Journal of Controlled Release 198 (2015) 71–79

Contents lists available at ScienceDirect

Journal of Controlled Release

j ourna l homepage: www.e lsev ie r .com/ locate / jconre l

Development of collagen–hydroxyapatite scaffolds incorporating PLGAand alginate microparticles for the controlled delivery of rhBMP-2 forbone tissue engineering

Elaine Quinlan a,b,c,1, Adolfo López-Noriega a,b,c,d,1, Emmet Thompson a,b,c, Helena M. Kelly a,d,Sally Ann Cryan a,b,d, Fergal J. O'Brien a,b,c,⁎a Tissue Engineering Research Group, Dept. of Anatomy, Royal College of Surgeons in Ireland, Dublin, Irelandb Trinity Centre for Bioengineering, Trinity College Dublin, Irelandc Advanced Materials and Bioengineering Research (AMBER) Centre, RCSI & TCD, Dublin, Irelandd School of Pharmacy, Royal College of Surgeons in Ireland, Dublin, Ireland

⁎ Corresponding author.E-mail address: fjobrien@rcsi.ie (F.J. O'Brien).

1 These authors contributed equally to the work.

http://dx.doi.org/10.1016/j.jconrel.2014.11.0210168-3659/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 October 2014Accepted 21 November 2014Available online 4 December 2014

Keywords:rhBMP-2PLGACalcium alginateMicroparticlesCollagen based scaffoldsBone tissue engineering

The spatiotemporally controlled delivery of the pro-osteogenic factor rhBMP-2 would overcome most of the se-vere secondary effects linked to the products delivering this protein for bone regeneration.With this inmind, theaim of the present work was to develop a controlled rhBMP-2 release system using collagen–hydroxyapatite(CHA) scaffolds, which had been previously optimized for bone regeneration, as delivery platforms to producea device with enhanced capacity for bone repair. Spray-drying and emulsion techniques were used to encapsu-late bioactive rhBMP-2 in alginate and PLGA microparticles, with a high encapsulation efficiency. After incorpo-ration of these microparticles into the scaffolds, rhBMP-2 was delivered in a sustained fashion for up to 28 days.When tested in vitro with osteoblasts, these eluting materials showed an enhanced pro-osteogenic effect. Fromthese results, an optimal rhBMP-2 eluting scaffold composition was selected and implanted in critical-sizedcalvarial defects in a rat model, where it demonstrated an excellent healing capacity in vivo. These platformshave an immense potential in the field of tissue regeneration; by tuning the specific therapeutic molecule tothe tissue of interest and by utilizing different collagen-based scaffolds, similar systems can be developed for en-hancing the healing of a diverse range of tissues and organs.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction

Bone morphogenetic proteins (BMPs) are the most importantgrowth factors (GFs) in bone formation and healing. It is well knownthat they enable skeletal tissue formation during embryogenesis andthroughout adulthood as well as bone growth and repair [1]. To date,more than 30 B.P. have been reported and among them BMP-2, -4,and -7 are each known to stimulate new bone formation at ectopicsites in vivo in critical sized defects [2,3]. In particular, BMP-2 isknown as a primary GF in bone formation involved in the commitmentofmultipotent stem cells to the osteogenic lineage. Additionally, recom-binant human BMPs (rhBMPs) have been isolated by purification andcloning and are widely available for regenerative applications [4].

BMPs were first described in 1965 but it was not until 1998 thatJohnson andUrist found the possibility of their clinical application bypu-rifying endogenous bone and utilizing it in human patients to heal longbone non-unions and segmental defects [5,6]. rhBMP-2was approved by

the FDA in 2002 to be used in spinal fusion. Based on this, Medtronic'sINFUSE®, a collagen sponge soaked with rhBMP-2, was commercializedthe same year. This product is the market leader in the bone graft fieldwith 44% of the $1.9 billion market; however, it has been associatedwith numerous complications such as heterotopic ossification,osteolysis, increased neurological deficits, retrograde ejaculation, or can-cer. These serious side effects have been linked to the uncontrolled andoffsite release of rhBMP-2 [7,8]. Thus while problems exist, due to theoutstanding pro-osteogenic effect of this growth factor, a major focusof research is on developing biomaterials for the spatiotemporally con-trolled delivery of rhBMP-2 to the defect site. In this sense, a myriad ofpolymeric matrices, ceramic scaffolds or hydrogel based materials havebeen tested in the last decade, showing very promising results [9–13].

We have recently developed a porous collagen–hydroxyapatite(CHA) scaffold with immense potential for bone repair as it is biode-gradable, biocompatible, osteoconductive and osteoinductive [14–16].The localized and sustained delivery of rhBMP-2 to the defect siteonce implanted would further enhance the regenerative capacity ofthis material which may be applied to an expanded range of bone re-generation applications. One way to incorporate rhBMP-2 into theseCHA scaffolds would be to pre-load the protein into polymeric carriers

72 E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

that could be subsequently integrated into the CHA network during fab-rication, an approach that has been successfully used in other collagen-based materials [17–20]. Ideally, these carriers would ensure the stabil-ity of the rhBMP-2 during the fabrication process and allow its con-trolled delivery after implantation while maintaining the existing porearchitecture, mechanical properties and proven biological activity ofthe underlying scaffold. In addition, the scaffold should act as a reser-voir, protecting the therapeutics from degradation in vivo.

With this in mind, the present work evaluated the potential of algi-nate and poly(lactic-co-glycolic acid) (PLGA)microparticles as rhBMP-2carriers and assessed the effect of the incorporation of differentamounts of these polymers into the CHA scaffold. The rhBMP-2 releasekinetics from these constructs was determined and the bioactivity ofthe growth factor following both the polymeric encapsulation and in-clusion into the CHA scaffold was tested in vitro on pre-osteoblasticcells. An optimal rhBMP-2 elutingmicroparticles/CHA scaffold composi-tion was selected and its capability for enhancing bone tissue regenera-tion in vivo was determined by implanting it into a critical size calvarialdefect of young male rats.

2. Materials and methods

2.1. Materials

Sodium alginate, calcium chloride, sodium chloride, polyvinyl alco-hol (PVA), sodium hydroxide, sodium dodecyl sulfate (SDS) and PLGA(50:50, Mw 24,000–38,000) were purchased from Sigma Aldrich(Arklow, Ireland). Acetonitrile and dichloromethane (DCM) were pur-chased from Fisher Scientific (Ballycoolin, Ireland). Physiogel® was ob-tained as a kind gift from Braun Medical (Emmenbrucke, Switzerland).RhBMP-2 was purchased from R&D Systems Europe Ltd. (UK). Allchemicals were used as obtained, without further purification.

2.2. Microparticle fabrication and characterization

Alginate and PLGA were selected as the polymers for encapsulatingrhBMP-2 as they have beenwidely used for drug delivery purposes; in ad-dition, both are biodegradable, biocompatible and classified as generallyregarded as safe (GRAS)by the FDA [21–23]. rhBMP-2 eluting alginatemi-croparticles were manufactured by spray-drying. In this procedure, apolymeric feed solution is atomized and passed through a drying cham-ber, solvent from thedroplets is evaporated andparticulate powder is col-lected. In our study, rhBMP-2 was added to a sodium alginate solution(0.5% w/v) at a protein:polymer ratio of 1 μg/mg. This mixture wasspray-dried using a Buchi Mini Spray-Dryer B290 with the following pa-rameters: compressed air 5–8 bar, air flow rate 400–600 L/h, inlet tem-perature 140 °C, aspirator 80% of maximum capacity and pump flowrate 15%. Alginate microparticles were crosslinked after fabrication; themicroparticleswere suspended in acetonitrile, poured into 1.2% (w/v) cal-cium chloridewhile stirring for 10min afterwhich theywere vacuum fil-tered through a 0.45 μm nylon membrane, washed twice with 10 mLdistilled water and freeze-dried at−56 °C overnight.

PLGA (50:50) microparticles were fabricated by a water–oil–water(W/O/W) emulsion method adapted from a previously describedmethod [24]. Briefly, 0.5 mL of an aqueous solution of 16 μg of growthfactor was added to 125 μL of Physiogel® (80 mg/mL of succinylatedgelatin) which has been reported to protect proteins from degradation(W1-Phase) [25]. PLGA (375 mg) was dissolved in 5 mL of dichloro-methane (O-Phase). After mixing the W1-Phase with the O-Phase, themixture was ultrasonicated to obtain the primary W1/O emulsion. ThisW1/O emulsion was added dropwise while sonicating to 30 mL of anaqueous solution containing 1 g of polyvinyl alcohol (PVA) and 1.13 gof NaCl. The resulting W1/O/W2 emulsion was dispersed in 70 mL ofthe same aqueous solution. This emulsion was stirred for 4 h at roomtemperature and the resulting microparticles were collected by centri-fugation (5000 rpm, 10 min). The particles were washed three times

with 1.13% (w/v) NaCl and freeze-dried at −56 °C overnight. Blank,non-rhBMP-2 loaded alginate and PLGA microparticles were fabricatedby adding distilled water instead of growth factor solutions during therespective procedures.

The encapsulation efficiency of both processes was determined bydigesting the microparticles and analyzing the concentration ofrhBMP-2 in the medium. With this aim, 5 mg of alginate microparticleswas dissolved in 10 mL of 0.1 M sodium citrate under magnetic stirringfor 30 min and 20 mg of PLGA microparticles were added to 3 mL of a0.1 M NaOH, 5% w/v sodium dodecyl sulfate (SDS) solution in water at37 °C while shaking (50 rpm, 2 h). Digestion media were centrifuged(10,000 rpm, 2 min) and the supernatant was analyzed for protein con-tent using a rhBMP-2 ELISA kit (R&D Systems UK) according to themanufacturer's instructions. The encapsulation efficiency was calculat-ed using the following formula:

EE %ð Þ ¼ amount of rhBMP−2 in the microparticles=rhBMP−2 loadedð Þ�100:

Microparticles were morphologically characterized with ScanningElectron Microscopy (SEM) using a Zeiss Supra Variable Pressure FieldEmission SEM. For SEM observations, samples weremounted on copperstuds with carbon glue and then sputtered with gold.

To examine protein release profiles, 20 mg of rhBMP-2 microparti-cles dispersed in 2 mL of phosphate buffer solution (PBS) was placedin a water bath at 37 °C under agitation. At certain timepoints, sampleswere centrifuged (10,000 rpm, 5 min) and the release mediumwas re-moved and replaced by fresh PBS. The concentration of rhBMP-2 in themedia was determined by ELISA as above.

2.3. In vitro bioactivity assessment of rhBMP-2 encapsulated in themicroparticles

MC3T3-E1 murine pre-osteoblasts were used to study the effects ofthe released rhBMP-2 on osteoblast proliferation and differentiationin vitro since they exhibit similar properties as osteoprogenitor celllines including high alkaline phosphatase (ALP) activity [26] and formcalcified bone tissue [27]. Cells were cultured in α-MEM supplementedwith 10% fetal bovine serum (FBS), 1% L-glutamine and 2% penicillin/streptomycin; the media were changed every 3 days. Cells weretrypsinized (trypsin–EDTA 0.25%), seeded on 6 well plates at adensity of 30,000 cells perwell and cultured for 24 h in standard growthmedia followed by growth media supplemented with ascorbic acid 2-P50 μM, dexamethasone 100 nM, β-glycerophosphate 10 mM, FBS 10%,100 U/mL penicillin, and 100 μg/mL streptomycin under standard con-ditions (37 °C, 5% CO2).

In order to assess the bioactivity of the rhBMP-2 after the encapsula-tion process, 20 mg of alginate or PLGA microparticles was dispersed in2 mL PBS and placed in a water bath at 37 °C under agitation. After 4 h,the samples were centrifuged (10,000 rpm, 5 min), the release mediawere collected and the concentration of rhBMP-2 was determined byELISA. Release medium was added to the osteogenic cell culture mediaat a final concentration of 50 ng/mL of rhBMP-2 for each formulation(Alginate BMP-2 (50 ng/mL), PLGA BMP-2 (50 ng/mL)). To determinethe effect of the dissolution product of blank microparticles on MC3T3bioactivity, the same amount of media from blank microparticles thathad been agitating for 4 h at 37 °C was added to another set of cells(blank alginate, blank PLGA). As a positive control, non-encapsulatedrhBMP-2 at the same concentration (50 ng/mL) was added to the cul-ture media (BMP-2 (50 ng/mL)); this bioactive concentration is capableof inducing an osteogenic response from these cells [28]. Cells in the ab-sence of rhBMP-2 or any additional media other than osteogenic differ-entiation media were used as negative control (cells alone).

DNA was quantified from MC3T3-E1 cells cultured after 7 days inthe presence of the different media. At the endpoint of the study cellswere washed in PBS, lysed using cell lysis buffer according to the

73E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

manufacturer's instructions (SensoLyte pNPP Alkaline PhosphataseAssay Kit). Briefly, the cell suspension was centrifuged at 2500 ×g for10 min at 4 °C. The supernatant was analyzed for dsDNA contentusing the PicoGreen assay (Quanti-iT TM PicoGreen dsDNA MolecularProbes, OR, USA) according to the manufacturer's instructions. To con-vert DNA content into cell number the DNA content per sample was re-lated to the amount of DNA quantified per cell of a fixed number of cells.

In order to assess the ability of rhBMP-2 released from themicropar-ticles to enhance osteogenesis, alkaline phosphatase (ALP), an earlymarker for osteoblast differentiation, was quantified from MC3T3-E1cells cultured in the presence of the different media [29]. With this pur-pose, after 7 days of culture,MC3T3-E1 cellswere lysed as above and thesupernatant was collected for quantification of ALP by a method utiliz-ing p-nitrophenyl phosphate (pNPP), which is hydrolyzed by ALP toproduce a yellowproduct. The amount of colored product is proportion-al to the amount of enzyme in the reaction.

2.4. Scaffold fabrication and characterization

The collagen–hydroxyapatite (CHA) scaffold is produced using apatented process developed within our laboratory [30]. Briefly, 1.8 g ofmicro-fibrillar bovine tendon collagen was added to 320 mL of 0.5 Macetic acid solution. This suspension was blended at 15,000 rpm for90 min using an overhead blender in a reaction vessel which was keptat 4 °C using a circulation system to prevent heat denaturation of thecollagen. Aliquots of 10 mL of a solution of 3.6 g of HA in 40 mL of0.5 M acetic acid solution were added every 60 min and the resultingslurry was blended for additional 60 min. The slurry was degassed byvacuum to remove air bubbles that could produce uncontrolled porosityin the final scaffold.

In order to assess the influence of the incorporation of differentamounts of microparticles on the characteristics of the scaffolds, sus-pensions of 50, 100 or 200 mg of blank alginate or PLGA microparticlesin 1 mL distilled water were mixed with 9 mL of CHA slurry, for finalconcentrations of 0.5, 1.1 and 2.2% w/v (weight of microparticles/vol-ume of CHA slurry). The mixtures were lyophilized at −10 °C using aconstant cooling freeze-drying protocol. In order to enhance the me-chanical and enzymatic resistance of the scaffolds theywere crosslinkedunder an ultra-violet (subtype C, 365 nm λ) lamp for 15 min [31,32].Subsequently, the scaffolds were sterilized under UV light for 4 h(254 nm λ); this method was chosen for this study as it presented thelowest risk of material and/or rhBMP-2 damaging compared to othersterilization techniques (dehydrothermal treatment, gamma radiation,ethylene oxide or ethanol soaking). The resulting scaffolds were denot-ed as alginate–CHA and PLGA–CHA and compared with a blank scaffoldfabricated with 1 mL of distilled water instead of the microparticlesuspension.

The effect of microparticle incorporation on the mechanical proper-ties of scaffolds was assessed and compared with blank CHA scaffolds.With this aim, wet compression testing was performed using a uniaxialtensile testingmachine (Z050, Zwick/Roell, Ulm, Germany) fittedwith a5 N load cell. Disks of scaffolds (n = 4) with diameters of 6 mm and adepth of 3 mm, obtained with a biopsy punch, were pre-hydratedprior to transfer to the testing rig. Uniaxial compression testingwas per-formed up to a maximum strain of 10%. The modulus was calculatedfrom the slope of the stress–strain curve over the range of 2–5% strain.Compressive testing results were analyzed using an in-house designedMicrosoft Excel macro.

SEM characterization of the surface and the cross-section of the dif-ferent composite scaffolds, carried out for assessing the distribution ofthe microparticles through the walls of the scaffolds, was performedusing a Zeiss Supra Variable Pressure Field Emission Scanning ElectronMicroscope. Samples were mounted on copper studs and weresputteredwith gold prior to observation. In order to examine the effectsof the inclusion of themicroparticles on the porous structure of the scaf-folds, they were embedded in JB-4-glycolmethacrylate® (Polysciences,

Inc., UK) using a previously established method and then sectionedinto 10 μm slices using a microtome [33]. Scaffold sections weremounted on glass slides and stained with Toluidine blue. Digital imageswere captured at 10× magnification using a microscope (Eclipse 90i,Nikon, Japan) and a digital camera (DS Ri1, Nikon, Japan). To examinepore diameters the images were evaluated usingMATLAB® pore topol-ogy analyzer software, where original images were transformed into bi-nary images [34]. The pore size was defined as the diameter of a circlewith a cross-sectional area equivalent to that of the best fit ellipse gen-erated by the software (at least 50 pores per sectionwere considered foreach analysis).

The porosity of the scaffolds with and without microparticles wasalso calculated using the following equation:

% porosity ¼ 100x 1− ρ actual=ρ theoreticalð Þ½ �:

The actual density (ρ actual) of the scaffoldswas calculated by divid-ing the actual mass of the scaffolds to the calculated volume of the scaf-folds. The theoretical density (ρ theoretical) values were taken from theliterature. Based on the results detailed in the next section, scaffoldswith 0.5% alginate and 2.2% (w/v) PLGA microparticles were deemedoptimal and rhBMP-2 eluting scaffolds were prepared with thosecompositions.

To examine protein release profiles, scaffolds (6mmdiameter) wereplaced in 2 mL of PBS, ensuring sink conditions, in a shakingwater bathat 37 °C. At given timepoints, release media were removed, replaced byfresh PBS and the protein concentration was determined by ELISA asoutlined above.

2.5. In vitro bioactivity assessment of rhBMP-2 delivered from scaffolds

In order to analyze osteogenesis of the rhBMP-2 eluting scaffolds,MC3T3-E1 cells were cultured as detailed above. In a 24 well-plate,6mmdiameter scaffoldswere seeded drop-wisewith 25 μL of a cell sus-pension and then placed in an incubator (37 °C, 5% CO2) for 15min, thedisks were then turned over and another 25 μL of the cell suspensionwas added drop-wise for a total of 50,000 cells per scaffold. After15 min of incubation, 1 mL of media was added to each well and thescaffolds returned to the incubator. At day 3 and day 7, constructswere rinsed twice in PBS, flash frozen in liquid nitrogen and stored at−80 °C until analysis. To analyze the influence of rhBMP-2 eluting scaf-folds on cell proliferation, DNA was quantified after 3 and 7 days of cul-ture. At those timepoints, the cell seeded scaffoldswere homogenized inlysis buffer using a rotor-stator homogenizer (Omni International,Germany), the cell/scaffold suspension was centrifuged (2500 ×g,10 min, 4 °C) and the supernatant was collected and used for DNA anal-ysis as described above. ALP was quantified from MC3T3 cells culturedin direct contact with scaffolds after 7 days. Scaffolds were homoge-nized in lysis buffer and assayed according to themanufacturer's proto-col (SensoLyte pNPP Alkaline Phosphatase Assay Kit) as describedpreviously.

2.6. In vivo assessment of rhBMP-2 releasing scaffolds to enhance bonerepair

In order to assess the ability of BMP-2 eluting PLGA–CHA scaffolds topromote bone regeneration, scaffolds (n = 8 per group) were im-planted in 7 mm critical-sized calvarial defects [35] using adult maleWistar rats (mean weight 375 g) (Harlan, UK). For comparative pur-poses, a second group received blank non-eluting CHA scaffolds, whilea third empty defect group received no scaffold, serving as a negativecontrol. An animal licensewas granted by the Irish GovernmentDepart-ment of Health (Ref. B100/4416) for carrying out this in vivo study con-ducted in accordance with protocols approved by the Research EthicsCommittee of the Royal College of Surgeons in Ireland (REC ApprovalNo. 662).

Fig. 1. SEM micrographs of rhBMP-2 eluting alginate and PLGA microparticles showingparticles within the 1–10 μm range.

Fig. 2. rhBMP-2 release from the microparticles, showing different profiles for eachpolymer.

74 E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

The generation of a 7-mm-diameter critical-size calvarial defect, theimplantation of the composite materials in the defect site, the proce-dures for induction of anesthesia and the procedures for postoperativeanimal care were performed using methods established in the TissueEngineering Research Group (TERG) of the Royal College of Surgeonsin Ireland (RCSI) [36–38]. A trephine burr drill was used to create a7mm circular transosseous defect in the left parietal bone of the rat cal-varium. The dura was identified to confirm that a transosseous defecthad been created. Scaffolds were implanted into the defect, afterwhich the periosteal and superficial connective tissue was closed inlayers with 3–0 absorbable monofilament sutures (Monocryl™) and fi-nally a topical skin adhesive (n-butyl cyanoacrylate) was applied. Post-operatively, rats were given a subcutaneous injection of the reversalagent atipamezole 1.2 mg/kg and placed in an incubator at 27 °C untilfully recovered. Animalswere housed (2 per cage) in the Biomedical Re-search Facility at the RCSI with ad libitum access to water, food and en-vironmental enrichment. At 8 weeks post-implantation, the animalswere euthanized by CO2 asphyxiation and cervical dislocation. A20 mm × 20 mm segment of calvarium containing the defect wasresected using a dental saw and the explants retrieved were stored in10% neutral buffered formalin for 4 days and then transferred to PBSprior to decalcification.

Explantswere decalcified for a period of 5weeks by immersing themin a solution of ethylenediaminetetraacetic acid (EDTA) 14% (w/v) atpH 7.4 with renewal of the solution 3–4 times per week. Following de-calcification, specimens were dehydrated in a graded series of ethanolsolutions using an automated tissue processor, bisected and embeddedin single paraffin wax blocks. 7 μm sections were cut from each sampleusing a rotary microtome and sections were mounted on L-polysinecoated glass slides. Qualitative and quantitative histological examina-tions were performed on explants in order to assess the levels of boneformation in the defect site. Each sectionwas stained with Hematoxylinand Eosin (H&E). Hematoxylin stains cell nuclei purple, eosin stains ex-tracellular matrix pink and bone appears dark pink/red. Images of eachspecimen were acquired and digitized using transmitted light andepifluorescence microscopic visualization (Nikon Microscope Eclipse90i with Nis Elements software v3.06, Nikon Instruments Europe, TheNetherlands). Quantitative analysis was performed on n = 4 sectionsfrom each specimen. Histomorphometrical analysis (blind) was carriedout in order to quantify the healing response from the H&E-stainedsamples. The defectmarginwas identified and any new bonewas quan-tified by measuring the area of bone nucleation sites (BNS) in each sec-tion and calculating the mean total area of these sites per group. Thearea of new bone formationwas calculated using Nis Elements softwarev3.06 (Nikon Instruments Europe B.V., The Netherlands).

2.7. Statistical analysis

The data are represented as means ± standard error of the mean(SEM). Statistics were carried out using GraphPad Prism softwareusing a general linear model ANOVA with Bonferroni post hoc analysis.All in vitro experiments were performed with a sample size of 3 pertreatment group and in vivo analysis was performed with a samplesize of 8 per treatment group. Statistical significance was taken atp ≤ 0.05 unless otherwise stated.

3. Results

SEM images, displayed in Fig. 1, show that the spraydrying and dou-ble emulsion processes, used for the fabrication of alginate and PLGAcarriers respectively, yield microparticles with a size within the 1–10 μmrange, results thatwere corroborated by laser diffraction analyses(see Supporting information). The morphology of the materials is verydifferent depending on the fabrication procedure used; alginate micro-particles are irregular in shape with a smooth surface while PLGA

microparticles are spherical and porous. These features have beenpreviously described for materials fabricated using similar procedures[39,40].

The encapsulation efficiency of rhBMP-2 for the processes, deter-mined fromdigestedmicroparticles,was 45% for the spraydried alginatemicroparticles and 83% for the PLGA microparticles fabricated by thedouble emulsion technique. Fig. 2 shows the release profile of the en-capsulated protein from these microparticles. The delivery of rhBMP-2from alginate microparticles was characterized by an initial quickrelease for 7 days followed by a slower, sustained release for another7 days. In contrast, the amount of rhBMP-2 delivered from PLGAmicro-particles was significantly lower than alginate. After an initial burst onthe first day, the growth factor was released in a sustained manner for14 days. After the initial burst, the delivery kinetics of rhBMP-2 fromthe polymers followed the Higuchi kinetics model of controlled release(see Supporting information). At the end of the release study, alginateand PLGA microparticles had released 46% and 12% of their contentrespectively.

Having determined the release kinetics of rhBMP-2 from the micro-particles, the next step was to examine if the encapsulation process hadaffected the bioactivity of the growth factor. With this aim in mind, thebioactivity of MC3T3-E1 cells cultured with rhBMP-2 released from themicroparticles and with the same concentration of non-encapsulatedrhBMP-2 was assessed and compared. The results of this study aredepicted in Fig. 3. As shown in Fig. 3, the effect of the eluted rhBMP-2after 7 days of culture on cell number was not significantly differentfrom that of non-encapsulated growth factor. However, although non-significant, there was a clear decrease in cell number when osteoblastswere seeded with medium where blank alginate microparticles hadbeen suspended for 4 h compared to the cells alone control. The expres-sion of ALP by cells cultured for 7 days with media from the differentgroups studied is displayed in Fig. 2. While ALP expression in cells cul-tured with rhBMP-2 released from both alginate and PLGA microparti-cles was significantly lower than the concentration in media where

Fig. 3.MC3T3 E-1 cell number (A) and ALP concentration (B) in cell culturemedium after 7 days of culturewithout rhBMP-2 andwith rhBMP-2 released from themicroparticles comparedto negative (cells alone) and positive controls (non-encapsulated BMP-2). rhBMP-2 released from the microparticles had a pro-osteogenic effect, as demonstrated by the enhanced ALPactivity. *** denotes p b 0.05.

75E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

non-encapsulated rhBMP-2 was used, it was significantly higher thanthat of cells cultured without rhBMP-2, confirming that the elutedrhBMP-2 had retained its bioactivity after the encapsulation process. In-terestingly, when the growth factor was released from PLGA micropar-ticles, it had a significantly higher effect on ALP production by the cellscompared to rhBMP-2 delivered by alginate microparticles. Mediafrom blank PLGA and alginate microparticles had no effect on ALP levelsin this study when compared to the control.

Once it was confirmed that it was possible to encapsulate and re-lease bioactive rhBMP-2 from alginate and PLGA microparticles withthe procedures detailed above, we then studied the effects of incorpo-rating these microparticles into a CHA scaffold during its fabrication

Fig. 4. SEMmicrographs of cross-sections of CHA scaffolds with different concentrations of PLGtegration of the microparticles within the walls of the scaffold.

process. With this aim in mind, three different concentrations of blankmicroparticles (0.5%, 1.1% and 2.2% weight of microparticles/volume ofCHA slurry) were tested.

SEM characterization of these constructs displayed in Fig. 4 showsthat the microparticles are integrated within the walls of the scaffold.In a parallel study, fluorescently labeled microparticles were also incor-porated into the CHA scaffolds and the constructswere characterized byfluorescentmicroscopy. Results from this study corroborate that themi-croparticles were homogeneously distributed within the CHA network(see video in Supporting information).

An increase in the microparticle concentration within the CHA ma-trix led to a decrease in the pore diameter of the scaffolds, as shown in

A and alginate microparticles. Images, taken at 5000× magnification, demonstrate the in-

Fig. 5. Pore diameter (A), porosity (B) and compressive modulus (C) of CHA scaffolds in-corporating different concentrations of alginate or PLGA microparticles. The inclusion ofthe microparticles does not negatively affect the mechanical and structural properties ofthe scaffolds for bone tissue purposes. *** denotes p b 0.05.

Fig. 6. (A) Cumulative release of rhBMP-2 from alginate and PLGA–CHA scaffolds,(B) MC3T3-E1 cell number after 3 and 7 days of culture in the scaffolds and (C) ALPconcentration in the culture medium after 7 days of culture in the rhBMP-2 eluting scaf-folds and the non-eluting negative control scaffolds (CHA control). rhBMP-2 elutingPLGA–CHA scaffolds elicit an enhanced pro-osteogenic effect on osteoblasts. *** denotesp b 0.01, ** denotes p b 0.02 and * denotes p b 0.05.

76 E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

Fig. 5A. This effect was also obviouswhen cross-sections of the scaffoldswere observed by SEM at low magnification (see Supporting informa-tion). Although the average pore diameter of scaffolds with 1.1% and2.2% microparticles was significantly lower than CHA scaffolds, it wasover 75 μm in all the studied materials. The incorporation of the micro-particles also induced a concentration-dependent reduction in the po-rosity of the scaffolds (Fig. 5B). However, porosity values were over96% for all the CHA-microparticle scaffolds well above that typically de-sired to promote cell infiltration and subsequent vascularizationin vivo.[36,41,42] The inclusion of the microparticles within the CHAstructure did not have a negative effect on the mechanical propertiesof the scaffolds. Indeed, as shown in Fig. 5C, the compressive modulusof 0.5% and 2.2% alginate–CHA scaffolds was significantly higher thanthat of the blank-CHA control.

Based on these results, and in order to have a similar amount ofrhBMP-2 released from the scaffolds, CHA scaffolds incorporating 0.5%alginate or 2.2% PLGA rhBMP-2 eluting microparticles were fabricated(for a final content of 1.5 μg and 6.6 μg rhBMP-2/6mmdiameter scaffoldrespectively) and their rhBMP-2 release profiles and in vitro bioactivityinvestigated. Fig. 6A shows the release profiles of rhBMP-2 from the twodifferent scaffolds. The growth factor delivery from the constructs fol-lows a different trend to that observed for their corresponding micro-particles alone, i.e., a high burst release followed by a sustained

release for only 7 days for BMP-2 eluting alginate–CHA scaffold and asustained release through the whole study for the BMP-2 elutingPLGA–CHA scaffold. In addition, while PLGA based constructs releaseda similar amount of rhBMP-2 to that delivered from the microparticlesalone in 28 days (ca. 12%), BMP-2 eluting alginate–CHA scaffolds onlyreleased 15% of the encapsulated growth factor. Interestingly, thealginate–CHA GF release follows the same kinetics model as the micro-particles alone after the initial burst (Higuchi model), however, the de-livery of rhBMP-2 fromPLGA–CHAfitswith a zero-order release kineticsprofile from the beginning of the study, demonstrating the sustaineddelivery of the protein with time (see Supporting information).

The results of the bioactivity tests of these scaffolds are summarizedin Fig. 6B and C. When compared to the non-eluting CHA control, thereis a significant increase in the number of cells after 3 and 7 days of cul-ture in both rhBMP2-eluting and blank PLGA–CHA scaffolds. In addition,after 7 days of culture, ALP expression in cells cultured in both rhBMP-2eluting PLGA–CHA and alginate–CHA scaffolds is higher than in the restof analyzed groups. However, comparing between growth factor releas-ing scaffolds, ALP expression is significantly higher in cells cultured inrhBMP-2 eluting PLGA–CHA scaffolds compared to those on alginate–CHA scaffolds.

Taken together, these results confirm that the rhBMP-2 eluted fromPLGA–CHAscaffolds induces a higher osteogenic response in osteoblastscompared to that delivered from alginate–CHA scaffolds. Thus, thehealing potential of these scaffolds was assessed in vivo. The results ofthis study, whichwas carried out by implanting thematerials in a crani-al critical defect in male Wistar rats for 8 weeks, are displayed in Fig. 7.

Fig. 7. (A) Representative images of sections of explants from7mm calvarial defects inWistar rats left empty or treatedwith CHA scaffolds or rhBMP-2 eluting PLGA–CHA scaffolds during8weeks and quantification of new area of bone. Qualitatively and (B) quantitatively, enhanced bone healing is observed in defectswhere the growth factor eluting scaffoldwas implanted.** denotes p b 0.01 and * denotes p b 0.05.

77E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

The histological images taken from sections of the explants of the de-fect site at the end of the in vivo study demonstrate different levels ofhealing among the tested groups. Thin fibrotic tissue, with some boneat the defect edges can be observed in samples where the defect wasleft empty demonstrating the critical nature of the defect created. De-fects treated with non-eluting blank CHA control scaffolds show a thinbone formation at the defect edges and in the center of the defect,with a good cell infiltration. Most encouragingly, rhBMP-2 eluting scaf-folds induced the formation of areas of bone throughout the defectspace and within the boundaries as evidenced bymore intense stainingindicative of bone nucleation sites. In these samples, some areas of ex-tracellularmatrix deposition could be observed. In addition, the scaffoldwas almost completely resorbed. The quantitative histomorphometricalanalysis of the area of newbone formed in all groups (Fig. 7B) supportedthese observations revealing significantly higher levels of new bone for-mation in animals treatedwith the rhBMP-2 eluting PLGA–CHA scaffoldafter 8 weeks compared to the empty defect (**p b 0.001) and CHA con-trol (*p b 0.01).

4. Discussion

The aim of the present study was to assess if rhBMP-2 eluting algi-nate and PLGA microparticles could be incorporated into CHA scaffoldsas ameans to control the release of the growth factor resulting in a ther-apeutic with enhanced capacity for bone repair. With this in mind, theeffect of incorporating different amounts of these microparticles onthe properties of the scaffolds was studied and the bioactivity of thegrowth factor in vitro post-incorporation into the scaffolds was ana-lyzed and shown to be retained. The extensive characterization carriedout in this work allowed for the design of a rhBMP-2 eluting PLGAmicroparticle-CHA scaffold composite with optimal structural charac-teristics for bone tissue regeneration. The regenerative capability ofthis controlled-delivery scaffold was tested in vivo, confirming its supe-rior bone healing properties.

The processes used formicroparticle fabrication, spray-drying for al-ginate and double emulsion for PLGA yielded microparticles with theexpectedmicrometric size. The loading of the growth factor into themi-croparticles was achieved with high encapsulation efficiencies. This isan important feature in the system as it minimizes the high cost ofrhBMP-2. However, only a fraction of the encapsulated protein was de-livered during the in vitro release studies. This might suggest that theprotein was degraded during the study in the release medium, or, inthe case of PLGA microparticles, rhBMP-2 was likely being retained by

the succinylated gelatine which was used for increasing the efficiencyof the growth factor loading process. Similar results have been obtainedin previous studies where this additive has shown to delay the deliveryof therapeutic proteins from PLGA microparticles [25]. Either way, theamount of growth factor released from themicroparticleswas consider-ably higher than the concentration range considered optimal for achiev-ing a pro-osteogenic effect [28]. In addition, rhBMP-2 encapsulated inthemicroparticles induced a pro-osteogenic response in preosteoblasts.The fact that the eluted growth factor did not elicit a pro-osteogenic ef-fect comparable to that of non-encapsulated rhBMP-2 directly added tothe cells suggests the degradation of the protein during the encapsula-tion process, probably due to thermal (spray-drying) and mechanical(double emulsion) stresses. One limitation of the present study wasthat the conformational integrity of rhBMP-2 post-encapsulation, priorto the bioactivity tests, was not assessed. However, more importantlyfrom a therapeutic perspective, despite this, released growth factorsled to a significant increase in ALP activity by the osteoblasts, which isindicative of an enhancement of cell differentiation towards a pro-osteogenic lineage. Interestingly, it was noted that ALP activity in cellsgrownwith rhBMP-2 eluted from PLGAmicroparticles was significantlyhigher than that of cells cultured in the presence of growth factor re-leased from the alginate microparticles. This result possibly suggeststhat the high temperatures from the spray-drying process may bedegrading BMP-2. A reduction in cell number was reported when oste-oblasts were seeded with the dissolution products of ionicallycrosslinked calcium alginate microparticles, which may be attributedto the release of calcium ions into the surrounding media as a result ofexchange with cations and protons from the medium [43]. Thus, thehigher ionic content of the medium, as well as the changes of pHresulting from the dissolution process could be affecting cell numbers.

Once itwas determined that it was possible to load bioactive rhBMP-2 into these polymeric carriers, a series of CHA scaffolds incorporatingdifferent amounts of PLGA and alginate microparticles was fabricatedwith the aim of determining the effect of the inclusion of the carrierson the properties of the materials. Specifically, CHA scaffolds with0.5%, 1.1% and 2.2% (weight of microparticles/volume of slurry) werefabricated and compared to blank CHA scaffolds. The microparticleswere incorporated into the scaffolds in an intermediate fabricationstep by mixing them with the CHA slurry, prior to the final freeze-drying step. Themicroscopic characterization of the scaffolds confirmedthat it is possible to integrate themicroparticles homogenously throughthewalls of the CHAnetwork. In addition, irrespective of the polymer orthe concentration of the microparticles, the resulting materials have

78 E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

high porosity, over 96%, with pore diameters over 75 μm, characteristicsthat are essential for bone tissue growth [36,41,42]. A small decrease inthe average pore size of the freeze-dried materials was observed withincreasing microparticle content. This result may be attributed to themicroparticles acting as additional nucleation sites for ice crystal forma-tion during lyophilisation from which pores are formed resulting insmaller, more compact porous architecture. The incorporation of PLGAmicroparticles did not negatively affect the scaffold mechanical proper-ties when compared tomicroparticle-free scaffolds. Interestingly, an in-crease in compressive modulus was actually seen in the alginate–CHAscaffolds with increasing concentrations of microparticles which maybe attributed to the hydrophilic microparticles reinforcing the scaffoldsin a hydrophilic environment. Additionally, the porous architecture ofthe scaffold may have influenced this result since high MP concentra-tions reduced scaffold pore size resulting in a denser and tightly packedmaterial. Based on these positive results, CHA scaffolds incorporating0.5% rhBMP-2 eluting alginate or 2.2% rhBMP-2 eluting PLGA micropar-ticles were fabricated, in order to compare the protein release kineticsand the in vitro bioactivity of materials potentially releasing the sameamount of growth factor.

Several differences were observed when studying the release kinet-ics of rhBMP-2 from the scaffolds compared to their corresponding mi-croparticles alone. For alginate, a higher concentration of the proteinwas released as a burst from the alginate–CHA constructs, and a muchlower concentration was released by the end of the study (only 15% ofthe incorporated rhBMP-2). These two phenomena might be explainedby the degradation of the outer layers of the microparticles when theywere immersed in the acidic slurry before lyophilisation. The majorityof the encapsulated rhBMP-2 is probably located in these outer layersof the microparticles; the aggregation of encapsulated molecules onthe surface of spray-dried microparticles is an expected consequenceof this fabrication process [44]. Thus, the partial degradation of the mi-croparticles may have released part of the encapsulated growth factorto the CHA matrix, with a double effect. On the one hand, free, non-encapsulated rhBMP-2 might have diffused quicker to the surroundingmedium leading to the burst delivery. On the other hand, the strong in-teraction of the rhBMP-2 with the hydroxyapatite component of theslurry may be hindering the delivery of a fraction of the protein,resulting in a much lower amount of protein released compared to themicroparticles alone. These results suggest that alginatemay be a usefulcarrier for other growth factors rather than BMP-2. In the case ofrhBMP-2 eluting PLGA–CHA scaffolds, the release is more sustainedfor the duration of the study, which may be due to the integration ofthemicroparticleswithin the scaffold; the CHAmatrix is acting as an ad-ditional diffusive barrier for the delivery of the protein.

Similarly rhBMP-2 eluting PLGA–CHA scaffolds induced a higherpro-osteogenic response in osteoblasts than rhBMP-2 eluting algi-nate–CHA scaffolds; rhBMP-2 released from PLGA–CHA scaffolds wasshown to significantly enhance the osteogenic differentiation of thecells. Interestingly, non-eluting blank PLGA–CHA scaffolds also inducedan increase in cell numbers compared to CHA scaffolds, suggesting ananabolic effect of the PLGA–CHA composites. This phenomenon waspreviously observed in PLGA–collagen based hydrogel materials andcan be attributed to the negatively charged carboxylate groups that re-sult from PLGA degradation, attracting calcium and facilitating en-hanced mineralization [45]. Taken together, these results demonstratethat it is possible to load and release, in a controlled fashion, a bioactivedosage of rhBMP-2 from CHA scaffolds by using PLGA as a polymericcarrier of the protein. The potential of these materials for enhancingbone tissue regeneration was then tested in vivo.

Qualitatively and quantitatively, the results from the in vivo study,carried out by implanting the scaffolds in a critical-sized bone defect,confirm that the rhBMP-2 eluting PLGA–CHA scaffolds enhance bone re-generation. Significantly greater bone regenerationwas observed in thisgroup when compared to empty defects and CHA control scaffolds, cor-roborating the bioactivity of the encapsulated rhBMP-2. This result also

reinforces the beneficial role of PLGA as growth factor carrier, as it isprotecting the protein from degradation in vivo, which otherwisewould occur very rapidly [46]. Histological analysis demonstrated theosteoconductivity of these materials, as newly formed bone could bedistinguished throughout the defect site. In addition the majority ofthe rhBMP-2 eluting PLGA–CHA scaffold has been infiltrated with cellsat 8weeks post-implantation. Furthermore, quantitativemeasurementsof bone formation indicated significantly higher levels of new bone inthe defects containing rhBMP-2 eluting PLGA–CHA scaffolds and to alesser extent the non-rhBMP-2 eluting CHA scaffold. This study thusconfirms the superior regenerative properties of these growth factoreluting scaffolds. According to previous studies, these results could beimproved even further if these materials showed an initial burst ofgrowth factor prior to the sustained release. It has been demonstratedthat BMP-2 releasing scaffolds that display initial burst release followedby a sustained release of BMP-2 have been shown to promote more sig-nificantly enhanced bone regeneration in some studies compared to asustained release alone [47]. This can be attributed to the burst releaseof BMP-2 promoting the recruitment of osteoprogenitors into the ma-trix. In this sense, a CHA scaffold with both alginate and PLGA growthfactor carriers could be an ideal platform for obtaining such releaseprofile.

The present work demonstrates that, with the procedures detailedabove, it is possible to fabricate growth-factor eluting microparticlesand incorporate them into a collagen-based scaffold fabricated by afreeze-drying process for the sustained and controlled delivery ofrhBMP-2, leading to materials with superior regenerative capabilities.This study has focused on a growth factor/scaffold composition and ar-chitecture optimized for bone tissue engineering. However, the poten-tial of the platform described in the present work is enormous. Bytuning the therapeutic molecule and the chemical nature of the matrix,similar systems can be developed for enhancing the healing of very di-verse tissues such as cartilage, cornea or peripheral nerves, which arealso regenerated using freeze-dried collagen based scaffolds [48–50].

5. Conclusions

The development of collagen–hydroxyapatite scaffolds incorporat-ing PLGA and alginate microparticles for the sustained and localized de-livery of rhBMP-2 was investigated. The inclusion of high amounts ofbothmicroparticles had aminimal effect on the structural andmechan-ical properties of the scaffolds. Bioactive rhBMP-2 was successfully re-leased from both sets of microparticles within the scaffolds but PLGA–collagen hydroxyapatite scaffolds showed a higher pro-osteogenic ef-fect in vitro than the growth factor delivered from alginate–collagen hy-droxyapatite scaffolds. The regenerative effect was demonstrated asPLGA–collagen hydroxyapatite scaffolds led to significantly higherbone regeneration after 8 weeks of implantation in a critical-sizedbone defect. These platformshave the potential to be used for deliveringtherapeutics to various tissues in addition to bone by adjusting the com-position of the scaffold and the growth factor released.

Acknowledgments

This publication has emanated from research conducted with the fi-nancial support of Science Foundation Ireland under Grant SFI 12/TIDA/B2383. A. L.-N. has been also supported by Enterprise Ireland under pro-ject number CFTD/2009/0127. E.Q. & F.J.O'B acknowledge the EuropeanResearch Council for funding (239685-CollRegen-ERC-2009-STG).

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jconrel.2014.11.021.

79E. Quinlan et al. / Journal of Controlled Release 198 (2015) 71–79

References

[1] A.H. Reddi, Regulation of cartilage and bone differentiation by bone morphogeneticproteins, Curr. Opin. Cell Biol. 4 (1992) 850–855.

[2] C.A. Kirker-Head, Potential applications and delivery strategies for bone morphoge-netic proteins, Adv. Drug Deliv. Rev. 43 (2000) 65–92.

[3] P.H. Warnke, I.N.G. Springer, J. Wiltfang, Y. Acil, H. Eufinger, M. Wehmoller, P.A.J.Russo, H. Bolte, E. Sherry, E. Behrens, H. Terheyden, Growth and transplantation ofa custom vascularised bone graft in a man, Lancet 364 (2004) 766–770.

[4] E.A.Wang, V. Rosen, J.S. Dalessandro, M. Bauduy, P. Cordes, T. Harada, D.I. Israel, R.M.Hewick, K.M. Kerns, P. Lapan, D.P. Luxenberg, D. McQuaid, I.K. Moutsatsos, J. Nove,J.M. Wozney, Recombinant human bone morphogenetic protein induces bone-formation, Proc. Natl. Acad. Sci. U. S. A. 87 (1990) 2220–2224.

[5] M.R. Urist, Bone — formation by autoinduction, Science 150 (1965) (893-&).[6] E.E. Johnson, M.R. Urist, One-stage lengthening of femoral nonunion augmented

with human bone morphogenetic protein, Clin. Orthop. Relat. Res. (1998) 105–116.[7] N.E. Epstein, Pros, cons, and costs of INFUSE in spinal surgery, Surg. Neurol. Int. 2

(2011) 10.[8] N.E. Epstein, Complications due to the use of BMP/INFUSE in spine surgery: the ev-

idence continues to mount, Surg. Neurol. Int. 4 (2013) S343–S352.[9] S.H. Jun, E.J. Lee, T.S. Jang, H.E. Kim, J.H. Jang, Y.H. Koh, Bone morphogenic protein-2

(BMP-2) loaded hybrid coating on porous hydroxyapatite scaffolds for bone tissueengineering, J. Mater. Sci. Mater. Med. 24 (2013) 773–782.

[10] D. Ben-David, S. Srouji, K. Shapira-Schweitzer, O. Kossover, E. Ivanir, G. Kuhn, R.Muller, D. Seliktar, E. Livne, Low dose BMP-2 treatment for bone repair using aPEGylated fibrinogen hydrogel matrix, Biomaterials 34 (2013) 2902–2910.

[11] S. Lu, J. Lam, J.E. Trachtenberg, E.J. Lee, H. Seyednejad, J.J.J.P. van den Beucken, Y.Tabata, M.E. Wong, J.A. Jansen, A.G. Mikos, F.K. Kasper, Dual growth factor deliveryfrom bilayered, biodegradable hydrogel composites for spatially-guidedosteochondral tissue repair, Biomaterials 35 (2014) 8829–8839.

[12] J. van der Stok, H.N. Wang, S.A. Yavari, M. Siebelt, M. Sandker, J.H. Waarsing, J.A.N.Verhaar, H. Jahr, A.A. Zadpoor, S.C.G. Leeuwenburgh, H.Weinans, Enhanced bone re-generation of cortical segmental bone defects using porous titanium scaffolds incor-porated with colloidal gelatin gels for time- and dose-controlled delivery of dualgrowth factors, Tissue Eng. A 19 (2013) 2605–2614.

[13] C.L. Huang, W.L. Lee, J.S.C. Loo, Drug-eluting scaffolds for bone and cartilage regen-eration, Drug Discov. Today 19 (2014) 714–724.

[14] J.P. Gleeson, N.A. Plunkett, F.J. O'Brien, Addition of hydroxyapatite improves stiff-ness, interconnectivity and osteogenic potential of a highly porous collagen-basedscaffold for bone tissue regeneration, Eur. Cells Mater. 20 (2010) 218–230.

[15] F.G. Lyons, J.P. Gleeson, S. Partap, K. Coghlan, F.J. O'Brien, Novel microhydroxyapatiteparticles in a collagen scaffold: a bioactive bone void filler? Clin. Orthop. Relat. Res.472 (2014) 1318–1328.

[16] C.M. Murphy, A. Schindeler, J.P. Gleeson, N.Y.C. Yu, L.C. Cantrill, K. Mikulec, L.Peacock, F.J. O'Brien, D.G. Little, A collagen–hydroxyapatite scaffold allows forbinding and co-delivery of recombinant bone morphogenetic proteins andbisphosphonates, Acta Biomater. 10 (2014) 2250–2258.

[17] Z.S. Patel, M. Yamamoto, H. Ueda, Y. Tabata, A.G. Mikos, Biodegradable gelatin mi-croparticles as delivery systems for the controlled release of bone morphogeneticprotein-2, Acta Biomater. 4 (2008) 1126–1138.

[18] M. Biondi, L. Indolfi, F. Ungaro, F. Quaglia, M.I. La Rotonda, P.A. Netti, Bioactivatedcollagen-based scaffolds embedding protein-releasing biodegradable microspheres:tuning of protein release kinetics, J. Mater. Sci. Mater. Med. 20 (2009) 2117–2128.

[19] F. Ungaro, M. Biondi, I. d'Angelo, L. Indolfi, F. Quaglia, P.A. Netti, M.I. La Rotonda,Microsphere-integrated collagen scaffolds for tissue engineering: effect of micro-sphere formulation and scaffold properties on protein release kinetics, J. Control. Re-lease 113 (2006) 128–136.

[20] C. Borselli, F. Ungaro, O. Oliviero, I. d'Angelo, F. Quaglia, M.I. La Rotonda, P.A. Netti,Bioactivation of collagen matrices through sustained VEGF release from PLGA mi-crospheres, J. Biomed. Mater. Res. A 92A (2010) 94–102.

[21] R.A. Jain, The manufacturing techniques of various drug loaded biodegradablepoly(lactide-co-glycolide) (PLGA) devices, Biomaterials 21 (2000) 2475–2490.

[22] S. Takka, F. Acarturk, Calcium alginate microparticles for oral administration: I: Ef-fect of sodium alginate type on drug release and drug entrapment efficiency, J.Microencapsul. 16 (1999) 275–290.

[23] T.P. Richardson, M.C. Peters, A.B. Ennett, D.J. Mooney, Polymeric system for dualgrowth factor delivery, Nat. Biotechnol. 19 (2001) 1029–1034.

[24] X. Wang, E. Wenk, X. Zhang, L. Meinel, G. Vunjak-Novakovic, D.L. Kaplan, Growthfactor gradients via microsphere delivery in biopolymer scaffolds for osteochondraltissue engineering, J. Control. Release 134 (2009) 81–90.

[25] L. Meinel, O.E. Illi, J. Zapf, M. Malfanti, H.P. Merkle, B. Gander, Stabilizing insulin-likegrowth factor-I in poly(D,L-lactide-co-glycolide) microspheres, J. Control. Release 70(2001) 193–202.

[26] V. Kartsogiannis, K.W. Ng, Cell lines and primary cell cultures in the study of bonecell biology, Mol. Cell. Endocrinol. 228 (2004) 79–102.

[27] H. Sudo, H.A. Kodama, Y. Amagai, S. Yamamoto, S. Kasai, In vitro differentiation andcalcification in a new clonal osteogenic cell-line derived from newborn mousecalvaria, J. Cell Biol. 96 (1983) 191–198.

[28] Y. Takuwa, C. Ohse, E.A. Wang, J.M. Wozney, K. Yamashita, Bone morphogeneticprotein-2 stimulates alkaline-phosphatase activity and collagen-synthesis in cul-tured osteoblastic cells, Mc3t3-E1, Biochem. Biophys. Res. Commun. 174 (1991)96–101.

[29] A. Yamaguchi, T. Komori, T. Suda, Regulation of osteoblast differentiation mediatedby bone morphogenetic proteins, hedgehogs, and Cbfa1, Endocr. Rev. 21 (2000)393–411.

[30] F.J. O'Brien, J. Gleeson, N. Plunkett, A collagen/hydroxyapatite composite scaffold,and the process for the production thereof, WO 2008096334 A3, 2008.

[31] K.S. Weadock, E.J. Miller, L.D. Bellincampi, J.P. Zawadsky, M.G. Dunn, Physical cross-linking of collagen-fibers — comparison of ultraviolet-irradiation anddehydrothermal treatment, J. Biomed. Mater. Res. 29 (1995) 1373–1379.

[32] K.S. Weadock, E.J. Miller, E.L. Keuffel, M.G. Dunn, Effect of physical crosslinkingmethods on collagen-fiber durability in proteolytic solutions, J. Biomed. Mater.Res. 32 (1996) 221–226.

[33] F.J. O'Brien, B.A. Harley, I.V. Yannas, L. Gibson, Influence of freezing rate onpore structure in freeze-dried collagen-GAG scaffolds, Biomaterials 25 (2004)1077–1086.

[34] M.G. Haugh, C.M. Murphy, F.J. O'Brien, Novel freeze-drying methods to produce arange of collagen–glycosaminoglycan scaffoldswith tailoredmean pore sizes, TissueEng. C Meth. 16 (2010) 887–894.

[35] F.G. Lyons, A.A. Al-Munajjed, S.M. Kieran, M.E. Toner, C.M. Murphy, G.P. Duffy, F.J.O'Brien, The healing of bony defects by cell-free collagen-based scaffolds comparedto stem cell-seeded tissue engineered constructs, Biomaterials 31 (2010)9232–9243.

[36] F.G. Lyons, A.A. Al-Munajjed, S.M. Kieran, M.E. Toner, C.M. Murphy, G.P. Duffy, F.J.O'Brien, The healing of bony defects by cell-free collagen-based scaffolds comparedto stem cell-seeded tissue engineered constructs, Biomaterials 31 (2010)9232–9243.

[37] M. Alhag, E. Farrell, M. Toner, N. Claffey, T.C. Lee, F. O'Brien, Evaluationof early healing events around mesenchymal stem cell-seeded collagen-glycosaminoglycan scaffold. An experimental study in Wistar rats, Oral Maxillofac.Surg. 15 (2011) 31–39.

[38] M. Alhag, E. Farrell, M. Toner, T.C. Lee, F.J. O'Brien, N. Claffey, Evaluation of the abilityof collagen-glycosaminoglycan scaffolds with or withoutmesenchymal stem cells toheal bone defects in Wistar rats, Oral Maxillofac. Surg. 16 (2012) 47–55.

[39] X.Wang, E.Wenk, X. Hu, G.R. Castro, L. Meinel, X.Wang, C. Li, H. Merkle, D.L. Kaplan,Silk coatings on PLGA and alginate microspheres for protein delivery, Biomaterials28 (2007) 4161–4169.

[40] K. Bowey, B.E. Swift, L.E. Flynn, R.J. Neufeld, Characterization of biologically activeinsulin-loaded alginate microparticles prepared by spray drying, Drug Dev. Ind.Pharm. 39 (2013) 457–465.

[41] D.W. Hutmacher, Scaffolds in tissue engineering bone and cartilage, Biomaterials 21(2000) 2529–2543.

[42] C.M. Murphy, F.J. O'Brien, Understanding the effect of mean pore size on cell activityin collagen–glycosaminoglycan scaffolds, Cell Adhes. Migr. 4 (2010) 377–381.

[43] K.Y. Lee, D.J. Mooney, Alginate: properties and biomedical applications, Prog. Polym.Sci. 37 (2012) 106–126.

[44] R. Vehring, Pharmaceutical particle engineering via spray drying, Pharm. Res. 25(2008) 999–1022.

[45] R.J. DeVolder, I.W. Kim, E.S. Kim, H. Kong, Modulating the rigidity andmineralizationof collagen gels using poly(lactic-co-glycolic acid) microparticles, Tissue Eng. A 18(2012) 1642–1651.

[46] M.C. Manning, K. Patel, R.T. Borchardt, Stability of protein pharmaceuticals, Pharm.Res. 6 (1989) 903–918.

[47] B. Li, T. Yoshii, A.E. Hafeman, J.S. Nyman, J.C. Wenke, S.A. Guelcher, The effects ofrhBMP-2 released from biodegradable polyurethane/microsphere composite scaf-folds on new bone formation in rat femora, Biomaterials 30 (2009) 6768–6779.

[48] A. Matsiko, C. Murphy, T. Levingstone, F.J. O'Brien, J. Gleeson, Development of a col-lagen–glycosaminoglycan scaffold for cartilage defect repair applications, J. TissueEng. Regen. Med. 6 (2012) 68–69.

[49] N.E. Vrana, N. Builles, H. Kocak, P. Gulay, V. Justin, A. Malbouyres, F. Ruggiero, O.Damour, V. Hasirci, EDC/NHS cross-linked collagen foams as scaffolds for artificialcorneal stroma, J. Biomater. Sci. Polym. Ed. 18 (2007) 1527–1545.

[50] S.W. Kim, H.K. Bae, H.S. Nam, D.J. Chung, P.H. Choung, Peripheral nerve regenerationthrough nerve conduit composed of alginate–collagen–chitosan, Macromol. Res. 14(2006) 94–100.