosis and treatment.

FIGURE 1. Schematic illustration of swarming magnetic PC-bots and their potential applications. When driven and navigated by a rotating B(t), the PC-bots consisting of pH-responsive hydrogel microspheres with encapsulated 1D periodic assemblies of Fe3O4 NPs can self-organize into swarms and move toward a specific target (e.g., tumor lesion). If the microenvironmental pH near the target is different from the medium, they can perceive local pH changes, and spontaneously perform on-the-fly visual pH detection employing their pH-responsive structural colors and self-regulated drug delivery utilizing their pH-dependent drug release. An ulcerated superficial tumor lesion (e.g., superficial lung, esophageal, gastric, or bladder tumor lesion) is depicted as a typical potential application scenario.

Fabrication and characterization

The magnetic PC-bots were fabricated by an emulsion droplet-templated-polymerization method assisted by a magnetic field, as shown in Figure 2A. In the fabrication process, a water phase containing monodispersed Fe₃O₄@poly(vinyl pyrrolidone) (Fe₃O₄@PVP) NPs and gelation reagents was added into mineral oil with Span 80, and then emulsified by mechanical stirring to obtain emulsion droplets. When the droplets were stably sitting at the bottom of the beaker, an external static magnetic field (B) was applied using a permanent magnet to induce the one-dimensional arrangement of Fe₃O₄@PVP NPs. Upon applying a B with sufficient strength, the color of the droplets changed from dark brown (intrinsic color of Fe₃O₄@PVP NPs) to green or red (structural colors), reflecting the periodic arrangement of Fe₃O₄@PVP NPs into 1D periodic assemblies in the droplets. After the color appearance, immediate UV illumination was quickly applied to initiate the polymerization of the gelation reagents in the droplets. Finally, the PC-bots were obtained after the droplets were transformed into solid polymer beads, in which the 1D periodic assemblies of Fe₃O₄ NPs were fixed in the poly(AA-co-AM) hydrogel networks.

FIGURE 2. Preparation and characterization of magnetic PC-bots. (A) Schematic demonstration of the preparation of the magnetic pH-responsive PC-bots via an emulsion droplet-templated-polymerization method combined with the magnetic assembly of Fe3O4@PVP NPs. (B–D) SEM (B), TEM (C), and optical microscopic (D) images of the prepared magnetic PC-bots. (E) FT-IR spectrum of the magnetic PC-bots. (F) Structural color of the magnetic PC-bots under dark-field optical microscopy. (G) Hysteresis loops for the magnetic PC-bots at room temperature.

The scanning electron microscope (SEM) image of the fabricated PC-bots is shown in Figure 2B, indicating that they have a spherical morphology and an average size of 3.0 μm. In the PC-bots, the parallel 1D periodic assemblies of Fe₃O₄ NPs were clearly observed under transmission electron microscopy (TEM) observation (Figure 2C). The 1D periodic assemblies are also visible as dark lines and dots in transparent spheres under bright-field optical microscopy when they were reoriented parallel (left panel in Figure 2D) or perpendicular (right panel in Figure 2D) to observing plane by a static B, respectively. The optical microscopic images also indicate that after being suspended in water, the PC-bots grew to an average size of 8.0 μm due to their swelling deformation after water absorption (Figure 2D). The chemical composition of the PC-bots was confirmed by Fourier transform infrared spectroscopy (FT-IR), as depicted in Figure 2E. The characteristic absorption band at 3347 and 3191 cm⁻¹ corresponds to the N–H in AM, the peak at 1662 cm⁻¹ represents the carbonyl group (C=O) in AA and AM, the peaks at 1417 and 1319 cm⁻¹ are the characteristic peaks of PVP, and the dip at 566 cm⁻¹ comes from Fe–O in Fe₃O₄.

Due to the photonic band resulting from the periodic arrangement of Fe₃O₄ NPs, the 1D periodic assemblies in the PC-bots would exhibit a bright structural color with a characteristic wavelength ($\lambda$) determined by the Bragg's law,⁶⁶ [Image Omitted. See PDF]where $d$ is the average distance between the Fe₃O₄ NPs (i.e., lattice constant) in the 1D periodic assembly, $n_{\mathrm{b}}$ refers to the average refractive index of the assembly, and $\theta$ is the angle between the incident light and the direction perpendicular to the long axis of the assembly. As shown in Figure 2F, the 1D periodic assemblies light the PC-bots up with bright red structural color under dark-field optical microscopy, in which each bright red dot represents a separated 1D periodic assembly in the PC-bots. The high-strength hydrogel skeleton can hold 1D periodic assemblies in place and prevent possible deformation and structural color changes by the applied external B (Figure S1). By adjusting the strength of the applied B in the UV polymerization process and the cross-linker concentration, the PC-bots with different structural colors (Figure S2) and different cross-linking degrees (Figure S3) can be fabricated. The magnetic PC-bots are superparamagnetic and have a good magnetic response, as verified by their magnetic hysteresis loops (Figure 2G). The saturated mass magnetization reaches 10.8 emu g^–1, while the coercivity is 0 mT at room temperature. Thus, the PC-bots can be easily magnetized and demagnetized, facilitating delicate control by the external B(t).

Magnetic propulsions of single PC-bots

When the PC-bots were dispersed in water, they would gradually settle near the glass substrate because of gravity sedimentation. Once a rotating B(t) is applied,[Image Omitted. See PDF]in which B₀ is the amplitude of the magnetic field, f is the rotation frequency of the field, and t is the time,⁴⁹ the magnetic PC-bot is subjected to a rotational torque (T_M), and then rotates near the substrate (Figure 3A). Due to the different distances to the hydrodynamic no-slip boundary (substrate wall), the top and bottom of the rotating PC-bot have different hydrodynamic mobilities. As a result, the hydrodynamic symmetry of the rotating PC-bot is broken, and its out-of-plane rotation then translates into an in-plane translational motion near the substrate with a theoretical speed (v) of $2\mathit{\pi fr}$, where r is the radius of the PC-bot. The detailed mechanism of the PC-bot is shown in Figure 3A, depicting that a PC-bot rotating about the x-axis with the rotating B(t) would move in the y-direction near the substrate.

FIGURE 3. Magnetically driven propulsions and pH responses of single PC-bots. (A) Schematic illustration of the magnetically driven rolling-while-slipping locomotion of magnetic PC-bots under a rotating B(t). (B) Time-lapse microscopic images of a PC-bot moving in a pre-designed zigzag trajectory. (C, D) Velocity (v) of the PC-bots as a function of the amplitude (B0) (C) and frequency (f) (D) of the applied rotating B(t), respectively. The f in C is kept at 2 Hz, and the B0 in D remains 20 mT. Each error bar was obtained from the standard deviation (SD) of the v of 5 PC-bots. (E) Time-lapse dark-field optical microscopic images of a PC-bot moving in a zigzag trajectory. (F) Schematic demonstration of the structural color changes of the PC-bot in response to environmental pH variation. (G, H) Time-lapse dark-field microscopic images depicting in situ structural color changes of the PC-bots when the pH of the medium was adjusted from 7.8 to 3.6 (G) and from 3.6 to 7.8 (H), respectively.

The translational motion of a typical PC-bot is shown in Figure 3B and Video S2, indicating that the PC-bot moves in a straight-line trajectory with a v of 2.5 μm s^–1 under a rotating B(t) with a B₀ of 20 mT and an f of 2 Hz (0–13 s in Figure 3B). It is noted that the experimental speed v is much lower than the theoretical one ($v=2\mathit{\pi fr}$, 50.2 μm s^–1). This phenomenon can be rationalized by the fact that hydrodynamic lubrication exists between the PC-bot and the substrate due to the large gap between them determined by the thermal motions of the PC-bot and electrostatic repulsions between them.⁶⁷ Therefore, the PC-bot moves forward in a “rolling-while-slipping” mode. During the magnetic propulsion, the periodical change of the direction of 1D periodic assemblies in the PC-bot with the rotating B(t) can be clearly observed (Video S2), further confirming its rolling-based motion behavior. When the rotating B(t) ceased, the PC-bot stopped immediately (13 s in Figure 3B), suggesting the precise “stop-and-go” control of the PC-bot. By changing the orientation (i.e., rotating axis) of the applied rotating B(t), the motion direction of the PC-bot can be adjusted. Thus, the PC-bot can be navigated to move along a pre-designed trajectory precisely. As shown in Figure 3B, under magnetic navigation, the PC-bot can move in a zigzag trajectory by making a right turn at 13 s and then a left turn at 31 s.

Besides the magnetic navigable directions, the speed of the PC-bots can also be precisely controlled by adjusting the B₀ and f of the applied rotating B(t). By increasing the B₀ from 0 to 8 mT while keeping the f at 2 Hz, the v of the PC-bots increased rapidly because the increasing B₀ could enhance the exerted magnetic torque T_M on the PC-bots for them to overcome the viscous fluid resistance (Figure 3C). Then, as the viscous fluid resistance was completely conquered when B₀ was over 8 mT, the PC-bots moved in a stable v of about 2.5 μm s^–1 with the increasing B₀. When keeping the applied B₀ at 20 mT, the translational speed v of the PC-bots increased linearly with the f of the rotating B(t) until a critical threshold (step-out frequency) was reached (Figure 3D). The maximum v of the PC-bots was found to be 13.3 μm s^–1 at a step-out frequency of 8 Hz. As the input f continued to increase, the rotation of the PC-bots became asynchronous with the rotating B(t) due to the increasing resistance moment, and thus the v gradually decreased with the f.⁴⁷ The controllable motion in direction and speed lays the foundation for the PC-bots to operate in complex environments.

pH responses of single PC-bots

As the PC-bots can diffract structural color at a certain wavelength, they can be tracked under dark-field optical microscopy. As shown in Figure 3E and Video S3, the position of the magnetic PC-bot can be tracked in real time utilizing its red structural color in the dark field, and thus it can be navigated to move along a pre-designed zigzag trajectory under dark-field optical microscopy. When the pH of the surrounding environment changes, the structural color of the PC-bots is expected to change accordingly because of their pH responses (Figure 3F). Specifically, when the pH value in the surrounding environment increases, carboxyl groups in the poly(AA-co-AM) hydrogel scaffold of the PC-bots deprotonate into carboxylates, resulting in the swelling of the hydrogel scaffold due to the higher solubility of the latter than the former.¹⁷ This, in turn, makes the interparticle distance d, namely, the lattice constant of the encapsulated 1D periodic assemblies, become larger, leading to the redshift of the diffracted color of the PC-bots. On the other hand, when the surrounding pH value decreases, carboxylates in the poly(AA-co-AM) hydrogel scaffold protonate to carboxyl groups, leading to hydrogel shrinking and the blue shift of the diffracted color of the PC-bots. As demonstrated in Figure 3G and Video S4, when the surrounding pH decreases from 7.8 to 3.6, the r of the PC-bots decreases gradually from 5.7 to 3.9 μm, and their structural color changes from red to blue accordingly. When the surrounding pH is adjusted to 7.8 again, the structural color of the PC-bots changes from blue to red again as they gradually swell (Figure 3H and Video S4), revealing their excellent reversibility in the pH response. More interestingly, the as-prepared PC-bots have a fast response to environmental pH changes, and their structural color change can be completed within 3 s (Figure 3G,H), facilitating their applications in fast pH microsensing.

Magnetic propulsions of swarming PC-bots

As local number density increases, intriguing collective behaviors emerge among the magnetic PC-bots (Figure 4A and Video S5). When driven by the rotating B(t), the gathered PC-bots rotate parallel to the floor (glass substrate surface) and produce a flow field around themselves when propagating forward. Due to local hydrodynamic interactions, the propagating front quickly becomes unstable in the direction transverse to propagation, leading to the appearance of density fluctuations (0–8 s in Figure 4A).⁶¹ Then, the unstable front with density fluctuations quickly splits into fingers, and each fingertip forms a swarm cohesively moving forward (21–36 s in Figure 4A). The formation of the swarms originates from the lateral hydrodynamic attraction and repulsion in the x–y plane. Under dark-field optical microscopy, the PC-bot swarm exhibits a “blinking” structural color when moving forward because of the periodic changes in the orientation of the encapsulated 1D periodic assemblies in individual rolling PC-bots (Figure 4B and Video S6).

FIGURE 4. Magnetically driven propulsions and pH responses of swarming PC-bots. (A) Time-lapse microscopic images showing the formation and collective motion of the PC-bot swarms. (B) The periodic “blinking” structural color of a typical PC-bot swarm within one rotation cycle (T) under dark-field microscopy when swarming leftward. (C) The velocity (U) of PC-bot swarms as a function of the area fraction (φ) under a rotating B(t) with a B0 of 20 mT and an f of 2 Hz. (D) The U of PC-bot swarms at different f of a rotating B(t) with a B0 of 20 mT. (E,F) Dark-field microscopic images (E) and corresponding reflection spectra (F) of the swarming PC-bots in the medium with different pH. (G) Schematic illustration (upper panel) and time-lapse microscopic images (lower panels) of the on-the-fly visual pH sensing of a PC-bot swarm when collectively moving from a pH 7.4 microwell toward a targeted capillary filled with pH 4.4 PBS buffer.

One striking collective effect of the swarming PC-bots is their density-dependent collective velocity (U). As shown in Figure 4C, the U of the swarming PC-bots increases with the increasing area fraction φ at fixed B₀ (20 mT) and f (2 Hz). The linear fitting revealed that the collective velocity U changed with φ in a function of U = αφ, where α was found to be 78 ± 4. Later, it was found that when the φ remained the same, the U of the swarming PC-bots increased with the f of the applied rotating B(t) until a step-out frequency of 10 Hz was reached (Figure 4D). Different from single PC-bots, the U of swarming PC-bots did not decrease but remained stable as the input f continued to increase to 15 Hz. In addition, the maximum U of the swarming PC-bots was found to be 180.0 μm s^–1, and this value was over 13 times higher than that of single PC-bots (13.3 μm s^–1), revealing the strong collective effect of the swarming PC-bots under the hydrodynamic interaction.

On-the-fly visual pH detection

Similar to a single PC-bot, the swarming PC-bots also show a pH-responsive structural color but with enhanced structural color brightness, expanded detecting area, and emergent pH-mapping performance. From the dark-field microscopy images in Figure 4E, the swarming PC-bots display a bright blue color in phosphate buffer solutions (PBS) with pH 3.6 when they were reoriented perpendicular to the imaging plane by a static B. With the increasing pH from 3.6 to 7.8, the diffracted color of the swarming PC-bots experiences a redshift, and finally turns to dark red at pH 7.8. The corresponding reflection spectra of the PC-bots at different pH are shown in Figure 4F. It can be seen that the PC-bots shift the diffraction peak (λ) up to 215 nm between pH 3.6 and 7.8. The sharp shift of the λ was observed between pH 4.0 and 4.8 because the ionization constant (pK_a) of AA is ~4.3. The cycling stability test demonstrated that the PC-bots could repeatedly and reversibly change their structural colors as the pH value varies. The values of λ at pH 7.8 and 3.6 remained stable with a deviation of <20 nm over 10 cycles (Figure S4). Moreover, even after 1 month, the PC-bots maintained their cycling stability in the pH range (Figure S4), indicating their robust and reversible pH-detection performance.

The shifting range in structural colors (or λ) of the PC-bots can be controlled by adjusting the cross-linking degrees and monomer ratios. As shown in Figure S5A, as the cross-linking degree increases from 0.5% to 1.5%, the λ-shifting range decreases from 215 to 110 nm when the pH changes from 3.6 to 7.8. This is because the increase in cross-linking degree makes the PC-bots stiffer and less sensitive to pH changes. When the monomer molecular ratio of AA to AM decreased from 7:3 to 3:7, the λ-shifting range decreased accordingly from 208 to 81 nm. This can be attributed to the reduced number of carboxyl groups in the hydrogel scaffold, which results in a decrease in pH-responsive volume change and subsequent adjustability of lattice spacing d in the PC-bots (Figure S5B). Further, due to the high structural color brightness, the pH in the surrounding environment is expected to be read by the naked eye (Figure S6).

With the combined collective motions and pH-responsive structural colors, the swarming PC-bots can perform motile on-the-fly visual pH detection. The motile visual pH detection was tested in a microfluidic chip with two microwells connected by a narrow channel and filled with pH 7.4 PBS buffer. The PC-bots were added to the left microwell, and a capillary filled with pH 4.4 PBS buffer was immersed in the right microwell (upper panel in Figure 4G). As shown in Figure 4G and Video S7, when driven and navigated by the rotating B(t), the swarming PC-bots can collectively move toward the capillary filled with pH 4.4 PBS buffer (0–12 s in Figure 4G). When they approach the region with a distance of ~800 μm to the capillary tip, the red PC-bots in the swarm front start to turn green (12 s in Figure 4G), and almost all PC-bots in the swarm turn green when they arrive at the region close to (~150 μm) the tip of the capillary (17–19 s in Figure 4G). In contrast, if the swarming PC-bots collectively move from a pH 4.4 microenvironment to a target region with pH 7.4, they gradually change their structural color from green to red (Figure S7 and Video S8). Meanwhile, when the swarming PC-bots approach the targeted region, a rainbow stripe can be observed in the swarm corresponding to the local pH gradient, suggesting their potential in motile pH mapping of microenvironments.

Self-regulated drug delivery

Besides the on-the-fly visual pH detection, the swarming PC-bots can also perform “motile-targeting” drug delivery (Figure 5). In pH 7.4 PBS buffer, the carboxyl groups on the PC-bot's hydrogel scaffold deprotonate to carboxylate anions. As the DOX mainly exists in a protonated cation form, the PC-bots can effectively anchor DOX molecules by employing electrostatic interactions between them, as verified by the bright-field and fluorescent microscopic images of the DOX-loaded PC-bots (left two panels in Figure 5A). The drug loading capacity of the PC-bots was confirmed to be up to 180 mg g^–1.

FIGURE 5. Self-regulated drug delivery of swarming PC-bots. (A) Schematic illustration of the tumor-targeted drug delivery of the swarming PC-bots in a microfluidic channel mimicking tumor microenvironment. The two left insets show bright-field and fluorescent microscopic images of the PC-bots loading with antitumor drug DOX. (B) Time-lapse microscopic images illustrating a swarm of PC-bots delivers DOX by traveling a long distance (10 mm) from one end of the microfluidic channel to targeted tumor sites (MCF-7 cells) at the other end. (C) Schematic of the pH-responsive DOX release from the PC-bot due to the pH-triggered volume change and the pH-responsive electrostatic interactions between DOX and the hydrogel scaffold. (D) Cumulative DOX release from PC-bots in the PBS buffer at different pH: (i) pH 7.4, (ii) pH 6.5, and (iii) pH 5.0. (E) MCF-7 cell viability after incubation for 24 h with bare PC-bots, DOX-loaded PC-bots at pH 7.4, DOX-loaded PC-bots at pH 6.5, and free DOX at different concentrations. Data are represented as mean ± SD. (F) Fluorescence images of MCF-7 cells after being incubated with bare PC-bots and DOX-loaded PC-bots for 24 h. The cells were stained by Calcein-AM/PI, green for alive cells and red for dead cells.

Utilizing the collective motion of the swarming PC-bots, they are expected to deliver the loaded drug to a specific tumor site in a “motile-targeting” behavior (in contrast to the passive and active targeting of immotile nanomedicines^68,69). To test the “motile-targeting” drug delivery of the swarming PC-bots, a tumor model based on microfluidic channels was used, in which MCF-7 cells embedded in hydrogel were placed in the right microwell (right panel in Figure 5A). When actuated by the rotating B(t), the swarming PC-bots loaded with DOX depart from the left microwell and pass through a narrow channel to the tumor-cell site, as shown in Figure 5B and Video S9. It is clearly demonstrated that the swarming PC-bots can travel a very long distance (over 10 mm) through the connecting microchannel to deliver the loaded drug.

In consideration of the pH response of the PC-bots and the acidic pH in the tumor microenvironment,⁵⁸ the swarming PC-bots are expected to spontaneously release the loaded antitumor drugs once they arrive at the tumor site, which is highly desirable for targeted tumor therapy. Specifically, as shown in Figure 5C, the PC-bots are in a swelling state at pH 7.4, and DOX molecules are anchored electrostatically to carboxylate sites of the hydrogel scaffold, making the loaded DOX molecules difficult to be liberated from the PC-bots. In a more acidic environment (pH 6.5 and 5.0), the electrostatically anchored DOX molecules can be released due to the deswelling of the PC-bots and the reduced electrostatic interactions after the protonation of carboxylate sites (Figure 5C). To evaluate their pH-dependent drug release behaviors, we monitored the DOX releasing dynamics from the PC-bots in PBS buffer at different pH (Figure 5D). At pH 7.4, the PC-bots released 14.2% of the loaded DOX in the first 4 h, and then the residual DOX was released at an extremely low rate ever after. In contrast, the DOX release from the PC-bots could reach 21.7% and 53.3% within 4 h at pH 6.5 and 5.0, respectively, revealing a high DOX releasing rate in the acidic environment. Therefore, the PC-bots can act as smart carriers for releasing the antitumor drug in a self-regulated manner according to local pH, without any external intervention.⁷⁰

Benefitting from “motile-targeting” drug delivery and self-regulated drug release in response to local pH, the swarming PC-bots are promising for tumor therapy. The therapeutic effect of the PC-bots on cancer cells was examined in vitro by a standard cell counting kit-8 (CCK-8) assay (Figure 5E). In the assay, MCF-7 cells were respectively exposed to bare PC-bots, DOX-loaded PC-bots at pH 7.4, DOX-loaded PC-bots at pH 6.5, and free DOX for 24 h. In these experiments, the concentrations of bare PC-bots were set at the same levels as the DOX-loaded PC-bots, and the dose of DOX was controlled by adjusting the concentration of the DOX-loaded PC-bots or free DOX in the medium.

Due to their biocompatible chemical compositions and soft hydrogel surface, the bare PC-bots show negligible cytotoxicity to MCF-7 cells across various concentrations (Figure 5E). In addition, they are also expected to cause minimal damage to normal cells, as evidenced by their low hemolysis rate (≤1%) even at a high concentration of 12.5 mg mL^–1 (Figure S8). In stark contrast, the cytotoxicity of free DOX and DOX-loaded PC-bots to MCF-7 cells increased with the increasing dose of the (loaded) DOX. At the same time, the anticancer efficacy of the DOX-loaded PC-bots is also strongly dependent on pH. The viability of MCF-7 cells exposed to the PC-bots with DOX decreases from 38.3% to 18.4% when the pH of the medium decreases from 7.4 to 6.5 because of the higher DOX release rate at lower pH. The acidic pH alone had a negligible effect on our experimental outcomes, as indicated by the high cell viability of 93.0% after incubation for 24 h at pH 6.5 (Figure S9). The therapeutic effect of the DOX-loaded PC-bots on cancer cells was further confirmed by fluorescence microscopy. As shown in Figure 5F, dead MCF-7 cells, which were stained with propidium iodide (PI), can hardly be found in the bare PC-bots group, while most cells are dead in the DOX-loaded PC-bots group (pH 6.5) due to the cytotoxicity of the released DOX. These results indicate that the developed PC-bots can not only deliver drugs to a targeted tumor site by their swarming motions, but also realize smart drug release in response to the acidic tumor microenvironments to kill tumor cells, showing great potential in “motile-targeting” tumor therapies.

DISCUSSION AND CONCLUSION

Swarming MNRs can execute complex assignments exceeding the capacity of single ones, such as self-adaptable motions through narrow cavities, cooperative large-cargo transport, and collective large-territory covering or mapping, because of their intriguing collective behaviors.^71–73 However, the developed swarming magnetic MNRs usually lack the capabilities to perceive and respond to chemical signals in local microenvironments, and are extremely challenging to perform autonomous theranostic tasks, especially when confronting unknown environments.⁴⁸ Here, inspired by chameleons, we propose swarming magnetic PC-bots by introducing magnetic photonic crystals into pH-responsive hydrogel materials. With the combined magnetic propulsion, responsive structural colors, and pH-dependent drug loading/releasing, the developed PC-bots could serve as an intelligent theranostic platform to spontaneously perform on-the-fly visual pH detection and self-regulated drug delivery by recognizing local pH conditions. Compared to the reported magnetic MNR microswarms,^43,48 the developed swarming PC-bots show two apparent characteristics. One lies in the unique periodic arrangement of encapsulated magnetic NPs, which provide the PC-bots with a photonic band. The other characteristic is that the PC-bots can perceive and respond to chemical signals in local microenvironments coming from the responsive hydrogel body.

The proposed swarming magnetic PC-bots could translate local microenvironmental pH into visual structural colors on the fly. Specifically, due to the unique structure of a pH-responsive hydrogel microsphere with encapsulated 1D periodic assemblies of Fe₃O₄ NPs, the PC-bots can diffract bright “blinking” structural colors when swarming under a rotating B(t), and further change their structural colors once perceiving pH changes, visualizing the local microenvironmental pH changes on the fly. In contrast to immotile sensors based on responsive photonic crystals,¹² the developed PC-bots show additional functions of active remote targetability, rapid deployment, and cross-region on-the-fly sensing. Compared to single sensing MNRs loaded with transducers of fluorophores, quantum dots, noble metal NPs, or catalysts,^24,25,74 the swarming PC-bots can rapidly cover a targeted territory, greatly eliminate the errors caused by individual differences, and dynamically monitor local pH changes with high reversibility and easy read-out (high imaging contrast). With the responsive visual structural colors, it is rational to envision that the swarming PC-bots can be controlled in a feedback manner in biomedical application scenarios. For instance, the swarming PC-bots can be navigated to explore an unknown region at certain B(t) parameters (input), and when reaching targeted microenvironments (e.g., tumor and inflammatory sites with low pH), they can report abnormal pH conditions there via visual structural-color changes (i.e., visual feedback). This visual feedback can be further used to guide the external B(t) controller to adjust magnetic propulsions of the swarming PC-bots, such as stop-and-go control, velocity adjustment, and rolling-to-rotating mode transition, realizing operation switching (e.g., from wide-range exploration to disease treatment on site). The main limitation of the on-the-fly visual pH detection by the PC-bots was that the reflective optical signal (structural color) from them is in the visible-light range and has a limited penetration depth in biological tissues. Thus, they can only be used in in-vitro microfluidic analytical devices and in-vivo transparent (e.g., eyeballs) or endoscopy-reachable organs (e.g., bladder and gastrointestinal tract).^33,75,76 To enable on-the-fly pH detection in deep tissues, future investigations may be required to broaden the Bragg diffraction of the PC-bots to the infrared light range¹⁸ and to explore their possible pH-responsive signals from ultrasound imaging, computed tomography, and magnetic resonance imaging.^59,77

The proposed swarming magnetic PC-bots could deliver anticancer drugs to targeted tumor sites in a “motile-targeting” manner and achieve self-regulated drug release by recognizing local pH. The pH-responsive PC-bots can load anticancer drug DOX with a high capacity of 180 mg g^–1, mainly through electrostatic interactions between the protonated DOX molecules and deprotonated carboxylate anions in the hydrogel scaffold. Under a rotating B(t), the PC-bots can rapidly transport a large amount of DOX to targeted tumor cells by swarming motions, and then release the drug at a different rate depending on local microenvironmental pH. Passive drug nanocarriers and single MNRs generally suffer from low drug-loading capacity and long accumulation time at target sites.^22,69 In this study, the swarming PC-bots in cohesive groups can simultaneously deliver a large dose of anticancer drugs within a short period of time to a targeted site. In addition, the stimuli-responsive hydrogel body not only largely preserves drugs before reaching the target site, but also endows the PC-bots with the intelligence to autonomously regulate drug release rate according to local pH conditions. This intelligence is of great importance for those application scenarios where the target site is unknown and dynamically changing. Nonetheless, the “motile-targeting” self-regulated drug delivery of the PC-bots is required to be further verified by in-vivo animal experiments.⁷⁸ Furthermore, different responsive hydrogels may be selected to design the PC-bots, facilitating them to act as a general motile self-regulated drug delivery platform for treating different diseases with other endogenous stimuli, such as glucose level, temperature, and redox potential.^79,80

In summary, we have developed swarming magnetic PC-bots capable of spontaneously performing on-the-fly visual pH detection and self-regulated drug delivery by perceiving local pH signals. By employing an emulsion droplet-templated-polymerization method combined with the magnetic assembly of Fe₃O₄@PVP NPs, the magnetic PC-bots consisting of pH-responsive poly(AA-co-AM) hydrogel microspheres with encapsulated 1D periodic assemblies of Fe₃O₄ NPs are successfully fabricated. When actuated by a rotating B(t), the magnetic PC-bots can move in a rolling-while-slipping mode and further self-organize into large swarms with a maximum collective velocity U of 180.0 μm s^–1. By adjusting B₀, f, and the orientations (i.e., rotating axis) of the applied rotating B(t), the velocity and motion direction of the PC-bots can be precisely controlled, respectively, and thus they can be navigated to move along a predetermined trajectory with tunable velocity. In addition, they show bright optically traceable “blinking” structural colors from the encapsulated 1D periodic assemblies when swarming. When surrounding pH changes, they show a rapid reversible response in volume changes and lattice-constant shift based on the protonation and deprotonation of the poly(AA-co-AM) hydrogel scaffold of the PC-bots. Thus, when reaching a microenvironment with different pH values (e.g., acidic tumor microenvironments), they can spontaneously perform on-the-fly visual pH detection by diffracting different structural colors and also self-regulated drug delivery utilizing their pH-dependent drug release. The integrated functions of controllable collective motions, responsive structural colors, and self-regulated drug delivery enable them to act as a promising “motile-targeting” theranostic platform for different diseases, such as cancer and inflammatory diseases. The proposed concept of endowing swarming PC-bots with the capability of perceiving surrounding chemical signals may evoke novel design strategies for intelligent and multifunctional MNRs.

EXPERIMENTAL SECTION Materials

All the chemicals used in this work were of analytical grade and were used as received without further purification. Acrylic acid (AA), methacrylic acid (MAA), acrylamide (AM), bis-acrylamide (BIS), 2-hydroxy-2-methylpropiophenone (HMPP), mineral oil, doxorubicin hydrochloride (DOX·HCl), and hexane were purchased from Aladdin and were used as received. Span 80 and ethanol were procured from Sinopharm Chemical Reagent Co., Ltd., China. The other chemicals and solvents without denotation were of analytical grade. Fetal bovine serum (FBS), DMEM high glucose medium, DMEM/F12, penicillin and streptomycin, Trypsin (0.25%), and PBS buffer were obtained from Hyclone.

Fabrication of PC-bots

The magnetic PC-bots were fabricated by an emulsion droplet-templated-polymerization method assisted by a magnetic field (Figure 2A). At first, Fe₃O₄@PVP NPs with a diameter of around 150 nm were synthesized according to the previously reported method.⁸¹ Then, 30 mg Fe₃O₄@PVP NPs were dispersed in a 2 mL aqueous solution with 0.432 g AA, 0.995 g AM monomers, 0.015 g cross-linker BIS, and 0.100 g photoinitiator HMPP. The aqueous mixture was vortexed for 15 min to form a homogeneous brown precursor solution (water phase). Next, 0.5 mL of the aqueous mixture was added to the 30 mL oil phase (the mixture of span 80 and mineral oil at a volume ratio of 1:50) and emulsified at a mechanical stirring speed of 500 rpm. After 15 min, 5 mL emulsion was transferred into a glass beaker (outer diameter 50 mm, height 72 mm) and placed in a static magnetic field (20 mT) for 1 min to induce the magnetic assembly of Fe₃O₄@PVP NPs in the emulsion droplets. Finally, the PC-bots were prepared by exposing the emulsion to a UV light source (Black-Ray B-100AP, China) for 4 min to trigger the polymerization of the emulsion droplets. The final product of the PC-bots was separated from the oil phase by magnetic separation and then washed with hexane and ethanol three times, respectively.

Characterization

SEM images were obtained using a Hitachi S-4800 field-emission SEM (Japan). TEM image was captured on a JEM-2100F instrument (JEOL, Japan) at an accelerating voltage of 200 kV. The FT-IR spectrum was obtained by using a 60-SXB FTIR spectrometer. The bright-field and dark-field optical microscopic images of the PC-bots were recorded with an optical microscope (Leica DMI3000M, Germany). The reflection spectra of the PC-bots were captured using a fiber-optic spectrometer (Ocean Optics USB2000+, UK).

Magnetically driven propulsions

A customized three-axis Helmholtz electromagnetic coil setup fixed on an inverted optical microscope was used to power the PC-bots. The electromagnetic coil setup was composed of three major parts, including electric current supplies (ATA-309 Power Amplifiers, China), a signal source (NI USB-6343, USA), and three pairs of electromagnetic coils generating rotating B(t) (Figure S10). The PC-bots dispersed in water or PBS buffer (1 mg mL^–1) were placed on the platform of the microscope for subsequent magnetic propulsion. The magnetic propulsion was observed and recorded through the optical microscope under bright-field and dark-field illumination. To investigate collective velocity of swarming PC-bots depending on the number density, the area fraction of the PC-bots in the field of view was adjusted by varying their concentration in the medium, and further determined using ImageJ software. All videos were analyzed using Video Spot Tracker V08.01 software.

pH detection

For the pH detection at static conditions, the optical microscope and the fiber-optic spectrometer were used to capture the structural colors and reflection spectra of PC-bots suspended in PBS buffers (1 mg mL^–1) with different pH from 3.6 to 7.8, after being reoriented to a θ of 90° by a static B. To test cycling stability, the reflection spectra of the PC-bots prepared on the first day and those after storage for 1 month were recorded when the pH of the medium was repeatedly adjusted to 3.6 and 7.8 using PBS buffers. To test on-the-fly visual pH detection, a capillary loaded with different pH buffers (pH = 4.4 or 7.4) was used as the test target. The PC-bots suspended in 10 μL aqueous suspension (1 mg mL^–1) were added to the left well (10 mm in diameter) of a microfluidic channel previously filled with water. Then, the PC-bots were driven by the rotating B(t) to reach the target placed in the right well after passing through a narrow channel (10 mm in length and 500 μm in width). The structural-color changes of the PC-bots during the targeted on-the-fly visual pH detection were recorded under dark field microscopy.

DOX loading and release

To load DOX, 10 mg PC-bots were mixed with 2 mL DOX aqueous solution (1 mg mL^–1). After shaking at 200 rpm for 24 h under dark conditions, the DOX-loaded PC-bots were collected by magnetic separation. To evaluate the DOX loading capacity, the supernatant was collected and the residual DOX was measured by UV/Vis spectroscopy at 480 nm (Shimadzu UV-2550, Japan). To investigate pH-responsive DOX-releasing behaviors, 10 mg DOX-loaded PC-bots were immersed in 2 mL PBS buffer with different pH values (pH 7.4, 6.5, and 5.0) and incubated at 37 °C. At regular time intervals, 1 mL PBS supernatant was taken from the mixture after centrifugation, and the concentration of the released DOX was detected by UV/Vis absorption spectra, and then fresh PBS (1 mL) was added for the continued drug release.

Antitumor assay of DOX-loaded PC-bots

MCF-7 cells (purchased from China Center for Type Culture Collection) were cultured in DMEM high glucose medium containing 10% of FBS, supplemented with 100 Um L^–1 penicillin and 100 U/mL streptomycin at 37 °C with 5% CO₂. Then, MCF-7 cells were seeded into a 96-well plate at a density of 5000 cells per well and cultured in 5% CO₂ at 37 °C for 24 h. Then, DOX-loaded PC-bots were added to the culture medium with a pH of 7.4 or 6.5, and incubated with the cells in 5% CO₂ at 37 °C for another 24 h, respectively. The concentrations of DOX (C_DOX) were controlled to be 3.125, 6.25, 12.5, 25, 50, and 100 μg mL^–1 by adjusting that of the DOX-loaded PC-bots (C_bot) in the medium, respectively. As control experiments, free DOX and bare PC-bots at the same C_DOX and C_bot levels were added to the culture medium and incubated with the cells, respectively. After 24 h of treatment, we used CCK-8 for the detection. After removal of the culture medium, the well was washed with pH 7.4 PBS buffer three times. Then, each well was incubated in complete DMEM with 10% v/v CCK-8 for 2 h at 37 °C. Afterward, we measured absorbance by a microplate reader (Multiskan GO, Thermo Scientific, USA) at 450 nm. Cell viability (%) was calculated as follows,[Image Omitted. See PDF]where ${\mathrm{OD}}_{\text{sample}}$ and ${\mathrm{OD}}_{\text{control}}$ denote absorbance values of the sample and control wells, respectively.

To explore the viability of the MCF-7 cells incubated with bare PC-bots and DOX-loaded PC-bots, live/dead staining was used. MCF-7 cells were seeded in a 24-well plate at a density of 8000 cells per well and cultured in 5% CO₂ at 37 °C for 24 h. Then, bare PC-bots and DOX-loaded PC-bots were added to the medium with pH 7.4 or 6.5, respectively, and incubated with the cells in 5% CO₂ at 37 °C for 24 h. After that, the medium was removed and washed with PBS three times, the cells were stained with 4.5 μM propidium iodide (PI) (dead cells were labeled red) and 2 μM Calcein-AM (live cells were labeled green) for 30 min at 37 °C, and subsequently analyzed by fluorescence microscopy (Olympus IX71, Japan).

Hemolysis assay

The EDTA-anticoagulated rabbit blood was purchased from Wuhan Chundu Biotechnology Co., Ltd., China. At first, the blood sample was added to 5 mL of PBS, and centrifuged to separate the red blood cells. After washing the red blood cells three times with 5 mL of PBS solution, the purified cells were diluted with the PBS solution to a volume of 1/10. Then, 0.5 mL of the diluted red blood cell suspension was mixed with PBS (as negative control), D.I. water (as positive control), and PC-bots suspension. All mixtures were vortexed and stored at room temperature for 3 h. Finally, the absorbance of the supernatant was measured at 540 nm. The hemolysis rate of the red blood cells was calculated as follows: Hemolysis rate = [(sample absorbance − negative control absorbance)/(positive control absorbance − negative control absorbance)] × 100%.

ACKNOWLEDGMENTS

This work was supported by the National key research and development program (Nos. 2021YFA1201400 and 2022YFB4701700), National Natural Science Foundation of China (Nos. 52073222 and 21875175, 52175009), Interdisciplinary Research Foundation of HIT (IR20211219), Natural Science Foundation of Chongqing (CSTB2022NSCQ-MSX0507), Natural Science Foundation of Heilongjian Province (YQ2022E022), and Fundamental Research Funds for Central Universities (2022IVA201).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.