Supramolecular hydrogels formed by β-cyclodextrin self-association and host–guest inclusion complexes

Abstract

Supramolecular hydrogels are highly interesting for drug delivery and tissue engineering applications, especially those systems that display a combination of tunable properties, high mechanical strength and easy preparation from well-available and biocompatible building blocks. In the present paper, we show that the combination of free β-cyclodextrin (βCD) and 8-arm or linear cholesterol-derivatized poly(ethylene glycol) (PEG–chol) in aqueous solution resulted in the formation of almost fully elastic gels with storage moduli in the range of 10–500 kPa. X-Ray diffraction measurements demonstrated the presence of crystalline βCD domains in the hydrogel networks. Rheological experiments further proved that hydrogel formation is based on inclusion complex formation between these βCD clusters and cholesterol coupled to the terminal end of PEG. The observation that the gels were weakened by addition of the competitive βCD–guest molecule adamantanecarboxylic acid (ACA) supported the proposed gelation mechanism. The gel mechanical properties were dependent on temperature, concentration of cholesterol-derivatized PEG and/or βCD, PEG's molecular weight and its architecture. This hydrogel system can be considered as an excellent candidate for future applications in the biomedical and pharmaceutical fields.

---

1. Introduction

During the last decades, hydrogels have attracted great attention as scaffolds for tissue engineering and as drug delivery matrices.^1–4 These materials are networks of hydrophilic polymers that can retain and/or absorb considerable amounts of water.⁵ In general, the high water content of hydrogels leads to a good biocompatibility, because their relatively soft and rubbery appearance minimizes irritation of the surrounding tissue.^6,7 In hydrogel networks, dissolution of the hydrophilic polymers is prevented by cross-links, which can be permanent or reversible.^1,8–12 A wide variety of chemical cross-linking methods is known to accomplish the formation of covalent bonds between hydrophilic polymer chains in hydrogels, e.g. radical,^13–15 high-energy¹⁶ or enzyme-mediated polymerization¹⁷ and click chemistry.^18,19 Because chemical cross-linking involves reaction conditions that might structurally modify encapsulated therapeutics and often require toxic cross-linking reagents, systems in which the network formation is based on reversible, physical interactions are presently preferred. Many physical interactions have been used to design these non-permanent networks, such as hydrophobic,^20,21 ionic,^22,23 hydrogen bonding^24,25 and biomimetic^26–28 interactions as well as the formation of crystalline domains²⁹ and stereocomplexes.^30–33

A physical driving force that has recently been exploited for the design of reversible, self-assembled hydrogels, is the inclusion complex formation between β-cyclodextrin (βCD) and lipophilic guest molecules.^34–41 βCD molecules are cyclic ‘doughnut’-shaped oligosaccharides composed of 7 dextrose units that are coupled via α-1,4-glucosidic linkages.⁴² All hydroxyl groups are located at the outer surface of the molecule, which renders the inner cavity of βCD relatively hydrophobic. Driven by hydrophobic and van der Waals interactions,^43,44 this cavity can act as a binding site for a wide range of lipophilic compounds (e.g. adamantane, cholesterol and aromatic compounds^34–41). Kretschmann et al. designed a thermoreversible hydrogel system based on adamantane-grafted N-isopropylacrylamide copolymers and a βCD dimer.³⁴ Other researchers prepared hydrogels by combining adamantane- and βCD-derivatized chitosan³⁵ or hyaluronic acid.^36,37 Molecules other than adamantane were utilized as guest compounds as well. For example, Hashidzume et al. described stimuli-responsive gels formed by poly(acrylamide) derivatized with either βCD or aromatic guest molecules.³⁸ Recently, our group reported on self-assembling hydrogels obtained by combining cholesterol- and βCD-derivatized 8-arm star-shaped poly(ethylene glycol) (PEG₈) in aqueous solution.^39–41 This gel system is highly versatile as its mechanical properties could easily be controlled by the polymer concentration, βCD–cholesterol stoichiometry, the PEG molecular weight and other parameters.^39,40 Degradation of these hydrogels is mainly mediated by a surface erosion process, which led to quantitative and nearly zero-order release of entrapped model proteins.⁴¹

Although the hosting properties of βCD have been investigated for more than four decades,^42,45 only recent studies have demonstrated that native, unmodified βCD molecules self-aggregate.^46–50 Light scattering experiments on aqueous solutions of βCD, in the concentration range of 4–12 mM, showed the formation of βCD nano-aggregates with average diameters of about 200 nm.^46,47 Bonini et al. visualized these nanostructures with transmission electron microscopy at cryogenic temperature (Cryo-TEM).⁴⁸ It was found that polydisperse nearly spherical structures with diameters in the range of 100–200 nm were predominant at low concentrations, while at higher concentrations (≥6 mM) micrometer planar aggregates became the main structures.

The ability of cyclodextrins to crystallize has previously been used to design so-called polypseudorotaxane-based hydrogel systems.^51–55 Polypseudorotaxanes are inclusion complexes in which cyclic molecules, e.g. αCD, a cyclodextrin subtype composed of 6 dextrose units, are threaded onto a polymer chain.⁵⁶ The dense packing of multiple αCD units onto the PEG chains leads to crystalline domains that act as physical cross-links in hydrogels.^51–55 Because of its larger inner cavity, βCD is not able to stably complex with PEG, but instead the bulkier polymer poly(propylene glycol) (PPG) was used as guest polymer to develop hydrogels.⁵⁷ Besides these polypseudorotaxane-based systems, no other hydrogels have been designed yet in which crystalline cyclodextrin domains are exploited.

In the present paper, we describe a novel self-assembled hydrogel system, which consists of cholesterol-derivatized 8-arm PEG and free βCD molecules in an aqueous solution. In these hydrogels, cholesterol–βCD inclusion complexes as well as βCD self-association are the driving forces in network formation. The presence of crystalline βCD nanoclusters was studied by X-ray diffraction measurements and the tailorability of the gel mechanical properties was investigated by rheological analyses.

2. Experimental part

Materials and polymer synthesis

All chemicals were used as received. Star-shaped 8-arm poly(ethylene glycol)s (PEG₈–OH) were purchased from JenKem Technology USA (Allen, USA). Products with different molecular weights were used: PEG₈10K–OH (M_n = 9656 Da (MALDI), PDI = 1.10), PEG₈20K–OH (M_n = 20 182 Da (MALDI), PDI = 1.08) and PEG₈40K–OH (M_n = 42 680 Da (MALDI), PDI = 1.06). Linear monomethoxy-poly(ethylene glycol) (mPEG5000–OH) and poly(ethylene glycol) (HO–PEG6000–OH) were obtained from Sigma-Aldrich (Zwijndrecht, The Netherlands) and Fluka (Buchs, Switzerland), respectively. The M_n's of these linear PEG's were determined by ¹H NMR spectroscopy using trichloroacetyl isocyanate (TCAI, Sigma-Aldrich, Zwijndrecht, The Netherlands) as a shift reagent.⁵⁸M_n values of the mPEG5000–OH and HO–PEG6000–OH products were 5.1 kDa and 6.8 kDa, respectively. The hydroxyl end groups of the star-shaped and linear PEG's were derivatized with either cholesterol or β-cyclodextrin (βCD) moieties using a biodegradable succinyl linker (SA), and the resulting polymers were characterized by ¹H NMR, GPC and polarimetry as previously reported.³⁹ A specific 8-arm polymer is referred to as PEG₈xxK–chol_y or PEG₈xxK–βCD_y, where xx represents the PEG MW (e.g., 40, 20 or 10 kDa), and y indicates the DS (degree of substitution, defined as the average number of either cholesterol or βCD groups per PEG molecule). Linear bifunctional and monofunctional PEG–chol polymers are denoted as PEG6K–chol_1.8 and mPEG5K–chol_0.9.

β-Cyclodextrin (βCD), 6-monodeoxy-6-monoamino-β-cyclodextrin (βCD–NH₂) and 1-adamantanecarboxylic acid (ACA) were provided by Sigma-Aldrich (Zwijndrecht, The Netherlands). Ammonium acetate (NH₄OAc) was purchased from Merck (Darmstadt, Germany) and 37% HCl solution in water was ordered from Acros Chimica (Geel, Belgium).

Hydrogel preparation and rheological characterization

Mixtures of 8-arm PEG–chol or linear PEG–chol with different amounts of βCD (0.5, 1, 2, 4 or 10 eq. βCD relative to cholesterol groups) were dissolved in 5 mM NH₄OAc buffer (pH 4.7) to obtain 2% (w/w) solutions. These solutions were subsequently lyophilized and the final hydrogels or aqueous mixtures (containing 2–25% (w/w) PEG–chol with 0.5–24.8% (w/w) βCD) were obtained by hydration of the lyophilized mixtures for 16 h at 4 °C with appropriate amounts of 100 mM NH₄OAc buffer (pH 4.7, buffer content: 52.7–97.5% (w/w)). During hydration, NH₄OAc buffer at pH 4.7 was used to minimize hydrolysis of PEG–chol's ester linkages.³⁹ Mixtures were also prepared containing non-derivatized PEG or monoamino-functionalized βCD (βCD–NH₂ · HCl) instead of native βCD. Adamantanecarboxylic acid (ACA) was used to competitively displace cholesterol from the βCD cavities. Therefore, 3 eq. ACA relative to the number of cholesterol moieties were added to a 2% (w/w) solution of 8-arm PEG–chol and βCD (2 eq. βCD relative to cholesterol). After lyophilization of the resulting mixtures, hydrogels were prepared as mentioned above.

Rheological characterization of the hydrogels was done with an AR-G2 rheometer (TA instruments, Etten-Leur, The Netherlands) equipped with a 1° steel cone geometry of 20 mm diameter and solvent trap. Using a spatula or pipette (for liquids), approximately 55 µL sample were placed between the pre-heated (40 °C) plates of the rheometer. Rheological gel characteristics were monitored by oscillatory time sweep experiments at 4, 20 and 37 °C and temperature sweep experiments. During time sweep experiments the G′ (shear storage modulus) and G″ (loss modulus) were measured for a period of 5 min. Temperature sweep experiments from 4 to 80 °C were done at a heating rate of 2 °C min⁻¹ (30 s equilibration per point). The point at which G″/G′ (= tan δ) = 1 is considered as the gel transition temperature (T_gel).⁵⁹ All experiments were performed at a frequency of 1 Hz and 1% strain.

X-Ray diffraction analysis (XRD)

X-Ray diffraction patterns of hydrogels as well as the lyophilized mixtures were recorded with a Nonius κ-CCD diffractometer using MoKα radiation (λ = 0.7107 Å) and a graphite monochromator. All patterns were recorded at a sample-to-detector distance of 75 mm; the maximum scattering angle (2θ) was 22°. Separate blank patterns were recorded to allow subtraction of air-scattering. The hydrogel data were also corrected with a background profile collected from deuterated water. The two-dimensional X-ray scattering images were transferred into one-dimensional intensity profiles with 2θ as x-axis.

3. Results and discussion

Hydrogel formation

To ensure homogeneity, hydrogels were prepared by first dissolving cholesterol-derivatized 8-arm or linear PEG and free βCD at a low concentration (2% (w/w)) in a 5 mM ammonium acetate buffer (pH 4.7). At this concentration, all components are well soluble. After lyophilization, the resulting mixture was hydrated in a smaller volume of 100 mM ammonium acetate buffer (pH 4.7) to obtain spontaneously formed hydrogels.

Fig. 1 shows a photograph of hydrated mixtures containing 10% (w/w) cholesterol-derivatized and non-derivatized 8-arm PEG with or without 2 eq. free unmodified βCD relative to the number of cholesterol or hydroxyl groups of the corresponding PEG component. A solution of non-derivatized 8-arm PEG (PEG₈20K–OH) behaved as a viscous liquid (D). 10% (w/w) PEG₈20K–OH in the presence of 2 eq. βCD led to a viscous suspension (C) likely due to βCD's low maximum solubility of ∼16 mM (1.8% (w/w)).⁴⁸ However, when using cholesterol-derivatized PEG (PEG₈20K–chol_5.6) with or without βCD, hydrogels were formed. Gels composed of PEG₈20K–chol_5.6 were almost transparent (A). Their formation is likely caused by the amphiphilicity of the polymer, where hydrophobic interactions occur between the PEG-bound cholesterol groups. Sample B in Fig. 1 shows that a white opalescent gel was formed by mixing PEG₈20K–chol_5.6 and 2 eq. βCD. This gel formation was somewhat surprising, since it was found in previous studies that disruption of self-assembling systems based on hydrophobic interactions occurred by adding small amounts of βCD (<16 mM).^37,60,61 For example, Akiyoshi et al. added βCD molecules to hydrogel nanoparticles based on cholesterol-grafted pullulan⁶⁰ or poly(L-lysine),⁶¹ causing the hydrophobic cholesterol groups to be captured by βCD complexation and consequently disrupting the nanogels. It should be mentioned that we used higher amounts of βCD (5.5% (w/w)), i.e. well above the solubility limit of βCD. This interesting phenomenon prompted us to investigate the gelation mechanism in more detail.

Fig. 1 Photographs of a 10% (w/w) PEG₈20K–chol_5.6 hydrogel (A), a hydrogel composed of 10% (w/w) PEG₈20K–chol_5.6 + 2 eq. βCD (B), an aqueous mixture of 10% (w/w) PEG₈20K–OH + 2 eq. βCD (C), and of a 10% (w/w) PEG₈20K–OH solution (D).

X-Ray diffraction analysis

The maximum solubility of βCD in water is ∼16 mM (1.8% (w/w)).⁴⁸ However, several studies have reported the formation of nano-aggregates below this concentration.^46–48 For most of the hydrogels investigated in this study (see ESI†, Tables S1 and S2), upon rehydration of the freeze dried mixtures, the βCD concentration is higher than its maximum solubility. This fact, together with the opalescent appearance of the PEG₈20K–chol_5.6–βCD based gels in comparison with the transparent PEG₈20K–chol_5.6 gels, supports the hypothesis of the presence of insoluble βCD clusters in the PEG₈20K–chol_5.6–βCD gels that might play a role in the observed gel formation.

The cluster formation of βCD was investigated by X-ray diffraction analysis. Fig. 2 shows that for the PEG₈20K–chol_5.6 powder (A) two major diffraction peaks were observed at 2θ (λ = 0.7107 Å) = 8.8° and 10.8° together with less intense peaks at 2θ = 12.3, 16.4, 18.7 and 20.6°. By using Bragg's law of diffraction⁶² (2d sin θ = λ, where d, θ and λ represent the crystal lattice spacing, the diffraction angle and the wavelength, respectively), this pattern can be attributed to the monoclinic unit cell of crystalline PEG.^63–65 The βCD powder (Fig. 2, trace B) also showed a high degree of crystallinity with many diffraction peaks, which were found to correspond to the reported X-ray structure of the βCD hydrate.⁶⁶ To estimate the size of the crystallite spheres in the βCD powder, the peak at 2θ = 5.7° was selected and the Scherrer equation⁶⁷ was applied:

where τ is the mean crystallite dimension, K is the shape factor (0.89), λ is the X-ray wavelength (0.7107 Å), β is the line broadening at half the maximum intensity and θ represents the diffraction angle. This resulted in an estimated crystallite diameter of approximately 100 Å. By comparing the volume of a crystallite sphere (volume = 4/3πr³, where r is the radius (50 Å)) with the volume of a single βCD molecule ((for simplicity) considered as a cylinder with volume = πr²h, where r and h are the radius (7.65 Å) and height (7.9 Å), respectively),⁴⁵ it is calculated that the structurally ordered domains are built up of ∼360 aggregated βCD units. Fig. 2 also displays the X-ray diffraction patterns of hydrogels composed of 15% (w/w) PEG₈20K–chol_5.6 with or without 1 or 2 eq. βCD. The PEG₈20K–chol_5.6 gel (E) was found to be amorphous, whereas in the PEG₈20K–chol_5.6–βCD gels (C–D) crystallinity was detected with a diffraction pattern that in the 2θ range of 2–12° can be mainly ascribed to the βCD powder. This confirms that the PEG₈20K–chol_5.6–βCD gels indeed contain crystalline domains of βCD molecules. Compared to the βCD powder, the X-ray diffraction peaks of the PEG₈20K–chol_5.6–βCD gels were broader. Because the peak width is inversely proportional to the crystallite size (eqn (1)), this suggests that in the hydrogels the crystallites are smaller than in the βCD powder. Although this cannot be directly related to the total size of the βCD clusters in the gels, due to potential amorphous clustering of the βCDs, it is legitimate to conclude that the structurally ordered part of the insoluble aggregates has a size <100 Å.

Fig. 2 X-Ray diffraction patterns of PEG₈20K–chol_5.6 (A) and βCD (B) powders and 15% (w/w) gels composed of PEG₈20K–chol_5.6 with 1 (C) or 2 (D) eq. βCD or without βCD (E). Patterns for (C), (D) and (E) were corrected with a background profile collected from deuterated water.

Gelation mechanism

Fig. 3 shows the rheological properties at 4 and 37 °C of hydrogels, composed of 10% (w/w) PEG₈20K–chol_5.6 or 10% (w/w) PEG₈20K–chol_5.6 with 2 eq. βCD relative to cholesterol. At both temperatures, viscoelastic hydrogels were formed with 10% (w/w) solutions of PEG₈20K–chol_5.6 with or without added βCD, i.e. the storage modulus (G′) exceeds the loss modulus (G″) in all cases. Interestingly, the addition of free βCD to a PEG₈20K–chol_5.6 solution led to a considerable increase in hydrogel strength as demonstrated by a ∼50 to 70-fold increase of G′ and a decrease in tan δ from 0.34 to 0.01 at 4 °C. Fig. 3 also shows that with increasing the temperature from 4 to 37 °C the G′ of both gels decreased, whereas tan δ increased (0.7 ± 0.2 and 0.14 ± 0.01 for, respectively, the PEG₈20K–chol_5.6 and PEG₈20K–chol_5.6 + 2 eq. βCD gels), indicating that the interactions which hold together the network structure became weaker at higher temperature.

Fig. 3 Rheological properties at 4 and 37 °C of mixtures containing 10% (w/w) PEG₈20K–chol_5.6 and 10% (w/w) PEG₈20K–chol_5.6 + 2 eq. βCD (measurement frequency = 1 Hz). Left axis: G′ and G″; right axis: tan δ. Data are shown as mean ± standard deviation, n = 3.

To further investigate the thermosensitivity of the viscoelastic gels, their G′ and G″ were measured while gradually increasing the temperature from 4 to 80 °C. With increasing temperature, G′ (and to a lower extent G″) gradually decreased, and at a certain temperature (T_gel) G′ equaled G″, which indicates a gel-to-sol transition. Above T_gel, G″ exceeded G′ and the mixtures behaved as viscous liquids. Besides higher gel strengths, the presence of βCD also gave the 10% (w/w) PEG₈20K–chol_5.6–βCD gel a higher T_gel of 53 ± 3 °C as compared to 42 ± 4 °C for the 10% (w/w) PEG₈20K–chol_5.6 gel (see ESI†, Table S3). We previously described a hydrogel system with a similar thermosensitive behavior.^39,40 In that system, composed of cholesterol- and βCD-derivatized 8-arm PEG (i.e. PEG₈20K–chol_6.1 and PEG₈20K–βCD_7.4; see Experimental part for polymer nomenclature), gel formation is driven by inclusion complex formation of cholesterol and βCD moieties coupled to the terminal ends of 8-arm PEG. An elevation of the temperature in this system resulted in a reduced number of βCD–cholesterol complexes, which ultimately led to a gel-to-sol transition. At a total PEG content of 10% (w/w), the T_gel of this particular gel system was 20 ± 1 °C, so that at 37 °C it behaved as a viscous liquid with low G′ and G″ between 2 and 35 Pa.³⁹ Compared to this previously reported PEG₈20K–chol_6.1–PEG₈20K–βCD_7.4 system, gels composed of PEG₈20K–chol_5.6 and free βCD showed higher T_gel and improved mechanical strength at elevated temperatures (G′ ≈ 500 kPa), which is beneficial for future applications.

The observation that an aqueous mixture of βCD molecules and cholesterol-derivatized 8-arm PEG forms gels that are stronger than the gels containing only PEG₈20K–chol suggests that the βCD molecules dramatically affect the hydrogel network structure. As discussed in the Introduction section, several groups have previously reported on the formation of βCD aggregates in water.^46–50 The increased G′ and lower tan δ of the PEG₈20K–chol_5.6/βCD gels as compared to PEG₈20K–chol_5.6 gels indicate that βCD increases the cross-link density of PEG₈–chol based hydrogels. This strongly suggests that the βCD aggregates form host–guest inclusion complexes with PEG-bound cholesterol moieties. In this way, the βCD nano-aggregates serve as cross-linkers in the polymer network, as depicted in Fig. 4.

Fig. 4 Schematic representation of 8-arm PEG–chol/βCD gels. Cholesterol groups at the termini of the 8-arm PEG's form inclusion complexes with crystalline nanoclusters of βCD's (A). βCD clusters with arbitrary size and crystal packing are shown. Dependent on the relative number of cholesterol and βCD moieties, hydrophobic cholesterol–cholesterol interactions might also occur (B).

Fig. 5 shows G′ and G″ at 37 °C of aqueous mixtures containing 20% (w/w) PEG₈20K–chol_5.6, linear bifunctional PEG6K–chol_1.8 or monofunctional mPEG5K–chol_0.9, combined with 2 eq. βCD (relative to PEG-bound cholesterol). This figure shows that mixtures based on monofunctional linear PEG–cholesterol and βCD did not result in the formation of viscoelastic networks (tan δ ≫ 1). This is not unexpected, because with only 1 interacting cholesterol group per PEG chain it is not possible to form a network structure. Fig. 5 also shows that the mixtures based on 8-arm PEG–cholesterol + 2 eq. βCD or linear bifunctional PEG–cholesterol + 2 eq. βCD both formed almost fully elastic and strong hydrogels with G′ values of 360 ± 13 kPa and 180 ± 20 kPa and tan δ values of 0.15 ± 0.01 and 0.21 ± 0.01, respectively. The approximately two-fold difference in G′ between these gels can be explained by the lower cross-link density in gels containing the linear PEG–cholesterol. To explain, it should be noted that the concentration of cholesterol (0.05 mmol g⁻¹ gel) and βCD (0.1 mmol g⁻¹ gel) was equal in both hydrogels and therefore, the difference in G′ and thus cross-link density between the 8-arm PEG₈20K–chol_5.6 and linear PEG6K–chol_1.8 containing gels are due to the architecture of the cholesterol-derivatized PEG's. The 8 arms of PEG₈20K–chol are covalently connected to a core unit, and consequently networks based on this polymer have a higher cross-link density, which leads to networks with higher G′ values. Fig. 5 also displays the G′ and G″ of mixtures in which the PEG–chol component was replaced by non-derivatized linear or 8-arm PEG (PEG₈–OH) and of an 11% (w/w) suspension of βCD. As represented by their very low G′ and G″, these mixtures behaved as Newtonian liquids. This demonstrates that the use of cholesterol-derivatized PEG with ≥2 functionalities per PEG molecule is crucial for network formation. A combination of inclusion complexes between cholesterol and βCD insoluble nano-aggregates (see also the above X-ray diffraction analysis) and hydrophobic cholesterol–cholesterol interactions is likely responsible for the network formation.

Fig. 5 Storage modulus (G′) and loss modulus (G″) measured at 37 °C (measurement frequency = 1 Hz) of mixtures of PEG₈20K–chol_5.6 + 2 eq. βCD, PEG6K–chol_1.8 + 2 eq. βCD, mPEG5K–chol_0.9 + 2 eq. βCD, non-functionalized PEG₈–OH or PEG6K–OH + 2 eq. βCD and an aqueous 11% (w/w) βCD suspension. For all mixtures, the concentration of derivatized (or non-derivatized) PEG was 20% (w/w). Data are shown as mean ± standard deviation, n = 3.

Fig. 6 shows the gel mechanical properties of an aqueous mixture of PEG₈20K–chol_5.6 with 1 eq. of a 6-monodeoxy-6-monoamino-functionalized βCD derivative (βCD–NH₂), in comparison with gels containing no or unmodified βCD. The figure demonstrates again the strengthening of a PEG₈20K–chol_5.6 containing gel in the presence of βCD. This figure also shows that the presence of βCD–NH₂ led to a collapse of the gel structure (tan δ > 1). The monoamino-derivatized βCD is able to form host–guest inclusion complexes,⁶⁸ however, its aqueous solubility is 40 times higher than that of βCD.⁶⁹ Furthermore, it likely has a lower tendency to aggregate, because the aminomethyl group is positively charged in the aqueous solutions (pK_a = 8.72),⁶⁹ which results in repulsion between the βCD–NH₂ molecules. The dissolution of the gel can therefore be explained by complexation of the PEG-bound cholesterol groups in the cavities of soluble βCD–NH₂, which does not result in network formation.

Fig. 6 Rheological characteristics (4 °C) of mixtures containing 10% (w/w) PEG₈20K–chol_5.6, 10% (w/w) PEG₈20K–chol_5.6 + 1 eq. βCD or 10% (w/w) PEG₈20K–chol_5.6 + 1 eq. βCD–NH₂ (measurement frequency = 1 Hz). Left axis: G′ and G″; right axis: tan δ. Data are shown as mean ± standard deviation, n = 3.

To get further insight into the gelation mechanism, an excess (3 eq. relative to the present cholesterol groups) of a competitive inclusion complex forming compound, 1-adamantanecarboxylic acid (ACA), was added to the 20% PEG₈20K–chol_5.6/2 eq. βCD gel system. The binding affinity of ACA towards βCD is 2× higher than that of cholesterol (K_a (in water): 3.2 × 10⁴ M⁻¹vs. 1.6 × 10⁴ M⁻¹).^60,70,71Fig. 7 shows that addition of the competitor substantially weakened the PEG₈20K–chol_5.6/βCD gel, as reflected by a 2-fold decrease in G′ and a higher G″ (resulting in a 3 times higher tan δ). The addition of ACA did not completely break the gel structure, but led to gel properties that approached those of a hydrogel composed of PEG₈20K–chol_5.6. Likely, when the cholesterol moieties are displaced from the βCD cavities by ACA, the hydrophobic interactions between these groups return. The gel strength of the PEG₈20K–chol_5.6/βCD/ACA mixture was slightly higher than that of the PEG₈20K–chol_5.6 gel (Student's t-test, P < 0.05), which is probably due to a remaining fraction of βCD-complexed cholesterol.

Fig. 7 Storage modulus (G′) and loss modulus (G″) at 4 °C of hydrogels containing 20% (w/w) PEG₈20K–chol_5.6, 20% (w/w) PEG₈20K–chol_5.6 + 2 eq. βCD or 20% (w/w) PEG₈20K–chol_5.6 + 2 eq. βCD + 3 eq. ACA (measurement frequency = 1 Hz). Left axis: G′; right axis: G″. Data are shown as mean ± standard deviation, n = 3.

In summary, the results in this section give evidence that the network formation of the hydrogel systems based on cholesterol-derivatized PEG combined with βCD is due to small crystalline βCD clusters (proven by X-ray diffraction measurements) combined with cholesterol–βCD inclusion complexes.

Mechanical gel properties as function of solid content, temperature, βCD concentration and the 8-arm PEG's molecular weight

Fig. 8 shows G′ of PEG₈20K–chol_5.6–βCD hydrogels (tan δ < 0.1) with increasing βCD contents. It demonstrates that with increasing molar ratios of βCD : cholesterol from 0–2, G′ significantly increased. Furthermore, an increase of βCD content led to higher T_gel as well (see ESI†, Table S3). This indicates that an increase of βCD concentration results in a tighter network, which can be ascribed to either a higher number or an increased size of the insoluble crystalline βCD domains that act as cross-links. An increased number of clusters results in a higher cross-link density, while larger clusters can accommodate more PEG-bound cholesterol groups and contribute to a denser network as well. At βCD contents higher than 2 eq. relative to cholesterol (12.4% (w/w)), G′ started to level off at 435 ± 30 kPa. This indicates that at 2 eq. βCD, all PEG-bound cholesterol moieties are complexed with βCD and consequently are involved in network formation. Therefore, a higher concentration of βCD is not expected to further influence the network structure.

Fig. 8 Storage modulus (G′) at 20 °C of hydrogels containing 22.5% (w/w) PEG₈20K–chol_5.6 with increasing amounts of βCD (measurement frequency = 1 Hz). For all gels, tan δ < 0.1. Data are shown as mean ± standard deviation, n = 3.

Fig. 9 shows the rheological properties at 4 and 37 °C of hydrogels composed of 8-arm or linear PEG–chol and 1 eq. βCD at 4 and 37 °C with increasing solid content of the gels. At solid contents <12.8% (w/w) (10% (w/w) PEG–chol; 2.8% (w/w) βCD), all mixtures showed liquid-like behavior with G′ < 0.3 kPa and tan δ > 1 (not shown in figure). Here, the concentration of interacting cholesterol and βCD groups is likely too low to obtain a strong physical network. However, at a solid content of 12.8% (w/w) or higher, G′ started to dominate G″. Furthermore, with increasing PEG–chol/βCD content, G′ (and also G″ to a lower extent) gradually increased. In agreement with these observations, also the observed T_gel's increased with higher solid contents, which reached values between 38 and more than 100 °C (T_gel's > 80 °C by extrapolation; see ESI†, Table S3). These observations are in line with expectations, as higher concentrations of PEG–cholesterol and βCD result in more interacting cholesterol moieties and βCD clusters and subsequently a higher cross-link density. At solid contents ≥28.7% (w/w) (22.5% (w/w) PEG–chol; 6.2% (w/w) βCD), both G′ and G″ leveled off. This can be ascribed to the solubility of both linear and 8-arm PEG–cholesterol, which approached its limit at concentrations ≥22.5% (w/w). Because at higher concentrations PEG–cholesterol is not fully dissolved, only the dissolved molecules contribute to network formation.

Fig. 9 Storage modulus (G′) and loss modulus (G″) at 4 and 37 °C of PEG₈20K–chol_5.6 + 1 eq. βCD (A) or PEG6K–chol_1.8 + 1 eq. βCD (B) hydrogels as a function of the percentage solid content (measurement frequency = 1 Hz). Data are shown as mean ± standard deviation, n = 3.

Fig. 10 shows the G′ of PEG₈–chol/βCD gels as a function of the 8-arm PEG's molecular weight. It shows that the highest gel strength was reached for mixtures based on 20 kDa star shaped PEG. At equal concentrations, the use of the higher molecular weight PEG₈–chol (PEG₈40K–chol_5.6) implies that the concentration of cholesterol moieties decreases, which in turn will lead to a lower number of cross-links. Based on the higher concentration of cholesterol groups present in the PEG₈10K–chol_5.6–βCD gels, stronger networks were expected. However, the use of the lower molecular weight PEG₈10K–chol_5.6 led to a 7-fold decrease of G′. As suggested previously,³⁹ this can be ascribed to the poor solubility of PEG₈10K–chol_5.6, and as a consequence not all cholesterol units are available to form inclusion complexes with βCD, which leads to weaker networks.

Fig. 10 Storage modulus (G′) at 37 °C of hydrogels containing 8.3% (w/w) βCD and 15% (w/w) PEG₈–chol_5.6±0.1 with different molecular weights (measurement frequency = 1 Hz). Data are shown as mean ± standard deviation, n = 3.

4. Conclusions

In this paper, we investigated a novel hydrogel system, containing cholesterol-derivatized 8-arm PEG or linear PEG supplemented with β-cyclodextrin molecules (βCD). The tendency of βCD molecules to form crystalline nano-aggregates combined with their ability to form host–guest inclusion complexes with PEG-bound cholesterol moieties resulted in strong, almost fully elastic hydrogels. The strength of these gels was several orders higher as compared to gels based on PEG–cholesterol, and a previously reported gel system in which not only the cholesterol but also the βCD units were connected to 8-arm PEG.^39,40 It was demonstrated that the gel mechanical properties could be tailored by the concentration of βCD, the gel's total solid content, the PEG molecular weight, and the architecture of the PEG–cholesterol. Although the physical nature of the networks resulted in a temperature-dependence of the gel strength, their viscoelastic properties remained at elevated temperatures. Besides their high strength and versatility, the gels are composed of biocompatible and well-available building blocks. This hydrogel system may therefore be very suitable for biomedical and pharmaceutical applications, such as tissue engineering scaffolds and drug delivery matrices.

Acknowledgements

This research was financially supported by a grant of the Ministry of Economic Affairs, The Netherlands (SenterNovem IS042016).

References

1. W. E. Hennink and C. F. van Nostrum, Adv. Drug Delivery Rev., 2002, 54, 13–36.
2. N. A. Peppas, J. Z. Hilt, A. Khademhosseini and R. Langer, Adv. Mater., 2006, 18, 1345–1360.
3. N. E. Fedorovich, J. Alblas, J. R. de Wijn, W. E. Hennink, A. J. Verbout and W. J. A. Dhert, Tissue Eng., 2007, 13, 1905–1925.
4. F. Brandl, F. Sommer and A. Goepferich, Biomaterials, 2007, 28, 134–146.
5. N. A. Peppas, P. Bures, W. Leobandung and H. Ichikawa, Eur. J. Pharm. Biopharm., 2000, 50, 27–46.
6. J. A. Cadée, M. J. A. van Luyn, L. A. Brouwer, J. A. Plantinga, P. B. van Wachem, C. J. de Groot, W. den Otter and W. E. Hennink, J. Biomed. Mater. Res., 2000, 50, 397–404.
7. H. Park and K. Park, Pharm. Res., 1996, 13, 1770–1776.
8. S. R. Van Tomme, G. Storm and W. E. Hennink, Int. J. Pharm., 2008, 355, 1–18.
9. N. M. Sangeetha and U. Maitra, Chem. Soc. Rev., 2005, 34, 821–836.
10. L. A. Estroff and A. D. Hamilton, Chem. Rev., 2004, 104, 1201–1217.
11. P. Terech and R. G. Weiss, Chem. Rev., 1997, 97, 3133–3159.
12. M. George and R. G. Weiss, Acc. Chem. Res., 2006, 39, 489–497.
13. W. N. E. van Dijk-Wolthuis, J. A. M. Hoogeboom, M. J. van Steenbergen, S. K. Y. Tsang and W. E. Hennink, Macromolecules, 1997, 30, 4639–4645.
14. W. N. E. van Dijk-Wolthuis, J. J. Kettenes-van den Bosch, A. van der Kerk-van Hoof and W. E. Hennink, Macromolecules, 1997, 30, 3411–3413.
15. E. Oral and N. A. Peppas, J. Biomed. Mater. Res., Part A, 2004, 68, 439–447.
16. J. B. Leach and C. E. Schmidt, Biomaterials, 2005, 26, 125–135.
17. L. Ferreira, M. H. Gil, A. M. S. Cabrita and J. S. Dordick, Biomaterials, 2005, 26, 4707–4716.
18. V. Crescenzi, L. Cornelio, C. Di Meo, S. Nardecchia and R. Lamanna, Biomacromolecules, 2007, 8, 1844–1850.
19. M. van Dijk, D. T. S. Rijkers, R. M. J. Liskamp, C. J. van Nostrum and W. E. Hennink, Bioconjugate Chem., 2009 DOI:10.1021/bc900087a.
20. T. Vermonden, N. A. M. Besseling, M. J. van Steenbergen and W. E. Hennink, Langmuir, 2006, 22, 10180–10184.
21. B. Jeong, Y. H. Bae, D. S. Lee and S. W. Kim, Nature, 1997, 388, 860–862.
22. S. R. Van Tomme, M. J. van Steenbergen, S. C. De Smedt, C. F. van Nostrum and W. E. Hennink, Biomaterials, 2005, 26, 2129–2135.
23. K. Y. Lee, J. A. Rowley, P. Eiselt, E. M. Moy, K. H. Bouhadir and D. J. Mooney, Macromolecules, 2000, 33, 4291–4294.
24. B. O. Haglund, R. Josi and K. J. Himmelstein, J. Controlled Release, 1996, 41, 229–235.
25. Y. J. Wang, L. M. Tang and H. Yu, J. Colloid Interface Sci., 2008, 319, 357–364.
26. K. L. Kiick, Soft Matter, 2008, 4, 29–37.
27. T. Miyata, N. Asami and T. Uragami, Nature, 1999, 399, 766–769.
28. S. Nagahara and T. Matsuda, Polym. Gels Networks, 1996, 4, 111–127.
29. R. J. H. Stenekes, H. Talsma and W. E. Hennink, Biomaterials, 2001, 22, 1891–1898.
30. S. J. de Jong, B. van Eerdenbrugh, C. F. van Nostrum, J. J. Kettenes-van den Bosch and W. E. Hennink, J. Controlled Release, 2001, 71, 261–275.
31. S. J. de Jong, S. C. De Smedt, M. W. C. Wahls, J. Demeester, J. J. Kettenes-van den Bosch and W. E. Hennink, Macromolecules, 2000, 33, 3680–3686.
32. C. Hiemstra, Z. Y. Zhong, P. Dijkstra and J. Feijen, Macromol. Symp., 2005, 224, 119–131.
33. H. J. Chung, Y. H. Lee and T. G. Park, J. Controlled Release, 2008, 127, 22–30.
34. O. Kretschmann, S. W. Choi, M. Miyauchi, I. Tomatsu, A. Harada and H. Ritter, Angew. Chem., Int. Ed., 2006, 45, 4361–4365.
35. A. Charlot, R. Auzély-Velty and M. Rinaudo, J. Phys. Chem. B, 2003, 107, 8248–8254.
36. A. Charlot and R. Auzély-Velty, Macromolecules, 2007, 40, 1147–1158.
37. A. Charlot and R. Auzély-Velty, Macromolecules, 2007, 40, 9555–9563.
38. A. Hashidzume, I. Tomatsu and A. Harada, Polymer, 2006, 47, 6011–6027.
39. F. van de Manakker, M. van der Pot, T. Vermonden, C. F. van Nostrum and W. E. Hennink, Macromolecules, 2008, 41, 1766–1773.
40. F. van de Manakker, T. Vermonden, N. el Morabit, C. F. van Nostrum and W. E. Hennink, Langmuir, 2008, 24, 12559–12567.
41. F. van de Manakker, K. Braeckmans, N. el Morabit, S. C. De Smedt, C. F. van Nostrum and W. E. Hennink, Adv. Funct. Mater., 2009, 19, 2992–3001.
42. T. Loftsson and D. Duchene, Int. J. Pharm., 2007, 329, 1–11.
43. D. Harries, D. C. Rau and V. A. Parsegian, J. Am. Chem. Soc., 2005, 127, 2184–2190.
44. M. V. Rekharsky and Y. Inoue, Chem. Rev., 1998, 98, 1875–1917.
45. S. Li and W. C. Purdy, Chem. Rev., 1992, 92, 1457–1470.
46. A. W. Coleman, I. Nicolis, N. Keller and J. P. Dalbiez, J. Inclusion Phenom. Mol. Recognit. Chem., 1992, 13, 139–143.
47. G. Gonzalez-Gaitano, P. Rodriguez, J. R. Isasi, M. Fuentes, G. Tardajos and M. Sanchez, J. Inclusion Phenom. Macrocyclic Chem., 2002, 44, 101–105.
48. M. Bonini, S. Rossi, G. Karlsson, M. Almgren, P. Lo Nostro and P. Baglioni, Langmuir, 2006, 22, 1478–1484.
49. S. Polarz, B. Smarsly, L. Bronstein and M. Antonietti, Angew. Chem., Int. Ed., 2001, 40, 4417–4421.
50. T. Loftsson, M. Masson and M. E. Brewster, J. Pharm. Sci., 2004, 93, 1091–1099.
51. K. M. Huh, Y. W. Cho, H. Chung, I. C. Kwon, S. Y. Jeong, T. Ooya, W. K. Lee, S. Sasaki and N. Yui, Macromol. Biosci., 2004, 4, 92–99.
52. K. M. Huh, T. Ooya, W. K. Lee, S. Sasaki, I. C. Kwon, S. Y. Jeong and N. Yui, Macromolecules, 2001, 34, 8657–8662.
53. J. Li, X. P. Ni and K. W. Leong, J. Biomed. Mater. Res., Part A, 2003, 65, 196–202.
54. E. Sabadini and T. Cosgrove, Langmuir, 2003, 19, 9680–9683.
55. H. Q. Yu, Z. G. Feng, A. Y. Zhang, L. G. Sun and L. J. Qian, Soft Matter, 2006, 2, 343–349.
56. T. Ooya and N. Yui, Crit. Rev. Ther. Drug Carrier Syst., 1999, 16, 289–330.
57. H. S. Choi, K. Kontani, K. M. Huh, S. Sasaki, T. Ooya, W. K. Lee and N. Yui, Macromol. Biosci., 2002, 2, 298–303.
58. J. Loccufier, M. Van Bos and E. Schacht, Polym. Bull. (Berlin), 1991, 27, 201–204.
59. H. H. Winter, Polym. Eng. Sci., 1987, 27, 1698–1702.
60. K. Akiyoshi, Y. Sasaki, K. Kuroda and J. Sunamoto, Chem. Lett., 1998, 27, 93–94.
61. K. Akiyoshi, A. Ueminami, S. Kurumada and Y. Nomura, Macromolecules, 2000, 33, 6752–6756.
62. P. W. Atkins and J. de Paula, in Atkins' Physical Chemistry, Oxford University Press, Oxford, 7th edn, 2002.
63. Y. Takahashi and H. Tadokoro, Macromolecules, 1973, 6, 672–675.
64. I. W. Hamley and M. J. Krysmann, Langmuir, 2008, 24, 8210–8214.
65. X. J. Loh, K. B. C. Sng and J. Li, Biomaterials, 2008, 29, 3185–3194.
66. T. Steiner and G. Koellner, J. Am. Chem. Soc., 1994, 116, 5122–5128.
67. H. P. Fink, D. Fanter and B. Philipp, Acta Polym., 1985, 36, 1–8.
68. M. V. Rekharsky and Y. Inoue, J. Am. Chem. Soc., 2002, 124, 12361–12371.
69. S. E. Brown, J. H. Coates, D. R. Coghlan, C. J. Easton, S. J. Vaneyk, W. Janowski, A. Lepore, S. F. Lincoln, Y. Luo, B. L. May, D. S. Schiesser, P. Wang and M. L. Williams, Aust. J. Chem., 1993, 46, 953–958.
70. K. Shiotani, K. Uehata, T. Irie, K. Uekama, D. O. Thompson and V. J. Stella, Pharm. Res., 1995, 12, 78–84.
71. M. Weickenmeier and G. Wenz, Macromol. Rapid Commun., 1996, 17, 731–736.

---

Footnote

† Electronic supplementary information (ESI) available: Studied hydrogel compositions with corresponding gel transition temperatures. See DOI: 10.1039/b916378f

---