Wake‐Riding Effect‐Inspired Opto‐Hydrodynamic Diatombot for Non‐Invasive Trapping and Removal of Nano‐Biothreats

Abstract Contamination of nano‐biothreats, such as viruses, mycoplasmas, and pathogenic bacteria, is widespread in cell cultures and greatly threatens many cell‐based bio‐analysis and biomanufacturing. However, non‐invasive trapping and removal of such biothreats during cell culturing, particularly many precious cells, is of great challenge. Here, inspired by the wake‐riding effect, a biocompatible opto‐hydrodynamic diatombot (OHD) based on optical trapping navigated rotational diatom (Phaeodactylum tricornutum Bohlin) for non‐invasive trapping and removal of nano‐biothreats is reported. Combining the opto‐hydrodynamic effect and optical trapping, this rotational OHD enables the trapping of bio‐targets down to sub‐100 nm. Different nano‐biothreats, such as adenoviruses, pathogenic bacteria, and mycoplasmas, are first demonstrated to be effectively trapped and removed by the OHD, without affecting culturing cells including precious cells such as hippocampal neurons. The removal efficiency is greatly enhanced via reconfigurable OHD array construction. Importantly, these OHDs show remarkable antibacterial capability, and further facilitate targeted gene delivery. This OHD serves as a smart micro‐robotic platform for effective trapping and active removal of nano‐biothreats in bio‐microenvironments, and especially for cell culturing of many precious cells, with great promises for benefiting cell‐based bio‐analysis and biomanufacturing.


Introduction
Nanoscale biological threats (nano-biothreats), such as viruses, mycoplasmas, and pathogenic bacteria, are notorious and adventitious contaminants of cell cultures and bio-microenvironments. These organisms can modify both the physiology of cultured cells and the structure of recombinant biomolecules. Due to the rapid reproductive capacity, even a small number of nano-biothreats DOI: 10.1002/advs.202301365 in bio-microenvironments can result in a great threat and even a disaster for different biomedical applications ranging from basic cell culturing to bio-analysis and biomanufacturing. [1] Such contamination is widespread in laboratories during cell culturing and can cause great economic losses in biomedical research. [2][3][4] Therefore, it is of great urgency to develop efficient tools for nano-biothreat removal and anti-bacteria that can be directly used in cell cultures. [5] To further facilitate cell-based biomanufacturing and therapeutics, such tools should be biocompatible and noninvasive to bio-microenvironments and cultured cells, which otherwise would generate adverse effects for further biomedical applications such as single-cell analysis and biomanufacturing. However, the design of effective methods combating the contamination and rapid spreading of nanobiothreats in bio-microenvironments and cell cultures, especially during culturing of many precious cells, such as pluripotent stem cells and primary neurons, is always a thorny problem. [6] Conventional nano-biothreat removal methods utilize either chemical (eg., 75% ethanol) or physical (eg., ultraviolet light) agents, which are the most widely used methods for cell cultures. [7,8] However, the effective removal and killing of nanobiothreats requires a high dosage of these sterilization agents, which lacks selectivity and can harm other biological samples, and thus cannot be applied during cell culturing and biomedical application processes. [9,10] Although the use of antibiotics has increased the selectivity for bacteria killing, antibiotics increase the risk of bacterial resistance. [11] In addition, it is not recommended to use antibiotics during neuron (especially primary neuron) culturing, since antibiotics can affect the living state of neurons. It is also not recommended to use antibiotics when performing gene transfection via cell culturing. Therefore, the design of nextgeneration platforms as nano-biothreat removal and antibacterial agents with high selectivity, efficiency, as well as high biocompatibility, that can actively work during cell cultures, especially for culturing of precious cells, such as pluripotent stem cells and primary neurons, is of great importance and urgency for many biomedical applications. [12][13][14][15] Different methods have been developed to meet this requirement. For example, the development of antibacterial nanomaterials has added a new dimension to nano-biothreat removal and antibacterial action. [16][17][18][19] However, many of the nanomaterial-based platforms are static and the antibacterial action is passive. On the other hand, micromotors (also called microrobots) that can convert external energy into motion hold great potential for controlled navigation and active operation in bio-microenvironments. [20][21][22][23] The dynamic and active feature provides a new suggestion for dynamic antibacterial and active nano-biothreat removal in bio-microenvironments. In particular, different strategies based on external energy sources, such as light, [24][25][26] magnetic, [27,28] and ultrasonic propulsion, [29] have been applied for the chemical-free navigation of micromotors in microenvironments. Among them, magnetic controlled micromotors are widely used for nano-biothreat removal. [30,31] However, all these magnetic-controlled micromotors necessitate additional specific magnetic materials to respond to magnetic sources for actuation. In another scenario, a micromotor platform directly using microorganisms with self-propelling capabilities, such as sperm, [32] green alga, [33] and rotifer, [34] can also be used for high-efficiency nano-biothreat removal. However, these motile organisms can affect cell growth and metabolism during culturing. In turn, the cell culture media can also affect and destroy the motility feature of the motile organisms, which makes it impossible to further work as a micromotor in cell cultures. Therefore, it is highly desirable to design an intelligent, active, and biocompatible platform that can be directly used in biomicroenvironments for nano-biothreat removal. Although optical tweezers (OT) provide huge potential for trapping and manipulating microscopic objects, [35] and recently shows the potential for micromotor actuation, [36,37] the diffraction limit makes it challenging for trapping nanoscale objects such as the nanobiothreats. In addition, the strong light intensity needed for stable trapping can induce potential harm to biological samples. On the other hand, hydrodynamic tweezers also show great potential for non-contact manipulation of micro/nano-objects via localized microfluidic flow. [38,39] However, neither of these two methods can achieve effective nano-biothreat removal due to their limited range of action.
In nature, dolphins often cruise the boat wakes to catch a free ride, allowing them to swim and migrate with far less energy than usual. This phenomenon is called the wake-riding effect. Such a wake-riding effect can be found at both low Reynolds numbers (Re) and high Re. For example, in the case of low Re, the wake-riding effect is utilized for colloidal particles transporting under the assistance of localized flow fields generated by front-moving particles. [40] In the case of high Re, during the process of human swimming, followers are shown to use the leader's wake to offset the wave, thereby reducing the resistance and achieving more efficient swimming. [41] Inspired by this wake-riding effect, in this work, by combing optical trapping and opto-hydrodynamic effect (OH effect), we report an optohydrodynamic diatombot (OHD) for trapping and removal of nano-biothreats in bio-microenvironments of cell cultures based on optical trapping-navigated triradiate Phaeodactylum tricornutum Bohlin (PTB, a widespread diatom in nature) (Figure 1a). It should be noted that we use the term wake-riding effect because of the similarity between nano-biothreats trapping by moving OHD and dolphins migrating with boats. Although the wakeriding effect for dolphins occurs at the water surface, our OHD can work inside fluids. Nano-biothreats near the wakes of the fast-rotating OHD are easily trapped via the OH effect, where the hydrodynamic resistance of nano-biothreats is significantly reduced. This effect greatly increased the nanoscale trapping efficiency of optical tweezers. These OHDs are capable of active and on-demand removal of different nano-biothreats in biomicroenvironments of cell cultures, such as viruses, mycoplasmas, and pathogenic bacteria, without affecting the viability of cultured cells including primary neurons. By coating with chitosan, these OHDs show remarkable antibacterial capability after bacteria collection, without affecting the viability of cultured cells and greatly increasing the survival rate of living cells. These features further facilitate targeted gene delivery. This OHD provides a new method for effective trapping of nano-objects in the biomicroenvironment and will serve as a new micro/nano-robotic platform for non-invasive, high-efficiency, and broad-spectrum removal of nano-biothreats during cell culturing, including many precious cells, such primary neurons, with great promises for benefiting single-cell analysis and cell-based biomanufacturing.

Experimental Setup for Wake-Riding Effect-Inspired OHD for Nanoscale Trapping
Natural PTB exhibits a triradiate morphology with three elongated diatom arms in the normal growth condition ( Figure 1b, the average size of the arm: 7.9 μm in length and 1.8 μm in width, average angle 120°, details see Figure S1, Supporting Information). This natural PTB, however, is an immotile diatom and is not able to locomotion. By applying an annularly scanning optical trap (scanning radius R: 10 μm, trapping power P, scanning frequency f) on the PTB (Figure 1c) based on a standard optical tweezers system (Aresis Tweez 250 si, operation wavelength: 1064 nm, details see Figure S2, Supporting Information), an immobile PTB can be turned into a controllably motile OHD (Movie S1, Supporting Information). A typical rotation trajectory of an OHD ( Figure S3a, Supporting Information). This rotation speed becomes stable with an optical power larger than 50 mW and a scanning frequency f larger than 8 kHz, and stabilizes at a maximum value of about 200 rpm ( Figure S3b,c, Supporting Information). However, it should be noted that a large optical power will inevitably cause optical damage and thermal damage to other biological samples in the bio-microenvironment. Therefore, to get an effective rotation and minimalize the light damage to the biological sample, we selected the parameters of P = 50 mW and f = 8 kHz to control the OHD in the experiments. The high-speed rotation of the OHD induces highly localized flow fields and hydrodynamic vortexes around the OHD arms. By moving the OHD along a predefined trajectory of laser trap, the OHD can be navigated to designated locations for targeted nano-biothreat removal as shown in Figure 1c.

Simulation Analysis
To show the nano-biothreat removal capability of the OHD, a numerical simulation based on the finite-element method was carried out. The geometry structure of the OHD in the simulation is  shown in Figure S4, Supporting Information, which is in accordance with the structure observed in the SEM images ( Figure 1b). Specific simulation details can be seen in Supporting Information. In the simulation, the OHD was subjected to a focused laser beam and locomoted along a predefined circular trajectory (radius R: 10 μm). The results show that the maximum velocity of OHD linearly increases with the increase in scanning frequency (f) (Figure 1d), whereas the nano-biothreats tend to be stable after a short acceleration process, due to the limited optical force. This shows that it is very difficult to remove nano-biothreat by the optical force alone. However, if the nano-biothreats are placed near the PTB and surf the flow around the PTB, it requires only a small compensation speed (Δv) to follow the PTB (Figure 1e), which can increase the removal efficiency. For the simulation in Figure 1e, the scanning frequency is kept at 8 kHz (the rotation speed of the OHD is 200 rpm), which is the same as that in the experiments. The situation that nano-biothreats moving along OHD by OH effect is similar to a dolphin riding the wake of a boat to catch a free ride, as shown in Figure 1f. At the microscale, the size difference between OHD and nano-biothreats is similar to that between a large boat and a dolphin. In both cases, a smaller object utilizes the flow field generated by a larger moving object to catch a free ride and labor-saving migration with less energy than usual. In order to show the OH effect of OHD on the nano-biothreats, the flow field distribution of an OHD-modeled microboat was simulated (Figure 1g), where the forward velocity of the microboat was set the same as the tangential velocity of the rotating OHD (209 μm s −1 ). It should be noted that the www.advancedsciencenews.com www.advancedscience.com vortex behind the microboat in Figure 1g schematically shows the vortex flow behind the microboat. Without the OH effect, fluidic resistance is increased with the increase in forward velocity. For the nano-biothreat (polystyrene particle, PS, 150 nm) with the forward velocity of 209 μm s −1 , the fluidic resistance is up to 295 pN. On the contrary, under the OH effect, the calculated maximum fluidic resistance of nano-biothreat is only 7.9 pN, which is 37 times smaller than that without the OH effect ( Figure 1h). Benefiting from the OH effect, a small optical force exerted on the nano-biothreat can thus overcome this fluidic resistance, resulting in the trapping and collection of the nano-biothreat by the OHD.
To further show the OH effect-assisted trapping of nanobiothreat by OHD, we analyzed the trapping efficiency of both sole optical trapping and OH-assisted trapping for 150-nm PS nanoparticles (Figure 2a). With a focused laser beam irradiated (power: 50 mW), microscopic objects can be trapped by optical force (F O ) via photon momentum and optical pressure change ( Figure 2bI). It can be seen that the optical pressure (P x ) on both sides of PTB is equal in value and opposite in direction, resulting in zero resultant optical force. However, as the PTB deviates from the optical axis, the symmetry destruction of the optical pressure generates a spring-like optical force (F O-PTB ) that tends to draw the PTB back onto the optical axis ( Figure 2bII). The maximum of F O-PTB (F Om-PTB = 1.99 pN) exists at the distance of 0.28 μm. The optical force (F O-NP ) exerted on the nanoparticle follows a similar law to PTB, but the value is more than two orders of magnitude smaller, and the maximum value (F Om-NP = 12.4 fN) exists at the distance of 0.6 μm. The results above indicate that the sole optical trapping has a weak confinement ability on nanoparticles. The motion of PTB and nanoparticles in static liquid is restricted by the fluidic resistance (f v ), which is linearly increased with the velocity (v) of the object (Figure 2bIII). Calculation results indicate that the maximum velocity (v max ) that can be driven by F Om-NP is 14 μm s −1 when trapping of 150-nm particle, while this v max for PTB is 209 μm s −1 (Figure 2bIII). Smaller particles move slower in liquids when the optical force and fluidic resistance are balanced during the trapping process. This is because the fluidic resistance is linearly increased with particle size, while the driving force F Om-NP is quadratically increased with an increase in particle size ( Figure 2bIV). However, in real experiments, considering the Brownian motion, the instantaneous velocity of random motion for nanoparticles can even be larger than the value of v max , and thus the particle can escape from the trap. Therefore, for sole optical trapping, it is challenging for effective trapping of nanoscale objects.
In the simulation, we further considered the nanoscale trapping in a moving OHD (v OHD = 209 μm s −1 , same as the calculated v max for PTB motion in Figure 2bIII) controlled by optical trapping (Figure 2c). The flow field around the moving OHD in water is depicted in Figure 2cI. Effective removal of nanoparticles requires the particles to follow the moving OHD with the same velocity. However, as previously calculated (Figure 2bIII), nanoparticles cannot travel at this speed by optical force. Fortunately, a localized flow field is generated around the PTB during PTB migration (Figure 2cI). For nanoparticles in such a flow field, a small optical trapping force (F O-NP ) can overcome this velocity difference, allowing the nanoparticles to remain synchronized with the OHD, which is similar to the wake-riding effect. Therefore, we are concerned about the velocity difference (Δv = v − v OHD ) to OHD as a reference (Figure 2cII). Combining the value of v max for different-sized particles that can be driven by optical force in Figure 2bIV and the contour line of Δv in Figure 2cII, we can obtain the effective capture range of particles with different sizes by the OHD in the flow field (Figure 2cIII). It can be seen that larger particles own a wider capture range. For sole optical trapping, the random motion of nanoparticles that can be suppressed and trapped by optical force is very limited as discussed in Figure 2b. For example, for a 150-nm nanoparticle, the maximum particle velocity is only 14 μm s −1 . However, due to the OH effect, the nanoparticles can be confined by the OHD, and rotated together with the OHD (Figure 2cIV), therefore, realizing the transition of trapping/removal velocity (v max ). This efficiency is more obvious for particles with smaller sizes. The Re number of the objects in our system can be estimated according to: Re = vd/μ = 0.01, where = 1 × 10 3 kg m −3 is the fluid density, v = 1 × 10 −3 m s −1 is the fluid velocity, d = 10 μm is the characteristic length, and μ = 1 × 10 −3 Pa s is the dynamic viscosity coefficients. As described by Fox et al., laminar flow exists for the Re number smaller than 2900, while for the Re number larger than 2900, the flow field changes from laminar flow to turbulent flow. [42] Therefore, the collection and removal of nano-biothreats by our OHD are in a laminar flow regime with a low Re number. In addition, we also numerically analyzed the removal rates with the flow fields changing from laminar to turbulent regime by changing the Re number with different dynamic viscosity coefficients. As shown in Figure S5, Supporting Information, with the flow field changing from laminar to the turbulent regime, the removal rate gradually decreases.
To further show the nanoscale trapping and removal capability of the rotating OHD, simulations on multiple 150-nm PS nanoparticle trapping and collection via both annularly scanning OT and OHD were both performed. In the simulation, the scanning frequency of both methods was kept the same at 8 kHz. As shown in Figure 2d, effective collection is achieved after two circles of OHD rotation, while no obvious collection is achieved even after three circles for OT (details see Movie S2, Supporting Information). This collection performance was also demonstrated experimentally. Effective collection and removal of Escherichia coli were realized at an operation time of t = 13.5 s for OHD, while 80% of the bacteria were still randomly distributed in the microchannel for OT ( Figure S6, Supporting Information). Repeated experiments show that about a 100% removal rate can be achieved after 14 s for OHD, with a much higher efficiency than that of OT at the same time ( Figure S7a, Supporting Information). Due to the synergic effect of optical force and hydrodynamic force, this removal capability also depends on the power of the trapping laser. As shown in Figure 2e, the simulated removal rate of OHD increases with the increase in optical intensity, and it is much higher than that for OT under the same optical intensity. From Figure 2e, it can also be seen that the removal becomes stable with the light intensity of 5 × 10 10 W m −2 at the focus. It should be noted that in the simulation, the light intensity shown in Figure 2e is the strongest intensity at the center of the focus. However, in the experiments, the focus size is about 2 μm in radius, and the optical power for effective OHD rotation and nano-biothreat collection is 50 mW. In this case, the average light intensity at the focus is estimated to be 3.98 × 10 9 W m −2 . This light intensity is within the typical light intensity range of optical tweezers, which is in the order of 10 9 W m −2 . [43] Compared with previously reported microalgae trapping and manipulation with optical power up to 150 mW, [36] the power of 50 mW we used for effective nano-biothreats trapping and collection is relatively low. In the experiments, we find that for optical power less than 40 mW, the removal rate is increased with the increase in optical power, and a removal rate of about 100% can be achieved with an optical power higher than 40 mW for OHD, while OT is only about 11% under the power of 40 mW ( Figure S7b, Supporting Information). Compared with other shaped algae or diatoms in nature, such as spherical or spindle-shaped ones, the www.advancedsciencenews.com www.advancedscience.com most prominent advantage of PTB is the high-efficiency trapping and collection capability of nano-biothreats resulting from its triradiate morphology with three elongated diatom arms. For example, for spherical algae (Chlamydomonas reinhardtii) and spindle diatom (Nitzschia Closterium), 70% and 90% of E. coli were still randomly distributed in the microenvironment after 30 s collection under the action of annular optical trap, respectively (Figure S8, Supporting Information), while for our PTB-based OHD, about 100% of the E. coli are trapped and collected. Both the simulation and experimental results show the wake-riding effect inspired by OHD exhibits a remarkable removal capability due to the synergic opto-hydrodynamic force and thus can be used for nano-biothreat trapping and removal.

Non-Invasive Trapping and Removal of Nano-Biothreats
As the three most widely existing nano-biothreats in cell cultures, the contamination of viruses, pathogenic bacteria, and mycoplasmas are great threats to many cell-based biomanufacturing and therapeutics. To show the non-invasive nano-biothreat trapping and removal capability of our OHD, a demonstration of nanobiothreat trapping and removal in a microfluidic channel was carried out (Figure 3a). Randomly distributed nano-biothreats are first trapped and collected around the OHD arms during OHD rotation (200 rpm). The collected nano-biothreats can be swept to a designated location via navigating the OHD. By turning off the trapping laser, the localized flow fields around the OHD arms disappear, and the collected nano-biothreats can be released at the designated location. Subsequently, the annular scanning optical trap is switched to a central trap with lower power (5 mW), and the OHD can be navigated to other positions for repeated use in subsequent experiments.
As some examples, Figure 3b-d shows the trapping and removal of adenoviruses (90-100 nm in diameter), [44] pathogenic bacteria (rod-shaped gram-negative E. coli), and mycoplasmas by using our OHD (P = 50 mW and f = 8 kHz) in a disc-shaped microfluidic channel (diameter: 100 μm, depth: 50 μm, fabrication details, see Experimental Section). As shown in Figure 3b, about 20 randomly distributed adenoviruses were completely trapped by the OHD at t = 20 s. By turning off the laser, the collected adenoviruses were released from the OHD and completely removed from the channel at t = 25 s, and the OHD was then moved to another position. Due to the small size of nano-biothreats, such as the virus and bacteria, and the diffraction limit, it is difficult to clearly observe the nano-biothreats. Therefore, we assigned yellow arrows at the positions of nano-biothreats for better identification. To avoid confusion caused by the yellow arrows, Figure S9, Supporting Information, shows a representative raw image, as well as that with the yellow arrow assigned during virus collection. Details of the trapping and removal of adenoviruses are shown in Movie S3, Supporting Information. To show the trapping and removal capacity of the OHD for different-sized nano-biothreats, PS particles with different sizes were used as the nano-biothreat models for demonstration. Experimental results show that our OHD is capable of the effective trapping and removal of PS particles with sizes from 100 nm to 2 μm (Figures S10-S12, Supporting Information). In addition to the effective removal of immotile abiotic particles that imitate nano-biothreats, importantly, our OHD can also be used for the effective removal of motile nano-biothreats, for example, pathogenic bacteria. Pathogenic bacteria are a common contamination during cell cultures. For example, contamination of a small number of pathogenic E. coli can result in the death of both HeLa cells and human promyelocytic leukemia cell line HL-60 within 12 h (Figure S13, Supporting Information), due to the rapid bacterial reproduction in the cell culture medium. Effective removal of rodshaped E. coli and spherical Staphylococcus aureus were realized, respectively (Figure 3c; details see Figures S14 and S15 and Movie S4, Supporting Information). In addition, in order to more clearly demonstrate the removal process of nano-biothreats, as an example, we performed additional experiments using fluorescent E. coli. The collection and removal process can be clearly observed under a fluorescent microscope ( Figure S16, Supporting Information).
In addition to the effective trapping and removal capability, the non-invasiveness and biocompatibility of the OHD are also very important for further cell-based biomedical applications. To show the non-invasiveness of the OHD during nano-biothreat removal, we further carried out experiments on the removal of mycoplasmas in the channel containing both mycoplasmas and cultured mammalian cells (HL-60). Mycoplasma is one of the most common contaminants during cell culture and can result in the destruction of healthy cells ( Figure S17, Supporting Information). Because of the small size and deformability, it is difficult to remove mycoplasma efficiently by traditional filtration methods. By using our OHD, it is capable of highly efficient and selective removal of mycoplasmas. As shown in Figure 3d, about 19 mycoplasmas were collected and removed at t = 32.6 s. Importantly, during the mycoplasmas collection and removal, the HL-60 cell can be avoided from the affection of the OHD, and it was kept intact during the mycoplasma removal process (details see Figure S18 and Movie S5, Supporting Information). This indicates that the OHD exhibits a non-invasive feature for nanobiothreat removal. The effective removal time of a single OHD for different nano-biothreats is different (Figure 3e). In addition, the removal efficiency of OHD for different nano-biothreats is also related to its rotation speed. As shown in Figure 3f, when the rotation speed reaches 200 rpm, the removal rate can reach the best of 100%. The collection and removal capacity is different for different nano-biothreats. As shown in Figure 3g, for the nano-biothreats we used, the saturation removal numbers of adenoviruses, E. coli, S. aureus, and mycoplasmas were about 39, 40, 45, and 71 with a single OHD, respectively. Importantly, our OHD can work in both very fluid microfluidic environments and really dense cell cultures. Our OHD can go into the dense cell cultures through the narrow gaps between neighboring cells to remove the invaded nano-biothreats. To show the applicability of our OHD for nano-biothreats trapping and removal in dense cell cultures, we have performed additional experiments in bacteria-contaminated real dense cell cultures. As an example, Figure 3h shows the trapping and removal of contaminated E. coli in real dense cell cultures (HeLa cells, cultured in a real petri dish with cell confluency of up to 90%). In these highly dense cell cultures, the OHD can also easily trap and remove the contaminated E. coli. As recommended in a cell culture guide posted in BiteSize Bio (https://bitesizebio.com/63887/cellconfluency/), the ideal cell confluency is 70-80% for real cell culture, and cells should be split at this confluency stage to improve the overall cell viability. These results demonstrate that our OHD can work in real dense cell cultures with cell confluency even higher than the ideal cell confluency for cell culture. After the trapping and collection by the OHD in a petri dish containing the cell culture, we can move the OHD with collected nano-biothreats to a designated position by optical tweezers, and then both the OHD and the collected nano-biothreats can be extracted outside from the cell culture media using a pipette. To show the biocompatibility of the OHD to the confluent cells, we have also performed additional experiments for cell viability tests in confluent cells treated with OHD. After culturing confluent cells with 24-h OHD treatment, as shown in Figure S19, Supporting Information, the viability of the cells was not affected, which was similar to cells without OHD treatment.
To further show the non-invasiveness and biocompatibility feature of the OHD, we tested the biocompatibility of OHD to www.advancedsciencenews.com www.advancedscience.com two different mammalian cell lines (adherent HeLa cells and suspending HL-60 cells). After co-culturing OHD with the cells for 24 h, both cell lines show no obvious decrease in cell viability ( Figure S20, Supporting Information, green fluorescence for living cells, blue fluorescence for nuclei). The cell viability is not affected by OHD, which is similar to that of normal cell culturing (Figure 3i). These results indicate that the PTB-based OHD is highly biocompatible with bio-microenvironments and mammalian cells.

OHD Array for Efficiency-Enhanced Removal in Cell Cultures
Although single OHD is capable of non-invasive nano-biothreat trapping and removal, the efficiency is limited by individual operation. As the number of nano-biothreat increases, the collection and removal capacity can be saturated, and the removal rate of the OHD is gradually decreased when exceeding the saturation number ( Figure S21, Supporting Information). Therefore, the formation of OHD arrays with highly reconfigurable and controllable capabilities is important for efficiency-enhanced multitask execution and manipulation, with higher speed and larger collection volume. Fortunately, our OHD can be extended into OHD arrays with high reconfigurability and controllability, and multiple OHDs can operate sequentially or simultaneously. By extending the single optical trap into trap arrays, multiple traps with designated patterns can be formed. Multiple PTB cells can then be turned into OHD arrays. The rotation of each OHD element in the array can be controlled similarly to that of a single OHD. As some examples, Figure 4a shows the formed OHD arrays with the pattern of "DIATOMBOT." These OHD arrays with designated patterns will provide more choices for cooperative robotic operation and on-demand task execution with higher efficiency. These OHD arrays can work independently and collaboratively for nano-biothreat removal with higher efficiency than that for a single OHD. For a single OHD, the trapping of 25 E. coli was completed at 14 s ( Figure S22, Supporting Information). The completion collection time was reduced to 7.3 s and 6.3 s for a two-OHD and three-OHD array, respectively (Figure 4b; details see Figure S22 and Movie S6, Supporting Information). The efficiency for a three-OHD array is more than twice that for a single OHD. A comparison of the collection and removal for a single OHD and OHD arrays are also shown in Figure 4h.
These OHD arrays can be directly used in cell cultures for highefficiency nano-biothreat removal without affecting the cultured cells. As shown in Figure 4c, for a single OHD, the removal of 14 E. coli in cultured HL-60 cells was completed within 15 s. For a three-OHD array, the removal efficiency was more than twice that for a single OHD. Removal of a similar number of E. coli with a three-OHD array was completed within only 7 s (Figure 4d). During the removal, the viability of the cultured cells was not affected. This capability is also applicable to precious cells, such as hippocampal neurons, which are very difficult to extract and cannot be passaged. As shown in Figure 4e, during the in vitro culturing of hippocampal neurons with a small amount of E. coli contamination, owing to the fast-multiplying ability of bacteria, the neurons were totally invaded and even lysed by the bacteria after 6 h. Although antibiotic treatment can effectively in-hibit bacteria proliferation during cell culturing, antibiotics can affect the living state of neurons and induce irreversible damage to the neurons, and therefore it is not recommended to use antibiotics during neuron (especially hippocampal neurons) culturing. As shown in Figure 4f, with the treatment of 1% penicillin/streptomycin (pen/strep), although an effective inhibition of bacterial proliferation was achieved, the neurons were irreversibly damaged by the antibiotics, leading to a shrinking in the cell body and inhibition of neurite outgrowth after cultured for 6 h. With the OHD array for bacterial removal, as shown in Figure 4g, the contaminated E. coli were effectively collected and removed in only 19 s. Importantly, after the removal, the cultured neurons remained in a good living state, as evidenced by the outgrowth of a higher number and longer neurites. Importantly, after the removal, the cultured neurons remained in a good living state with new branches growing out. These results indicate that our OHD can be directly used for biothreat removal during neuron cell culturing. Due to the collection capacity limit of a single OHD, the total removal capacity for different OHD arrays is also different, and this capacity is increased with the increase in OHD number in the array (Figure 4i,j). For bacteria number that exceeds the removal capacity of a single OHD, although a single OHD cannot completely remove the bacteria, OHD arrays can get a perfect removal efficiency (Figure 4i and Figure S23, Supporting Information). For a given number of E. coli, the removal time is decreased with the increase in the number of OHD in the array (Figure 4j). These results indicate that we can build OHD arrays with more OHDs to further get a higher removal efficiency and capacity.

Non-Invasive Antibacterial Capability for Enhanced Gene Delivery
Despite the non-invasive and high-efficiency nanobiothreat collection and removal capability of OHD in bio-microenvironments, the further non-invasive and efficient killing of contaminated nano-biothreats in the biomicroenvironments are very important to ensure further singlecell analyses. Importantly, in addition to the nano-biothreat removal, our OHD is also capable of non-invasive bacterial killing and antibacterial treatment. As shown in Figure 5a, the OHD is modified with a chitosan (Chi) layer as a micro-robotic strategy for antibacterial treatment in cell culturing microenvironments. Chitosan has garnered increasing interest in the field of antibacterial as a renewable material due to its unique properties such as high biocompatibility, ease of decomposition, and low toxicity. [45] Our antibacterial strategy relies on the combination of the efficient nano-biothreat removal capability of OHD and the strong bactericidal activity of the chitosan layer. Taking E. coli as an example, naked OHD can realize E. coli removal. Nevertheless, it does not exhibit antibacterial properties, and there still exists an infection risk for the cultured cells. In this case, the final fluorescence of the OHD with collected live E. coli stained with a bacterial viability kit (details, see Experimental Section) is yellow. However, by coating a layer of chitosan, the collected E. coli can be killed by the OHD, demonstrating the good antibacterial effect of OHD ( Figure S24, Supporting Information). Therefore, there is no infection risk for cultured cells. Since the OHD is fluorescent  red and the dead E. coli also has red fluorescence, the measured fluorescence intensity also reflects the enhanced antibacterial effect of the chitosan-coated OHD (chi-OHD, Figure 5b).
In order to show the antibacterial ability of the chi-OHD, we carried out a series of different controlled experiments. To show the bacterial viability after different treatments, a commercial viability kit based on two dyes (DMAO: 9-Octadecen-1-amine,N,Ndimethyl-,(9Z)-and EthD-3: Ethidium Homodimer 3) was used for E. coli staining. The DMAO dye (green) was used to label live bacteria, while the EthD-3 dye (red) could only penetrate damaged bacteria and was used to label dead E. coli. As shown in Figure 5c, after the collection of E. coli for 10 min, the bare OHD fluoresces yellow, which is the combination of red (for PTB) and green (for live E. coli) fluorescence. This phenomenon indicates that bare OHD cannot result in E. coli killing. For the treatment with chitosan only, the fluorescence of about 40% E. coli is red, while the others are still green after 10 min treatment, indicating only about 40% of the bacteria are killed by the chitosan solution. However, for chi-OHD, all the fluorescence is red, indicating all the bacteria are killed after 10 min. The high antibacterial efficiency obtained by the chi-OHD reflects the key role of the combination of chitosan and the effective rotation of OHD for antibacterial activity. The OHD rotation increases chitosan-bacteria interaction, and thus results in a high-efficiency antibacterial performance. Figure 5h,i show the comparison of the antibacterial efficiency for naked OHD, chitosan solution, and chi-OHD. This antibacterial performance is dependent on the concentration of the chitosan solution for OHD coating. When the chitosan concentration reaches 0.2 mg mL −1 , the antibacterial efficiency of the OHD reaches 98% (Figure 5j).
This antibacterial capability is of great importance for cell culturing and further for single-cell analysis and cell-based biomanufacturing. As shown in Figure 5d, for HL-60 cells contaminated with E. coli, without any treatment, once the living HL-60 cells (green) are infected by active bacteria, the cell is dead after 60 min (red). However, by using chi-OHD for bacteria removal and antibacterial treatment, the HL-60 cell is not infected by bacteria even though the microenvironment is contaminated with E. coli, and the cell viability is not affected after 60 min (green fluorescence, Figure 5e). Although the chi-OHD can kill bacteria, chi-OHD is highly biocompatible and noninvasive to the cultured cells ( Figure S25, Supporting Information). This non-invasive antibacterial feature further facilitates the study of enhanced drug delivery and single-cell-based therapy. To show this capability, Cy3-labeled small interfering RNA mimics (siRNAs) that fluoresce red were loaded into mesoporous silica particles (details, see Experimental Section) and added to the cultured HL-60 cells contaminated with E. coli. The bacteria contamination and infection result in cell death (red fluorescence), and siRNA cannot be delivered into the HL-60 cells (Figure 5f). However, with the treatment of chi-OHD, the contaminated E. coli are completely removed and killed. The cell viability is thus not affected by the contaminated E. coli even after 2 h, which is similar to the cells without any contamination (Figure 5k). Since the cell viability is not affected by the contaminated E. coli, the siRNA is thus delivered into the cell in 90 min (Figure 5g). These results indicate that the non-invasive antibacterial capability of chi-OHD can be directly used to remove and kill the contaminated nano-biothreats in cell cultures, and further for enhanced drug delivery and subsequent single-cell analysis.

Conclusion
In summary, inspired by the wake-riding effect, we created a noninvasive and active OHD based on a natural diatom for nanobiothreat removal, which can be directly used in cell cultures for enhanced gene delivery. The synergic optical trapping and hydrodynamic force of OHD results in high-efficiency trapping and removal of different nano-biothreats, including adenoviruses, pathogenic bacteria, and mycoplasmas, in cellular environments without affecting culturing cells including precious primary neurons. This removal capability was further enhanced by reconfigurable OHD array formation. Efficient and non-invasive antibacterial performance was demonstrated in a cellular environment with chitosan-coated OHD, which further facilitated single-cell gene delivery.
This OHD provides a new method for the effective trapping of nano-objects in bio-microenvironment, and serves as a smart micro-robotic platform for non-invasive and active nanobiothreat removal and antibacterial treatment in cellular environments, offers a seamless interface among optical, microrobotic, and biological worlds. Due to the non-invasiveness and high biocompatibility, this OHD can be directly used in cell cultures for the removal of nano-biothreat contamination, without affecting the cultured cells including precious cells. This feature ensures subsequent cell-based biomanufacturing and bio-analysis after nano-biothreats removal and antibacterial treatment. OHD arrays ensure the capability of largerscale removal of nano-biothreats via the cooperation of microrobot individuals. This OHD provides a biocompatible and smart platform for non-invasive, high-efficiency, and broadspectrum removal of nano-biothreats, especially the rapid spreading drug-resistant bacteria, in bio-microenvironments of cell culturing of precious cells, such as neurons, with great promises for benefiting cell-based biomanufacturing and single-cell analysis.

Experimental Section
Experimental Setup of Optical Tweezers: A scanning optical tweezers system (Aresis, Tweez 250Si, Section S3, Supporting Information) was built on an inverted optical microscope (Nikon Eclipse Ti-U) using a continuous wave solid-state laser with 1064-nm operation wavelength (maximum output power: 5 W). The laser beam was refocused into the sample chamber after passing through a 60× water immersion inverted objective (numerical aperture: 1.0) for optical manipulation. Multiple trap sequences could be constructed in a loop manner by a computer-interfaced AOD system, and the high switching rate (maximum 100 kHz) ensured stable capture and precise manipulation of multiple targets at the same time. The real-time image was recorded by a computer-interfaced highspeed CCD camera. Details of the setup can be seen in Figure S2, Supporting Information.
PTB Preparation and OHD Construction: PTB diatom solution was purchased from Zhuhai Xinrui Trading Co., Ltd, China. PTB cells were obtained through a 47 mm × 5 μm mixed cellulose ester filter, and the PTB suspension was then prepared by resuspending PTB cells into 1 mL phosphate-buffered saline (PBS) buffer, with a final concentration of about ≈2 × 10 4 -4 × 10 4 cell/mL. This PTB solution was then transferred to the