Here, we report non-invasive strategy for isolating cancer cells by autonomously propelled carbon nanotube (CNT) microrockets. H₂O₂-driven oxygen (O₂) bubble-propelled microrockets were synthesized using CNT and Fe₃O₄ nanoparticles in the inner surface and covalently conjugating transferrin on the outer surface. Results show that self-propellant microrockets can specifically capture cancer cells.

Self-propelled micro-motors have paved the way to exciting applications in biomedical field such as delivering drugs,^[1] nanoscale transport and assembly,^[2] motion-based biosensing disease markers and microrobotics.^[3] The usage of such micromachines to transport diverse payloads is one of the next prospects for nanomotor development.^[4-6] Particularly chemically powered micro-/nanomotors based on different chemical compositions and structures, that are capable of moving autonomously in the presence of hydrogen peroxide fuel are being given emphasis.^[2,3,7,8] Among these, self-propelled microshuttle is lucrative for practical biomedical applications.^[2-7]. Especially, fabrication of nano and micropropellant systems featuring specific cell recognitions in shortest time frame is highly anticipated and yet challenging.

We report for the first time a chemically-powered CNT based magnetic micromachine for isolating and transporting cancer cells. Described microrocket technology could find a potential application as a bio-analytical micromachine for selective and rapid isolation of circulating cancer cells (CTCs). Detecting CTCs is a challenge due to the extremely low occurrence (10-100 per milliliter of blood) of CTCs among a large number of hematologic cells in the blood (10⁹ mL^-1).^[10,11] Several strategies, involving immunomagnetic beads or microfluidic devices, have been designed for isolating and accounting CTCs from blood.^[12-14] However, most of these approaches are limited by their slow rate and low CTC-capture yield.^[13,14] Currently, only one technology is available commercially (CellSearch^®) to identify CTCs from the blood of cancer patients.^[9]

We recently reported rapid and specific isolation of CTC based on magneto-dendritic nanosystem ^[12] Here we show the preparation of bio-functionalized microrockets and their application for rapid isolation of cancer cells. Microrocket system consists of three functional components: (i) CNT, (ii) iron oxide (Fe₃O₄) nanoparticles for magnetic isolation, and (iii) Tf ligand (Tf) for specific targeting. Fe₃O₄ nanoparticles were loaded in the inner surface of CNTs as described previously with some modifications.^[15] The unique advantages of the microrocket platform are: i) low density of microparticles ii) instantaneous propellant motion of microparticles (downward and upward) with an ability to ‘strike’ cancer cells in its path, iii) rapid capture (~5 min) of TfR-overexpressing (TfR⁺)cancer cells at the clinically relevant concentrations (approximately 1 CTC per 10⁵ blood cells), iv) specific targeting ability due to presence of Tf ligand, which is generally used to capture cancer cells overexpressing TfR on their membranes, v) magnetic isolation of the captured cells owing to the presence of Fe₃O₄ nanoparticles and (v) large surface area and internal volume allows multiple components to be loaded onto the nanotube.

We followed a multi-step process (Figure 1A) to synthesize the Tf-CNT-Fe₃O₄ platform. Figure 1A displays the TEM image of Fe₃O₄-CNTs. We observed that Fe₃O₄ particles deposited inside CNT had a uniform size of ~ 6 nm (see supporting information; Figure S1). Moreover, the image indicates the successful assembly of the Fe₃O₄ nanoparticles in the nanotubes with 4~8 nm i.d.Coupling of Tf on CNT-Fe₃O₄ composite was confirmed by Fourier transformed infrared (FTIR) spectroscopy (see supporting information; Figure S2). Further, Tf attachment quantified by a modified Bradford procedure was found to be ~ 0.2 mg of Tf per g of CNT-Fe₃O₄. The magnetic property of Tf-CNT-Fe₃O₄ particles was also visually evaluated in aqueous medium by placing it next to a permanent magnet (see supporting information; Figure S3).

Figure 1B illustrates the self-propulsion of suspended Tf-CNT-Fe₃O₄ microparticles mimicking microrocket in a solution containing H₂O₂. Microrockets initially propelled instantaneously towards the bottom of the tube and gradually reverted direction upwards. We noticed that the speed of the Tf-CNT-Fe₃O₄ microrocket was strongly dependent on the H₂O₂ concentration (Figure S4). As expected, the microrockets display highest speed at the maximum H₂O₂ (8%) concentration tested in aqueous solution. The upward directional speed of the microrockets decreases gradually from 0.90 mm•s^-1 (at 8%) to 0.68 mm • s^-1 (at 4%), and subsequently to 0.56 mm • s^-1 (at 1%) per unit area of the particles. We hypothesize that microrocket motion is influenced by the surrounding H₂O₂ concentration as it affects the rate of the Fe³⁺ reduction to Fe²⁺.

Notably such chemically powered micromotors are commonly incompatible with the high ionic strength environment of biological fluids and extending the scope of such microrockets to physiological conditions is a key challenge. So we studied the ability of our microrockets to propel in biological fluid such as Dulbaco’s modified eagle medium (DMEM) cell media. Interestingly, the microrockets were found to propel efficiently even in DMEM containing 4% of H₂O₂. Figure 1B shows images of the microrocket at different vertical positions during its motion for a complete cycle. Initially, it was observed that microrockets, of smaller dimension, floated at the meniscus due to the low density and attached O₂ bubbles, formed by Fe₃O₄ nanoparticle catalyzed decomposition of H₂O₂. Once the smaller bubbles dispersed, a downward movement was observed (Figure 1B) due to the gravitational force. In the meantime, more O₂ bubbles formed and adhered to the Tf-CNT-Fe₃O₄ particle and in few seconds, the adhered O₂ bubble grew larger by coalescence of several smaller bubbles. Eventually the total volume of the bubble was sufficiently high, so that the buoyancy force balanced the gravitational and viscous forces and the microrocket moved upward (Figure 1B).^[16,17] Figure 2A shows the tracking trajectory of a microrocket in cell media, indicating a vertical motion. It is noted that the microrocket moved with high average speed of 0.38 mm • s^-1 in the downward direction and 0.62 mm • s^-1 in the upward movement per unit area of the particle (about 12 and 16 times its body length per second). This corresponds to a

large driving force of over 231 and 300 pN, based on the drag force F=6πμrv, where v is the speed, and μ is the viscosity of the medium and r is the radius of the microrocket. Further, when the concentration of H₂O₂ in cell media was varied there was notable effect on the speed as shown in Figure 2B. The speed of the microrocket at 8% H₂O₂ concentration was ~1.9 and~1.4 times higher in the downward and upward direction than at 1% of H₂O₂ concentration. Also, the number of times the microrocket moved up and down changed. At high H₂O₂ concentration (8%) the total distance moved by the microrocket was ~2 times more than that at lower concentration (1%) (Figure S5 and S6). Moreover, it was also revealed that the speed of the microrockets was only slightly affected in DMEM. The speed of the microrocket was 10% less in DMEM cell media compared to aqueous media containing 4% H₂O₂. Partial blocking of Fe₃O₄ surface by adsorbed proteins and increased solution viscosity may be responsible for this moderated speed.

We validated the application of microrockets for selective and rapid isolation of cancer cells from a heterogeneous population by separating HCT116 cells from an artificial CTC suspension. We placed cell suspension in DMEM at the concentration of (1 x 10⁶ cells mL^-1) containing 4% H₂O₂. Tf-CNT-Fe₃O₄ microrockets were incubated in cell suspension for 5 min to target and isolate the HCT116 cells. We envisioned that Tf-functionalized microrockets could strike and selectively bind suspended HCT116 cells through the TfRs and finally transport them at the top of the tube from where they can be retrieved. Figure 3A illustrates the pick-up and transport of a cancer cell by a microrocket. On the other hand, Figure 3B shows cancer-cell-loaded microrocket. The force necessary for moving a relatively large (~16 mm) cancer cell is considerably high. The minimum force necessary for transporting such large cells at one body length per second in DMEM estimated from Stokes’ law is 1.88 pN. The high speed of the microrocket is slightly affected by the cell loading (e.g., decreasing from 0.62 to 0.50 mm•s^-1 in cell media), reflecting its high towing force. We observed that Tf-CNT-Fe₃O₄ microrockets can efficiently pick-up and transport HCT116 cancer cells. We observed rapid action and selective targetability of Tf-CNT-Fe₃O₄ microrocket in capturing cells from a suspension. It also confirmed that the isolated cells remained normal for ~30 min and could be used for further studies (see supporting information; Figure S7). In addition, the HCT116 cell viability was also evaluated in DMEM containing 4% H₂O₂ after 1 hour. The study showed majority of the cells (78%) remained viable (Figure S7).

Furthermore, in order to confirm that the specific interaction of Tf-CNT-Fe₃O₄ microrocket with TfR⁺ cells (such as, HCT116) is because of Tf, we used CNT- Fe₃O₄ microrocketwithout Tf to capture TfR⁺ cells (as a negative control). We found that no cells were attached to the CNT-Fe₃O₄ microrocket even after 5 min incubation. Hence, experiments with HCT116 confirm that after 5 min incubation, there is very little non-specific interaction of cells with the Tf-CNT-Fe₃O₄microrocket. In addition, to evaluate the effect of self propulsion of Tf-CNT-Fe₃O₄ microrocket on HCT116 capture, Tf-CNT (without Fe₃O₄) was used. It was found that Tf-CNT could capture only ~ 22% HCT116 cells (see supporting information; Figure S8). The study shows that self-propulsion of Tf-CNT-Fe₃O₄ microrocket plays a significant role in cell capture and isolation.

Our final goal was to quantify the capture efficiency of microrocket while targeting cancer cells from an artificial CTC suspension. We spiked human peripheral blood mononuclear (hPBMC) cells with TfR⁺ GFP-labeled HCT116 cells in various hPBMC: HCT116 ratios (e.g. 1 x 10³:1, 1 x 10⁴:1 and 1 x 10⁵:1). Cell suspensions of different ratios were incubated with Tf-CNT-Fe₃O₄ microrocket for 5 min prior to magnetic isolation. Following isolation, both the captured and the residual cell suspensions were imaged to estimate the number of captured and uncaptured HCT116-GFP cells in each sample. Assuming the number of captured HCT116-GFP cells to be N_C, the number of uncaptured HCT116-GFP cells to be N_U and following the method of Zheng et al, the capture efficiency (C.E.) for each dilution was estimated as:^[18

(1)

Figure 4A shows image of captured cells (green) attached to the Tf-CNT-Fe₃O₄ particles (black). As expected, Tf-CNT-Fe₃O₄ particles are seen to aggregate following magnetic separation. Figure 4B shows an image of the residual cell suspension containing hPBMC cells and, in this particular case, a single uncaptured HCT116-GFP cell is seen (shown in red circle). Figure 4C shows a plot of recovered vs. spiked HCT116 cells. The % of recovered HCT116 cells for the clinically relevant range of HCT116:hPBMC ratios (1:1 x 10⁴ to 1:1 x 10⁵) are highlighted separately in the histogram of Figure 4D. Cell capture studies confirmed that Tf-CNT-Fe₃O₄ particles can successfully capture ~ 85% of the cancer cells within 5 min from a background of hPBMNCs even when the abundance of cancer cells is as low as ~0.001% of total number of cells.

Experimental Section

Preparation of CNT-Fe₃O₄ microparticle: Purification and oxidation of CNT were carried out using a previously described literature procedure.^[18] Oxidized CNTs (AO-CNT) were then mixed with ethylene glycol and water. FeCl₃⋅6H₂O and FeCl₂⋅4H₂O were added to the mixture and sonicated for 2 h, followed by vigorous stirring for 4 h. The pH was adjusted to ~10 using NH₄OH. The resulting CNT-Fe₃O₄ microparticles were isolated by magnetic separation, washed with distilled water and dried overnight under vacuum.

Conjugation of Tf with CNT-Fe₃O₄: 2 mg of Tf was dissolved in 2 mL of D.I. water and 1.0 mg of the CNT-Fe₃O₄ was added to it. EDC.HCl was added to this reaction mixture (adding the same number of moles of EDC.HCl as Tf) and constantly stirred at room temperature for 4 h. The reaction time was limited to 4 h to avoid any possible intramolecular cross linking. Resulting Tf conjugated CNT-Fe₃O₄ microparticles were isolated by magnetic separation and dried at room temperature under vacuum.

Tf-CNT-Fe₃O₄-cell imaging: HCT116-GFP cells were plated at a density of 2 x 10⁵ mL^-1 on glass coverslips in 35 mm culture dishes. After 24 h, HCT116 cells in 1 mL DMEM were treated with 500 µg mL^-1 of Tf-CNT-Fe₃O₄ in a NMR tube for 5 min and then subjected to magnetic field separation. The cell pellet obtained after a minute in the strong magnetic field was removed from the remaining cell suspension (containing uncaptured cancer cells). The captured cell pellet and the residual cell suspension was imaged by a Zeiss microscope (Zeiss, Observer.Z1) fitted with a 20 × objective using both bright field and fluorescence channels and the number of cells were counted using ImageJ plugin.^[19

Estimation of capture efficiency from artificial CTC suspension: Artificial CTC samples were prepared by spiking hPBMCs with GFP-labeled HCT116 cells at specific ratios (1:10³-1:10⁵ for HCT116:hPBMC respectively). 500 µg mL^-1 Tf-CNT-Fe₃O₄ was added to the 1 mL of artificial CTC mixture incubated for 5 min and then subjected to magnetic field separation. The cell pellet obtained after a minute in the strong magnetic field was removed from the remaining cell suspension (containing hPBMCs and uncaptured cancer cells). The captured cell pellet and the residual cell suspension was imaged and the number of HCT116 cells were counted using ImageJ cell counter plugin.^[19

Motion parameters, cell culture, isolation of human hPBMC and characterization are included in the supporting information.

Conclusions

We demonstrated a novel CNT based microrocket that propels efficiently by the thrust of O₂ bubbles. The new self-propelled microrockets display ultrafast propulsion in aqueous solution as well as in DMEM. The microrocket displayed a driving force of over 231 and 300 pN in DMEM containing 4% H₂O₂. The speed and the distance travelled by the microrocket can be manipulated by changing the H₂O₂ concentration. The designed multifunctional microrocket has the ability to (i) rapidly target (~5 min) and efficiently capture (~85%) TfR⁺ cancer cells from an artificial CTC-like suspension, (ii) magnetic isolation of the captured cells from peripheral blood cells and (iii) subsequent high resolution imaging. We envision that such self-powered micromotors may provide a new and unique approach for rapid and efficient extraction of CTCs from biological fluids and hence for the early diagnosis of cancer and its recurrence.