Lipid Nanotubes and Microtubes: Experimental Evidence for

Mar 12, 2017 - Lipid Nanotubes and Microtubes: Experimental Evidence ... 305-8565, Japan, and CREST, Japan Science and Technology Agency (JST),...
0 downloads 0 Views 733KB Size
Download PDF
Langmuir 2004, 20, 5969-5977

5969

Lipid Nanotubes and Microtubes: Experimental Evidence for Unsymmetrical Monolayer Membrane Formation from Unsymmetrical Bolaamphiphiles Mitsutoshi Masuda* and Toshimi Shimizu Nanoarchitectonics Research Center (NARC), National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba Central 5, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan, and CREST, Japan Science and Technology Agency (JST), Tsukuba Central 4, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8562, Japan Received April 10, 2004 Unsymmetrical bolaamphiphiles, ω-[N-β-D-glucopyranosylcarbamoyl]alkanoic acids, with even-numbered oligomethylene chains (12, 14, 16, 18, and 20 carbons) self-assembled in water to form lipid nano- and microtubes. The tubular assemblies were separated by centrifugation and examined by transmission electron microscopy, X-ray diffraction, and Fourier transform infrared spectroscopy to study the molecular packing within the tubular membranes. The nanotubes encapsulated the staining reagent phosphotungstate, which revealed them to be hollow cylinders up to several hundred micrometers long with 30-43-nm outer diameters and 14-29-nm inner diameters. By comparing the membrane stacking periodicity obtained from powder X-ray diffraction analysis of the dehydrated tubes with the molecular packing within single crystals, we found that the nanotubes consist of an unsymmetrical monolayer lipid membrane (MLM) in which the molecules are packed in a parallel fashion. This suggests that the inner surface of the nanotubes is covered with carboxy headgroups and the outer surface with 1-glucosamide headgroups. The inner diameters of the lipid nanotubes could be controlled in the range 17.7-22.2 nm in steps of ∼1.5 nm/two carbons by varying the oligomethylene spacer length. The microtubes had three types of molecular arrangements. The first type was a symmetrical MLM in which the molecules were packed in an antiparallel fashion. The other two types had unsymmetrical MLM stacking with head-to-head and head-to-tail motifs. Increasing the number of oligomethylene spacers stabilized the unsymmetrical MLM structure in both nano- and microtubes.

Introduction Lipid nanotubes with different inner- and outermost lipid membrane surfaces are intriguing nanoarchitectures applicable to the specific modification of internal and external surfaces,1-3 selective filling of nanomaterials into hollow cylinders,4-6 controlled release,7 and creation of templates for the fabrication of inorganic nanomaterials.8-12 The most direct way to construct an unsymmetrical lipid membrane is to make use of the asymmetry of heteroditopic 1,ω-amphiphiles having headgroups differing in size or properties, that is, “unsymmetrical bolaamphiphiles” (Figure 1).13-16 This is because bolaamphiphiles are able to form a monolayer lipid membrane (MLM).13,14 * Corresponding author. E-mail: [email protected]. (1) Reches, M.; Gazit, E. Science 2003, 300, 625. (2) Dujardin, E.; Peet, C.; Stubbs, G.; Culver, J. N.; Mann, S. Nano Lett. 2003, 3, 413. (3) Ringler, P.; Mu¨ller, W.; Ringsdorf, H.; Brisson, A. Chem.sEur. J. 1997, 3, 620. (4) Fuhrhop, J.-H.; Tank, H. Chem. Phys. Lipids 1987, 43, 193. (5) Djalali, R.; Chen, Y.; Matsui, H. J. Am. Chem. Soc. 2003, 125, 5873. (6) Yang, B.; Kamiya, S.; Yoshida, K.; Shimizu, T. Chem. Commun. 2004, 500. (7) Schnur, J. M. Science 1993, 262, 1669. (8) van Bommel, K. J. C.; Friggeri, A.; Shinkai, S. Angew. Chem., Int. Ed. 2003, 42, 980. (9) Small, D. M. The Physical Chemistry of Lipids; Plenum Press: New York, 1986; Vol. 4. (10) Jung, J. H.; Kobayashi, H.; van Bommel, K. J. C.; Shinkai, S.; Shimizu, T. Chem. Mater. 2002, 14, 1445. (11) Jung, J. H.; Lee, S.-H.; Yoo, K.; Yoshida, K.; Shimizu, T.; Shinkai, S. Chem.sEur. J. 2003, 9, 5307. (12) Ji, Q.; Iwaura, R.; Kogiso, M.; Jung, J. H.; Yoshida, K.; Shimizu, T. Chem. Mater. 2004, 16, 250. (13) Fuhrhop, J.-H.; Wang, T. Chem. Rev., published online Mar. 12, http://dx.doi.org/10.1021/cr030602b. (14) Fuhrhop, J.-H.; Fritsch, D. Acc. Chem. Res. 1986, 19, 130.

Figure 1. Lipid nanotube consisting of monolayer lipid membranes of an unsymmetrical bolaamphiphile.

MLMs can be classified as unsymmetrical or symmetrical (so-called polymorphs) depending on whether their component lipids are packed in a parallel or antiparallel fashion (Figure 2a). In multilayer stacked MLMs, two additional kinds of membrane stacking motifs (polytypes) can be defined depending on the molecular orientation at the interface between the two surfaces: head-to-head (H-H) or head-to-tail (H-T) (Figure 2b). Unsymmetrical MLMs with H-T stacking can form lipid nanotubes having different inner and outer surfaces. Indeed, the lipid membranes of thermophilic and acidophilic archaebacteria are known to be composed of natural unsymmetrical bolaamphiphiles. These bacterial membranes are unsymmetrical. To date, many attempts have been made to form nanotubes,4,17-19 nanofibers,20-23 vesicles,24-26 and (15) Fuhrhop, J.-H.; Ko¨ing, J. Membranes and Molecular Assemblies: The Synkinetic Approach, 1st ed.; The Royal Society of Chemistry: Cambridge, U.K., 1994. (16) Claussen, R. C.; Rabatic, B. M.; Stupp, S. I. J. Am. Chem. Soc. 2003, 125, 12680. (17) Fuhrhop, J.-H.; Spiroski, D.; Boettcher, C. J. Am. Chem. Soc. 1993, 115, 1600.

10.1021/la049085y CCC: $27.50 © 2004 American Chemical Society Published on Web 06/02/2004

5970

Langmuir, Vol. 20, No. 14, 2004

Masuda and Shimizu Chart 1

Figure 2. (a) Schematic illustration of monolayer lipid membranes (MLMs) formed from unsymmetrical bolaamphiphiles and (b) the resulting four types of multilayer structures according to the stacking motif of the MLM. The smaller of the two hydrophilic groups is considered to be the tail.

crystals27-30 from synthetic unsymmetrical bolaamphiphiles. Except for limited examples of vesicular membranes in the fluid state,24-26 fibrous assemblies in the solid state,21 and crystal structures,27 most attempts have resulted in the formation of symmetrical MLMs28,29 in which the molecules are packed in an antiparallel fashion. Alternatively, molecular orientation in MLMs is random or unknown. Recently, we reported that the unsymmetrical 1-galactosamide bolaamphiphile 2(14) (Chart 1) forms unsymmetrical MLMs within a crystal lattice27 and that symmetrical 1-glucosamide bolaamphiphiles produce MLM-based nanofibers.31 Both findings suggest that the hydrogen-bond network of sugar hydroxy and amide groups plays a significant role in controlling molecular orientation within the MLMs and in stabilizing the assemblies.15,17 In the present article, we describe micro- and nanotube formation from synthetic unsymmetrical bolaamphiphiles 1(n) (Chart 1) having a 1-glucosamide moiety at one end of an oligomethylene chain and a carboxylic acid group at the other end. On the basis of the MLM thicknesses estimated from both powder X-ray diffraction (XRD) (18) Sirieix, J.; Lauth-de Viguerie, N.; Rivie`re, M. R.; Lattes, A. New J. Chem. 2000, 24, 1043. (19) Prata, C.; Mora, N.; Polidori, A.; Lacombe, J.-M.; Pucci, B. Carbohydr. Res. 1999, 321, 15. (20) Guilbot, J.; Benvegnu, T.; Legros, N.; Plusquellec, D. Langmuir 2001, 17, 613. (21) Schneider, J.; Messerschmidt, C.; Schulz, A.; Gnade, M.; Schade, B.; Luger, P.; Bombicz, P.; Hubert, V.; Fuhrhop, J.-H. Langmuir 2000, 16, 8575. (22) Song, J.; Cheng, Q.; Kopta, S.; Stevens, R. C. J. Am. Chem. Soc. 2001, 123, 3205. (23) Jaeger, D. A.; Guowen, L.; Subotkowski, W.; Carron, K. T. Langmuir 1997, 13, 5563. (24) Fuhrhop, J.-H.; David, H.-H.; Mathieu, J.; Liman, U.; Winter, H.-J.; Boekema, E. J. Am. Chem. Soc. 1986, 108, 1785. (25) Fuhrhop, J.-H.; Mathieu, J. Chem. Commun. 1983, 144. (26) Liang, K.; Hui, Y. J. Am. Chem. Soc. 1992, 114, 6588. (27) Masuda, M.; Shimizu, T. Chem. Commun. 2001, 2442. (28) Szafran, M.; Dega-Szafran, Z.; Katrusiak, A.; Buczak, G.; Glowiak, T.; Sitkowski, J.; Stefaniak, L. J. Org. Chem. 1998, 63, 2898. (29) Aigouy, P. T.; Costeseque, P.; Sempere, R.; Senac, T. Acta Crystallogr., Sect. B 1995, 51, 55. (30) Sim, G. A. Acta Crystallogr. 1955, 8, 833. (31) Shimizu, T.; Masuda, M. J. Am. Chem. Soc. 1997, 119, 2812.

patterns of the tubular assemblies and single-crystal analyses of 1(12)32 and 2(14),27 we provide evidence, for the first time, that these nanotubes are composed of unsymmetrical MLMs. In addition, we obtained three types of microtubes based on symmetrical or different types of unsymmetrical MLMs. Furthermore, we could control the inner diameter of the unsymmetrical MLM-based nanotubes, composed of wedge-shaped bolaamphiphiles, merely by varying the oligomethylene spacer length. Controlling the diameter of lipid nanotubes in the range 1-20 nm is crucial for tuning the properties of confined nanospaces because current advanced microfabrication systems cannot operate with this much precision. Although it is possible to control the morphology of lipid nanotubes,33-36 there have been no reliable approaches for controlling their diameters until now.18,37-39 Results and Discussion Self-Assembly and TEM Observations. A series of unsymmetrical bolaamphiphiles 1(n) was synthesized and subjected to self-assembly in pure water. Hot (>97 °C) aqueous solutions of even-numbered 1(n) compounds (n ) 14, 16, 18, and 20; 0.1-1.5 mg mL-1) became slightly opaque and viscous upon cooling (0.1 °C min-1), which indicated self-assembly. On the other hand, the oddnumbered compound 1(13) yielded only colorless crystalline solids, suggesting the odd-even effect on the selfassembled morphology.21,31 Transmission electron microscopic (TEM) observations revealed that the 1(n) compounds with n ) 14, 16, 18, and 20 formed micrometer-sized tubes (microtubes) along with relatively thinner nanometer-sized fibers having a highaxial-ratio morphology, which were eventually proven to be nanotubes, as mentioned below (Figure 3). The microtubes were 118-190 nm in outer diameter and 60-90 nm in inner diameter, had walls 20-40 nm thick, and were up to several hundred micrometers long. We also observed similar images of the microtubes in water and the nanotubes on a mica substrate under optical- and atomic force microscopy (AFM), respectively, indicating that both the tubular assemblies were actually formed in water.32 The tapelike forms with shallow contrast observed in Figure 3 are also microtubes with thin walls. Interest(32) See the Supporting Information. (33) John, G.; Jung, J. H.; Minamikawa, H.; Yoshida, K.; Shimizu, T. Chem.sEur. J. 2002, 8, 5494. (34) Georger, J. H.; Singh, A.; Price, R. R.; Schnur, J. M.; Yager, P.; Schoen, P. E. J. Am. Chem. Soc. 1987, 109, 6169. (35) Spector, M. S.; Selinger, J. V.; Singh, A.; Rodriguez, J. M.; Prince, R. R.; Schnur, J. M. Langmuir 1998, 14, 3493. (36) Oda, R.; Huc, I.; Schmutz, M.; Candau, S. J.; MacKintosh, F. C. Nature 1999, 399, 566. (37) Thomas, B. N.; Lindemann, C. M.; Corcoran, R. C.; Cotant, C. L.; Kirsch, J. E.; Persichini, P. J. J. Am. Chem. Soc. 2002, 124, 1227. (38) Singh, A.; Wong, E. M.; Schnur, J. M. Langmuir 2003, 19, 1888. (39) Porrata, P.; Goun, E.; Matsui, H. Chem. Mater. 2002, 14, 4378.

Lipid Nanotubes and Microtubes

Langmuir, Vol. 20, No. 14, 2004 5971

Figure 3. TEM pictures of unstained self-assembled structures showing a mixture of micro- and nanotubes formed from (a) 1(14), (b) 1(16), and (c) 1(18). The numerical values indicate the outer and (in parentheses) inner diameters of the nano- and microtubes indicated by arrows.

ingly, we were able to separate the nanotubes, since only the microtubes were sedimented from the dispersion by centrifugation. This enabled us to analyze the nanotubes selectively by TEM, XRD, and Fourier transform infrared (FT-IR) spectroscopy. Negative staining of the separated nanotubes before TEM observation revealed them to be tubular, since each hollow cylinder was filled with the staining reagent (Figure 4a-f). We have observed the effective filling of HAuCl4 solution into other glycolipid nanotubes via capillary force only when the nanotubes were emptied by freeze-drying.6 Thus, efficient staining of the hollow cylinder will be explained by the fact mentioned above. The nanotubes were 30-43 nm in outer diameter, 14-29 nm in inner diameter, had walls ∼6-7 nm thick, and were up to several hundred micrometers long. The outer diameter was in good agreement with that measured from tapping mode AFM.32 This finding indicated that TEM observation is a useful methodology for evaluating the dimensions of the nanotubes. Both hydrated and dehydrated nanotubes were stable for more than a year when stored at 15-30 °C and atmospheric pressure. The shortest bolaamphiphile, 1(12), self-assembled exclusively into nanotubes in pure water but below pH 3 partly transformed into single crystals suitable for X-ray analysis. This transformation indicates that the degree of ionization of the carboxy group affects the self-assembled morphology, as reported previously.40,41 Single-Crystal Structure Analysis of 1(12) and 2(14). Figure 5 shows the molecular packing within the crystal lattice of 1(12), as determined by single-crystal X-ray analysis (Figure 5a), together with that of 2(14), as previously reported27 (Figure 5b). The bolaamphiphile 1(12) is packed into a symmetrical MLM, but 2(14) packs into an unsymmetrical MLM with a head-to-tail (H-T) interface between its layers (the larger sugar moiety of the two hydrophilic groups is considered the head). On the basis of a comparison of both packing patterns in the crystal lattice, we found that the MLM thickness (d) (2.66 nm) of the symmetrical MLM of 1(12) is greater than its extended molecular length (L) (2.52 nm) and that the MLM (40) Kogiso, M.; Ohnishi, S.; Yase, K.; Masuda, M.; Shimizu, T. Langmuir 1998, 14, 4978. (41) Estroff, J. A.; Hamilton, A. D. Angew. Chem., Int. Ed. 2000, 39, 3447.

thickness (d) (2.07 nm) of the unsymmetrical MLM of 2(14) is smaller than its extended molecular length (L) (2.77 nm). The greater d value of the 1(12) MLM is due to the pseudo-interdigitated structure of the bolaamphiphile, a structure in which only the sugar moieties occupy the MLM’s interfaces and the carboxy moieties are retracted from it. Despite a molecular tilt of 33° with respect to the normal to the layer plane, the d value is larger than the L value in the symmetrical MLM motif (Figure 5a). The lower d value of the unsymmetrical MLM of 2(14) is due to its large molecular tilt (52° with respect to the normal to the layer plane). The relationship between d and L observed in the crystal lattice is used below to define the MLM arrangement within the tubular assemblies. Powder XRD Measurements. The micro- and nanotubes separated by centrifugation were freeze-dried, and their molecular packing was characterized by XRD. The diffraction patterns obtained (Figures 6 and 7) revealed that the nano- and microtubes are composed of MLM stacking structures that differ both in molecular packing within the MLM (polymorph) and in the MLM stacking motif (polytype), as defined in Figure 2. That is, the nanotubes gave a single diffraction peak ascribable to the MLM stacking periodicity, hereafter termed “stacking periodicity”, in the small-angle region (Figure 7), whereas the microtubes had at least three diffraction peaks in this region (Figure 6). This finding clearly indicates that the nanotubes possess only one type of MLM having a single stacking motif, but the microtubes have a mixture of polymorphs or polytypes of MLMs. Characterization of Microtubes. To clarify the MLM polymorph and polytype in the microtubes on the basis of the relationship between d and L mentioned above, we plotted the stacking periodicity value obtained from the small-angle XRD peaks, as well as from the calculated length for each extended molecule (L), against the oligomethylene chain length (n) (Figure 8). The stacking periodicity values indicated that the MLMs of microtubes belong to three categories: thin, medium, and thick. Thin, medium, and thick membranes were observed for 1(14) and 1(16), but only thin and thick membranes were observed for 1(18) and 1(20). The stacking periodicity values of the 1(13) crystalline solids, 1.97 and 3.06 nm, fell into the same categories as a thin and a medium membrane, respectively. The thin membrane consists of

5972

Langmuir, Vol. 20, No. 14, 2004

Masuda and Shimizu

Figure 4. TEM pictures of the nanotubes formed from (a) 1(12), (b) 1(14), (c) 1(16), (d) 1(18), (e) 1(20), and (f) the sodium salt of 1(18). Each sample was centrifuged and negatively stained with phosphotungstate. The numerical values indicate the outer and (in parentheses) inner diameters of the nanotubes indicated by arrows.

an unsymmetrical MLM with head-to-tail (H-T) orientation at the interfaces (Figure 8d), because the stacking periodicity value for 1(14) is similar to the MLM thickness (d) of its epimer, 2(14), in the crystal lattice (marked with a triangle at n ) 14). The thick membrane consists of an unsymmetrical MLM with H-H orientation at the interface (Figure 8a), because the stacking periodicity values correspond to exactly twice the d value of the unsymmetrical MLM with H-T orientation for n ) 14, 16, 18, and 20.42 The microtube’s medium membrane appears to be a symmetrical MLM (Figure 8b), on the basis of the similarity of its stacking periodicity value to the d value of the symmetrical MLM of 1(12) in the crystal lattice (marked with a triangle at n ) 12). However, whether the molecules have H-T or H-H orientation at the interfaces is still unknown. (42) In the XRD patterns of the microtube (Figure 6), the (001) diffraction peaks of the unsymmetrical MLM with H-T orientation (d ) 2.22, 2.36, 2.60, and 2.70 nm for n ) 14, 16, 18, and 20, respectively) may overlap with higher-order (002) diffraction peaks of that with H-H orientation, which should appear at angles corresponding to half of 4.35, 4.26, 5.15, and 5.24 nm for n ) 14, 16, 18, and 20, respectively, although the overlapping was negligible because the (001) peak was stronger than the (002) peak in general.

Characterization of Nanotubes. On the basis of the assignment of the microtubes described above, we found that, except for the 1(12) nanotube, the nanotube MLMs (open squares, Figure 8) could be attributed to an unsymmetrical motif with sharp curvature (Figure 8c). That is, each d value of the nanotubes was estimated to be equal to or slightly smaller than L. The criterion for distinguishing MLMs with a symmetrical motif from those with an unsymmetrical motif is whether the d value is larger or smaller than the L value (open circles, Figure 8). The d values of the nanotubes, except for 1(12), are 5-10% shorter than their L values, indicating unsymmetrical MLMs with a molecular tilt. On the basis of the d and L values, we estimated that the molecules are tilted by 17-26° with respect to the radius of the nanotube. In contrast, the tilt angles of the molecules in the unsymmetrical MLM-based microtubes were 39-44°. Instead of the severe molecular tilt possible in a less-curved MLM, the MLMs of the nanotubes show sharp membrane curvature. However, the d value (3.02 nm) of the 1(12) nanotube, which is longer than its L value (2.62 nm), corresponds to the d value of symmetrical MLMs (2.623.0 nm), suggesting that, at shorter chain lengths, the

Lipid Nanotubes and Microtubes

Langmuir, Vol. 20, No. 14, 2004 5973

Figure 7. XRD patterns for the dehydrated nanotubes of 1(n) separated by centrifugation.

Figure 5. Symmetrical and unsymmetrical MLM packing of (a) 1(12) and (b) 2(14), respectively, based on single-crystal structure analysis. The calculated molecular length (L) and MLM thickness (d) estimated from single-crystal structure analysis are shown.

Figure 6. XRD patterns for the dehydrated microtubes of 1(n) separated by centrifugation.

nanotube membrane motif starts to change from the unsymmetrical motif to the more energetically favored symmetrical motif observed in the 1(12) crystals. FT-IR Measurements of Micro- and Nanotubes. The FT-IR spectra of the separated micro- and nanotubes revealed the extended molecular structures within the

MLM of the micro- and nanotubes (Figure 9 and Table 1). A folded molecular conformation is ruled out because the CH2 antisymmetric and symmetric stretching band peaks of the tubular assemblies had lower frequencies (2916.02918.0 cm-1 and 2848.4-2849.8 cm-1, respectively) ascribable to an all-trans conformation. No shoulder peaks at the higher frequencies attributable to a gauche conformation were seen (Table 1).43,44 The amide I and II band frequencies suggest that the amide linkage in 1(n) forms a linear hydrogen-bond chain that stabilizes the MLM structures in both micro- and nanotubes. We previously found a similar stabilization for twisted fibrous assemblies formed from dumbbellshaped bolaamphiphiles.31 The nanotubes must have relatively weaker amide hydrogen bonds than the microtubes because the amide I bands of the nanotubes (1634-1640 cm-1 and 1665-1668 cm-1) appeared at higher frequencies than those of the microtubes (16341638 cm-1). The weakly hydrogen-bonded amide band at 1665 cm-1 observed for the 1(14) and 1(16) microtubes was probably due to contamination of the nanotube, even after centrifugation. Broad shoulders around 1696-1714 cm-1 are attributable to the antisymmetrical stretching band of the carboxy group, suggesting that the carboxyl headgroups form hydrogen bonds that stabilize the interfaces in the MLMs. It was difficult to characterize the hydrogen bonds into the cyclic (1708-1711 cm-1) and lateral (1722-1725 cm-1) arrangements reported previously45-47 because these bands also include other carboxy-based IR bands associated with hydrogen bonds to sugar hydroxyl groups. Indeed, the bolaamphiphiles 2(14) and 1(12) form such hydrogen bonds in the crystal lattice and have IR bands at 1704 and 1696 cm-1, respectively. The carboxy headgroup of 1(n) probably forms a hydrogen bond with other (43) Masuda, M.; Vill, V.; Shimizu, T. J. Am. Chem. Soc. 2000, 122, 12327. (44) Mantsch, H. H.; McElhaney, R. N. Chem. Phys. Lipids 1991, 57, 213. (45) Dong, J.; Ozaki, Y.; Nakashima, K. Macromolecules 1997, 30, 1111. (46) Sun, L.; Crooks, R. M.; Ricco, A. J. Langmuir 1993, 9, 1775. (47) Kogiso, M.; Hanada, T.; Yase, K.; Shimizu, T. Chem. Commun. 1998, 1791.

5974

Langmuir, Vol. 20, No. 14, 2004

Masuda and Shimizu

Figure 8. MLM stacking periodicity in the micro- and nanotubes of 1(n) calculated from XRD data as well as the calculated molecular length (L) of 1(n) plotted against the chain length (n). The same calculations were also performed for a symmetrical MLM of 1(12) (triangle at n ) 12) and an unsymmetrical MLM of 2(14) having H-T orientation (triangle at n ) 14). Table 1. IR Absorption Bands of Isolated Self-Assembled Structures Formed from Bolaamphiphiles 1(n) and 2(14) bolaamphiphile 1(12) 1(13) 1(14) 1(16) 1(18) 1(20) 2(14)

Figure 9. FT-IR spectra of the COOH and the amide I and II band regions for the nano- and microtubes formed from 1(n) (n ) 12, 14, 16, 18, and 20) and for powdered crystalline 1(12) and 2(14).

carboxy headgroups as well as with sugar hydroxy groups to stabilize the MLM structures in the nano- and microtubes. Relation between the Chain Length and the MLM Polymorph. The series of XRD patterns of the microtubes indicated that the chain lengths of 1(n) remarkably affect the MLM polymorph in the tubular assemblies. When the chain length is increased from n ) 12 to n ) 20, the smallangle diffraction peaks (3.02 nm for n ) 12 in Figure 7 and 3.06, 3.23, and 3.34 nm for n ) 13, 14, and 16, respectively, in Figure 6) corresponding to the symmetrical

IR data (cm-1) morphology nanotube single crystal solid nanotube microtube nanotube microtube nanotube microtube nanotube microtube single crystal

νas(CH2) νs(CH2) 2916.8 2917.5 2916.0 2916.6 2916.2 2917.3 2917.1 2916.2 2916.3 2916.5 2918.0 2918.4

2849.7 2849.6 2849.8 2849.5 2849.3 2850.1 2849.3 2849.5 2848.4 2849.6 2849.4 2849.3

amide I

amide II

1634, 1668 1655 1661, 1683 1640, 1665 1634 1634, 1667 1633, 1670 1638, 1667 1632 1637, 1666 1633 1632

1542 1545 1533, 1547 1541 1543 1548 1547 1543 1544 1542 1548 1547

MLM relatively decrease in intensity, while the peak intensity for the unsymmetrical MLM (the other peaks in the small-angle region in Figure 6) become stronger. Thus, 1(18) and 1(20), having the longest spacer chains among those examined, form only unsymmetrical MLMs in both nano- and microtubes, and the shortest, 1(12), forms only symmetrical MLMs. The observed chain length effect on the MLM’s polymorph is due to the difference in membrane fluidity, which depends on the oligomethylene chain length (n). Shortening the oligomethylene chain increases the molecular fluidity within MLMs. With very short chain lengths, the MLM transforms into a random orientation, as found for the shortest-chain nanotube, 1(12) (Figure 10a). Alternatively, the molecule forms a symmetrical MLM, as in the 1(12) crystal, in which antiparallel packing best compensates for the dipole moment and the void arising from the bulky headgroup (Figure 10b). Thus, the effect of the chain length on the type of MLMs formed by 1(n) can be well explained. In contrast to the MLMs of the shorter chains, those of the longer hydrophobic chains tend to have abrupt curvature, as observed for the series of nanotubes (Figure 10c). Alternatively, the hydrophobic chains tilt, as in the microtubes and the crystal lattice of 2(14), adjusting for the difference between the cross-sectional area per

Lipid Nanotubes and Microtubes

Langmuir, Vol. 20, No. 14, 2004 5975

Figure 10. Schematic illustration of the effect of the chain length on molecular packing in tubular assemblies and crystals of unsymmetrical bolaamphiphiles.

Figure 12. Observed inner diameter (Dobs-in) values of lipid nanotubes formed from (a) 1(12), (b) 1(14), (c) 1(16), (d) 1(18), and (e) 1(20) evaluated from negatively stained TEM images. Figure 11. (a) Schematic models of unsymmetrical bolaamphiphiles 1(n) with a short and a long oligomethylene chain and (b) tubular assemblies self-assembled from each model.

This relation means that, if the as and al values are constant, the innermost diameter is proportional to the molecular length (L), that is, proportional to the oligomethylene chain length of the bolaamphiphiles. We can estimate the headgroup areas, al and as, of 1(n) from both the cross-sectional area of the 1-glucosamide headgroup observed in a crystal lattice49 and the molecular area of stearic acid on a Langmuir-Blodgett (LB) film at pH 850,51 (0.295 and 0.221 nm2, respectively). We calculated the

molecular length (L) values on the basis of CoreyPauling-Koltun (CPK) molecular modeling. The crosssectional area of the oligomethylene chain (0.203 nm2 for hexagonal chain packing)52 is too small to perturb the packing of either headgroup. Approximately 250 nanotubes were picked randomly from the TEM images to evaluate the distribution and average inner diameter (Dobs-in) of the nanotubes. The obtained data are shown in Figure 12 and are summarized in Table 2 together with the inner diameter (Dcalc-in) values calculated from eq 1. Upon an increase in the spacer carbon number from n ) 14 to n ) 20, the average Dobs-in of the nanotubes increased from 17.7 nm to 22.2 nm in 1.5-nm steps. Flattening of nanotubes onto the TEM carbon grid did not occur because the height and periodic width of the nanotube arrays as observed using tapping mode AFM were similar.32 We obtained only one exception, for 1(12), which formed a relatively wide nanotube with Dobs-in ) 20.6 nm, comparable to that of 1(18). This anomaly probably occurred because the MLM became distorted from the unsymmetrical type, as mentioned above, which produced a Dobs-in value higher than the calculated Dcalc-in value (Figure 12a and Table 2, run 1). Thus, the inner diameter of these lipid nanotubes could be controlled merely by changing the oligomethylene

(48) The cross-sectional area of the sugar headgroup was estimated from single-crystal analysis of 2(14). (49) Masuda, M.; Shimizu, T. Carbohydr. Res. 1997, 302, 139. (50) Dhathathreyan, A. Colloids Surf. 1988, 29, 425. (51) Tomoaia-Cotisel, M.; Zsako, J.; Mocanu, A.; Lupea, M.; Chifu, E. J. Colloid Interface Sci. 1987, 117, 464.

(52) Ruocco, M. I.; Shipley, G. G. Biochim. Biophys. Acta 1982, 691, 309. (53) Popovitz-Biro, R.; Hill, K.; Shavit, E.; Hung, D. J.; Lahav, M.; Leiserowitz, L.; Sagiv, J.; Hsiung, H.; Meredith, G. R.; Vanherzeele, H. J. Am. Chem. Soc. 1990, 112, 2498. (54) Puggelli, M.; Gabrielli, G. Colloid Polym. Sci. 1983, 261, 82.

molecule of the sugar headgroup and the methylene chain, ∼0.295 and ∼0.20 nm2, respectively (Figure 10d).9,48 Controlling the Inner Diameter of Nanotubes. If a nanotube consists of unsymmetrical MLMs in which relatively larger headgroups occupy the outer surfaces of the MLM (Figure 11), the innermost diameter (Dcalc-in in Figure 11) can be defined by eq 1,32 where as and al are the cross-sectional areas of the small and large headgroup, respectively, and L is the molecular length.

Dcalc-in ) 2asL/(al - as)

(1)

5976

Langmuir, Vol. 20, No. 14, 2004

Masuda and Shimizu

Table 2. Inner Diameters of Lipid Nanotubes Based on Calculations and TEM Observations

run 1 2 3 4 5 6

chain length of 1(n)

molecular lengtha (L, nm)

calculated inner diameterb (Dcalc-in, nm)

observed inner diameterc (Dobs-in, nm)

n ) 12 n ) 14 n ) 16 n ) 18 n ) 20 peptideamphiphiled

2.62 2.87 3.12 3.38 3.64 4.33

15.6 17.1 18.6 20.2 21.7 73e

20.6 ( 1.9 17.7 ( 1.6 18.7 ( 1.6 20.8 ( 2.3 22.2 ( 2.1 50-70d

a The molecular lengths were calculated from CPK molecular modeling. b The Dcalc-in values were calculated by substituting each value of L, al ) 0.295 nm2, and as ) 0.221 nm2 into eq 1. The al and as values were evaluated on the basis of crystal analysis of the 1-D-glucosamide bolaamphiphile49 and the π-A isotherms of the LB films, respectively.50,51 c Dobs-in is the number-average inner diameter estimated from more than 250 nanotubes randomly chosen on TEM images. d Reference 17. e al and as were estimated as 0.255 and 0.228 nm2, respectively, on the basis of the π-A isotherms of the LB films made from peptide analogue lipids and stearylamine.53,54

spacer length (at least between n ) 14 and n ) 20), since the average of the observed Dobs-in values agrees well with the calculated Dcalc-in values. We similarly calculated the Dcalc-in value for a peptide-based lipid nanotube previously reported by Fuhrhop et al.17 The Dcalc-in value was also found to be consistent with Dobs-in (Table 2, run 6), indicating that eq 1 can be rationalized for the nanotubes consisting of unsymmetrical MLMs or unsymmetrical MLMs with H-T orientation at the interface. Effect of Carboxylic Headgroup Ionization on Self-Assembly Morphology. Self-assembly of the sodium salt of 1(18) revealed a remarkable effect of ionization of the carboxy headgroup on the tubular morphology and diameter (Figure 4f). The sodium salt of 1(18) (pH 7.9 at 0.2 mM) self-assembled to form a mixture of vesicles and nanotubes (white and black arrows, respectively, in Figure 4f). Furthermore, the tubular morphology was less uniform and appeared somewhat disordered, and the aspect ratio was less than that for the un-ionized 1(18) (Figure 4d). The Dobs-in value slightly increased (24.8 nm on average for 50 nanotubes) at the higher pH values. This is because ionic repulsion between the headgroups disorders the MLM by destroying the intermolecular hydrogen bonds. In addition, ionization increases the as and also the Dcalc-in values. Considering our previous results on proton triggered self-assembly of dicarboxylic oligopeptide bolaamphiphiles,40 we found that the protonation behavior of the carboxy headgroup of 1(n) remarkably affects the fine morphology of the present nanotubes. We previously found that the longer the oligomethylene spacer is, the more easily the bolaamphiphile can be protonated.40 The pKa value for the C12-linked dicarboxylic bolaamphiphiles was shifted to ∼6.2, clearly different from those of general carboxylic acids (pKa ∼ 3). Most of the carboxy headgroups in nanotubes prepared from salt-free 1(n) (n ) 14, 16, 18, and 20) seemed to be protonated because the pH of the nanotube dispersions was in the neutral range (6.0-7.2). Therefore, the as value of the longer-chain 1(n) compounds is constant in pure water and the Dobs-in value fit well with the calculated Dcalc-in value. Only the aqueous dispersion of the 1(12) nanotubes showed a weakly acidic pH value (4.7), probably due to ionization of the headgroup. This finding indicates that the carboxy headgroups of the other nanotubes are protonated.

In conclusion, we have found that unsymmetrical 1-glucosamide bolaamphiphiles 1(n) with even-numbered oligomethylene chains self-assemble in water to form nanotubes and three types of microtubes. The ratio of the molecular length (L) to the MLM thickness (d) within both tubular assemblies and single crystals indicates that the nanotubes are composed of unsymmetrical MLMs. This ratio also enabled us to determine the polymorph or polytype of the microtubes. The oligomethylene chain lengths of 1(n) greatly affected the MLM polytype or polymorph upon self-assembly. Unsymmetrical MLMbased nanotubes of 1(n) possess both an outer surface covered with sugar hydroxyl groups and an inner surface covered with carboxy groups. We were able to control the inner diameters of unsymmetrical MLM nanotubes by varying the length of the spacer chain. Experimental Section 1H

General. The and 13C NMR spectra were recorded with a JEOL 600 (600 MHz) or a JEOL GSX 270 (270 MHz) NMR spectrometer. Preparative column chromatography was performed on silica gel. The chromatographic purity of the intermediates was monitored by thin-layer chromatography (Kiesel gel F 254, Merck). The compounds were visualized by spraying the plates with 5% sulfuric acid in methanol and then charring them on a hot plate. The molecular lengths were estimated by molecular mechanics calculations performed with the CONFLEXMM2 force field as implemented in CAChe, version 4.1.1, Fujitsu Co. Ltd., Japan. The molecular length (L) was defined as the center-to-center distance between the hydrogen atoms of the terminal carboxy group and the sugar hydroxyl group at the C4 position, plus twice the van der Waals radius of a hydrogen atom (0.05 nm). Synthesis of ω-[(N-β-D-Glucopyranosyl)carbamoyl]alkanoic Acid, 1(n). All amphiphiles and intermediates were synthesized on the basis of the method previously reported.55 The details of the stoichiometry, procedure, and the chemical and physical properties are described below. 12-[(2,3,4,6-Tetra-O-acetyl-N-β-D-glucopyranosyl)carbamoyl]dodecanoic Acid, 3(12). Platinum(IV) oxide (90.8 mg, 0.4 mmol) was added to a solution of 2,3,4,6-tetra-O-acetyl-βD-glucopyranosyl azide55 (0.523 g, 2.0 mmol) in methanol (130 mL) under a nitrogen atmosphere. Hydrogen was introduced to the solution for 3 h. After filtration and evaporation of methanol, the residue was dissolved in dimethylformamide (DMF) (20 mL) containing pyridine (1.582 g, 20 mmol). To the mixture was added a dichloromethane (5 mL) solution of 1,12-dodecandioic acid dichloride55 (2.362 g, 8 mmol) at -10 °C. The reaction mixture was stirred at -10 °C for 1 h and at 20 °C for 20 h and was then poured into water. After this mixture was stirred for 2 h, the solvent was removed under reduced pressure. To the residue was added chloroform (200 mL), and the organic layer was washed with 5% sodium hydrogen carbonate aqueous solution (100 mL), 5% citric acid (100 mL), and water (100 mL). The organic layer was separated, dried over anhydrous sodium sulfate, and concentrated in a vacuum. The residue was subjected to column chromatography. Elution with a chloroform/methanol mixture (1% methanol in chloroform to 8% methanol in chloroform) gave tetraacetylated 3(12) as an amorphous solid (0.470 g, 34%): Rf ) 0.75 (chloroform/methanol ) 9:1, v/v); 1H NMR (in CDCl3, 22 °C) δ 6.30 (d, J ) 9.2 Hz, 1H, NH), 5.31 (t, J ) 9.5 Hz, 1H, H-3), 5.27 (t, J ) 9.5 and 9.2 Hz, 2H, H-1), 5.07 (t, J ) 9.9 and 9.5 Hz, 2H, H-4), 4.92 (t, J ) 9.9 and 9.5 Hz, 2H, H-2), 4.32 (dd, J ) 12.5 and 4.4 Hz, 1H, H-6a), 4.08 (dd, J ) 12.5 and 1.8 Hz, 1H, H-6b), 3.84 (ddd, J ) 12.5, 4.4, and 1.8 Hz, 1H, H-5), 2.34 (t, J ) 7.3 and 7.7 Hz, 2H, -CH2COO-), 2.18 (m, 2H, -CH2CONH-), 2.08, 2.06, 2.05, and 2.04 (s, 3H, CH3CO-), 1.63 (m, 2H, -CH2CH2COO-), 1.33 (m, 2H, -CH2CH2CONH-), 1.25 (m, 16H, -CH2-). 13-[(2,3,4,6-Tetra-O-acetyl-N-β-D-glucopyranosyl)carbamoyl]tridecanoic Acid, 3(13). Amorphous solid (42%): Rf ) 0.85 (chloroform/methanol ) 9:1, v/v); 1H NMR (in CDCl3, 22 (55) Masuda, M.; Shimizu, T. J. Carbohydr. Chem. 1998, 17, 405.

Lipid Nanotubes and Microtubes °C) δ 6.30 (d, J ) 9.2 Hz, 1H, -NH-), 1.25 (m, 18H, -CH2-); otherwise similar data as those for 3(12). 14-[(2,3,4,6-Tetra-O-acetyl-N-β-D-glucopyranosyl)carbamoyl]tetradecanoic Acid, 3(14). Amorphous solid (40%): Rf ) 0.82 (chloroform/methanol ) 20:1, v/v); 1H NMR (in CDCl3, 22 °C) δ 6.30 (d, J ) 9.2 Hz, 1H, -NH-), 1.25 (m, 20H, -CH2-); otherwise similar data as those for 3(12). 14-[(2,3,4,6-Tetra-O-acetyl-N-β-D-galactopyranosyl)carbamoyl]tetradecanoic Acid, 4(14). The details of the stoichiometry and procedure of the synthesis were the same as those for 3(14) except we started with 2,3,4,6-tetra-O-acetyl-β-Dgalactopyranosyl azide in this case.55 Amorphous solid (28%): Rf ) 0.24 (chloroform/methanol ) 20:1, v/v); 1H NMR (in CDCl3, 22 °C) δ 6.28 (d, J ) 9.2 Hz, 2H, -NH-), 5.44 (dd, J ) 2.3 Hz, 1H, H-4), 5.27 (t, J ) 9.2 Hz, 1H, H-1), 5.15 (dd, J ) 9.9 and 2.3 Hz, 1H, H-3), 5.10 (dd, J ) 9.9 and 9.2 Hz, 1H, H-2), 4.14 (ddd, J ) 9.9, 6.6, and 0.7 Hz, 2H, H-5), 4.06 (ddd, J ) 10.2, 9.9, and 6.6 Hz, 2H, H-6), 2.34 (dd, J ) 7.6 and 7.6 Hz, 2H, -CH2COO-), 2.20 (dd, 2H, J ) 7.3 and 7.6 Hz, -CH2CONH-), 2.15, 2.06, 2.04, and 2.00 (s, 3H, CH3CO-), 1.60 (m, 4H, -CH2CH2CONH- and -CH2CH2COO-), 1.25 (m, 12H, -CH2-). 16-[(2,3,4,6-Tetra-O-acetyl-N-β-D-glucopyranosyl)carbamoyl]hexadecanoic Acid, 3(16). Amorphous solid (33%): Rf ) 0.72 (chloroform/methanol ) 9:1, v/v); 1H NMR (in CDCl3, 22 °C) δ 6.30 (d, J ) 9.2 Hz, 1H, -NH-), 1.25 (m, 24H, -CH2-); otherwise similar data as those for 3(12). 18-[(2,3,4,6-Tetra-O-acetyl-N-β-D-glucopyranosyl)carbamoyl]octadecanoic Acid, 3(18). Amorphous solid (28%): Rf ) 0.4 (chloroform/methanol ) 9:1, v/v); 1H NMR (in CDCl3, 22 °C) δ 6.30 (d, J ) 9.2 Hz, 1H, -NH-), 1.25 (m, 28H, -CH2-); otherwise similar data as those for 3(12). 20-[(2,3,4,6-Tetra-O-acetyl-N-β-D-glucopyranosyl)carbamoyl]icosanoic Acid, 3(20). Amorphous solid (25%): Rf ) 0.40 (chloroform/methanol ) 9:1, v/v); 1H NMR (in CDCl3, 22 °C) δ 6.30 (d, J ) 9.2 Hz, 1H, -NH-), 1.25 (m, 32H, -CH2-); otherwise similar data as those for 3(12). 12-[N-β-D-Glucopyranosylcarbamoyl]dodecanoic Acid, 1(12). The tetraacetylated bolaamphiphile 3(12) (0.147 g, 0.5 mmol) in anhydrous methanol (5 mL) was treated with 0.05 M sodium methoxide (11 mL) at room temperature for 5 h with monitoring by thin-layer chromatography. The reaction mixture was neutralized with ion-exchange resin (Amberlite IR-120, H+ form), filtered, and concentrated. The resulting bolaamphiphile was dried at room temperature at 5 mmHg for 12 h and used for self-assembly without further purification (crystalline powder, 1.95 g, 89%); Rf ) 0.52 (chloroform/methanol/water ) 75:22.5:2.5, v/v/v); mp, 168.0-173.0 °C; 1H NMR (in DMSO-d6/ D2O ) 98:2, v/v, 60 °C) δ 4.64 (d, J ) 8.8 Hz, 2H, H-1), 3.63 (dd, J ) 11.7 and 1.8 Hz, 1H, H-6a), 3.43 (dd, J ) 11.7 and 4.8 Hz, 1H, H-6b), 3.18 (dd, J ) 9.5 and 7.7 Hz, 1H, H-4), 3.13 (dd, J ) 9.5 and 8.8 Hz, 1H, H-2), 3.10 (m, 1H, H-5), 3.07 (dd, J ) 9.5 and 7.7 Hz, 1H, H-3), 2.08 (m, 2H, -CH2CONH-), 2.01 (dd, J ) 7.3 Hz, 2H, -CH2COO-), 1.46 (m, 4H, -CH2CH2CO-), 1.25 (m, 16H, -CH2-). Anal. Calcd for C20H37N1O8‚1/2H2O: C, 56.06; H, 8.94; N, 3.27. Found: C, 56.19; H, 8.73; N, 3.05. 13-[N-β-D-Glucopyranosylcarbamoyl]tridecanoic Acid, 1(13). Crystalline powder, 92%; Rf ) 0.45 (chloroform/methanol/ water ) 65:30:5, v/v/v); mp, 167.3-175.0 °C; 1H NMR (in DMSOd6/D2O ) 98/2, v/v, 60 °C) δ 1.25 (m, 18H, -CH2-); otherwise similar data as those for 1(12). Anal. Calcd for C21H39N1O8: C, 58.18; H, 9.07; N, 3.23. Found: C, 58.19; H, 8.93; N, 3.16. 14-[N-β-D-Glucopyranosylcarbamoyl]tetradecanoic Acid, 1(14). Crystalline powder, 98%; Rf ) 0.55 (chloroform/methanol/ water ) 65:30:5, v/v/v); mp, 164.6-166.0 °C; 1H NMR (in DMSOd6/D2O ) 98/2, v/v, 60 °C) δ 1.25 (m, 20H, -CH2-); otherwise similar data as those for 1(12). Anal. Calcd for C22H41N1O8: C, 59.04; H, 9.23; N, 3.13. Found: C, 59.20; H, 9.30; N, 3.08. 14-[N-β-D-Galactopyranosylcarbamoyl]tetradecanoic acid, 2(14). The details of the stoichiometry and procedure of the synthesis were the same as those for 1(14) except we started with 4(14) in this case. Crystalline powder, 99%; Rf ) 0.44 (chloroform/methanol/water ) 65:30:5, v/v/v); mp, 176.4 °C; 1H NMR (in DMSO-d6/D2O ) 98/2, v/v, 23 °C) δ 4.67 (d, J ) 9.2 Hz, 1H, H-1), 3.71 (dd, J ) 2.6 and 3.3 Hz, 2H, H-4), 3.51 (dd, J ) 11.0 and 6.2 Hz, 1H, H-6a), 3.42 (dd, J ) 11.0 and 6.2 Hz, 1H,

Langmuir, Vol. 20, No. 14, 2004 5977 H-6b), 3.39 (t, J ) 9.2, 1H, H-2), 3.35 (dd, J ) 6.6 and 2.6 Hz, 1H, H-5), 3.32 (dd, J ) 9.2 and 3.3 Hz, 1H, H-3), 2.16 (t, J ) 7.3 and 7.7 Hz, 2H, -CH2CONH-), 2.08 (t, J ) 7.3 and 7.7 Hz, 2H, -CH2COO-), 1.49 (m, 4H, -CH2CH2CO-), 1.25 (m, 20H, -CH2). Anal. Calcd for C22H41N1O8: C, 59.04; H, 9.23; N, 3.13. Found: C, 59.20; H, 9.30; N, 3.08. 16-[N-β-D-Glucopyranosylcarbamoyl]hexacanoic Acid, 1(16). Crystalline powder, 96%; Rf ) 0.72 (chloroform/methanol/ water ) 65:30:5, v/v/v); mp, 168.7-169.5 °C; 1H NMR (in DMSOd6/D2O ) 98/2, v/v, 60 °C) δ 1.25 (m, 24H, -CH2-); otherwise similar data as those for 1(12). Anal. Calcd for C24H45N1O8: C, 60.61; H, 9.54; N, 4.01. Found: C, 60.47; H, 9.41; N, 2.86. 18-[N-β-D-Glucopyranosylcarbamoyl]octadecanoic Acid, 1(18). Crystalline powder, 98%; Rf ) 0.85 (chloroform/methanol/ water ) 65:30:5, v/v/v); mp, 169.8-172.6 °C; 1H NMR (in DMSOd6/D2O ) 98/2, v/v, 60 °C) δ 1.25 (m, 28H, -CH2-); otherwise similar data as those for 1(12). Anal. Calcd for C26H49N1O8: C, 61.56; H, 9.82; N, 2.76. Found: C, 61.69; H, 9.70; N, 2.65. 20-[N-β-D-Glucopyranosylcarbamoyl]icosanoic Acid, 1(20). Crystalline powder, 98%; Rf ) 0.83 (chloroform/methanol/ water ) 65:30:5, v/v/v); mp, 167.1-174.9 °C; 1H NMR (in DMSOd6/D2O ) 98/2, v/v, 60 °C) δ 1.25 (m, 32H, -CH2-); otherwise similar data as those for 1(12). Anal. Calcd for C28H53N1O8: C, 63.25; H, 10.05; N, 2.63. Found: C, 63.12; H, 10.13; N, 2.61. Separation of Nano- and Microtubes. The aqueous dispersions of the nano- and microtubes were transferred to a centrifuge tube (5 mL) and centrifuged (2000g, 30 min at 15 °C) to separate the nanotube components, which stayed in the supernatants, from the microtubes. TEM Observations. The aqueous dispersions of the nanoand microtubes (0.1 mg mL-1) were dripped onto an amorphous carbon grid, and excess water was blotted with filter paper. After drying in air for 5 min, the sample was negatively stained with phosphotungstate solution (2 wt %, pH adjusted to 2 for acidform nanotubes and to 8 for salt-form nanotubes). TEM was done with a Carl-Zeiss LEO912 instrument operated at 50 keV. Images were recorded on an imaging plate (Fuji Photo Film Co. Ltd. FDL5000 system) with 20-eV energy windows at 3000250 000× and were digitally enlarged. FT-IR Measurements. The FT-IR spectra of the separated nano- and microtubes were measured with a Fourier transform IR spectrometer (JASCO FT-620) operated at 4-cm-1 resolution with an unpolarized beam and attenuated total reflection (ATR) accessory system (Diamond MIRacle, horizontal ATR accessory with a diamond crystal prism, PIKE Technologies, USA) and a mercury cadmium telluride (MCT) detector. Several drops of the aqueous dispersions of the nano- and microtubes (0.1 mg mL-1) were dripped onto the prism and dried under nitrogen stream prior to measurement. XRD Measurements. The XRD of a freeze-dried sample was measured with a Rigaku diffractometer (Type 4037) using graded d-space elliptical side-by-side multilayer optics, monochromated Cu KR radiation (40 kV, 30 mA), and an imaging plate (R-Axis IV). The typical exposure time was 10 min with a 150-mm camera length. Freeze-dried nanotubes from 1(n) were vacuum-dried to constant weight and then put into capillary tubes, without being powdered.

Acknowledgment. We thank Dr. K. Yoza at Bruker AXS Co. Ltd. Japan for X-ray analysis of 1(12). Calculations were performed on computers of the Tsukuba Advanced Computing Center (TACC) at the National Institute of Advanced Industrial Science and Technology, Japan. Supporting Information Available: Optical microscopic images of the microtubes of 1(14) and 1(18) in water, AFM images of the nanotubes of 1(14) and 1(18) on mica, crystallographic data for 1(12), and details of the estimation of diameters based on the dimensions of unsymmetrical bolaamphiphiles. This material is available free of charge via the Internet at http://pubs.acs.org. LA049085Y