Self-organized optical device driven by motor proteins

Significance

Certain nanomaterials, such as protein molecules, produce various advanced functions when incorporated into an ordered system and, therefore, have large potential for use in engineering devices. To explore how to assemble a functional device from protein components, we have tried to create a molecular device inspired by a fish pigment cell, “melanophore.” We induced ordered assembly of protein molecules through self-organization of the proteins in a specific artificial microstructure and thereby succeeded in producing a melanophore-like optical device. We believe that self-organization of molecules in microstructures can be a powerful method for assembling functional molecular systems in future nanotechnology.

Abstract

Protein molecules produce diverse functions according to their combination and arrangement as is evident in a living cell. Therefore, they have a great potential for application in future devices. However, it is currently very difficult to construct systems in which a large number of different protein molecules work cooperatively. As an approach to this challenge, we arranged protein molecules in artificial microstructures and assembled an optical device inspired by a molecular system of a fish melanophore. We prepared arrays of cell-like microchambers, each of which contained a scaffold of microtubule seeds at the center. By polymerizing tubulin from the fixed microtubule seeds, we obtained radially arranged microtubules in the chambers. We subsequently prepared pigment granules associated with dynein motors and attached them to the radial microtubule arrays, which made a melanophore-like system. When ATP was added to the system, the color patterns of the chamber successfully changed, due to active transportation of pigments. Furthermore, as an application of the system, image formation on the array of the optical units was performed. This study demonstrates that a properly designed microstructure facilitates arrangement and self-organization of molecules and enables assembly of functional molecular systems.

Within a cell, motor proteins work as mechanical components that efficiently convert chemical energy to mechanical energy. Major motor proteins, such as myosin, kinesin, and dynein, travel unidirectionally along specific filamentous protein polymers, actin filaments, or microtubules, using the chemical energy derived from ATP. Although the action of motor proteins itself is rather simple, they are involved in numerous functions in living cells such as cell division, muscle contractions, ciliary beating, and melanophore color changes (1). These diverse and elaborate functions are realized through highly ordered molecular systems that consist of not only the motor proteins but also various types of protein molecules. For example, myosin and actin form alternatively arranged bundles with tens of other proteins to construct aligned sarcomeres, the basic units of the muscle, which produce efficient contractions under strict Ca²⁺ regulation (1). Likewise, in the cilium or flagellum, dynein molecules are integrated into the “9 + 2” arrangement of microtubules and generate oscillatory bending (1). Thus, diversity of in vivo functions of motor proteins is achieved by the variety of manners in which motor proteins are organized into specific higher order systems.

In the last decade, remarkable progress has been made in the applications of motor proteins in microscale and nanoscale engineering, which has enabled the control of motor protein movements and the transport of artificial objects by motor protein (2–14). These microtransportation systems are expected to be a shuttle for micrototal analysis systems and other simple tools (15–17). To fully use the potential of motor proteins in artificial systems, it is necessary to develop higher-level functional systems. However, simply mixing protein components rarely forms an ordered system, and, therefore, organizing motor protein molecules with other associated proteins into highly ordered structures is a key bioengineering challenge. To explore the methods required to create such systems, we decided to create unique optical devices, inspired by fish melanophores.

Some species of fish, such as killifish and zebrafish, change their skin color depending on their surroundings (Fig. 1 A and B). This camouflage phenomenon is related to the pigment cells, “melanophores,” that exist on the fish skin. The cell has a radial array of microtubules that elongate from the center; the minus ends of the microtubules are located at the center and the plus ends are located along the periphery (18). Along these microtubule networks, specific motor proteins deliver black pigment granules, “melanosomes,” and alter their distribution. When cytoplasmic dynein (a minus end-directed motor) is activated, the pigment granules are transported to and concentrated at the center of the cell, and the melanophore becomes transparent. In contrast, activation of kinesin II (a plus end-directed motor) induces dispersion of the pigments, which darkens the melanophore (19–21) (Fig. 1 C–F and Movie S1).

Fig. 1.

Molecular system of fish camouflage and design of an optical system. (A and B) Zebrafish in a dark and a bright field. (Scale bar: 1 cm.) (C and D) Microscopic images of the skin of a zebrafish. Pigment cells (melanophores) change the color of the skin by dispersion (C) or aggregation (D) of pigment granules (melanosomes). (Scale bar: 50 µm.) (E and F) The mechanisms of color change of a melanophore. In a melanophore, motor proteins transport pigment granules along radially arranged microtubules, which induces dispersion (E) or aggregation (F) of pigments. (G) Design of an optical system mimicking a fish melanophore. With the support of a microstructure, the protein components form a melanophore-like ordered system in a self-organizing manner. (H) The concept of a protein-based display. In an array of melanophore-like optical units, activation of a specific group of units produces a picture.

In this study, using a microstructure that supports the arrangement and self-organization of protein components, we created a melanophore-like optical device, “artificial melanophore” (Fig. 1G). The combination of microelectromechanical systems (MEMS) technology and self-organization of the proteins enables the easy fabrication of thousands of identical artificial melanophores. We tried to create images on the array of artificial melanophores by controlling the pigment distribution of each unit (Fig. 1H).

Results and Discussion

Design and Fabrication of Supporting Microstructures.

A key structure of a melanophore is the radial array of microtubules surrounded by the cell membrane. To construct this microtubule structure, we referred to a previous microtubule-arranging method, in which microtubules were elongated from short microtubules (microtubule seeds) fixed on a glass surface (22). When the microtubule seeds are fixed within a small spot area, the spot should function as an artificial microtubule-organizing center (MTOC). By growing microtubules from those MTOCs, radially arranged microtubules can be obtained (23, 24). We have adopted this strategy and fabricated microchambers that included a scaffold to fix microtubule seeds at the center (details of fabrication methods are available in SI Materials and Methods).

By reference to the size of fish melanophores, we designed a microchamber hexagonally surrounded by partition walls whose radius and height were 26 and 7 µm, respectively, and photolithographically fabricated the chambers of thick photoresist SU8 on a glass surface (Fig. S1A). To create scaffolds for seeds, we overlaid the removable photoresist on the chamber structures and executed oxygen plasma etching through circular patterns of overlaid photoresist (Fig. S1 B–D). Removing the overlaid photoresist, we eventually obtained hydrophobic microchambers, each containing a hydrophilic circular scaffold area (5.5-µm radius) at the center (Fig. 2 A1 and B and Fig. S1E; the hydrophilic scaffold area cannot be discerned in Fig. 2B). The difference between hydrophilic and hydrophobic surfaces was used to control the selectivity of protein attachment to the surfaces in the later processes.

Fig. 2.

Fabrication of a radial microtubule array in a microchamber. (A) Fabrication process of a radial microtubule array (side view of a wafer). (A1) Partition walls of the microchambers were photolithographically fabricated on a glass surface. The fabricated walls formed hexagonal chambers and each chamber had a hydrophilic pattern (seeding zone) at the center. (A2) Nonmotile mutant kinesin (blue dots), which serves as an anchor between the surface and the seeds, was selectively attached to the seeding zone in the presence of a surfactant. (A3) Microtubule fragments (red dots) were fixed via the anchor molecules as seeds of microtubule assembly. (A4) Microtubules (green lines) were grown from seeds, which made a radial microtubule array in the chamber. (B) An SEM image of the fabricated microchambers. (Scale bar: 10 µm.) (C) Microtubules (green lines) elongated from the microtubule seeds (yellow dots). Note that the microtubules were grown only from one end (the plus end) of short microtubules. (D) A histogram of the length of the microtubules that were polymerized for different times. (E) The time course of the length (the average and SD) of microtubules in tubulin polymerization. (F) Microscopic images of a radial microtubule array fabricated in a chamber. T93N kinesin-CFP (blue; Left), microtubule seeds (red; Center Left), a microtubule array (green; Center Right), and a merged image (Right) are shown. Hexagonal fluorescent pattern is the autofluorescence of photoresist. (Scale bar: 10 µm.)

Assembly of a Radial Array of Microtubules.

We prepared microtubule fragments (seeds) by polymerizing tubulin with guanosine-5′-(α,β-methyleno)triphosphate (GMPCPP), which promotes nucleation of tubulin but inhibits the elongation of microtubules (25). Because the microtubule fragments could hardly get attached to the glass surface of circular hydrophilic patterns due to their negative charge, we used a nonmotile mutant kinesin T93N, which binds to both microtubules and hydrophilic surfaces without motility (26). By including a surfactant (0.1% Brij35), which selectively blocks protein attachment to hydrophobic surfaces (4), we were able to attach the T93N kinesin molecules selectively to the circular hydrophilic patterns (Fig. 2 A2 and F, Left, and Fig. S1F). We subsequently introduced microtubule seeds and anchored hundreds of them on the circular hydrophilic patterns via the mutant kinesin molecules (Fig. 2 A3 and F, Center Left, and Fig. S1G). We will refer to the circular hydrophilic patterns as seeding zone.

Next, we polymerized tubulin from the fixed seeds to assemble the radial microtubule arrays. In the ordinary tubulin polymerization method, microtubules are grown toward both plus and minus ends (27), which causes a mixed polarity in the resultant radial microtubule arrays. To avoid this problem, we polymerized tubulin in the presence of N-ethylmaleimide–modified tubulin, which inhibits microtubule growth from the minus ends (28, 29) so that the microtubules elongated only toward their plus ends (Fig. 2C). Because length of microtubules was almost proportional to polymerization time (Fig. 2 D and E and Fig. S2A), we optimized the polymerization time so that the plus ends of the majority of microtubules just reach the edge of the chambers. As a result, radial arrays of polarity-arranged microtubules were successfully assembled in the chambers (Fig. 2 A4 and F, Center Right and Right, Figs. S1H and S2B, and Movie S2 and Movie S3). In a typical radial microtubule array, the number of microtubules was 200–500, as estimated from the fluorescence intensity of the microtubule seeds. The length of the microtubules was about 15–20 µm. Most of the microtubules (about 70%) were elongated approximately horizontally (<30°) and were contained in the chamber.

Preparation of Motor-Associated Pigment Granules.

Previously, many researchers have succeeded in transporting artificial microobjects, such as polystyrene beads, Au- or Q-dots, or photolithographically fabricated microstructure, with motor protein physically absorbed or chemically linked to the surface of them (3, 6, 7, 10, 13). However, our system required all granules to be transported along a microtubule toward its minus end without detachment, which necessitated developments of unique, robustly motile granules. This prompted us to focus on the complex of the axonemal dynein and the microtubules. As reported previously, flagellar outer-arm dynein molecules are aligned on a microtubule in a self-organizing manner with the interval of 24 nm (30, 31). Furthermore, the dense cluster of dynein molecules on a microtubule enables the complex to travel a long distance along another microtubule (>10 µm) toward the minus end in the presence of ATP (32, 33). For example, ∼40 dynein molecules can be mounted linearly even on 1-µm microtubule fragments and interact with a track microtubule cooperatively. We thought such dynein–microtubule complexes could be used as robustly motile granules suitable in our system (Fig. 3A).

Fig. 3.

Transportation of the pigment granules along the radial microtubule array. (A) Illustration of the pigment transportation on an elongated microtubule. Dynein molecules transport a pigment (fluorescent microtubule fragment) toward the minus end of the microtubule (toward the center of the chamber). (B) Fluorescence images of pigment attachment to a radial microtubule array. A radial microtubule array (Left) and pigments attached to the array (Center) and a merged image (Right) are shown. (C) Time-lapse microscopic image sequence of transportation of pigments in a radial microtubule array. ATP was generated by photolysis of caged ATP, which was included in the chamber, to activate the motor protein dynein. The number in each picture represents the time (in seconds) after the UV flash. (Scale bar: 10 µm.)

We prepared fluorescently stained microtubule fragments from fluorescent tubulin as “pigments.” We mixed the microtubule fragments with crude axonemal dynein extracted from Chlamydomonas. At this point, outer-arm dynein molecule should be associated with a microtubule at its ATP-dependent motor stalk head and/or at its ATP-independent stem (31). To dissociate the ATP-dependent binding, we added ATP to the mixture of dynein and microtubules. The resulting complexes should interact with a track microtubule at the motor domain of dynein molecules.

Immediately after the removal of the ATP by ATPase activity of apyrase, we introduced the motile pigment granules to the chambers containing a radial microtubule array. The pigments were randomly attached to the arranged microtubules, mimicking the basic structure of a melanophore consisting of a radial microtubule array, motor proteins, and pigment granules (Fig. 3B and Fig. S1I). In our optimized condition, about 0 to three pigments were attached to each elongated microtubule, and the total number of pigments in a single chamber was estimated to be about 500–1,000. Despite the high density of arranged microtubules at the center, the distribution of pigments was not much biased probably because the high density of microtubules prevented the pigments from diffusing into the dense array.

Color Pattern Change in a Melanophore-Like System.

To activate the dynein-associated pigments without exchanging the solutions, we added photoreleasable ATP (0.5 mM caged ATP) to the buffer. When ATP was released by 350-nm UV light irradiation, almost all of the pigment granules were transported toward the seeding zone at the center of the chambers along the radially arranged microtubules, although some pigments were motionless or dissociated from the microtubules (Fig. 3C and Movie S4). Velocity of pigment transportation was about 13 µm/s, and almost all motile pigments aggregated in the seeding zone within 10 s. The aggregated pigment granules continued to move in random directions within the seeding zone as long as ATP existed, which presumably resulted from repetition of transportation, dissociation, and reassociation of pigment granules in the dense microtubule network. When ATP was exhausted by dynein and apyrase, pigments were immobilized in the seeding zone with the rigor cross-bridges between dynein molecules and microtubules. As a result of pigment aggregation, distribution of fluorescence was changed in the chambers, and the brightness was greatly increased only at the center of the chambers (Fig. S3). We call this system “artificial melanophore.”

Portrayal of Pictures on an Array of Optical Units.

As an application of the artificial melanophore, we tried to create an image display device. We fabricated a honeycomb array of about 7,500 artificial melanophores on a 4 mm × 4 mm area of a glass surface. In this array, each melanophore chamber, activatable with a UV flash, acts as one pixel of a display.

We aimed to create an image on the melanophore array by irradiating selected artificial melanophores with UV light (Fig. 4A). For that purpose, we first prepared several photomasks through which UV was flashed (Fig. S4A). When an artificial melanophore array was exposed to patterned UV through a mask, the mask pattern was copied on the array transiently, but the picture was rapidly blurred presumably due to diffusion of released ATP (Fig. S4 B, D, and F). Therefore, we covered the top of chambers with pentadecane oil and obtained an array of artificial melanophores that were separated from each other (Fig. S1J). Using this array, we succeeded in changing the color patterns only in the chambers that were exposed to UV (Fig. 4B, Figs. S1 K and L and S4 C, E, and G, and Movie S5). Thus, we were able to accurately copy the mask pattern, such as simple graphics or English letters, on the screen of the artificial melanophores (Fig. 4C and Fig. S5 A and B).

Fig. 4.

Image portrayal on an array of artificial melanophores. (A) Schematic demonstrating the technique used to display a picture on an array of artificial melanophore chambers. When the UV light was flashed through a patterned photomask on a melanophore layer, which contained caged ATP, the color patterns in the chambers exposed to UV were changed, thereby making a copy of the mask pattern on the melanophore screen. (B) The boundary between the UV-exposed area and the UV-unexposed area of an artificial melanophore array. The pigment granules aggregated only in the UV-exposed area (Right) because the chambers were sealed with oil to prevent the ATP from diffusing. This figure is a magnified image of a part of C. (C) Biodisplay composed of thousands of artificial melanophore chambers. Each 50-µm chamber functioned as a pixel, and the two colors in the chambers created the pictures on the screen (Fig. S5 A and B). The picture was captured as ∼300 contiguous images and was subsequently reconstituted (Materials and Methods or SI Materials and Methods), and the figure image has “seams” as the boundaries between capture images. (Scale bar: 1 mm.)

Contrast of Images Displayed on the Artificial Melanophore Array.

The contrast between the bright and dark pixels of displayed images depended on several factors, related to microtubule arrays and pigment granules that composed each artificial melanophore (Fig. S6). Regarding a microtubule array, the number and length of aligned microtubules were important. A low number (<100 per chamber) of microtubules resulted in paucity of attachment of the pigment granules, which attenuated the color change of a chamber. To avoid this problem, we had to fix a sufficient number of microtubule seeds on the seeding zone by adjusting the concentration and the size of the seeds. In addition, the appropriate length of microtubules was required so that the array just fits in the chamber (25-µm radius). On one hand, too short microtubules generated areas devoid of pigments along the periphery of the chamber. On the other hand, when microtubules were too long, some pigments lost motility, presumably because long microtubules protruded into the oil area. Although we were able to control the average length of microtubules by changing the time of polymerization, the size of the seeding zone (5.5-µm radius) created an additional problem. The distance from the periphery of a chamber to the near edge of the seeding zone was 19.5 µm, whereas that to the far edge was 30.5 µm, so that the optimum length of a microtubule varies depending on where and to which direction it elongates within the seeding zone. Smaller seeding zones have less of this problem but result in fewer attached seeds and, consequently, fewer microtubules. The size of the current seeding zone (5.5 µm) was chosen as a compromise between those two factors that influence the contrast of the images. The number of pigment granules also significantly affected the clarity of images: not only the paucity of pigments but also their excess degraded the contrast of the picture image. This is because too many pigment granules were not contained within the seeding zone after aggregation, which resulted in poor convergence of brightness. The current number (0 to three pigments per microtubule), achieved by adjusting the concentration of pigments and incubation time for attachment, was optimized for the 5.5-µm seeding zone.

Our present optical system has three main areas for improvement of image quality, which, in fish melanophores, are properly solved. One is that the total amount of brightness in each chamber does not change much before and after the aggregation of pigments, attenuating the contrast of images. In fish melanophores, the total amount of brightness changed significantly by the aggregation of pigment granules because the color change is generated by movements of granules that absorb light, rather than those that emit light. Hence, the development of nonluminous, light-absorbing “pigment” in its original sense should greatly improve the image quality in our system. Second, a lower density of arrayed microtubules in the peripheral region of a chamber is reducing the magnitude of color changes in that area. In fish melanophores, network of actin filaments is used to fill the gaps between the microtubules, along which pigments associated with myosin V are transported (34, 35). Although it is not easy to incorporate actin–myosin motile system in our current system, it may be feasible to form similar microtubule networks that fill the gaps between arrayed microtubules. For example, γ-tubulin adds branched microtubules on existing microtubules (36), which will capture pigments in the peripheral region and support their transport to the seeding zone when ATP is supplied. Third, the ratio between the area of seeding zone and the total chamber area should affect the extent of color change. The ratio in our system (10–15%) was similar to or smaller than that in fish melanophores (10–25%). Therefore, the superior performance of the fish melanophores indicates that factors mentioned above have greater impact on the performance than the area ratio.

Further Sophistication of Our Optical Device.

In this study, we successfully assembled an optical device from protein components combined with MEMS technologies. However, there are a number of ways for further sophistication in terms of assembly methods and the design.

We assembled the radial microtubule arrays by polymerizing tubulin from artificial MTOCs. We think that this is one of the most useful ways to assemble microtubule arrays. In fact, the aligned microtubules had the same polarity as that of living cells and functioned well as a fixed track for transportation. In addition, the number, the length, or the position of microtubules is controllable to some extent. However, our method consists of several distinctive steps and is not very simple to perform. A possible alternative to assemble microtubule arrays is to induce self-organization of microtubules with oligomers of motor protein. Nédélec et al. (37) reported that a mixture of microtubules and kinesin oligomers tethered with tetrameric streptavidin could be dynamically arranged into a radial microtubule array in a microchamber. Furthermore, in this system, motor oligomers aggregate at the center of the microtubule array during the assembly processes. Therefore, simply associating pigment granules with the motor oligomers may enable the motile system to accomplish color change. This is certainly a simple and attractive alternative approach that is worth pursuing in the future.

As to the system design, reversibility is the most important subject to be challenged. Our artificial melanophore system can display any monochrome image but cannot show a second image after erasing the first image. To make this display reversible, we need to add a system to redisperse pigment granules. In a fish melanophore, a pigment granule is associated with not only dynein, but also kinesin and myosin, which enables bidirectional transport of pigments in the network of cytoskeleton and thereby accomplishes reversible color change of the cell (19, 34, 35). Although it is currently difficult to disperse pigments actively using motor protein, which requires independent regulation of multiple motor proteins within the chamber, Brownian diffusion can also be used to disperse pigments. For example, our pigment granules should be detached from track microtubules by addition of ATP analogs, such as ADP-vanadate or adenosine-5′-(β,γ-imido)triphosphate (AMPPNP), because binding affinity between an axonemal dynein molecule and a microtubule significantly decreases in the presence of those compounds (38). If the concentration of those chemicals can be changed by external stimuli such as light, temperature, or magnetic field, pigment granules will be dispersed by diffusion throughout the chamber within 10 min. Thus, the system will be reset to the original state and become ready to show a second image.

Potentials of Molecular Assembly Supported by Artificial Microstructures.

In living organisms, functional molecular systems are assembled in a self-organizing manner involving many regulatory proteins. However, it is extremely difficult to artificially build protein-based functional systems through similar processes because the biological assembly process is complicated and is not yet adequately understood. Hence, we have induced arrangement and self-organization of protein molecules with the support of an artificial microstructure. In this method, once the supporting microstructures are prepared, functional molecular systems can be created by simple addition of components. We believe that molecular assembly supported by artificial microstructures will enable us to fabricate a variety of micromachines in broad engineering fields.

Materials and Methods

Details on experimental materials and methods are presented in SI Materials and Methods. A summary is given below.

Partition wall of the hexagonal chambers was fabricated by photolithography of negative photoresist SU8-5 (MicroChem). Hydrophilic circular patterns located at the center of the chamber were created by O₂ plasma etching through a temporal patterned layer of S1818 photoresist (Rohm and Haas Electronic Materials). The gene of T93N mutant human conventional kinesin was obtained using QuikChange mutagenesis kit (Stratagene), and was expressed in Escherichia coli Rosetta (DE3) pLysS (Novagene). Dynein was extracted from the flagellar axoneme of Chlamydomonas reinhardtii 137c in 0.6 M KCl buffer. Tubulin was extracted from porcine brains by cycles of polymerization and depolymerization, and was purified using phosphocellulose P11 (Whatman). Purified tubulin was stained with 10 mM reactive fluorescent dye carboxytetramethylrhodamine (Invitrogen; C1171) or Alexa Fluor 488 (Invitrogen; A20000). The tubulin was polymerized in buffer including 1 mM GMPCPP and was stabilized with 10 µM Taxol to obtain microtubule fragments (short microtubules). Artificial melanophores were fabricated in the microchambers by sequential introduction of protein components (see Assembly of a Radial Array of Microtubules and Preparation of Motor-Associated Pigment Granules). UV light, which was used to photolyze caged ATP and activate the artificial melanophores, was flashed through an objective lens (UPlanFLN60X or UPlanSApo4X; Olympus) from a mercury arc lamp equipped with an electronic shutter. The change in fluorescence of the artificial melanophores was observed using an inverted microscope (IX71; Olympus) equipped with a highly sensitive color CCD camera (MLX, nac). The pictures displayed on the 4 × 4 mm array of the artificial melanophores were captured as 300 contiguous images and were subsequently reconstituted.