Received August 22, 2007

Revised October 18, 2007

Introduction

There is a high need for materials and technologies for the development of suitable pharmaceutical formulations of the modern generation of biotherapeutics such as peptides, proteins, and plasmid DNA.¹ In this way, the active compounds are protected against premature degradation and released in a controlled manner. Both natural and synthetic polymers, as well as lipids have been used to create delivery vehicles in all shapes and sizes, e.g., hydrogels, microspheres, liposomes, micelles, etc.² Hydrogels are hydrophilic polymeric networks capable of imbibing substantial amounts of water. Dissolution of the polymer chains is prevented by chemical or physical cross-linking.³ Their high water content makes them attractive candidates for protein delivery and tissue engineering applications.^4–6

Numerous hydrogels have been developed, mainly based on chemical cross-linking of, e.g., modified dextrans^7–10 and poly(ethylene glycol)s (PEG).¹¹ Despite the favorable characteristics of these matrices regarding biodegradability and biocompatibility, the cross-linking conditions might adversely affect the structure and activity of the entrapped bioactive molecules. Therefore, physically cross-linked hydrogels have gained increasing interest in recent years as an alternative for the chemically cross-linked gels as a delivery system for, among others, pharmaceutically active proteins. These hydrogels are formed through physical entanglements of the polymer chains, avoiding the use of potentially harmful cross-linking agents. Nowadays, most promising are those hydrogels that are formed at the site of injection, i.e., in situ. Various strategies have been developed to create in situ forming delivery systems. Chitosan,^{12, 13} hyaluronic acid,^{14, 15} and PEG-based polymers¹⁶ have been successfully used as photopolymerizable systems in tissue engineering and drug delivery. Alginates spontaneously gellify in the presence of divalent cations like Ca²⁺ and have, among others, been used for ophthalmic delivery¹⁷ and as synthetic extracellular matrices.¹⁸ In situ gelation can also occur in response to a certain trigger, e.g., temperature,^19–21 pH,^{22, 23} or after some time after mixing the components. The latter can be the result of, e.g., hydrophobic interactions,²⁴ ionic interactions,²⁵ or stereocomplexation.^{26, 27} It has been reported that stereocomplexation between oligolactate grafts of opposite chirality on dextran resulted in hydrogels after mixing aqueous solutions of dex-L-lactate and dex-D-lactate. Extensive in vitro and in vivo studies have shown that this system is promising as protein delivery system.^{26, 28–30} Although the stereocomplexed dex-lactate hydrogels have successfully been used as injectable matrices,²⁹ the high viscosity of the two components used to prepare the gels limits their practical applicability.

The objective of this work was to develop a novel in situ forming hydrogel, potentially suitable for the controlled delivery of pharmaceutical proteins, in which the injectability of dextran microspheres is combined with physical cross-linking between the microspheres. For this purpose, we have made use of the favorable properties of lactic acid oligomers to connect the individual microspheres, creating a macroscopic gel. Lactic acid is a chiral molecule, which leads to the formation of stereocomplex crystallites between the L- and D-lactic acid oligomers. Therefore, oligolactate chains of opposite chirality were covalently grafted onto microspheres that were formed beforehand by polymerization of (hydroxy ethyl methacrylate)-substituted dextran. The focus of this manuscript is on the preparation and characterization of the “dex-HEMA-lactate” microspheres as well as the self-assembling macroscopic hydrogels.

Materials and Methods

Materials. Dextran T40 (from Leuconostoc spp), N,N,N′,N′-tetramethyl ethylenediamine (TEMED), 2-hydroxyethyl methacrylate (HEMA, 95%), and lysozyme (from hen egg white, M_w 14000 g/mol) were provided by Fluka (Buchs, Switzerland). Poly(ethylene glycol) (PEG) 10 kDa and potassium peroxodisulfate (KPS, >99%) were purchased from Merck (Darmstadt, Germany). L-Lactide ((3S-cis)-3,6-dimethyl-1,4-dioxane-2,5-dione, >99.5%) and D-lactide ((3R-cis)-3,6-dimethyl-1,4-dioxane-2,5-dione, >99.5%), poly(L-lactic acid) (PLLA, intrinsic viscosity 0.59 dL/g), and poly(D-lactic acid) (PDLA, intrinsic viscosity 2.39 dL/g) were obtained from Purac Biochem BV (Gorinchem, The Netherlands) and used without further purification. N-2-hydroxyethyl piperazine-N′-2-ethanesulfonic acid (Hepes), 1,1′-carbonyl diimidazole (CDI, 98%), 4-N,N-dimethyl aminopyridine (DMAP, 99%), and poly(vinyl alcohol) (PVA, 88% hydrolyzed) were purchased from Acros Chimica (Geel, Belgium). Stannous octoate (tin(II) bis(2-ethyl hexanoate) SnOct₂, 95%) and S-ethyl lactate (98%) were provided by Sigma-Aldrich (Zwijndrecht, The Netherlands). The bicinchoninic acid (BCA) protein assay kit was purchased from Interchim (Montluçon, France).

Preparation of Dex-HEMA Microspheres. Hydroxyethyl methacrylate-derivatized dextran (dex-HEMA) with a substitution degree of 10 (DS, i.e., the number of HEMA groups per 100 glucopyranose units) was synthesized and characterized according to van Dijk-Wolthuis et al.³¹ Microspheres were prepared through radical polymerization of dex-HEMA in an all-aqueous environment, as previously described.^{32, 33}

Synthesis of Polydisperse Enantiomeric Oligolactate. L- and D-Lactic acid oligomers were synthesized by a ring-opening polymerization reaction of lactide with ethyl lactate as initiator, according to de Jong et al.³⁴ The average degree of polymerization (DP_av) of the formed oligolactates was controlled by the ethyl lactate/lactide ratio. The DP_av used in this study were 5 and 13.

Grafting Lactic Acid Oligomers to Dex-HEMA Microspheres. To couple the lactic acid oligomers to dex-HEMA microspheres, the hydroxyl groups of the oligomers were activated using N,N′-carbonyldiimidazole (CDI) as described by de Jong et al.²⁶ Next, the activated oligomers (ethyl lactate-CI) were grafted to dex-HEMA microspheres, essentially as described for grafting of lactic acid oligomers to soluble dextran with some minor modifications. In short, lyophilized dex-HEMA microspheres (500 mg) were dispersed in dry dimethyl sulfoxide (DMSO) (10 mL). Next, DMAP (100 mg) was dissolved in the mixture and subsequently, ethyl lactate-CI (e.g., 230 mg to obtain a theoretical maximum DS of 8 (defined as the number of oligolactate grafts per 100 glucopyranose units and further referred to as “DS_theor”)) was added and the vials were placed on a roller bench for 10 days. Finally, the microspheres were washed with DMSO (3 times) and dichloromethane (DCM) (3 times) and subsequently dried overnight by evaporation of the DCM. Microspheres with different theoretical oligolactate substitution degrees were prepared (DS_theor 2, 4, and 8). As a control, the dex-HEMA microspheres were treated with non-CDI-activated oligolactate under the same conditions.

Preparation of PLLA and PDLA Microspheres. PLLA en PDLA microspheres were prepared using an oil-in-water (o/w) solvent evaporation technique.³⁵ Briefly, 3 g PLLA or PDLA was dissolved in 13 mL DCM and subsequently poured into 180 mL PVA solution (1% in water (w/v)). The o/w emulsion was stirred for 4 h (mechanical stirrer, 1200 rpm) during which the DCM was allowed to evaporate. After three washing and centrifugation steps (10 min, 3000 rpm), the microspheres were lyophilized.

Fourier Transform Infrared Spectroscopy (FT-IR). Successful modification of the dex-HEMA microspheres with oligolactates was studied with FT-IR (Biorad FTS 6000 spectrometer (Varian Inc., Palo Alto, CA)). Microsphere dispersions were prepared (5% w/w in Hepes buffer, 100 mM, pH 7.0) of which 10 µL was brought into a liquid sample cell with a CaF₂ window and a path length of 7 µm. For a single spectrum, 1024 scans were accumulated, and subsequently, the spectra were corrected for water vapor. Before and after calculation of the second derivative, the spectra were smoothed using a second-order, seven-points Savitzky−Golay smoothing.³⁶

Raman Spectroscopy. To confirm the oligolactate grafting of the microspheres, Raman spectra of dry dex-HEMA microspheres, dex-HEMA-lactate microspheres, and oligolactate were recorded at room temperature with a Kaiser RXN dispersive Raman spectrometer. A 532 nm (60 mW) laser was used for excitation and a Peltier element-cooled Andor CCD camera was applied for detection.

X-Ray Photoelectron Spectroscopy (XPS). XPS was performed to analyze the surface composition of the oligolactate-grafted microspheres. The XPS data were obtained with a Vacuum Generators XPS system using a CLAM-2 hemispherical analyzer for electron detection. Nonmonochromatic Al (Kα) X-ray radiation was used for generating the photoelectrons at an anode current of 20 mA at 10 KeV. The pass energy of the analyzer was set at 50 eV. The survey scan was taken with a pass energy of 100 eV. Dry microspheres were attached to double-sided tape before mounting in the analysis chamber. Both C(1s) and O(1s) spectra were recorded.

Lactic Acid Determination. To investigate the extent of grafting of oligolactates to the dex-HEMA microspheres, the lactic acid content of degraded microspheres was measured. In detail, dex-HEMA-lactate microspheres (10 mg) were hydrolyzed in alkaline conditions at room temperature overnight (1 mL 0.1 N NaOH), and subsequently, the solution was neutralized (100 µL 1 N HCl). Analysis of the amount of lactic acid in the samples was carried out on a Waters HPLC system (Waters Associates Inc., Milford, MA) consisting of a quaternary gradient pump (model 600E), an autosampler (model 717), and a variable wavelength absorbance detector (model 486). An Alltech Prevail Organic acid column (5 µm, 250 mm × 4.6 mm) was used with KH₂PO₃ buffer (25 mM, pH 2.5) and acetonitrile/water (95/5 w/w) as eluents A and B, respectively, at a flow rate of 1 mL/min. A gradient was run from 0 to 60% B in 10 min, followed by an elution with 60% B for 10 min, with a total run time of 30 min. The injection volume was 50 µL, and the detection wavelength was 210 nm. The calibration curve was linear between 0.01 and 4 µmol lactic acid.

Rheology of Microsphere-Based Hydrogels. Rheological analysis of macroscopic hydrogels, comprising dex-HEMA-lactate microspheres, was performed using a controlled stress rheometer (AR1000-N, TA Instruments, Etten-Leur, The Netherlands), equipped with an acrylic flat plate geometry (20 mm diameter) and a gap of 500 µm. A solvent trap was used to prevent evaporation of the solvent. Gels (200 mg) were prepared by dispersing the dry microspheres in buffer (Hepes, 100 mM, pH 7.0). When both oligo-L-lactate and oligo-D-lactate substituted microspheres were used, the particles were thoroughly mixed in the dry state, using a vortex, before hydration with buffer. The microspheres were allowed to hydrate overnight at 4 °C. The viscoelastic properties of the samples were determined by measuring the G′ (shear storage modulus) and G′′ (loss modulus) at 20 °C with a constant strain of 1% and constant frequency of 1 Hz. The extent of recovery of the material after deformation was evaluated during creep experiments. A shear stress (varying from 1 to 100 Pa) was applied while the strain was monitored. After 1 min, the stress was removed and the recovery of the sample was monitored by measuring the strain during 2 min. To mimic the behavior of the system during injection, the oscillatory stress was gradually increased until the gel started to flow. Subsequently, a constant stress was applied during a time sweep experiment monitoring the recovery of the network after flow. The influence of the solid content (10–20%), the chain length of the oligolactate grafts (DP_av 5 and 13), and the degree of oligolactate substitution (DS_theor 2–8) on the network properties was investigated.

For comparison, PLLA and PDLA microspheres were mixed, hydrated (Hepes buffer 100 mM pH 7.0) (20–60% solid content (w/w)) and investigated for possible network formation.

In Vitro Lysozyme Release from Microsphere-Based Hydrogels. Protein-loaded hydrogels (1 mg protein/100 mg gel) were prepared by hydration of a mixture of dry dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres (DS_theor 8, DP_av 13) with lysozyme solution (in Hepes buffer 100 mM pH 7.0). The in vitro protein release was studied as described previously.³⁷ In short, hydrogels (500 mg, 15% solid (w/w)) were transferred into a release device consisting of a gel and a release compartment, with a diameter of respectively 8.5 mm and 15 mm and length of 8.8 mm and 30 mm. As a result, cylindrical gels of 8.5 mm × 8.8 mm (diameter × length) were obtained. Release buffer (3 mL, 100 mM Hepes pH 7.0, 0.02% NaN₃, 150 mM NaCl) was added to each gel, and the device was incubated on a shaking plate at 37 °C. Samples of 0.5 mL were taken at regular time intervals and replaced by an equal volume of fresh buffer. The protein content of the release samples was determined using the BCA protein assay.³⁸ Additionally, the enzymatic activity of the released lysozyme was determined. The assay is based on the hydrolysis of the outer cell membrane of Micrococcus lysodeikticus, resulting in solubilization of the affected bacteria and consequent decrease of light scattering.³⁹ The percent remaining enzyme activity was obtained by comparing the activity to that of a reference lysozyme solution.

Results and Discussion

Preparation and Oligolactate Substitution of Dex-HEMA Microspheres. Internally cross-linked dex-HEMA microgels with an equilibrium water content of 70% and a mean volume diameter of 9 µm were prepared by making use of an aqueous PEG/dextran two-phase system.^{32, 33} Next, the microsphere particles were dried, swollen with DMSO, and then grafted with oligolactate chains (Figure 1).

Figure 1. Schematic representation of oligolactate grafting on dex-HEMA microspheres, resulting in dex-HEMA-lactate microspheres. Click to Enlarge

In this way, structures are created consisting of hydrophilic, dextran-based particles, substituted with hydrophobic oligolactate. Derivatization of the microspheres did not affect the size, but the wettability visibly decreased, as evidenced from their poor dispersability in water. These findings indicate that water penetration into the microspheres was retarded due to an increased hydrophobicity of the microsphere surface.

FT-IR and Raman Spectroscopy. The presence of the oligolactate chains in modified dex-HEMA microspheres was confirmed by FT-IR and Raman spectroscopy (Figure 2, 3). The second derivative FT-IR spectrum of dex-HEMA microspheres showed a peak around 1750 cm⁻¹, attributed to vibrations of carbonyl groups in the HEMA chains.⁴⁰ The grafted microspheres (dex-HEMA-D-lactate) showed an increase in carbonyl peak height due to an increased intensity of the carbonyl vibration, dependent on the degree of lactate substitution (DS_theor 8 > DS_theor 4 > DS_theor 2) (DP_av 13) (Figure 2). The control samples, consisting of microspheres that were treated with nonactivated oligolactate, showed a slight increase of the carbonyl peak, indicating that some aspecific binding to the microparticles had occurred.

Figure 2. Second derivative FT-IR spectra of dispersions consisting of dex-HEMA microspheres (▼) or dex-HEMA-D-lactate microspheres of different oligolactate substitution degrees (DStheor 2 (■), DStheor 4 (−), DStheor 8 (●); DPav 13) in buffer. Dex-HEMA microspheres treated with nonactivated oligolactate were measured as a control (*). Click to Enlarge

Figure 3. Raman spectra (region 2700–3100 cm−1) of unmodified dex-HEMA microspheres (A), oligolactate (●), and dex-HEMA-L-lactate microspheres (DStheor 8, DPav 13) (−) (B). Click to Enlarge

Figure 3A shows the region 2700–3100 cm⁻¹ of the Raman spectrum of unmodified dex-HEMA microspheres. The observed peaks correspond to CH and CH₂ stretching vibrations.⁴⁰ In Figure 3B the Raman spectra (region 2700–3100 cm⁻¹) of oligolactate and dex-HEMA-L-lactate microspheres (DS_theor 8) are depicted. The Raman spectrum of oligolactate showed 3 dominant peaks at 2885, 2950, and 3000 cm⁻¹, corresponding to CH and CH₃ stretching vibrations,⁴⁰ which were also detected in the spectrum of the oligolactate-substituted dex-HEMA microspheres. They indicate the presence of oligolactate chains in the microspheres, most likely as a result of the grafting reaction. Both FT-IR and Raman spectroscopy results point toward a successful grafting of oligolactates to the dex-HEMA microspheres.

Lactic Acid Determination of Degraded Microspheres. To quantify the extent of grafting, the lactic acid content of the microspheres was determined. Therefore the dex-HEMA-lactate microspheres were incubated for 16 h under alkaline conditions to yield dextran, poly(HEMA),⁴¹ and lactic acid.^{28, 42} Degradation of the dex-HEMA-lactate microspheres and subsequent lactic acid quantification revealed that 10–13% and 16–20% (for DP_av 13 and 5, respectively) of the amount of oligolactate added was grafted to the microspheres. The relatively low grafting efficiency might indicate that the OH groups present at the outer layer of the microspheres are preferentially modified, by which reaction of the OH groups buried inside the particles with oligolactate is prevented.

XPS of Oligolactate-Grafted Dex-HEMA Microspheres. XPS provides information on chemical bonding and elemental composition of the outer layer (a few nm) of a sample. Because the lactic acid quantification of the oligolactate-grafted microspheres suggested that mainly the outer part of the dex-HEMA microspheres is substituted with oligolactate, the microsphere surface was scanned during XPS measurements. Figure 4 shows representative C(1s) spectra of oligolactate, dex-HEMA microspheres, and modified dex-HEMA microspheres. In the spectrum of oligolactate, 3 peaks at 291, 293, and 295 eV are visible, originating from C−H, C−O, and CO bonds, respectively. The dex-HEMA microspheres (without and with oligolactate grafts) only showed one peak, at 293 eV, originating from C−O bonds.^{43, 44} C−O groups are the most abundant ones in dex-HEMA microspheres. Grafting of the particles with oligolactate led to an increase in intensity of other groups (CO and C−H), but due to the excessive amount of C−O groups, only one broad peak, ascribed to C−O, was detected. For all microspheres, the O(1s)−C(1s) was ~246.5 eV, indicating that the shifts in peak maxima of both O(1s) and C(1s)⁴³ were not due to specific properties of the samples, but due to the charging of the products.

Figure 4. XPS C(1s) spectra (region 285–305 eV) of oligolactate (---), dex-HEMA microspheres (▼), and dex-HEMA-L-lactate microspheres (DStheor 8, DPav 13) (−). Click to Enlarge

The full width at half-height (fwhh) of the peaks of the different microsphere samples was calculated. Grafting of the microspheres with oligolactate resulted in an increased fwhh from 2.00 eV (±0.05 eV) (dex-HEMA microspheres) to 2.43 eV (±0.05 eV) (dex-HEMA-L-lactate microspheres, average of DS_theor 2–8). This is presumably due to shouldering of the peaks toward a higher binding energy as a result of an increasing amount of CO groups present at the surface. Keeping the detection limit of peak shifts and peak separations in mind (±0.2 eV),⁴⁵ the differences between the substituted microspheres of various DS were too small to detect. These XPS results indicate that, presumably, preferential grafting of the outer surface layer has taken place, leading to microspheres with comparable surface composition.

Rheological Characterization of the Self-Assembling Networks. Rheological studies were done to evaluate the ability of the oligolactate-substituted microspheres to interact and form macroscopic hydrogels. Figure 5 illustrates the possible interactions that may occur upon hydration of the dex-HEMA-lactate microspheres. The individual microspheres grafted with L- or D-oligolactate could assemble due to hydrophobic interactions between the lactate-enriched domains on their surface. However, when microspheres, derivatized with oligolactate of opposite chirality, are mixed, stereocomplexes could be formed, resulting in relatively stronger interactions between the particles.

Figure 5. Schematic representation of the concept of the self-assembling hydrogel (grafts are not on scale). Click to Enlarge

Figure 6A shows a picture of a macroscopic hydrogel that is formed by the self-assembly of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres. For comparison, the possibility of PLLA and PDLA microspheres to self-assemble through stereocomplexation was also investigated. As evidenced by the image in Figure 6B, no gelation occurred. Rheological experiments on the latter system, did not show, even in high concentrations (solid contents up to 60% (w/w)), the formation of a network.

Figure 6. (A) Hydrogel composed of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres. (B) Dispersion containing PLLA and PDLA microspheres (in both cases: 15% solid (w/w) in Hepes buffer 100 mM pH 7.0). Click to Enlarge

Rheological analysis of dispersions of dex-HEMA microspheres (15% solid w/w) as well as microspheres treated with nonactivated oligo-L- or D-lactate behaved like a Newtonian liquid, showing that the particles did not interact with each other to form a network (data not shown). Although FT-IR analysis (Figure 2) and Raman spectroscopy (Figure 3) suggested that some oligolactate chains might be bound noncovalently to the microspheres, it appeared to be insufficient to lead to network formation with neighboring microspheres.

Figure 7 depicts the effect of the solid content of the gels (defined as the % dry microspheres w/w) on the hydrogel properties of mixtures of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres (DS_theor 8, DP_av 13). In these hydrogel systems, water forms a continuous phase in which the microparticles, the solid components, are dispersed. A water content higher than 90% prevented the microspheres from interacting, and no gel formation could be observed. A strong increase of the storage modulus from about 850 to 12 700 Pa was observed when the solid content was increased from 10 to 20%. Within this range, the tan(δ) (= G″/G′) was lower than 0.1, indicating that the loss modulus (G″), which is a measure for the viscosity of the material, is very low, demonstrating that almost fully elastic gels were formed. The discontinuous semisolid phase imparts the gel strength because a higher concentration of microspheres leads to a denser packing of the particles and more opportunities to interact.

Figure 7. Storage modulus (G′) at 20 °C of dispersions consisting of a mixture of equal amounts of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres (oligolactate DPav 13 DStheor 8) as a function of the solid content (w/w) (n = 3). Click to Enlarge

Frequency sweeps were performed to check whether the applied 1 Hz was in the linear range. It was found that G′ remained constant between 0.1 and 10 Hz, in other words, the storage modulus was frequency independent, again indicative of an almost fully elastic network.

Interestingly, dispersions (15% solid (w/w), corresponding to 50% (w/w) of free water) of the individual components dex-HEMA-L-lactate or dex-HEMA-D-lactate microspheres (DS_theor 2, DP_av 13) also showed gel-like behavior with tan(δ) < 0.1 and storage moduli (G′) of 800 and 600 Pa, respectively (Figure 8). The G′ increased to 1900 and 2400 Pa when the DS_theor was increased from 2 to 8 for dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres, respectively. Obviously, a higher number of oligolactates present on the microspheres led to more physical cross-links between the microspheres and thus stronger hydrogels. No significant differences in gel properties were observed between oligo-L- or oligo-D-lactate grafted microspheres for the different DS_theor. Apparently, hydrophobic interactions between the oligolactate chains on different microspheres are sufficient to interconnect the microspheres and create an almost fully elastic network (tan(δ) < 0.1). This behavior was different from that of noncross-linked, water soluble, oligo-L- or oligo-D-lactate substituted dextran with a high DP and DS, which behaved as a typical viscoelastic material due to association of longer oligolactates with oligomers of the same chirality.²⁶

Figure 8. Storage modulus (G′) of hydrogels (15% solid (w/w)) consisting of dex-HEMA-L-lactate microspheres (L), dex-HEMA-D-lactate microspheres (D) or a mixture of both (L + D) at 20 °C as a function of the DS (DStheor 2: gray bars; DStheor 4: black bars; DStheor 8: white bars) (DPav 13) (n = 3). Click to Enlarge

Importantly, for each DS_theor, mixing equal amounts of microspheres substituted with oligolactates of opposite chirality led to stronger hydrogels, corresponding with G’s of 1400 and 4400 Pa for DS_theor 2 and 8, respectively (DP_av 13), which is almost double as strong as those hydrogels comprising only oligo-L- or oligo-D-lactates (Figure 8). This is attributable to stereocomplex formation between L- and D-lactate (Figure 5). This has been described previously for self-assembling hydrogels based on oligolactate-grafted dextran, which formed an almost elastic network as well.²⁶

Microspheres were grafted (DS_theor 2 and 8) with oligolactate of either DP_av 5 or 13, and the influence of the oligolactate chain length on the rheological behavior was monitored (Figure 9). The microspheres grafted with the shortest oligolactate chains (DP_av 5) (15% solid (w/w)) were able to interact and form hydrogels, but the gel strength could not be altered by varying the DS_theor (2 vs 8). It was previously shown that enantiomeric monodisperse lactic acid oligomers of DP 7 are amorphous, whereas blends of the corresponding L- and D-forms show crystallinity.³⁴ When grafted to dextran, however, a DP lower than 11 did not result in hydrogel formation.²⁶ Taking this into account, it is clear that, when using oligolactate chains with a DP_av 5, only a fraction of the oligomers will be sufficiently long to be able to interact with lactic acid oligomers on other microspheres to yield stereocomplexes, so the possibilities to tailor the network properties diminish. For hydrogels (15% solid (w/w)) composed of dex-HEMA-L- and dex-HEMA-D-lactate microspheres with a DS_theor 8, an increase of DP_av from 5 to 13 led to a strong increase in G′ from 1000 to 4500 Pa (Figure 9), most likely due to the presence of stereocomplexes acting as physical cross-links between the microspheres.

Figure 9. Storage modulus (G′) of dispersions containing equal amounts of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres with varying oligolactate substitution degrees (DStheor 2: white bars; DStheor 8: black bars) as a function of the DPav of the oligolactate chains (15% solid (w/w), 20 °C) (n = 3). Click to Enlarge

Creep experiments were performed to investigate the extent of recovery of the hydrogels after deformation (Figure 10). Dispersions (15% solid) containing dex-HEMA microspheres showed Newtonian behavior with a % strain of 8000 during the retardation phase and no recovery after withdrawal of the shear stress (results not shown). In contrast, under the same experimental conditions, hydrogels (15% solid (w/w)) composed of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres (DS_theor 8, DP_av 13) reached a plateau deformation value of 1.2% and recovered almost completely when the stress was removed (Figure 10). Dispersions containing solely oligo-L- or oligo-D-lactate grafted microspheres displayed the same deformation profile (not shown). These results demonstrate the almost full elasticity of the self-assembling microsphere-based hydrogels.

Figure 10. Creep experiment on a hydrogel consisting of a mixture of equal amounts of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres (15% solid (w/w), applied stress 50 Pa, 20 °C). Click to Enlarge

The reversibility of gelation was studied by measuring the gel properties at a constant strain (1%, corresponding to a shear stress of 6 Pa) during 20 min and by subsequently increasing the oscillatory stress until the network was broken (tan(δ) >1). Immediately hereafter, the network properties were monitored during a second time sweep. When the stress on the gels (equal amounts of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres, 15% solid (w/w), DP_av 13, DS_theor 8) was increased, the gel was broken and flow started to occur, shown by the sudden increase of tan(δ) at approximately 350 Pa (Figure 11). Upon removal of the stress, the network rebuilt almost instantaneously, as indicated by the recovery of the initially observed G′ (~5000 Pa) and low tan(δ) (<0.1). These results indicate that, upon shear, the interactions between the microspheres can be temporarily and reversibly broken. This is an important feature for application as an injectable system because the gels need to exhibit a reversible yield point, i.e., they have to flow when a certain stress is applied, after which the network needs to be rebuilt. The yield point is dependent on the initial gel strength and can be tailored by varying the solid content, the number of lactate grafts, and their DP_av. Some representative examples are listed in Table 1.

Figure 11. Storage modulus (G′) (−) and tan(δ) (---) of a dispersion containing equal amounts of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres (DStheor 8, DPav 13) (15% solid (w/w), 20 °C) as a function of the oscillatory stress. Click to Enlarge

In Vitro Lysozyme Release. Release experiments were performed to evaluate the potential of these self-assembling microsphere-based hydrogels for controlled protein delivery. Protein loading was done by simply mixing the microspheres with a protein solution, entrapping the proteins in the pores between the microspheres. In this way, protein loading is accomplished at neutral pH and at room temperature, avoiding potential stress factors, such as organic solvents, cross-linking agents, etc., which might lead to protein denaturation. A continuous lysozyme release was obtained from hydrogels (15% solid (w/w)) composed of equal amounts of dex-HEMA-L-lactate and dex-HEMA-D-lactate (DS_theor 8, DP_av 13) (Figure 12). Fifty percent of the entrapped protein was released in 5 days, and after 30 days, an almost quantitative release was observed. Furthermore, the bioactivity of the released lysozyme was found to be the same as that of native lysozyme. It can therefore be concluded that neither degradation nor aggregation of the lysozyme during preparation of the gel and/or during release had occurred, emphasizing the protein-friendly technology.

Figure 12. Cumulative lysozyme release (%) from hydrogels (15% solid (w/w)) composed of equal amounts of dex-HEMA-L-lactate and dex-HEMA-D-lactate microspheres (DStheor 8, DPav 13). Mean ± SD of three independent measurements. Click to Enlarge

Conclusions

This paper reports on the possibility of creating macroscopic hydrogels based on self-assembling dextran microspheres. Hydrophilic dex-HEMA microspheres were grafted with enantiomeric oligolactate chains, creating surface hydrophobized microgels. Hydrophobic interactions between oligolactate chains on the microspheres led to the formation of macroscopic networks. An even stronger type of physical cross-links, stereocomplexes, was introduced by addition of oligolactate-substituted microspheres of opposite chirality. The network properties of these self-assembling hydrogels showed to be tailorable by modifying the grafting density of the microspheres, the degree of polymerization of the oligolactate grafts, and the solid content of the gels. Interestingly, for application of these systems as an injectable matrix, it was found that, upon exposure to shear forces, the interactions between the microspheres were temporarily broken, allowing the gel to be injected, after which the network was rebuilt. It can be foreseen that this novel microsphere-based system will possess an improved injectability compared to the dex-lactate hydrogels.⁴⁶ Protein release experiments showed a continuous lysozyme release during 30 days with full preservation of its enzymatic activity. The degradation kinetics of the system and the nature of the formed degradation products are the subject of a present investigation, but it can be expected that the same degradation products will be found as for dex-HEMA and dex-lactate hydrogels.^{7, 9, 41} The protein-friendly loading of the gel, the biocompatible nature of the material, and possibility of tailoring the gel properties makes this hydrogel attractive in a variety of pharmaceutical applications.

* Corresponding author. E-mail: W.E.Hennink@uu.nl. Telephone: +31 30 253 6964 . Fax: +31 30 251 7839.

† Department of Pharmaceutics, Utrecht Institute for Pharmaceutical Sciences, Utrecht University.

‡ Inorganic Chemistry and Catalysis group, Department of Chemistry, Faculty of Science, Utrecht University.