A dual-application poly (dl-lactic-co-glycolic) acid (PLGA)-chitosan composite scaffold for potential use in bone tissue engineering

Abstract The development of patient-friendly alternatives to bone-graft procedures is the driving force for new frontiers in bone tissue engineering. Poly (dl-lactic-co-glycolic acid) (PLGA) and chitosan are well-studied and easy-to-process polymers from which scaffolds can be fabricated. In this study, a novel dual-application scaffold system was formulated from porous PLGA and protein-loaded PLGA/chitosan microspheres. Physicochemical and in vitro protein release attributes were established. The therapeutic relevance, cytocompatibility with primary human mesenchymal stem cells (hMSCs) and osteogenic properties were tested. There was a significant reduction in burst release from the composite PLGA/chitosan microspheres compared with PLGA alone. Scaffolds sintered from porous microspheres at 37 °C were significantly stronger than the PLGA control, with compressive strengths of 0.846 ± 0.272 MPa and 0.406 ± 0.265 MPa, respectively (p < 0.05). The formulation also sintered at 37 °C following injection through a needle, demonstrating its injectable potential. The scaffolds demonstrated cytocompatibility, with increased cell numbers observed over an 8-day study period. Von Kossa and immunostaining of the hMSC-scaffolds confirmed their osteogenic potential with the ability to sinter at 37 °C in situ.


Introduction
There is an urgent need for alternative approaches for the regeneration of bone following fracture or orthopaedic damage in lieu of traditional methods, and these alternative approaches constitute an important tissue engineering application [1]. The current 'gold standard' therapy is the bone graft procedure, which involves taking autologous bone, usually harvested from the iliac crest of the patient, and implanting it into their defect site [2,3]. Alternatively, allograft bone from donors or cadavers can be extracted from the femoral heads or extremities of other long bones [4]. This implanted tissue acts as a scaffold for the existing bone tissue to infiltrate and deposit extracellular matrix (ECM), leading to the remodelling of the fractured bone [5]. Numerous drawbacks are associated with the above procedures, including the limited supply of autologous bone, complications at the donor site and high surgical costs [2]. Furthermore, in large defects, resorption may occur before osteogenesis has been completed [6]. Allograft bone usage is associated with incompatibility with the host, and the possible transmission of diseases and infections such as hepatitis and HIV [1,5,7,8]. The risk of disease transmission from allograft bone can be minimised by processing or devitalization via freeze-drying or irradiation; however, this may reduce the osteoinductivity and mechanical strength [9,10]. Other options include the usage of bone morphogenetic proteins (BMPs), distraction osteogenesis and bone cement; however, these are also not ideal [3]. The shortcomings in the current clinical options have led to concerted efforts in search of alternative strategies for the repair of bone.
The combination of porous and non-porous microspheres, which are able to sinter at body temperature, enables the introduction of porosity within injected scaffolds, hence, allowing proliferating cells access to nutrients [13,14]. Simultaneously, the delivery of growth factors such as BMPs to the growing cells is also facilitated. BMPs have been studied for their use in non-union bone defects, spinal fusion and open tibial fractures [14]. Furthermore, it has been reported that one such BMP, BMP-2, is present during the initial phase of fracture repair, and during chondrogenesis and osteogenesis [15].
Various strategies have been utilized for the sintering of microspheres into scaffolds. These include the incorporation of plasticizers in order to reduce polymer glass transition temperatures [16], the addition of organic solvents such as dichloromethane [11,17] and the application of heat [4,7,8]. Although the use of high temperatures and organic solvents result in mechanically strong scaffolds, these conditions are not ideal for the body and so are not suitable for sintering in situ. Therefore, a system capable of sintering at 37 °C in situ would be extremely beneficial.
Protein-loaded PLGA microspheres often exhibit an initial burst release [14,18] which is not ideal for an intended controlled release of BMP-2 at a defect site. A number of strategies have been employed to control the release of proteins from PLGA microspheres. These include varying the polymer molecular weight [14], the inclusion of additives such as poloxamer 188 and the use of a PLGA-PEG-PLGA triblock polymer [9,19].
Chitosan is a natural polysaccharide derived from chitin and is popular in tissue engineering applications for a variety of reasons, which include its cytocompatibility and ability to promote cell adhesion [3]. Chitosan microspheres show promise for use in the encapsulation of proteins and have previously been shown to retain the activity of a neural growth factor [20]. Moreover, due to its cationic nature and propensity to slow degradation, chitosan-based materials are able to sustain the release of growth factors [21]. Chitosan has been used in combination with PLGA in various forms, including by embedding PLGA microspheres into chitosan scaffolds [19,20,22]. PLGA/chitosan microspheres can be formulated in a variety of ways. These include the use of supercritical fluid technology, the double emulsion method [23,24], the solvent evaporation technique [25], an electro-dropping layer-by-layer approach [26] and conjugation and adsorption methods [27]. Porous microspheres have also been treated with chitosan [27], whilst others have encapsulated protein-loaded chitosan microspheres into large porous PLGA microspheres [18].
In a previous study, we reported the formulation of a novel PLGA scaffold delivery system based on porous and protein-loaded microspheres that sintered at 37 °C [14]. There have been a number of reports utilising composites of PLGA/chitosan microspheres for use in bone tissue engineering [26][27][28][29][30][31]. In the present work, we report the development of a 'dual-application' PLGA/chitosan composite scaffold formulation which sinters at 37 °C when injected through a hypodermic needle as well as when implanted as a paste. Furthermore, we aimed to control the release kinetics of a model protein for BMP-2 (BMP-2 itself was not used due to the cost implications) from this system, via the inclusion of chitosan, and to investigate its cytocompatibility and osteoinductive capabilities on primary human mesenchymal stem cells (hMSCs).

Formulation of PLGA microspheres
Porous PLGA microspheres were prepared using the double emulsion solvent evaporation method as described in detail elsewhere [13,14]. Briefly, a 250-μl aliquot of PBS was added to a 20% w/v PLGA/DCM solution and homogenised at 9000 rpm using a Silverson L5 M homogeniser (East Longmeadow, MA, U.S.A.). This was added to 200 ml of 0.3% w/v PVA solution and homogenised at 4000 rpm and then stirred at 300 rpm for 4 h. The microspheres were washed with distilled water and then exposed to ethanolic-NaOH in order to enhance the surface porosity. They were then sieved (40 μm) and washed using distilled water. Non-porous microspheres were prepared in a similar way using 100 μl of 100 mg/ ml BSA solution or 100 μl of distilled water, instead of 250 μl of PBS. BSA was chosen as a model protein as it is compatible with chitosan and has previously been used as a substitute for growth factors [32][33][34].
Non-porous PLGA/chitosan composite microspheres were prepared similarly; however, instead of using 200 ml of 0.3% w/v PVA solution, the aqueous phase comprised 150 ml of 0.4% w/v PVA solution containing 0.05 g of TPP. The primary emulsion, in addition to 50 ml of 0.25% w/v chitosan solution in 2% v/v acetic acid, was added to the external aqueous phase simultaneously and homogenised. All microspheres were freeze-dried using a Thermo Fisher Scientific FR-Drying Digital Unit (Waltham, MA, U.S.A.) for 48 h and stored at −20 °C until use.

Scanning electron microscopy and size analysis
The freeze-dried samples were mounted onto aluminium stubs (Agar Scientific, U.K.) and gold-coated using a Balzers SCD030 gold sputter coater (Balzers Union Ltd., Lichtenstein). The morphology and surface topography of the microspheres were observed using a Jeol 6060L scanning electron microscopy (SEM) imaging system (Jeol Ltd., Hertfordshire, U.K.) at 10 kV. The particle size distribution and mean microsphere diameter were determined using a Coulter LS230 particle size analyser (Beckman, U.K.).

Fourier transform infrared spectroscopy
Fourier transform infrared (FTIR) spectra of the microspheres and their constituents were obtained using a Spectrum RX 1 FTIR spectrophotometer (Perkin Elmer, Waltham, MA, U.S.A.). Samples were mixed with potassium bromide (KBr) and compressed using a 5-tonne force into disks; 256 scans were acquired from 400 to 4000 cm −1 .

Preparation of 3D scaffolds
PLGA and PLGA/chitosan composite scaffolds were previously prepared in our laboratories [14]. A 1:1 mass ratio of porous to non-porous microspheres was mixed in a weighing boat followed by mixing with PBS (pH 7.4) at a ratio of 0.25:1 (PBS to microspheres) to form a paste. The paste was packed into a 6-mm diameter and 12-mm height polytetrafluorethylene (PTFE) mould using a spatula, and then stored in a sealed de-humidifying chamber at 37 °C for 17 h.

Time of flight secondary ion mass spectrometry
The presence and distribution of the chitosan coating on the scaffold surfaces was assessed using a time of flight secondary ion mass spectrometer (time of flight secondary ion mass spectrometry [ToF-SIMS] IV, ION-TOF GmbH, Munster, Germany). Scaffolds were placed on the ToF-SIMS stage and secured with metal clips. A 25-keV Bi 3 + primary ion source was used to scan a 256 × 256 pixel raster, while simultaneously not exceeding the limit of static, as described by Rafati et al. [35]. Surface charge due to the primary ion beam on the insulating sample surface was compensated using a flood gun generating low energy electrons (20 eV). Negative and positive polarity data for 500 × 500 μm areas were analysed using the SurfaceLab 6 software (IONTOF, Germany). PLGA was identified by the presence of C 3 H 3 O 2 − (m/z = 71) and C 3 H 5 O 2 − (m/z = 73) [35]. Diagnostic secondary ion peaks for chitosan were identified as CN − (m/z = 26) from the negative polarity data, in addition to CH 4 N + (m/z = 30) and C 4 H 5 N 2 + (m/z = 81) from the positive polarity data. For a semi-quantitative analysis, each area was split into four regions of interest, and the ion intensity data for these peaks of interest were exported and normalised to the total ion intensity.

Encapsulation efficiency (%EE) of BSA within microspheres and scaffolds
The %EE of BSA within the non-porous PLGA and PLGA/chitosan composite microspheres and scaffolds were determined by gently stirring 10 mg of the microspheres or one scaffold in 750 μl or 13 ml of DMSO, respectively, for 1 h. This was followed by the addition of 2.15 ml or 37.27 ml of 0.02% w/v SDS in 0.2 M NaOH to the microspheres or scaffolds, respectively. The solution was left to stand at room temperature for 1 h. Standard concentrations of BSA were calibrated with a BCA reagent so that the sample absorbance could be matched with standard concentrations on an Infinite 200 plate reader (Tecan, Switzerland) at 562 nm. The %EE of BSA within the microspheres and scaffolds was then calculated using Equation (1).

Release of BSA from microspheres and scaffolds
Release studies of BSA from the PLGA and PLGA/chitosan composite microspheres were carried out by submerging 50 mg of microspheres in 1.5 ml of PBS in a micro-centrifuge tube. The tubes were incubated at 37 °C. At predetermined time intervals, the PBS supernatant was removed and replaced with fresh buffer. Aliquots (150 μl) were withdrawn from the supernatant and assayed for the presence of BSA at 562 nm on the microplate reader using the BCA assay kit. BSA release from scaffolds was studied in 4 ml of PBS and assayed as described above.

Preparation of 3D scaffolds post-injection
Microsphere mixtures were prepared at a 1:1 ratio of porous to non-porous microspheres and an approximate BSA loading of 3 mg of BSA/g of mixture, as described in Section 2.5. The mixture was suspended in PBS at a concentration of 50 mg of microspheres/ml, vortex-mixed briefly and then drawn into a 1-ml syringe (BD Fine) fitted with a 19-G needle Actual mass of BSA in 10 mg of microspheres OR 1 scaffold Theoretical mass of BSA used for 10 mg of microspheres OR 1 scaffold × 100 (1.1 × 50 mm, BD fine, Franklin Lakes, NJ, U.S.A.). Finally, the contents of the syringe were injected into the PTFE scaffold mould.

Compressive strength of scaffolds
The compressive strength of the scaffolds was assessed using a TA.HD+ texture analyser (Stable Microsystems, U.K.) equipped with a 50-kg load cell at a speed of 0.04 mm/s over a contact area of approximately 28.75 mm 2 . Dry PLGA and PLGA/chitosan BSA-loaded scaffolds prepared as described in Sections 2.5 and 2.9 were tested, and the compressive strength was determined as the stress at the maximum strain.

Cell culture and seeding onto scaffolds
Primary hMSCs were cultured in hMSC basal media supplemented with the contents of an MSCGM hMSC SingleQuot kit. The cells were maintained in a humidified tissue-culture incubator at 37 °C in 5% CO 2 . The cytocompatibility test was carried out on BSA-free scaffolds. Scaffolds were prepared directly into a 24-well plate in a manner similar to that described in Section 2.5. A 1:1 porous to non-porous microsphere mixture was UV sterilised for 80 min [36] and then transferred to the well. Basal growth medium was then added at a ratio of 0.25:1 (medium to microspheres). After 17 h of sintering, each scaffold was seeded with 1 × 10 5 hMSCs and incubated for 2 h, followed by the addition of 1 ml of media to each scaffold/well. The cell-seeded scaffolds were maintained at 37 °C with 5% CO 2 . For all cell experiments, either 5 replicates or 2 independent repeats each comprising at least 3 replicates was carried out.

Cell viability assay
Each scaffold was submerged in 1 ml of media and 111 μl of Presto Blue reagent and the cell viability was determined at day 1, 3, 6 and 8 post-seeding using the Presto Blue cell viability reagent. The well plate was protected from light and incubated at 37 °C for 25 min. Aliquots of 100 μl were withdrawn from each well in triplicate and the absorbance was read on an infinite 200 plate reader (Tecan, Switzerland) at excitation and emission wavelengths of 560 and 590 nm, respectively. The Presto Blue reagent was replaced with fresh media and the scaffolds were placed back in the incubator. On day 8, after measuring the cell viability, the scaffolds were washed with PBS and the cells were fixed with 10% v/v buffered formalin solution for 20 min. Fixed hMSC-scaffold constructs were viewed under the SEM.

Assessment of mineralisation
In order to determine the degree of mineralisation on the scaffolds, the von Kossa assay was utilized. Cells were seeded onto scaffolds as described in Section 2.11 and incubated in basal growth media for 21 days. On day 21, cells were fixed with 10% v/v buffered formalin for 20 min and thoroughly washed with PBS. A 450-μl aliquot of 1% w/v silver nitrate solution was added to each scaffold and incubated under a UV light source for 1 h. The solution was then removed and the scaffolds were washed three times with deionized water. This was followed by treatment with sodium thiosulphate solution for 5 min in order to remove any excess silver nitrate solution. The scaffolds were then washed with PBS prior to imaging under a dissection microscope (Leica, Germany).

Osteocalcin immunostaining
Cells were seeded onto scaffolds as described in Section 2.11. The scaffolds were incubated in basal growth media for 21 days after which they were fixed using 10% . Four images were taken of four different areas on each scaffold and then converted into binary formats so that the stained areas could be calculated.

Statistical analyses
A statistical analysis of the data was carried out using Microsoft Excel. An unpaired t test and the ANOVA procedure were used and the results were deemed significant when p < 0.05.

Physical characterisation of PLGA/chitosan composite microspheres and scaffolds
BSA-encapsulated PLGA/chitosan composite microspheres were formulated using TPP as a cross-linker as detailed in Section 2.2. Both the PLGA and composite microspheres appeared smooth, as shown in the SEM images in Figure 1(A) and (B), respectively. Thus, the addition of chitosan cross-linked with TPP did not alter the superficial appearance of the microspheres and no unprocessed, free chitosan is visible from the SEM images. Size analysis revealed the average diameters of the PLGA and composite microspheres to be 69.75 ± 21.47 μm and 66.85 ± 22.68 μm, respectively (Figure 1(C)). The FTIR spectra of the raw materials and microspheres are presented in Figure 2. The chitosan spectrum shows a high-intensity peak at 3400 cm −1 , which corresponds to stretching vibrations of the O-H and N-H bonds, in addition to hydrogen bonding in the  backbone [37]. The characteristic peak at 1647 cm −1 is a result of the amide functionality and may be present as a consequence of the axial deformation of the C=O bond [37] and strong N-H bending [38]. Peaks present at 1019 and 1086 cm −1 (corresponding to C-O stretch vibrations), and 1152 cm −1 (asymmetric stretch of the C-O-C bond) are also indicative of chitosan [37].
The TPP spectrum, similarly, shows a peak of significant intensity at 3390 cm −1 , corresponding to the stretching vibrations of the O-H bond. Peaks around the 1095 cm −1 region are an indication of the P=O phosphate group. The PLGA spectrum presents a peak at 3473 cm −1 , which is indicative of vibration of the terminal O-H groups. Other peaks that indicate PLGA are present at 743 cm −1 (C-H bend), 1086 and 1180 cm −1 (C-O stretch), 1381 cm −1 (C-H bend), 1771 cm −1 (the carbonyl C=O) and 2876 cm −1 (CH 2 bend) [39]. Both PLGA and PLGA/chitosan composite BSA-loaded microspheres show peaks at identical wavelengths, which suggests that the microspheres are predominantly PLGA. Moreover, the spectra of PLGA and PLGA/chitosan composite microspheres show peaks at 1621 and 1639 cm −1 , respectively, which are attributed to the C=O bond of the amide groups that are found both in BSA and chitosan. However, there does appear to be a slightly more pronounced peak at 1639 cm −1 on the spectrum of the PLGA/chitosan composite microspheres, which corresponds to the amide C=O bond suggesting the presence of chitosan in the formulation.
The ToF-SIMS analysis was carried out in order to ascertain the presence of chitosan on the scaffold surfaces. BSA-free scaffolds were analysed based on the overlap of chitosan and BSA secondary ion peaks (discussed in Section 2.6). Intensities of nitrogen-containing positive secondary ion peaks CH 4 N + (m/z = 30) and C 4 H 5 N 2 + (m/z = 81), as well as the negative ion peak CN − (m/z = 26) were all significantly higher in the composite PLGA/ chitosan scaffolds when compared to the chitosan-free scaffolds, as shown in Figure 3(A). However, there was no significant difference between the profiles of diagnostic PLGA ion peaks for the PLGA and composite scaffolds (Figure 3(B)).
The incorporation of chitosan did not elicit a significant change in the encapsulation efficiency of BSA in the microspheres, with 80.58 ± 17.06% and 81.57 ± 3.06% of the protein being encapsulated into the PLGA and PLGA/chitosan composite microspheres, respectively. Moreover, there was no statistical difference in the encapsulation efficiencies of the PLGA and composite scaffolds (2.81 mg/g [93.68% ± 3.50%] and 2.52 mg/g [84.02% ± 12.08%] for the PLGA and PLGA/chitosan composite scaffolds, respectively).

Release of BSA from microspheres and scaffolds
The release profile of BSA was mapped over 28 days from both microspheres and scaffolds sintered at 37 °C (Figure 4). The initial burst release after 24 h from the PLGA microspheres was significantly higher than from the PLGA/chitosan composite microspheres, 0.93 ± 0.06 μg/mg and 0.57 ± 0.03 μg/mg, respectively (p < 0.05). After 28 days, 1.72 ± 0.23 μg/ mg of BSA was released from the PLGA microspheres, which was higher in comparison to 1.20 ± 0.05 μg/mg from the PLGA/chitosan composite microspheres (p = 0.05) Similarly, there was a significant retardation of the initial burst release from the scaffolds containing PLGA/chitosan composite microspheres, 0.10 ± 0.02 μg/mg, in comparison to the PLGA scaffolds, 0.16 ± 0.01 μg/mg (p < 0.05, Figure 4(B)).

Sintering of microspheres into scaffolds
In order to study the effect of the scaffold preparation method on their subsequent morphology and mechanical strength, the PLGA and PLGA/chitosan composite scaffolds were prepared using two different methods. Firstly, a paste was formed from the microspheres as previously reported [14]. In the second method, we aimed to study the ability of the microspheres to sinter post-injection through a 19-G needle into a scaffold mould. This was then followed by a 17-h incubation period at 37 °C. Photographs of the resulting scaffolds and their compressive strengths are presented in Figure 5(A) and (B), respectively. The sintering process results in the expulsion of water so that the components within close proximity. We believe that this favours 'fusion' and bond formation within the scaffolds. This approach to scaffold sintering at 37 °C is superior to the more harsh methods employing elevated temperatures and reagents. The overall appearances of PLGA and composite scaffolds were very similar ( Figure 5(A)). However, when comparing scaffolds prepared using the paste method, the compressive strength of PLGA/chitosan composite scaffolds was significantly higher (0.846 ± 0.272 MPa) than the PLGA scaffolds (0.406 ± 0.265 MPa, p < 0.05). Figure 5(A) shows that it was possible to successfully sinter a microsphere suspension post-injection, thus, forming intact scaffolds that retained their shape when removed from the mould. This confirms the injectable potential of the microspheres. When scaffolds were sintered as a suspension post-injection, there was no significant difference between the compressive strengths of the PLGA and PLGA/chitosan composite scaffolds, 0.086 ± 0.068 MPa and 0.048 ± 0.00096 MPa, respectively (p > 0.05, Figure 5(B)); however, it is likely that the compressive values may be below the lower limit of threshold of the machine.

Cell proliferation on scaffolds
The culturing of primary hMSCs on the scaffolds was used as a means to test their cytocompatibility. Cell proliferation was assessed using the Presto Blue viability reagent on day 1, 3, 6 and 8 ( Figure 6(A)).
Cell proliferation increased over time on both scaffold types. On day 1, the cell numbers on PLGA and PLGA/chitosan composite scaffolds were 1.06 × 10 4 and 1.03 × 10 4 , respectively. Both types of scaffolds exhibited a very similar cell growth profile with no statistically significant difference found between them (p > 0.05) on day 1, 3 and 6. However, the cell number on day 8 was significantly higher on the PLGA scaffolds (p < 0.05) at 6.25 × 10 4 and 4.45 × 10 4 for PLGA and PLGA/chitosan composite scaffolds, respectively. SEM images of the cell-scaffold constructs on day 8 are shown in Figure 6(B) and (C), with cells visibly distributed between microspheres in both scaffold types.

Assessment of mineralisation
The extent of mineralisation on the scaffolds after 21 days in culture media was assessed using the von Kossa assay as described in Section 2.14. Dark brown/black nodules (indicated by the white arrow in Figure 7(B)) are visible on the scaffolds and represent positive staining. A qualitative analysis shows that there are more nodules on the PLGA/chitosan composite scaffolds (Figure 7(B)), which appear darker in the figure, in comparison to the PLGA scaffolds (Figure 7(A)).

Osteocalcin immunostaining
The presence of the bone marker protein, osteocalcin, was detected using the immunostaining technique described in Section 2.14. The data obtained was processed using ImageJ, which allowed us to quantify the amount of stain present on each scaffold. The results of this analysis show that there was an increase in osteocalcin staining on the composite scaffolds when compared to the PLGA scaffolds (p < 0.05, Figure 8(A)). When osteogenic media was used (data not shown), the osteocalcin staining on the PLGA/chitosan and PLGA scaffolds was not significantly different (p > 0.05). Processed, merged images are shown in Figure  8

Discussion
Scaffolds made from biodegradable microspheres are a promising approach for bone regeneration. However, there are several features to consider when developing such systems. These include the incorporation of porosity and growth factors into the scaffolds, whilst at the same time providing mechanical strength to enable the microspheres to be injectable and sinter in situ. Some research groups have developed scaffolds with some of these properties; however, most groups do not take into account all desirable features in one system. In the present study, we propose a novel dual-application PLGA/chitosan composite scaffold system with the potential to meet all of the above desirable criteria. The system comprises porous and non-porous protein-loaded microspheres with the ability to sinter at 37 °C and release protein. The mechanical strength of the system is dependent upon its mode of application, with a higher compressive strength achievable when it is applied as a paste, and sufficient strength to maintain the shape (as evident from the fact that the microspheres sintered at 37 °C and were subsequently removed from the mould intact) when injected as a suspension. The cytocompatibility and osteogenic potential of the formulation were evaluated and compared with our previously reported system [14]. Protein-loaded microspheres were formulated using PLGA and chitosan, where the chitosan was cross-linked using TPP. There were no observable differences in the morphology and size of the composite microspheres when compared with PLGA microspheres. The presence of chitosan within the composite scaffolds formed via the paste method was confirmed by ToF-SIMS, suggests that chitosan is formed as part of the microstructure of the particles. Furthermore, the composite scaffolds demonstrated higher compressive strength than the PLGA scaffolds. In this regard, chitosan contributes to the mechanical strength of the scaffolds, due to interactions between the negatively charged PLGA [40] and the protonated amine groups in the chitosan structure. Moreover, the compressive strength demonstrated by the composite scaffolds fell within an acceptable range as reported by Misch et al. [38].
The chitosan coating attenuated the initial burst release from the microspheres and scaffolds, and this reduction may partly be attributed to chitosan complexing with BSA (isoelectric point, approximately 5), thereby, impeding its release. The ability of chitosan, a natural polyelectrolyte, to non-covalently bind to negatively charged proteins has been reported [41]. A similar observation of a reduced burst release was made for the same system when encapsulated with lysozyme, which is positively charged at a neutral pH (data not shown). This suggests that other factors contribute to the reduction in burst release. It has been reported that the burst release of proteins from PLGA microspheres is usually due to protein residing near, or on the surface of, the delivery system. We believe that the formation of a chitosan -TPP matrix layer slows the release of the protein and significantly contributes to the attenuation in the initial burst release. This effect has been demonstrated in PLGA/chitosan microspheres encapsulated with a non-protein drug, rifampicin, in which the addition of chitosan caused a reduction in the burst release [42]. The slower, steadier release of BSA from the microspheres and scaffolds containing PLGA/ chitosan is desirable in BMP-2 applications. The controlled release reduces the need for supra-physiological loadings, which are necessary when there is a huge initial loss via a burst release [19].
The system described herein possesses dual-applicability arising from the formulation's potential of having two application modes (i) a paste that is implanted within a degenerated bone tissue, takes the shape of the defect area and then sinters at 37 °C, and (ii) the injection of the microsphere suspension directly into the defect area. The former would be useful in applications requiring a relatively stronger scaffold, such as the regeneration of cancellous bone for which the ultimate compressive strength has been reported to range from 0.22 to 10.44 MPa [38]. However, the latter is more suited to applications in which the delivery system may be injected and remain in one location, hence, allowing the controlled delivery of a specific, known dose of protein to the site. To our knowledge, this is the first time that the ability of microspheres to sinter at 37 °C, post-injection, has been demonstrated.
The ability of cells to attach and grow on the scaffolds is paramount in the development of protein delivery systems in regenerative medicine. For this reason, the cytocompatibility of the scaffolds with hMSCs was investigated. The cell number increased on the composite scaffolds over the 8-day period from 1.03 × 10 4 on day 1, to 4.45 × 10 4 on day 8. There was no significant difference between cell numbers on the composite and PLGA scaffolds, except on day 8, by which time the cell numbers were higher on PLGA scaffolds (p < 0.05). Although previous studies have investigated the cytocompatibility of sintered composite PLGA/chitosan microspheres scaffolds with other cell types, these formulations were not capable of sintering in situ [18].
The potential of the scaffold material to promote the differentiation of hMSCs is another key factor that is crucial for the production of a successful biomaterial. Although the presence of BMP-2 has been shown to promote osteogenic differentiation, the intrinsic ability of the material itself to promote the process is also of interest. Chitosan has been reported to have numerous biomedical properties, including its ability to improve osteogenesis in animal bone defect models [43]. In this study, we investigated the cell response to protein-free scaffolds in basal media in order to study the effect of the scaffold material on osteogenesis. The presence of a calcified ECM is a reliable way of confirming osteogenesis [44]. Nodules were observed on both composite and PLGA scaffolds based on von Kossa staining, which indicates the presence of calcium. To provide further confirmation of the deposition of a calcified matrix, the presence of osteocalcin, a late protein marker of osteogenic differentiation, was determined. Its expression is known to rise with an increase in mineralisation [45]. The composite scaffolds showed a significantly higher degree of osteocalcin staining when compared to the PLGA scaffolds. Previous studies have demonstrated the ability of chitosan-containing scaffolds to induce differentiation in the presence of osteogenic media [46], which we also confirmed (data not shown). However, relatively little evidence has demonstrated this in basal growth media. Therefore, these results suggest that the inclusion of chitosan in PLGA microspheres enhanced the osteogenic capacity of the resultant scaffolds.

Conclusion
In this study, a novel, dual-application composite microsphere system was developed with the ability to fuse together as a paste, thereby forming an intact scaffold in the body at 37 °C. Furthermore, the ability of a suspension of the microspheres to sinter post-injection was also demonstrated. Composite PLGA/chitosan microspheres were shown to attenuate the initial burst release and elicited a steady, slow release of protein over 28 days. The scaffold's cytocompatibility and ability to promote osteogenesis were also demonstrated. This technology, therefore, exhibits potential as a scaffold for bone regeneration and is an excellent candidate for further in vitro and in vivo testing.