Biotemplated flagellar nanoswimmers

Since Albert R. Hibbs first proposed the use of microscopic surgical machines to Richard Feynman in the 1950s,¹ there has been increasing interest in the potential use of untethered micro- and nano-scale robots for biomedical applications. Over the past decade, a number of research groups have demonstrated the potential of small scale robots to be used for in vivo imaging/sensing,² targeted delivery,³ assisted fertilization,⁴ and minimally invasive surgery.⁵ To achieve these tasks, various small scale robots have been proposed, including chemically driven propellers,⁶ biologically actuated microrobots,⁷ and flexible swimmers.⁸ However by far, magnetically actuated helical micro-robots have been the most investigated and have shown the most promise for potential use in vivo.² The advantage in using these fuel-free robots stems from their ability to efficiently convert non-invasive wirelessly transmitted signals into locomotion. Thus far there have been a number of proposed manufacturing methodologies to develop these swimmers including self-scrolling,⁹ shadow-growth,¹⁰ and direct laser writing.¹¹ While these top-down methods have been shown to be effective at producing small helical structures, they often require expensive equipment and can involve complex fabrication steps which limit their potential for large-scale mass-production. An alternative, more cost-effective method for fabrication is to use chemically or biologically self-assembled structures as templates for micro- and nano-structured helices. Thus, here we report the fabrication of biotemplated hollow porous magnetic nano-helices and demonstrate their ability to swim using rotating magnetic fields.

One of the major challenges in micro- and nanorobotics is the necessity to overcome the dominant viscous forces which prevail at a low Reynolds number. Many bacterial species, such as Escherichia coli, propel themselves in their seemingly inertialess environments by rotating chiral filaments, breaking the time reversal symmetry of Stokes flow, to generate thrust. The non-reciprocal motion generated by these microorganisms has been the inspiration for artificial helical swimmers. More specifically, it is the bacterial flagellum and the flagellar motor, which are mimicked by these robots. The natural bacterial flagellum itself is a unique self-assembled nanohelix which bacteria rotate to generate thrust.¹² Structurally, bacterial flagella consist entirely of a single sub-unit protein, called flagellin. The flagellin units are held together through non-covalent bonds and are arranged in repeated helical patterns which give the flagellum its overall shape.¹³ However, in contrast to most forms of eukaryotic flagella, the shape of bacterial flagella is not fixed; the presence of a molecular switch present with flagellin enables the flagellum to undergo discrete reversible changes in helical geometry, i.e., polymorphic transformations, in response to physical, chemical, and electromagnetic stimuli.¹⁴ The ability of the flagellum to change shape and handedness enables the flagellum to transfer stress generated during swimming that would otherwise cause the flagellum to yield.¹⁵ From an engineering perspective, polymorphism makes these natural nanostructures a promising template for the fabrication of inorganic nanohelices.

So far, there have been a number of reports on biotemplated one-dimensional micro- and nanostructures, including tobacco mosaic virus (TMV),¹⁶ M13 bacteriophages,¹⁷ Spirulina,¹⁸ and vascular plants,¹⁹ and tubular phospholipids.²⁰ These templated nanostructures have been used for a number of applications including active surface coatings¹⁶ and electrodes for batteries.¹⁷ Bacterial flagella have also been used as templates for titania,²¹ silver,²² and silica nanotubes.²³ For biological applications, flagella templated silica nanotubes are particularly interesting due to their biocompatibility and unique physiochemical characteristics, such as large specific surface area and mesoporous structure.²⁴ Combining these properties with the ability for chiral geometries to swim at a low Reynolds number presents the potential use of these structures as functional nanorobotic devices. Of critical importance, in the fabrication of helical swimmers, is the ability to finely control helical geometry; the pitch and radius of a helical swimmer to a large degree determine its swimming characteristics.²⁵ For wild-type bacterial flagella, this geometry can be controlled by modulating the ionic strength, pH, and other solvent properties.¹⁴ However, under the typical reaction conditions used in templating flagella with silica, usually only one non-helical polymorph is observed,²⁶ termed s-circular, which due to its achiral shape cannot be used for swimming. In an effort to fabricate magnetic silica templates with the nominal helical pitch observed in swimming bacteria, we describe a method to retain specific polymorphic forms and demonstrate the ability to actuate these swimmers using uniform rotating magnetic fields.

Bacterial flagella used for the templating process were obtained from a log-phase culture of Salmonella typhimurium (SJW1103). Flagella from these peritrichous bacteria were harvested and repolymerized using a modification of Asakura’s method.^14,27 Before templating, the outer surface of flagella was labeled with Cy3 dye and observed using fluorescence imaging. The transcription of amorphous silica on the surface of flagella was achieved using a modified sol-gel method that was first reported by Wang et al.²³ [Fig. 1(a)]. Details of the templating process are described in the supplementary material. Briefly, a solution of repolymerized flagella is gently mixed with an aqueous solution containing (3-aminopropyl)triethoxysilane (APTES) and then placed in an ice-water bath. In solution, the primary amine of the silane interacts, through hydrogen bonding and/or electrostatic bonds, with surface accessible proteins of the flagellar surface and subsequently undergoes hydrolysis with the amino acids to form silicic acid. Next, tetraethyl orthosilicate (TEOS) is directly added to the reaction mixture, resulting in the formation of silica as the polycondensation of TEOS occurs on silicic acid nucleating sites. This temptation process is similar to the metallization used for other templates;²⁰ however it does not require the addition of a metal catalyst (e.g., Pd) for synthesis.

During the templating process flagella undergo a polymorphic transformation from a left-handed helix to a compressed circular form. As observed from florescent imaging during this reaction process, the initial helical geometry of the flagella at room temperature is unchanged in the presence of both APTES, which acts as a weak base, and TEOS. However, at 4 °C, flagella undergo a polymorphic transformation to the circular polymorph. This transformation is presumably due to the effect of both the pH and temperature inducing a discrete conformation change in the β-harpin region of flagellin monomers.²⁸ This transformation occurs before significant silica growth, and thus the geometry of the flagella template takes the polymorphic form set by the fluid environment. To obtain flagella with helical geometries, it has previously been demonstrated that by simply adjusting pH between 4 and 10, the flagellar form can be altered;²⁹ however we observed that fixed templated polymorphs could also be achieved by using extreme pH values (>pH 10). Specifically, it was observed that by adjusting the pH of reaction solutions to 11, 11.7, and 12.1, three different templated helical forms are obtained.

After obtaining nanotubes, a thin film of ferromagnetic material was deposited on templated flagella to impart magnetic properties to the silica templates. This was done by thermal evaporation of nickel onto a monolayer of dried templates. Note that the nickel thickness was limited to approximately 10% of the thickness of silica nanotubes to limit potential magnetic aggregation. The nickel coated nanotubes were then magnetized in a 160 mT magnetic field. The structures were released from the substrate by gentle agitation in deionized water and by using low amplitude sonication.

The general morphology of the silica templated flagella was inspected optically, with scanning electron microscopy (SEM) and transmission electron microscopy (TEM). From bright field imaging of the material obtained at the three pH values, most of the templated flagella displayed morphology similar to that displayed by bare flagella at the given pH. Specifically, from florescence imaging, repolymerized flagella were observed to have left-handed helical forms with pitches of 2.23 μm and 0.81 μm, for pH values of 11 and 11.7, and a right-handed form with a pitch of 1.14 μm for pH 12.1. The polymorphic forms observed at pH 11, 11.7, and 12.1 have been termed normal, coiled, and curly, respectively.³⁰ Due to the optical transparency of silica, the polymorphic form of fluorescently labeled flagella inside of the silica nanotube can be observed (see the insets of Fig. 2). SEM images in Figs. 2(a)–2(c) show individual silica nanotubes that have an outer diameter of approximately 200 nm, although this can be adjusted by changing the amount of added TEOS.²⁶ Also, the surface of all nanotubes was rough, with a pearl-necklace-like structure, which can be attributed to the non-uniform growth of silica on silicic acid nucleating sites.²³ In addition to nanotube structures, all samples also had a significant number of silica particles/aggregates present. It is hypothesized that these particles are the result of depolymerized flagellin or short flagellar fragments being templated; it has been previously reported that above pH 10 flagella begin to depolymerize.³⁰ The observed number of silica particles increased with increasing pH, which provides some evidence of this hypothesis. However, fluorescence imaging did not show the presence of flagella/flagellin within these particles. Also, spherical silica particles, and aggregates of these particles, present in pH 11.7 and 12.1 samples were found to often obscure the presence of nanotubes. Furthermore, unlike the majority of nanotubes formed at pH 11 (normal-like morphology), many nanotubes formed at the other pH values (i.e., coiled and curly like morphologies) appeared to be linked to particle aggregates and not suitable for use as nanoswimmers. TEM imaging of the normal-like silica nanotubes shown in Fig. 2(e) reveals an inner pore, approximately 20 nm in diameter, where the flagella initially resided;²⁶ during washing, flagella depolymerized into flagellin and was removed. These silica nanotubes were also analyzed for their elemental composition through energy dispersive X-ray (EDX) spectroscopy. EDX spectra revealed 86% of the sample to be silicon, 10% carbon, 4% oxygen, by weight. The presence of carbon may be due to unremoved flagellin present within the silica nanotube.

To evaluate their swimming characteristics, nanoswimmers, approximately 5 μm in length, were exposed to low strength (<20 mT) uniform rotating magnetic fields generated by an approximate Helmholtz electromagnetic coil system.³¹ Swimming experiments were conducted in a cylindrical polydimethylsiloxane gasket (2 mm × 2 mm) sealed on both sides by no. 1 thickness glass coverslips. Templated nanoswimmers with normal-like morphology, dispersed in deionized water, were recorded at 30 frames/s as the rotational frequency of the external field (5 mT) was increased from 0 to 20 Hz. As displayed in Fig. 3, swimmers displayed a linear increase in translational velocity, which is an indication that the swimmer surface roughness did not have a significant effect on swimming motion. Also, while no apparent asynchronous rotation or step-out point was apparent in the frequencies explored, from previous reports of magnetically actuated helical swimmers,¹¹ it would be expected that such a point would be reached at higher frequencies, after which a non-linear decrease in swimming speed with increasing frequency would result. At these higher frequencies, surface roughness may lead to non-idealities in swimming motion. In comparison to flagellated swimming bacteria, such as E. coli and S. typhimurium, which rotate their flagella at frequencies on the order of 100 Hz,³² the frequencies used for templated nanoswimmer experiments were much smaller. However, fitting a line to the magnetic nanoswimmer data yields a slope of ∼0.2, and thus given enough magnetic torque, it would be expected that at 112.5 Hz rotation rate (average flagellar motor rate) swimmers would move at a linear speed of approximately 22 μm/s, which is similar to the swimming speed of flagellated bacteria in water, ∼25 μm/s.³³ The similarity in swimming speeds can be attributed to both swimmers having approximately the same helical pitch, ∼2 μm.

For application of these devices for micro- and nanoscale tasks, it is necessary to wireless steer nanoswimmers. Here positional control is demonstrated by changing the direction of rotation (θ) of the external field, which causes swimmers to turn in order to maintain the dipole alignment. Using this method, swimmers were manually controlled to move in an approximant figure-eight pattern. For steering, the frequency was held at 50 Hz and resulted in an average swimming speed of 9 μm/s, which is in agreement with the velocity profile in Fig. 3 if the data are extrapolated. Overlaid images of normal templated swimmers moving in this pattern are shown in Fig. 4. Due to the significant effect of Brownian motion, swimmers occasionally moved in and out of the focal plane, resulting in occasional out-of-focus images. Also, note that while only a single swimmer was present in the immediate viewing area, other swimmers within the sample chamber also moved in the same pattern as they were exposed to the global control input signal.

In summary, we have developed a process for generating rigid hollow helical magnetic nanoswimmers using bacterial flagella as biotemplates. These devices were prepared by transcribing silica onto the surface of repolymerized flagella and then depositing a thin layer of nickel on the silica surface. This fabrication method can be readily modified to produce swimmers with different characteristics; modulating initial TEOS concentration, pH, and amount of deposited nickel alters the swimmer thickness, helical geometry/handiness, and magnetic responsiveness, respectively. In a low strength uniform magnetic field, templated swimmers were able to act as mechanical transducers, converting rotational motion into translation, and were controllably steered. Although these helices are approximately ten times thicker than bacterial flagella (∼20 nm), the relationship between the rotation rate and linear speed is similar to that of flagellated bacteria. The approach of using bacterial flagella as templates for biomimetic swimmers is relatively inexpensive and can be easily developed for mass-production of very large populations of nanoswimmers, considering that cell densities over 10⁹/ml can be obtained in few hours, with multiple flagella generated per cell. In addition to low Reynolds number swimmers, templated flagella and their synthesis can be applied to other biologically inspired templates as well as other applications, including filtration, single molecule detection, and nanostructured electronics, such as dye-sensitized solar cells and lithium ion batteries. Furthermore, by templating flagella with biocompatible materials such as silica, these devices have significant potential to be used for biomedical applications.

See supplementary material for flagella repolymerization, silica transcription, and characterization details.

This work was funded by the National Science Foundation (DMR 1306794 and CMMI 1634726) and the Korea Evaluation Institute of Industrial Technology (KEIT), funded by the Ministry of Trade, Industry, and Energy (MOTIE) (NO. 10052980) awards to M.J.K., and with Government support under and awarded by the DoD, Air Force Office of Scientific Research, National Defense Science and Engineering Graduate (NDSEG) Fellowship, 32 CFR 168a, awarded to J.A.