Transient serum exposure regimes to support dual differentiation of human mesenchymal stem cells

Human mesenchymal stem cells (MSCs), which can generate both osteoblasts and chondrocytes, represent an ideal resource for orthopaedic repair using tissue‐engineering approaches. One major difficulty for the development of osteochondral constructs using undifferentiated MSCs is that serum is typically used in culture protocols to promote differentiation of the osteogenic component, whereas existing chondrogenic differentiation protocols rely on the use of serum‐free conditions. In order to define conditions which could be compatible with both chondrogenic and osteogenic differentiation in a single bioreactor, we have analysed the efficiency of new biphasic differentiation regimes based on transient serum exposure followed by serum‐free treatment. MSC differentiation was assessed either in serum‐free medium or with a range of transient exposure to serum, and compared to continuous serum‐containing treatment. Although osteogenic differentation was not supported in the complete absence of serum, marker expression and extensive mineralization analyses established that 5 days of transient exposure triggered a level of differentiation comparable to that observed when serum was present throughout. This initial phase of serum exposure was further shown to support the successful chondrogenic differentiation of MSCs, comparable to controls maintained in serum‐free conditions throughout. This study indicates that a culture based on temporal serum exposure followed by serum‐free treatment is compatible with both osteogenic and chondrogenic differentiation of MSCs. These results will allow the development of novel strategies for osteochondral tissue engineering approaches using MSCs for regenerative medicine. Copyright © 2012 John Wiley & Sons, Ltd.


Introduction
In the natural joint, articular cartilage plays an important role, protecting the subchondral bone from high stresses as well as reducing nominal contact pressure and allowing low-friction movements (Mow et al., 1993;Little, 1969). Both the articular cartilage and the underlying subchondral bone are often subjected to structural damage as a result of trauma and disease, leading to severe pain, joint deformity, loss of motion and the need for surgical procedures (Tampieri et al., 2008;Swieszkowski et al., 2007). Due to the avascular, non-innervated nature of cartilage, and its limited capacity to repair itself, there is an increased demand to engineer osteochondral grafts that overcome the problems of standard chondral grafts, such as donor site morbidity and poor anchorage between the native tissue and the chondral construct (Cancedda et al., 2003). Typical osteochondral grafts consist of a superficial cartilaginous layer and an underlying calcified tissue, representing cartilage and bone, respectively, which often delaminate when subjected to mechanical loads and stresses of the body (Martin et al., 2007). Approaches considered for joint tissue repair include cell-free composites, e.g. ceramic-based devices, and scaffolds loaded with a single cell type or with more complex multilineage cell populations (Tampieri et al., 2008;Vinatier et al., 2009). Although more physiologically relevant, scaffolds engineered to host multiple cell types are technically challenging to seed and populate homogeneously. An optimal solution would be to engineer a homogeneous scaffold using a single cell source that has both osteogenic and chondrogenic differentiation potential, thus overcoming complications of multicellular and multimedia approaches that can lead to delamination of the bone and cartilage regions, and ultimately contribute largely to the failure of the construct.
Mesenchymal stem cells (MSCs) have been utilized in many tissue-engineering applications for several years (Caplan and Bruder, 2001), due to their ability to differentiate towards multiple lineages, including osteoblasts, chondrocytes and adipocytes (Pittenger et al., 1999;Mackay et al., 1998;Young et al., 1998). MSCs are the primary cell source for endochondral and intramembranous bone formation in vivo (Caplan and Pechak, 1987), making them an ideal resource for the production of cell-seeded osteochondral constructs that can be utilized for bone repair. In addition, MSCs represent a potential autologous cell source, which would overcome graft rejection issues previously seen in the repair of chondral defects (Csaki et al., 2008;Bruder et al., 1994;Deans and Moseley, 2000), and autologous MSCs have already been successfully used in a clinical setting for bone repair (Quarto et al., 2001). Conditions for the osteogenic and chondrogenic differentiation of MSCs have been described for many years in the literature (Bruder et al., 1994(Bruder et al., , 1997Pittenger et al., 1999;Sottile et al., 2002).
One key difference in the culture conditions used to stimulate osteogenic and chondrogenic differentiation is the use of serum: osteogenic treatments typically include 10% serum, whereas chondrogenic differentiation protocols are carried out in a serum-free medium. The presence of serum, an ill-defined cocktail of growth factors, proteins, toxin scavengers and other source-dependent substances (Liu et al., 2007), is thus considered an unsuitable component to support the simultaneous differentiation into these two lineages.
In order to design treatment regimes which could resolve this incompatibility and provide a basis for the differentiation of both osteogenic and chondrogenic lineages in MSC-seeded constructs cultured in a single chamber, we investigated the use of new biphasic treatments. The research presented here examines a new approach to MSC differentiation, using culture conditions involving transient FCS exposure, and reports on the efficacy of these new treatments for the differentiation of MSCs towards osteogenic and chondrogenic lineages.

Materials and methods
All reagents were purchased from Invitrogen (Paisley, UK) unless otherwise stated.

Cell culture
A culture of a human bone marrow-derived mesenchymal stem cell (hMSC) line (Okamoto et al., 2002;Sottile et al., 2003) was maintained in a monolayer culture in complete Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% v/v fetal calf serum (FCS), 1% v/v L-glutamine, 1% v/v non-essential amino acids (NEAA) and 1% v/v penicillin/streptomycin, in a humidified atmosphere at 37 C and 5% CO 2 . For the purposes of the work presented here, this medium will be referred to as standard culture (SC) medium. Monolayer cultures were passaged by enzymatic digestion, using 0.02% trypsin-EDTA, producing a single cell suspension, and cells were passaged using 1 in 4 dilutions for culture maintenance. Live cell counts carried out using trypan blue exclusion were used to determine the final cell-seeding density (2 Â 10 4 cells/ml). For mineral deposition analysis, Thermanox TM tissue culture plastic discs (Nunc TM , Fisher Scientific, UK; 13 mm diameter) were placed into 24-well plates and sterilized under UV for 1 h. The wells were washed with phosphate-buffered saline (PBS; (Oxoid, UK) and 1 ml standard MSC culture medium before 1 ml MSC cell suspension was added (2 Â 10 4 cells/ml). The cells were allowed to attach to the Thermanox discs in SC for 24 h before starting the differentiation experiment.

Osteogenic differentiation
To evaluate the effects of different culture conditions on osteogenic differentiation, two medium preparations were tested: (a) OS + FCS [serum-containing osteogenic medium: DMEM supplemented with 10% v/v fetal calf serum (FCS), 1% v/v L-glutamine, 1% v/v NEAA, 1% v/v penicillin/streptomycin, dexamethasone (0.1 mM), ascorbic acid phosphate (50 mM) and b-glycerophosphate (10 mM; Sigma-Aldrich)] (Sottile et al., 2002); (b) OS -FCS (osteogenic supplement only: recipe as stated in (a) but with no FCS). Cells (passage 37) were treated with OS + FCS for 1, 2, 3, 4 and 5 days before switching to OS -FCS for the remainder of the 21 day culture period. Control wells were treated in parallel with OS + FCS or OS -FCS for the 21 days without switching. An additional control culture was also treated with SC medium in the absence of osteogenic supplements for the duration of the assay. A summary of the different culture conditions used is presented in Figure 1a. The medium was changed every day for the first 5 days of culture and every 3 days thereafter.

Chondrogenic differentiation
To determine the influence of serum exposure on the chondrogenic differentiation of hMSCs, two media were tested: (a) CS -FCS [serum-free chondrogenic medium: DMEM supplemented with 1% v/v L-glutamine, 1% v/v NEAA, 1% penicillin/streptomycin, dexamethasone (0.1 mM), ascorbic acid phosphate (50 mM), sodium-pyruvate (1 mM), L-proline (40 mg/ml), TGF-b (10 ng/ml) and ITS + (Sigma-Aldrich)]; (b) CS + FCS (chondrogenic medium containing FCS: recipe as stated in (a) with the inclusion of 10% v/v FCS). Cell pellets were formed with 2.5 Â 10 5 cells (passage 39) and cultured in a polystyrene Falcon tube as previously described (Sottile et al., 2002). One set of pellets was treated with CS + FCS for 5 days and then switched to CS for the remainder of the 21 day culture period ('CS 5 '), and another set was treated for the 21 days with CS + FCS medium ('CS + FCS'). Pellets maintained for 21 days in CS -FCS medium ('CS -FCS') or SC medium in the absence of all chondrogenic supplements ('SC') were used to provide positive and negative controls, respectively. A summary of the different culture conditions used is presented in Figure 1b. The medium was changed every day for the first 5 days of culture and every 3 days thereafter.

Alamar blue assay
Cell metabolic activity was evaluated using an Alamar blue assay after 21 days culture. This assay is a nonendpoint, non-toxic assay that measures mitochondrial metabolic activity. After 21 days of culture, the medium was removed and the samples washed three times with PBS at ca. 37 C; 500 ml Alamar blue solution [AbD Serotec, UK; 1:10 Alamar blue:warm Hanks' balanced salt solution (HBSS)] were added to each well. Three empty wells were filled as blanks, and the plate incubated for 80 min at 37 C and 5% CO 2 , then wrapped in aluminium foil and put on a plate shaker for 10 min. A 100 ml extract from each sample well was transferred to a 96-well plate in triplicate. Fluorescence was measured at 530/25 nm excitation and 590/35 nm emission, using a Bio-TEK FLx800 microplate fluorescence reader.

Hoechst 33258 DNA assay
Cell proliferation was determined by assaying for total DNA content after the 21 day culture period. The fluorochrome Hoechst 33258 (Sigma-Aldrich) binds cellular DNA, resulting in enhanced fluorescence, which is directly proportional to the DNA content (Rago et al., 1990). To lyse the cells, the culture medium was replaced with 1 ml sterile distilled water and the well plate stored at À20 C. Once frozen, the well plate was placed into the incubator at 37 C to thaw, and the cycle was repeated 3 times. Aliquots of 100 ml from each well were transferred to a 96-well plate along with 100 ml Hoechst stain, used at a working dilution of 1:50 in TNE buffer [10 mM Tris (hydroxymethyl)methylamine, 1 mM EDTA, 2 mM NaCl in distilled water, pH 7.4]. The plate was gently agitated and fluorescence measured at 360 nm excitation and 460 nm emission, using a Bio-TEK FLx800 microplate fluorescence reader. A standard curve of DNA was produced using known concentrations of DNA from calf thymus (Sigma-Aldrich), reconstituted in 0.01 M NaCl at a concentration of 20 mg/ml.

DMMB assay
Binding of 1,9-dimethylmethylene blue (DMMB) to sulphate and carboxylate groups was used to quantify glycosaminoglycans (GAG) (Farndale et al., 1982). Briefly, samples were incubated with 1 ml papain solution overnight in a waterbath at 60 C [0.1% papain (Sigma-Aldrich): papain buffer (sodium phosphate 0.1 M, cysteine hydrochloride 0.005 M, EDTA 0.005 M and distilled water)]. Papain-digested pellet samples were diluted 1:100 with heat-treated papain solution and aliquots of 20 ml were taken in triplicate from each well and transferred to a 96-well plate. A GAG reference calibration curve was generated using chondroitin 4-sulphate (Sigma-Aldrich) and 20 ml DMMB solution was added to each well before optical density measurement at 540 nm (Bio-TEK ELx800 microplate reader).

Safranin-O staining
After 21 days of culture, the pellets were washed in PBS and fixed in 4% PFA. After washing again in PBS, they were submerged in 15% sucrose in PBS for several hours and then transferred to a 30% sucrose solution and left over night. Pellets were then embedded in an OCT compound (Tissue-Tek, USA) and frozen. Samples were loaded into the cryostat, cut into sections 5 mm thick and collected on SuperFrost W glass slides (VWR). Pellet sections were rinsed in PBS to remove OCT prior to being washed with distilled water and rehydrated through a series of methanols (100%, 90% and 70% v/v) and rinsed in tap water. The sections were then incubated in Mayer's haematoxylin for 5 min, rinsed under running water until the nuclei turned blue and rinsed in distilled water. The slides were immersed in 0.1% safranin-O for 10 min and then dehydrated through an increasing series of methanols (70%, 90% and 100% v/v). The sections were finally cleared in xylene and mounted using DPX prior to viewing under a phase-contrast microscope.

ELISA for osteocalcin
An osteocalcin ELISA kit (Invitrogen, UK) was used to measure intact osteocalcin concentrations, following the manufacturer's instructions. The absorbances of the samples (proportional to the human osteocalcin concentration) were read at 450 nm, using a colorimetric plate reader (Bio-TEK Elx800 microplate reader).

Immunohistochemistry
Cells were fixed in ice-cold 4% PFA for 15 min and washed in PBS + 0.1% Tween 20 (PBT) for 5 min. Nonspecific binding was blocked by incubation with 0.75% FCS in PBT for 1 h. The cells were then exposed to the primary antibody (bone alkaline phosphatase and osteonectin antibodies from DSHB; osteocalcin antibody from Abcam) at a dilution of 1:200 in PBT and stored at 4 C overnight. After washing three times in PBT at 20 min intervals, the cells were incubated with a 1:200 dilution of Texas red-conjugated secondary antibody (Vector Laboratories) for 1 h in the dark, before washing four times in PBS for 15 min each. The samples were finally mounted onto microscope slides, using Vectashield mounting medium containing DAPI (Vector Laboratories) and analysed using fluorescent light microscopy.

Alizarin red staining and quantification
After 21 days in culture, cells were fixed with 4% paraformaldehyde (PFA) for 15 min. After washing three times with PBS, 1 ml 1% aqueous alizarin red (Sigma-Aldrich) was added to each well and left for 10 min, prior to multiple washes with distilled water (Tataria et al., 2006). Once the excess stain had been removed, the bone nodules were viewed using a phasecontrast microscope. Quantitative analysis of alizarin red staining was achieved by the extraction of the stain using a destaining solution (Tataria et al., 2006; 20% methanol, 10% acetic acid in water) for 15 min, whilst the well plates were gently agitated using a plate shaker. Alizarin red concentrations were determined using a colorimetric plate reader at a wavelength of 405 nm (Bio-TEK Elx800 microplate reader). The values obtained were normalized to DNA content/well.

ESEM and EDX; mineralization, identification and quantification
The cell cultures were washed three times with warm PBS and fixed in 3% glutaraldehyde in 0.1 M cacodylate buffer for 30 min, then washed in 0.1 M cacodylate buffer containing 7% sucrose. The cells were then viewed and analysed in a Philips FEG environmental scanning electron microscope (ESEM) operated at 10 kV with a field emission gun in wet mode, allowing analysis to be carried out on unprocessed samples. The resulting energy-dispersive X-ray (EDX) spectra were used to identify elements present as a result of matrix mineralization.

FIB-SEM and FIB-TEM
Focused Ion Beam-scanning electron microscopy (FIB-SEM; for lift-out) and Focused Ion Beam-transmission electron microscopy (FIB-TEM; for evaluation) were carried out using protocols detailed elsewhere (Edwards et al., 2009). In brief, prior to sectioning using an FEI Quanta 200 3D FIB-SEM, samples were fixed in ice-cold 4% PFA and dehydrated. A small strip of copper conducting tape was put on the surface of the sample to provide a conductive path from a region adjacent to the region to be milled and the sample support. Ion beam-induced deposition of tungsten (W) on the surface connected the copper tape to the region to be milled, whilst a further W deposition using a gas-injection system provided a~2 mm thick protective layer over the region to be sectioned. A section was cut from the sample and transferred using an Omniprobe manipulator needle to a copper TEM support grid for final thinning, using lowering current steps and finishing at 30 kV and 30 pA. Electron-transparent specimens were then examined using scanning transmission electron microscopy (STEM) imaging and EDX analysis, both carried out on a Jeol 2100F TEM.

Statistics
All experimental data represent mean AE SEM of two repeats of experiments, with four wells/pellets seeded for each assay. Triplicates from each well were taken for quantitative assays. Statistical comparisons for all experimental sets were based on one-way ANOVA, using Tukey's post-test for pairwise comparisons, with p < 0.05 considered as significant unless otherwise stated.

Transient serum exposure supports MSC metabolic activity
To evaluate the effects of varying FCS exposure regimes on OS-treated MSC cultures (Figure 1a), we performed a proliferation assay comparing the DNA content present in cultures (Figure 2a). Proliferation was increased with prolonged exposure to serum in the culture medium. There was no significant difference between cells maintained in SC medium and OS 0 cells that were not exposed to FCS. There was also no significant difference in the DNA content from cultures OS 1 , OS 2 , OS 3 and OS 4 . However, a significant increase was observed after 5 days of exposure to FCS in the OS 5 culture (p < 0.05), which in turn showed no significant difference to OS 21 (p > 0.05). In addition, the metabolism in MSCs cultured under OS conditions with varying FCS exposure periods was measured using the alamar blue assay (Figure 2b). The development of fluorescence represented increased metabolic activity, indicating that the cells were healthy and viable for proliferation. Alamar blue reduction increased with exposure to FCS and was significantly higher in all conditions compared to SC conditions (p < 0.001). No significant difference was detected between OS 0 and the cultures exposed to FCS for 1, 2 and 3 days. There was a significant increase in metabolic rate after 4 days of exposure, but no significant difference was found for exposures longer than 4 days (p > 0.05). Analysis of cell viability by propidium iodide staining confirmed that MSCs were not adversely affected when exposed to FCS for 5 days compared to standard FCS-containing OS treatment (see Supporting information, Figure S1).

Transient exposure to FCS supports expression of osteogenic markers
Expression of the osteogenic markers bone-specific alkaline phosphatase (ALP) and osteocalcin (OCN) was analysed Metabolic activity in cultures after variable exposure to OS + FCS over a 21 day period. Alamar blue reduction increased with longer exposure to OS + FCS. There was a significant increase after 4 days of exposure but no significant difference thereafter. Bars = mean AE SEM (n = 4); *p < 0.05, **p < 0.001. Culture conditions were tested as defined in Figure 1 using RT-PCR (Figure 3a). The expression of ALP was higher in all conditions containing OS supplements compared to the SC control. Although there was no difference between OS 0 , OS 1 and OS 2 , exposure to FCS for 4 days or more saw a significant increase of the ALP signal, with an expression level observed for 5 day exposure to FCS comparable to that observed for OS 21 . OCN upregulation was also seen with increased exposure to FCS. While no clear difference was seen between OS 0 and 1 day of exposure to FCS, both showing low levels of OCN, a significant increase was seen after 4 days of exposure to FCS, with no substantial difference between OS 4 , OS 5 and OS 21 conditions. To confirm this result, the concentration of OCN protein present in selected cultures was quantified using ELISA ( Figure 3b). A significant increase in OCN concentrations was seen with increased exposure to FCS. There was no significant difference detected between 5 days of exposure to FCS and OS 21 (p > 0.05). Comparable to previous data, there was a significant difference between this and OS 0 , which was not exposed to FCS (p < 0.001), which in turn was expressing higher levels of OCN than the SC control. Trace amounts of OCN protein were detected in the SC control.
Immunohistochemical staining was used to further assess the presence of alkaline phosphatase (ALP), osteocalcin (OCN) and osteonectin (OSN) in hMSC cultures (Figure 3c). Staining was observed for all three markers in conditions exposed to OS supplement, with a qualitative increase in  expression with increased exposure to FCS. A strong ALP, OCN and OSN signal was detected in OS 5 and OS 21 . In comparison, expression was much lower is conditions treated without FCS, but still higher than the SC control. Minimal ALP and OCN expression was detected in the SC control.

Transient serum exposure supports mineral deposition
To further evaluate the effect of transient serum exposure on MSC differentiation, cultures were stained with alizarin red to analyse mineral deposition. The SC control showed only minimal background staining. In the presence of OS supplements, nodular aggregates were visible under all conditions, indicating the presence of calcium phosphate deposits resulting from a calcifying extracellular matrix (ECM) (Figure 4). Quantification of the alizarin red staining in parallel cultures showed that there was no significant difference between OS 0 and OS 1 , OS 2 and OS 3 (p > 0.05). However, a significant increase was seen after 4 days of exposure to FCS (p < 0.001) and after 5 days of FCS exposure there was no significant difference compared to the OS 21 positive control. In order to better characterize the mineral deposited in the cultures, a more detailed microscopy analysis was performed using ESEM ( Figure 5). All cells treated with OS, with and without serum, displayed an enlarged  Alizarin red staining was stronger with increased exposure to OS + FCS. There was a significant increase after 4 days of exposure (**p < 0.001) but no significant difference thereafter (p > 0.05). Bars = mean AE SEM (n = 4). Culture conditions were tested as defined in Figure 1 stellate structure, characteristic of osteogenically differentiated MSCs compared to the SC control, which retained the small, elongated fibroblastic shape typical of undifferentiated MSCs. All samples except the negative control (SC) showed evidence of matrix mineralization when analysed using ESEM (Figure 5a). Two distinct crystal phases of the mineral were present, one visually characteristic of crystalline hydroxyapatite (HA; dendritic structure) and the other of amorphous HA (spherical, globular structure). Further evaluation using EDX analysis confirmed that the matrix was rich in calcium phosphate, giving Ca:P ratios in the range 1.53-1.59 (Figure 5b). FIB-TEM was used to examine the composition of the ECM of the multilayered cell culture close to the Thermanox disc, to ensure that matrix mineralization was taking place throughout the culture and not just on the surface (Figure 6). The four predominant elements detected by EDX mapping of the STEM section were found to be carbon, calcium, oxygen and phosphorous. Carbon and oxygen were identified to spatially occupy the same area, whilst calcium and phosphate were shown to be co-localized (Figure 6b). The carbon and oxygen signals are typical of general organic cellular matter. Trace amounts of oxygen were also detected in the same regions as the calcium and phosphorus, suggesting regions of mineralized matrix.

Transient serum exposure is compatible with chondrogenic differentiation
To assess whether the short serum exposure used in the above described protocol would be compatible with chondrogenesis, we performed a micromass pellet chondrogenesis assay. hMSCs were subjected to the chondrogenesis protocol in standard serum-free conditions (CS), or following a 5 day transient serum exposure (CS 5 ). In both cases, hMSCs showed a similar level of chondrogenic response, as shown by the measurement of GAG production ( Figure 7a) and by strong safranin-O staining (Figure 7b). In comparison, hMSC pellets treated in the continuous presence of serum (CS 21 ) failed to differentiate and remained negative, as was observed for hMSCs maintained in control medium devoid of chondrogenic cues (SC).

Discussion
In this study we have investigated the effect of reduced exposure to FCS on the differentiation of hMSCs. Although efficient differentiation was not achieved in the absence of serum throughout the treatment, our  Figure 5. ESEM analysis locating calcium phosphate depositions within the mineralized matrix after 21 days of culture in control medium (SC) or osteogenic medium with variable exposure to FCS (OS 5 , OS 21 , OS 0 ). ESEM images (a) show mineral deposition and a change in morphology typical of osteogenically differentiated MSCs. Dendritic and globular mineral deposits visible in OS 5 , OS 21 and OS 0 cultures are typical structures of crystalline and amorphous hydroxyapatite, respectively. (b) EDX spectra and Ca:P ratios confirming the presence of calcium and phosphate in the mineralized matrix after 21 days of culture with variable exposure to OS + FCS. Culture conditions were tested as defined in Figure 1 findings demonstrate that it is possible to achieve osteogenic differentiation, producing mineralizing osteoblastlike cells, without FCS continuously present in the osteogenic culture medium. These results support other studies suggesting that MSC cultures can be maintained in reduced-serum conditions (Gronthos and Simmons, 1995;Kuznetsov et al., 1997). Although serum-free formulations for the maintenance of MSC cultures are the subject of ongoing investigations (Sotiropoulou et al., 2006;Berger et al., 2006), existing studies indicate that human bone marrow stromal cells (BMSCs) demonstrated more rapid proliferation in FCS-containing medium, while little proliferation occurred in serum-free conditions, suggesting that the presence of high FCS was more effective than serum-free solutions (Kuznetsov et al., 2000). Results from our cellular assays also showed that the efficacy of the cultures with regard to proliferation and metabolism was greater in FCS-containing medium; however, the use of a 5 day boost with FCS followed by serum-free medium for the remaining culture period produced results with no significant difference compared to FCS exposure for the full 21 day period. The metabolic activity and proliferation rates of the OS 0 culture were still higher than the SC control, but significantly lower than those exposed to OS 21 , in agreement with studies showing that cells still proliferate without serum in the medium but often experience senescence, lowering the overall proliferation rates (Agata et al., 2009). With regard to osteogenic differentiation, the literature is unclear regarding serum-free treatments of MSCs without the use of serum replacements. Montzka et al. (2010) investigated the effect of reducing the serum content of the osteogenic medium to 2% from the standard 10%, and found osteogenic differentiation to be successful without the need for additional growth factors to the culture medium. It has so far remained unclear whether temporally reducing exposure to serum would enable successful differentiation. Our study indicates that with 5 days of exposure to FCS, efficient differentiation is obtained with no significant difference for both mineralization and osteogenic marker expression when compared to cultures with serum-containing osteogenic medium for the complete 21 day period. This underlines the important role of FCS in the initiation process and indicates that differentiation is not solely dependent on osteogenic supplement (ascorbic acid, dexamethasone and b-glycerophosphate).
Bone-specific alkaline phosphatase (ALP) plays a major role in bone calcification, specifically matrix mineralization, and therefore its expression during osteogenic differentiation is considered to be a useful marker of successful matrix mineralization, indicating the presence of osteoblastlike cells (Siffert, 1951). ALP expression was significantly lower in serum-free conditions and increased with exposure to FCS, fitting closely with the data from cell proliferation and metabolism assays. Similar results produced by Agata et al. (2009) show that ALP activity in serum-free expanded BMSCs was significantly lower than that of BMSCs expanded with serum, regardless of the duration of the osteogenic induction. In combination with the proliferation data from the current study, this suggests that OS 0 , OS 1 and OS 2 conditions may lead to many cells remaining in an undifferentiated state within the culture, explaining the low levels of ALP produced. Similar observations were reported when Friedman et al. (2006) found that lower ALP expression was seen in serum-free conditions before the addition of extra growth factors, such as BMP, to the culture medium. During the late stages of osteogenic differentiation, before the initiation of mineral deposition, ALP expression peaks and is usually followed by an increase in the expression of OCN, which occurs concurrently with matrix deposition and is therefore tightly associated with mineralization (Lian et al., 1999). The trend observed for OCN expression is similar to that seen for ALP expression, confirming that the production of osteoblast-like cells is higher in cultures exposed to FCS for longer, and that a 5 day boost gives statistically similar data to OS 21 culture conditions. This suggests that there may be many cells in an undifferentiated state remaining in the conditions with minimal or no exposure to FCS.
Interestingly, it was previously reported that when bone marrow stromal cells precultured in serum-containing medium were trypsinized and treated with osteogenic supplement without FCS for 3 days before transplantation, bone formation was sustained in vivo when compared to cultures treated with serum-free medium throughout, suggesting that biphasic culture approaches for osteogenic stimulation would be compatible with efficient differentiation in vivo (Kuznetsov et al., 2000). Our observations further indicate that both proliferation rates and osteogenic differentiation can be maintained in vitro at levels similar to FCS-containing conditions if the serum boost is exercised for 5 days before switching to serum-free conditions. Analysis of the Ca 2+ incorporation into the extracellular matrix with alizarin red (Titorencu et al., 2007) confirmed that 5 days of FCS inclusion in the treatment successfully produced a heavily mineralized matrix made up of calcium phosphate at day 21. It will be interesting to further evaluate whether a similar outcome can be achieved at earlier time points, such as day 14, in order to further optimize the differentiation protocol. Although the complete exclusion of FCS from OS treatment produced some calcium deposits, quantification showed that mineralization was significantly lower, in agreement with data from RT-PCR, ELISA and immunostaining, suggesting that little mineralization was achieved in the absence of any serum exposure. In addition to the presence of calcium nodules confirmed by staining, ESEM revealed two different phases of the mineralized matrix in MSC cultures: a visually crystalline phase with a sharp dendritic structure, and a globular non-crystalline phase. The use of EDX confirmed that these nodules were made up of calcium phosphate with Ca:P ratios close to that of hydroxyapatite (~1.67), indicating that osteogenic differentiation and successful matrix mineralization had been achieved with transient serum treatment.
Furthermore, our results show that the inclusion of the 5 day phase of FCS exposure supports both mineralization and successful chondrogenic differentiation of hMSCs, indicating that temporal restriction can overcome the negative impact of long-term serum use on chondrogenic cultures (Wang et al., 2010;Lee et al., 2009;Malpeli et al., 2004). Further evaluation of this transient protocol will be useful to establish the effect of short serum exposure on the levels and kinetics of chondrogenic matrix component synthesis such as collagen II and X. The results presented here underline the benefit of developing new approaches involving transient exposure regimes in order to support dual differentiation from a single cell source. Further improvements to this approach will seek to address remaining issues related to the use of biological products such as serum, including risks associated with possible pathogens, including prions and viruses, as this will be critical for subsequent translational applications (Liu et al., 2007).
These results provide the basis for the development of a cellularized scaffold from a single material, in a single chamber, with the aim of creating two distinct regions harbouring osteochondral and osteogenic lineages. The identification of a culture regime compatible with both lineages allows the design of new tissue-engineering approaches, in which naive MSCs seeded on a scaffold and maintained in this biphasic culture medium can be directed towards both osteogenic and chondrogenic fates by physical (such as graduated pore size) and biochemical (such as growth factor release) cues provided locally within the scaffold. Such strategies combining a united supportive differentiation medium and localized fate regulators can enable the production of single scaffolds with gradual osteochondral properties through a natural interface, as seen in the osteochondral region of a joint.

Conclusions
In conclusion, we have identified a culture protocol that allows successful osteogenic differentiation of hMSCs with osteogenic supplement but without the need for long-term exposure to FCS. Our data show that it is possible to treat the cells with osteogenic supplement and FCS for 5 days and then remove the serum, achieving a level of differentiation that is comparable to that observed in cells cultured in osteogenic supplement with FCS for the full culture period. Our study also shows that this strategy of short-term serum treatment successfully maintains the chondrogenic potential of MSCs. These findings represent the first step for the development of new culture approaches aiming to achieve the combined osteogenic and chondrogenic differentiation of naive MSCs seeded in single scaffolds, in a single bioreactor, for joint repair. This strategy can now be refined and integrated with other cues provided to the cells within the scaffold itself, such as localized growth factor release or surface chemistry, in order to allow the biological patterning of cellularized scaffolds for the repair of small skeletal defects.