Introduction

One of the greatest findings in the 1990s was the self-assembly of cyclic molecules onto a linear polymeric chain, represented by α-cyclodextrin (α-CD) onto poly(ethylene glycol) (PEG)1, 2 and β-cyclodextrin (β-CD) onto poly(propylene glycol),3, 4 leading to the development of supramolecular materials based on cyclodextrins5, 6, 7, 8 and others.9, 10 In the assembly, CD molecules formed intermolecular hydrogen bonds2, 4, 6, 7 with each other to stay on the chain while threading and dethreading were competing, and finally a supramolecular assembly called pseudopolyrotaxane was obtained as a kinetic product. As a matter of course, some thermodynamic advantages accompanied this process, such as enthalpic gain on forming hydrogen bonds,6 and total entropic changes on assembly of the components, as well as desolvation. It seems that PEG and poly(propylene glycol) chains provided a suitable guide for α-CD and β-CD to fill the size-matched cavity and align cooperatively through formation of hydrogen bonds. In recent years, Takata et al. reported an excellent synthetic approach for yielding pseudopolyrotaxanes based on self-assembly of modified CD molecules onto a linear polymeric chain without relying on forming hydrogen bonds, but using heterogeneous systems in which permethylated α-CD and poly(tetrahydrofuran) or PEG were used in hydrocarbon solvents11 as well as in water.12 Even though the initial 1:1 complex is not necessarily preferred to form, threading occurs sequentially and the complexation reaction proceeds because dethreading is retarded through kinetic and/or thermodynamic stabilization of intermediary complexes in a cooperative manner, which seems unfavorable for a size-mismatched combination (Scheme 1a).2 Each cyclic molecule enters on threading and leaves at the end of a chain. Setting up a barrier at each end of a chain affects both entering and leaving,6, 7 and brings about changes in the ratio of kn/k−n. A precise barrier ideally allows a cyclic molecule to remain on any chain, instead of relying on specific combinations of cyclic and linear components with size adequacy (Scheme 1b). In addition, cyclic molecules would stay on chains cooperatively when they were crosslinked by a spacer, leading to a supramolecular network through the pseudopolyrotaxane formation (Scheme 1c).13, 14, 15, 16, 17

Here we report the roles of barriers at the ends of a chain, and crosslinking of cyclic molecules in self-assembly of size-mismatched components, PEG and β-CD.17, 18, 19 We investigated systematically the pseudopolyrotaxane formation of a series of dumbbell-shaped PEG chains 1a–e with native β-CD 2 and its permethylated derivative 320 (Scheme 1b and Figure 1). Two crosslinked β-CD derivatives, ditopic 417 and tetratopic 5, were used to form a supramolecular network on mixing with 1 in water (Scheme 1c and Figure 1).

Figure 1
figure 1

Chemical structures of dumbbell-shaped poly(ethylene glycol) chains 1a–e, β-cyclodextrin 2, permethylated β-cyclodextrin 3, ditopic crosslinker 4 and tetratopic crosslinker 5.

Full size image

Experimental procedure

Materials

Poly(ethylene glycol)s (average Mn: 380–420 and 950–1050) were purchased from Aldrich Co. Ltd. (St Louis, MO, USA), and used as a starting material to prepare PEG bis(prop-2-ynyl) ether 7 (matrix-assisted laser desorption/ionization time-of-flight mass spectrometry m/z 513 ([M(m=9)+Na]+, BP),21 and PEG bis(2-aminoethyl) ether (matrix-assisted laser desorption/ionization time-of-flight MS m/z 1096 ([M(n=22)+Na]+, BP).22 Pentaerythritol ethoxylate 16 (average Mn: ∼797) was purchased from Aldrich Co. Ltd., and derived to quadruply-azidated 10. Heptakis(2,3-di-O-methyl)-hexakis(6-O-methyl)-6-monodeoxy-6-monoazido-β-CD 8 and heptakis(2,3-di-O-methyl)-hexakis(6-O-methyl)-6-monodeoxy-6-monoamino-β-CD 3 were prepared according to literature.20 Synthetic procedures and spectral data of new compounds (1, 5, 9, 10, 12, 13 and 17), and experimental details of complexation study are described in Supporting Information.

Measurements

The frequency dependence of the oscillatory shear moduli, such as shear storage modulus G′ and loss modulus G″ in aqueous solutions, was measured by a cone-and-plate rheometer (MR500, UBM, Kyoto, Japan) at 25±1 °C. The cone angle was 5° and the diameter was 25 mm.

X-ray diffraction measurement was performed with a powder diffractometer (RINT2000, Rigaku, Tokyo, Japan) using graphite-monochromatized CuKα radiation (λ=1.542 ).

Matrix-assisted laser desorption/ionization time-of-flight mass spectra were recorded in the reflector mode on a mass spectrometer (Voyager DE RP, PerSeptive Biosystems, Framingham, MA, USA) using a nitrogen laser (337 nm) and an accelerating potential of 25 kV. α-Cyano-4-hydroxycinnamic acid was used as a matrix.

Results and discussion

Preparation of dumbbell-shaped PEG chains 1 and supramolecular crosslinkers, ditopic 4 and tetratopic 5

The dumbbell-shaped PEG chains 1a–e were prepared by a reaction of PEG (average n=22) bis(2-aminoethyl) ether with corresponding benzoyl chlorides 6 that have one or two alkyloxy group(s) at 2-, 4- or 6-position (Scheme 2). The ditopic crosslinker 4 was prepared by attachment of monoaminated permethylated β-CD to each end of PEG (average l=7) bis(carboxymethyl) ether through a condensation reaction.17 The tetratopic crosslinker 5 was prepared by repeated 1,3-dipolar cycloaddition reactions23, 24 of monoazidated permethylated β-CD 820 and quadruply-azidated pentaerythritol ethoxylate 10 (average a+b+c+d=15) at the respective ends of PEG (average m=9) bis(prop-2-ynyl) ether 7 (Scheme 3).

Inclusion complexation of dumbbell-shaped 1 with β-CD 2

Complexation of the dumbbell-shaped PEG chains 1 with native β-CD 2 was first investigated by mixing in water at room temperature and at 60 °C (Scheme 4 and Table 1). Mixing of 1a with 2 in water at room temperature for 10 days successfully gave a white solid. The 1H nuclear magnetic resonance spectrum of the isolated solid measured in DMSO-d6 indicated a simple integration of 1a and 2 at a ratio of 1 to 10. An indistinguishable solid was given on mixing at 60 °C from the solid formed at room temperature. The powder X-ray diffraction pattern for the solid of 1a and 2 is quite similar to that of β-CD itself.4, 25 Mixing of 1a with permethylated 3 failed to produce any solid, and just gave a transparent solution at both room temperature and 60 °C. If we consider the hydrophobic inherence in 1a, whose solution is turbid in water at any temperature, the solubilization would be accounted for by assuming inclusion complexation with 3 at least at the end of 1a.17 It would be difficult to judge whether the methylated β-CD was trapped onto 1a or not from only the appearance due to high water solubility of 3.18 Similar results were obtained from mixing of 1b with 2 or 3 qualitatively, in terms of precipitation or solubilization (Table 1). Note that mixing at 60 °C accelerated precipitation,7 and gave an indistinguishable solid in a similar yield to that given at room temperature. A longer time was required for precipitation that resulted in a higher yield for 1a [R2=OPr, R4=OBn] than for 1b [R2=OEt, R4=Obn] (Table 1). All solids obtained here dissolved in DMSO or in an excessive amount of water, indicating formation of hydrogen bonds as commonly seen in an inclusion complex of native cyclodextrins.1, 2, 3, 4, 26, 27 As previously reported, the results of mixing PEG bis(2-aminoethyl) ether, which has no bulkiness at the ends, with α-CD1, 2 and β-CD3, 4 are also listed in Table 1 for reference. As in the case of the mixing of 1a with 3, no precipitates were formed from a transparent solution of 1c [R2=OPr, R4=OPr], 1d [R2=OPr] or 1e [R2=OPr, R6=OPr] in the presence of 2 at any temperature (Scheme 4).

Table 1 Conditions (concentration and temperature) for complexation of 1a, 1b or PEGBA with 2, 3 or α-CD,a and results (time for precipitation, ratiob of PEG to CD in an isolated solid and yieldc)
Full size table

Supramolecular crosslinking of dumbbell-shaped 1 with ditopic 4 and tetratopic 5

Complexation of dumbbell-shaped 1 with the ditopic crosslinker 4 was then investigated in terms of the zero-shear viscosity (η0) of a transparent solution in water under several conditions (Table 2). If each of the two permethylated β-CD molecules in 4 is discretely assembled onto a chain, and the supramolecularly crosslinked complexes are stable, this will lead to an increase in viscosity.28 Mixing of 1a/b (46 mM) with 4 (0.25 M) at room temperature for 10/5 days led to no change in viscosity (Figures 2a and b), but the solution appeared to be similar to a solution of 1a and permethylated 3. Alternatively, heating these systems induced increases in viscosity over time, which attained an order of 105 Pa•s after periods of 6 days for 1a [R2=OPr, R4=OBn], and 3 days for 1b [R2=OEt, R4=OBn] (Scheme 5 and Figures 2a and b). If the increase in viscosity was achieved through inclusion complexation with 4 only at the ends of 1a/b, the solution viscosity would decrease with heating.29, 30, 31, 32 It is noteworthy that a gradual increase in viscosity was also found in a solution of 1c (46 mM) [R2=OPr, R4=OPr] in the presence of 4 (0.24 M) at 40 °C (Scheme 5 and Figure 2c), if we consider the failure of precipitation to occur on mixing of 1c with native β-CD 2 at any temperature (Scheme 4). The viscosity did not increase at room temperature as in the case of mixing of 1a or 1b with 4 (Figure 2). Also note that these increased viscosities still remained for weeks after the system cooled down, and returned to solutions with low viscosity (∼10−2 Pa•s) by triple dilution with water followed by shaking at room temperature for 5 days for 1a, for 2 days for 1b and within 1 day for 1c (Scheme 5). These results indicate that the pseudopolyrotaxanes were formed on heating,18 and were kinetically stable during the supramolecular crosslinking. It is important for the system to be heated to form an assembly through the supramolecular crosslinking, although slipping on/off is allowed even at room temperature (see above). The values of η0 measured at low concentrations ([1a–c]<20 mM, [4]<0.10 M) were on the order of ∼100 Pa•s. The values increased with concentration as follows: η0=2.1 × 103 Pa•s for 1a and 4, and 1.9 × 102 Pa•s for 1b and 4 ([1a], [1b]=31 mM, [4]=0.16 M) and so on, measured after maintaining a solution at 40 °C for 8 days. Solutions with so high concentrations are no longer allowed to measure nuclear magnetic resonance with suitable quality. In a transparent solution of 1d, 1e or PEG bis(2-aminoethyl) ether in the presence of 4, the viscosity remained on an order of 10−2 Pa•s under any conditions (Scheme 5). These results are also summarized in Table 2.

Table 2 Conditions (concentration,a temperature and time) for mixing of 1a–e or PEGBA with 4 or 5, and zero-shear viscosity
Full size table
Figure 2
figure 2

Zero-shear viscosities (η0) of aqueous solution of (a) dumbbell-shaped PEG 1a, (b) 1b and (c) 1c (46 mM) in the presence of ditopic crosslinker 4 (0.25 M), determined by measurement of complex viscosities (η*) at 25±1 °C, followed by calculation according to the equation η0≡limω → 0(G″/ω). For a solution of 1b stirred at 40 °C for 6 days (Figure 2b), the viscosity was not determined due to disappearance of the terminal region.

Full size image

A solution with increased viscosity becomes harder to stir. In fact, a stirring bar stopped on the 8th/6th day from the time a solution of 1a/b and 4 was heated to 40 °C, whereas it had been continuing in a solution of 1c and 4 for 8 days at that temperature. As shown in Figure 3, the solution of 1a and 4 appears deceptively something similar to a gel in the photograph (Figure 3a), and is fluent in the long term (∼2 days; Figure 3b). We describe these macroscopic observations through dynamic viscoelastic measurement.33 For the solution of 1a/b and 4 stirred at 40 °C for 8/6 days, we can find the rubbery region in high frequency, indicating that the solution property changed from viscous to elastic (Figures 4a and b). Alternatively, for the solution of 1c and 4, the terminal region was clearly detected (Figure 4c). Changes over time in the oscillatory shear moduli are summarized in Supplementary Figures S1–S3.

Figure 3
figure 3

Photographs of aqueous solution of dumbbell-shaped poly(ethylene glycol) 1a (46 mM) and ditopic crosslinker 4 (0.25 M), stirred at 40 °C for 8 days, followed by inverting and standing for (a) 0 and (b) 24 h.

Full size image
Figure 4
figure 4

Frequency dependency of storage modulus (G′, red circle) and loss modulus (G″, blue square) of aqueous solution of (a) dumbbell-shaped poly(ethylene glycol) 1a in the presence of ditopic crosslinker 4 stirred at 40 °C for 8 days, (b) 1b in the presence of 4 stirred at 40 °C for 6 days and (c) 1c in the presence of 4 stirred at 40 °C for 8 days ([1]=46 mM, [4]=0.25 M), measured at 25±1 °C.

Full size image

We found a remarkable change in the dynamic viscoelastic property when the tetratopic crosslinker 5 was used in complexation with 1a. In a solution of 1a (19 mM) and 5 (52 mM) stirred at 40 °C, a stirring bar stopped on the 3rd day. The frequency dependence of the oscillatory shear moduli showed plateaus at low frequency, and indicated that a network structure had formed in the transparent solution (Figure 5a).33 Once formed, returning to a low viscosity solution required a 7-day shaking after triple dilution. Replacing the tetratopic 5 with the ditopic 4 (0.10 M) in the above solution resulted in the value of η0 lying on the order of 100 Pa•s, although almost the same amount of permethylated β-CD was present in the solution (Table 2). In a solution of 1c (19 mM) and 5 (52 mM) stirred at 40 °C for 3 days, the terminal region was found (Figure 5b), which was not detected when the ditopic 4 (0.10 M) was used in place of 5 (Table 2).

Figure 5
figure 5

Frequency dependency of storage modulus (G′, red circle) and loss modulus (G″, blue square) of aqueous solution of (a) dumbbell-shaped poly(ethylene glycol) 1a in the presence of tetratopic crosslinker 5, and (b) 1c in the presence of 5 stirred at 40 °C for 3 days ([1]=19 mM, [5]=52 mM), measured at 25±1 °C. In a solution of 1a/c (19 mM) itself or 5 (52 mM) itself, the values of G′ and G″ were on the order of ∼10−2 Pa, measured under the same conditions.

Full size image

Conclusions

A series of dumbbell-shaped PEG chains 1 were designed and prepared to investigate systematically the role of barriers at the ends of 1 through the self-assembling of size-mismatched cyclic and linear components (pseudopolyrotaxane formation). Differences in chemical structure at the ends of a chain were projected on macroscopic behaviors: substituents on 2,4-positions (a, b or c) affected the rate and yield of precipitation in the complexation with native β-CD that forms hydrogen bonds, and the rates of assembly as well as dissociation in complexation with a ditopic/tetratopic supramolecular crosslinker 4/5 that kinetically stabilizes the assembly without relying on hydrogen bonds. In addition, the self-assembling of 1a–c with 4/5 was induced only at 40 °C or higher, although a number of supramolecular assemblies dissociate on heating.1, 29, 30, 31, 32, 34, 35, 36, 37, 38 Substituent(s) on 2- or 2,6-positions (d or e) were too small or bulky to work well. The system was also well modulated by crosslinking of a cyclic component, as demonstrated by combinations of 1a–c and monotopic 3/ditopic 4, or 1a/c and ditopic 4/tetratopic 5. A precise design of end groups according to a cyclic component would broaden the range of possibilities in selecting a chain without relying on a specific interaction between the components.

scheme 1

(a) Pseudopolyrotaxane formation between size-matched cyclic and linear components, (b) pseudopolyrotaxane formation between size-mismatched cyclic and dumbbell-shaped linear components and (c) supramolecular crosslinking based on pseudopolyrotaxane formation between size-mismatched crosslinked cyclic and dumbbell-shaped linear components.

scheme 2

Preparation of dumbbell-shaped poly(ethylene glycol) chains 1. PEGBA, poly(ethylene glycol) bis(2-aminoethyl) ether

scheme 3

Preparation of tetratopic crosslinker 5.

scheme 4

Mixing of 1 with 2 in water leading to precipitation (1a, 1b) at room temperature (r.t.) or elevated temperature, or no precipitation (1c, 1d and 1e), and dissociation by dilution.

scheme 5

Mixing of 1 with 4 in water leading to increase in viscosity only at elevated temperature (1a, 1b and 1c) or no increase (1d, 1e), and dissociation by dilution.