2020, Vol.49, No.10

This review highlights the latest research on protein-based smart microtubes (MTs) and nanotubes (NTs) as ultrasmall biomaterials that are designed for use in biochemical and biomedical applications. These practical cylinders were prepared using an alternate layer-by-layer assembly of proteins (including enzymes and antibodies) and oppositely charged poly(amino acids) into a track-etched micro/nanoporous polycarbonate (PC) membrane with the subsequent dissolution of the template. The multilayered tube structure resembles a rolled sandwich. Various functionalities can be introduced into the inner surface, cylindrical wall, and outer surface. The inner surface-modified NTs can capture nanoparticles (NPs), viruses, and bacteria in capillaries. The tubules containing enzymes and AuNPs in stratiform walls act as microreactors for hydrolysis and polymerization. Moreover, MTs with PtNPs or catalase as an interior surface are self-propelled in an aqueous H2O2 solution by jetting O2 bubbles from the open-end terminus. These swimming tubes efficiently captured Escherichia coli. like a micromop. Avidin-coated MT motors caught biotinylated particles. Enzyme-coated swimming tubes accelerated catalytic reaction by their stirring motion. Urease-interior MTs swim slowly without bubble ejection. A perspective of the practical use of protein-based MTs and NTs with a good biofriendly nature is also described.

Micrometer- and nanometer-scale cylindrical hollow structures that are made of soft materials, i.e., microtubes (MTs) and nanotubes (NTs), exhibit several benefits compared to spherical particles (Figure 1). First, tubes have different interior and exterior surfaces that are independent. Therefore, it is possible to assign multiple tasks to three separate structural parts, i.e., inner surface, cylindrical wall, and outer surface. Second, tubes have open-end termini, which are useful for delivery applications. Guest molecules can be readily loaded and released without structural change. Third, tubes have long circulation persistence in vivo. Discher et al. have reported that polymer tubes (8-µm length) remained in the bloodstream of rats up to one week, which was approximately 10 times longer than that of spherical particles.1 Many investigators have synthesized tubular architectures by self-assembly techniques.28 Template-assisted synthesis using a micro/nanoporous membrane is an ingenious procedure for preparing uniform hollow cylinders.924 The outer diameter (O.D.) and tube length (T.L.) can be precisely modulated according to the pore diameter and thickness of the membrane used. Specifically, the alternate layer-by-layer (LbL) assembly technique, which involves a multilayer build-up onto the channel surface, allows to create various smart MTs and NTs that are composed of many combinations of soft materials.1024 A typical example is polymer cylinders made of oppositely charged synthetic polyelectrolytes that interact via electrostatic interaction.1015 Proteins, polypeptides, DNAs, RNAs, and polysaccharides are naturally occurring polyelectrolytes that show versatile biochemical reactivity. Thus, protein MTs and NTs have attracted considerable attention owing to their potential applications1322 such as drug delivery, biomolecular separation, and enzymatic reactions. Martin et al. have first reported the synthesis of glucose oxidase (GOD) NTs using a hard porous Al2O3 membrane.16 Each protein layer was cross-linked by glutaraldehyde, and the template was dissolved by chemical etching with 5% phosphoric acid. Unfortunately, many tubes collapsed during the membrane dissolution process. Li et al. have prepared (cytochrome c/PSS)5 NTs [PSS: poly(styrenesulfonate)] using an Al2O3 membrane, which should be also dissolved under acidic conditions.17 In 2010, we reported an efficient synthesis of protein NTs comprising a combination of negatively charged human serum albumin (HSA, Mw: 66.5 kDa) and positively charged poly-l-arginine (PLA, Mw: ca. 70 kDa) using a track-etched nanoporous polycarbonate (PC) membrane (Figure 2).22,25 The template was dissolved in N,N-dimethylformamide (DMF), and the liberated tubes were freeze-dried to yield a lyophilized powder that was readily dispersed in water. The multilayered tube structure resembles a “rolled sandwich”. An important benefit of this wet-template-assisted synthesis using alternate LbL assembly in a PC membrane is that one can easily design each layer by changing the filtration material. Specifically, the one-dimensional (1-D) pore space interior of the tube can be tailored by the deposition of the last filtered component. Using this method we generated a series of protein MTs and NTs and characterized their capabilities in molecule capturing, virus trapping, bacteria trapping, enzyme reaction, and polymerization. Furthermore, we developed self-propelled protein MT motors with biocatalytic propulsion. Structural and functional diversity of our protein cylinders is a major advantage over self-assembling protein NTs68 and naturally occurring NTs.2629 In this review, we describe the recent results from our research on protein-based smart MTs and NTs as ultrasmall biomaterials designed for use in biochemical and biomedical applications. Our studies constitute a new chemistry of protein architectures that may allow development of new biodevices.

We established an efficient synthesis procedure of protein MTs and NTs using the LbL deposition technique with a micro/nanoporous PC membrane (Figures 2A, B).25 Typically, aqueous solutions of positively charged PLA and negatively charged HSA were alternately filtered through a track-etched PC membrane (400-nm pore diameter) (Figure 2A). Three-cycle depositions of each component conferred (PLA/HSA)3 multilayers onto the pore walls. PLA was exploited as an electrostatic glue for the mutual cohesion of protein layers. The adsorbed materials on the top and bottom surfaces of the PC template were mechanically removed using a cotton swab with water. The subsequent dissolution of PC template in DMF and freeze-drying of the precipitates yielded small pieces of thin films. These pieces were composed of arrays of (PLA/HSA)3 NTs (HSA NTs) with 407 ± 13-nm O.D. and 50 ± 4-nm wall thickness (W.T.), as determined by scanning electron microscopy (SEM) observations (Figure 2C).25 The maximum T.L. (ca. 9 µm) corresponded to the PC membrane pore depth. When the NT-embedded PC membrane was immersed into the DMF solution, the PC framework instantly dissolved. The tube damage in DMF, which is a polar amide solvent, was negligible compared to that in other possible reagents such as CH2Cl2.30

We proposed a six-layered cylinder model, in which each HSA layer had a single-protein thickness (Figure 2B). This model was based on the general principle of LbL membrane growth. The negatively charged HSA binds to the positively charged surface of the PLA layer via electrostatic attraction and changes the surface charge to negative. Subsequently, the next PLA solution makes a new cationic surface on the HSA layer. The average thickness of a PLA/HSA bilayer in HSA NTs is 16.7 nm. Assuming that the dimensions of HSA are 8 nm,31,32 the PLA layer thickness is 8.7 nm. Normally, less filtration, (PLA/HSA)2, was insufficient to obtain robust NTs.

The lyophilized HSA NTs powder was suspended in a sodium phosphate buffered (PB) solution (pH 7.0), yielding a slightly turbid dispersion. To observe the morphology and stability of HSA NTs in water, the aqueous dispersions were freeze-dried under reduced pressure. The SEM measurements of the resultant powder showed that all tubules swelled considerably.25 Their W.T. doubled (104 ± 5 nm) compared to that of the original shape (Figures 2D). Of note, the direction of swelling was toward the inside of the hollow, and the O.D. remained unchanged. Thus, a drastic reduction in the inner diameter (I.D.) was observed. These morphologies were stable for 24 h. The average thickness of a PLA/HSA bilayer in a swollen state was estimated to be 33 nm; thus, the PLA layer thickness was 25 nm. The swelling ratio of the PLA layer was estimated to be 2.5, which was similar to published values for general synthetic polyelectrolytes (1.2–4.0).33,34

In addition, we performed scanning force microscopy (SFM) measurements of HSA NT arrays.35 Although the height of NTs (∼9 µm) is considerably high for the SFM tip and cantilever, the open-end terminus of the hollow cylinders can be clearly visualized (Figure 2E). The images show round chimney-like architectures.

Ligand Capture and Nanoparticle Trap.

HSA is the most prominent plasma protein in human blood, and it acts as a transporter or depot of insoluble endogenous and exogenous compounds such as fatty acids, hemin, bilirubin, thyroxine, metal ions, and a broad range of drugs.36,37 Uranyl ion (UO22+), a widely used negative staining agent for transmission electron microscopy (TEM), strongly binds to the subdomain IIB of HSA.38 The TEM observations of stained HSA NTs using uranyl acetate yielded positive images of the cylindrical walls (Figure 2F), which implied that UO22+ bonded to NTs.25 Remarkably, the inner wall of the cylinder was clearly dark, which indicated that the sixth and most accessible layer was constituted of HSA. Moreover, we determined that a cyanine dye, Zn-porphyrin, and a fatty acid were also bound to the HSA components in the cylindrical wall of HSA NTs.25

To promote the biospecific capturing ability of protein NTs, we introduced avidin (Avi, Mw: 68 kDa) as the last layer of the tubular wall. Avi from egg white binds four biotins with the highest affinity of any known protein (K > 1015 M−1).39 We prepared (PLA/HSA)2PLA/PLG/Avi NTs (Avi NTs) using the same procedure (Figure 3A).25 Because Avi is a basic glycoprotein (isoelectric point: 10.0–10.5) and has a positive net charge at pH 7.0, an anionic poly-l-glutamic acid sodium salt (PLG: Mw: ca. 70 kDa) was deposited on the fifth PLA layer to fix the final Avi layer. SEM images showed the formation of uniform cylinders with 403 ± 10-nm O.D., 59 ± 5-nm W.T., and ca. 9-µm T.L.

As a ligand for these hybrid protein NTs, we exploited a biotinylated fluorescein (bFL). The fluorescence intensity of an aqueous bFL solution disappeared after incubation with Avi NTs.25 The bleaching of color was clearly observed, which indicated that bFL was bound to the Avi layer. The binding ratio of bFL and Avi was estimated to be ca. 3.4/1 (mol/mol). Biotinylated fluorescent nanoparticles (bFNPs) were also captured in the 1-D pore space interior of Avi NTs (Figure 3A).25 The fluorescence intensity of the PB solution containing 100-nm bFNPs decreases after incubation with Avi NTs, which suggested that bFNPs were incorporated into the hollows. In contrast, a decrease in fluorescence was negligible after incubation with 250-nm bFNPs.25 This difference showed that larger bFNPs could not enter the tubes because the swollen NT’s I.D. (ca. 200 nm) was smaller than the particle diameter. The average number of 100-nm bFNPs captured inside the tube were calculated to be 161. This suggests that the occupancy ratio of the particles in the volume was 29%.

Virus Trap.

Hepatitis B virus (HBV) infection is a serious global health problem. Despite the existence of vaccination, 350 million people suffer from chronic HBV worldwide.40 Infectious HBV, so-called Dane particle (DP, 42-nm diameter), comprises a core nucleocapsid containing genome DNA covered with an envelope, which consists of a hepatitis B surface antigen (HBsAg, Mw: ca. 25 kDa). If one can synthesize a unique nanotubular trap for infectious DP, it would have an important impact not only on bioseparation chemistry but also on medical applications.

We prepared HSA-based NTs with an HBsAg antibody (HBsAb) as the last layer of the wall, using a nanoporous PC membrane (400-nm pore diameter), (PLA/HSA)2PLA/PLG/HBsAb NTs (HBsAb NTs, Figure 3B).41 The dimensions of the tube were observed by SEM measurements (414 ± 16-nm O.D., 59 ± 4-nm W.T., and ca, 9-µm T.L.). To evaluate the DP-trapping capability of HBsAb NTs, DNA quantification assays of supernatants were conducted by PCR. As expected, DNA concentrations considerably decreases after incubation with HBsAb NTs.41 The trapping ratio reached 99.9%. We reasoned that DPs were perfectly ensnared into the 1-D pore space of NT. TEM observation demonstrated the entrapping of DPs into the hollow (Figure 3B). The removal efficiency by a single HBsAb NTs treatment reached a −3.9log order.

Another challenging target is influenza viruses. Among the three types of influenza viruses (A, B, and C), type A is a virulent pathogen that induces a severe disease.42 In 2009, a swine-origin influenza A (H1N1) virus spread worldwide and caused a human pandemic. The influenza A virus infection is initiated by the specific binding of viral hemagglutinin to a cellular surface receptor, containing sialyloligosaccharide with N-acetyl neuraminic acid (Neu5Ac) terminals.43 Fetuin from fetal calf serum (Mw: 48.4 kDa) is a heavily glycosylated protein with tribranched oligosaccharides including Neu5Ac residues.44 Influenza viruses are well known to interact with this natural glycoprotein.45

We prepared HSA-based NTs with an innermost layer of fetuin using a nanoporous PC membrane (800-nm pore diameter), (PLA/HSA)5PLA/fetuin NTs (Fetuin NTs).46 SEM measurements demonstrated the formation of homogeneous hollow cylinders with 811 ± 13-nm O.D., 97 ± 3-nm W.T., and ca. 15-µm T.L. For trap experiments, we exploited the influenza A virus [Puerto Rico/8/34 (PR8, H1N1)] with a diameter of 94 ± 15 nm. The I.D. of Fetuin NTs is sufficiently large to accommodate PR8 virions. The virus trapping capability of NTs was examined by the direct enzyme-linked immunosorbent assay of the remaining PR8. The results showed that the PR8 solution treated with Fetuin NTs became completely virus-free.46 We inferred that influenza A virus PR8 diffused into the hollow space of NTs and bonded to the inner surface wall on the basis of specific binding of hemagglutinin on the virus surface to the sialyloligosaccharide chain of fetuin. The removal efficiency by a single treatment with NTs was on the −5.0 log order.

These two astonishing results for HBsAb NTs41 and Fetuin NTs46 provoke the development of a new field of nanometer-scale virus detection and removal devices. For example, the elimination of small viruses (hepatitis E viruses and human parvo B19 viruses) and corona viruses (SARS-CoV-2) that caused pandemic in 2020 would be of tremendous medical importance.

Bacteria Trap.

The next challenge is the trapping of a living organism. Escherichia coli (E. coli), a rod-shaped gram-negative bacterium (0.4–0.7-µm width, 2–4-µm length), is the smallest organism of the micrometer-scale world. Many strains are harmless, but some serotypes (e.g., enterohemorrhagic E. coli. O157) can induce severe poisoning in humans.47 (PLA/HSA)9 MTs (HSA MTs) were fabricated using a PC membrane (1.2-µm pore diameter).48 SEM measurements revealed the formation of uniform hollow cylinders with 1.0 ± 0.02-µm O.D., 148 ± 5-nm W.T., and ca. 25-µm T.L.

Then, we measured the E. coli capture capability of HSA MTs. The dimensions of E. coli K12 were as follows, 425-nm width and ca. 2–3-µm length. The colony incidence of the sample solution after mixing with MTs was assayed using the standard plate count method. There were fewer colonies appearing on the plate (Nc) compared to those of identically treated E. coli without the tubes.48 Surprisingly, Nc became completely zero after 30 min of mixing with HSA MTs; the disappearance yield reached 100%. However, incubation with slender HSA NTs exhibited no change in the colony number. We reasoned that E. coli entered the pores of swollen MTs (ca. 500-nm I.D.), but they were too large to enter the narrow NT’s channel (ca. 200-nm I.D.). The efficiency of removal by a single HSA MT treatment was over the −7log order.

The incorporation of E. coli into MTs was confirmed by confocal laser scanning microscopy (CLSM). The CLSM images of the mixture solution of fluorescein-labeled HSA MT and CTC (5-cyano-2,3-ditolyl tetrazolium chloride)-labeled E. coli demonstrated that CTC-E. coli (red) was loaded into the 1-D pore space of the MT (green) (Figure 3C).48 It is remarkable that the edge of E. coli appeared from the mouth of the tube in most cases, thereby forming a “bacterium-corked MT”. Because the E. coli width is only slightly smaller than the pore size, it cannot enter deep inside the tubule.

Interestingly, E. coli trapped in the HSA MT was metabolically inactive. After 1 h of mixing, almost all E. coli lost cell proliferation ability.48 The E. coli growth can be prevented by stopping cell deviation. In the narrow space of MTs, bacteria cannot reproduce through asexual reproduction by binary fission. This remarkable result will serve as a trigger to make novel bacteria removing devices.

HSA NTs and α-Glucosidase NTs.

HSA possesses a weak esterase activity for hydrolysis of 4-nitrophenylacetate (4NPA).36 The reaction occurs in the subdomain IIIA of HSA. If HSA components in HSA NTs retain their original property, the tubules are also able to show the same enzyme activity. First, we evaluated an esterase activity of HSA NTs.35 After adding 4NPA to the NT dispersion, absorbance increased at 440 nm, which was based on the product, 4-nitrophenol (4NP), absorption. Therefore, we suspected that 4NPA diffused into the cylindrical wall and was hydrolyzed to generate 4NP. The enzyme activity parameters [Michaelis constant (Km) and catalytic constant (kcat)] for the esterase activity of HSA NTs were similar to those of free HSA (Table 1).

Table 1. Enzyme activity parameters of protein MT and NT reactors
Table 1. Enzyme activity parameters of protein MT and NT reactors
  Substrates Km/mM kcat/s−1
HSAa 4NPA 0.81 0.05
HSA NTa 4NPA 1.26 0.05
αGDb MUGlc 0.18 16
αGD NTb MUGlc 0.2 0.42
GODc βGlc 13.6 300
GOD1 MTc βGlc 13.4 328
GOD2 MTc βGlc 13.8 158
GOD3 MTc βGlc 14.7 141

aIn PBS dispersion (pH 7.4, 22 °C) (ref 35). bIn 100-mM PB dispersion (pH 7.0, 22 °C) (ref 50). cIn 10-mM PB dispersion (pH 7.0, 25 °C) (ref 51).

α-d-glucosidase from Saccharomyces cerevisiae (αGD, Mw: 68.5 kDa) is a typical exo-type carbohydrase, which is used to cleave an α(1→4) glucosyl linkage in the non-reducing terminal of the substrate.49 We evaluated the enzyme activities of multilayered protein NTs with an αGD interior surface, (PLA/HSA)2PLA/αGD NTs (αGD NTs) (413 ± 17 nm-O.D., 52 ± 3-nm W.T., and ca. 9-µm T.L.).50 In an aqueous medium, NTs captured a fluorogenic glucopyranoside, 4-methylumbelliferyl-α-d-glucopyranoside (MUGlc), into their 1-D pore space and hydrolyzed the substrate with a release of α-d-glucose (αGlc). A non-competitive-type inhibition was observed in the enzyme parameters.50 The probability of four subunits of adhered αGD facing toward the aqueous solution is half or one-fourth on the curved inner wall. This statistic geometry may be one of the reasons for the lower enzyme activity.

Glucose Oxidase NTs.

To elucidate the relation between enzyme activity and stratified wall structure (enzyme layer number and position) in detail, we prepared three different MTs, in which the GOD layer was introduced as the innermost layer, intermediate layer, or all-internal protein layers.51 (i) After the 7.5-cycle LbL growth of PLA/HSA, GOD was deposited onto the fifteenth PLA layer, which provided (PLA/HSA)7PLA/GOD MTs (GOD1 MTs). (ii) GOD was introduced into the eighth layer, which yielded (PLA/HSA)3PLA/GOD(PLA/HSA)4 MTs (GOD2 MTs). (iii) Moreover, using GOD as all-internal protein layers produced (PLA/GOD)7PLA/HSA MTs (GOD3 MTs). All tubules exhibited almost the same morphologies (ca. 1.2-µm O.D., ca. 135-nm W.T., and ca. 23-µm T.L.).

Upon the addition of β-d-glucose (βGlc) into the tube dispersion containing peroxidase and o-dianisidine, a color change was clearly visible, and absorbance at 460 nm increased.51 Three-types of MTs catalyzed the oxidation of βGlc to β-d-glucono-1,5-lactone and H2O2. We inferred that βGlc molecules diffused into the swollen tube wall and were oxidized by the GOD components. It is remarkable that the Km and kcat values of GOD1 MTs were equivalent to those of the GOD solution (Table 1).51 The immobilized enzymes on the interior surface wall retained their original catalytic activity. In contrast, non-competitive-type inhibition was observed in GOD2 and GOD3 MTs. A decrease in kcat values is attributed to the low diffusion of the substrate in the cylindrical wall and blocked ligand sites by upper and lower PLA layers. We concluded that the performance of the innermost GOD layer was almost identical to that of a free GOD molecule; however, the catalytic activity of the intermediate layer declined by ca. 40–50% compared to that of monomeric GOD. Overall, enzymatic reactions, which occurred in protein MTs and NTs, are affected by several factors such as diffusion of substrates in the 1-D pore space and positions and geometries of the enzymes in layered walls.

Lipase NTs for Polymerization.

There are no studies reporting enzymatic polymerization in NTs. If a nanotubular reactor for polymer synthesis were to be constructed, it would have a great impact on advanced material chemistry. Kobayashi et al. first reported ring-opening polymerization of lactones catalyzed by lipases in 1993.52 They determined that immobilized Candida antarctica lipase B (CalB, Mw: 33 kDa) was effective for polymerization.53 We prepared HSA-based NTs with an inner wall composed of CalB, (PLA/HSA)2PLA/CalB NTs (CalB NTs) (393 ± 9-nm O.D., 47 ± 5-nm W.T., and ca. 9-µm T.L.) and investigated their catalytic performance toward the ring-opening polymerization of lactone.54 12-Dodecanolactone (DDL) was added to an aqueous dispersion of CalB NTs. The mixture was continuously stirred at 22 °C for 24 h. Then, the resultant suspension was through to collect the white solid product. MALDI-TOF mass spectroscopy demonstrated the formation of oligomers with molecular weights of 638–2816.54 The degree of oligomerization was calculated to be 3–14. Of note, the molecular weights of oligoDDL produced in aqueous CalB NTs were higher than those prepared in the CalB solution. The superior catalytic performance of CalB NTs may be due to the incorporation of a water-insoluble DDL monomer in the hollow. The contact area of the interface between the monomer droplet and the CalB wall increases in the cylinder, which accelerated the oligomerization process.

Hybrid NTs with Inorganic Materials.

Metal NPs are also exploited as a layered wall component. Hybrid NTs combine the merits of both biomolecules and inorganic nanomaterials with the potential for widely diverse applications. We chose gold nanoparticles (AuNPs) because they have received a considerable interest owing to their catalytic activity as well as tunable optical and electronic properties.55,56 Hybrid NTs were prepared by an alternate LbL assembly of a AuNP and HSA admixture ([HSA]/[AuNP] = (1 mg/mL)/(0.9 mg/mL) and PLA into a track-etched PC membrane (400-nm pore diameter) (Figure 4).57 SEM images showed the formation of uniform hollow cylinders of (PLA/AuNP-HSA)3 NTs (AuNP-HSA NTs) with 426 ± 12-nm O.D., 65 ± 7-nm W.T., and ca. 9-µm T.L.) (Figure 4A). TEM and energy dispersive X-ray (EDX) measurements supported the high loading of AuNPs in the tubular wall (Figure 4B).57 The core–shell AuNP-HSA corona, in which the protein is bound to the NP surface, is essential for robust NT formation (Figure 4C). AuNP-HSA NTs serve as a heterogeneous catalyst for the reduction of 4NP with sodium borohydrate.57 The superior performance of AuNP-HSA NTs is probably attributed to the more homogeneous distribution of AuNPs in the swollen HSA layers. We reasoned that HSA in NT equally dispersed AuNP units in the tube wall and concentrated the 4NP substrate in the cylinder from the bulk aqueous solution.

Interestingly, the calcination of AuNP-HSA NTs at 500 °C in air yielded red solid NTs that were composed of thermally fused AuNPs.57 The tubular structure remained intact, although the shrinkage of morphology was observed: 195 ± 10-nm O.D., 41 ± 4-nm W.T., and ca. 4-µm T.L. (Figure 4D). A detailed inspection revealed that the cylindrical wall consists of numerous AuNPs with a diameter of ca. 13 nm. AuNP NTs showed the same catalytic activity; however, the efficiency was lower than that of AuNP-HSA NTs.

Moreover, we demonstrated the template-assisted synthesis of AuNP-HSA cylinder arrays that were covalently attached to the glass surface.57 Stiff AuNP-HSA NTs allowed us to create a “protein NT forest” on the planar substrate (Figure 4E). The NT density (7.9 × 107 tubes/cm2) showed good agreement with the porosity number of the PC membrane (7.9 × 107 pores/cm2).

Solid NTs comprising hematite nanoparticles (α-Fe2O3NPs) can be prepared from an iron-storage protein ferritin (Mw: 460 kDa).58 The initial (PLA-ferritin)3 NTs (Ferritin NTs) precursors were fabricated using a track-etched PC membrane (400-nm pore diameter). The obtained uniform cylinders (410 ± 14-nm O.D., 61 ± 5-nm W.T., ca. 9-µm T.L.) were calcined at 500 °C in air, which yielded reddish-brown iron oxide NTs. The hollow structure remained perfect, but its O.D., W.T., and T.L. were considerably smaller (211 ± 8 nm, 29 ± 3 nm, and ca. 5 µm, respectively). The disappearance of the protein shell and PLA layers was confirmed using infrared and EDX spectroscopies. Subsequent SEM, TEM, and X-ray photoelectron spectroscopy showed that tubular walls comprise fine α-Fe2O3NPs with a 5-nm diameter. Iron oxide NTs demonstrated superparamagnetic properties with a blocking temperature of 37 K and efficient photocatalytic activity for the degradation of 4-chlorophenol.58

Autonomous Propulsion.

Self-propelled MT motors have attracted considerable interest during the last decade.5969 Many metal-based capillaries containing an internal wall of a Pt thin film have been fabricated by rolling-up processing with photolithography or templating synthesis with electrochemical deposition. A disproportionation reaction of H2O2 (2H2O2 → 2H2O + O2) occurs on the Pt surface with the subsequent continuous ejection of O2 microbubbles from the terminal opening. Thus, the tube gains thrust in the direction opposite to the bubble discharge. The O2 bubble expulsion is responsible for autonomous propulsion. These swimming MTs have been demonstrated to be useful for environmental applications63,64,66 such as pollutant cleaners and separation devices. Other goals of tubular motor development are practical life science applications for human health6769 such as on-demand vehicles for targeted drug delivery and elements of microdiagnostic devices. These smart MTs must have biocompatibility, which is the most important feature. However, the inorganic rigid micropipes with Pt walls are not biodegradable. As an alternative, citrate-stabilized Pt nanoparticles (PtNPs) have been widely used in the medical field as a strong quencher for O2•− and H2O2.70 The PtNP itself shows almost no cytotoxicity.71 We prepared HSA-based MTs with an interior surface composed of PtNPs by templating synthesis using a track-etched PC membrane (1.2-µm pore diameter), (PLA/HSA)8PLA/PtNP MTs (PtNP MTs).72 SEM and TEM observations revealed the formation of uniform hollow cylinders (1.16 ± 0.02-µm O.D., 147 ± 11-nm W.T., and ca. 23-µm T.L.) (Figures 5A, B). In aqueous H2O2 media (pH 7.0, 5-wt % H2O2, 1-wt % SDS). PtNP MTs are self-propelled by jetting O2 bubbles from the open-end terminus (Figure 5C, Movie S1).72 The disproportionation of H2O2 was catalyzed by the innermost PtNP layer. The continuous O2 bubble releases from one end induced self-pumping of peroxide fuel in the hollow space. The resulting liquid flow in the capillary propels the tube (Figure 5C). The swimming trajectories of PtNP MTs mostly show a turning motion, which is attributable to the asymmetrical shape of the tube’s body and mouth. The PtNP MTs velocity depended on the H2O2 concentration of 1–10 wt %, which typically was 273 ± 26 µm/s (12.3 body lengths/s) in 5-wt % H2O2. By adding a magnetite nanoparticle (Fe3O4NP) layer in the tube wall, the direction of MT movement can be manipulated to become a straight motion using magnetic field guidance.72

Because the exterior surface of MTs is positively charged, negatively charged bacteria would be adsorbed on the tube. We exploited a genetically engineered E. coli in which green fluorescent protein (GFP) was secreted. Fluorescent E. coli was added to the self-propelled PtNP MT dispersion. PtNP MTs swam in the E. coli suspension by spouting out O2 microbubbles. After 15 min, the tubules were collected, and the remaining free E. coli was lysed to obtain a clear GFP solution. The fluorescence intensity of GFP in the lysate was markedly low (only 1%) compared to the value of the identically treated lysate without the tubes.72 The 99% removal of E. coli was achieved by single treatment with self-propelled MTs. The moving motion is essential for accelerating the contact frequency to catch the bacteria. Our results showed that swimming MTs captures E. coli efficiently and function as an “automatic micromop”.

Transportation, Separation, and Stirring.

To provide additional functionality to swimming MTs, it is required to use the outer surface of the cylinder. Nevertheless, a shortcoming of the wet-template synthesis lies in the difficulty of changing the exterior surface. The outside wall, which is composed of the initial filtration materials, must be PLA in our procedure. To overcome this difficulty, we produced an electrostatic LbL coating of released MTs with Avi protein in bulk aqueous solution.73 Avi–biotin complexation allows swimming MTs to capture various biotinylated substances.

We dispersed PtNP MTs in a PB solution and electrostatically wrapped them with Avi by LbL coating, which yielded Avi/PLG(PLA/HSA)8PLA/PtNP MTs (Avi/PtNP MTs, ca. 1.2-µm O.D. and ca. 24-µm T.L.).73 Avi/PtNP MTs were self-propelled in an H2O2 solution by the ejection of O2 bubbles with an average velocity of 141 ± 8.7 µm/s (5.9 body lengths/s) (pH 7.0, 5-wt % H2O2, 0.2-wt % Triton X-100).

To confirm the exterior surface of Avi and to evaluate the selective biotin binding ability, we added biotinylated fluorescein (bFL) to the PB solution of the tubes. Swimming Avi/PtNP MTs could efficiently capture bFL and strongly fluoresced.73 The swimming motion can be observed using optical microscopy under fluorescence mode (Figure 6A, Movie S2). Even with the coexistence of different dyes, bFL was specifically captured by Avi/PtNP MTs.

The next challenge is the transportation of nanometer- and micrometer-size particles. Upon addition of biotinylated fluorescent nanoparticles with a diameter of 15 nm (bFNP15) to the PB solution of swimming Avi/PtNP MTs, the particles immediately adhered to the tubes’ exterior surface (Figure 6B).73 Swimming bFNP15-bound Avi/PtNP MTs emitted red fluorescence. The larger bFNP100s (100 nm) and bFMP4s (4 µm) were also caught by the tubular motors.73

Enzyme immobilization on the outer surface would allow swimming tubes to have a biocatalytic activity. Its self-propulsive motion is expected to facilitate the enzyme reaction. We again used the αGD enzyme. Biotinylated αGD (bαGD) was added to the PB solution of Avi/PtNP MTs yielding αGD-covered Avi/PtNP MTs [bαGD/Avi/PLG(PLA/HSA)8PLA/PtNP MTs (bαGD-Avi/PtNP MTs)] (ca. 1.2-µm O.D. and ca. 24-µm T.L.) (Figure 6B).73 After the addition of MUGlc to the swimming tube dispersion, the blue fluorescence of the product, 4-methylumbelliferon (MU), was visible. The enzyme activity of bαGD-Avi/PtNP MT motors was approximately 30% of that of free bαGD.73 This restricted activity likely occurred because αGD molecules densely adhered to the exterior surface with statistic geometries. The result is consistent with that of αGD NTs with an enzyme interior surface.50 Our results demonstrated that self-propelled protein MTs act as ultrasmall transporters, removers, separators, and stirrers.

Catalase-powered MT Motors.

Catalase (Cat, Mw: 240 kDa) is a common hemoprotein enzyme that is capable of decomposing H2O2 with high efficiency in antioxidant defense systems. Several successful experiments related to Cat-powered polymer MTs have been reported,7476 although they still show poor biocompatibility. If one were able to immobilize Cat on the inner surface of protein MTs, the obtained cylinders could possess both self-propulsion capability and biodegradability. We prepared protein MT motors containing Cat as an interior surface, (PLA/HSA)7PLA/PLG/Avi/bCat MTs (Cat MTs).77 Unfortunately, the direct LbL synthesis of designed tubes failed. Therefore, we employed a two-step preparation using Avi–biotin interaction with biotinylated Cat (bCat). Optical microscopy revealed that the O.D. and T.L. of Cat MTs were ca. 1.2 µm and ca. 24 µm, respectively.

Upon addition of H2O2 into the aqueous Cat MT dispersion, the tubes were self-propelled by jetting O2 bubbles from the terminal opening with the average velocity of 71 ± 20 µm/s (3.0 body lengths/s) (pH 7.0, 5-wt % H2O2, 0.2-wt % Triton X-100) (Figures 7A, B, Movie S3A).77 Azide anion is known to coordinate to the heme groups of Cat and inhibit enzyme activity. Actually, the addition of NaN3 stopped the movement of tubes. The cylindrical wall of protein multilayers must be soft, although the MT structure was sufficiently stable to endure the bubble-propelled motion in water.

To our surprise, the migration of O2 bubbles was readily visible at the center of the capillary because protein walls are transparent and reasonably long (Figure 7A bottom, Movie S3B).77 Careful inspection revealed that two expulsions every 45 ms (22.2 Hz) from the tube’s mouth create one bubble in the outer aqueous phase (Figure 7C). For example, the first expulsion at 5 ms created one small bubble (ca. 3-µm diameter) in the outer aqueous phase (red arrow). The second expulsion at 50 ms enlarged this bubble to 5 µm. The third expulsion at 90 ms released the bubble (red arrow) and generated another new one (light blue arrow).

The pH and temperature greatly affect the enzyme activity. Indeed, the swimming speed of Cat MTs was affected by pH (pH 6.0–9.0). It is particularly interesting that the pH dependence of velocity exhibited a maximum peak at pH 7.0, which corresponded to the optimum pH of Cat (Figure 7D).77,78 As expected, the temperature dependence of the tube’s speed (15–50 °C) also coincided with the Cat activity curve.77 We concluded that the Cat MT motor velocity could be regulated by modulating enzyme activity with pH and temperature.

Cat MTs consist only of proteins and polypeptides. Therefore, the tubular wall can be easily decomposed by acid and protease. By adding Pronase (a cocktail of several proteases) into the Cat MT dispersion, the hollow column morphology completely disappeared.77

Urease-powered MT Motors.

Cat enzyme is a promising alternative to inorganic catalysts. However, some concerns still remain. (i) The H2O2 fuel is a powerful oxidative reagent that is toxic to cells and tissues. (ii) Gas bubble formation may be undesirable in an in vivo environment. Recently, Sánchez et al. have reported that silica NTs with an internal wall bearing urease (Ure) were self-propelled in water without bubble spouting.79 The NTs (220-nm O.D.) were driven by the turnover of the urea substrate triggered by the Ure reaction (Urea + H2O → 2NH3 + CO2). Urea is essential in the metabolism of nitrogen-containing compounds by animals. The normal range of urea is 2.1–7.1 mM in human blood and 100–450 mM in human urine.80

Using Jack bean Ure (Canavalia ensiformis, Mw, 480 kDa), we prepared HSA-based MTs containing a Ure inner wall, (PLA/HSA)7PLA/PLG/Avi/bUre MTs (Ure MTs).81 Optical microscopic observations demonstrated that the homogeneous cylinder structure was preserved after biotinylated Ure (bUre) binding (ca. 1.2-µm O.D. and ca. 24-µm T.L.). Ure MTs swam very smoothly with nonbubble propulsion in a PB solution containing physiological concentration of urea (100 mM) (Figure 8, Movies S4).81 The swimming trajectory of the tube showed a turning motion (Figure 8A, Movie S4A) or a straight motion (Figure 8B, Movie S4B). The average velocity was 2.7 ± 0.2 µm/s. The decomposition of urea occurs on the innermost Ure layer. The products NH3 and CO2 are diffused from the 1-D pore space to the bulk aqueous solution (Figure 8C). This diffusiophoresis may be responsible for autonomous propulsion. In addition, Ure MTs entirely disappeared in the aqueous solution of protease, which shows their adequate biodegradability.81

In this study, we showed that template-assisted synthesis using a track-etched PC membrane combined with the dissolution of template with DMF and freeze-drying of released cores could be used to produce various protein MTs and NTs. Multiple functions can be conferred to the three parts of the tubule, i.e., inner surface, cylindrical wall, and outer surface. The target ligand and substrate molecules can diffuse into the 1-D pore space and/or fully swollen walls, where they are captured by protein components. Specifically, the simple deposition of Avi layers as internal or external walls is useful for loading biotin derivatives. As practical applications, MTs and NTs act as size-selective traps for viruses and bacteria. The perfect elimination of influenza viruses, corona viruses, and enterohemorrhagic E. coli O157 using protein tubules is expected to be of incredible medical importance. Lipase NTs perform as nanochannel reactors for polymer syntheses. The use of PtNP and catalase as the innermost layer allows protein cylinders to swim in an aqueous H2O2 solution by jetting O2 microbubbles. Enzyme-wrapped PtNP MTs act as self-propelled biocatalysts that enhance enzyme reaction by the stirring motion. This enhancement is likely to be applicable to any enzyme. Cat MT motors were decomposed by protease, which showed perfect biodegradability. Ure MTs swam smoothly in an aqueous solution containing nontoxic urea fuel. The striking benefit of our “all-protein MTs and NTs” is their potential for immobilizing different enzymes with desired hierarchy in the cylindrical wall, which could accomplish a functional relay of sequential enzymatic reactions. It is possible to sandwich glucoamylase and GOD as intermediate layers into Cat MT; thus, starch fuel-driven MT motors can be created. Tailor-made protein MTs and NTs that are biofriendly will serve as a trigger to establish a new field in biomaterial chemistry and to create innovative ultrasmall devices for use in many biological and biomedical scenarios.

This work was supported by Grants-in-Aid for Scientific Research (B) (No. 15H03533 and No. 18H01833) from JSPS, Grant-in-Aid for Exploratory Research (No. 26600030) from JSPS, and the Science Research Promotion Fund from Promotion and Mutual Aid Corporation for Private Schools of Japan.

Supporting Information is available on https://doi.org/10.1246/cl.200433.

Teruyuki Komatsu received his Ph.D. from Waseda University under the guidance of Prof. Eishun Tsuchida in 1994. After postdoctoral research work with Prof. Jürgen-Hinrich Fuhrhop at Freie Universität Berlin as a JSPS fellow for research abroad (1995–1997), he returned to the Research Institute for Science and Engineering, Waseda University, where he was appointed as a Lecturer (1997) and was promoted to Associate Professor in 2003. From 2006 to 2010, he held additional post of a PRESTO researcher of JST. In 2010, he moved to Chuo University as a full professor of Department of Applied Chemistry. His major research interests are synthesis and development of smart biomaterials and their practical applications.