Mesoporous bioactive nanocarriers in electrospun biopolymer fibrous scaffolds designed for sequential drug delivery

Abstract

Here we communicate a novel design to deliver multiple drugs from scaffolds which have special therapeutic efficacy for the repair and regeneration of hard tissues. A sequential release of multiple drugs (a rapid release of drug 1 accompanied by a slow release of drug 2) was enabled by pre-loading drug 2 within mesoporous bioactive glass nanospheres (mBGn) which were added up to 30% to a polymer (polycaprolactone–gelatin) fiber matrix that has also encapsulated drug 1. In particular, excellent bioactive properties of mBGn, i.e., induction of bone mineral-like apatite formation and release of therapeutic ions (calcium and silicon) potentiate the usefulness of the mBGn-added scaffolds for bone regeneration. Proof-of-concept study utilizing two model drugs within the mBGn-added fiber (procaine hydrochloride (PCH) in mBGn and tetracycline hydrochloride (TCH) in nanofiber) demonstrated a typical sequential release pattern of the drugs, i.e., a rapid release of TCH within 24 h while a sustainable and long-term release of PCH over weeks to a month. Although biological efficacy of the drug-delivering scaffolds warrants further study, this finding suggests the mBGn-added polymer fiber may be a potential therapeutic matrix for bone regeneration.

---

1. Introduction

Development of scaffolds that can properly deliver therapeutic molecules is currently considered an important issue to improve the tissue regenerative capacity.^1,2 Many technological advances have been put forward to engineer scaffolding materials to deliver drugs and growth factors sustainably, timely and controllably, and even targeting the tissue site of action.^3,4

Among else, a recent emerging concept lies in the use of dual and even multiple drugs and growth factors that have different and time-dependent actions in the tissue repair and regeneration processes.⁵ Therefore, drugs delivered sequentially to the injured tissues in concern will act more effectively than when treated at the same time. One example is the requirement of angiogenic factors at the early stage of tissue repair while osteogenic drugs function at a much later phase of bone formation.⁶ Anti-inflammatory drugs are often necessitating at the early phase of injured tissue, and tissue specific differentiating factors are largely effective in tissue formation at a later stage.^7,8

Some other strategies to design scaffolds for this purpose have also been proposed; an example is a layered structure with a gradient composition to hold different type of drugs.⁹ A biphasic design incorporating two different materials is another promising way to deliver drugs sequentially.¹⁰ While the geometrical factors, including scaffolding shape and size, should first be considered, the composition of scaffolds in use, particularly for the case of degradable materials, is also important parameter as the drug release profile primarily depends on the degradation of matrix.

Here we propose a composite scaffold that is considered to deliver two different drugs in a sequential way. Moreover, the composition of the scaffold was chosen for the purpose of bone regeneration. As electrospun fibers have already demonstrated a potential scaffolding matrix for the repair and regeneration of tissues including bone, we used the electrospun fiber as the scaffolding matrix and to encapsulate drug 1. Next, an inorganic mesoporous nanocarrier is required to contain drug 2; mesoporous silica nanoparticles are well known drug nanocarriers^11–16 while lacking the potential for bone regeneration due to their slow and poor hydroxyapatite forming ability. Therefore, a nanocomposite composition, comprising of biopolymer and “bioactive” inorganic nanoparticles, specifically polycaprolactone–gelatin (PCL–Gel) blend polymer with bioactive glass nanoparticle (BGn) was used. BGn has recently been demonstrated to be excellent inorganic nanomaterials for hard tissue regeneration.¹⁷ In particular, we developed here the mesoporous BGn which is purposed to load drug molecules.

Hence, the current scaffold is in the form of electrospun fiber and the composition is PCL–Gel matrix with incorporated mesoporous BGn (mBGn). In order to realize the sequential dual drug delivery, one type of drug is contained within the biopolymer fiber for rapid release while another type of drug is loaded within the mBGn for more sustained release. We describe the processing of the composite scaffolds and their characteristics, and perform a proof-of-concept study to show the sequential drug delivery capacity of the scaffolds using two model drugs.

2. Experimental section

2.1. Materials

Tetraethyl orthosilicate (TEOS, C₈H₂₀O₄Si, 98%), Calcium nitrate tetrahydrate (Ca(NO₃)₂·4H₂O, 99%), poly(ethylene glycol), (PEG, (C₂H₄O)_nH₂O, M_n = 10 000), ammonium hydroxide (NH₄OH, 28.0% NH₃ in water, ≥99.99% metal basis), methanol anhydrous (CH₃OH, 99.8%), Gelatin (Gel, type B from bovine skin), poly(ε-caprolactone) (PCL, (C₆H₁₀O₂)_n, M_n = 70 000–90 000), 2,2,2-trifluroethanol (TFE, CF₃CH₂OH, ≥99%), tetracycline hydrochloride (TCH, C₂₂H₂₄N₂O₈·HCl), procaine hydrochloride (PCH, C₁₃H₂₀N₂O₂·HCl), high purity chemicals for simulated body fluid (SBF), tris(hydroxymethyl aminomethane) (Tris–buffer), 1 N hydrochloric acid (IN HCl) and phosphate buffered saline (PBS) tablets were all purchased from Sigma-Aldrich and were used as-received without any further purification. Ultrapure deionized water (18.2 MΩ cm, Millipore Direct-Q system) was used throughout.

2.2. Sol–gel synthesis of mesoporous bioactive glass nanospheres

Mesoporous bioactive glass nanospheres (mBGn) with well-developed spherical morphology and mesoporous structure were prepared using a new ultra-sound assisted base-catalyzed sol–gel method and PEG as a template.¹⁷ A composition of 75% SiO₂ and 25% CaO (mol%) was used for the glass preparation. In a typical synthesis, 5 g PEG was dissolved by magnetic stirring in 120 ml absolute methanol and then the solution pH was adjusted to 12.5 by adding NH₄OH. To this clear solution, 0.299 g Ca (NO₃)₂·4H₂O was added and dissolved upon magnetic stirring. In a separate vial, 0.793 g TEOS was diluted with 30 ml absolute methanol and then added drop-wise to the pH-adjusted solution while vigorously stirring and simultaneously applying a high power ultra-sound for 20 min. The white precipitate obtained after 24 h of vigorous magnetic stirring was separated and washed by water–ethanol in 3 centrifugation/redispersion cycles at 5000 rpm for 5 min and then dried at 70 °C overnight. To remove the PEG template, the dried powder was calcined at 600 °C in air for 5 h. The calcined nanopowder was stored under vacuum for further uses.

2.3. Electrospinning of PCL–Gel fibers incorporating mBGn

The PCL–Gel and PCL–Gel–mBGn fibers were prepared by electrospinning technique. For PCL–Gel fibers, 10 wt% Gel and 10 wt% PCL solutions in TFE were separately prepared under magnetic stirring at 40 °C and then both solutions were mixed at 1 : 1 weight ratio and vigorously stirred at room temperature to allow homogenization. For PCL–Gel–mBGn fibers, mBGn were first dispersed separately in both 10 wt% Gel and 10 wt% PCL solution in TFE at proper amounts (mBGn : Gel or PCL = 1 : 1, 2 : 1 or 3 : 1 by wt%) and the two mixtures were homogenized. The prepared solutions were introduced into a 10 ml plastic syringe with a 23-gauge needle tip and then engaged to an injection pump (adjusted at 1 ml h⁻¹ injection rate) of the electrospinning machine operated at 10.5 kV using a high-voltage DC power supply. The electrospun fibers were collected on an aluminum-foil sheet attached to a rotating cylinder (220 rpm) located at distance of 14.5 cm from the tip of the syringe needle. Finally, the prepared electrospun fibrous sheet was allowed to completely evaporate any residual TFE solvent and then kept in a desiccator for further uses.

2.4. Characterization

The mesoporous structure, particle size and the dispersion of the mBGn within the polymer fiber matrix were observed by a high-resolution transmission electron microscopy (HR-TEM, JEM-3010, JEOL, Japan) equipped with an energy-dispersive X-ray spectrometer (EDX, Oxford Instruments). The morphology of the prepared mBGn and fibrous scaffolds were observed by field-emission scanning electron microscopy (FE-SEM, Tescan, MIRA II LMH, Czech Republic). Before observation, the scaffolds were coated with platinum for 120 s using automatic magnetron sputter coater (Cressington 108 Auto sputter coater, UK). The diameter of the fiber was measured from the SEM micrographs using image analysis software. The phase of the mBGn and fiber samples before and after immersion in SBF was analyzed with X-ray diffraction (XRD, Rigaku, Ultima IV, Japan) using CuKα radiation (λ = 1.5418 Å). X-ray was generated at 40 mA and 40 kV, and data were obtained at diffraction angles (2θ) from 4° to 70° with a step size of 0.02° and a scanning speed of 2° min⁻¹.

Specific surface area, pore volume and pore size of the prepared mBGn were obtained from N₂ adsorption–desorption measurements. The N₂ adsorption–desorption isotherms were obtained at −196.15 °C on an automated surface area and pore size analyzer (Quadrasorb SI, Quantachrom instruments Ltd., USA.). Samples were degassed under vacuum at 300 °C for 12 h prior to analysis. The specific surface area was calculated according to the Brunauer–Emmett–Teller (BET) method. The pore size distribution was determined from the N₂ desorption branch of the obtained N₂ adsorption–desorption isotherms on the basis of the non-local density functional theory (NLDFT) method. Attenuated total reflectance-Fourier transform infrared spectra (ATR-FTIR; Varian 640-IR, Australia) of the mBGn before and after SBF immersion and fiber samples with different mBGn contents were obtained with a resolution of 4 cm⁻¹ in the range 4000–400 cm⁻¹ using GladiATR diamond crystal accessory (PIKE Technologies, USA). Thermogravimetric analysis of the drug loaded mBGn was carried out on a thermogravimetric analyzer (TGA N-1500, Scinco Ltd., South Korea) at a heating rate of 10 °C min⁻¹ and a nitrogen flow rate of 40 ml min⁻¹. The cumulative release of calcium and silicon ions from the nanoparticles was measured at 37 °C for up to 14 days. Nanoparticles of 20 mg were immersed in 10 ml Tri-HCl buffered solution at pH 7.4. At predetermined time points, samples were taken and centrifuged at 15 000 rpm for 15 min and the supernatants were collected for characterization with inductively coupled plasma atomic emission spectrometry (ICP-AES; OPTIMA 4300 DV, Perkin-Elmer, USA). Three replicate samples were evaluated and the average value was recorded.

2.5. In vitro bioactivity assessment

The in vitro hydroxyapatite forming ability of the mBGn and the fibrous scaffolds were tested in Kokubo's simulated body fluid (SBF, pH 7.4) at 37 °C.¹⁸ SBF was prepared by dissolving NaCl (142.0 mM), NaHCO₃ (4.2 mM), KCl (5.0 mM), K₂HPO₄·3H₂O (1.0 mM), MgCl₂·6H₂O (1.5 mM), CaCl₂ (2.5 mM), and Na₂SO₄ (0.5 mM) into deionized water buffered to pH 7.4 with Tris–HCl buffer at 37 °C.¹⁹ Fibrous webs were cut into circular shape samples of 15 mm diameter and subsequently placed in 6-well plates containing 10 ml SBF in each well and were incubated at 37 °C for 28 days with refresh of the SBF solution every day. After incubating the samples, the fibrous scaffolds were removed from SBF, washed with distilled water, and dried. Formation of bone like apatite layer on the surface of fibrous scaffolds was determined by XRD, SEM and EDS (Bruker).

2.6. Contact angle measurement

Surface hydrophilicity of the fibrous scaffolds was assessed by water contact angle measurements at room temperature by the sessile drop method using a benchtop Phoenix contact angle measurement system (PHX300, SEO, South Korea). The fibrous scaffolds were placed on a sample holder stage and one drop of deionized water (2 μL) automatically delivered from a needle connected to a syringe pump was dropped onto the sample surface. Then, images of the water droplets were recorded within 15 s using a video camera system (CCD camera) and the surface contact angles were calculated based on these images using Image XP software which gives numerical values of the left and right contact angles. The average values of contact angle of tested samples were used.

2.7. Drug loading and in vitro release study

Drug 1 (Tetracycline Hydrochloride, TCH) was loaded directly onto the PCL–gelatin polymer matrix at 5 wt%, where TCH was well solubilized in TFE by a magnetic stirring under dark conditions prior to dissolving the polymers. To determine the amount of TCH loaded in the PCL–Gel matrix, 10 mg of accurately weighed samples was dissolved in TFE and the drug content was quantified by UV measurement at a characteristic detection wavelength, λ_max = 357 nm using the TCH calibration curve.

Drug 2 (Procaine Hydrochloride, PCH) was loaded onto the mBGn and the loading capacity was investigated by dispersing the mBG nanospheres at concentration of 10 mg ml⁻¹ in aqueous solutions (pH 3) containing different concentrations of PCH (5, 10, 20, and 25 and 30 mg ml⁻¹). The PCH loaded amounts were determined from TGA, and were represented by the adsorption isotherm, i.e., plotting the PCH quantity loaded within the mBGn with respect to the concentration of drug initially added into the medium. The PCH loading amount was plotted according to the following mass balance equation: q_e = (C₀ − C_e) × (V/W), where q_e is the amount of drug (in mg) adsorbed per mg mBGn, C₀ and C_e are the initial and equilibrium concentrations of PCH, respectively (mg ml⁻¹), V is the volume of solution (ml), and W is the weight of the mBGn used (mg). After plotting the q_e-versus-C_e curves, a modified Langmuir isotherm model was applied for curve-fitting, according to the following equation: q_e = q_mKC_e/(1 + KC_e),²⁰ where q_m is the maximum amount loaded and K is unknown parameter (kinetic constant) which can be determined.

The mBGn loaded with 21.5% PCH were selected for the preparation of the drug loaded PCL–Gel–mBGn fibers. To avoid any possible interference during the UV-VIS assay of each drug, two composite solutions of PCL–Gel–mBGn were prepared; one containing 10 wt% PCH-loaded mBGn and the other containing 10% mBGn and 5 wt% TCH. The solutions were electrospun under the same conditions as mentioned above and finally dried and stored in a desiccator kept in dark place.

The in vitro release of drugs from the bare mBGn and the fibrous scaffolds was carried out in PBS solution (pH 7.4) and the release amount was determined using a UV-VIS spectrometer (Libra S22, Biochrom, UK). A series of standard PCH and TCH solutions in PBS were prepared and the calibration curves of PCH and TCH were determined by taking absorbance values at λ_max vs. PCH and TCH concentrations between 0 and 30 μg ml⁻¹. For this concentration range, the calibration curves fit the Lambert and Beers' law: A₂₉₀ = 0.0727C − 0.0412 (R_PCH² = 0.999), A₃₅₇ = 0.0316C (R_TCH² = 0.998) where A is absorbance at λ_max and C designates concentration (μg ml⁻¹). For the determination of PCH release from mBGn, 5 mg of mBGn loaded with 21.5% PCH was dispersed in 25 ml PBS and incubated at 37 °C for different time periods. For the determination of either PCH or TCH released from fibers, 10 mg of accurately weighed samples was immersed in 10 ml PBS and incubated at 37 °C for different time periods. At each pre-determined time point, 1 ml release medium was withdrawn for the analysis.

3. Results and discussion

3.1. Design and characteristics of mBGn

As depicted in Fig. 1a, the nanofiber composite scaffold is designed to perform a sequential drug delivery which is composed of polymer fiber matrix with mesoporous nanospheres (mBGn). Drug 1 is directly dispersed in the fiber matrix which is to release rapidly at the initial period, while drug 2 is incorporated within mBGn to profile more sustained and long-term release, as drawn in the release profile (Fig. 1b).

Fig. 1 (a) Schematic diagram, showing sequential drug releasing therapeutic scaffolds made of polymer nanofiber incorporating drug 1 and drug 2-loaded mBGn, where the drug 1 is to release rapidly while drug 2 is to come out more sustainably to show a sequential release profile, as depicted in (b). (c–i) Physico-chemical characteristics of mBGn: (c) TEM image taken at low and high magnification, showing monodispersed mesoporous nanospheres, (d) size distribution of the mBGn measured from TEM images, showing an average particle size of 62.7 nm, (e) TEM-EDS analysis revealing an atomic composition of Si : Ca = 74 : 26, (f) mesoporous properties of mBGn, represented as BET N₂ adsorption–desorption curve, and summarized with parameters including specific surface area (47.2 m² g⁻¹), pore volume (0.116 cm³ g⁻¹) and mesopore size (3.54 nm), and (i) hydroxyapatite forming ability of mBGn in SBF, as revealed by TEM image, showing a number of mineralized nanocrystallites developed at 3 days of SBF-test, and the XRD pattern change with test period (1, 3 and 7 days), showing characteristic apatite peaks.

To realize this system, we first developed the mBGn with composition of 75SiO₂–25CaO (mol%) using a new sono-reaction assisted sol–gel process under alkaline conditions and PEG as a template. The produced nanoparticles were seen under TEM (shown in Fig. 1c). Uniform-sized nanoparticles were well developed by this new process. Higher magnification of TEM image revealed the mesoporous structure. The nanoparticle size analyzed by the TEM images gave an average size of 62.7 nm ± 12.3 nm with a narrow distribution (Fig. 1d). TEM-EDS analysis showed an atomic composition of the mBGn with Si/Ca = 74/26 (Fig. 1e), being matched to the intended composition (75SiO₂–25CaO).

The mesopore structure of the mBGn was analyzed by BET method. The N₂ adsorption–desorption curve of the mBGn showed a typical hysteresis loop of mesoporous nanomaterials (Fig. 1f). The mesopore size distribution of the mBGn exhibited a short range of mesopores mostly in 3–4 nm size. The specific surface area, pore volume and mesopore size were calculated to be 47.2 m² g⁻¹, 0.116 cm³ g⁻¹, and 3.54 nm, respectively (summarized in Fig. 1g). This high mesoporosity of the mBGn developed herein is considered to take up a large quantity of drug molecules within the structure.

As the composition of mBGn (75SiO₂–25CaO) was developed to be bone-bioactive, i.e., forming a bone mineral-like hydroxyapatite (HA) phase in simulated body fluid (SBF). The in vitro HA forming ability of the prepared mBGn was tested in SBF. TEM image of mBGn immersed for 3 days in SBF showed a significant change in morphology, flowering of flake-like nanocrystallites through on the nanoparticles (Fig. 1h), which is a typical morphology of biomimetic apatite formed in SBF. XRD pattern of the SBF-treated mBGn for varying time point (1, 3 and 7 days) showed the development of peaks at the corresponding diffraction angles, including a main peak at 2θ ∼32°, which being a characteristic of HA mineral particularly poorly crystallized (Fig. 1i).

Based on this, it is summarized that the developed mBGn has a size of ∼60 nm with a high mesoporosity and a mesopore size of ∼3.5 nm, and holds excellent bone bioactivity to form HA mineral phase. Next the mBGn will be utilized as drug loading vehicles and for producing nanocomposite fibrous scaffolds by electrospinning.

3.2. Fibrous scaffolds of PCL–Gel incorporating mBGn

The developed mBGn were incorporated at varying contents (0–40 wt%) within a polymeric solution made of PCL–Gel to implement an electrospinning process. The electrospinnability of the nanocomposite solutions into a fibrous web was first examined. The PCL–Gel fiber was generated well with uniform-sized (Fig. 2a). When 10% mBGn was incorporated, the fiber was well produced with a smooth surface (Fig. 2b). When 20% mBGn was added, fiber was also generated well to form a similar morphology to 10% mBGn case (Fig. 2c). Some of the surface (20% mBGn) appeared to be protruded by the nanoparticles present inside. When 30% mBGn was used, the fiber generation was also successful without a bead formation (Fig. 2d). When 40% mBGn was incorporated, a bead formation started which was due to the agglomerated nanoparticles, in which case, the fiber sizes were not uniform showing ultrathin fibers along the beads (Fig. 2e). TEM image of the 30% mBGn sample was taken representatively to examine the internal nanostructure of the fiber (Fig. 2f). Mesoporous nanospheres were well distributed within the polymer matrix. The fiber size measured from the SEM images showed a slight increase with increasing mBGn content (Fig. 2g), which was due to the possible increased viscosity of the nanocomposite solutions. The FT-IR spectra of the fiber samples showed the chemical bond structure of the compositions, where the 30% mBGn sample showed a substantial level of additional bands (Si–O–Si) related with mBGn (Fig. 2h).

Fig. 2 (a–e) SEM images showing the electrospun fibers incorporating various contents of mBGn: (a) 0, (b) 10, (c) 20, (d) 30 and (e) 40%. Over 40%, fiber morphology started to be disintegrated with bead formation. The mBGn dispersed within polymer matrix (30% mBGn) are revealed on TEM images (f). (g) Fiber size measured from SEM images, and (h) FT-IR spectra of the fibers incorporating mBGn (0 and 30% mBGn are shown representatively).

Results demonstrated that the mBGn added up to 30% preserved well the electrospun fiber morphology with uniform sizes, featuring a typical nanocomposite, i.e., nanoparticles-distributed polymer matrix. However, the addition over 40% disintegrated the fiber generation showing the formation of lots of beads which resulting from highly agglomerated nanoparticles. For the further studies, the compositions containing up to 30% mBGn will be used.

The nanocomposite fibrous scaffolds were further characterized in terms of their properties that are meaningful for the biomedical uses particularly for cell culture and bone regeneration. First, the hydrophilic property of the fiber scaffolds was measured by a contact angle test. Although the PCL–Gel composition shows high hydrophilicity with a contact angle of ∼24° which is due to the added gelatin at an equivalent amount to PCL, the addition of mBGn further improved the property, with an ongoing decrease of contact angles with mBGn additions (16° and 10°, for 10% and 20% mBGn, respectively), and finally a complete wetting with 30% mBGn addition (Fig. 3a).

Fig. 3 Physico-chemical properties of the mBGn-incorporating fiber scaffolds: (a) contact angle measurement revealing a substantial increase in hydrophilicity with mBGn addition. (b and c) SBF immersion test for 28 days, showing the role of mBGn in improvement of apatite forming ability of the fiber scaffolds; (b) XRD patterns of 30% mBGn sample before and after SBF test (HA crystal planes are also indexed), and (c) SEM images of samples (0% and 30% mBGn after SBF test), as well as EDS analysis of the sample (30% mBGn after SBF test).

Next, the apatite forming ability of the nanocomposite fiber scaffolds was tested in SBF. After incubation in SBF for 4 weeks, the XRD pattern of the scaffolds was analyzed (Fig. 3b). The 4 weeks, a relatively longer period than general SBF-study periods, was chosen to see the complete and uniform coverage of apatite crystals through the fiber. While there were only PCL peaks before immersion, very strong HA peaks at the corresponding crystal planes were developed on the 30% mBGn sample after the SBF immersion. SEM images of the scaffolds during the SBF test were examined. After the SBF immersion, there was no noticeable mineralization on 0% mBGn scaffold, however, a substantial level of mineralization occurred on the 30% mBGn scaffold, covering the whole surface of the fiber, as clearly revealed on a high magnification SEM image (Fig. 3c). Furthermore, the Ca and P atomic peaks were strongly signaled by EDS analysis with Ca/P ratio of 1.63, which is very close to stoichiometry of HA (Ca/P = 1.67). This mineralization behavior of the fiber scaffolds demonstrated well the role of the mBGn component in improving the surface activity, i.e., bone-bioactivity, which will be favorable for the cellular interactions with the interface of a scaffold, particularly in the osteogenesis and later stage of cellular mineralization.²¹ This HA mineralization on the surface of the scaffolds was primarily driven/accelerated by the possible release of ions particularly calcium ions which were contained in the mBGn. The released ions will supersaturate the SBF condition with respect to calcium ions which accelerates the calcium deposition first onto the PCL–Gel surface and then phosphate and carbonate ions will follow to form calcium phosphate compound and further to crystallize into HA phase with time.²² As the PCL–Gel did not show any mineral inductions during the test, it is considered that PCL–Gel blend surface has very low innate ability to induce CaP mineralization. As the SBF medium was refreshed daily, the condition can be considered semi-dynamic which is slightly different from a static condition. While in case of keeping the SBF medium not refreshed, the ions in SBF will stably be induced to form mineral, which however, was largely limited in this study by the continual medium change.

The possibility of ion release from the mBGn-added scaffolds was explained in terms of ionic release from the bare mBGn. The Ca and Si ions released from 20 mg of the nanoparticle sample immersed in 10 ml of Tri-buffered medium at pH 7.4, were measured using ICP-AES analysis (shown in Fig. 4). Ca ions released ∼8 mM for the 1^st day while Si ions released a half of that for 1 day. After this, the Ca ions were released continuously profiling an almost linear pattern up to 14 days recording ∼15 mM; on the other hand, Si ions were not released so notably, but there was only slight change during 14 days, recording ∼5 mM. It may be assumed that the release from the nanoparticles can be extrapolated to the nanocomposite scaffolds. For example, when we consider 10 mg of the 30% mBGn-added scaffold sample, it has ∼3 mg of mBGn which initially can release Ca ions of ∼1.16 mM and Si ions of 0.56 mM within 1 day followed by linear Ca ions release in the range from ∼1.16 to ∼2.22 mM and Si ions release in the range from ∼0.56 to ∼0.74 mM during 14 days. However, the extrapolation is rather simplified and ambiguous as the mBGn is embedded within the fiber matrix and thus the ionic release will be more slowed down due to PCL–Gel diffusion barrier, thus the values considered will be at most the upper limit. As far as the hypothesis is taken, this concentration range of Ca ions should definitely influence the supersaturation of the medium and consequently the mineralization behavior of the scaffolds. Another point to highlight in the ionic release is that the released Ca and Si ions will affect the cellular behaviors, such as cell proliferation and osteogenic differentiation which will be helpful for utilizing the nanocomposite scaffolds for bone regeneration. Several recent studies have investigated the effects of Ca and Si ions on the cell responses, where the concentration range effective for osteogenic differentiation was 0.2–10 mM for Ca and 0.02–2 mM for Si, a much narrower window being noticed for Si.^23–27 The recorded values herein, though resulting from the bare mBGn case, are considered to be within the recommended concentration region. Further cellular experiments will thus be needed to confirm this possible effective role.

Fig. 4 Ionic release of mBGn, as analyzed by ICP-AES, showing a continual release of calcium and silicon ions during test period.

3.3. Dual drug release profiles from scaffolds

Overall, the mBGn-added polymer fibrous scaffolds were developed to implement sequential delivery of therapeutic drugs. A proof-of-concept study was carried out using two model drugs, PCH and TCH, introduced within mBGn and polymer matrix, respectively. PCH is recently known for its anticancer activity while TCH is a well-known antibiotic.^28,29 Both drugs have some similarities in the chemical properties, molecular size and charge property, that allows us to record the dual release profiles but from separate gains and enabling to investigate drug-independent matrix effects. The PCH loading capacity of mBGn and its in vitro release behavior were investigated. PCH was initially pooled within distilled water at varying concentration (0–30 mg ml⁻¹), and 10 mg of mBGn was added to incorporate the drug molecules. The loading quantity was measured and presented as the mg PCH per mg mBGn (Fig. 5a). Results exhibited a loading of 0.175 mg when 5 mg PCH was used, which gave ∼17.5% with respect to host carrier mBGn. With higher concentration of PCH, the loading quantity increased very slowly, 0.215, 0.26, 0.285 mg for 10, 20 and 25 mg PCH used respectively, which gave ∼22–28.5% with respect to mBGn. The loading curve appeared to show a plateau at the highest PCH concentration (30 mg), and thus the maximal loading quantity was achieved to be 0.289 mg mg⁻¹ when 30 mg PCH was used, which was ∼29% loading with respect to mBGn. This loading capacity of mBGn for PCH is considered to be very high, suggesting the mBGn is highly effective for delivering small drug molecules like PCH. Next a well-known modified Langmuir isotherm model was applied to the experimental data, for the curve-fitting of the graph. The experimental isotherm curve was shown to fit pretty well to the model (R² = 0.991), with K and q_m determined to be 0.342 and 0.314, respectively.

Fig. 5 (a) Drug PCH loading onto mBGn and (b) the subsequent release profile from mBGn. Curve fitting using modified Langmuir model is included in (a). Model equation and parameters obtained are also provided. PCH release from mBGn appeared to almost complete within 12 h.

The in vitro release of PCH from mBGn was then recorded (Fig. 5b). In this case, a sample loaded with PCH at 0.215 mg mg⁻¹ mBGn (21.5% PCH loaded onto mBGn) was used as a representative sample. The release of PCH was very fast, profiling an almost complete release within 24 h. It is considered that PCH drug molecules form very weak chemical interactions with the mBGn, and thus an ionic exchange within a saline medium happens easily, liberating the PCH quickly. It is also deemed that the ionic dissolution and material degradation of mBGn should affect the release of the PCH drug. The relatively fast ionic releases within a day (recalling Fig. 4) should also accelerate the liberation of the drug molecules adsorbed on the surface.

Based on this drug release behavior from mBGn, it was believed that when the mBGn–drug complex is incorporated within the fiber matrix, the release behavior will be further slowed-down. Therefore, a design of the mBGn–polymer composite fibrous scaffolds was made to deliver two drugs, PCH within mBGn and TCH within polymer matrix. The PCH loading concentration was made at 21.5%, a condition same as that used for the mBGn release test. The mBGn–PCH content within the fiber matrix was made at 10% while TCH at 5% (all by weight). Similar morphological features were observed for the drug-loaded composite fibers and the drug-free scaffolds, confirming the drug molecules did not hamper the fiber formation via an electrospinning. The TCH and PCH drug release patterns were investigated in Fig. 6. TCH drug was released very rapidly, recording almost complete release within ∼48 h, which demonstrated the hydrophilic TCH drug molecules easily diffused out through the PCL–Gel matrix.

Fig. 6 Dual-drug release profile from the 10% mBGn-added nanocomposite fiber scaffold, performed as a proof-of-concept study. Two model drugs loaded either within mBGn (drug 2, PCH) or within nanofiber (drug 1, TCH) were released in PBS for different periods. While TCH releases rapidly with almost complete release within 48 h from the fiber scaffold, PCH which was pre-loaded in mBGn releases much more sustainably profiling over 3 weeks. PCH release was relatively rapid initially up to 1 week and then slowed down thereafter. Data well fitted according to the modified model of Ritger–Peppas empirical equation, and kinetics parameters are also presented in the graph. M_t/M_∞ = Ktⁿ where M_t and M_∞ are the absolute amount of drug released at time t and infinite time (∞), respectively, and K is released rate constant for each equation, incorporating structural and geometric characteristics of the drug delivery device, and n is the released exponent, indicative of the drug release mechanism.

On the other hand, PCH release pattern was in striking contrast to the TCH release. PCH release showed a long-term release pattern, i.e., drug molecules initially released rather rapidly with ∼60.5% of drug loaded being released within a week, and then released very sustainably showing almost linearly to the test period (over 3 weeks). The PCH release event is very complicated as the PCH should diffuse out through mBGn pore channels and then through the polymer matrix, i.e., influenced both by mBGn and polymer matrix. As soon as the PCH molecules diffuse out from the mBGn and present on the polymer matrix, the dissolution of and diffusion through the polymer should affect the drug release. As we have seen already for the case of TCH, the drug release through the polymer matrix is considered to be fast. Therefore, the PCH release through the polymer matrix will not be a rate determining process. Rather, the PCH liberation from the mBGn will more dominantly control the release rate. Although, we have observed that PCH from bare mBGn was rapidly liberated, this, however, will not occur similarly in case of embedding the mBGn within the polymer matrix. The liberation process should be substantially slowed down due to the limited action of water molecules which have to diffuse into the polymer networks and further to liberate the PCH drug molecules that are adsorbed onto the mBGn surface. Consequently, the polymer matrix was considered to play some crucial roles in slowing down the release of PCH drug, providing a diffusional barrier for water come-in as well as a liberation reaction barrier between water and PCH drug, consequently sustaining the drug release over weeks to a month. The experimental result was fitted using a modified model of Ritger–Peppas empirical equation, i.e., M_t/M_∞ = Ktⁿ, where M_t and M_∞ are the amount of drug released at time t and infinity (∞), respectively, and K is release rate constant for each equation, incorporating structural and geometric characteristics of the drug delivery device, and n is the released exponent, indicative of the drug release mechanism. Data fitted pretty well with R² = 0.983, giving K value of 12.5 and n value of 0.3. As to the diffusional exponent characteristic, the exponent n is defined to be 0.45 when drug is released from a swelling structure containing hydrophilic polymers, while the degradable matrices show anomalous behaviors with much higher n values.^30,31 However, in our case, a much lower exponent value (n = 0.3) was obtained, which being possibly due to that the drug diffusion-out from the mesopore channels of mBGn was highly sustained. It has been also stated that n ≤ 0.5 in Peppas model is indicative of a diffusion-controlled mechanism.^30,31

Although here we used the two model drugs that are not exactly relevant for bone regeneration, the design concept of sequential drug delivery was considered to be proved. As such, we consider applying other drugs that are favorable for tissue response and bone regeneration after implantation in bone defects. One possibility is to use anti-inflammatory drug (as drug 1) together with osteogenic drug (as drug 2).

4. Conclusions

Collectively, the current proof-of-concept study clearly demonstrated the sequential release strategy of two different drugs using mBGn–polymer nanocomposite fiber scaffolds. As the mBGn-added scaffolds have high potential for bone bioactivity and possible bone regenerating capacity, which though requiring further studies, an improvement of the scaffolds with this delivery concept will create promising therapeutic bone scaffolds.

References

1. T. H. Kim, J. J. Kim and H. W. Kim, Biotechnol. Lett., 2013 DOI:10.1007/s10529-013-1366-4.
2. S. Srouji, D. Ben-David, R. Lotan, E. Livne, R. Avrahami and E. Zussman, Tissue Eng., Part A, 2011, 17, 269–277.
3. Z. Zhang, J. Hu and P. X. Ma, Adv. Drug Delivery Rev., 2012, 64, 1129–1141.
4. S. H. Shin, O. Purevdorj, O. Castano, J. A. Planell and H. W. Kim, J. Tissue Eng., 2012, 3, 2041731412443530.
5. O. Qutachi, K. M. Shakesheff and L. D. K. Buttery, J. Controlled Release, 2013, 168, 18–27.
6. Q. Cui, A. S. Dighe and J. J. Irvine, Curr. Pharm. Des., 2013, 19, 3374–3383.
7. V. Mouriño and A. R. Boccaccini, J. R. Soc., Interface, 2010, 43, 209–227.
8. A. Martins, A. R. C. Duarte, S. Faria, A. P. Marques, R. L. Reis and N. M. Neves, Biomaterials, 2010, 31, 5875–5885.
9. T. Okuda, K. Tominaga and S. Kidoaki, J. Controlled Release, 2010, 143, 258–264.
10. N. Alhusein, I. S. Blagbrough and P. A. D. Bank, Drug Delivery Transl. Res., 2012, 2, 477–488.
11. S. Kwon, R. K. Singh, R. A. Perez, E. A. Abou Neel, H. W. Kim and W. Chrzanowski, J. Tissue Eng., 2013, 4, 2041731413503357.
12. B. Song, C. Wu and J. Chang, Acta Biomater., 2012, 8, 1901–1907.
13. Z. Hou, X. Li, C. Li, Y. Dai, P. Ma, X. Zhang, X. Kang, Z. Cheng and J. Lin, Langmuir, 2013, 29, 9473–9482.
14. K. Qiu, C. He, W. Feng, W. Wang, X. Zhou, Z. Yin, L. Chen, H. Wang and X. Mo, J. Mater. Chem. B, 2013, 1, 4601–4611.
15. M. Vallet-Regi, F. Balas and D. Arcos, Angew. Chem., 2007, 40, 7548–7558.
16. M. Manzano and M. Vallet-Regi, Prog. Solid State Chem., 2012, 40, 17–30.
17. A. El-Fiqi, T. H. Kim, M. Kim, M. Eltohamy, J. E. Won, E. J. Lee and H. W. Kim, Nanoscale, 2012, 4, 7475–7488.
18. T. Kokubo, H. Kushitani, S. Sakka, T. Kitsugi and T. Yamamuro, J. Biomed. Mater. Res., 1990, 24, 721–734.
19. T. Kokubo and H. Takadama, Biomaterials, 2006, 27, 2907–2915.
20. I. Smirnova, J. Mamic and W. Arlt, Langmuir, 2003, 19, 8521–8525.
21. H. S. Yu, S. J. Hong and H. W. Kim, Mater. Chem. Phys., 2009, 113, 873–877.
22. F. Yang, S. K. Both, X. Yang, X. F. Walboomers and J. A. Jansen, Acta Biomater., 2009, 5, 3295–3304.
23. V. G. Varanasi, K. K. Leong, L. M. Dominia, S. M. Jue, P. M. Loomer and G. W. Marshall, J. Oral Implantol., 2012, 38, 325–336.
24. K. J. Kim, Y. J. Jeon, J. H. Lee, S. T. Ahn, S. H. Lee, D. W. Cho and J. W. Rhie, Tissue Eng. Regener. Med., 2010, 7, 171–177.
25. W. Zhai, H. Lu, C. Wu, L. Chen, X. Lin, K. Naoki, G. Chen and J. Chang, Acta Biomater., 2013, 9, 8004–8014.
26. P. Han, C. Wu and Y. Xiao, Biomater. Sci., 2013, 1, 379–392.
27. O. Tsigkou, J. R. Jones, J. M. Polak and M. M. Stevens, Biomaterials, 2009, 30, 542–550.
28. A. Villar-Garea, M. F. Fraga, J. Espada and M. Esteller, Cancer Res., 2003, 63, 4984–4989.
29. S. S. DeRossi and E. V. Hersh, Dent. Clin. North Am., 2002, 46, 653–664.
30. U. Gbureck, E. Vorndran, F. A. Müller and J. E. Barralet, J. Controlled Release, 2007, 122, 173–180.
31. S. Hesaraki, A. Zamanian and M. Khorami, J. Australas. Ceram. Soc., 2011, 47, 32–36.

---