Generation of stable microtubule superstructures by binding of peptide-fused tetrameric proteins to inside and outside.

Microtubules play important roles in biological functions by forming superstructures, such as doublets and branched structures, in vivo. Despite the importance, it is challenging to construct these superstructures in vitro. Here, we designed a tetrameric fluorescent protein Azami-Green (AG) fused with His-tag and Tau-derived peptide (TP), TP-AG, to generate the superstructures. Main binding sites of TP-AG can be controlled to the inside and outside of microtubules by changing the polymerization conditions. The binding of TP-AG to the inside promoted microtubule formation and generated rigid and stable microtubules. The binding of TP-AG to the outside induced various microtubule superstructures, including doublets, multiplets, branched structures, and extremely long microtubules by recruiting tubulins to microtubules. Motile microtubule aster structures were also constructed by TP-AG. The generation of various microtubule superstructures by a single type of exogenous protein is a new concept for understanding the functions of microtubules and constructing microtubule-based nanomaterials.

INTRODUCTION

Microtubules are tubular cytoskeletons composed of tubulins, which are used as components of motile nanomaterials, such as active matters (–) and molecular robots () by complexation with motor proteins. Although the microtubules produced in vitro usually have singlet structures, various microtubule structures are formed in vivo, including structures with different lengths and protofilament numbers, multiplets (doublets or more) (), and branched structures (, ). A singlet microtubule is defined as a single hollow tube composed of protofilaments made of tubulins, while a doublet microtubule consists of a complete singlet microtubule (A-tubule) and an incomplete microtubule (B-tubule) tethered to the A-tubule. These microtubule superstructures are formed as needed to perform various functions in vivo. For instance, flagella and cilia have complex microtubule structures, including doublet and branched microtubules, and these microtubule superstructures are thought to contribute to the mechanical strength and motility of flagella and cilia (–). Microtubule-organizing centers (MTOCs), known as the centrosomes in animal cells, contain microtubule nucleators to control the nucleation, stabilization, and orientation of microtubules for the formation of complex microtubule assemblies, such as spindles and asters (, ). One of the ways that the structures of microtubules are controlled in vivo is by the binding of microtubule-associated proteins (MAPs) (). Although MAPs that bind to the outer surface of microtubules are well known, various microtubule inner proteins (MIPs) that bind to the inner surface of microtubules have recently been reported (, –). The formation of doublet microtubules is thought to occur by (i) the binding of MIPs to the inside of the complete A-tubule for induction of a stabilized formation of an incomplete B-tubule on the outer junction and (ii) the binding of other proteins, such as PACRG and FAP20, to the inner junction to close the B-tubule (). The branching of microtubules is mainly induced by centrosomes and the γ-tubulin ring complex (γ-TuRC) (, ). Branching structures arise by the nucleation of new microtubules from the sides of existing microtubules via γ-TuRC and other proteins (–).

Although complex microtubule superstructures observed in nature are supposed to have unique functions, such as mechanical strength, distinct from singlet microtubules, it is difficult to artificially construct the microtubule superstructures in vitro. Recently, the formation of doublet microtubules from singlet microtubules in vitro has been reported by deletion of the C-terminal tails of microtubules to allow the access of new tubulin (). However, this method does not allow the branching of doublet microtubules into two singlet microtubules as observed in human sperm flagella (). If these microtubule superstructures can be constructed by exogenous molecules in vitro, then they would be useful for understanding their properties/functions in vivo and as new types of microtubule building blocks for nanotechnological applications including double-track railways (). For this purpose, the formation of stable microtubules as a nucleus and the recruitment of new tubulin to the formed microtubules are thought to be important ().

As a binding motif for the internal surface of microtubules, we have previously developed a Tau-derived peptide (TP; CGGGKKHVPGGGSVQIVYKPVDL) (). TP was designed using a repeat domain of Tau that is involved in binding to the inner surface of microtubules through interaction with a taxol (paclitaxcel)–binding pocket on β-tubulin (). The binding of TP to the interior of microtubules can be achieved by the preincubation of TP with tubulin, followed by the polymerization of the TP-tubulin complex. Thus, the conjugation of exogenous molecules, such as green fluorescent protein (GFP) () and magnetic cobalt-platinum nanoparticles (), to TP can result in their encapsulation inside microtubules (). The TP motif will be useful not only to encapsulate exogenous molecules in microtubules but also to recruit tubulins to preexisting microtubules by displaying TP on the outer surface.

Here, using the tetrameric protein Azami-Green (AG), which shows fluorescence by tetramerization (), we developed AG protein fused with His-tag and TP (TP-AG) that can bind both to the inside and outside of microtubules (Fig. 1A). TP-AG induced microtubule formation and stabilization and increased the rigidity of the microtubules, enabling the formation of a variety of microtubule superstructures, including doublets, multiplets, branched structures, extremely long microtubules, and aster structures. Because various microtubule structures could be generated by one type of exogenous protein, we proposed a simple design principle for forming diverse microtubule superstructures in vitro.

Fig. 1.

Binding of TP-AG to microtubules.

(A) Schematic diagram of TP-AG. (B) Preparation of TP-AG–incorporated microtubules by the Before and After methods and (C) the confocal laser scanning microscopy images. Preparation concentrations: [tubulin] = 3.2 μM, [tubulin-TMR] = 0.8 μM, [TP-AG] = 8 μM, and [GMPCPP] = 0.2 mM. Scale bars, 10 μm. (D) TP-AG–incorporated TMR-labeled GMPCPP microtubules using subtilisin-treated tubulin prepared by the Before and After methods (left). Scale bars, 10 μm. The IAG/ITMR ratio for each type of microtubule determined from the confocal images (right). Error bars represent the SEM (N = 30). *P < 0.0001, two-tailed Student’s t test. Preparation concentrations: [tubulin] = 19.2 μM, [tubulin-TMR] = 4.8 μM, [TP-AG] = 8 μM, [subtilisin] = 1 μM, and [GMPCPP] = 0.2 mM. (E) TP-AG–incorporated GMPCPP microtubules prepared by the Before and After methods and after further treatment with a TMR-labeled anti-AG antibody (left). Scale bars, 10 μm. The ITMR/IAG ratios for each type of TP-AG–incorporated microtubule treated with TMR-labeled anti-AG antibody, determined from the confocal images (right). Error bars represent the SEM (N = 30). *P < 0.0001, two-tailed Student’s t test. Preparation concentrations: [tubulin] = 4 μM, [TP-AG] = 8 μM, [TMR-labeled anti-AG antibody] = 1.2 μM, and [GMPCPP] = 0.2 mM.

RESULTS

Design and binding analysis of TP-AG to microtubules

The tetrameric structure of AG is useful to display multiple peptides on the surface for the efficient binding to microtubules. In addition, AG (~6.1 × 6.7 nm), which is slightly larger than the tubulin monomer, is small enough to be encapsulated within a microtubule. Some microtubule-binding proteins have higher affinities for microtubules with attached His-tags because of the interactions between the His-tags and the C-terminal tail of tubulin located on the outer surface of microtubules (, ). We hypothesized that adding several His-tags to nonmicrotubule-related proteins may give the proteins the ability to interact with microtubules. Thus, we attached a His-tag to the N terminus of AG protein so that there would be four His-tags per tetrameric AG (table S1). Although AG without a His-tag does not have the ability to interact with microtubules (), the binding of His-tagged AG (referred to as AG hereafter) to the outer surface of microtubules was confirmed (see Supplementary Text). Next, we attached the TP sequences to the C terminus of AG to form TP-AG, allowing binding to the inner surface of microtubules (Fig. 1A and table S1). The formation of a TP-AG tetramer was confirmed by native polyacrylamide gel electrophoresis (PAGE) and dynamic light scattering (DLS) measurements (fig. S1, B and C). The binding of TP-AG to microtubules was evaluated by two methods: the “Before” and “After” methods (Fig. 1B). In the Before method, tubulin was first incubated with TP-AG, and then microtubules were formed by the addition of guanosine 5′-triphosphate (GTP) or guanosine-5′-[(α,β)-methyleno]triphosphate (GMPCPP). In the After method, TP-AG was incubated with preassembled microtubules. In the present study, GMPCPP was mainly used for analysis of the binding of TP-AG and the effects on the microtubule structures, and GTP was mainly used for analysis of the effects of TP-AG on the formation and stability of the microtubules. The binding of TP-AG to tetramethylrhodamine (TMR)–labeled GMPCPP microtubules was observed in both the Before and After methods by confocal laser scanning microscopy (Fig. 1C). Because the original AG showed fluorescence upon tetramerization (), TP-AG is thought to preserve its tetrameric form when bound to a microtubule. We analyzed the binding ratio of TP-AG to GMPCPP microtubules by a cosedimentation assay and found that almost all TP-AG was bound to microtubules at high concentrations of tubulin, with dissociation constant (Kd) = 11 μM in the Before method (fig. S3, A to C). When the tubulin was treated with subtilisin that digests the C-terminal tails of tubulin (, ), high binding of TP-AG to microtubules was observed in the Before method unlike for AG (fig. S2A), whereas the binding was low in the After method (Fig. 1D). In addition, the binding of a TMR-labeled anti-AG antibody to TP-AG–incorporated microtubules was low in the Before method, whereas more binding was observed in the After method (Fig. 1E). These results showed that most of the TP-AG was bound to the inner surface of microtubules in the Before method and to the outer surface of microtubules in the After method (Fig. 1B), similar to that observed for TP-conjugated GFP (). Electron microscopy (EM) also showed inclusions inside the microtubules and decoration of the outside of the microtubules in the presence of TP-AG, further supporting the binding of TP-AG to both the inside and outside of the microtubules (fig. S4). The binding of TP-AG to intracellular microtubules was also suggested (fig. S3D), indicating that TP-AG is potentially useful for applications in living cells. The binding of a monomeric version of AG (mAG) () fused with TP (TP-mAG) to microtubules was much weaker than for tetrameric TP-AG (table S1 and fig. S5, A and B). Thus, the tetrameric structure of TP-AG is important for the binding to microtubules, and the binding of TP-AG to the inside or outside of the microtubules can be controlled by changing the polymerization method.

Effects of TP-AG on the polymerization and stability of microtubules

Because GTP microtubules are typically unstable in vitro because of the hydrolysis of GTP to guanosine diphosphate, the effects of TP-AG on the polymerization and stability of GTP microtubules were analyzed. A considerable increase in the turbidity of the tubulin solution with TP-AG suggested that TP-AG caused a marked enhancement in microtubule formation, which was much more effective compared with the microtubule-stabilizing anticancer drug, taxol (Fig. 2A). When only TP was used, the turbidity was not increased as much as with taxol (). Microtubule formation was not facilitated by AG or TP-mAG, indicating that both the tetramerization of TP-AG and the binding of TP-AG to the inside of the microtubules were necessary for the induction of microtubule formation. The formation efficiency of GTP microtubules prepared by the Before method was evaluated by SDS-PAGE of the separated free tubulin and microtubules (Fig. 2B). Increased formation of microtubules was observed by the treatment of TP-AG, and the microtubule structures were partially retained even after treatment at 4°C for 90 min (depolymerization conditions). GTP microtubules with TP-AG prepared by the Before method were stable with longer lengths compared with taxol-stabilized microtubules, and only aggregates were observed after treatment with AG or without any additives (Fig. 2C). Treatment with only TP also induced the formation of aggregates (). To examine the effects of TP-AG on the physical stability of the microtubules, EM measurement was performed (Fig. 2D). In the negative-stain EM images, the microtubules with TP-AG had a straighter appearance compared with normal microtubules, which appeared to have a winding shape (Fig. 2D, left), indicating that TP-AG physically strengthened the microtubule structures. This result was confirmed by analyzing the ratio of the contour and end-to-end lengths of the microtubules (Fig. 2D, right). TP-mAG also increased the straightness of the microtubules, but the effect tended not to be as high as that of TP-AG (fig. S5, C and D). Although there are proteins known to fix the tubulin lattice into straight conformation (, , ), there have not been exogenous or native proteins that can straighten the whole microtubule structures. These results indicated that the binding of TP-AG to the inside of the microtubules formed rigid and stable microtubules and considerably promoted microtubule formation. The microtubules with TP-AG prepared by the After method also showed straighter structures compared with the normal microtubules (Fig. 2D). It is possible that the binding of TP-AG to the inner and outer surfaces had similar strengthening effects on the microtubules or that the partial binding of TP-AG to the outer surface in the Before method was enough to strengthen the microtubules.

Fig. 2.

Facilitation of the polymerization and stabilization of microtubules by TP-AG.

(A) Changes in turbidity over time caused by tubulin polymerization in the presence of TP-AG, AG, TP-mAG, or taxol at 37°C. The thick lines represent the mean values, and the colored areas show the standard errors of the mean (N = 3). OD, optical density. (B) Effect of TP-AG on the formation of GTP microtubules prepared by the Before method. Samples were prepared by polymerization at 37°C for 30 min (37°C) and subsequent depolymerization at 4°C for 90 min (4°C). SDS-PAGE results of the supernatants (S) and the pellets (P) (top). The microtubule formation efficiency as calculated by the ratio of S and P (bottom). Error bars represent the SEM (N = 3). (C) GTP microtubules prepared by the Before method with TP-AG, AG, taxol, and without any additives. Scale bars, 10 μm. The lengths of the microtubules (means ± SD, N = 30) are shown. (D) Effect of TP-AG on the straightness of GMPCPP microtubules prepared by the Before and After methods evaluated by negative-stain EM (left). Red arrowheads indicate doublet-like microtubules. Scale bars, 500 nm. Box plot of the ratios of the contour and end-to-end lengths of microtubules obtained from the negative-stain EM images (N = 100) (right). *P < 0.0001, two-tailed Student’s t test. The concentrations for each experiment are described in Materials and Methods.

Microtubule superstructures induced by TP-AG

In the negative-stain EM observations of TP-AG–treated GMPCPP microtubules, there were doublet-like microtubules and branched microtubules present, not just singlet microtubules, especially in the After method (Fig. 2D and fig. S6A). With further incubation at 37°C for 30 min in the After method, the populations of doublet microtubules (44.0%) and branched microtubules (5.6%) were increased in the negative-stain EM images compared with the microtubules without further incubation (Fig. 3, A and B). There were also regions where multiplets were formed (Fig. 3A, orange arrowheads) and the formation of multiplet and branched structures was taking place at the same time (Fig. 3A, cyan arrowhead). The doublet microtubules, branched microtubules, and multiplets were also observed by cryo-EM (Fig. 4, A and B, and fig. S6B). The structures of the doublet microtubules induced by TP-AG were similar to the doublet microtubules isolated from cilia (fig. S6C) and doublet microtubules constructed in vitro (), which consisted of complete singlet A-tubules and incomplete B-tubules attached to the A-tubules. Cryo-EM observation showed a characteristic feature of the doublet as characterized by thinner B-tubule diameter compared with the A-tubule (the B-tubule was responsible for 36.4% of the diameter of the doublet; Fig. 4, B and C). To analyze the TP-AG reconstituted doublet microtubules, the diameters of the TP-AG reconstituted doublets, TP-AG A-tubules, and TP-AG B-tubules were measured (Fig. 4C). The diameter of the TP-AG doublet was larger than that of the singlet (Fig. 4C) but thinner than that of the bundle of two singlets (fig. S4D). Furthermore, the diameter of the doublets observed in the presence of TP-AG was very similar to the diameter of native doublet microtubules from cilia (Fig. 4C). Thus, we believe that the reconstituted doublet has essentially the same structure as the native doublet microtubule. Because most microtubules (97.1%) were singlet microtubules in the absence of TP-AG and the population of the doublet microtubules with TP-AG in the Before method (9.0%) was less than in the After method (Fig. 3B), it was thought that these doublet microtubules were formed via TP-AG bound to the outside of the preassembled microtubules. From the EM images, the branching was caused by the B-tubule detaching from the A-tubule of doublet microtubules. The method of the branching was different from that caused by the nucleation factor SSNA1 in which the tubulin lattice of one singlet microtubule is separated into two singlet microtubules (). The number of branched microtubules was increased by further incubation at 37°C for 30 min, supporting the concept that the branched structures were grown from doublet microtubules under the polymerization conditions (Fig. 3B). Because the lengths of the B-tubules of the doublet microtubules were similar with (0.47 ± 0.25 μm) and without (0.40 ± 0.18 μm) incubation (Fig. 3C), when the B-tubules were partially grown, the microtubules might split into two separate singlets rather than keep growing as a doublet. Because the doublet microtubules formed by subtilisin treatment did not form branched structures (), it is possible that the remaining C-terminal tails of tubulin under the conditions used in the present study pushed the formed B-tubule off and thereby branched structures grew from the doublet microtubules. The branches had a mean angle of 4.9° ± 2.5° (Fig. 3D), and this small angle was similar to the branched microtubules that are split from doublet microtubules in cilia () rather than the branches induced by SSNA1, which have wider angles (47° ± 15°) (). In the Before method, main binding site of TP-AG is the inside but the partial binding to the outer surface could induce the doublet microtubule formation and branching. TP-mAG also induced the formation of doublet and branched microtubules; however, the efficiency was not as high as that of TP-AG (fig. S5E). Only singlet microtubules were observed when the microtubules were treated with AG only (fig. S2C), indicating that the TP moiety and tetramerization were important for the efficient formation of doublet and branched microtubules.

Fig. 3.

Various microtubule structures induced by TP-AG.

(A) Negative-stain EM images of TP-AG–incorporated GMPCPP microtubules prepared by the After method with additional incubation at 37°C for 30 min. General view (top) and selected EM images (bottom) of microtubules. Singlet microtubules (black arrowheads), doublet microtubules (red arrowheads), branched microtubules (blue arrowheads), multiplet microtubules (orange arrowheads), and branching and the formation of multiplets in the same area (cyan arrowhead) were observed. There were substructures observed within the microtubules (magenta arrowhead). Scale bars, 500 nm. (B) Quantification of microtubule types prepared by the Before method with buffer (N = 248) and TP-AG (N = 290), by the After method with buffer (N = 171), TP-AG (N = 212), and TP-AG with further incubation at 37°C for 30 min (N = 539) from the negative-stain EM images. Multiplet microtubules are counted as “Doublet” here. The percentages of microtubule types are summarized in table S5. (C) Histogram of the lengths of the B-tubules of doublet microtubules with (red) and without (yellow) incubation at 37°C for 30 min. The curves are Gaussian fits to the histogram (with incubation: 0.47 ± 0.25 μm, N = 235; without incubation: 0.40 ± 0.18 μm, N = 86). (D) Histogram of the distribution of the angles of the branched microtubules. The curve is a Gaussian fit to the histogram (4.9° ± 2.5°, N = 33). Data presented as means ± SD. Preparation concentrations: [tubulin] = 10 μM, [TP-AG] = 20 μM, and [GMPCPP] = 0.2 mM.

Fig. 4.

Cryo-EM observation of TP-AG–induced doublet and branched microtubules.

(A) Representative cryo-EM images of TP-AG–incorporated GMPCPP microtubules prepared by the After method with additional incubation at 37°C for 30 min. A general view (left) and each type of microtubule (right) are shown. Red arrowheads, doublet microtubules; blue arrowheads, branched microtubules. Scale bars, 250 nm. (B) Magnified view of the doublet indicated in (A). Red arrows indicate representative particles lined within the B-tubule. These particles are slightly larger than tubulin, possibly corresponding to TP-AG. Scale bar, 50 nm. (C) Mean diameters of singlet microtubules, TP-AG reconstituted doublets, TP-AG reconstituted A-tubules, TP-AG reconstituted B-tubules, native ciliary doublets, native ciliary A-tubules, and native ciliary B-tubules as measured from cryo-EM images. The values are as follows: singlet, 21.0 ± 1.5 nm (N = 32); TP-AG doublet, 35.3 ± 5.0 nm (N = 27); TP-AG A-tubule, 22.4 ± 3.1 nm (N = 27); TP-AG B-tubule, 12.9 ± 3.4 nm (N = 27); cilia doublet, 32.7 ± 3.0 nm (N = 28); cilia A-tubule, 19.7 ± 2.1 nm (N = 28); cilia B-tubule, 14.6 ± 3.1 nm (N = 28) (means ± SD). The statistical parameters are summarized in table S2.

To further understand the mechanisms, the formation of doublet and branched microtubules induced by TP-AG was observed by total internal reflection fluorescence microscopy (TIRFM). TMR-labeled GMPCPP microtubules (red) were immobilized on a substrate and then Alexa Fluor 405–labeled tubulin (blue), TP-AG (green), and GTP were added to induce the formation of doublet and branched microtubules (Fig. 5A). Without TP-AG, we observed elongation of blue microtubules only from both ends of template red microtubules (fig. S7). In contrast, the colocalization of red and blue microtubules was observed in the presence of TP-AG (Fig. 5, B and C), indicating that the binding of TP-AG induced the formation of doublet microtubules composed of red and blue microtubules. The branched structures composed of TP-AG and blue tubulin were also observed (Fig. 5C), and time-lapse images showed that the branched microtubules were grown from the microtubule seeds (Fig. 5D and movie S1). The branched structures were similar to that grown from doublet microtubules observed in the EM images (Figs. 3 and 4). The efficiency of the branching was higher under the conditions for TIRFM measurements, compared with those for EM measurements, probably because excess amounts of TP-AG and tubulin were added to the microtubule seeds in TIRFM. In a previous report, the digestion of the C-terminal tails of microtubules by subtilisin made free tubulins accessible to induce the formation of doublet microtubules; however, branched microtubules were not formed in this method (). In our system, the remaining flexible C-terminal tails of the microtubules were important for the formation of branched microtubules. It is plausible that TP-AG bound to the outer surface of the microtubules covered the C-terminal tails of the microtubules and then the exposed TP moieties recruited free tubulin to induce the formation of doublet microtubules (Fig. 5E). The B-tubules of the doublet microtubules would be flexible because of the presence of flexible C-terminal tails, which can stochastically induce the formation of branched microtubules. The tetrameric structure of TP-AG with four His-tags and four TPs is possibly important for the efficient binding to the outer surface of microtubules and the recruitment of tubulin.

Fig. 5.

Observation of the formation of TP-AG–induced doublet and branched microtubules.

(A) Visualization of TP-AG–induced doublet and branched microtubules using TIRFM. TP-AG, blue tubulin, and GTP was added to the red microtubule seeds, and the formation of doublet and branched microtubules was observed. (B) General view and (C) selected TIRFM image of microtubules. The inset (dashed line) indicates the location of (C). (D) Time-lapse images showing the growth of branched microtubules observed by TIRFM. The blue arrowheads indicate the grown branched microtubules. The images were cropped from movie S1. Scale bars, 20 μm (B) and 5 μm (C and D). (E) Model for the formation of doublet and branched microtubules with TP-AG.

The effect of TP-AG on GTP microtubules was also assessed by negative-stain EM. When 10 μM tubulin was used, only aggregates were observed in the absence of TP-AG (Fig. 6A). Increasing the amount of TP-AG in the Before method increased the amount of microtubules formed (Fig. 6, B to F). When the amount of TP-AG was low, most of the microtubules were singlet microtubules, whereas complex structures, such as doublet microtubules, multiplet microtubules, and branched microtubules, were observed by increasing the equivalents of TP-AG (Fig. 6, G and H). The formation of doublet microtubules may require a certain level of stabilization of the microtubules because it has been reported that doublet microtubules were formed only when stable GMPCPP microtubules were used and not when unstable GTP microtubules were used (). In our system, first TP-AG binds to the inner surface of microtubules to stabilize singlet microtubules, and then additional TP-AG binds to the outer surface to induce the formation of doublet microtubules. When the ratio of tubulin to TP-AG was increased to 1:4, extremely long and winding microtubules were observed (Fig. 6H). This result was consistent with the confocal images in which TP-AG–encapsulated microtubules were longer than taxol-stabilized microtubules (Fig. 2C). These extremely long microtubules are thought to be formed by a combination of the formation of doublet (or multiplet) microtubules, splitting of the doublet, and the weak bundling activity of TP-AG. This complex microtubule structure is thought to lead to the high stability of the TP-AG–treated microtubules.

Fig. 6.

Extremely long microtubules induced by TP-AG.

(A to F) Representative negative-stain EM images of GTP microtubules treated with TP-AG by the Before method with increasing amounts of TP-AG as indicated above. The top row shows lower magnification with the carbon hole, and the bottom row shows higher magnification EM images. Without TP-AG, 10 μM tubulin alone did not yield microtubules (A). With a lower concentration of TP-AG, most of the microtubules were singlet microtubules (black arrowheads). When higher concentrations of TP-AG were added, more complicated microtubule structures were formed, including doublet microtubules (red arrowheads) and multiplet microtubules (orange arrowheads). The inset (dashed line) indicates the location of (G). Scale bars, 10 μm (top images) and 500 nm (bottom images). (G) Detail of multiplet microtubule formation. Even within one microtubule, there were both singlet areas (black arrowheads) and doublet areas (red arrowheads). Scale bar, 500 nm. (H) Example of a branched microtubule. The microtubule was prepared with two equivalents TP-AG as in (E). EM image of one continuous microtubule structure with overlapping areas. A singlet microtubule (black arrowhead) was split into two singlets at the locations indicated by the blue arrowheads. These two singlet microtubules were intertwined together (orange arrowheads). Scale bar, 500 nm.

Motile properties of TP-AG–incorporated microtubules

The in vitro motility of microtubules driven by adenosine 5′-triphosphate (ATP) on a kinesin-coated substrate is fundamental property of microtubules for constructing dynamic systems (–). To investigate the effects of TP-AG to the motility of microtubules, the gliding properties of microtubules were evaluated (Fig. 7A and movie S2). The velocity of TP-AG–incorporated GMPCPP microtubules prepared by the Before method was increased (1.3-fold) compared with normal microtubules. The persistence length (Lp) of the microtubules, an indicator of the rigidity, was also increased (3.3-fold) with TP-AG, which corresponded with the negative-stain EM images (Fig. 2D). Because rigid GMPCPP microtubules move faster than flexible GTP microtubules (), it was assumed that the increased rigidity of the TP-AG–incorporated microtubules was responsible for the enhanced velocity.

Fig. 7.

Motile properties of TP-AG–incorporated microtubules and formation of aster-like structures.

(A) Effect of TP-AG on the velocity and persistence length (Lp) of GMPCPP microtubules prepared by the Before method on a kinesin-coated substrate. Fluorescence microscopy image of moving microtubules using a temporal-color code (left). Histogram of the velocity of the microtubules (center, N = 120). The curves are Gaussian fits to the histogram. Lp of microtubules estimated using the curves that are derived from Eq. 3 in Materials and Methods (right, N = 120). Data presented as means ± SD. The difference in velocities and Lp of microtubules treated with TP-AG and buffer was found to be statistically significant (P < 0.0001, t test). Preparation concentrations: [tubulin] = 6.4 μM, [tubulin-TMR] = 1.6 μM, [TP-AG] = 0.8 μM, and [GMPCPP] = 0.2 mM. See also movie S2. (B) Fluorescence microscopy image of moving aster structures of TP-AG–incorporated GTP microtubules prepared by the Before method on a kinesin-coated substrate using a temporal-color code. To visualize the movement of microtubules in (A) and (B), the time series images were color coded and superimposed on one another. The color scales are shown at the bottom of the image. (C) Time-lapse images showing the motility of the aster structures of TP-AG–incorporated GTP microtubules. Preparation concentrations: [tubulin] = 4 μM, [tubulin-TMR] = 1 μM, [TP-AG] = 1 μM, and [GTP] = 1 mM. See also movie S3. Scale bars, 20 μm.

Microtubule aster structures have been constructed using microtubule nucleating structures, such as DNA origami () and beads coated with antibodies to aurora kinase A, which plays a key role in microtubule nucleation (, ). Because TP-AG has microtubule polymerization activity, we further explored whether there are conditions under which TP-AG can generate the aster-like microtubule assembly. The formation of aster-like structures composed of the GTP microtubules treated with 0.2 equivalents of TP-AG to tubulin was observed in the Before method (Fig. 7, B and C, and movie S3). At the center of the asters, a strong accumulation of TP-AG was observed, and the centers showed no motility. Accumulated microtubules with less TP-AG at the periphery of the asters showed motility that was released and translocated away from the aster. This radially organized motility from the asters was similar to that observed with the asters from MTOCs reconstituted from Xenopus sperm (). It is possible that the microtubules with accumulated TP-AG in the present study serve as microtubule nucleating centers to trap microtubules with less TP-AG, followed by the release of the trapped microtubules from the center. To our knowledge, this is a first example that a single kind of exogenous protein works as a nucleating center. When an increased amount of TP-AG was added, aster structures were not formed, and accumulated structures with no motility were formed instead (fig. S8). Thus, we were able to form different microtubule superstructures by balancing the content of TP-AG–incorporated and non–TP-AG–incorporated microtubules.

DISCUSSION

Using a simple design based on fusing a His-tag and TP to tetrameric AG, we succeeded in stabilizing microtubules and inducing the formation of various microtubule superstructures, including doublets, multiplets, branched microtubules, and extremely long microtubules, by binding TP-AG to both the inside and outside of microtubules. In addition, the formation of MTOC-like motile aster structures was achieved. The obtained microtubule superstructures were largely controlled by the polymerization conditions, including the polymerization type (Before or After), the nucleotides (GTP or GMPCPP), and the molar ratios as summarized in table S5. Notably, the addition of excess TP-AG and tubulin to microtubule seeds induced the branching structures in an efficient manner (Fig. 5). Introducing both a His-tag and a TP sequence to the tetrameric scaffold (TP-AG) gave rise to functions that were not achieved using the His-tag and TP sequence with the monomeric form (TP-mAG) or by His-tag in the tetrameric form (AG). Therefore, we were able to tune the design of the protein to accomplish specific purposes. These results provide design guidelines for the recruitment of exogenous proteins to microtubules and the construction of microtubule superstructures. In the future, modification of TP-fused proteins could be used to induce specific microtubule superstructures. For instance, the binding motifs to the C-terminal tails of microtubules, the protein scaffolds, and the linkers between TP and the protein scaffolds could affect the binding affinity of the protein toward microtubules and the orientation on the microtubules. Optimization of these parameters will be useful to construct specific microtubule superstructures in a highly efficient and controlled manner. In the current method, the binding of TP-AG is roughly regulated to inside and outside by the polymerization method. In future, this should be strictly regulated by further modification of the method.

Here, we discuss some possible applications of our system in biophysical studies and nanobiotechnology. The straightening of whole microtubule should be useful for the study of MAPs that sense the curvature of microtubules, such as doublecortin (). The straightening of microtubules could aid the structural analysis of singlet microtubules with other microtubule binding proteins using cryo-EM. Straight microtubules are also useful as the components of motile nanomaterials. Because the morphology of the microtubules affects the cooperative assembly behavior of the microtubules (), our stable and straight microtubules will be useful for the construction of microtubule-based active matters and molecular robots. In addition, straight microtubules could be used as markers in a similar way as actin filaments have been used to observe the rotation of F1–ATPase (). Our method does not require the severing of C-terminal tail of the tubulin; therefore, the method should be compatible with different posttranslational modifications of the C terminus of tubulin. Thus, the reconstitution of doublet microtubules with different posttranslational modifications on the A- or B-tubules may be possible, similar to the native doublet microtubules from cilia (). These reconstituted doublets will be useful to study the mechanisms of intraflagellar transport that use doublet microtubules as a natural double-track railway (). Recently, branching DNA origami has been used as a track for the sorting of cargos by motor proteins (). By combining the posttranslational modification of tubulins with our branched doublet, it may be possible to sort cargos using different motor proteins that have different affinities toward the posttranslational modifications (). Microtubules have been previously used as templates to fabricate metal nanowires (, ), and our microtubule superstructures may be used as such templates to construct metal nanowires, including branched and long forms. The high microtubule polymerization activity of TP-AG surpasses that of taxol, a well-known anticancer drug. TP-AG could be used as an anticancer drug that prevents cell division by stabilizing the cellular microtubules. In addition, the generation of microtubule superstructures in living cells and tissues may be used to modulate cellular structures and functions.

Induction of the microtubule superstructures consisting natural tubulins with the C-terminal tails by exogenous proteins provides important insights into the formation principles of doublets and branched structures in vivo, such as in flagella and cilia. In our system, TP-AG generated doublet microtubules similar to the native doublet structures. Therefore, the formation of doublet and branched microtubules with TP-AG in the presence of the C-terminal tail is a model of the formation of doublet and branched microtubules in vivo. TP-AG is thought to generate doublet microtubules by (i) stabilizing the microtubule structure, (ii) suppressing the C-terminal tails of tubulin by His-tags, and (iii) recruiting tubulin molecules to the surface of the preassembled microtubules by TPs. In vivo, different types of MIPs may work together to stabilize microtubules and shield the C-terminal tail of tubulin and recruit tubulins to generate doublet microtubules. Notably, we were able to achieve this result using a single type of protein. In vivo, the loss of certain MIPs near the tip of the cilia leads to loss of this balance and hence the doublet was split into two singlets as has been previously observed ().

In conclusion, we have developed a new tool that can be used to generate different microtubule structures by varying the conditions. Our tool has possible applications in both material and biological sciences. Our work will be useful for the design of more sophisticated tools that can be tuned to construct a specific microtubule structure in the future.

MATERIALS AND METHODS

Equipment and materials

Ultraviolet-visible (UV-vis) spectra were obtained using a Jasco V-630. Ultracentrifugation was performed using an Optima MAX-TL ultracentrifuge (Beckman Coulter) using TLA 120.2 rotor. Tubulin was purified from porcine brain by a reported procedure (). Tubulin-TMR and tubulin–Alexa Fluor 405 were prepared following a standard protocol (). Recombinant kinesin-1 consisting of the first 573 amino acid residues of human kinesin-1 was prepared according to a reported procedure (). In the SDS-PAGE analysis, the proteins were mixed with 2× Laemmli buffer, heated at 95°C for 5 min, and then loaded onto SDS-PAGE gel (15% acrylamide). In the native PAGE analysis, the proteins containing 20% glycerol were loaded onto native PAGE gel [6% acrylamide, 60 mM tris, and 40 mM N-cyclohexyl-3-aminopropanesulfonic acid (pH 9.4)], and electrophoresis was carried out at 200 V (constant voltage) in a buffer containing tris and N-cyclohexyl-3-aminopropanesulfonic acid (pH 9.4). The DLS measurements were carried out using a Zetasizer Nano ZS (Malvern) instrument with an incident He-Ne laser (633 nm) at 25°C. Correlation times of the scattered light intensities G(t) were measured several times, and the means were calculated for the diffusion coefficient. Hydrodynamic diameters of the scattering particles were calculated using the Stokes-Einstein equation. The reagents were purchased from Watanabe Chemical Ind. Ltd., Tokyo Chemical Industry Co., Dojindo Laboratories Co. Ltd., Wako Pure Chemical Industries, and Sigma-Aldrich and used without further purification. For negative-stain EM, holey carbon girds (U1013, EMJapan Co. Ltd.) were prehydrophilized using an E-1010 ion sputter (Hitachi). Negatively stained grids were observed in an H-7100 electron microscope (Hitachi) equipped with AMT XR41 digital CCD (charge-coupled device) camera (Advanced Microscopy Techniques). For cryo-EM, holey carbon grids (Quantifoil R2/1 grid: Cu 300 mesh, M2951C-1-300) were hydrophilized by a JEC-3000FC Auto Fine Coater (JEOL) with a target of aluminum. Vitrification of the grids were performed using a Vitrobot Mark IV (Thermo Fisher Scientific), and vitrified grids were observed in a Glacios cryo–transmission electron microscope (cryo-TEM) (Thermo Fisher Scientific) equipped with Falcon III direct electron detector (Thermo Fisher Scientific) at SPring-8. Confocal laser scanning microscopy measurement was carried out using a FluoView FV10i (Olympus). For TIRFM measurement, samples were illuminated with a 15-mW Nikon Ti3 laser source and visualized using a three-color TIRFM (Eclipse Ti2, Nikon) equipped with an oil-coupled Plan Apo 60× 1.40 objective (Nikon). The 401-nm (15% intensity), 488-nm (10% intensity), and 561-nm (5% intensity) lines of the Nikon Ti3 laser were used to observe the tubulin–Alexa Fluor 405, TP-AG, and tubulin-TMR, respectively. Images were captured using a cooled complementary metal-oxide semiconductor (CMOS) camera (Neo CMOS, Andor) connected to a PC. In the motility assay, samples were illuminated with a light-emitting diode light source and visualized using an epi-fluorescence microscope (Eclipse Ti2-E, Nikon) using an oil-coupled Lambda S 60× objective [numerical aperture (NA) 1.4] (Nikon). For the observation of aster structures, samples were illuminated with a 100-W mercury lamp and visualized using an epi-fluorescence microscope (Eclipse Ti, Nikon) using an oil-coupled Plan Apo 60× objective (NA 1.4) (Nikon). Concentrations of TP-AG and AG are described as monomer concentrations.

Construction of plasmids

The TP-AG, AG, and TP-mAG expression vectors were constructed on the basis of the pET-29b(+) expression vector (Merck, Darmstadt, Germany). The TP-AG, AG, and TP-mAG genes were synthesized to optimize codon usage for Escherichia coli. The open circular pET-29b(+) vector was synthesized by inverse polymerase chain reaction (PCR). PCR was performed in the TaKaRa PCR Thermal Cycler Dice Touch (TaKaRa, Shiga, Japan) using the Tks Gflex DNA Polymerase (TaKaRa). The synthetic genes coding TP-AG, AG, and TP-mAG were ligated into the open circular pET-29b(+) using the In-Fusion HD Cloning Kit (TaKaRa) to generate the recombinant protein expression vectors. The plasmids were sequenced by Eurofins Genomics (Tokyo, Japan).

Expression and purification of proteins

The pET-29b(+) vectors coding TP-AG, AG, and TP-mAG were transformed into E. coli BL21(DE3) strain by a heat-shock procedure. Bacterial cells were spread on Luria-Bertani– ampicillin (LBA; 100 μg/ml) agar and grown overnight at 37°C. A single transformant colony was grown in LBA medium at 37°C overnight. The culture was diluted 100-fold by addition to fresh LBA medium and grown at 37°C until an absorbance of 0.5 was noted at 600 nm (midlogarithmic phase), and then the culture was incubated with 0.1 mM isopropyl β-d-1-thiogalactopyranoside at 20°C. After 18 hours of incubation, cells were harvested by centrifugation at 8000 rpm for 10 min. The cell pellets were suspended in Ni-affinity binding buffer [50 mM tris-HCl (pH 7.4), 150 mM NaCl, and 20 mM imidazole] on ice. The cells were lysed by sonication. After centrifugation at 13000 rpm for 10 min, the supernatant was loaded onto 1 ml of Ni-affinity column (GE Healthcare). After washing with the same buffer and the storage buffer [50 mM tris-HCl (pH 8.0) and 300 mM NaCl], the protein was eluted from the column using the Ni-affinity elution buffer [50 mM tris-HCl (pH 8.0), 300 mM NaCl, and 250 mM imidazole]. The eluted sample was dialyzed against the storage buffer [Spectra/por7; cutoff molecular weight (Mw), 8 kDa; Spectrum Laboratories Inc.) at 4°C. The purity of proteins was evaluated by SDS-PAGE (fig. S1A). The concentration of proteins was determined using reported molar extinction coefficients ().

Construction of TP-AG–incorporated GMPCPP microtubules

In the typical Before method, TP-AG (10 μM) was mixed with a solution containing tubulin (4 μM) and tubulin-TMR (1 μM) in BRB80 [80 mM Pipes (pH 6.9), 1.0 mM MgCl2, and 1.0 mM EGTA] and incubated at 25°C for 30 min in the dark. Then, 2 μl of GMPCPP premix (1 mM GMPCPP and 20 mM MgCl2 in BRB80) was added to the mixture (8 μl) and incubated at 37°C for 30 min in the dark (final concentrations: [tubulin] = 3.2 μM, [tubulin-TMR] = 0.8 μM, [TP-AG] = 8 μM, and [GMPCPP] = 0.2 mM). As a control, AG or TP-mAG (8 μM) was used instead of TP-AG.

In the typical After method, GMPCPP premix (2 μl) was mixed with a solution (6 μl) containing tubulin (4 μM) and tubulin-TMR (1 μM) in BRB80. The mixture was incubated at 37°C for 30 min in the dark. Then, TP-AG (2 μl) was added to the mixture and kept at 25°C for 30 min in the dark (final concentrations: [tubulin] = 3.2 μM, [tubulin-TMR] = 0.8 μM, [TP-AG] = 8 μM, and [GMPCPP] = 0.2 mM). As a control, AG or TP-mAG (8 μM) was used instead of TP-AG.

Binding analysis of TP-AG to GMPCPP microtubules using subtilisin-treated tubulin

C-terminal tails of tubulin were removed by the treatment of subtilisin according to the reported procedure with modification (). Mixture of tubulin (64 μM) and tubulin-TMR (16 μM) were treated with subtilisin in a subtilisin:tubulin ratio of 1:100 (w/w) at 4°C for 30 min. Then, the subtilisin activity was inhibited by incubation with 12.5 mM phenylmethylsulfonyl fluoride at 4°C for 10 min. After centrifugation at 21,000 rpm at 4°C for 20 min, the supernatants were collected and used as subtilisin-treated tubulin. The removal of C-terminal tails of tubulin was confirmed by SDS-PAGE (fig. S2A). TP-AG–incorporated GMPCPP microtubules were prepared by the Before or After method as above using subtilisin-treated tubulin or untreated tubulin (final concentrations: [tubulin] = 19.2 μM, [tubulin-TMR] = 4.8 μM, [TP-AG] = 8 μM, and [GMPCPP] = 0.2 mM). As a control, AG (8 μM) was used instead of TP-AG. The mixture was used for confocal laser scanning microscopy imaging. Fluorescence intensity per microtubule was calculated from the fluorescence images by subtracting the background intensity using ImageJ software (National Institutes of Health). The background-subtracted AG fluorescence intensity per TMR fluorescence intensity for each microtubule (N = 30) were calculated.

Binding analysis of anti-AG antibody to TP-AG–incorporated GMPCPP microtubules

Anti-AG polyclonal antibody (1.3 mg/ml, 190 μl) in phosphate-buffered saline (PBS; 137 mM NaCl, 8.1 mM Na2HPO4, 2.68 mM KCl, and 1.47 mM KH2PO4) was mixed with 5-carboxytetramethylrhodamine, succinimidyl ester (10 mM, 10 μl) in dimethyl sulfoxide (DMSO). The mixture was incubated at 37°C for 15 min in the dark. The TMR-labeled anti-AG antibody was obtained by washing with PBS seven times by the ultrafiltration (cutoff Mw, 10 kDa). The concentrations of TMR and the anti-AG antibody of the TMR-labeled antibody were determined by UV-vis spectrometry. The modification ratio of TMR per the anti-AG antibody was determined as 2.2. TP-AG–incorporated GMPCPP microtubules were prepared by the Before or After method as above. Then, the TMR-labeled anti-AG antibody (6 μM, 2 μl) was added to the TP-AG–incorporated microtubule solution (8 μl) and incubated at 25°C for 1 hour in the dark (final concentrations: [tubulin] = 4 μM, [TP-AG] = 8 μM, [TMR-labeled anti-AG antibody] = 1.2 μM, and [GMPCPP] = 0.2 mM). The mixture was used for confocal laser scanning microscopy imaging. Fluorescence intensity per microtubule was calculated from the fluorescence images by subtracting the background intensity using ImageJ. The background-subtracted TMR fluorescence intensity per AG fluorescence intensity for each microtubule (N = 30) were calculated.

Binding analysis of TP-AG to GMPCPP microtubules by cosedimentation assay

TP-AG–incorporated GMPCPP microtubules were prepared by the Before method as above (final concentrations: [tubulin] = 5, 10, 20, and 30 μM; [TP-AG] = 37 μM; and [GMPCPP] = 0.2 mM). As a control, AG or TP-mAG (37 μM) was used instead of TP-AG. After ultracentrifugation of the mixture at 50,000 rpm at 37°C for 5 min, the supernatants and the pellets were collected and analyzed by SDS-PAGE. The ratio of TP-AG bound to microtubules (θ) was calculated as the ratio of the density of TP-AG bands in the supernatants and the pellets. The results were treated with the following Hill equation (Eq. 1) to give the dissociation constant (Kd) and the Hill coefficient (n; fig. S3C)

By the fitting, Kd and n were determined as 11 μM and 2.2, respectively.

Binding analysis of TP-AG in living cells

Human hepatoma HepG2 cells were purchased from the RIKEN BioResource Research Center (Ibaraki, Japan) and cultured in Dulbecco’s modified Eagle’s medium containing 10% (v/v) fetal bovine serum, streptomycin (50 μg/ml), penicillin (50 μg/ml), neomycin (100 μg/ml), 1 mM sodium pyruvate, and 1% (v/v) minimum essential medium nonessential amino acids (M7145, Sigma-Aldrich). HepG2 cells were seeded onto a single-well glass bottom dish at 2.0 × 104 cells per well in a final volume of 100 μl and incubated at 37°C for 24 hours at 5% CO2. According to the protocol of Lipofectamine LTX (Thermo Fisher Scientific), TP-AG (40 μM, 10 μl), opti-MEM (188 μl), LTX (2 μl), and PLUS (0.4 μl) were mixed and incubated at 25°C for 15 min. After removal of the medium, the mixture was added to the cells and incubated at 37°C for 1 hour, 5% CO2. After washing with PBS, the cells were incubated with Hoechst 33342 (10 μg/ml) and 1 μM Tubulin Tracker Deep Red (Thermo Fisher Scientific) in the medium at 37°C for 30 min at 5% CO2. After washing with probenecid (7.7 μg/ml) in Hanks’ balanced salt solution, the medium with probenecid (7.7 μg/ml) was added to the cells, and confocal images were acquired.

Construction of TP-AG–incorporated GTP microtubules

In the typical Before method, TP-AG (6.25 μM) was mixed with a solution containing tubulin (4 μM) and tubulin-TMR (1 μM) in BRB80 and incubated at 25°C for 30 min in the dark. Then, 4 μl of GTP premix (5 mM GTP, 20 mM MgCl2 in BRB80, and 25% DMSO) was added to the mixture (16 μl) and incubated at 37°C for 30 min in the dark (final concentrations: [tubulin] = 3.2 μM, [tubulin-TMR] = 0.8 μM, [TP-AG] = 5 μM, and [GTP] = 1 mM). As a control, AG or taxol (5 μM) was used instead of TP-AG.

Evaluation of the effect of TP-AG on the formation of GTP microtubules

TP-AG–incorporated GTP microtubules were prepared by the Before method as above (final concentrations: [tubulin] = 30 μM, [TP-AG] = 34 μM, and [GTP] = 1 mM). As a control, taxol (34 μM), AG (34 μM), or the storage buffer was used instead of TP-AG. For the “37°C” samples, the samples were polymerized by incubating at 37°C for 30 min upon addition of GTP premix. For the “4°C” samples, the samples were polymerized by incubating at 37°C for 30 min upon addition of GTP premix and then further incubated at 4°C for 90 min. After ultracentrifugation at 50,000 rpm at 37°C for 5 min for the 37°C samples and at 35,000 rpm at 4°C for 5 min for the 4°C samples, the supernatants and the pellets were collected and analyzed by SDS-PAGE. The microtubule formation efficiency was calculated as the ratio of the density of tubulin bands in the supernatants (S) and the pellets (P) using the following Eq. 2.

Confocal laser scanning microscopy measurement

The glass bottom dishes (Matsunami, Osaka, Japan) were coated by poly-l-lysine (1 mg/ml; Mw, 30,000 to 70,000; Sigma-Aldrich) at room temperature for 1 hour and then removed and dried. The microtubule samples were put on the plate and kept at room temperature for 0.5 to 1 hour and then observed by confocal laser scanning microscopy. Tubulin-TMR was excited with 550 nm and observed through a 574-nm emission band-pass filter (red). TP-AG, AG, and TP-mAG were excited with 493 nm and observed through a 505-nm emission band-pass filter (green). Hoechst 33342 was excited with 352 nm and observed through a 455-nm emission band-pass filter (cyan). Tubulin Tracker Deep Red was excited with 653-nm laser and observed through a 668-nm emission band-pass filter (blue).

Turbidity measurement

Turbidity measurements were performed with tubulin (4 μM) and GTP (1 mM) in the absence or presence of TP-AG, AG, TP-mAG, or taxol (10 μM) in BRB80 at 37°C. The optical density at 350 nm was monitored with a UV-vis spectrometer for 60 min at 1-min intervals. When TP-AG, AG, and TP-mAG were used, the absorbance of these proteins at 350 nm was subtracted.

Negative-stain EM measurement

Samples (3.5 μl) with different polymerization methods were applied to prehydrophilized carbon grids and negatively stained with 1.5% uranyl acetate. The specimens were examined in an H-7100 electron microscope (Hitachi) operating at 75 kV. EM images were recorded by AMT XR41 digital CCD camera (Advanced Microscopy Techniques). Microtubule lengths and angles of branches were measured by ImageJ using EM images taken at a nominal magnification of ×30,000. To evaluate the straightness of microtubules, EM images were taken at a nominal magnification of ×20,000. A homemade ImageJ macro was used to obtain the coordinates of start and end points of one microtubule and segmented length of the same microtubule at the same time. The end-to-end distances of microtubules were calculated, and the ratio with the segmented lengths of corresponding microtubules was obtained. To analyze the proportions of the microtubule types, EM images were taken at a nominal magnification of ×30,000. Visual inspection was performed to categorize microtubule types. For both measurement of straightness and analysis of proportions of microtubule types, microtubules with at least 500 nm long were used.

Cryo-EM measurement

For cryo-EM, polymerized GMPCPP microtubules (10 μM) were incubated with TP-AG (20 μM) at 25°C for 30 min and further incubated at 37°C for at least 30 min. A total of 3.5 μl of TP-AG–treated microtubule samples was applied to a glow discharged holey carbon grid (Quantifoil R2/1). The grids were blotted and plunged into liquid ethane using the Vitrobot Mark IV (Thermo Fisher Scientific) at 30°C and 100% humidity with a blot force of 3 and a blot time of 5 s. The grids were observed in the Glacios cryo-TEM (Thermo Fisher Scientific) operating at 200 kV. EM images were taken with Falcon III direct electron detector (Thermo Fisher Scientific) in a linear mode with the pixel size of 4.0 Å per pixel. Total dose was ~18 e/Å2, and the defocus was set to −2.0 μm. The diameters of singlets, doublets, bundles of singlets, and multiplets were measured by Fiji using cryo-EM images. For measuring A-tubule and B-tubule diameters, the same doublet was used for the analysis.

TIRFM measurement

TMR-labeled seed microtubules were obtained by polymerizing 15 μM tubulin mix (tubulin-TMR: nonlabeled tubulin = 1:5 in molar ratio) at 37°C for 30 min using a polymerization buffer containing 80 mM Pipes, 1 mM EGTA, 1 mM GMPCPP, and 4 mM MgCl2. The prepared microtubule solution was diluted 50-fold using BRB80. A flow cell with approximate dimensions of 2 mm by 9 mm by 0.1 mm (W × L × H) was prepared by placing a cover glass (9 mm by 18 mm; Matsunami) on a glass slide (40 mm by 50 mm; Matsunami) using a double-sided tape as a spacer. First, 10 μl of BRB80 was passed through the flow cell. Next, the flow cell was filled with 5 μl of protein A (5 mg/ml; BioVision Inc., USA) and incubated for 5 min, followed by a wash with 20 μl of BRB80. Then, 5 μl of anti-tubulin antibody (73 mg/ml; T3526, Sigma-Aldrich) was passed through the flow cell and incubated for 5 min. After washing with 20 μl of BRB80, the flow cell was incubated for another 15 min with 1% pluronic F-127 (P6866, Invitrogen) in water, and the GMPCPP seed microtubules in BRB80 were applied and allowed to bind to the anti-tubulin antibody–coated substrate for 10 min. Last, to allow the seed microtubules for showing dynamic instability, a polymerizing mixture [5 mM GTP, d-glucose (9 mg/ml), glucose oxidase (100 U/ml), and catalase (100 U/ml)] in BRB80 containing 20 μM tubulin mix (tubulin–Alexa Fluor 405: nonlabeled tubulin = 1:1 in molar ratio) and 40 μM TP-AG was applied into the flow cell. Then, after ~30 min, observation was performed under TIRFM at 25°C.

Motility assay

TP-AG–incorporated GMPCPP microtubules were prepared by the Before method as above (final concentrations: [tubulin] = 6.4 μM, [tubulin-TMR] = 1.6 μM, [TP-AG] = 0.8 μM, and [GMPCPP] = 0.2 mM). BRB80 was used instead of TP-AG as a control (buffer). The microtubules were diluted 16-fold by BRB80. Flow cells were prepared by making a narrow channel on a 24-mm by 60-mm coverslip covered with a 18-mm by 18-mm coverslip (Matsunami, Osaka, Japan) using double-sided tape as a spacer. First, casein (0.5 mg/ml) in BRB80 was introduced into the flow cells and incubated for 3 min. Then, the solution was exchanged with wash buffer [casein (0.5 mg/ml), d-glucose (4.5 mg/ml), glucose oxidase (50 U/ml), catalase (50 U/ml), 1.0 mM dithiothreitol, and 1.0 mM MgCl2 in BRB80] containing 800 nM kinesin and incubated for 3 min. After washing with wash buffer, the solution was exchanged with microtubule solution and incubated for 3 min. After washing with wash buffer, the solution was exchanged with wash buffer containing 5.0 mM ATP and 1.0 mM Trolox. Then, the motility of microtubules was imaged every 10 s. All the experiments were performed at room temperature. Temporal-color coded images of microtubule movement were generated by superimposing 13 images using the Temporal-Color Code plugin in the Fiji image processing software package based on ImageJ. The images obtained by the fluorescence microscopy were analyzed to determine the velocity, end-to-end length, and contour length of each microtubule using ImageJ. The contour length along each microtubule and the end-to-end distance of the same microtubule were measured. Persistence length (Lp) was determined by fitting the data to Eq. 3 using Excel and Solverwhere is the mean squared end-to-end distance and L is the contour length ().

Observation of aster structures

TP-AG–incorporated GTP microtubules were prepared by the Before method as above (Final concentrations: [tubulin] = 20 μM, [tubulin-TMR] = 10 μM, [TP-AG] = 4 μM, [GTP] = 1 mM). The microtubules were diluted 80-fold by a solution containing 1 mM GTP, 4 mM MgCl2 in BRB80, 5% DMSO. Flow cells were prepared by making a narrow channel on a 24 mm by 60 mm coverslip covered with 18 mm by 18 mm coverslip (Matsunami, Osaka, Japan) using double-sided tape as a spacer. First, casein (0.5 mg/ml) in BRB80 was introduced into the flow cells and incubated for 3 min. Then, the solution was exchanged with wash buffer containing 800 nM kinesin and incubated for 3 min. After washing with wash buffer, the solution was exchanged with the diluted microtubule solution and incubated for 3 min. After washing with wash buffer, the solution was exchanged with wash buffer containing 5.0 mM ATP, 1 mM GTP, 4 mM MgCl2, and 5% DMSO. Then, the motility of microtubules was imaged every 30 s. All the experiments were performed at room temperature. A temporal-color coded image of microtubule movement was generated by superimposing 21 images using the Temporal-Color Code plugin in the Fiji image processing software package based on ImageJ.

Statistical analysis

The analytical information (the SEM, SD, the N values, and analytical methods) are shown in the figure legends. All figures were obtained from a minimum of three experiments. Box plots in Fig. 2D and fig. S5D are presented with the elements: center line, median; box limits, Q1 and Q3; whiskers, “adjacent points,” analyzed by Igor Pro 9 software. For statistical analyses, Igor Pro 9 software and Prism software were used.