**PLGA Based Drug Carrier and Pharmaceutical Applications**

Editors

**Maria Carmo Pereira Joana A. Loureiro**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editors* Maria Carmo Pereira Biotechnology and Energy Portugal

Joana A. Loureiro Biotechnology and Energy Portugal

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *Pharmaceutics* (ISSN 1999-4923) (available at: https://www.mdpi.com/journal/pharmaceutics/ special issues/PLGA Pharm).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Volume Number*, Page Range.

**ISBN 978-3-0365-0220-5 (Hbk) ISBN 978-3-0365-0221-2 (PDF)**

© 2021 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

### **Contents**




### **About the Editors**

**Maria Carmo Pereira** received her PhD in Chemical Engineering from the University of Porto, Portugal, in 1998. She is a Professor at the Department of Chemical Engineering, Faculty of Engineering, University of Porto, Portugal. Currently, she coordinates the "Supramolecular Assemblies" research group at LEPABE. Her research fields span biophysics, supramolecular interactions, including novel nano-engineered biomaterials for therapeutic applications, and environmental sciences. Her scientific activity is focused on the design and preparation of drug delivery systems—polymeric nanoparticles, liposomes, and conjugated systems for biomedical applications—and the development of immunosensors as biomarkers for neurodegenerative diseases. M.C. Pereira has published more than 160 papers in international peer-reviewed journals and 50 conference proceedings.

**Joana A. Loureiro** is an MSc in Chemical Engineering and in Pharmaceutical Sciences and received her PhD in Chemical Engineering from the University of Porto (Portugal) in 2013. She is a researcher at LEPABE, Department of Chemical Engineering, Faculty of Engineering, University of Porto, Portugal. In recent years, her scientific research has been directed at the design and production of drug delivery systems (DDS) functionalized with targeting molecules for drug targeting and pharmaceutical applications, mainly related to neurological diseases. Her scientific activity also focuses on the study of drug–membrane interactions using biomimetic in vitro models to predict drug behavior. J.A. Loureiro is the author or co-author of 40 publications in high impact international peer-reviewed journals and 32 conference proceedings.

### *Editorial* **PLGA Based Drug Carrier and Pharmaceutical Applications: The Most Recent Advances**

**Joana Angélica Loureiro \* and Maria Carmo Pereira \***

LEPABE, Department of Chemical Engineering, Faculty of Engineering of the University of Porto, s/n, R. Dr. Roberto Frias, 4200-465 Porto, Portugal

**\*** Correspondence: joana.loureiro@fe.up.pt (J.A.L.); mcsp@fe.up.pt (M.C.P.)

Received: 17 September 2020; Accepted: 21 September 2020; Published: 22 September 2020

Poly(lactic-*co*-glycolic acid) (PLGA) is one of the most successful polymers that has been used to produce medicines, such as drug carriers (DC). This is one of the few polymers that the Food and Drug Administration (FDA) has approved for human administration due to its biocompatibility and biodegradability [1]. DCs produced with PLGA have gained enormous attention over recent years for their ability to be versatile vehicles to transport different type of drugs, e.g., hydrophilic or hydrophobic small molecules or macromolecules, and protect them from degradation and uncontrolled release [2–6]. These drug delivery systems (DDS), including micro and nanoparticles, have the potential to modify their surface properties and improve interactions with biological materials. Furthermore, they can also be conjugated with specific target molecules to reach specific tissues or cells [7,8]. They are being used for different therapeutic applications, such as vaccinations or as treatments for cancer, neurological disorders, inflammation, and other diseases [9–12].

This Special Issue aims to focus on the recent progress of PLGA as a drug carrier and its new pharmaceutical applications. It comprises an exciting series of 19 research articles on the recent advances in the field.

In the first research study presented in this Special Issue, Ho et al. developed polymeric microspheres which contain micronized triamcinolone acetonide (TA) in order to increase the drug retention time in joints after intra-articular administration [13]. Poly(lactic-*co*-glycolic acid)/poly(lactic acid) (PLGA/PLA) carriers were prepared through spray-drying to incorporate the microcrystals that were previously prepared by ultra-sonication. In vivo testing in rat models was demonstrated to prolong drug retention in joints. The TA remained there for over 28 days, which was more 21 days compared with the TA-free group. Furthermore, these nanocarriers were demonstrated to be stable for one year.

The group of Peula-García used PLGA nanoparticles to carry bone morphogenetic protein (BMP-2) [14]. The nanocarriers were synthetized by a double-emulsion (water/oil/water, W/O/W) solvent evaporation technique, using the surfactant Pluronic F68 as a stabilizer. The BMP2-loaded nanocarriers presented positive results when evaluated using mesenchymal stromal cells from human alveolar bone regarding their proliferation, migration, and osteogenic differentiation. Another strategy to encapsulate BMP-2 was conducted by Minardi et al. [15]. PLGA multistage vector composite microspheres were used as carriers that demonstrated a good capacity for BMP-2 encapsulation and did not present toxicity for the rat mesenchymal stem cells. García-García et al. applied a combined strategy to regenerate tissue defects [16]. They used BMP-2- and 17β-estradiol-loaded microspheres, PLGA-based, in a sandwich-like system produced by a hydrogel core.

In another study, Hwang et al. fabricated PLGA carriers combined also with a hydrogel matrix. They produced oxaliplatin-loaded PLGA microparticles using a double emulsion technique and then loaded them into hyaluronic acid and carboxymethyl cellulose sodium-based cross-linked hydrogels [17]. This drug delivery system was analyzed in rat models and a substantial improvement was observed in terms of bioavailability and the mean residence time of the microparticle-loaded hydrogels.

Kim et al. developed an original system to be used in the topical delivery of trolamine salicylate (TS), a topical anti-inflammatory analgesic used for the treatment of small joint pain [18]. Here, the optimization process was done using different amounts of PLGA, ethyl 2-cyanoacrylate, poly (ethylene glycol) (PEG) 400, and TS. The researchers proved that the produced formulations enhanced the delivery of TS into and across the skin.

Duse et al. used PLGA nanoparticles to encapsulate curcumin, a well know natural compound that present anticancer benefits [19]. It was shown that the use of PLGA nanoparticles improves the bioavailability and site-specific drug uptake. The nanoparticles' efficacy was tested using SK-OV-3 human ovarian adenocarcinoma cells and demonstrated to be very efficient in transporting curcumin. Furthermore, with the same objective to treat the cancer, our research group used factorial design as a tool to optimize the co-encapsulation of temozolomide and o6-benzylguanine in PLGA nanoparticles [20]. The produced nanoparticles, rather than demonstrating stability for several days, presented optimal physicochemical properties for brain delivery, including a size lower than 200 nm and a negative surface charge. In the same research line, demonstrating the potential of the co-encapsulation, Bazyli ´nska et al. encapsulated a hydrophobic porphyrin photosensitizing dye—verteporfin—in combination with low-dose cisplatin, a hydrophilic cytostatic drug [21]. Different coatings were applied to the PLGA nanoparticles, PEG, or folic acid functionalized. Those nanoparticles proved to have an increased internalization and efficiency regarding anticancer activity.

Another interesting study proposed by Varga and colleagues, who contributed with an interesting study of nanoparticle design and optimization where the (±)-α-Tocopherol (TP) with vitamin E activity was encapsulated in PLA and PLGA nanoparticles [22]. To stabilize the nanoparticles, the non-ionic stabilizing surfactant Pluronic F127 was used. Several techniques were used to characterize these novel nanoparticles, such as transmission electron microscopy (TEM), dynamic light scattering (DLS), and infrared spectroscopy (FT-IR).

Morelli et al. improved paclitaxel delivery in the gastro-intestinal tract by encapsulating the drug in PLGA nanoparticles coated with PEG [23]. The nanoparticles demonstrated stability in the gastric tract and simply penetrated inside carcinoma colon 2 (CaCo2) cells.

With the objective to overcome the undesired lag time of the commercially available risperidone, Janich et al. encapsulated this drug in PLGA–lipid microcapsules and PLGA–lipid microgels [24]. The carriers were evaluated regarding their physio-chemical properties and the first formulation was demonstrated to achieve a nearly zero order release without a lag time over 2 months.

A research work using PLGA nanoparticles for ocular application was also collected. Ryu et al. produced rapidly dissolving dry tablets containing alginate and dexamethasone-loaded PLGA nanoparticles [25]. These nanoparticles presented sustained drug release for 10 h. In vivo experiments showed their efficiency and make this DDS a promising strategy for aseptic and accurate dose delivery of ophthalmic drugs.

An interesting approach based on a combination of cell and drug delivery for the treatment of Huntington's disease (HD) was proposed by André et al. [26]. The authors used laminin-coated PLGA nanoparticles to transport brain-derived neurotrophic factor (BDNF). The nanoparticles/cell complexes were evaluated in an ex vivo model of HD. Promising results were obtained by the researchers, encouraging them to go further in their research with this system.

Two works lead by Roing and Wacker present new theranostic PLGA-based nanoparticles. In the first one, biodegradable and photoluminescent polyester (BPLP) with PLGA polymer was used to fabricate biocompatible photoluminescent nanocapsules [27]. Additionally, superparamagnetic iron oxide nanoparticles (SPIONs) were incorporated into the polymeric shell to transform the particles into a magnetic resonance/photoluminescence dual-model imaging theranostic platform. The particles demonstrated good uptake and biocompatibility with hCMEC/D3 endothelial cells. In the second study, three different technologies for the encapsulation of sorafenib into PLGA and PLGA–PEG copolymers were adopted [28]. Those nanoparticles presented size ranges between 220 and 240 nm. In order to transform those nanoparticles in a theranostic medicine, gadolinium complexes were covalently attached to the nanoparticles' surface. That way, the nanoparticles could be located using magnetic resonance imaging.

PLGA toxicity was investigated by Bakhaidar et al. [29]. Here, the researchers studied the impact of size-selected PLGA–PEG nanoparticles on platelet activation and aggregation. The results demonstrated that nanoparticles of all sizes are associated with the surface of platelets leading to possible internalization. Furthermore, the NP–platelet interaction proved to not conduct platelet aggregation, making these PLGA nanoparticles promising delivery systems for targeted drug delivery to platelets.

Another relevant study was performed by Operti et al., who used microfluidics technology as a tool to manufacture particles in a highly controllable way [30]. In their study, they produced PLGA particles at diameters ranging from sub-micron to micron using a single microfluidics device. Through modification of flow and formulation parameters, the nanoparticle size changed substantially. Furthermore, in this study, the researchers proved how the particle size influences the release characteristics, cellular uptake, and in vivo clearance of these particles.

Finally, a research study regarding the importance of new techniques to characterize PLGA nanoparticles was included in this special edition. Shmool et al. investigated the dynamics of PLGA microspheres prepared by freeze-drying [31]. The water-oil-water (w/o/w) double-emulsion technique was selected for the production of the microspheres. Their molecular mobility at lower temperatures, leading to the glass transition temperature, using temperature-variable terahertz time-domain spectroscopy (THz-TDS), was evaluated. THz-TDS records show distinct transition processes, one in the range of 167–219 K, associated with local motions, and the other in the range of 313–330 K, associated with large-scale motions.

The papers presented in this Special Issue represent a small part of the research that is ongoing in the field of PLGA nanocarriers all over the world. The huge potential of PLGA nanoparticles make them a promising drug delivery system with outstanding properties and with much more potential for exploring in the coming years. With this Special Issue, the editors expect that the readers from the field find it stimulating and contributing more ideas or methodologies for their future work.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Design and In Vivo Pharmacokinetic Evaluation of Triamcinolone Acetonide Microcrystals-Loaded PLGA Microsphere for Increased Drug Retention in Knees after Intra-Articular Injection**

**Myoung Jin Ho 1, Hoe Taek Jeong 1, Sung Hyun Im 1, Hyung Tae Kim 1, Jeong Eun Lee 1, Jun Soo Park 1, Ha Ra Cho 1, Dong Yoon Kim 1, Young Wook Choi 2, Joon Woo Lee 3, Yong Seok Choi 1,\* and Myung Joo Kang 1,\***


Received: 22 July 2019; Accepted: 12 August 2019; Published: 19 August 2019

**Abstract:** A novel polymeric microsphere (MS) containing micronized triamcinolone acetonide (TA) in a crystalline state was structured to provide extended drug retention in joints after intra-articular (IA) injection. Microcrystals with a median diameter of 1.7 μm were prepared by ultra-sonication method, and incorporated into poly(lactic-*co*-glycolic acid)/poly(lactic acid) (PLGA/PLA) MSs using spray-drying technique. Cross-sectional observation and X-ray diffraction analysis showed that drug microcrystals were evenly embedded in the MSs, with a distinctive crystalline nature of TA. In vitro drug release from the novel MSs was markedly decelerated compared to those from the marketed crystalline suspension (Triam inj.®), or even 7.2 μm-sized TA crystals-loaded MSs. The novel system offered prolonged drug retention in rat joints, providing quantifiable TA remains over 28 days. Whereas, over 95% of IA TA was removed from joints within seven days, after injection of the marketed product. Systemic exposure of the steroidal compound was drastically decreased with the MSs, with <50% systemic exposure compared to that with the marketed product. The novel MS was physicochemically stable, with no changes in drug crystallinity and release profile over 12 months. Therefore, the TA microcrystals-loaded MS is expected to be beneficial in patients especially with osteoarthritis, with reduced IA dosing frequency.

**Keywords:** triamcinolone acetonide; microcrystal; PLGA microsphere; local delivery; spray-drying technique; intra-articular injection; joint retention; systemic exposure

#### **1. Introduction**

Intra-articular (IA) injection of corticosteroids, such as triamcinolone acetonide (TA) crystalline suspension, is commonly recommended to alleviate pain and inflammation in knee joints [1,2]. Marketed injectable TA suspensions are intended to be slowly dissolved in the synovial fluid, and the glucocorticoid molecules steadily bind to and activate the glucocorticoid receptors, obstructing the production of inflammation mediators, including prostaglandins, leukotrienes, and pro-inflammatory cytokines [3]. Nevertheless, the analgesic and/or anti-inflammatory effect of the TA crystalline suspension are reported to be weakened within two weeks following IA injection because of a rapid efflux of the drug from the arthritic joint [4–6]. As the synovial lining is ultra-structured with permeable

intercellular gaps measuring 0.1–5.5 μm, therapeutic agents injected via the IA route tend to easily escape from the joint [7,8]. Moreover, the steroidal compound is quite soluble in aqueous media (21 μg/mL in phosphate buffered saline at 25 ◦C) [9], and TA crystals that have decreased to a few microns in size might be readily translocated into systemic circulation. In vitro release experiments showed that the marketed TA crystalline suspension was completely dissolved within 2 h [10].

Several pharmaceutical approaches, including hydrogels, liposomes, nanoparticles, and microparticles (MPs), have been explored to prolong the retention time of steroidal compounds in the synovial tissues, minimizing systemic exposure following IA injection [2,10–15]. When considering the leaky structure of the synovium, one of the sound strategies for localized delivery to the synovial tissues is administering the therapeutic agents to IA after entrapping in a micro-sized carrier system. Biocompatible and biodegradable polymeric MPs generally larger than 10 μm have been reported to be effective for remaining in the synovial cavity and providing a sustained-release profile in the joint [2,9,16]. Actually, a single IA injection of the PLGA MPs was clinically demonstrated to be effective in providing an extended retention time of the corticosteroid in joints while reducing the drug distribution in the bloodstream [17].

We previously formulated a PLGA MS system containing TA as in a stable crystalline form, to achieve the sustained-release profile in joints after IA injection [18]. Compared to the polymeric MS containing TA in an amorphous state, the TA crystals-loaded MSs prepared by layering the suspended drug crystals with PLGA polymer exhibited excellent physicochemical stability under storage condition (25 ◦C/60% R.H.) in terms of drug crystallinity and drug content in MSs. However, the MS system containing TA crystals with a median diameter of 7 μm could not effectively retard drug release, as the coating layer on the irregular TA crystals was unfair, with insufficient coating thickness; the extent of drug released from the MSs reached 75% for 12 h under sink condition. Thus, we assumed that the additional micronization process of the TA crystals might be beneficial, prior to encapsulation process, to obtain satisfactory coating thickness on the TA crystals, providing sustained-release profile in joints.

Herein, the goals of this study were to construct uniform TA microcrystals and embed the microcrystals in the polymeric MS, providing a prolonged retention profile in the joint following IA injection. The uniform drug microcrystals were prepared by an ultra-sonication method and then encapsulated into the polymeric MS by a spray-drying technique. The physicochemical characteristics of the microcrystals-loaded MSs were evaluated in terms of outer and inner structures, particle size, the drug loading amount, and loading efficiency. In vitro drug release patterns from the MSs were regulated by adjusting the ratio of PLGA to PLA polymers and the ratio of drug to the polymer. The in vivo concentration profile of TA in the plasma and joint tissue following IA injection of the microcrystals-loaded MSs were comparatively evaluated with those of the marketed TA crystalline suspension in rats.

#### **2. Materials and Methods**

#### *2.1. Materials*

TA powder and Triam inj.® were kindly provided by Shinpoong Pharmaceutical Co. (Seoul, Korea). PLGA polymer with a lactide/glycolide ratio of 50:50 (5050DLG 4A, molecular weight 38,000–54,000 kDa) and PLA polymer (R203H, 18,000–24,000 kDa) were purchased from Lakeshore Biomaterials (Birmingham, AL, USA). Polysorbate 20, polysorbate 80, Sorbian monolaurate (Span 20), poly(ethyleneglycol) 4000 (PEG 4000), polyethylene-polypropylene glycol 188 (Poloxamer 188), cholesterol, benzalkonium chloride (BKC), sodium lauryl sulfate (SLS), and phosphate buffered saline tablets were purchased from Sigma Chemical Co. (St. Louis, MO, USA). Lecithin (l-α-phosphatidylcholine) was obtained from Avanti Polar Lipids, Inc. (Alabaster, AL, USA). Gelatin and glycerol were provided from TCI Chemicals, Co. (Tokyo, Japan). Acetonitrile (ACN), ethyl alcohol, and methanol of HPLC grade were obtained from J.T. Baker (Phillipsburg, NJ, USA). All other reagents were analytical grade.

#### *2.2. Preparation of TA Microcrystals Using an Ultra-Sonication Method*

Drug microcrystal suspension was fabricated using an ultra-sonication method previously reported with slight modifications [19]. Different kinds of stabilizers (polysorbate 20, polysorbate 80, span 20, PEG 4000, poloxamer 188, cholesterol, and BKC) were dissolved in ACN in the concentration ranges from 0.05 to 0.5 *w*/*v* % as shown is Table 1. TA powder (100 mg) was then added to the solution and vigorously vortexed for 5 min to disperse the drug powder homogeneously. The ultra-sonicator (Model Vibracell VC-505, Sonics and Materials Inc., Newtown, NC, USA) equipped with a 1/2-inch (13 mm) probe was placed into the TA suspension and it was sonicated for 3 min at 40% amplitude with 3 s pulses (on/off alteration). To prevent temperature elevation, the samples were located inside an ice bath during the sonication procedure. Prepared TA microcrystal suspensions were then stored at room temperature for further experiments.

#### *2.3. Preparation of TA Microcrystals-Loaded MSs Using a Spray-Drying Technique*

TA microcrystals-loaded MSs were fabricated by a spray-drying technique with a Buchi mini-spray dryer (Model B-290, Buchi Labortechnik AG, Flawil, Switzerland). The feeding solution was prepared by subsequently dissolving PLGA/PLA polymers and lecithin into the BKC-stabilized TA microcrystal suspension. The composition of each microcrystals-embedded MSs is represented in Table 2. The feeding solution was then pumped into the spray dryer nozzle at a feeding rate of 3 mL/min and a stirring rate of 250 rpm. The inlet and outlet temperatures were set to 70 ◦C and 45 ◦C, respectively, to evaporate the organic solvent. The atomizing air flow was 246 L/h and the aspirator capacity was 100%. Prepared microcrystals-loaded MS powders were collected and stored in a desiccator (Model OH-3S, As-one, Seoul, Korea) at 25 ◦C for 24 h to remove the residual organic solvent.

#### *2.4. Morphological Features of TA Microcrystals and Microcrystals-Loaded MSs*

#### 2.4.1. Appearance f TA Microcrystals and Microcrystals-Loaded MSs

Morphological features of raw material, TA microcrystals, and polymeric MSs were observed by SEM (Model Sigma 500, Carl Zeiss, Oberkochen, Germany). Drug powder and MSs samples were placed on a carbon tape and fixed onto an aluminum stub. The TA microcrystal suspension was dropwise loaded on the carbon tape and then dried for 6 h at room temperature to remove the aqueous vehicle. The platinum coating procedure was then conducted using an automatic sputter coater (Model 108Auto, Cressington, UK) at 15 mA. Appearance of samples was scrutinized by an electron microscope at an accelerated voltage of 15 kV.

#### 2.4.2. Cross-Sectional Image of TA Microcrystals-Loaded MSs

The internal structure of the microcrystals-loaded MSs was scrutinized by SEM after fixing the MSs in the gelatin blocks. At first, the gelatin medium was prepared by dissolving gelatin (20 *w*/*v* %) and glycerin (5 *w*/*v* %) in distilled water [20]. Approximately 40 mg of the MSs powder was dispersed in the 3 mL of gelatin medium inside the polystyrene disposable base mold (Tissue-Tek®, 15 <sup>×</sup> <sup>15</sup> <sup>×</sup> 5 mm) at 37 ◦C. The mold was then placed in a deep freezer maintained at −70 ◦C for 12 h. The frozen gelatin block was mounted to the cryostat stub (Model CM3050S, Leica Microsystems, Wetzlar, Germany) using an optimum cutting temperature compound (Sakura Finetechnical Co., Ltd., Tokyo, Japan). The MS-loaded block was then sectioned at a thickness of 20 μm at −20 ◦C and was immediately placed on the double-sided carbon tape. Samples were defrosted for 1 h at room temperature and coating and observation procedures were conducted using the same method as described above.

#### 2.4.3. Hyperspectral Mapping Images of TA Microcrystals-Loaded MSs

A hyperspectral microscopy imaging system (Model CytoViva®, Cytoviva Inc., Auburn, AL, USA) was employed to visualize TA microcrystals inside the MSs. The Cytoviva® system included a BX-41 microscope (Olympus Corporation, Tokyo, Japan), a visible-near infrared hyperspectral imaging system, dual fluorescence module, and high-resolution adaptor. Approximately 10 μL of TA microcrystal suspension, blank MSs, and TA microcrystals-loaded MSs suspended in 1 mL of 0.5 *w*/*v* % polysorbate 80 solution was dropped onto a cover glass and the hyperspectral spectra were analyzed. The mapping process was performed on the TA microcrystals-loaded MS image, with acquired spectra of TA microcrystals and blank MSs (HyperVisual Software ENVI 4.8, ITT Visual Information Solutions, Boulder, CO, USA). The spectra corresponding to TA microcrystals and the blank MSs were expressed as red and yellow, respectively, in the hyperspectral image of TA microcrystals-loaded MSs.

#### *2.5. Physicochemical Characterization TA Microcrystals and Microcrystals-Loaded MSs*

#### 2.5.1. Crystallinity Analysis

The crystalline state of TA powder, TA microcrystals, blank MSs, and microcrystals-loaded MSs was analyzed using an X-ray diffractometer (XRD, Model Ultima IV, Rigaku, Tokyo, Japan) at 25 ◦C. For the TA microcrystal suspension, the aqueous vehicle was removed by centrifuging the suspension at 3500 g for 10 min, and subsequently, oven drying at 40 ◦C for 12 h. Each sample was put on the glass sample plate and the diffraction pattern over a 2θ range of 5–35◦ was determined using a step size of 0.02◦. Voltage, current, and scan speed were set to 40 kV, 30 mA, and 1 s/step, respectively.

#### 2.5.2. Particle Size Analysis

Size distributions of TA microcrystals and MSs were determined by Mastersizer MS 2000 (Malvern Instruments Ltd., Worcestershire, UK) equipped with a Hydro 2000 S automatic dispersion unit. Prior to analysis, MSs powder was suspended in the aqueous medium consisting of 1 *w*/*v* % PEG 4000 and 0.5 *w*/*v* % polysorbate 20. The suspended samples were then dropwise added to an automatic dispersion unit to obtain a 10–15% range of obscuration. Sample and background measurement times were set to 5 s and 10 s, respectively, and 5 runs were conducted for each measurement. Mie theory was applied to calculate the size distribution by volume with the refractive index value of 1.52. The resultant particle sizes of the three batches were averaged and presented as mean ± standard deviation (SD) (*n* = 3). The d0.5, d0.9, and d0.1 indicated the median value defined as the diameter where 50%, 90%, and 10% of the population were below this value, respectively. SPAN value was an indicator representing the homogeneity of the particle size and was calculated by dividing the difference of d0.9 and d0.1 by d0.5 [21].

#### 2.5.3. Determination of Loading Amount and Efficiency of TA Microcrystals in MSs

To dissolve TA microcrystals-loaded MSs, 10 mg of MSs were added to 1 mL of dimethyl sulfoxide and then sonicated with a bath-type sonicator (Model 5510E-DTH, Bransonic, USA) for 10 min. The opaque solution was diluted with ACN and distilled water mixture (3:2 *v*/*v*) and was subsequently centrifuged at 16,000 *g* for 10 min to remove the precipitates. The concentration of TA in the supernatant was determined by a Waters HPLC system (Waters Corporation, Milford MA, USA) comprised of a pump (Model 515), auto sampler (Model 717 plus), UV detector (Model 486), and equipped with a Capcell Pak C18 column (150 mm × 2.0 mm, 3 μm, Shiseido, Tokyo, Japan). The mobile phase consisted of ACN and distilled water at a volume ratio of 3:2 and was eluted with a flow rate of 1.0 mL/min. The detection wavelength was set to 254 nm. The calibration curve of TA was linear in the concentration range of 1–100 μg/mL, with r<sup>2</sup> values of 0.999. The drug loading amount and loading efficiency were calculated as follows [22]:

$$\text{Drug loading amount} = \text{WL/WT},\tag{1}$$

Drug loading efficiency (%) = (WL/WF) × 100, (2)

where WL, WT and WF represent the weight of TA in microcrystals-loaded MSs (mg), total weight of microcrystals-loaded MSs (mg) and feeding weight of TA (mg).

#### *2.6. In Vitro Drug Release Profiles and Morphological Changes of Microcrystals-Loaded MSs*

In vitro release profiles of TA from the novel MSs were comparatively evaluated with that of a marketed product under accelerated test conditions (45 ± 0.5 ◦C). To guarantee sink condition during the experiment, 0.5 *w*/*v* % of SLS and 0.05 *w*/*v* % of poloxamer 188, were added to 10 mM phosphate buffered saline (pH 7.4). MSs or the marketed product (Triam inj.®, TA 40 mg/mL) containing 20 mg of TA were immersed into 200 mL dissolution medium maintained at 45 ± 0.5 ◦C and then shaken with an agitation speed of 100 rpm. At predetermined intervals, 1 mL of the release medium was withdrawn and centrifuged at 16,000 *g* for 10 min. The supernatant was diluted two-fold with the mobile phase and TA concentration in the aliquot was determined by HPLC as described above. The equivalent volume of fresh pre-warmed dissolution medium was replenished to maintain a constant medium volume.

The morphological changes of the novel MSs during the in vitro release test were scrutinized by SEM. At predetermined intervals, MSs prepared with the PLGA:PLA ratio of 4:0 (F1), 1:3 (F4), and 0:4 (F5) were withdrawn and centrifuged at 900 *g*. The pelletized MSs were stored at −70 ◦C for 24 h and lyophilized for 24 h. The appearance of the lyophilized MS samples was observed by SEM with the same procedure described earlier.

#### *2.7. In Vivo Systemic Exposure and Joint Retention of TA after IA Injection in Rats*

#### 2.7.1. Animals and Experimental Protocols

In Vivo pharmacokinetic studies were performed after approval from the Institutional Animal Care and Use Committee (IACUC) of Seoul National University Bundang Hospital (approval number: BA1608-206/050-01, date of approval: August 9, 2016). Six-week-old male Sprague-Dawley rats (250 ± 20 g) were acquired from Samtako (Kyungki-do, Korea). Four or five rats were housed in each cage and kept in a temperature- and relative humidity-controlled room (23 ± 1 ◦C and 50 ± 5%, respectively) with a 12-h light-dark cycle. During the acclimatization period, rats were allowed free access to tap water and standardized chow.

After at least three days of the acclimatization period, rats were divided into three groups (*n* = 9 per group) by a stratified randomization scheme for similar body weights groups. The hair on both hind knee joints was removed using hair removal cream. Prior to IA injection, spray-dried MS (F4 and F8) were re-dispersed in the sterile diluent composed of 0.66 *w*/*v* % sodium chloride, 0.63 *w*/*v* % carboxymethylcellulose sodium, and 0.04 *w*/*v* % polysorbate 80 at the same drug concentration (2.5 mg/mL as TA). Each group received 50 μL of the marketed product, F4, and F8, respectively, using an insulin syringe (31 G) in both knee joints, to administer 125 μg of TA per knee. At the predetermined time, blood samples (approximately 0.2 mL) were collected from the submandibular vein using a 26 G heparinized syringe. Blood samples were centrifuged at 16,000 *g* for 10 min. The obtained plasma samples were then stored at −70 ◦C until being analyzed by LC-MS/MS assay.

Apart from the systemic exposure evaluation, knee samples were collected to estimate the level of TA in joint tissues. After 3, 7, 21, 28, and 42 days of the IA injection in both knees, two animals from each group were sacrificed, and both knees were removed using bone cutters. After removing any residual substances and adhered tissues, knees were accurately weighed and stored at −70 ◦C until LC-MS/MS analysis.

#### 2.7.2. LC-MS/MS Analysis of TA Concentration in Plasma and Knee Tissues

The TA concentrations in rat plasma or joint tissue were determined using the LC–MS/MS assay previously reported [23]. In brief, thawed plasma (100 μL) was mixed with 900 μL of methanol and vigorously vortexed for 10 min, to precipitate protein. After centrifuging at 16,000 g, a supernatant

(10 μL) was analyzed through an LC-MS/MS system (Model LC-20 Prominence HPLC, Shimadzu and Model API 2000, AB/SCIEX, Foster City, CA, USA). In the case of articular samples, the frozen knee tissues were thawed and immersed in 2 mL of ACN and shaken overnight to extract TA from the tissue. The extracted solution was centrifuged at 16,000 *g* for 5 min and the supernatant was injected into the LC-MS/MS system. The transitions of 435.1/415.0/15 precursor ion (*m*/*z*)/product ion (*m*/*z*)/collision energy (V) were then monitored for TA. Data acquisition/analyses were conducted using Analyst® version 1.5.2 software (ABSciex, Concord, ON, Canada). The assay was validated thoroughly and showed acceptable precision and accuracy, with a lower limit of quantification of 0.2 ng/mL in both rat plasma and knee tissue extract.

#### 2.7.3. Pharmacokinetic Parameters from TA Concentration Profile in Plasma

Pharmacokinetic parameters such as area under the plasma concentration versus time curve (AUC0–7days), maximum plasma concentration (Cmax), time needed to reach the maximum plasma concentration (Tmax), and terminal half-life (T1/2) in plasma were calculated using the linear trapezoidal rule in the BA Calc 2007 pharmacokinetic analysis program (Korea Food & Drug Administration, Seoul, Korea).

#### *2.8. Physicochemical Stability of TA Microcrystals-Loaded MSs*

The long-term storage stability of the novel MSs was evaluated in terms of drug crystallinity, drug content, and in vitro release profile. TA microcrystals-loaded MS (F4) power was placed into the scintillation vial and was stored in the chamber maintained at 25 ◦C and 60% R.H. After 12 months of storage, the drug crystallinity, drug content, and in vitro release profile were evaluated with the same method as previously described.

#### *2.9. Statistical Analysis*

Each experiment was performed at least thrice and the data are presented as the mean ± SD. Statistical significance was determined using a one-way analysis of variance (ANOVA) test and was considered to be significant at *p* < 0.05 unless otherwise indicated.

#### **3. Results and Discussion**

#### *3.1. Formulation and Physical Characteristics of TA Microcrystals*

Various stabilizers were screened to micronize TA powder in the organic solvent using a probe type ultra-sonicator (Table 1). ACN was employed as the vehicle as it exhibited low solubility for the steroidal compound (<1 mg/mL), and high solvation capacity for PLGA and PLA polymers [18]. When the surface stabilizer was not included in the vehicle, TA powder was not uniformly dispersed in the medium after the homogenization process, rather forming large precipitates. The addition of steric stabilizers, such as polysorbate 20, polysorbate 80, PEG 4000, poloxamer 188, and cholesterol could not provide a uniform dispersion of the split TA microcrystals, forming drug aggregates within 24 h. On the other hand, when BKC was included in the organic vehicle at a concentration of 0.1 to 0.5 *w*/*v* %, TA microcrystals with a median size below 2.1 μm were shaped with re-dispersibility in the organic solvent (Table 1 and Figure 1A). When the concentration of the cationic surfactant was less than 0.05 *w*/*v* %, it could not afford the re-dispersibility of TA microcrystals. Thus, BKC at the concentration of 0.1 *w*/*v* % was employed for further preparation of TA microcrystal suspension in ACN.

The morphological feature of the TA microcrystals stabilized by 0.1 *w*/*v* % BKC was observed by FE-SEM. TA raw material showed characteristic crystal forms, such as hexahedron, octahedron, and dodecahedrons, with different sizes in the range from 2 to 20 μm (Figure 1C). Whereas, the crystal size was markedly decreased to 1–3 μm by the ultra-sonication process (Figure 1D), coinciding with the crystal size as determined by Mastersizer (d0.5, 1.7 μm). In spite of crystal size reduction, no noticeable change to the shape was observed in the TA microcrystals. The crystalline state of TA microcrystals

was further evaluated by comparing the X-ray diffraction spectrum of TA microcrystals with that of the raw material (Figure 1B). The spectrum of TA microcrystals was identical to that of drug powder, exhibiting distinctive diffraction peaks at 2θ equal to 9.9◦, 14.5◦, 17.6◦, and 24.7◦. On the other hand, the cationic surfactant showed no distinctive diffraction peaks over the 2θ range of 5–35◦. Taken together, we concluded that TA powder was effectively micronized to 1–3 μm, with no crystalline changes during the ultra-sonication process.

**Table 1.** Effects of kinds of stabilizers on size, homogeneity, and dispersibility of TA microcrystals in ACN.


<sup>1</sup> Weight per volume concentration in ACN. <sup>2</sup> Expressed as mean ± SD (*n* = 3). <sup>3</sup> Indicates the volume weighted diameter below which 50% of the total particle. <sup>4</sup> Calculated by dividing the difference between d0.9 and d0.1 by d0.5. d0.9 and d0.1 indicate the volume weighted diameters below which 90% and 10% of the total particle, respectively. <sup>5</sup> Visually evaluated after 24 h storage at room temperature. <sup>6</sup> Indicates without stabilizer.

**Figure 1.** Morphological and physical characteristics of TA microcrystals. (**A**) Size distribution of the drug microcrystals stabilized by BKC, (**B**) XRD patterns of (a) TA raw material, (b) BKC powder, and (c) TA microcrystals, SEM images of (**C**) TA raw material, and (**D**) TA microcrystals stabilized by BKC at the concentration of 0.1 *w*/*v* %.

#### *3.2. Formulation and Physical Characteristics of TA Microcrystals-Loaded MSs*

Various TA microcrystals (1.7 μm median size) or intact TA powder (7.2 μm)-embedded MSs were fabricated using the spray-drying technique, and these particle size, homogeneity, drug loading amount, and loading efficiency are represented in Table 2. The median particle size of MSs prepared ranged from 15.8 to 18.8 μm, with a narrow size distribution possessing a SPAN value below 2.1. The formulation variables, such as TA crystal size, the ratio of PLGA and PLA polymers, and the ratio of drug to polymer, did not cause marked differences in size and homogeneity of TA-loaded MSs (Table 2). The particle size of the novel MS was considered to be suitable for IA prolonged delivery, preventing the trans-synovial efflux of injected TA microcrystals [17,24,25]. There was also no remarkable difference in the drug loading efficiency in the polymeric MSs between the formulas, exhibiting more than 90% drug loading efficiency in all formulations. The absence of the external phase during the spray-drying process might prevent distribution and/or diffusion of TA microcrystals during the external phase, and thus promote TA crystals to be located in the polymeric matrix after solvent evaporation, irrespective of composition variables. On the other hand, the loading amount of TA in MS was adjusted from 0.09 to 0.31 *w*/*w*, by controlling the drug to polymer weight ratio from 1:2 to 1:10.


**Table 2.** Compositions and physicochemical characteristics of TA microcrystals-loaded MSs.

<sup>1</sup> Presented as d0.5 value; the volume weighted diameter below 50% of the total particle. <sup>2</sup> Calculated by dividing the difference between d0.9 and d0.1 by d0.5: d0.9 and d0.1 by d0.5: d0.9 and d0.1 are the volume weighted diameters below 90% and 10% of the total particle, respectively. <sup>3</sup> Calculated by dividing the weight of TA in microcrystals-loaded MSs by total weight of microcrystals-loaded MSs (mg). <sup>4</sup> Expressed as the percentage (%) after dividing the weight of TA loaded in MSs by total fed weight of TA (mg). <sup>5</sup> Expressed as mean ± SD (*<sup>n</sup>* <sup>=</sup> 3); Note: Lecithin was included in all formulations at the weight ratio of 5 *w*/*w* % to the total amount of polymers.

The novel TAs-loaded MSs were further characterized in terms of outer and internal structures and drug crystallinity in MS (Figure 2). The MSs (F4) prepared by the spray-drying technique was highly spherical, with a smooth and homogeneous surface (Figure 2A). In the cross-sectional image, the microcrystals showed different textures from the polymeric matrix and were found to be uniformly embedded in the polymeric matrix (Figure 2B). The number of microcrystals loaded per MS was elucidated by translating the loading amount into the number of MS and TA microcrystals. In the process of converting the weight to a number, the volume of single MS and TA microcrystal was calculated with the assumption that the shape of the MS and microcrystal were spherical and cubic, respectively, with both having a density of 1.0. In the MS formulations prepared with the drug to polymer ratio of 1:2 (F6), 1:3 (F7), 1:5 (F4), and 1:10 (F8), the number of microcrystals embedded in each MS was calculated as 215, 94, 82, and 50, respectively. In the hyperspectral image (Figure 2C), TA microcrystals (red color) were observed to be principally located inside the PLGA/PLA MS (yellow color). However, individual microcrystals were not separately spotted in the image, probably because of the low resolution of Cytoviva®. The characteristic peak of TA microcrystals was identically detected in the microcrystals-load MS (Figure 2D), denoting that TA microcrystals stabilized by BKC were successfully incorporated in the MSs, with no crystalline changes during the fabrication process.

**Figure 2.** Morphological and physical characteristics of the microcrystals-loaded MSs. Representative micrograph of (**A**) intact and (**B**) cross-sectioned microcrystals-loaded MS (F4), (**C**) hyperspectral image of microcrystals-loaded MSs, and (**D**) XRD patterns of (a) TA microcrystals, (b) blank MS, (c) F4 MS (drug:polymer = 1:5), and (d) F8 MS (1:10); Notes: Inset in (**B**) is ×5000 magnified images and the arrow points to the TA crystal surrounded by the polymeric matrices. In the hyperspectral image (**C**), PLGA/PLA polymers and TA microcrystals are colored as yellow and red, respectively.

#### *3.3. In Vitro Drug Release and Degradation Profiles of TA Microcrystals-Loaded MSs*

In Vitro drug release profiles from the marketed product, the drug powder- or micronized TA crystals-loaded MSs were evaluated under accelerated dissolution conditions. Although the synovial fluid does not assure the sink condition for the drug dissolution, the in vitro release test under sink conditions was favored for quicker comparison between the formulations. Moreover, as the drug release profiles from PLGA/PLA MSs could be retarded from days to months at body temperature (37 ◦C), the liberation pattern of the steroidal compound from the MSs was further facilitated by elevating the temperature of the dissolution media (45 ◦C), promoting the degradation and/or hydrolysis of the biodegradable polymers [26–28]. Actually, the accelerated test at high temperature was reported to be beneficial for faster comparison of release behavior between MS formulas, with high correlation with that obtained at 37 ◦C [27]. Shen and Burgess (2012) revealed that the time required to reach 100% drug release from the MSs prepared with PLGA polymer with the glass transition temperature (Tg) of 44–48 ◦C was determined to be 10, 5, 3, and 1.3 days, respectively, at temperatures of 45, 50, 53, and 60 ◦C. Herein, the temperature of the dissolution medium was set to 45 ◦C, which did not exceed the Tg values of both polymers (46–52 ◦C). When the polymeric MSs were exposed to the medium at temperatures above the Tg of the polymer, the drug diffusion coefficient was drastically increased [29], diminishing the difference between the release profiles between the polymeric particulates.

Under the accelerated condition, the marketed product containing 13 μm-sized TA crystals stabilized by polysorbate 20 and sodium carboxymethyl cellulose was rapidly liquefied in the aqueous medium, showing complete drug release within 90 min (Figure 3A). It coincided with a previous report that showed that the TA crystal suspension was readily dissolved in phosphate buffered saline within 2h[10]. The drug release from the MS with 7.2-μm-sized TA crystals (F0) was not markedly retarded compared to the intact drug crystal, releasing over 95% of TA within 3 h. The incomplete and/or erratic coating thickness of the polymeric layer on the TA crystals probably could not effectively restrain the dissolution and diffusion procedures of the TA crystals into the aqueous media. On the other hand, drug release from the MSs containing smaller TA crystals (median size of 1.7 μm) was markedly impeded compared to the marketed product or 7.2-μm-sized TA crystal-loaded MS, especially as the ratio of PLA increased in the MS (Figure 3A).

**Figure 3.** In Vitro release profile of TA from the novel MSs under accelerated conditions (45 ◦C). (**A**) Drug release profile from the marketed product (×), 7.2 μm-sized TA crystals-loaded MS (F0, –), and 1.7 μm-sized TA crystals-loaded MS prepared with different PLGA:PLA ratios; 4:0 (F1, -), 3:1 (F2, ), 2:2 (F3, ), 1:3 (F4, ), and 0:4 (F5, -) and (**B**) drug release profile from MSs prepared with different drug:polymer ratios; 1:2 (F6, ), 1:3 (F7, -), 1:5 (F4, ), and 1:10 (F8, ); Notes: The inset graph (**A**) is the magnified release profile from the marketed product and 7.2 μm-sized TA crystals-loaded MS. Data are expressed as the mean value (*n* = 3) and error bars are SDs.

As the glycolic acid has faster hydration/swelling behavior compared to lactic acid [30], MSs prepared with over 50% of PLGA polymers (F1, F2, and F3) showed higher burst release, with over 60% of drug released within 2 days. After the initial burst release, the extent of TA liberated from the F1, F2, and F3 polymers continuously rose, exhibiting over 90% release after 7 days under sink condition. On the other hand, the F4 formula with a PLGA:PLA ratio of 1:3, exhibited a more protracted release profile compared to that of F1, F2, and F3, exhibiting a linear release pattern for 21 days after a 53% initial release in the first 2 days. F5 (PLGA:PLA ratio of 0:4) showed the slowest release profile, displaying only 52% of the accumulated drug release after 21 days. Although there was marked difference in release profile depending on the ratio of PLGA and PLA polymer, the drug release pattern from novel drug microcrystals-loaded MSs were characterized by initial burst release and subsequent slow release profile, which is consistent with the typical release pattern of PLGA/PLA based MPs previous reported [31,32]. In the early phase, TA microcrystals located on or inner compartment near the surface of the MS might be rapidly dissolved by surrounding and/or penetrated aqueous media, and released from polymeric matrix mainly by diffusion mechanism. After initial burst release, the drug release rate tended to be retarded, due to the increased diffusion distance. Afterward, and the remaining steroidal compound in the MSs might be liberated by polymeric degradation and erosion and/or collapse of polymeric MSs.

Different drug release patterns depended on the PLGA:PLA ratio were highly consistent by the morphological changes of TA microcrystals-loaded MSs. As shown in Figure 4, because of the rapid swelling and hydrolysis nature of PLGA polymer, the PLGA MSs (F1) began to collapse and was excavated within three days. Thus, the drug microcrystals embedded in the PLGA MS might be readily exposed to aqueous media, and immediately dissolved under sink condition. On the contrary, because of the greater hydrophobicity of the PLA polymer compared to the PLGA polymer, the hydrolytic degradation of PLA MS (F5, PLGA:PLA = 0:4) progressed slowly. When the MSs were scrutinized at 3 and 7 days, fine pores were formed on the roughed surface and the pore size was gradually enlarged

as time elapsed. Nevertheless, the overall globular shape and dimension of MS were retained even at 21 days, supporting the slow and incomplete release profile of TA from the MS (F7). The degradation pattern of F4 prepared with the PLGA:PLA ratio of 1:3 was intermediate between those of MSs prepared with PLGA or PLA polymer F1 and F5, respectively. After surface erosion and pore formation at three days, the MS was then gradually collapsed over 21 days. The drug release rate from F4 was markedly delayed compared to that from F1 but was much faster and higher than that from F5, releasing over 80% of the drug loaded for 21 days. The ratio of PLGA to PLA polymers was fixed to 1:3, expecting the prolonged release pattern for further investigation.

**Figure 4.** Morphological changes of TA microcrystals-loaded MSs prepared with different PLGA:PLA ratios, 0:4 (F1), 1:3 (F4), 0:4 (F5) under the accelerated release conditions (45 ◦C); Note: Scale bars in each image indicate 2.0 μm.

The in vitro release profile of TA from the PLGA/PLA MSs prepared with different drug to polymer ratios was further evaluated. As shown in Figure 3B, the initial drug release from the novel MSs were gradually decreased, as the drug to polymer ratio was increased. When the ratio of drug to polymer was 1:2 (F6) or 1:3 (F7), the percentage of drug released for 24 h had reached approximately 86% and 78%, respectively. On the other hand, in the formulations of the drugs:polymer ratio of 1:5 (F4) and 1:10 (F8), the release of the steroidal compound from the MSs were markedly retarded, exhibiting less than 60% of drug release over 5 days. Drug release from the F4 or F8 was steeped after 7 days, probably because of erosion and/or collapse of the MSs, but prolonged for 21 days. Formulas F4 and F8 were further exploited for in vivo pharmacokinetic study in rats, expecting an extended release profile over one month in the knee joint.

#### *3.4. In Vivo Systemic Exposure and Ioint Retention of TA after IA Injection in Rats*

The systemic exposure and local bioavailability of TA following a single IA injection of the marketed product or the novel MSs (F4 and F8) were evaluated in normal rats. The IA dose of TA treated in all groups was same to 0.25 mg per knee, which was well tolerated in rats [10]. The plasma levels of TA as a function of time following IA injection of the marketed product, F4, and F8 are represented in Figure 5 and the relevant PK parameters are summarized in Table 3. It is recommended that the exposure of the steroidal compound in blood be minimized, as the exogenous corticosteroid can cause Cushing syndrome, incurred impaired wound healing, infection, and muscle weakness [10,33,34]. However, unfortunately, the plasma level of TA was drastically elevated after administration of the marketed product, reaching Cmax value of 218.7 ng/mL after 3.7 h. This rapid redistribution of TA into the bloodstream is in agreement with earlier reports that intra-articularly injected TA crystals were rapidly absorbed, with a Tmax value of 4 h in patients with osteoarthritis [6,35,36]. This rapid drug efflux from the knee joint is also correlated with in vitro release profiles, denoting that TA crystalline suspension injected in the joint might be rapidly dissolved and passed out the gap in the synovial membrane. After reaching a Cmax of 3.7 h, the plasma level of TA sharply decreased below 30 ng/mL after 12 h post-administration of the marketed product.

**Figure 5.** Plasma concentration–time profiles of TA following IA administration of the marketed product, F4 (drug:polymer = 1:5) and F8 (drug:polymer = 1:10) in rats at a dose of 0.5 mg/kg; Note: Each point represents mean ± SD (*n* = 5).

**Table 3.** Pharmacokinetic parameters of TA in plasma following IA administration of the marketed product, F4, and F8 in rats.


<sup>1</sup> Calculated from the plasma concentration–time curve from Tmax to 24 h after IA injection; Notes: Data are expressed as mean ± SD (*n* = 5). Statistical analysis was performed using the one-way ANOVA test; \* significantly different from the marketed product (*p* < 0.05), \*\* significantly different from F4 (*p* < 0.05); Abbreviations: AUC0–7days, area under the plasma concentration–time curve until day 7; Cmax, maximum plasma concentration; Tmax, time to reach maximum plasma concentration; T1/2, elimination half-life of the drug.

In contrast, the systemic exposure of the exogenous corticosteroid was markedly decreased following IA injection of the TA microcrystals-loaded MSs (F4 and F8). The Cmax values of TA following IA injection of F4 and F8 were determined to 75.6 ng/mL and 32.2 ng/mL, respectively, which were only 34% (*p* < 0.05) and 15% (*p* < 0.05) that of the marketed product. Correspondingly, AUC0–7days values in the F4- and F8-treated groups were drastically decreased to less than 54% (*p* < 0.05) and 37% (*p* < 0.05) of that obtained from the marketed product, respectively. These pharmacokinetic data indicated that the novel MSs remarkably lessened the redistribution of dissolved and/or micronized compound into the bloodstream, prolonging the retention time of TA in the knee. Between the two groups treated with MSs prepared with loading amount of 0.16 (F4) and 0.09 (F8), respectively, the drug exposure to blood was much lowered in the F8-treated group, showing 68% and 42% decreased AUC0-7days and Cmax values compared to those obtained from the F4-treated group. This pharmacokinetic tendency is explainable with the in vitro release test results, which revealed that the extent of drug released from the polymeric matrix declined as the drug to polymer ratio increased.

The drug remaining in the joint tissue following IA single injection of each formula was further assessed in normal rats (Figure 6). After the IA injection of the marketed product, the drug concentration in the joint at three days post-dosing was only 5.6 μg/g because TA crystals were quickly effluxed from the joint tissue. The percentages of the drug remaining in the joint tissue at 3 and 7 days were calculated to be only 4.5% and 2.4%, respectively. After 21 days, the drug concentration in the joint tissue was below the limit of detection. This result is in line with a previous report that only two of eight patients with osteoarthritis had quantifiable synovial TA concentration at week 6, following IA injection of the marketed product [37]. On the other hand, the novel MS formulations exhibited a markedly profound and prolonged concentration profile in joint tissue compared to the marketed product, exhibiting quantifiable TA concentration over 28 days. Three days after the single administration of F4 or F8, the drug concentration in joints was determined to be 45 μg/g and 67 μg/g, respectively, which is one-third and one-half of the initial dose. In both MSs-treated groups, the TA concentration in the tissue gradually decreased as time elapsed, but approximately 5% of the initial dose was still detected at 28 days. These findings suggested that the retention time of TA in the joint tissue was extended with the sustained-release pattern of the novel MSs.

**Figure 6.** TA remains in rat joint tissue following IA administration of the marketed product, F4, and F8 at a dose of 125 μg of TA per knee; Notes: Vertical bars represent mean ± SD (*n* = 6). Statistical analysis was performed using the one-way ANOVA test; \* significantly different from the marketed product (*p* < 0.05), \*\* significantly different from F4 (*p* < 0.05). BLOQ value of LC-MS/MS analysis was determined to 0.2 ng/mL.

#### *3.5. Long-Term Stability of TA Microcrystals-Loaded MS*

The physicochemical stability of the novel MS was evaluated after 12 weeks of storage under ambient conditions (25 ◦C, 60% RH). The storage condition of the MSs was set to ambient condition, as there was morphological change above 40 ◦C, due to softening of the polymer over Tg. At first, the drug crystallinity in the MS was assessed using XRD because a change in drug crystalline nature may occur during storage, affecting the drug chemical stability and release pattern from the MS. Under

ambient conditions, the crystallinity of TA microcrystals embedded in the MS was stably maintained over 12 months, with no changes in diffraction pattern (Figure A1). There was also no change in drug content in F8 MS, displaying over 97% drug content after 12 months of storage. The in vitro dissolution pattern was also comparable with that of MSs immediately prepared, exhibiting a sustained-release profile of over 21 days (Figure A1). From these findings, we concluded that the novel MS system was physicochemically stable at least for one year under ambient conditions.

#### **4. Conclusions**

A novel parenteral sustained-release system of TA was successfully prepared by micronizing TA powder into 1.7 μm-sized microcrystals, and subsequently embedding into PLGA/PGA polymeric MSs using a spray-drying technique. TA microcrystals were efficiently entrapped into the polymeric MSs, preserving their distinctive crystalline nature. In vitro drug release from the novel MSs was markedly retarded compared to the marketed product and even 7.2 μm-sized TA crystal-loaded MSs, exhibiting a prolonged release profile over 21 days under accelerated conditions (45 ◦C). In an in vivo pharmacokinetic study in normal rats, the duration that the TA remained in the joint tissue was markedly extended, providing profound drug remains at 28 days following IA single injection. Moreover, TA microcrystals-loaded MSs drastically decreased the systemic exposure of the steroidal compound compared to the marketed product. Thus, the novel IA long-acting system could be a valuable tool, providing both increased drug retention in the knee and diminished systemic exposure of TA following a single administration.

**Author Contributions:** Conceptualization, M.J.H. and M.J.K.; Data curation, M.J.H., H.T.J. and J.S.P.; Formal analysis, H.R.C., D.Y.K. and Y.S.C.; Funding acquisition, M.J.K.; Investigation, S.H.I., H.T.K. and J.S.P.; Methodology, M.J.H., H.T.J. and J.W.L.; Project administration, Y.W.C., Y.S.C. and M.J.K.; Resources, H.T.J., S.H.I., H.T.K., J.E.L. and J.S.P.; Supervision, Y.W.C., Y.S.C. and M.J.K.; Validation, H.R.C., D.Y.K. and Y.S.C.; Visualization, M.J.H.; Writing—original draft, M.J.H.; Writing—review & editing, Y.W.C., J.W.L., Y.S.C. and M.J.K.

**Funding:** This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2016R1C1B1010687).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

**Figure A1.** Physicochemical stability of TA microcrystals-loaded MS (F4) after 12 months of storage under ambient conditions. (**A**) XRD patterns of F4 MS (a) after preparation and (b) after 12 months of storage, (**B**) in vitro release profile of F4 after preparation and after 12 months storage under accelerated conditions; Notes: XRD pattern and release profile of F4 MS after preparation are identical to those depicted in Figures 2D and 3B, respectively. Each point in (**B**) in vitro release profile represents mean ± SD (*n* = 3).

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Formulation, Colloidal Characterization, and In Vitro Biological E**ff**ect of BMP-2 Loaded PLGA Nanoparticles for Bone Regeneration**

**Teresa del Castillo-Santaella 1, Inmaculada Ortega-Oller 2, Miguel Padial-Molina 2, Francisco O'Valle 3, Pablo Galindo-Moreno 2, Ana Belén Jódar-Reyes 1,4 and José Manuel Peula-García 1,5,\***


Received: 20 June 2019; Accepted: 31 July 2019; Published: 3 August 2019

**Abstract:** Nanoparticles (NPs) based on the polymer poly (lactide-co-glycolide) acid (PLGA) have been widely studied in developing delivery systems for drugs and therapeutic biomolecules, due to the biocompatible and biodegradable properties of the PLGA. In this work, a synthesis method for bone morphogenetic protein (BMP-2)-loaded PLGA NPs was developed and optimized, in order to carry out and control the release of BMP-2, based on the double-emulsion (water/oil/water, W/O/W) solvent evaporation technique. The polymeric surfactant Pluronic F68 was used in the synthesis procedure, as it is known to have an effect on the reduction of the size of the NPs, the enhancement of their stability, and the protection of the encapsulated biomolecule. Spherical solid polymeric NPs were synthesized, showing a reproducible multimodal size distribution, with diameters between 100 and 500 nm. This size range appears to allow the protein to act on the cell surface and at the cytoplasm level. The effect of carrying BMP-2 co-adsorbed with bovine serum albumin on the NP surface was analyzed. The colloidal properties of these systems (morphology by SEM, hydrodynamic size, electrophoretic mobility, temporal stability, protein encapsulation, and short-term release profile) were studied. The effect of both BMP2-loaded NPs on the proliferation, migration, and osteogenic differentiation of mesenchymal stromal cells from human alveolar bone (ABSC) was also analyzed in vitro.

**Keywords:** BMP-2; PLGA nanoparticles; Pluronic F68

#### **1. Introduction**

In the context of nanomedicine, tissue regeneration using colloidal micro- and nano-structures having unique size and surface activity has received increasing attention over recent years. Many efforts have been made to improve the engineering of these nano-systems in order to reach a "smart" delivery of bioactive molecules in order to optimize their therapeutic advantages and minimize harmful side effects [1]. With this aim, a broad spectrum of biocompatible nanocarriers has been described, showing properties suitable for different biological and therapeutic applications [2]. Among these varied proposals, polymeric nanosystems represent a major group in which poly lactic-co-glycolic acid (PLGA) is one of the most widely used due to its biocompatibility, biodegradability, and low cytotoxicity, gaining the approval from different drug agencies for human use [3,4].

PLGA-based structures are described as micro- and nanocarriers to deliver a wide variety of active molecules and drugs, synthetic or natural molecules with hydrophilic or hydrophobic properties, and biomolecules from proteins to nucleic acids [5–7]. PLGA micro- and nanosystems can be set up using different formulation techniques, with the possibility of a systemic or local distribution. These systems can be applied not only in tissue regeneration but also in very diverse therapies: Anticancer drug delivery, infections, inflammatory diseases, or gene therapy [3]. Despite this great potential, certain applications, especially in protein encapsulation, are hindered by problems, such as an uncontrolled release profile and protein denaturation [8–11].

The water-in oil-in water (W/O/W) double emulsion method is an "emulsion solvent evaporation" technique frequently used to encapsulate hydrophilic molecules as proteins in PLGA NPs [6,12]. The appropriate choice of organic solvents, the use of polymer-surfactant blends, and the addition of stabilizer-protective agents have proved to be key aspects for optimizing the resulting systems [9,11]. Additionally, a surface specific functionalization can be used to improve their versatility, allowing the chemical surface immobilization of different molecules in order to confer targeting or adhesive properties to these nanocarriers [13].

Within tissue engineering, bone regeneration has a broad range of applications, mostly in the field of dentistry, where PLGA is suggested as a reference polymer to formulate NPs with bone-healing uses [14]. The literature describes the delivery of bioactive molecules, normally growth factors, using polymeric microparticles (MPs) and NPs with PLGA as the main component [13]. Among the bone morphogenetic growth factors, BMP-2 (bone morphogenetic protein 2) has been the most frequently cited, with many examples in which encapsulation or surface adsorption enables adequate entrapment efficiency and diverse release patterns [15–19]. For proteins with a very short half-life, such as BMPs, biodegradable PLGA nanosystems provide protection and optimal dosage for an adequate stimulation of cell differentiation [20,21].

Thus, within this scenario, in the present work, we seek to optimize a nano-particulate system in order to carry out and control the release of BMP-2 using as a starting point the synthesis procedure of a lysozyme-loaded NP system, previously described for the encapsulation of that model protein [11]. Also, to encapsulate BMP-2, we prepared a second system in which this protein was co-adsorbed with bovine serum albumin onto the surface of empty NPs. The size and morphology, the protein encapsulation efficiency, the surface characteristics, and the colloidal and temporal stability were studied to complete the physico-chemical characterization of both NP systems.

The release profile of BMP-2 indicates the potential of a PLGA nanocarrier for bone regeneration and depends heavily on the polymer degradation by hydrolysis [22]. However, over the short term, during which the release does not depend on this chemical degradation, proper control of release is necessary in order to modulate other physical processes. Thus, we focused our release experiments on the short-term using different techniques to compare the two NP samples and establish the corresponding BMP-2 release profiles. Finally, the biological activity (cell migration, proliferation, and osteogenic differentiation) was tested in vitro using mesenchymal stromal cells (MSCs) derived from alveolar bone [23].

#### **2. Materials and Methods**

#### *2.1. Nanoparticle Synthesis*

#### 2.1.1. Formulation

Poly(lactide-co-glycolide) acid (PLGA 50:50) ([C2H2O2]x[C3H4O2]y), x = 50, y = 50 (Resomer® 503H, (Evonik, Essen, Germany), 32–44 kDa was used as the polymer, and polymeric surfactant Pluronic F68 (Poloxamer 188) (Sigma-Aldrich, St. Louis, MO, USA) as the emulsifier. Their structure, based on a poly(ethylene oxide)-block-poly(propylene oxide)-block-poly(ethylene oxide), is expressed as PEOa-PPOb-PEOa with a = 75 and b = 30. Human recombinant bone morphogenetic protein, rhBMP-2 (Sigma-H4791), was used as therapeutic biomolecule. Water was purified in a Milli-Q

Academic Millipore system. A double-emulsion synthesis method was used following a procedure previously described with slight modifications [11]. In this method, 100 mg of PLGA and 3 mg of deoxycholic acid (DC) were dissolved in a tube containing 1 mL of ethyl acetate (EA) and vortexed. In total, 40 μL of a buffered solution at pH 12.8, with or without rhBMP-2 (200 μg/mL), were added and immediately sonicated (Branson Ultrasonics 450 Analog Sonifier) for 1 min (Duty cycle dial: 20%, Output control dial: 4) with the tube surrounded by ice. This primary W/O emulsion was poured into a plastic tube containing 2 mL of a buffered solution (pH 12) of F68 at 1 mg/mL, and vortexing for 30 s. Then, the tube surrounded by ice was sonicated at the maximum amplitude for the micro tip for 1 min (Output control: 7). This second W/O/W emulsion was poured into a glass containing 10 mL of the buffered F68 solution and kept under magnetic stirring for 2 min. The organic solvent was then rapidly extracted by evaporation under vacuum to a final volume of 8 mL. The resulting empty and BMP-2 encapsulated NP systems were named NP and NP-BMP2, respectively. A detailed scheme of the synthesis procedure, with a yield based on the PLGA component always higher than 85%, is shown in Figure S1 of the Supplementary Materials.

#### 2.1.2. Cleaning and Storage

After the organic solvent evaporation, the sample was centrifuged for 10 min at 20 ◦C at 12,000 rpm. The supernatant was filtered using Millipore nanofilters, 0.1 μm for measuring the free non-encapsulated protein. The pellet was then resuspended in phosphate buffer (1.15 mM NaH2PO4), PB, to a final volume of 4 mL and kept refrigerated at 4 ◦C. Under these conditions, the systems kept colloidal stability at least for one month.

#### 2.1.3. Protein Loading and Encapsulation Efficiency

The initial protein loading was optimized for the nanoparticle formulation, preserving the final colloidal stability after the evaporation step and taking into account the amounts shown in the literature for this growth factor when encapsulated inside PLGA NPs [24,25]. Thus, we chose 2 μg as the initial total mass of rhBMP-2, which means a relation of 2 <sup>×</sup> 10−5% *w*/*w* (rhBMP-2/PLGA). The amount of encapsulated rhBMP-2 was calculated by measuring the difference between the initial added amount, and the free non-encapsulated protein present in the supernatant after the cleaning step, which was tested by a specific enzyme-linked immuno-sorbent assay following the instructions of the manufacturer (ELISA, kit RAB0028 from Sigma-Aldrich, St. Louis, MO, USA). Then, protein-encapsulation efficiency (EE) was calculated as follows:

$$EE = \frac{M\_I - M\_F}{M\_I} \times 100$$

where *MI* is the initial total mass of rhBMP-2, and *MF* is the total mass of rhBMP-2 in the aqueous supernatant.

#### 2.1.4. Physical Protein Adsorption

Bovine serum albumin (BSA) and rhBMP-2 were coupled on the empty nanoparticle surface by a physical adsorption method. The appropriate volume of an aqueous protein solution containing 0.5 mg of BSA and 2 μg of rhBMP-2 was mixed with 5 mL of acetate buffer (pH 5) containing empty NPs with 12.5 mg of PLGA. This provided a starting amount of proteins corresponding to 0.04% *w*/*w* (protein/PLGA), while the mass relation between proteins was 0.4 *w*/*w* (rhBMP-2/BSA). This solution was incubated at room temperature for 2 h under mechanical stirring. The nanoparticles were separated from the buffer solution by centrifugation, and after the supernatants were filtered (Millipore nanofilters, 0.1 μm), they were qualitatively analyzed by gel electrophoresis while the protein quantification was made by a bicinchoninic acid protein assay (BCA) (Sigma-Aldrich, St. Louis, MO, USA) for BSA and the specific ELISA for rhBMP-2. The nanoparticle pellet was resuspended in phosphate buffer (pH 7.4) and stored at 4 ◦C. This system was named NP-BSA-BMP2.

#### 2.1.5. Protein Separation by Gel Electrophoresis, SDS-PAGE

The protein-loaded NPs and different supernatants were treated at 90 ◦C for 10 min in the following buffer: 62.5 mM Tris-HCl (pH 6.8 at 25 ◦C), 2% (*w*/*v*) sodium dodecyl sulfate (SDS), 10% glycerol, 0.01% (*w*/*v*) bromophenol blue, 40 mM dithiothreitol (DTT). Samples were then separated by size in porous 12% polyacrylamide gel (1D SDS polyacrylamide gel electrophoresis), under the effect of an electric field. The electrophoresis was run under constant voltage (130 V, 45 min) and the gels were stained using a Coomassie Blue solution (0.1% Coomassie Brilliant Blue R-250, 50% methanol and 10% glacial acetic acid) and destained with the same solution lacking the dye.

#### *2.2. Nanoparticle Characterization: Morphology, Size, Concentration, and Electrokinetic Mobility*

NPs were imaged by scanning electron microscopy (SEM) with a Zeiss SUPRA 40VP field-emission scanning electron microscope from the Scientific Instrumentation Center of the University of Granada (CIC, UGR).

The hydrodynamic size distribution of the NPs was evaluated by nanoparticle tracking analysis (NTA) with a NanoSight LM10-HS (GB) FT14 (NanoSight, Amesbury, UK) and an sCMOS camera. The particle concentration according to the diameter (size distribution) was calculated as an average of at least three independent size distributions. The total concentration of NPs of each system was determined in order to control the number of particles used in cell experiments. The measurement conditions for all samples were 25 ◦C, a viscosity of 0.89 cP, a measurement time of 60 s, and a camera gain of 250. The camera shutter was 11 and 15 ms for the empty and BMP-loaded NPs, respectively. The detection threshold was fixed at 5.

The electrophoretic mobility of the NPs was determined using a Zetasizer® NanoZeta ZS device (Malvern Instrument Ltd., Malvern, UK) working at 25 ◦C with an He-Ne laser of 633 nm, and a 173◦ scattering angle. Each data point was taken as an average over three independent sample measurements. For each sample, the electrophoretic mobility distribution and the average electrophoretic mobility (μ-average) were determined by the technique of laser Doppler electrophoresis.

#### *2.3. Colloidal and Temporal Stability in Biological Media*

The average hydrodynamic diameter and the polydispersity index (PDI) by dynamic light scattering (DLS) of each NP system were measured in different media (phosphate buffer (PB) saline phosphate buffer (PBS), and cell culture medium: Dulbecco's modified Eagle's medium, DMEM (Sigma)). Also, data on temporal stability were gathered by repeating these analyses at different times after synthesis (0, 1, and 5 days) and after 1 month under storage conditions.

In vitro release experiments were conducted as follows: 1 mL of each sample for each incubation time was suspended in PBS at 37 ◦C. After the corresponding time (24, 48, 96, 168 h), NPs were separated from the supernatant of released proteins by centrifugation for 10 min at 14,000 rpm (10 ◦C). The NP pellet was suspended in 1 mL of 0.05 M NaOH and stirred for 2 h for a complete polymer degradation. The alkaline protein solution was assayed by BCA and ELISA to quantify the unreleased amount. The protein released was calculated taking into account the total encapsulated amount. All experiments were made in triplicate.

#### *2.4. Cell Interactions*

For all biological in vitro studies, a cell population cultured from the maxillary alveolar bone was used. This population was previously characterized and confirmed to present all characteristics of a mesenchymal stromal cell population (MSC) [23]. Cells were taken from healthy human donors after the approval from the Ethics Committee for Human Research from the University of Granada (424/CEIH/2018). Regular Dulbecco's modified Eagle's medium (DMEM) with 1 g/L glucose (DMEM-LG) (Gibco), 10% fetal bovine serum (FBS) (Sigma-Aldrich, St. Louis, MO, USA), 1:100 of non-essential amino acid solution (NEAA) (Gibco), 0.01 μg/mL of basic fibroblast growth factor (bFGF) (PeproTech, London, UK), 100 U/mL of penicillin/streptomycin, and 0.25 μg/mL of amphotericin B was used as culture medium for all experiments. Cultures were maintained at 37 ◦C in a 5% CO2 atmosphere (2000 cells/well). All biological experiments were repeated in triplicate at least 3 times per condition.

#### 2.4.1. Cell Migration

A cell-migration assay was conducted as previously described [26,27]. Briefly, MSCs were distributed on to three wells for each condition and allowed to grow to a cell confluency close to 99%, in 24-wells/plate at 3000 cells/cm2, and in each well three different scratches were made. Then, cells were starved for 24 h by adding culture medium without serum. A scratch was made using a pipette tip along the diameter of the well. A wash step with PBS was performed to remove the scratched cells. Fresh complete culture media was added and supplemented depending on the assigned group (BMP-2, NP- BMP2, and NP-BSA-BMP2 at 1.25, 2.5, and 5 ng/mL of BMP-2). Afterwards, nine images were taken from the same area in each condition until 48 h later. On these images, the scraped area was measured by ImageJ software (National Institute of Health, Bethesda, MD, USA; http://rsbweb.nih.gov/ij/). The reduction in the scratched area over time was measured considering the area at time 0 as 100% open.

#### 2.4.2. Cell Proliferation

Proliferation was evaluated by a sulphorhodamine (SRB) assay [28]. The assay was conducted by seeding the cells at 1500 cells/cm2 in a 96-well plate at a confluence not higher than 50%. After cell attachment, the different supplements were added (BMP-2, NP- BMP2, and NP-BSA-BMP2 at 1.25, 2.5, and 5 ng/mL of BMP-2) and the cells were maintained in culture for up to 7 days. At each time point, the cells were washed with 1X PBS and fixed by adding ice-cold 10% trichloroacetic acid for 20 min at 4 ◦C. Then, the cells were washed 3 times with dH2O and dried until all time points were collected. Each well received 0.4% SRB in 1% acetic acid for 20 min at room temperature with gentle shaking. The staining was finished by washing each well 3 times with 1% acetic acid and drying it at room temperature for 24 h. The dye was retrieved from the cells by adding 10 mM Tris Base at pH 10.5 and gently shaking for 10 min. The solution recovered was then distributed in a 96-well plate and the optical absorbance was read at 492 nm.

#### 2.4.3. Osteogenic Differentiation

Osteogenic differentiation was evaluated by adding osteogenic media to the cell culture in combination with free BMP-2, NP-BMP2, and NP-BSA-BMP2 at the highest dosages used in previous experiments. Cells were seeded at 3000 cells/cm2 and cultured to reach an 85% to 90% confluency. This was followed by the addition of induction media containing 10 mM of β-glycerophosphate (Fluka, 50020), 0.1 μM of dexamethasone (Sigma-Aldrich, D2915) and 0.05 mM of L-ascorbic acid (Sigma-Aldrich, A8960). Cell cultures were maintained for 7 days to analyze early activity. At day 7, cells were collected in 1 mL of TRIzol®. Then, RNA was extracted and converted to cDNA. Alkaline phosphatase (ALP) was then evaluated, expression being calculated relative to glyceraldehyde-3-phosphate dehydrogenase protein (GAPDH) by the 2−ΔΔCt method. These procedures were conducted as described elsewhere [23]. Forward and reverse primer sequences were AGCTCATTTCCTGGTATGACAAC and TTACTCCTTGGAGGCCATGTG for GAPDH, and TCCAGGGATAAAGCAGGTCTTG and CTTTCTCTTTCTCTGGCACTAAGG for ALP.

#### 2.4.4. Statistical Evaluation

Cell migration and proliferation were evaluated by ANOVA followed by Tukey multiple comparisons test for pairwise analysis. Comparison between the levels of ALP at 4 vs. 7 days were analyzed by paired Student's *t* test. In all cases, a *p* value lower than 0.05 was established as statistical significance.

#### **3. Results and Discussion**

#### *3.1. Nanoparticle Formulation*

Double emulsion-solvent evaporation has been described as a robust and frequently used method to produce biomolecule-loaded PLGA NPs [6,12,13,29]. A formulation previously optimized by our group enabled the preservation of the biological activity of encapsulated biomolecules using a slightly aggressive organic solvent. Moreover, deoxycholic acid has been used in the first step of the formulation in order to improve the colloidal stability of NPs and, simultaneously, to obtain NP surfaces enriched with carboxylic groups, improving their versatility and allowing a subsequent chemical immobilization of different specific ligands [30]. By means of this improved formulation, in the present work, we developed empty nanoparticles (NPs) or nanoparticles encapsulating rhBMP-2 (NP-BMP2). A schematic description of the synthesis procedure is shown in Figure S1 of the Supplementary Data. For NP-BMP2, we achieved a protein-encapsulation efficiency (EE) of 97 ± 2%. This result is consistent with the literature in which several authors have reported similarly high values encapsulating this protein inside PLGA nano- and microparticles [31,32]. Our formulation has several factors leading to this very high EE value: The low protein/polymer relation in mass [33], the affinity of rhBMP-2 to an unspecific interaction with hydrophobic surfaces [31], or the addition of stabilizers (poloxamer) in the second step of the double-emulsion procedure [13]. The absence of rhBMP-2 in the supernatant resulting from the centrifugation step in the cleaning process was verified by ELISA and SDS-PAGE, in which a clear band corresponding to 14 kD of rhBMP-2 polypeptidic chains is shown for lane A in Figure 1, corresponding to NP-BMP2. The mass of protein encapsulated, around 2 μg, is similar to that of different PLGA micro- and nanosystems described in the literature [18,34,35]. Taking into account the storage conditions for our samples, this corresponds to 500 ng/mL, which represents a sufficient concentration for practical applications since this growth factor shows in vitro biological activities at very low dosages (5–20 ng/mL) [13].

**Figure 1.** SDS polyacrylamide gel electrophoresis (SDS-PAGE) analysis under reducing conditions of solid PLGA Nanoparticles (PLGA NPs) and liquid (supernatant) fractions of different NP systems. Lane **P**: Protein standards; lane **A**: NP-BMP2 (bone morphogenetic protein); lane **B**: supernatant of NP-BMP2 after synthesis and encapsulation of rhBMP-2; lane **C**: NP after physical adsorption of BSA/rhBMP-2; lane **D**: supernatant after physical adsorption of BSA(bovine serum albumin)/rhBMP-2 on NP system.

On the other hand, a second nanosystem resulted, modifying the way in which rhBMP-2 is incorporated in the nanocarrier. There are several examples of surface adsorption of different growth factors in micro- and nanoparticles [35–37], and surface immobilization over the encapsulation has recently been proposed as a way to modulate the later release of biomolecules. This process, which depends on the slow diffusion of biomolecules through the polymeric matrix, is consequently highly influenced by the protein–polymer interaction [38,39] and polymer degradation [3,6]. Thus, this new focus on the use of PLGA NPs for biomolecule delivery was explored by immobilizing the protein rhBMP-2 on the surface of empty NPs by means of simple physical adsorption. This process is known to be governed by electrostatic and hydrophobic interactions between protein molecules and NP surfaces [40].

For this, the surface-charged groups, the hydrophilicity, the net charge of the protein molecules, and the characteristics of the adsorption medium are the reference parameters. Thus, we designed a co-adsorption experiment in which a mixture of rhBMP-2 and BSA (0.4% *w*/*w*, rhBMP-2/BSA) interact simultaneously with the PLGA NP surface. Albumins are routinely used as protective proteins when growth factors are incorporated in PLGA NPs [13,19]. Moreover, a surface distribution of BSA molecules can improve the colloidal stability of NPs at physiological pH due to their net negative charge under these conditions [41]. Figure S2 from Supplementary Materials shows a scheme of the co-adsorption process. The adsorption efficiency is higher than 95% and in SDS-PAGE from Figure 1, two bands characteristic of both proteins can be seen in lane C, corresponding to the NP-BSA-BMP2 nanosystem. However, lane D, corresponding to the run of the supernatant from the centrifugation of the nanosystem after adsorption processes, shows the absence of any protein. This result is fully explained by taking into account the pH of the medium (pH 5.0), near the isoelectric point of BSA, where the adsorption of this protein onto negatively charged nanoparticles presents a maximum [40,42]. The immobilization of rhBMP-2 on the negatively charged surface of NPs proves they are electrostatically favored due to the positive net charge of this protein at acid and neutral pH.

#### *3.2. Nanoparticle Characterization*

#### 3.2.1. Nanoparticle Size

SEM and STEM micrographs (Figure 2) show that the samples consist of spherical particles of different diameters (between 150 and 450 nm), a range similar to that found in a previous work in which NPs were loaded with lysozyme following a similar synthesis protocol [11]. In that work, the DLS technique failed to provide a reliable size distribution. Therefore, the NTA technique was directly used to determine the hydrodynamic size of the BMP2-loaded NPs (see NTA video in the Supplementary Material).

The size distributions for empty (NP) and BMP-loaded NPs (NP-BMP2) from NTA (Figure 3 and videos S1, S2) were consistent with the SEM images. Particles with diameters between 100 and 500 nm were found to have the highest particle concentration at around 200 nm. The loading with BMP had an effect on the size distribution, leading to more defined peaks. These measurements enabled us to determine the concentration of particles in the measured sample: 6.88 <sup>±</sup> 0.09 <sup>×</sup> 108 pp/mL and 5.19 <sup>±</sup> 0.12 <sup>×</sup> 10<sup>8</sup> pp/mL for NP and NP-BMP2 nanosystems, respectively. These values were used (by taking into account the corresponding dilution) to control the number of particles added in the cell experiments.

**Figure 2.** Scanning electron microscopy (SEM) micrograph of rhBMP-2-loaded nanoparticles (NP-BMP2).

**Figure 3.** Hydrodynamic diameter distribution of NP (circles) and NP-BMP2 (thick black line) measured at pH 7.0 (phosphate buffer) by nanoparticle tracking analysis (NTA).

#### 3.2.2. Electrokinetic Mobility and Colloidal Stability

The surface charge of nanoparticles can be analyzed using an electrokinetic study by measuring the electrophoretic mobility (μe) under different conditions. Figure 4 shows the μ<sup>e</sup> and zeta potential values for the three nanosystems: NP, NP-BMP2, and NP-BSA-BMP2, at low ionic strength and different pH values. The electric surface charge of NPs resides in the carboxylic groups of the uncapped PLGA and deoxycholic acid molecules. These functionalized groups are additionally useful due to the possibility of a chemical surface vectorization in order to develop directed delivery nanocarriers [43]. It was previously confirmed that protonation of these acidic surface groups at pH values under their pKa value was tightly correlated with a loss of surface charge and consequently a reduction (in

absolute value) of the electrophoretic mobility of the colloidal system [44,45]. Usually, when colloidal particles are coated by protein molecules, the μ<sup>e</sup> values change markedly compared with the same bare surfaces and are influenced by the electrical charge of the adsorbed protein molecules [46,47]. The electrokinetic behavior of the NP-BMP2 system remains similar to that of NP, and encapsulation of rhBMP-2 does not affect the surface charge distribution. A similar result was reported by d'Angelo et al. on encapsulating different growth factors in PLGA-poloxamer blend nanoparticles in the same proportion *w*/*w* of protein/polymer [24]. This may be due to the low amount of encapsulated protein and its distribution in the inner part of the NPs (far from the surface). In our system, this internal distribution may be favored by the encapsulating conditions where the basic pH (pH 12.0) of the water phase containing rhBMP-2 allows a negative charge of these protein molecules, thereby preventing their electrostatic specific interaction with acidic groups of the NPs.

**Figure 4.** Electrophoretic mobility and zeta potential vs. pH in buffered media of low salinity (ionic strength equal to 0.002 M) for the different nanosystems: (black square) NP; (blue triangle) NP-BMP2; (red circle) NP-BSA-BMP2.

The electrokinetic distribution for the NP-BSA-BMP2 system radically changes. As previously shown, the very high adsorption efficiency leads to NPs with both proteins adsorbed around their surface. This situation is closely correlated with the μ<sup>e</sup> values from Figure 4. Taking into account the *w*/*w* relation between adsorbed proteins (250 times higher for BSA), albumin molecules modulate the behavior at pH values below their isoelectric point (pI 4.7), where the positive net charge of BSA masks the original surface charge of NPs and even changes their original values to positive ones. This is a typical result found for this protein-covering colloidal particles [42,48]. At neutral and basic pH values, BSA molecules have a negative net charge, and the slight decrease in the absolute μ<sup>e</sup> values could be due to the reduction of the negative net surface charge of NPs, which may be shielded, at least in a small part, by the positive charge of rhBMP-2 molecules under their basic isoelectric point (pI 9.0).

The colloidal stability for the different nanosystems (NP, NP-BMP2, and NP-BSA-BMP2) was determined by analyzing the size distributions in various media (PB, PBS, and DMEM) at different times after synthesis (0, 1, and 5 days). Size distributions similar to the original ones were found for the two formulations, NP and NP-BMP2, in all the media analyzed. This result was similar to that previously found for these types of NPs encapsulating lysozyme [11], in which the combination of electrostatic and steric interactions generated by surface chemical groups of NPs confer the stability mechanism that prevents colloidal aggregation [33]. The decrease of the absolute value of the zeta potential for the NP-BSA-BMP2 system as a consequence of surface protein distribution does not affect its colloidal stability. This system also maintains the same size distribution in the different media. It is commonly accepted that a zeta potential higher than +30 or −30 mV will give rise to a stable colloidal system [49] and the zeta potential value for NP-BSA-BMP2 is above −30 mV. Colloidal stability in PBS and DMEM, typically used media for the development of scaffold or cell interactions, respectively, assures the potential use of these nanosystems for in vitro or in vivo living environments. Additionally, these systems maintained their size under storage in PB, at 4 ◦C for at least 1 month (data not shown), showing this to be an adequate medium for sample storage.

#### 3.2.3. Protein Release

One of the main problems for micro- or nanosystems of PLGA drug delivery is to find the appropriate release pattern for encapsulated/attached protein molecules. A wide spectrum of formulations modulates this property by the use of different types of synthesis processes, PLGA polymers, co-polymers, and stabilizers [3,13]. An adequate limitation and control in the burst release is critical for BMPs in order to ensure long-term continuous release that, favored by the polymer degradation, provides better in vivo action in driving bone and cartilage regeneration [20]. Therefore, we previously developed a dual PLGA nanosystem for controlled short-term release, where protein diffusion and protein–polymer interaction are the main factors governing this process [11].

In the present work, NP-BMP2 and NP-BSA-BMP2 nanosystems represent two different ways in which rhBMP-2 was incorporated into the nanocarrier. Figure 5A shows the cumulative release of both proteins, rhBMP-2 and BSA, for different systems as a function of time in a short-term period (7 days). The encapsulated rhBMP-2 protein reaches an amount released of around 30% of the initial encapsulated one while adsorbed rhBMP-2, despite its surface distribution, is three times lower. However, BSA shows released amounts up to 80% of the initial adsorbed ones. In all cases, error bars correspond to the standard deviations from three independent experiments. Under these conditions, the growth factor encapsulated in NP-BMP2 presents a release pattern similar to that previously found with the same formulation but using lysozyme as the protein [11]. Poloxamer in the water phase of the synthesis process can be key in modulating both specific and unspecific interfacial protein interactions [50]. Thus, the relation between protein–polymer interaction and protein diffusion appears to be well balanced, preventing an excessive initial burst and simultaneously maintaining the needed protein flux to release around a third of the encapsulated rhBMP-2 in 7 days. Although an excessive initial burst has been widely reported for PLGA NPs related with protein molecules close to the surface [6], this situation did not appear for the NP-BMP2 system, this being consistent with the electrokinetic behavior that did not show the presence of protein near surface. The literature offers some examples with reduced short-term release of BMP-2 using more hydrophilic PLGA-PEG co-polymers [16] or a different synthesis process [25].

The release performance for the NP-BSA-BMP2 system, also shown in Figure 5A, presents notable differences. The electrokinetic profile has previously justified the surface location of BSA and rhBMP-2 on the surface, which could lead to a fast release of both proteins. However, results from Figure 5A,B show this trend only for the BSA protein that is released from NPs, with about 20% of the initial amount remaining after seven days. However, up to 90% of the initial load of rhBMP-2 protein, unlike BSA, remains attached to the surface. The NP surface with hydrophilic groups form poloxamer molecules and a negative charge due to the abundant presence of carboxylic groups (end-groups of PLGA and deoxycholic acid molecules) favor a desorption process for BSA, whose molecules have a negative charge under release conditions (physiological pH). This agrees with the results of other authors who, even after encapsulating BSA in PLGA-poloxamer blend NPs, achieved a fast burst release of above

40% to 50% of the initial protein amount [33]. Moreover, the co-encapsulation of albumins with growth factors could strongly affect its release profile, causing an initial burst [21,24]. Otherwise, the specific electrostatic attraction between positive rhBMP-2 molecules and negative surface groups slows down the short time release of this protein. This result is in agreement with the low release of adsorbed BMP previously found using PLGA micro- and nanoparticles with uncapped acid end groups [38,51]. Thus, the combination of different methods for trapping BMP-2 into and around NPs shows up the possibility of attaining a properly controlled release, balancing the interactions between polymers, stabilizers, and protein.

**Figure 5.** (**A**) Cumulative release of rhBMP-2 for NP-BMP2 (black square) and NP-BSA-BMP2 (red circle) systems; and cumulative release of BSA for NP-BSA-BMP2 (blue triangle) system, incubated for different times at 37 ◦C in saline phosphate buffer (pH 7.4). (**B**) SDS-PAGE analysis under reducing conditions of solid fraction of NP-BSA-BMP2 after release at different times where the number of each lane corresponds to the time in hours.

#### *3.3. Biological Activity and Interactions*

#### 3.3.1. Cell Migration

Cell migration is the first and necessary step in tissue regeneration [52]. Thus, a regenerative agent must accelerate cell migration or, at least, not interfere with it. In the present study, we found no differences between the groups, doses, and control in terms of closure of a scratched area (ANOVA with Tukey multiple comparisons test) (Figure 6). In contrast to our findings, previously published data suggests a positive effect of BMP-2 on cell migration [53,54]. However, in those studies, the doses applied, and the cell types were different than in the current experiments. We used lower doses of BMP-2 in order to test whether, even at low dosages, BMP-2 could still provide benefits if protected in a nanoparticle system. As mentioned, we demonstrated no negative effect of the system on cell migration. Our results nonetheless support the idea that BMP-2 activity is mediated by the activation of the phosphoinositide 3-kinase (PI3K) pathway, a common group of signaling molecules that participate in several process with BMP-2 and other molecules [26,54]. It should also be mentioned that the timeframe of a migration assay is short. Thus, the potential advantages of a controlled-release system as the one under study might be limited. That is, the release of BMP-2 from the nanoparticles, as demonstrated in Figure 5, is limited to the first 48 h. Thus, a sustained positive effect on migration activity over time could be hypothesized.

**Figure 6.** Migration assay. Percentage of scratched area closure at 24 and 48 h on different groups and doses.

#### 3.3.2. Cell Proliferation

Proliferation is another of the cell activities required for tissue regeneration. However, this property must be balanced with both migration and differentiation, and not all three characteristics increase at the same time and with the same ratios [55]. In fact, reportedly, when a dose of BMP-2 induces higher proliferation, it decreases differentiation [56]. This property has been extensively analyzed but discrepancies can still be detected in the literature. Therefore, Kim et al. analyzed different doses of BMP-2 and its effect on cell proliferation and apoptosis. It was confirmed in vitro that high doses, but still lower than those used clinically, reduce cell proliferation and increase apoptosis [57]. This should be avoided. We have found that although free BMP-2 does not induce higher proliferation than the control at any of the doses applied nor time points (ANOVA with Tukey multiple comparisons test), the same amount of BMP-2 encapsulated or adsorbed onto PLGA nanoparticles boosts proliferation, this being statistically significant when using a dose of 2.5 ng/mL or higher (ANOVA with Tukey multiple comparisons test) (Figure 7). These dosages are still lower than those suggested in previous studies. Apart from that difference, a positive effect on proliferation was still achieved. Moreover, following the release pattern from Figure 5, more BMP-2 is expected to be released over time beyond the 7-day time frame. Thus, a sustained induction effect could be expected as well until full confluency of the cell culture.

**Figure 7.** Proliferation of human mesenchymal stromal cells (MSCs) as measured by sulphorhodamine (SRB) absorbance. Results were normalized to T0 in each group.

#### 3.3.3. Osteogenic Differentiation

It has been confirmed that cell differentiation induced by BMP-2 needs the presence of permissive osteoinductive components. Particularly, β-glycerophosphate has been shown to exert a synergistic effect with BMP-2 in inducing cell differentiation [56]. Thus, to test for osteogenic differentiation, we analyzed the expression of ALP mRNA. Maximum ALP activity was found to occur 10 days after stimulation with PLGA-based microparticles containing BMP-2 in co-encapsulation with human serum albumin [16]. Although other tests could have been used to reinforce our findings, ALP is known to modulate the deposition of mineralized nodules, thus indicating osteoblastic activity. For all of this, we supplemented the differentiation media with β-glycerophosphate and either free BMP-2, NP-BMP2, or NP-BSA-BMP2 for 4 and 7 days so that we could capture the early dynamics of the expression of the gene. In our study, we identified an increase in the expression of ALP in all groups from day 4 to day 7 (Figure 8). Although ALP at day 7 in the BMP-2 group appears to be higher than for the other two groups, the change did not prove significant. In fact, differences between groups were not statistically significant within any time period. Noteworthy though, the increase was not significant within the BMP-2 group (*p* = 0.141, Student's *t* test), but it was significant within the other two groups (*p* = 0.025 and *p* = 0.003; NP-BMP2 and NP-BSA-BMP2 groups, respectively). This, again, could be taken as a confirmation of the sustained release of the protein from the nanoparticle system beyond the earlier time points.

This and both the migration and proliferation studies described below lead us to confirm that the system proposed can maintain a proper release of BMP-2 over time, sustaining a positive effect on cell migration and proliferation with initial reduced doses of BMP-2. The fact that the excessive initial burst is prevented is important for the application of this nanotechnology in bone regeneration, as in dentistry. In this way, the negative effects of initial high doses of BMP-2 are avoided at the same time as the molecule is protected from denaturalization inside the NP. Thus, the regenerator effects are maintained over time.

**Figure 8.** Relative fold change in the expression of ALP mRNA (control group: BMP2 at 4 days). \* = Statistical significance of the comparison over time (*p* = 0.025 and *p* = 0.003, Student's *t* test; NP-BMP2 and NP-BSA-BMP2 groups).

#### **4. Conclusions**

In this work, a delivery PLGA-nanosystem previously developed for model proteins was chosen as the reference system to carry and deliver the growth factor BMP-2. This NP system, with a dual size distribution, was developed following a double-emulsion formulation in which the process and the components used were optimized to reach the appropriate colloidal and biological behavior. Encapsulation and adsorption are two different processes to load BMP-2 in PLGA NPs. Both were tested to elucidate the factors controlling them and their influence in the physico-chemical and biological properties of nanosystems. We verified that protein–polymer specific interactions have a major role in the way that protein molecules are carried and delivered from NPs. In vitro experiments showed that BMP-2-loaded PLGA NPs are the nanocarriers with the best release profile over the short-term without an initial burst and with moderate and sustained release of active protein before the onset of polymer degradation. Therefore, the biological activity is positive with no negative interaction with migration or proliferation but rather the induction of cell differentiation through the expression of ALP.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1999-4923/11/8/388/s1, Figure S1. Scheme of the formulation of NP-BMP2; Figure S2: Scheme of the protein adsorption process for NP-BSA-BMP2; Video S1. NTA experiments for NP-BMP2; Video S2. NTA experiments for empty NPs.

**Author Contributions:** Conceptualization, J.M.P.-G. and P.G.-M.; methodology, J.M.P.-G., A.B.J.-R. and M.P.-M.; investigation, T.d.C.-S., I.O.-O., J.M.P.-G., A.B.J.-R. and M.P.-M.; resources, A.B.J.-R., P.G.-M., F.O.-R.; writing—original draft preparation, J.M.P.-G. and M.P.-M.; writing—review and editing, J.M.P.-G., M.P.-M., A.B.J.-R., T.d.C.-S.; supervision, J.M.P.-G., P.G.-M. and F.O.-R.; funding acquisition, A.B.J.-R. and P.G.-M.

**Funding:** This research was funded by the Consejería de Economía, Innovación, Ciencia y Empleo de la Junta de Andalucía (Spain) through research groups FQM-115 and CTS-1028, by the following research project: MAT2013-43922-R—European FEDER support included—(MICINN, Spain) and by MIS Ibérica S.L.

**Acknowledgments:** The authors wish to express their appreciation for the technical support to D. Darío Abril-García.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Biocompatible PLGA-Mesoporous Silicon Microspheres for the Controlled Release of BMP-2 for Bone Augmentation**

**Silvia Minardi 1,**†**, Joseph S. Fernandez-Moure 2,**†**, Dongmei Fan 1, Matthew B. Murphy 1, Iman K. Yazdi 1, Xuewu Liu 3, Bradley K. Weiner 1,4,\* and Ennio Tasciotti <sup>1</sup>**


Received: 14 December 2019; Accepted: 28 January 2020; Published: 1 February 2020

**Abstract:** Bone morphogenetic protein-2 (BMP-2) has been demonstrated to be one of the most vital osteogenic factors for bone augmentation. However, its uncontrolled administration has been associated with catastrophic side effects, which compromised its clinical use. To overcome these limitations, we aimed at developing a safer controlled and sustained release of BMP-2, utilizing poly(lactic-*co*-glycolic acid)-multistage vector composite microspheres (PLGA-MSV). The loading and release of BMP-2 from PLGA-MSV and its osteogenic potential in vitro and in vivo was evaluated. BMP-2 in vitro release kinetics was assessed by ELISA assay. It was found that PLGA-MSV achieved a longer and sustained release of BMP-2. Cell cytotoxicity and differentiation were evaluated in vitro by MTT and alkaline phosphatase (ALP) activity assays, respectively, with rat mesenchymal stem cells. The MTT results confirmed that PLGA-MSVs were not toxic to cells. ALP test demonstrated that the bioactivity of BMP-2 released from the PLGA-MSV was preserved, as it allowed for the osteogenic differentiation of rat mesenchymal stem cells, in vitro. The biocompatible, biodegradable, and osteogenic PLGA-MSVs system could be an ideal candidate for the safe use of BMP-2 in orthopedic tissue engineering applications.

**Keywords:** BMP-2; silicon; microsphere; PLGA; bone regeneration; controlled release

#### **1. Introduction**

Autologous bone grafting remains the gold standard for spinal fusion, traumatic non-union and total hip arthroplasty complicated by osteolysis [1–3], yet it comes with morbidity (separate incision, graft site pain, potential infection, etc.) and might provide insufficient volumes of bone for complex or multilevel reconstruction [4]. Bone morphogenetic protein-2 (BMP-2) is a transforming growth factor known to play a key role in the development and repair of bone and cartilage [5]. Initially, it appeared to provide an ideal solution [6–8] to enhance bone growth but as its clinical use has expanded, multiple complications associated with BMP-2 use have come to light including local wound problems, chemical radiculitis, bony overgrowth into the canal or foramen, osteoclast activation with associated bony resorption and device displacement, and, possibly, cancer when used at very high doses in an off-label

manner [1,9–11]. It is thought these complications were associated with the uncontrolled release and systemic distribution of supraphysiologic doses of this potent growth factor [1,12]. One promising way to avoid these complications is via the controlled local delivery of very small but effective doses using the combination of growth factors with controlled drug delivery vehicles [13–17].

To date, however, systems designed to provide this controlled release have been limited by: (a) a burst release phenomena which leads to similar supraphysiologic dosing, inefficiently sustained dosing, and uncontrolled delivery as that seen with delivery systems currently used clinically; (b) the inability to preserve the quaternary structure of drugs following release from the delivery system; and (c) delivery system (polymer) degradation byproducts that have a secondary negative impact on the structure of the drugs released [16,18]. Therefore, a new carrier system capable of sustained, regulated, local release of small but effective doses that do not impact BMP-2 functionality are needed to allow the avoidance of the biological complications associated with burst supraphysiologic dosing [16].

With the development of nanomaterials, several types of particles, both nano and micro, have been used as growth factor delivery carriers [19]. Of these, nanoporous silicon has emerged as a material uniquely capable of the preservation of protein stability and function with predictable degradation properties in physiologic fluids and systems [16,17,19–22]. The breakdown product, orthosilic acid, has also been shown to stimulate mineralization by osteogenic cells while retaining the ability to buffer the breakdown products of coating polymer (poly(lactic-*co*-glycolic acid), PLGA) [19,23].

Recently, by integrating the drug preserving and encapsulating capabilities of nanoporous silicon with the further controlled release capabilities achieved by polymer encapsulation (using PLGA), we optimized a double controlled delivery system for the controlled and sustained temporospatial release of growth factors [17]. The platforms consisted of the mesoporous silicon-based multistage vector (MSV), encapsulated within a PLGA microparticle (PLGA-MSV). In our previous study, we demonstrated that PLGA-MSV is able to efficiently load a growth factor (i.e., PDGF-BB) and release it in a controlled fashion in vivo, with a significant reduction of the initial burst release, while preserving its functionality (i.e., inducing vascularization) [17].

The aim of the current study was to optimize the release of BMP-2 through the PLGA-MSV delivery system and assess its effectiveness in inducing osteogenesis in vitro.

#### **2. Materials and Methods**

#### *2.1. Preparation of PLGA-MSV Microspheres*

The PLGA-MSV microspheres were fabricated by a modified S/O/W emulsion method as in our previously published studies [7]. Briefly, PLGA (Sigma Aldrich, St. Louis, MO, USA was dissolved in dichloromethane (DCM; Fisher Scientific, Loughborough, UK) to form PLGA/DCM organic phase solution (10% and 20% *<sup>w</sup>*/*v*). BMP-2 loaded particles (8 <sup>×</sup> 107) were suspended into 1 mL of PLGA/DCM solutions (10% and 20% *w*/*v* respectively) by vortex mixing and sonication for 2 min. The organic phase containing the MSV particles was transferred into 3 mL of PVA (2.5% *w*/*v*) solution and emulsified for 1 min by vortex mixing. The primary emulsion was then gradually dispersed into 50 mL of PVA solution (0.5% *w*/*v*). The resulting suspension was stirred continually for 2 h under a biochemical hood, and the DCM evaporated rapidly during the stirring process. PLGA-MSV microspheres were washed with deionized water 3 times and lyophilized overnight. The freeze-dried BMP-2 loaded PLGA-MSV microspheres were then stored at −80 ◦C.

#### *2.2. Characterization of PLGA-MSV Microspheres*

The morphology of the microspheres was characterized by scanning electron microscope (SEM; Nova NanoSEM 230, FEI, Lincoln, NE, USA) and confocal microscope (Nikon A1 laser confocal microscope). The samples were sputter coated with 8 nm of platinum (Pt; Cressington sputter coater 208 HR System, Ted Pella, Inc., Watford, UK) and examined by SEM under a voltage of 3 kV, spot size 3.0, and a working distance of 5 mm.

#### *2.3. Loading of BMP-2 into Nanoporous Silicon Particles (MSV)*

Two hundred microliters of BMP-2 growth factor solution (Peprotech) was added into 8 <sup>×</sup> 107 oxidized nanoporous silicon particles in an Eppendorf tube. The suspension was mixed throughout by vortex mixing and sonication. The tube was gently rotated on a rotator at room temperature for 2 h to allow the adsorption of BMP-2 into the MSV particles. The BMP-2 loaded particles were then spun down by centrifugation (Sorvall Legend X1R Centrifuge, Thermo Scientific, Waltham, MA, USA) (4500 rpm for 5 min), lyophilized overnight, and stored at −80 ◦C for future use. The concentrations of the BMP-2 loading solution and the supernatant were measured by Elisa assay to determine the amount of BMP-2 loaded into the MSV particles.

#### *2.4. Evaluation of Growth Factor (BMP-2) In Vitro Release*

The BMP-2 loaded PLGA-MSV microspheres (10% and 20% *w*/*v*) containing 8 <sup>×</sup> 107 of MSV particles were dispersed into 0.5 mL of 1% BSA solution at 37 ◦C. The BMP-2 loaded PLGA microspheres (10% and 20% *w*/*v*) were used as a control. At predetermined time intervals, the suspension was spun down at 4500 rpm for 5 min and 0.5 mL of each supernatant was collected, and replaced with 0.5 mL of fresh 1% BSA solution. The amount of BMP-2 released from BMP-2 loaded PLGA-MSV microspheres was detected using an enzyme-linked immunosorbent assay kit (BMP-2 ELISA, R&D Systems, Minneapolis, MN, USA).

#### *2.5. Cell Isolation and Culture*

The study protocol and all operations were reviewed and approved by the Houston Methodist Research Institute's Institutional Animal Care and Use Committee (IACUC, protocol IS00003525, 18 August 2010). All investigators complied with the National Research Council's Guide for the Care and Use of Laboratory Animals. Male Sprague Dawley rats (*N* = 10) with an average weight of 310 g were used in the study. The rodents underwent a mandatory 48 h acclimation time prior to any surgical procedures and were housed in pairs with ad libitum water and chow until the study period began. Bone marrow stromal cells (BM-MSCs) were isolated from male Sprague Dawley rats as previously described. Briefly, femora and tibiae bones were removed from male Sprague Dawley rats (100–125 g) sacrificed by CO2 overdose under isoflurane anesthesia. Bones were cleaned of connective tissues, ligaments, and muscle by scalpel, washed thoroughly in phosphate-buffered saline (PBS, Invitrogen, Carlsbad, CA) containing 2% penicillin and streptomycin (P/S, Invitrogen). The proximal and distal ends of each bone were removed and the marrow was gently flushed out with PBS containing 1% fetal bovine serum (FBS, HyClone, ThermoFisher 124 Scientific, Logan, UT, USA) and 1% P/S. The diaphysis regions were crushed using a mortar 125 and pestle while submerged in PBS washes until the PBS appeared clear, indicating complete removal of the remaining marrow and perivascular cells. The total BM fraction was kept on ice for up to 1 h prior to further purification. The total BM cell fractions was counted by hemocytometer and then purified to mononuclear cells by centrifugation on Ficoll (150× *g* for 30 min without brake). The mononuclear BM populations were counted, resuspended in standard media (alpha-MEM (αMEM, Invitrogen) with 20% FBS, 1% P/S, 1% sodium pyruvate and 1% GlutaMAX (Invitrogen)). Cells were seeded in T75 flasks at a density of approximately 105 cells/cm<sup>2</sup> and cultured in hypoxic conditions (5% O2, 5% CO2) to maintain their multipotency. Upon reaching 80% confluency, cells were passaged and split 1:4 in new flasks.

Cells were cultured in α-minimum essential medium (αMEM; Invitrogen) supplemented with 20% (*v*/*v*) defined fetal calf serum (Invitrogen), 1%, l-glutamine (Invitrogen), 1% sodium pyruvate (Invitrogen), 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen) as the standard growth media. Osteogenic growth media included 10 mM β-glycerophosphate, 0.1 mM ascorbate-2-phosphate, and 100 nM dexamethasone. Cells were maintained at 37 ◦C in a humidified 5% CO2 atmosphere. Cell culture media was changed every 3 days.

#### *2.6. Cell Metabolic Activity—MTT Assay*

MTT assay of BM-MSC treated with PLGA-MSV microspheres was performed to quantify metabolic activity [8]. Two thousand and five hundred BM-MSCs were seeded and cultured in a 24-cell culture well plate in the presence of the PLGA and PLGA-MSV microspheres (BM-MSCs:Particles, 1:5). Cells only were used as a control. MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) assay was performed on day 1, 4, and 7. Cell culture media was removed from cell culture wells and 500 μL of MTT working solution (0.5 mg/mL) were added into the wells. The cells were incubated in the MTT working solution at 37 ◦C for 4 h. The solution was removed from the cell culture wells and replaced with 500 μL of dimethyl sulfoxide (DMSO; Sigma Aldrich). The cells were incubated with DMSO at room temperature for 30 min. The solutions were transferred to a 96 well plate and the absorbance of the colored solutions was quantified by a spectrophotometer (Synergy H4 Hybrid Reader, BioTek, Winooski, VT, USA) at 570 nm. DMSO was used as blank. Cells only and PLGA particle wells were used as controls.

#### *2.7. Alkaline Phosphatase (ALP) Activity Assay*

Osteogenic differentiation was measured by ALP activity, a biomarker of osteoblastic differentiation. The assay was carried out according to manufacturer guidelines for the spectrophotometric procedure using ALP reagent (Vector Laboratories, Inc. Burlingame, CA, USA). The BM-MSCs were plated into a 24 well cell culture plate at a density of 2500 cells per well with the PLGA-MSV microspheres (20% *w*/*v*) loaded with BMP-2 was used as the experiential group. Cells alone, cells cultured with BMP-2, and empty PLGA-MSV microspheres without BMP-2 (20% *w*/*v*) were used as controls. Cell culture media was changed every 3 days. Cells were cultured in α-minimum essential medium (αMEM; Invitrogen) supplemented with 10% (*v*/*v*) fetal calf serum (Invitrogen), 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen) as the standard growth media. Culture conditions were 37 ◦C in a humidified 5% CO2 atmosphere. The osteogenic growth media included 10 mM β-glycerophosphate, 0.1 mM ascorbate-2-phosphate, and 100 nM dexamethasone. Cell culture media was replenished twice weekly. Cells were cultured in standard media until 60% confluence and then switched to osteogenic media. ALP assays were performed at week 1, 2, and 3. The medium was aspirated and 1 mL of PBS was added into each well to wash the cells. Cells were washed 3 times with PBS, and fixed in 10% buffered formalin for 15 min. Cells were then washed twice in deionized (DI) water and covered in ALP stain made fresh. ALP staining stock solution was made from ALP substrate kit III (Vector Laboratories. Inc., Burlingame, CA, USA) into 5 mL of 100 mM tris-HCl (pH = 8.2) solution. Absorbance of cells after staining was used to quantify ALP activity. The experiments were performed in triplicate.

#### *2.8. Von Kossa Co-Staining*

Cell cultures were co-stained for calcium-triphosphate mineral deposition by Von Kossa staining. Co-staining of the cell culture was performed to simultaneous both quantify ALP activity and provide a qualitative assessment of mineral deposition. Following ALP staining, cells were washed twice in DI water and soaked in in 1% aqueous silver nitrate (AgNO3) and placed under ultraviolet (UV) light for 60 min and then rinsed with DI water. To remove unreacted silver, 5% sodium thiosulfate was added for 5 min, removed and cells rinsed in DI water. Following this cell nuclei were counterstained with Nuclear Fast Red (Sigma Aldrich) for 5 min, rinsed in DI water, and serially dehydrated prior to characterization.

#### **3. Results**

#### *3.1. PLGA-MSV Characterization*

The PLGA-MSVs were characterized by SEM-energy-dispersive X-ray (EDX) and confocal microscopy. SEM images show that the spherical-shaped PLGA-MSVs with smooth surfaces had a wide distribution from a few microns to approximately 50 μm, with an average diameter of 23 ± 3 (*n* = 50; Figure 1A). EDX spectrum showed the presence of a Si peak (Figure 1B), which indicates the presence of MSV particles inside of the PLGA microspheres. The full encapsulation of MSV was further confirmed by optical microscopy, where hemispherical MSVs appeared yellow in color (Figure 1C). MSVs were also loaded with a reporter protein (FITC-BSA) and imaged by confocal microscopy. Imaging clearly demonstrated that the FITC-BSA loaded MSVs (in green) were fully encapsulated (Figure 1D–F).

**Figure 1.** SEM and confocal images of PLGA-multistage vector (MSV). Representative SEM image of PLGA-MSV (**A**). EDX spectrum of PLGA-MSV microspheres showing the presence of the silicon peak corresponding to MSV (**B**). Confocal laser microscopy Z-stack of PLGA-MSVs loaded with FITC-BSA (in green), showing that MSVs were fully encapsulated into the PLGA microsphere (**D**,**E**). Close up of a PLGA-MSV (**C**,**F**).

#### *3.2. PLGA-MSV Loading with BMP-2 and In Vitro Release*

Recombinant human BMP-2 was used in this study. BMP-2 is a 29 kDa protein with an isoelectric point of 8.21 (Peprotech). The BMP-2 loaded PLGA-MSV microspheres (with a PLGA coating of 10% and 20% *w*/*v*) were used for in vitro sustained release studies. The loading efficiency of BMP-2 into MSV particles is shown in Figure 2A. Consistent with mass transport theory, the greater number of particles added to the solution of BMP-2, the greater amount of loading achieved and thus a higher loading efficiency. The in vitro release profiles of BMP-2 from different types of microspheres were monitored for 40 days. A delayed burst release over three days was demonstrated by 10% *w*/*v* PLGA-MSV microspheres (Figure 2B). When monitored to 40 days 90% of release was achieved within 10 days and sustained release equilibrium near 24 days (Figure 2C). In contrast, BMP-2 release from 20% *w*/*v* PLGA-MSV microspheres showed a more linear-like release with a more linear early release profile (Figure 2B). This slower rate of BMP-2 release continued until equilibrium was reached at 41 days (Figure 2C).

**Figure 2.** The in vitro loading and release profiles of BMP-2 using PLGA-MSV microspheres (10%, 20% *w*/*v*; *n* = 3). (**A**) The loading profiles of BMP-2 into MSV particles with fixed BMP-2 concentrations and mass, and varied MSV microparticle number. Cumulative release profile of BMP-2 from different 10% and 20% PLGA-MSV in the first three days (**B**) and over 41 days, (**C**). Values are reported as mean ± standard deviation. A value of *p* < 0.05 was considered statistically significant: \*\* *p* < 0.01 (calculated by *t* test analysis).

#### *3.3. Cell Metabolic Activity—MTT Assay*

Cell metabolic activity in the presence of the PLGA and PLGA-MSV microspheres was analyzed using MTT assay. Figure 3 shows the results of MTT assay for BM-MSCs cultured with PLGA-MSV microspheres at 1, 4, and 7 days. Metabolic activity among cells alone (control) and cells with the PLGA-MSV microspheres shared the similar trend and there was no significant difference among of the three groups at 1, 4, or 7 days. Cells in each group achieved a comparable metabolic activity and reached equilibrium at each time point.

**Figure 3.** MTT assay for rat bone marrow stromal cells (BM-MSCs) cultured with 10% PLGA-MSV and 20% PLGA-MSV microspheres over 7 days (*n* = 3). Cell viability is reported as absorbance at 490 nm. The cell metabolic activity of BM-MSCs in the presence of both formulations of PLGA-MSV was comparable to that of untreated cells (CTRL). Values are reported as mean ± standard deviation. No statistical significance was found among selected experimental conditions.

#### *3.4. PLGA-MSV-BMP2 Enhances Alkaline Phosphatase Activity (ALP)*

ALP is synthesized by osteoblasts and is a biomarker of presumed ECM production during BM-MSC differentiation [24]. As shown in Figure 4, at weeks 1 and 2, BMP-2 loaded 20% PLGA-MSV(PLGA-MSV-BMP2) microspheres showed higher ALP activities compared to cells alone and cells treated with BMP-2 (BMP2). At week 2, cells treated with PLGA-MSV-BMP2 showed the highest ALP activity during the duration of the investigation. The ALP activities decreased to a fairly low level for all controls and experimental groups after 3 weeks. Although, interestingly, the ALP activity of the control remained elevated at this time point.

**Figure 4.** ALP activity of untreated BM-MSCs (CTRL), BM-MSCs treated with soluble BMP-2 or 20% PLGA-MSV loaded or non-loaded with BMP-2 over 3 weeks (mean ± SD, *n* = 3). Values are reported as mean ± standard deviation. Data was analyzed through 2-way ANOVA. A value of \* *p* < 0.05 was considered statistically significant (\*\*\*\* *p* < 0.0001).

#### *3.5. ALP Staining and Von Kossa Staining*

All groups were co-stained for ALP and calcium using standard ALP staining and Von Kossa stain. ALP staining data of BM-MSC treated with BMP2 and PLGA-MSV-BMP2 correlated with ALP quantitative activity data. BMP2 and PLGA-MSV-BMP2 showed qualitatively increased staining at 1 and 2 weeks while untreated cells showed increased ALP staining at 3 weeks (Figure 5).

**Figure 5.** ALP and Von Kossa co-staining of BM-MSCs and BM-MSCs with experimental groups. ALP activities increased from week 1 to week 2 and then decreased from week 2 to week 3 in BMP2 and BMP2-PLGA-MSV. The cells in the presence of the PLGA-MSV-BMP-2 microspheres at week 2 showed the strongest ALP activity (red circles) among all of the control and experimental groups over the 3-week culture period. Von Kossa staining progressively increased in all groups with BMP2 and BMP2-PLGA-MSV showing the most staining at week 3 (Scale bar: 60 μm).

Von Kossa staining was performed to assess functional activity of BM-MSC osteoinduction. Calcium deposition was delayed compared to ALP staining. Calcium (black stain) is seen in the PLGA-MSV-BMP2 group as early as one week and continues to increase throughout the 3 weeks having the most deposition of all groups. In the control and BMP2 group calcium begins to be seen at 2 weeks yet the deposition of calcium only continues to increase in the BMP2 group. PLGA-MSV did not demonstrate appreciable deposition until week 3 (Figure 5).

#### **4. Discussion**

This study reports the development and investigation of a PLGA based protein delivery system, PLGA-MSV microspheres, for controlled protein release to stimulate cell differentiation. Herein, we studied the loading and release of BMP-2 from MSV, the relationship of PLGA:BMP concentration on BMP-2 release, and the effect of long-term constant BMP-2 stimulation on BM-MSCs. BMP-2 was successfully encapsulated into the microsphere system and the performance was tested in vitro. The experiments show that: (1) BMP/PLGA-MSV microspheres had the ability to release small but effective doses of BMP-2 for 40 days in a controlled, linear fashion; (2) the release profile of the BMP/PLGA-MSV microspheres was dependent on the PLGA coating of 10% or 20%; (3) the PLGA-MSV system did not impact cell metabolic activity; and (4) BMP-2 released from PLGA-MSV microspheres was capable of osteoinduction of BM-MSCs.

Previously, published reports with the clinical and pre-clinical use of BMP-2 have demonstrated bone augmentation, but the growth occurred in an uncontrolled and non-physiologic fashion. Partly due to the short half-life and need for supraphysiologic dosing, this resulted in heterotopic bone formation. Clinically this has translated in to the development of multiple complications [1,10]. Currently formulated BMP devices must be used at high concentrations resulting in problems with protein release timing stability, and need for accessory factors [11,12]. Additionally, the bone regenerative process under these conditions is very different from the normal physiologic processes in normal bone healing in which multiple BMPs and other growth factors are present in overlapping temporal patterns [13,14]. Here we reported a biomolecular release vehicle capable of controlling release kinetics of an encapsulated molecule for localized protein delivery, in this case BMP2. This phenomenon is affected by the percentage of PLGA used indicating an added level of spatiotemporal control of the system. Increasing PLGA percentage slowed release kinetics from a burst release to a more constant release and could theoretically lead to more prolonged and slowed release kinetics [25,26]. The slower release kinetics may be more biologically accurate and induce a more sustained response in contrast to the massive burst release kinetics often seen from PLGA alone [27]. Here we demonstrated this system is capable of long-term delivery as well as in vitro osteoinduction.

As many of the issues attributed to the use of growth factors in regenerative medicine have been centered on the lack of controlled release, this model presents a system for localized delivery within a clinically relevant scenario such as a healing fracture. MSCs are the major regenerative cell involved in the fracture healing process [28]. Their stimulation is paramount for the successful regeneration of tissue and organized bone formation. Without their coordinated efforts through properly induced pathways, dysregulated and disorganized growth occurs leading to weakened fracture callus formation and recurrent injuries [15,16]. Previously, investigators have demonstrated the efficacy of BMP2 release on the activity of MSCs [17]. Here we show that normal cellular function is not inhibited by the PLGA-MSV system and that sustained BMP-2 release results in MSC osteoinduction and increased ALP activity compared to BMP-2 alone. Cells can proliferate normally in the presence of the PLGA-MSV microspheres, which, indicates that these materials were nontoxic to cells and compatible with normal cell function. When loaded with BMP, the PLGA-MSV system was used to stimulate BM-MSC in vitro. ALP enzyme activity was increased in the PLGA-MSV-BMP-2 group and this correlated with ALP staining. These findings translated to a persistent and increasing deposition of calcium seen in the Von Kossa co-stain. The delayed deposition of calcium is consistent with normal biological function and an expected outcome [29]. The PLGA-MSV-BMP system induced early and robust ALP activity

at one week compared to BMP2 alone. At this same time point calcium is already detected in the Von Kossa stain of the PLGA-MSV-BMP2 group. This suggests that the slowed release may in fact have greater biological activity compared to high dose free BMP. This localized effect may provide insights into the further concomitant use of MSCs and PLGA-MSV vectors for regenerative purposes in other applications.

There are several limitations with the current study. First the lack of free PLGA-BMP2 makes direct comparison difficult. There have been several studies looking at the use of PLGA as a protein carrier and more specifically a BMP2 carrier. These studies reported the release kinetics of BMP2 from PLGA and its impact in vitro on osteoinduction. Given the historical findings with PLGA alone as the carrier we chose to focus on the use of MSV encapsulated in PLGA. Second, while ALP activity, staining and calcium deposition all form part of the evaluation of BM-MSC osteoinduction further studies in the transcriptional changes and more quantitative histological assessments could be performed. This was out of the scope of this initial work and is the focus of subsequent works in progress. Third, an additional comparison group looking into the effects of PLGA-MSV ratio on bioactivity toward BM-MSC is needed. While there were significant differences between the release profiles between 10% and 20% PLGA-MSV-BMP2 no biologic data was obtained. This is currently a subject of current investigation. Lastly, this is only an in vitro study with one cell line. The biologic milieu of the in vivo micro-environment is composed of a complex interplay of signaling and cell types. While our in vitro findings are promising they warrant analysis in an in vivo fracture model to assess the efficacy of localized controlled release.

#### **5. Conclusions**

In conclusion, this biocompatible, biodegradable, and osteogenic PLGA-MSV microsphere system holds promise as a candidate for the delivery of safe, local, small but effective doses of bioactive proteins for pharmaceutical induction of osteoregeneration. The local and long-term delivery of proteins and other bioactive molecules for orthopedic applications may avoid the complications associated with burst release and supraphysiologic dosing of BMP-2 currently afforded by available delivery systems.

**Author Contributions:** Conceptualization, S.M., E.T., D.F.; methodology, S.M., D.F., M.B.M., I.K.Y., X.L.; data curation and analysis: S.M., D.F., M.B.M., I.K.Y.; writing—original draft preparation, S.M., D.F, J.S.F.-M. writing—review and editing, S.M., J.S.F.-M.; supervision, B.K.W., E.T.; project administration, B.K.W., E.T.; funding acquisition, B.K.W., E.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Defense Advances Research Projects Agency (DARPA), grants W81XWH-14-1-0600, Log #SC130156; W81XWH-15-1-0718, Log #4170002.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **PLGA-BMP-2 and PLA-17**β**-Estradiol Microspheres Reinforcing a Composite Hydrogel for Bone Regeneration in Osteoporosis**

**Patricia García-García 1, Ricardo Reyes 2,3, Elisabet Segredo-Morales 1, Edgar Pérez-Herrero 1,2, Araceli Delgado 1,2,\* and Carmen Évora 1,2,\***


Received: 23 October 2019; Accepted: 29 November 2019; Published: 3 December 2019

**Abstract:** The controlled release of active substances—bone morphogenetic protein 2 (BMP-2) and 17β-estradiol—is one of the main aspects to be taken into account to successfully regenerate a tissue defect. In this study, BMP-2- and 17β-estradiol-loaded microspheres were combined in a sandwich-like system formed by a hydrogel core composed of chitosan (CHT) collagen, 2-hidroxipropil γ-ciclodextrin (HP-γ-CD), nanoparticles of hydroxyapatite (nano-HAP), and an electrospun mesh shell prepared with two external electrospinning films for the regeneration of a critical bone defect in osteoporotic rats. Microspheres were made with poly-lactide-*co*-glycolide (PLGA) to encapsulate BMP-2, whereas the different formulations of 17β-estradiol were prepared with poly-lactic acid (PLA) and PLGA. The in vitro and in vivo BMP-2 delivered from the system fitted a biphasic profile. Although the in vivo burst effect was higher than in vitro the second phases (lasted up to 6 weeks) were parallel, the release rate ranged between 55 and 70 ng/day. The in vitro release kinetics of the 17β-estradiol dissolved in the polymeric matrix of the microspheres depended on the partition coefficient. The 17β-estradiol was slowly released from the core system using an aqueous release medium (*De*ff <sup>=</sup> 5.58·10−<sup>16</sup> <sup>±</sup> 9.81·10<sup>−</sup>17m2s−1) and very fast in MeOH-water (50:50). The hydrogel core system was injectable, and approximately 83% of the loaded dose is uniformly discharged through a 20G needle. The system placed in the defect was easily adapted to the defect shape and after 12 weeks approximately 50% of the defect was refilled by new tissue. None differences were observed between the osteoporotic and non-osteoporotic groups. Despite the role of 17β-estradiol on the bone remodeling process, the obtained results in this study suggest that the observed regeneration was only due to the controlled rate released of BMP-2 from the PLGA microspheres.

**Keywords:** BMP-2-microspheres; hydrogel system; 17-βestradiol release; bone regeneration; osteoporosis; poly-lactide-*co*-glycolide; polylactic acid

#### **1. Introduction**

Regeneration of bone critical defects is still a challenge in the orthopedic field. Local treatment with bone morphogenetic protein (BMP-2) incorporated in different biomaterial scaffolds has demonstrated to be efficient to induce new bone formation for critical bone defect in several animal models. Nowadays, collagen sponges loaded with recombinant BMP-2 are clinically available as bone graft substitutes

for the treatment of nonunion and critical-sized bone defects. Although BMP-2 is a potent osteogenic agent, a controlled release profile is required for safety and efficacy. Thus, the scaffold to fill the bone defect should not only be designed to act as support and guide for tissue growth, but also to control the release rate of active substances. Although many materials and structures have been proposed to construct these scaffolds, the control of the release rate has not always been taken into account. In fact, in some cases the protein is incorporated in the material by incubation and, unless a material-protein interaction occurs, a burst release at an early period would be expected. Consequently, a high dose of BMP-2 would be in blood circulation, leading to a high risk of side effects and, at the same time, a significant loss of the protein in the site of action. In addition, some authors showed that BMPs in these cases might also stimulate bone resorption due to the high dose of BMP-2 associated to uncontrolled release [1]. To minimize the osteoclastic effect of BMPs, some authors proposed the addition of an anti-catabolic agent. The most frequently studied combination has been BMP-2 with bisphosphonates as anti-resorption agents such as alendronate [2,3] and zolendronate [4]. The results indicated good bone regeneration with improved bone quality and mineralization in different localizations compared to BMP-2 alone [5,6].

Among the several natural polymeric scaffolds prepared for bone regeneration, chitosan is a biomaterial frequently used for this purpose. In a recent extensive review [7] based on chitosan (CHT) applied in bone tissue regeneration, different advantageous aspects were showed and discussed such as biocompatibility, capacity for BMPs sustained release, improvement of cell proliferation, and increase of in vitro and in vivo differentiation and mineralization. However, bone graft materials to simulate bone structure, e.g., collagen, the major ECM of bone tissue, and hydroxyapatite (HAP), a mineral component of the bone, have also been widely studied [6,8–14].

Although the aforementioned studies indicated positive results, the mentioned strategies applied in osteoporosis (OP) conditions have not always been effective [15]. According to recent reports, the prolonged subcutaneous administration of alendronate and the low level of estrogen in OP alters the evolution of calvarial bone repair due to estrogen, Transforming Growth Factor beta 1 (TGF-β1), and α-estrogen receptor (α-ER) interaction [16]. Menopausal women are the population most affected by this disease due to estrogen deficiency. Previous reports revealed that local implantation of scaffolds loaded with combinations of BMP-2 and 17β-estradiol formulated in microspheres of polylactic acid (PLA) or PLGA, in rat calvaria critical defects increased the bone repair in OP rats, but the new bone that refilled the defect was less mineralized compared to non-OP groups [17,18].

As some biomaterials may promote bone regeneration, controlled release of the active substances is required for efficient and safe bone regeneration [19,20]. In this study, we propose a loaded BMP-2 and 17β-estradiol sandwich-like system, comprising two polymeric external films and a core of a biocomposite hydrogel containing microspheres, to provide sustained release of active substances. The core system composed of CHT, collagen, HAP nanoparticles (nano-HAP), polyethylene glycol (PEG-400) and 2-Hidroxipropil γ-Ciclodextrin HP-γ-CD was previously characterized in terms of composition, rheological behavior and mass-transfer using RITC-dextran as macromolecule model and 17β-estradiol in microspheres [21]. In the present study, we aim to study the influence of the release rate of 17β-estradiol on the osteogenic effect induced by BMP-2 released from PLGA microspheres within core system after sandwich-like system implantation in a OP rats critical size defect. Therefore, 17β-estradiol was incorporated into the system in 3 forms: free and dispersed in the core, encapsulated in microspheres prepared with a mixture of PLA and PLGA dispersed within the hydrogel and lastly encapsulated in the PLGA films shell prepared by electrospinning technique.

#### **2. Materials and Method**

#### *2.1. Materials*

PLGA 75:25 (Resomer® RG755-S), PLGA 50:50 (Resomer® RG504), PLGA 85:15 (Resomer® RG858-S), and PLA (Resomer® RG203-S) were supplied by Evonic Industries (Darmstadt, Germany). Chitosan

(Protasan® UP-CL-213) was purchased from NovaMatrix (Sandvika, Norway). 2-Hidroxipropil γ-Ciclodextrin (CAVASOL® W8 HP), was supplied by Wacker Chemical (Burghausen, Germany). The bovine collagen type I was purchased from CellSystems Biotechnologie (Vertrieb GmbH, Germany). Riboflavin (RB), Poly(ethylene glycol) 400, Poly(vinyl alcohol) (PVA, Mw 33–70 kDa; 87–90% hydrolyzed), 17β-estradiol, and all the other reagents were purchased from Sigma-Aldrich, (St. Louis, MO, USA). The recombinant human bone morphogenetic protein 2 (BMP-2) was bought from Biomedal Life Sciences (Sevilla, Spain). Citrate-coated carbonated apatite nanoparticles (nano-HAP) were kindly donated (Jaime Gómez-Morales, PhD, Laboratory of Crystallographic Studies, CSIC, Granada, Spain).

#### *2.2. Microspheres Preparation and Characterization*

The BMP-2 microspheres were prepared by the double emulsion method (w/o/w) previously described [22]. Briefly, 200 μL of an aqueous solution (0.2% PVA) of BMP-2 (260 μg) was emulsified with 1 mL of a PLGA mixture (150 mg) of RG504 and RG858 [4:1] in methylene chloride (DCM) by vortexing 1 min (position 10, Genie® Industries 2, Sciencies Industries Inc. USA). Then, this emulsion was poured into 10 mL of 0.2% PVA solution vortexed 15 s, then poured into 100 mL of 0.1% PVA and kept under magnetic stirring for 1 h for solvent evaporation.

The 17β-estradiol microspheres were prepared by a modified solvent evaporation method previously described [17]. Briefly a mixture of 17β-estradiol (4 mg), PLA-S RG203-S (160 mg) and PLGA RG858 (40 mg) dissolved in 0.6 mL of DCM:Methanol (DCM:MeOH) (80:20) was emulsified with 4 mL of 1% PVA aqueous solution by vortexing 1 min (position 10), and then added to 100 mL of 0.16% PVA solution, under magnetic stirring for organic solvent evaporation (1 h).

Both type of microspheres were collected by filtration (Pall Corporation, pore size 45 μm, Sigma-Aldrich, USA), lyophilized, and stored at 4 ◦C until use.

Microspheres were characterized in terms of size (Mastersizer 2000, Malver Instruments, Malvern, UK) and morphology (SEM, Jeol JSM-6300, Tokyo, Japan). To determine the BMP-2 encapsulation efficiency and to carry out the BMP-2 release assays, some batches were prepared with 125IBMP-2. The BMP-2 was labeled with 125INa (Perkin-Elmer) by the iodogen method [23]. The content of 17β-estradiol in the microspheres was determined spectrophotometrically at λ = 280 nm previous dissolution in a mix of DCM:MeOH (80:20).

To determine the solubility of 17β-estradiol in the polymer matrix of the microspheres, differential scanning calorimetry (DSC 025, TA Instruments, New Castle, DE, USA) was performed. 17β-estradiol and lyophilized microspheres were analyzed after drying in an oven at 37 ◦C overnight. In addition, samples of polymer blends (RG 203-S and RG 858, 4:1) and samples of the polymer blend with excess 17β-estradiol (8.5%) were dissolved in DCM:MeOH (80:20) and maintained in a hood for 24 h. Then, samples were placed 48 h more in a vacuum desiccator to complete the evaporation of the organic solvent. The analysis of all samples was performed with the same thermal program in two thermal cycles under a nitrogen atmosphere (50 mL/min). In the first cycle, temperature was increased to 40 ◦C (10 ◦C/min) and then cooled to −20 ◦C (5 ◦C/min) to avoid possible water interference. Once the samples were stabilized, they underwent a final heating cycle from −20 ◦C to 270 ◦C (10 ◦C/min).

#### *2.3. Fabrication and Characterization of the Film*

The film was fabricated by a previously described electrospinning method [24]. Briefly, 7 mg of 17β-estradiol and 300 mg of a mixture of PLGAs, RG755-S and RG858 [4:1] were dissolved in 2 mL of hexafluoroisopropanol (Sigma-Aldrich, Steinheim, Germany) and electrospun at 7kV; flow rate of 3.0 mL/h and 10 cm of distance from the collector.

The film quality was checked in terms of porosity, thickness and fiber diameter using helium pycnometer (AccuPyc 1330, Micromeritics, Norcross, GA, USA), stereo microscopy (Leica M205C, Leica Las, v3 sofware), and SEM (Jeol, JSM-6300, Tokyo, Japan), respectively.

#### *2.4. Core System Preparation and Characterization*

To prepare the core system, approximately 20 mg of microspheres were dispersed in 50 μL of the hydrogel composed by a mixture of collagen type I (5 mg/mL), HP-γ-CD (34 mg/mL), RB (0.4 mg/mL), CHT (5 mg/mL), PEG-400 (150 mg/mL), and 5 mg of nano-HAP. Then, the hydrogel was cross-linked with 5% *w*/*w* TPP sterile aqueous solution (0.5 μL/μL of hydrogel) and visible light blue at 468 nm (Dental device) for 3 min [21]. The dose of BMP-2 was 6 μg in microspheres and the total 17β-estradiol dose was 200 μg in three different forms: electrospun films, microspheres, or dispersed into the gel.

The quality of the core system was checked according to its rheological characteristics and porosity previously described [21]. In addition, water uptake and mass loss assays were carried out by incubation of aliquots of 300 μL of the core system in 5 mL of sterile MilliQ water (37 ◦C) under orbital agitation (25 rpm). At specific times, six samples were withdrawn, we then removed excess water, weighed, and freeze-dried the samples. Then, three samples were visualized by SEM (Jeol JSM-6300) to see the evolution of the internal structure after incubation. The other three samples were used to record the dried weight and calculate the percentage of mass loss and water uptake, applying Equations (1) and (2), respectively, where *W*<sup>0</sup> is the initial weight of the sample and *Ww* and *Wd* are the weights of the wet and dried sample, respectively, at the different times tested.

$$\text{Mass loss} \left( \% \right) = \frac{\left( W\_0 - W\_d \right)}{W\_0} \times 100 \tag{1}$$

$$\text{Water update}(\%) = \frac{(\mathcal{W}\_{\text{iv}} - \mathcal{W}\_d)}{\mathcal{W}\_d} \times 100\tag{2}$$

To test the syringeability of the core system two syringes of 1 mL were loaded with a suspension of microspheres of 17β-estradiol in the hydrogel, up to 0.5 mL. Then, 4 doses of 50 μL each were unloaded from both syringes assayed for fluidity through a 20G needle and dose uniformity. For this, the discharged samples were lyophilized, and the 17β-estradiol content evaluated by spectrophotometry at 280 nm, after dissolution in DCM: MeOH (80:20).

#### *2.5. In Vitro Release Assays*

BMP-2 in vitro release assays were carried out by incubating an amount of 125I-BMP-2 microspheres and an amount of core system with the same amount of 125I-BMP-2 microspheres in sterile MilliQ water at 37 ◦C and 25 rpm. The amount of BMP-2 released was calculated by measuring the radioactivity of supernatant samples with a gamma counter (Cobra® II, Packard).

The in vitro release of 17β-estradiol from the different formulations (dispersed in the core system, microspheres, microspheres incorporated to the core system, and electrospun films) was carried out at 37 ◦C and 25 rpm using two release media: an aqueous solution of sodium lauryl sulfate (SLS) 1% [25] and MeOH:water (50:50) [26,27]. The released 17β-estradiol was measured in the supernatant using the spectrophotometric method. The effective diffusion coefficient, *De*ff in the matrix of the microspheres and the mass transfer coefficient of the drug in the boundary layer *h*, were calculated according to the non-steady-state Fick law, as previously described in detail [21]. Whether or not the released fraction of 17β-estradiol from the microspheres dispersed in the core system was analyzed, and Equations (3)–(6) were applied for *De*ff and *h* calculation.

$$\frac{M\_t}{M\_{\infty}} = 1 - \sum\_{n=1}^{\infty} \frac{6 \, L^2}{\beta\_n^2 \left(\beta\_n^2 + L^2 - L\right)} \exp\left(-\frac{\beta\_n^2}{R^2} \, D\_{eff} \, t\right) \tag{3}$$

where *Mt* and *M*<sup>∞</sup> are the total mass of drug released to the media at time *t* and at the end of the experiment, respectively. The β*ns* are the infinite roots (eigenvalues) of the Equation (4):

$$
\beta\_{\rm nl} \cot \beta\_{\rm nl} + L - 1 = 0 \tag{4}
$$

*L* is the dimensionless mass transfer Biot number Equation (5):

$$L = \frac{h \, R}{D\_{eff}} \tag{5}$$

For large values of *L*, the roots of Equation (4) are multiples of the number *pi* and Equation (3) can be simplified in the Equation (6), that is, a simplified solution of non-steady-state Fick law:

$$\frac{M\_{\rm f}}{M\_{\rm ox}} = 1 - \frac{6}{\pi^2} \sum\_{n=1}^{\infty} \frac{1}{n^2} \exp\left(-\frac{n^2 \cdot \pi^2}{R^2} \; D\_{eff} \; t\right) \tag{6}$$

As stated previously, to minimize the residual sum of squares, "genetic algorithms" already implanted in R software (R Foundation for Statistical Computing, version 3.6.1., 2019, Vienna, Austria) were used [21].

#### *2.6. Animal Experiments*

All animal experiments were carried out in conformity with the European Directive (2010/63UE) on Care and Use in Experimental Procedures. Furthermore, the animal protocols were approved on 5 November 2014 by the Ethics Committee for Animal Cares of the University of La Laguna (CEIBA) with identification code CEIBA2014-0128. All surgical procedures were made under aseptic conditions.

#### 2.6.1. Animal Models

Forty female adult Sprague-Dawley rats approximately 12 weeks old, weighing 200–250 g, were divided in 4 groups of 10 each. The experimental osteoporosis was induced in 3 groups by three different protocols, OVX, chronic administration of DEX and OD. The forth group was the sham, non-osteoporotic control group (non-OP). The bilateral ovariectomy was carried out under isoflurane anesthesia, via dorsal approach to the animals of OVX and OD groups. Analgesia consisted in buprenorphine administered subcutaneously (0.05 mg/kg) before surgery and paracetamol (70 mg/100 mL) in the water, for 3 days post-surgery. The DEX group received 0.3 mg/kg body weight of dexamethasone-21-isonicotinate (Deyanil retard, Fatro Ibérica, Barcelona, Spain) administered subcutaneously once ievery two weeks [28] up to the time of euthanasia. Then, two weeks after the ovariectomy, the rats of group OD were chronically treated with DEX as the DEX group. The 40 rats were sacrificed after 12 weeks and the calvaria and femurs were extracted to be histologically analyzed. The results of these analyzes were used to evaluate the 3 protocols tested to induce OP.

#### 2.6.2. Animals Groups

Fifty female Sprague-Dawley rats (12 weeks old), weighing 200–250 g, were divided into 2 groups of 25 each: OP and non-OP. The rats of the OP group were ovariectomized and the rats of the non-OP group underwent similar surgery but the ovaries were not resected. Twelve weeks post-surgery, 8 mm critical size cranial defects were created surgically with a trephine burr in the rats under isoflurane and the systems were placed into the defects [20]. Analgesia treatment was administrated.

Female rats were divided into 5 groups of 10 rats each—5 OP and 5 non-OP—and the applied regenerative treatment is reflected in Table 1. The implantation of the systems was carried out following a two steps procedure. First, a layer of film (bottom film), previously soaked in the blood produced during the surgery, was placed in the defect then 50 μL of the hydrogel mixed with the microspheres and partially cross-linked with UV light, was discharged. Second, the hydrogel was completely cross-linked by dripping 25 μL of sodium tripolyphosphate (TPP) forming the core system, after 5 min, a second layer of film (soaked in blood) was placed on the top, like a sandwich, and the wound was then closed.


**Table 1.** Experimental groups to evaluate regenerative efficiency.

#### 2.6.3. 125I-BMP-2 in Vivo Release Assay

The BMP-2 release kinetics was monitored periodically by measuring the remaining 125I-BMP-2 at the rat calvarial defect site (*n* = 5) using an external probe-type gamma counter (Captus ®, Capintec Inc., Ramsey, NJ, USA), as previously described and validated [29].

#### *2.7. Rat Mesenchymal Stem Cells (rMSCs) Osteogenic Di*ff*erentiation*

The rMSCs were obtained by centrifugal isolation as previously described [30] from the bone marrow of the femur of OVX female Sprague-Dawley rats. Briefly, the cells were resuspended in high glucose DMEM (HyClone® Utah) supplemented with 10% fetal bovine serum (Biowest, South America Origin), 1% penicillin–streptomycin (PAA, Pasching, Austria), and 2 mM l-Glutamine stable (Biowest, France) (Complete Culture Medium, CCM). Then, cells were cultured in flasks of 75 cm<sup>2</sup> and subcultured by incubating at 37◦C and 5% CO2. The culture medium was changed every 2–3 days.

To test the osteogenic differentiation, 50,000 cells (passage 2) in 20 μL of CCM were added over aliquots of 300 μL of the core system (hydrogel with microspheres) with and without nano-HAP and incubated at 37 ◦C and 5% CO2 for 1.5 h for cell adhesion. The homogeneous cell distribution was checked by light microscopy. Afterwards, 500 μL of CCM were added to each well, after 3 days incubation the medium was changed to CCM supplemented with 10 mM β-glycerol phosphate, 10–7 M dexamethasone and 50 μM ascorbate-2-phosphate. At 7, 14, and 21 days of culture, three wells of each time point were washed (2 times) with Hank's balanced salt solution (HBSS 1x) and cooled at 4 ◦C. Then, 500 μL of 0.1 M buffer Tris-HCl, 0.1M NaCl, and 0.05 M MgCl2 (pH = 9.2–9.5) containing Nitro blue tetrazolium chloride (NBT, Roche Diagnostics, Mannheim, Germany) and 5-Bromo-4-chloro-3-indolyl phosphate (BCIP, Roche Diagnostics, Mannheim, Germany) were added and incubated at 37 ◦C and 5% CO2 under soft agitation for 1.5 h. Then, the NBT/BCIP was removed and the cells were fixed with a solution of 3.7–4% p-formaldehyde buffered to pH = 7.0 (Panreac®, Barcelona, Spain) during 30 min. After this, the formaldehyde was removed, and the wells were washed 3 times with HBBS 1x. Immediately after this, cells were visualized by stereo microscopy (Leica M205C, Leica Las, v3 software). In addition, samples were dehydrated in a graded series of ethanol before being embedded in Paraplast® and microtome (Shandon Finesse 325, Thermo Fisher Scientific, Madrid, Spain) sections were observed by light microscopy (LEICA DM 4000B, Barcelona, Spain).

#### *2.8. Histology, Immunohistochemical, and Histomorphometrically Evaluation*

First, to check the osteoporotic-like condition, 12 weeks after the 12 rats undergone the different protocols were sacrificed and the femurs and calvaria were analyzed. The femurs and calvaria were fixed (4% paraformaldehyde solution), decalcified in Histofix® Decalcifier (Panreac, Barcelona, Spain) and prepared for histological analysis as previously described [20,22].

Bone morphology was analyzed by hematoxylin–erythrosin staining. The histomorphometric analyze was carried out in femurs by measuring the following parameters; thickness of the cortical bone (Ct.Wi) and number (Tb.N), width (Tb.Wi) and separation (Tb.Sp) of the trabeculae in cancellous bone. In the calvaria bone, the histomorphometric analysis was carried out by measuring the following parameters, cortical bone thickness (CBT) and intercortical space thickness (IST) occupied by trabecular bone in transversal sections of calvaria.

To determine the capacity of the bone active substances, so as to regenerate the critical size defect practiced in the calvaria of the rats, samples of the 10 groups of 5 rats each were examined.

Samples were processed as previously described [22]. New bone formation was identified by hematoxylin–erythrosin staining. Bone mineralization was assessed with VOF trichrome stain, in which red and brown staining indicates advanced mineralization, whereas less mineralized, newly formed bone stains blue [31]. Sections were analyzed by light microscopy (LEICA DM 4000B, Barcelona, Spain). Computer-based image analysis software (Leica Q-win V3 Pro-Image Analysis System, Barcelona, Spain) was used to evaluate all sections. A region of interest (ROI) within the defect (50 mm2) for quantitative evaluation of new bone formation was defined. New bone formation was expressed as a percentage of repair with respect to the original defect area within the ROI. From the total bone repair, the areas of mature bone (MB) and immature bone (IB) were determined, and the MB/IB ratio for each experimental group as well as between non-osteoporotic and osteoporotic-like animals was calculated.

For immunohistochemical analysis, sections were deparaffined and rehydrated in Tris-buffered saline (TBS) (pH 7.4, 0.01 M Trizma base, 0.04 M Tris hydrochloride, 0.15 M NaCl), which was used for all further incubations and rinse steps. Sections were incubated in citrate buffer (pH 6) at 90 ◦C for antigen retrieval, followed by incubation in 0.3% hydrogen peroxide in TBS buffer for 20 min. After a rinse step, sections were blocked with 2% FBS in TBS–0.2% Triton X-100 (blocking buffer). The indirect immunohistochemical procedure was carried out by incubating the sections with osteocalcin (OCN) polyclonal antiserum (1/100) (Millipore, Barcelona, Spain) in blocking buffer overnight at 4 ◦C. Sections were rinsed three times, then incubated with biotin-SP-conjugated donkey anti-rabbit F(ab) fragment (1/200) (Millipore, Barcelona, Spain) in blocking buffer for 1 h followed, after another rinse step, by incubation in peroxidase-conjugated streptavidin (1/300) (Millipore, Barcelona, Spain) for 1 h. Peroxidase activity was revealed in Tris–HCl buffer (0.05 M, pH 7.6) containing 0.05% of 3,3 -diaminobenzidine tetrahydrochloride (Sigma, Poole, UK) and 0.004% hydrogen peroxide. Reaction specificity was confirmed by replacing the specific antiserum with normal serum or by pre-adsorption of the specific antiserum with the corresponding antigen.

OCN staining was evaluated using computer-based image analysis software (ImageJ, NIH, Bethesda, MD, USA). OCN staining was measured by applying a fixed threshold to select for positive staining within the ROI. Positive pixel areas were divided by the total surface size (mm2) of the ROI. Values were normalized to those measured from blank scaffolds and are reported as relative staining intensities.

Statistical analysis was performed with SPSS.25 software. We compared the distinct treatments by means of a one-way analysis of variance (ANOVA) with a Tukey multiple comparison post-test. Significance was set at *p* < 0.05. Results are expressed as means ± SD.

#### **3. Results**

#### *3.1. Sandwich-Like System Characterization*

The characteristics of the microspheres, electrospun film, and core system are shown in Table 2.


**Table 2.** Characteristics of the component of the sandwich-like system. Microspheres: size and encapsulation efficiency. Electrospun film: thickness and porosity of the film and average diameter of the fibers). Core system: porosity freshly prepared and lyophilized and water uptake and mass loss after incubation in MilliQ water, 37 ◦C, and 25 rpm.

SEM image of the microspheres is shown in Figure 1A and the differential scanning calorimetry thermograms of the 17β-estradiol microspheres components are plotted in Figure 2. The glass transition temperature (Tg) of the mixture of polymers RG 203-S and RG 858 [4:1] was located at 52–58 ◦C, in the temperature range of the PLA (RG203-S) and PLGA (RG858). The DSC analysis of pure 17β-estradiol showed three endothermic peaks (Figure 2A), the first two at 118.1 ◦C and 174.4 ◦C, previously attributed to the partial and complete loss of hydrogen-bound water and reticular water, respectively. The third at 179.4 ◦C corresponds to the melting point [32]. This last peak, characteristic of the crystalline structure of 17β-estradiol, was not detected in the spectrum of the polymers and 17β-estradiol blend or in the thermogram of the microspheres (Figure 2B). These results indicated that 17β-estradiol was dissolved in the polymer by at least 8.5%.

**Figure 1.** SEM images (**A**) microspheres, (**B**) hydrogel, (**C**) hydrogel high magnification detail, (**D**) core system freshly prepared, (**E**) internal structure of the core system after 4 weeks incubation in water at 37 ◦C and 25 rpm, and (**F**) high magnification detail of Figure 1E. Scale bars: (**A**–**E**) 100 μm, (**C**) 1 μm, (**F**) 20 μm.

The integrity of the system was assayed throughout the 4-week test duration. The SEM images of the internal structure of the hydrogel (Figure 1B,C), the freshly prepared core system (Figure 1D), and after 4 weeks incubation showed that the microspheres were homogeneously dispersed in the hydrogel and are trapped in the core system during incubation (Figure 1E,F). The core system absorbed a significant amount of water during the first days of incubation which was maintained over time. Contrarily, the system lost little mass: less than 20% during the first week and approximately 35% after 4 weeks (Table 2). In addition, the core system flew well through the 20 G needle and the average dose discharged was 83.5 ± 6% of the loaded dose.

**Figure 2.** Differential scanning calorimetry thermograms. (**A**) Curve of pure 17β-estradiol. (**B**) Curves of PLA (RG203-S); PLGA (RG858), polymer blend (RG203-S:RG858, [4:1]), and polymer blend with 8.5% of 17β-estradiol, previously dissolved in DCM:MeOH (80:20) and the curve of the microspheres of 17β-estradiol (**B**).

#### *3.2. Osteogenic Di*ff*erentiation*

The alkaline phosphatase positive (ALP+) cell count, in the hydrogels pre-seeded with rMSCs and cultured in the osteogenic differentiation culture medium, showed a discrete and progressive increase in the number of cells between 7 and 21 days of culture in the hydrogels without and with nano-HAP; the number being significantly higher in those containing nano-HAP (Figure 3). Likewise, qualitatively, the cells presented, in the scaffolds with nano-HAP, greater intensity of color, suggesting greater ALP activity (Figure 3) than without nano-HAP.

**Figure 3.** Alkaline phosphatase (ALP) activity in rMSCs cultures. (**a**) Representative images from hydrogels 20 days after cultured showing the AP-positive (ALP+) cells' morphology (arrowheads) in hydrogel without (non-HAP) and with nanohydroxyapatite (HAP). (**b**) Graphic showing the number of APL+ cells/ microscopic field at different time points of analyses (7, 14, and 21 days) after culture in each system. Scale bar = 20 μm. The identical symbol on different bars indicates significant differences.

#### *3.3. Release Profiles of 125I-BMP-2 and 17*β*-Estradiol.*

Although the hydrogel provoked a strong reduction of the burst effect, the in vitro release of BMP-2 from the microspheres and from the core system showed a biphasic profile. During the first 24 h, approximately 7% of the protein was release from the core system versus 27% from the microspheres directly dispersed in the medium. Afterward, the release rate was kept in the range of 60 to 55 ng/day (Figure 4). The in vivo release profile was also biphasic, with a first phase that lasted up to 7 days whereas approximately 50% of the protein released. Then, the release rate was reduced to 70 ng/day, which is slightly higher than the in vitro rate.

**Figure 4.** BMP-2 release assays. In vivo release profile of 125I-BMP-2 from PLGA-microspheres in the system after implantation in the rat calvaria defect (*n* = 5). In vitro release of 125I-BMP-2 (incubation in water at 37 ◦C and 25 rpm) from PLGA-microspheres and from the PLGA microspheres dispersed in the system.

The in vitro release rate of 17β-estradiol in the aqueous medium depended on the formulation (Figure 5B); 100% and 70% of the 17β-estradiol dispersed in the hydrogel and in the electrospun film were detected in the medium after 4 weeks incubation, respectively. The release profile was characterized by a high burst effect; approximately half of the dose was released during the first day. By contrast, the 17β-estradiol release rate was extremely slow from the microspheres alone and from the microspheres included in the hydrogel (core system). Both release profiles were similar: less than 20% was delivered in 4 weeks. However, in the MeOH: water (50:50) medium there were not differences in the transfer profiles. The presence of MeOH modifies the solubility of 17β-estradiol, showing a strong burst effect that varied in a range between 70 and 85% in the first 24 h (Figure 5A).

**Figure 5.** In vitro 17-β-estradiol release profiles in different release media at 37 ◦C and 25 rpm. (**A**) In MeOH:water (50:50). (**B**) In aqueous solution of SLS 1%. (ED) 17β-estradiol dispersed in the core system; (EF) 17β-estradiol in the electrospinned film; (microspheres) 17β-estradiol pre-encapsulated in microspheres and (core system) 17β-estradiol microspheres in the hydrogel.

The estimated values of *De*ff and *h* for 17β-estradiol release in the different media are showed in Table 3. Although the value of *De*ff of 17β-estradiol in SLS was significantly lower compared to the MeOH:water (50:50), there were not differences for *h* regardless of the release media used. *h* is the contribution of hydrogel, as part of the boundary layer, to the whole mass transfer process, and the value of this coefficient should not change by varying the release media. The values of *R*<sup>2</sup> (Table 3), together with the comparison of experimental and predicted values of the released fractions shows a good fit of the data to the proposed model for both release media.

**Table 3.** Estimated values of effective diffusion coefficient (*De*ff) and mass transfer coefficient (*h*) for 17β-estradiol release in different media applying Equations (3)–(6).


*3.4. Histology, Immunohistochemical, and Histomorphometrically Evaluation*

#### 3.4.1. Osteoporotic Model

The osteoporotic model was assessed in both long bone (femur) and flat bone (calvaria). The histological analysis of calvaria showed evident changes in the structure and microarchitecture of the bone among the different experimental groups. Although the non-OP animals showed a normal bone structure in cortical bone (CB) and trabecular bone (TB), in the intercortical space (ICS), for the different OP models (DEX, OVX, and OD (Figure 6a), it was observed a progressive decrease in cortical bone thickness (CBT) and an increase in the intercortical space thickness (IST), being the group OD the one that presented greater alteration of the tissue bone structure (Figure 6b).

**Figure 6.** Validation of the OP model in calvarial bone. (**a**) Representative images in transversal section of calvaria in non-osteoporotic animals (non-OP) and in each of the different experimental models of osteoporosis tested showing the bone structure in each one. (**b**) Histomorphometric analysis of the cortical bone thickness and intercortical space thickness evaluated in calvaria in the different models of osteoporosis. CB: cortical bone; BMa: bone marrow; ICS: intercortical space; TB: trabecular bone. Scale bar = 100 μm. The identical symbol on different bars indicates significant differences.

The histological analysis of the femurs showed evident changes in the structure and microarchitecture of the bone among the different experimental groups. Although the non-OP animals showed a normal bone structure, in the different OP models (DEX, OVX and OD), structural changes were observed, both at the level of cortical and cancellous bone, showing a less compact bone and with a more porous structure (Figure 7a). The histomorphometric analysis revealed differences in the parameters measured in cancellous bone (Tb.N, Tb.Wi., and Tb.Sp.), with a significant reduction in all of them, in OVX and OD compared to the non-OP and DEX animals (Figure 7b–d). The cortical bone thickness (Ct.Wi.), although it showed a slight reduction in the groups (OVX and OD) with respect to the non-OP and DEX groups, was not significant (Figure 7e).

**Figure 7.** Validation of the OP model in long bone (femur). (**a**) The left column shows representative panoramic images in longitudinal section of rat femur in non-osteoporotic animals (non-OP) and in each of the different experimental models of osteoporosis tested. The right column shows high magnification images of the distal portion of the femur, showing differences in the microarchitecture of the bone in the different models. The column on the left shows detail of the boxed areas in which the structural characteristics of the compact and trabecular bone can be observed in each of the models. Histomorphometric analysis of the different parameters evaluated in femur in the different models of osteoporosis (**b**), Tb. N (mm), (**c**) Tb. Wi (μm), (**d**) Tb. Sp. (μm), and (**e**) Ct. Wi (μm). CB: cortical bone; TB: Trabecular bone; BMa: bone marrow. Scale bars: Left column 250 μm. Right column 50 μm. The identical symbol on different bars indicates significant differences.

#### 3.4.2. Calvarial Critical Size Defect

The histological analysis at the level of the calvarial defect showed a few new bone formations in the blank groups (B and B HAP), being limited to the margins of the defect in both, non-OP, and OP groups (Figure 8a). The groups implanted with BMP-2 + 17β-estradiol in the three different formulations showed a greater area of newly formed bone in the defect area, not only in the margins, but also in inner zone of the defect (Figure 8a). The newly formed bone in the different experimental groups of non-OP and OP animals showed a normal morphology and VOF staining, revealing significant areas of mineralization, slightly higher in the groups of non-OP animals (Figure 8a).

**Figure 8.** Repair process in calvarial defect. (**a**) Representative images in horizontal section of calvarial critical size defects in non-OP and OP rats showing the repair response at the defect level in the different experimental groups 12 weeks post-implantation. (**b**) Histomorphometrical analysis comparing of the degrees of repair (%) among the different experimental groups in non-OP and OP rats 12 weeks post-implantation. (**c**) Histomorphometric analysis showing the ratio between mature bone and immature bone (MB/IB) among the different experimental groups (**d**) and between non-OP and OP rats, estimated using VOF staining. Bars represent means ± SD (*n* = 4). The identical letter on different bars indicates significant differences. BMa: bone marrow; CT: connective tissue; NB: newly formed bone; DS: defect site. Scale bar = 1 mm. The identical symbol on different bars indicates significant differences.

The histomorphometric analysis showed little repair response in the blanks groups (B and B HAP) of non-OP and OP animals, with repair percent between 6 and 8%. The groups implanted with BMP-2 + 17β-estradiol in the three different formulations, on the contrary, showed a significantly higher repair response of 38–45%, with no differences being observed between non- OP and OP animals (Figure 8b).

The histomorphometric analysis of mature and immature bone showed a higher quantity of mature bone, and therefore with a greater degree of mineralization, in the experimental groups of non-OP with respect to OP animals. The ratio between mature and immature bone (MB/IB) showed individually higher values in all non-OP with respect to OP groups as well as on the whole, with values of 1.47 in non-OP animals and 0.99 in OP (Figure 8c,d).

The immunohistochemical analysis of osteocalcin (OCN), a marker of late osteogenesis and mineralization, showed a low immunoreaction in the blank groups (B and B HAP) in both non-OP and OP animals, with no differences between them (Figure 9a). In the groups implanted with BMP-2 + 17β-estradiol in the three different formulations, the immunoreaction was higher and more intense with respect to the blank groups, in this case being slightly higher in the non-OP animals (Figure 9a).

**Figure 9.** OCN relative expression. (**a**) Representative images in horizontal section of calvarial critical size defects in non-OP and OP rats showing OCN immunoreactivity in the different experimental groups 12 weeks post-implantation. (**b**) Histomorphometric analysis showing the relative staining values for OCN-ir. Bars represent means ± SD (*n* = 4). The identical letter on different bars indicates significant differences. CT: connective tissue; NB: newly formed bone. Scale bar = 100 μm. The identical symbol on different bars indicates significant differences.

The histomorphometric analysis confirmed the histological data, showing slightly higher relative staining values in the BMP-2 +17β-estradiol groups in non-OP animals (Figure 9b).

#### **4. Discussion**

In the present study, a BMP-2-17β-estradiol hydrogel system, with porosity of approximately 72%, was evaluated for regeneration of a critical size defect in rat calvaria. Although the system had already been characterized in terms of rheological behavior, porosity, interactions between components, mass transfer parameters, and cell viability, here the good injectability of the system was showed, and its characterization has been completed by testing the water uptake and mass loss as well as differentiation studies in cultures of osteoporotic rMSCs. In vitro release profiles of 17β-estradiol in two media and the in vitro and in vivo release profile of BMP-2 were also analyzed.

The osteogenic differentiation of osteoporotic rMSCs seeded on the system was assessed in order to test the effect of the incorporation of the nano-HAP on the cell behavior. The results showed greater ALP activity and a greater number of differentiated cells in the systems with nano-HAP. Therefore, nano-HAP systems were subsequently used in the in vivo experiments.

First, a histological evaluation of the femur and calvaria of rats suffering the three treatments for osteoporosis induction was carried out. OP is a systemic bone disease characterized by the increase of bone porosity, loss of bone mass and changes in the microstructure of the skeletal. Consequently, the OP population has an increased risk of fracture.

Despite the high number of OP studies and the several publications dedicated to tissue repair in non-OP specimens, very few reports devoted to bone defect regeneration in OP have been published. As OP might be primary post-menopausal or secondary, due to corticoid chronic administration, three animal models were used for OP induction: OVX, chronic glucocorticoid treatment, and the combination of both. In previous reports, a combination OD rat model was used. However, the high deterioration observed in the animals, the risk of induced additional disorders on the skeletal and the fact that bone loss reverses after corticoid stop [33] justify the bone histological study of the different treatment, in order to simplify the model and improve animal welfare. In general, OP condition is established through the analysis of long bones and lumbar spine but few data of the effect on the calvaria of the animals used as OP models are available [34]. Most of the publications on the regeneration of calvaria critical size defect in OP animals do not report the effects of OP in the calvaria [35,36]. In the present study, the data comparing the response of the femurs and calvarias to the three treatments revealed that the effect of OVX was similar to OD combination, consequently the fabricated system was tested in OVX rats.

As some authors have observed a delay in bone consolidation of OVX rats [37] and as this combination of BMP-2 and 17β-estradiol, formulated in microspheres, when applied to a critical calvaria defect, improved bone healing in OP rats, but the new bone was less mineralized [17,18], we tried to prolong the release of active substances to cover this delay. The drugs were incorporated to the hydrogel system pre-encapsulated in microspheres for prolonged controlled release. To reduce the release rate of the active substances the microspheres were prepared with a mixture of polymers, 25% of the RG 858 was incorporated to the RG 504 for BMP-2 microspheres as well as to the RG203-S for 17β-estradiol microspheres. RG 858 is a PLGA 85:15, of high molecular weight with a degradation rate lower than that of RG 504 and RG203-S. The BMP-2 release profiles showed a two-phase behavior with a weak burst effect that coincides with the period in which the system loses mass and uptakes a high amount of water. However, the burst effect of BMP-2 that can be seen from the microspheres was damped by the hydrogel, probably due to the interaction of the protein with the chemical groups of the HAP [6,13]. Afterwards, the second phases were practically parallel, which indicated a mass transfer process controlled by the access of water inside the microspheres, dissolution and diffusion of the protein throughout the porous of the polymeric matrix.

The release profile of 17β-estradiol as liposoluble drug in a MeOH:water (60:40) release medium was previously characterized [21]. By contrast, here two release media were used, MeOH:water (50:50) and an aqueous solution of SLS, because we suspected that the solvent affected the release rate of the lipophilic substance. In fact, the release of 17β-estradiol in MeOH:water was fast regardless of the formulations. However, the in vitro release of 17β-estradiol was formulation-dependent when an aqueous medium was used; a high decrease of the release rate of the drug from the microspheres was observed. As DSC results indicated, 17β-estradiol formed a solid solution in the microspheres, which indicates that the release process takes place by molecular diffusion of 17β-estradiol within the microspheres, governed by the partition coefficient, and consequently the aqueous medium dissolved it very slowly. In addition, the higher estimated values of *De*ff in the MeOH:water compared with the aqueous medium, confirmed the dependency on the release media. The calorimetric analysis of the electrospun sheet was not carried out because the amount of 17β-estradiol would have had to be increased to be detected and its characteristics would have been modified. Although 17β-estradiol is expected to also be dissolved in the polymer of the film, the large specific surface area that microfibers expose to the medium causes the drug to release rapidly. Similarly, 17β-estradiol dispersed in the hydrogel was very little retained. Obviously, neither media are physiological, but it seems more correct to use an aqueous medium to predict release in vivo. Although, with reservations, one would also expect a slightly faster release rate in vivo as the biological components present in the tissue could accelerate the drug release from the system.

Despite the beneficial role of the nano-HAP controlling the BMP-2 burst effect as well as its positive effect on the proliferation and osteogenic differentiation of rMSC, which justifies the use of nano-HAP in the system, the reparative effect of the blank scaffolds with and without nano-HAP was not enough to be considered useful. Unlike that observed in this study, other authors showed better bone repair in different bone defects practiced in osteoporotic goats implanted with a system of type I collagen containing nano-HAP than without [38]. By contrast, another study [6] found, as in the present study, that the use of nano-HAP and calcium sulfate bone substitute scaffolds in rat critical calvaria defect showed no effect on repair and mineralization at 8 and 12 weeks with respect to the empty defect. In both studies, the systems were loaded with BMP-2 combined with 17β-estradiol or zolendronic acid [6] that might abolish the effect of the HAP observed in vitro.

Although previously our group reported [17,18] a better result in bone regeneration of OP animals combining BMP-2 and 17β-estradiol, in this study, the repair effect observed has been similar to that observed in non-OP groups. The three combinations of BMP-2 with 17β-estradiol in each of the three formulations used showed the same effect at 12 weeks. However, the ratios of mature and immature bone in normal and osteoporotic animals showed significant differences, indicating that the quality of the repaired bone, at least after 12 weeks, was better in normal animals.

Although these results coincide with previous work, in the present study it seems that the mineralization of the bone formed slightly improved in the OP animals. The difference in the relative osteocalcin expression was not statistically significant. These results might suggest that a longer release of BMP-2 together with the composition of the system, presence of collagen and nanoHAP favor the mineralization process. In any case, it would be necessary in future to conduct studies aimed at discerning the role of each of these components in the process. However, we have not been able to reproduce the positive effect of 17β-estradiol combined with BMP-2 in a hydrogel composed of Pluronic, Tetronic and, cyclodextrin, with other scaffolds of different composition [18]. In addition, according to the present study, the fact that the different release profiles of 17β-estradiol had no effect on the repair of the defect indicates that 17β-estradiol, when applied locally, and regardless of the release rate (available dose) and of the obvious role that it plays on bone remodeling, does not justify its inclusion as active substance in the repair of bone defects neither in normal animals nor in osteoporotic ones. Therefore, new strategies and alternative drugs are currently being designed trying to accelerate mineralization of new bone in OP groups.

#### **5. Conclusions**

The prepared hydrogel system resulted to be easily injectable and solidified fast due to crosslinking of collagen and chitosan chains. The system helps control the burst effect of BMP-2 pre-encapsulated in PLGA microspheres, probably due to the nano-HAP. Release of 17β-estradiol from PLA-PLGA microspheres was more complex and is governed by the partition coefficient of the drug which is in solid dissolution in the microspheres. The system was biocompatible both in vivo and in vitro. However, the regenerative effect detected in the critical bone defect of both OP and non -OP rats was mainly due to the osteogenic effect of BMP-2 released in a controlled rate for 6 weeks. A delay in the mineralization of the new bone which fills the defect in OP animals was observed. 17β-estradiol released from different formulations and included in the system does not improve bone repair.

**Author Contributions:** Conceptualization, E.P.-H., A.D., and C.É.; methodology, E.P.-H., A.D., C.É., R.R., E.S.-M., P.G.-G; validation, R.R.; formal analysis, E.P.-H., A.D., C.É., R.R., P.G.-G.; investigation, E.P.-H., A.D., C.É., R.R. and P.G.-G.; resources, A.D. and C.É.; writing—original draft preparation, R.R., A.D., and C.É.; writing—review and editing, A.D. and C.É.; supervision, A.D. and C.É.; funding acquisition, A.D. and C.E. All authors have made a substantial contribution to the work.

**Funding:** This research and the APC was funded by Ministry of Science and Technology (MAT2014-55657-R).

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
