Biomaterials 31 (2010) 8684e8695 Contents lists available at ScienceDirect Biomaterials journal homepage: www.elsevier.com/locate/biomaterials A comparison of bioreactors for culture of fetal mesenchymal stem cells for bone tissue engineering Zhi-Yong Zhang a, c, Swee Hin Teoh a, b, **, Erin Yiling Teo a, b, Mark Seow Khoon Chong c, Chong Woon Shin d, Foo Toon Tien d, Mahesh A. Choolani c, Jerry K.Y. Chan c, e, f, * a Centre for Biomedical Materials Applications and Technology (BIOMAT), Department of Mechanical Engineering, Faculty of Engineering, National University of Singapore, Singapore National University of Singapore Tissue Engineering Programme (NUSTEP), National University of Singapore, Singapore Experimental Fetal Medicine Group, Department of Obstetrics & Gynaecology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore d Bioengineering Laboratory, Technology Centre for Life Sciences, Singapore Polytechnic, Singapore e Department of Reproductive Medicine, KK Women’s and Children’s Hospital, Singapore f Cancer and Stem Cell Biology Program, Duke-NUS Graduate Medical School, Singapore b c a r t i c l e i n f o a b s t r a c t Article history: Received 27 May 2010 Accepted 28 July 2010 Available online 24 August 2010 Bioreactors provide a dynamic culture system for efficient exchange of nutrients and mechanical stimulus necessary for the generation of effective tissue engineered bone grafts (TEBG). We have shown that biaxial rotating (BXR) bioreactor-matured human fetal mesenchymal stem cell (hfMSC) mediated-TEBG can heal a rat critical sized femoral defect. However, it is not known whether optimal bioreactors exist for bone TE (BTE) applications. We systematically compared this BXR bioreactor with three most commonly used systems: Spinner Flask (SF), Perfusion and Rotating Wall Vessel (RWV) bioreactors, for their application in BTE. The BXR bioreactor achieved higher levels of cellularity and confluence (1.4e2.5x, p < 0.05) in large 785 mm3 macroporous scaffolds not achieved in the other bioreactors operating in optimal settings. BXR bioreactor-treated scaffolds experienced earlier and more robust osteogenic differentiation on von Kossa staining, ALP induction (1.2e1.6, p < 0.01) and calcium deposition (1.3 e2.3, p < 0.01). We developed a Micro CT quantification method which demonstrated homogenous distribution of hfMSC in BXR bioreactor-treated grafts, but not with the other three. BXR bioreactor enabled superior cellular proliferation, spatial distribution and osteogenic induction of hfMSC over other commonly used bioreactors. In addition, we developed and validated a non-invasive quantitative micro CT-based technique for analyzing neo-tissue formation and its spatial distribution within scaffolds. Ó 2010 Elsevier Ltd. All rights reserved. Keywords: Bioreactor Bone tissue engineering Micro CT Mesenchymal stem cells 1. Introduction Bone graft is the second most transplanted tissue in the world, with more than 1.5 million transplantation performed in United States annually [1,2]. However, this demanding need for effective bone grafts to treat non-union fractures cannot be fulfilled by existing bone grafts. This is due to the limited availability and donorsite morbidity associated with the use of autografts [3,4], the * Corresponding author. Experimental Fetal Medicine Group, Department of Obstetrics & Gynaecology, Yong Loo Lin School of Medicine, National University of Singapore, Singapore 119228. Fax: þ65 6779 4753. ** Corresponding author. Centre for Biomedical Materials Applications and Technology (BIOMAT), Department of Mechanical Engineering, Faculty of Engineering, National University of Singapore, Singapore. E-mail addresses: mpetsh@nus.edu.sg (S.H. Teoh), jerrychan@nus.edu.sg (J.K.Y. Chan). 0142-9612/$ e see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.biomaterials.2010.07.097 significant risk of disease transmission and immune reaction with the use of allografts [5]and vulnerability of fatigue in synthetic grafts secondary to their inability to remodel [6]. Bone tissue engineering (BTE) has been proposed as a promising strategy to develop tissue engineered bone grafts (TEBG), which are not only available off-theshelf like allo- and synthetic grafts, but also have potent bone repair capacity like autografts. The success of BTE strategy requires the integrated technological advances from research fields of scaffold, stem cell and bioreactor culture system [7]. Mesenchymal Stem Cells (MSC) are readily isolated, nonimmunogenic and have well defined osteogenic differentiation pathways, and have been investigated as osteogenic cell sources for BTE [8e11]. However, the clinical use of human adult MSC has been hindered by their low frequencies in vivo, limited proliferation capacity and early senescence in culture, and hence has limited capacity for generating adequate cell numbers for clinical applications, especially in the older age groups [12]. More recently, human Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 fetal MSC (hfMSC) have been identified and characterized from ontologically primitive sources [13,14]. In a head-to-head comparative study, hfMSC have significantly higher proliferative and osteogenic potential and lower immunogenicity than MSC types isolated from perinatal or postnatal origins, suggesting their greater potential as allogeneic osteogenic cellular source for BTE [15]. Dynamic bioreactor culture systems are essential for the in vitro cultivation and maturation of TE bone grafts, especially for larger grafts where the core of the scaffold is more than 200 mm from the surface. Bioreactors improve the mass transport of nutrients and allow the diffusion limitation of traditional static culture, which is generally taken to be around 200 mm, to be overcome [16,17]. In addition, the dynamic media flow applies a mechanical stimulus to the cells, enhancing cellular osteogenesis and mineralization through triggering of mechano-transduction signaling pathways [18,19]. Currently, several types of bioreactors have been developed for BTE applications. This includes the Spinner Flask (SF) bioreactor, Perfusion bioreactor and Rotating Wall Vessel (RWV) bioreactor, all of which have been shown to promote cellular proliferative and osteogenic differentiation [18,20e22]. Notably, most of these bioreactors are uni-axial in design, which may place a constraint on the homogenous flow pattern of media. In order to overcome this, we have developed a Biaxial Rotating (BXR) bioreactor which integrates a biaxial rotating wall vessel design with a media perfusion system, evidenced with improved flow dynamics over uni-axial rotation under in-silico simulation [23]. The dynamic culture and osteogenic priming of hfMSC-mediated macroporous polycaprolactone/ tri-calcium phosphate (PCL-TCP) scaffolds in this BXR bioreactor generated an effective TEBG which resulted in the healing of a critical sized defect [24,25]. While the use of bioreactors for the culture of MSC for BTE has been reported by several groups, there is a paucity of studies which have compared different bioreactors in an optimised manner with the use of relevant osteogenic cell sources loaded onto macroporous scaffolds suitable for BTE. In this study, we performed a head-to-head comparison of this BXR bioreactor with SF, Perfusion and RWV bioreactors for BTE. After optimization of each individual parameter for BXR, SF and Perfusion bioreactors, we conducted a systematic comparison of the bioreactors and their effect on the proliferation and osteogenic differentiation of hfMSC seeded within macroporous PCL-TCP scaffolds. 2. Materials and methods 2.1. Samples and ethics Fetal tissue collection was approved by the Domain Specific Review Board of National University Hospital (D06/154), Singapore in compliance with international guidelines regarding the use of fetal tissue for research [26]. Pregnant women gave separate written consent for the clinical procedure and for the use of fetal tissue for research purposes. And fetal tissues were collected from fetuses after clinically indicated termination of pregnancy. Fetal gestational age was determined by crownrump length measurement. In this study, a fetal sample at 14þ2 (weeks þ days) gestations was utilized for both preliminary and comparative experiments. 2.2. Isolation and characterization of hfMSCs hfMSC were isolated through plastic adherence, and culture expansion, and characterized through immunophenotyping, colony-forming capacity and trilineage differentiation into osteoblasts, adipocytes and chondrocytes as previously described [15,24]. Briefly single-cell suspensions of fetal bone marrow were prepared by flushing the marrow cells out of humeri and femurs using a 22-gauge needle into Dulbecco’s modified Eagle’s medium (DMEM, Sigma, USA)eGlutaMAX (GIBCO, USA) supplemented with 10% fetal bovine serum (FBS), 50 U/ml penicillin, and 50 mg/ml streptomycin (GIBCO, USA) (referred as D10 medium), and then plated onto 100 mm dishes at 106 mononuclear cells/ml in D10 medium. Media were changed every 2e3 days and non adherent cells were removed, and sub-cultured at 104/cm2 to sub-confluence. hfMSCs at passage 3 were used for characterization and at passage 4 were used for both preliminary and comparative experiments. 8685 2.3. Experimental design 2.3.1. Generation of macroporous PCL-TCP scaffolds A bioactive PCL-TCP composite scaffold with high porosity was fabricated using the fused deposition modeling technique, which results in a honey-comb-like pattern of triangular pores with a porosity of 70%, average pore size of 0.523 mm, and a lay-down pattern of 0/60/120 as previously described (Fig. 1A) [15]. 2.3.2. Seeding hfMSC to PCL-TCP scaffold hfMSC were seeded onto the porous scaffolds by adding cell suspension media to scaffolds (seeding density 2500 cells/mm3 scaffold), placed in 24-well culture plates, and incubated for 3 h in an incubator to allow for the initial cellular attachment to the scaffolds. Thereafter, 3 ml of D10 medium was slowly added to each well and hfMSC cellular scaffolds were incubated in a humidified atmosphere at 37  C and 5% CO2 for 1 week with D10 medium changes 3 times a week to acclimatize the cellular scaffolds before dynamic culture (Fig. 1B). 2.3.3. Optimisation of bioreactor parameters First, a “Preliminary Experiment” was designed to optimize the culture conditions and setting of the bioreactor parameters for BXR, Perfusion and SF bioreactors (Fig. 1B). These were a biaxial rotation speed of 2, 5, 10 rpm with a fixed perfusion flow rate of 3.5 ml/min for BXR bioreactor; a perfusion flow rate of 0.2, 2.0 and 5.0 ml/min for Perfusion bioreactor; and a stirring speed of 10, 20 and 30 rpm for SF bioreactor. After seeding hfMSC to PCL-TCP scaffolds (4  4  4 mm, Fig. 1A) in a dropwise manner and pre-cultured for one week as above, hfMSC cellular scaffolds were then transferred to BXR, Perfusion and SF bioreactors operated in different culture conditions for 1 week and assessed for cellular adhesion, proliferation and viability (n ¼ 4). 2.3.4. Comparative experiment After the optimal culture condition has been determined, a comparative experiment was conducted to compare BXR, Perfusion, SF, and RWV bioreactors for BTE application (Fig. 1B). During the experiment, BXR, Perfusion and SF bioreactors were operated at the optimal conditions derived from the initial experiment, while RWV bioreactor was operated at the rotation speed of 47 rpm and perfusion flow rate of 3.5 ml/min, which achieved free suspension of the scaffolds in the media. We use large cylindrical shaped scaffolds (10 mm in diameter and height, Fig. 1A) for this comparative study. After the loading of the cells and pre-culture in static condition for 1 week as above, hfMSC cellular scaffolds were randomly divided into four groups and cultured in osteogenic induction media (D10 medium supplemented with 10 mM b-glycerophosphate, 10 8 M dexamethasone and 0.2 mM ascorbic acid) within the four different bioreactors over 4 weeks. Each group was compared for cellular adhesion, proliferation, viability, distribution within the scaffold, osteogenic differentiation and mineralization. In all bioreactors, media were changed once a week over four weeks, with 28 ml of medium change per scaffold per week being kept as a constant between groups to ensure identical access to nutrients. The bottom end of each cylindrical scaffold was defined as the side of the scaffold sitting on the culture plate during the initial static loading of the scaffold. 2.3.5. Bioreactor setups The BXR bioreactor consists of a spherical culture chamber, where the cellular scaffolds are anchored to the cap of bioreactor by pins, a medium reservoir and a perfusion system, which connects culture chamber and medium reservoir, as previously described [23,24]. The spherical culture chamber is designed to rotate in two perpendicular axes (Y and Z, as indicated in blue block arrows, Fig. 1C) simultaneously and perfused with media flow circulating between culture chamber and medium reservoir (as indicated by red arrows, Fig. 1C). The perfusion bioreactor consists of a number of perfusion culture chambers, where cellular scaffolds are placed, a medium reservoir and a peristaltic pump (Masterflex, USA). Culture chambers, reservoir and pump are connected to each other in series, with medium circulating among each other (as indicated by red arrows, Fig. 1C). The SF bioreactor (Bellco Biotechnology, USA) consists of a culture chamber, where the cellular scaffolds are anchored to the cap of culture chamber by pins, and a magnetic stir bar, which can rotate in the Z-axis (as indicated by blue block arrows, Fig. 1C). Finally, an RWV bioreactor (Synthecon, USA), consisting of a horizontally rotated culture chamber, where the cellular scaffolds are suspended, and a perfusion system with media continuously flowing through the culture chamber (as indicated in red arrows, Fig. 1C) was used. The culture chamber can rotate in the X-axis (as indicated in blue block arrows, Fig. 1C) at certain speed to suspend the cellular scaffolds in a free floating culture condition. All four bioreactor systems were placed within an incubator during the culture. 2.4. Cellular adhesion and proliferation of hfMSC cellular scaffolds The morphology of hfMSC in 3D culture, cellular adhesion and extracellular matrix (ECM) production were examined weekly by Phase Contrast Light Microscope (PCLM) over 4 weeks. The qualitative analysis of cell viability in 3D was performed by fluorescein di-acetate/propidium iodide (FDA/PI) staining, where FDA stains viable cells green, and PI stains necrotic and apoptotic cell nuclei red. Cellular 8686 Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 Fig. 1. PCL-TCP scaffolds, experimental design and bioreactor illustrations. (A) Macroporous PCL-TCP 3D scaffolds as shown by the SEM image were used for this study: cubic scaffolds (4  4  4 mm) were used for preliminary experiment, while cylinder-shaped scaffolds with a diameter of 10 mm and height of 10 mm were used for comparative experiment. (B) Initially, a preliminary experiment was done to determine the optimal culture conditions for BXR, Perfusion and SF bioreactors. This was followed by a comparative experiment, the four types of bioreactors were compared for BTE applications using the optimal condition as determined. (C) Photographs and schematic illustrations of their design and working patterns of the four bioreactor systems used in this study. scaffolds were rinsed in PBS and stained with FDA/PI as previously described [15], and viewed under a confocal laser microscope (Olympus, FV1000 Fluoview, Japan). The total cell number in the 3D cellular scaffold on week 0 (the time of transfer to the bioreactors, as illustrated by Fig. 1B) 1, 2, 4 (n ¼ 3) were estimated by quantifying the double stranded DNA (dsDNA) content of each scaffold using a PicoGreen dsDNA Quantification Kit (Molecular Probes, USA). The total dsDNA was extracted through enzymatic digestion as previously described [15] and assayed by following the manufacturer’s instruction. The proliferation of the hfMSC inside 3D scaffold was interpreted by changes of dsDNA quantity. 2.5. Neo-tissue formation and spatial distribution analysis by micro CT Micro CT, which is usually used for hard tissue imaging and analysis [27], was explored in this study to investigate the neo-tissue distribution within the large scaffolds. After 4-week bioreactor culture, cellular scaffolds (n ¼ 4 per group) were harvested and fixed in 10% neutral buffered formalin, placed in the sample holder and scanned through 180 with a rotation step of 1 at a spatial resolution of 35 mm in a micro CT machine (Skyscan, Belgium). An averaging of 5 and a 1 mm aluminum filter were used during the scanning. The scan files were reconstructed at a step size of 1 using a modified Feldkamp algorithm as provided by Skyscan (Belgium). The reconstructed data were loaded onto the 3D modeling software, VGstudio (Volume Graphics GmbH, Germany) to stack the 2D image into a 3D model for quantitative histomorphometric analysis; a low global thresholdings were applied to visualize the neo-tissue together with scaffold material. Four acellular scaffolds, cultured in osteogenic induction media for 4 weeks, were utilized as negative controls for the neo-tissue volume quantification through subtraction of the scaffold material. Furthermore, the neo-tissue volumes of different regions (Top, Middle and Bottom regions, Fig. 4D) of large scaffolds were calculated respectively in order to investigate the spatial distribution of neo-tissue within the large scaffolds. 2.6. Osteogenic differentiation and mineralization of cellular scaffolds 2.6.1. ALP activity assay The intracellular ALP of hfMSC scaffolds under different bioreactors culture were compared at week 0, 1, 2, 4 (n ¼ 3 per group/time point). Cell lysates were Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 8687 Fig. 1. (continued). tested for ALP activity using SensoLyteÔ pNPP Alkaline Phosphatase Assay Kit (AnaSpec USA) and the ALP activities were normalized to the total protein content determined using the Bradford assay (Bio-Rad Laboratories, US) as previously described [15]. 2.6.2. von Kossa staining hfMSC cellular scaffolds were stained with von Kossa at week 4. Briefly samples were gently rinsed twice with PBS then fixed in 10% formalin for 1 h, and washed in water. Finally, they were stained with freshly made 2% silver nitrate in water (w/v) for 10 min in the dark and expose to sun light for 3 h. 2.6.3. Calcium content assay The calcium content of the hfMSC cellular scaffolds at week 1, 2 and 4 (n ¼ 3) were assayed as previously described [15]. Briefly, the calcium deposition is dissolved in 0.4 ml 0.5 N acetic acid and determined by a colorimetric assay using calcium assay kit (BioAssay Systems, USA). Control cell-free empty scaffolds cultured as above (n ¼ 3) were used as a negative control to offset the elution of calcium from the tri-calcium phosphate component in the PCL-TCP scaffold. 2.6.4. Scanning electron microscope (SEM) and energy dispersive X-ray spectrometer (EDX) analysis Cellular scaffolds were dehydrated, gold sputtered, viewed under the SEM and element component of crystal-like structure inside the samples was analyzed by EDX as previously described [15]. 2.7. Statistical analyses All the data have been represented as mean  SD, and compared using either two-way ANOVA or student t-test. A value of P < 0.05 was taken as significance. 3. Results 3.1. Characterization of hfMSC hfMSC isolated from the bone marrow grew as spindleshaped cells as monolayers. They expressed a typical MSC 8688 Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 immunophenotype, being positive for mesenchymal markers CD105 (SH2), CD73 (SH3, SH4); intracellular marker vimentin and laminin; cell adhesion molecules CD29, CD44, CD 106, & CD 90; and negative for haemopoietic and endothelial markers CD14, CD34, CD45, CD31, vWF and HLA II. Moreover, small subpopulations of MSC (22e55%) stained positively for HLA I and the embryonic stem cell markers Oct-4 and Nanog as previously described [15,24,25,28]. Under permissive culture conditions, hfMSC differentiated into osteogenic, adipogenic and chondrogenic lineages. 3.2. Optimization of bioreactor parameters In order to optimize the bioreactors for their individual maximal performance, we subjected macroporous hfMSC-PCL/TCP scaffolds (4  4  4 mm3 ¼ 64 mm3) over a range of parameters to achieve maximal cellular proliferation. hfMSC in the BXR bioreactor proliferated most rapidly and reached confluence within one week when rotated at 5 rpm biaxially compared to 2 and 10 rpm as evidenced with both FDA/PI staining and PCLM (Fig. 2A). This was corroborated by the measurement of total dsDNA content within scaffolds (Picogreen dsDNA assay), which showed a 1.3 and 1.7 fold higher cellularity at 5 rpm over 2 and 10 rpm conditions respectively (Fig. 2A, p < 0.05; p < 0.01, n ¼ 4). FDA/PI staining showed a high degree of cellular viability preserved in cellular scaffolds under all culture conditions. Similarly, for the optimization studies of Perfusion and SF bioreactors, maximal cellular proliferation was achieved at 2.0 ml/min perfusion rate (Perfusion bioreactor) and 20 rpm stirring speed (SF bioreactor) with high cellular viability in both groups (Fig. 2B and C). The microscopy findings were in line with Picogreen dsDNA assay, which showed significantly more cellular amount in the above-mentioned optimal culture condition compared to the other conditions (1.3e1.5 for Perfusion bioreactor; 1.4e2.1 for SF bioreactor; p < 0.05, p < 0.01; n ¼ 4; Fig. 2B and C). RWV bioreactor settings were placed to ensure that the scaffolds were freely floating. 3.3. Cellular adhesion, proliferation and viability Using the optimal settings for each bioreactor, we examined the performance of all four bioreactors with large volume scaffolds (cylindrical shaped, 785 mm3). Under BXR bioreactor culture, hfMSC proliferated rapidly, adhered and saturated the pores within the scaffolds by the end of one week, as seen under PCLM (Fig. 3A), while the Picogreen dsDNA assay showed a further increase of dsDNA content from week 1 to 2 before plateauing between week 2 and 4 (Fig. 3B). In contrast, cellular confluence in Perfusion, SF and RWV bioreactors were incomplete, with large pores within the scaffolds devoid of cells being found in all three groups at the end of the experiment at week 4 (red circles, Fig. 3A). Compared to the BXR bioreactor, Perfusion, SF and RWV bioreactors cultured cellular scaffolds resulted in lower cellularity at 1 (0.4e0.6, dsDNA analysis, p < 0.05, n ¼ 3, Fig. 3B) and 4 weeks of culture (0.6e0.7, p < 0.05, n ¼ 3, Fig. 3B). Dynamic culture in all four bioreactor systems were capable of increasing mass transfer of nutrients as evidenced by the high degree of cellular viability deep in the core of these large scaffolds (5000 mm from the surface of the scaffolds, FDA/PI staining, Fig. 4A). Z-stack confocal imaging of FDA/PI staining showed the highest cellular content in scaffolds cultured in BXR bioreactor among the four groups. After four weeks of dynamic culture, scaffolds in the Perfusion, SF and RWV bioreactors were still not confluent, with the presence of defects within the scaffolds (red circles, Fig. 4A), corroborating the earlier PCLM and Picogreen findings (Fig. 3). 3.4. MicroCT assessment of neo-tissue formation and spatial distribution A micro CT was explored to qualitatively and quantitatively analyze the formation and spatial distribution of neo-tissue composed of hfMSC and ECM within the large scaffolds. By applying a low global threshold of 26, the scaffolds and neo-tissue can be visualized and analyzed without picking up empty parts of the scaffold. After four week of dynamic culture in BXR bioreactor, there was complete occupancy of the macroporous structure, as seen in all the cross-section views of different regions, except for a central defect which accommodated the anchoring pin (Fig. 4B). On the other hand, Perfusion, SF and RWV bioreactor cultured scaffolds demonstrated significantly lower amounts of neo-tissue (0.66e0.76, p < 0.05, n ¼ 4, Fig. 4B and C), which was in keeping with the Picogreen dsDNA assays (Fig. 3B). Quantitative measurement of neo-tissue amount in different cross-sectional regions of scaffolds was possible, and demonstrated a homogenous pattern in BXR bioreactor cultured scaffolds, with similar amounts of neo-tissue being distributed in the top, middle and bottom regions of these large scaffolds. This was not the case in scaffolds grown in Perfusion, SF and RWV bioreactor cultures, where the neo-tissue had a predilection towards the bottom region of scaffolds compared to other regions (Fig. 4D). 3.5. Osteogenic differentiation and mineralization of hfMSC in bioreactors BXR bioreactor cultured scaffolds deposited higher quantities of extracellular calcium crystals compared to other bioreactors cultured scaffolds by week 2 (red arrows, Fig. 3A), and limited the passage of light through scaffolds by week 4 (Fig. 3A). This observation was corroborated by the darker von Kossa staining of the BXR bioreactor cultured scaffolds over other bioreactors cultured ones at week 4 (Fig. 5A). SEM revealed higher levels of crystal-like ECM deposition in cellular scaffolds cultured under BXR bioreactor versus other bioreactors (Fig. 5B). These deposits (red circle, Fig. 5B) were calcium phosphate salts as shown through element component analysis by EDX to consist mainly of P, Ca and O elements (Fig. 5C). Quantitatively, BXR bioreactor cultured scaffolds expressed higher levels of ALP activity, a key indicator for osteoblastic induction, than other bioreactors at week 2 and 4 (1.2e1.6, p < 0.05, p < 0.01, n ¼ 3, Fig. 5D). BXR bioreactor cultured scaffolds laid down the highest amount of calcium crystals, followed by Perfusion, SF, and RWV bioreactor cultured scaffolds at week 4 (1.3e2.3, p < 0.01, n ¼ 3, Fig. 5E). 4. Discussion Bioreactors have been widely used in BTE for their capacity to improve mass transfer in large cellular scaffold constructs, and to apply favorable mechanical stimuli to facilitate the osteogenic differentiation of cellular scaffolds [20,29]. While many different bioreactors have been developed and investigated for bone tissue engineering applications, there had been no studies which compared them systematically in a head-to-head manner after optimising their culture parameters. Hence, it is not known how different bioreactors will perform against each other. In this present study, we compared four types of bioreactors regarding their potentials for BTE application. Firstly, we optimised each bioreactor for cellular proliferation within macroporous scaffolds in order to avoid bias due to sub-optimal settings. This was followed by a comparative study with hfMSC-mediated large PCL-TCP scaffolds achieving significantly higher cellular proliferation, increased neotissue formation with uniform spatial distribution, and more robust Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 8689 Fig. 2. Optimisation of culture conditions for BXR, Perfusion and SF bioreactors (preliminary experiment). (A) In optimising the various bioreactors, BXR bioreactor achieved the highest cellular proliferation at 5 rpm rotational speed while (B) a 2 ml/min perfusion rate was optimal for the Perfusion bioreactor as assessed through FDA/PI confocal microscopy (confocal z-stack images, constructed from 44 horizontal image sections with 300 mm in depth Mag. 100) PCLM and Picogreen DNA Assay. (Green dye is taken up by live cells) (C) The SF bioreactor achieved maximal cellular proliferation at 20 rpm rotation speed (*p < 0.05; **p < 0.01, n ¼ 4). FDA/PI (Florescein Di-Acetate/Propidium Iodide). 8690 Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 8691 Fig. 4. Cellular viability and micro CT analysis of neo-tissue formation and distribution. (A) All the bioreactor cultured hfMSC cellular scaffolds maintained a high cellular viability throughout the 4 weeks, while cellular scaffolds under BXR and Perfusion bioreactor culture showed more cellularity than the other two bioreactors culture by FDA/PI staining (confocal z-stack images, constructed from 44 horizontal image sections with 300 mm in depth Mag. 100). (B) Micro CT were used to analyze the growth and spatial distribution of neo-tissue (cells and ECM) in the large scaffolds; BXR bioreactor culture led to the fully confluence of neo-tissue throughout the whole scaffolds (the central defect was created by the anchoring pin), while other bioreactor cultured scaffolds resulted in lower levels of cellular confluence and non-homogenous spatial distribution within the large scaffolds, as seen in the cross-section images at different region of scaffolds (threshold ¼ 26). (C) Quantitative analysis by micro CT showed a significant higher neo-tissue volume in BXR bioreactor group compared to others (*p < 0.05, threshold ¼ 26, n ¼ 4); and (D) Quantitative analysis of neo-tissue amount in three different regions of scaffolds showed a uniform spatial distribution in BXR bioreactor cultured scaffolds, while cellular distribution favoured the bottom end of the scaffold, as defined during static culture when cultured in the other three bioreactors (threshold ¼ 26, n ¼ 4). osteogenic differentiation and mineralization under BXR bioreactor culture over the other three groups. We have also demonstrated the feasibility of using a micro CT technique as a simple and nondestructive modality to analyze the formation and spatial distribution of neo-tissue in the 3D scaffolds qualitatively and quantitatively. Among a number of bioreactors which have been developed, SF, Perfusion and RWV bioreactors are most widely used and Fig. 3. Cell adhesion and proliferation of hfMSC cellular scaffolds cultured in different bioreactor systems. (A) BXR bioreactor cultured cellular scaffolds achieved complete cellular confluence in the first week of culture, deposited extracellular crystals (red arrows) in the second week and highly laden with extracellular minerals which affected the passage of light through scaffolds at week 4. Scaffolds cultured in the other three bioreactors experienced slower cellular proliferation and less crystal deposition (red arrows) comparatively. After four weeks of culture in the Perfusion bioreactor, small perfusion holes (red circles) were found in the majority of the pores of cellular scaffolds. (B) Picogreen assay showed highest cellularity of cellular scaffolds cultured in BXR bioreactor throughout the whole study, followed by the Perfusion bioreactor group, with SF and RWV bioreactors culture resulting in the lowest dsDNA content (*p < 0.05, n ¼ 3). 8692 Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 Fig. 4. (continued). investigated for BTE applications, which allowed promotion of cellular proliferation, and in vitro and in vivo osteogenesis [22,30e32]. Comparisons among these bioreactors have been reported in several studies [33e36]. However, their results have been controversial. For example, Meinel et al. work [35] suggested that SF bioreactor culture was more favorable for osteogenic differentiation when compared to Perfusion bioreactor culture, which was at odds with Goldstein et al. report [34]. While the differences between the studies could be due in part to the use of different cells and scaffolds, a more likely reason could be the comparison of bioreactors using non-optimised parameters. This could easily lead to bias where the optimal condition of one bioreactor was compared with non-optimal conditions in another. Hence, we meticulously mapped the optimal conditions for the proliferation of hfMSC in each bioreactor, with the parameters set to achieve free suspension culture in the RWV bioreactor. In addition, we controlled the amount of medium available to each scaffold to ensure identical access to nutrient availability. Our comparative experiment showed macroporous hfMSC loaded-scaffolds cultured under BXR bioreactor achieved the best cellular proliferation, followed by Perfusion bioreactor, while both SF and RWV bioreactor culture resulted in the least cellular proliferation. This was in turn mirrored by the degree of osteogenic induction and differentiation, as measured quantitatively by induction of ALP and laying down of extracellular minerals, which was more robust in scaffolds cultured in the BXR bioreactor, then SF and Perfusion bioreactors, and finally the RWV bioreactor. These findings are in agreement with Sikavitsas et al. report which demonstrated that SF bioreactor performed better than RWV bioreactor for the proliferation and osteogenesis [33]. This also compared well with Goldstein et al. report, which showed higher ALP expression in rat MSC embedded in large cylindrical scaffolds (12.7 mm diameter  6 mm height) in both Perfusion and SF bioreactors culture compared with RWV bioreactor culture [34]. Our results however, conflicted with Meinel et al. study [35], which documented better cellular proliferation of human adult MSC seeded onto 3 mm thick collagen scaffolds in SF compared to Perfusion bioreactors. It is however, not clear whether optimised parameters for flow and spin settings were used in Meinel’s study. In addition, the volume of medium was not equally supplied in both groups. Moreover, Meinel el al. had used smaller microporous scaffolds of only 3 mm in thickness. In contrast, we studied the use of hfMSC seeded into larger macroporous osteoconductive scaffolds (10 mm in diameter and 10 mm in thickness), with carefully optimised bioreactor parameters, allowing insights towards the potential use of these bioreactors for clinical applications to be gained. The SF bioreactor is designed to improve the mass transport at the construct surface by continuous stirring of the culture medium (Fig. 1C). It has been widely used in BTE because of its simple design and effectiveness in promoting cell proliferation and osteogenic differentiation [22,32,35,37,38]. However, the strong turbulent flow generated by the stirring have been shown to be detrimental to the cells and newly formed ECM [19,33]. In our present study, the compromised cell growth and ECM deposition could thus be due to the continuous exposure to strong turbulent flow, which has been shown in other’s study [33]. The RWV bioreactor provides a dynamic culture environment by rotating at a speed which allows the free suspension of the cellular constructs (Fig. 1C). This induces a high mass transfer while maintaining a low shear stress suitable for BTE applications [21,39e40]. However, this bioreactor is beset with problems of random collisions occurring between scaffolds themselves and the culture chamber. This is especially so with the culture of larger scaffolds, resulting in cellular damage and disruption of cellular attachment and matrix deposition onto the scaffolds. This could possibly account for its worst performance reported here and in a number of other studies [34,41]. As the collisions between the scaffolds and the culture vessel wall is a random occurrence, a nonhomogenous distribution of neo-tissue is found within each scaffold, which has been observed in a few other reports [19,34]. Finally, the microgravity environment generated through the RWV bioreactor has been implicated in the disruption of osteoblastic function and maturation [42,43], which may contribute to its poor performance in the induction of osteogenic differentiation of hfMSC (Fig. 5). Perfusion bioreactor enhances the mass transfer of cellular scaffolds not only at the periphery but also within its internal pores of the scaffolds (Fig. 1C) [30], and has been shown to enhance cellular proliferation and osteogenic stimulation of cellular scaffold constructs [6,44e46]. However, one major concern relates to a “washout” phenomenon at the front of the scaffolds facing the oncoming flow, resulting in a less homogenous cellular distribution [47]. In keeping with this, we have observed less neo-tissue distributed at the front zone of the scaffolds as compared to the middle and rear zones (Fig. 4B). In addition, we observed the presence of “perfusion holes” after 4-weeks of Perfusion bioreactor culture (Figs. 3A and 4A), which contributed to the non-homogenous and non-confluent scaffold. The BXR bioreactor has a hybrid design derived from conventional RWV bioreactor and Perfusion bioreactors, with improved design features to address specific deficiencies as described above (Fig. 1C). Firstly, the culture chamber is designed to rotate in two perpendicular axes simultaneously, which should generate a more homogenous media flow than with uni-axial rotation, as shown with in-silico simulations [23]. Secondly, a perfusion system was incorporated to circulate the media between the culture chamber and reservoir, allowing increased mass transfer with low shear Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 8693 Fig. 5. Osteogenic differentiation and mineralization of hfMSC cellular scaffolds. (A) von Kossa staining demonstrated higher levels of mineralization of scaffolds cultured in BXR bioreactor than others, as shown in darker von Kossa staining at week 4. (B) SEM images of cellular scaffolds at week 4 of culture showed more ECM and higher degree of mineralization in BXR cultured scaffolds compared to others. (C) EDX analysis of the element components revealed the mineralized nodules as calcium phosphate salts, consisting of P, Ca and O elements. (D) Quantitative ALP activity assay the highest level of ALP activity in BXR bioreactor cultured scaffolds than others at week 2 and 4, although the ALP activity of scaffolds cultured in SF bioreactor was the highest in the first week (*p < 0.05, **p < 0.01, n ¼ 3). (E) Analysis of calcium deposition in the scaffolds revealed significantly higher calcium deposition in BXR bioreactor cultured scaffolds than others, and the least calcium deposition in RWV bioreactor cultured scaffolds after the four-week bioreactor culture (**p < 0.01, n ¼ 3). 8694 Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 Fig. 5. (continued). stress. However, unlike the conventional Perfusion bioreactor, cultured scaffolds in BXR bioreactor are not located directly in the flow stream, thus avoiding the “cellular washout” problem found in Perfusion bioreactors. Lastly, the BXR bioreactor utilizes the anchorage culture of cellular scaffolds, which are secured by pins, instead of the free suspension culture found in RWV bioreactor, eliminating the risk of scaffold collisions with the chamber walls and overcoming the adverse effect of microgravity on the proliferation and maturation of osteogenic cells [43,48]. These design features are likely to be responsible for its superior performance for the culture and osteogenic induction of hfMSC-mediated large macroporous TEBG over other bioreactors. The formation and spatial distribution of neo-tissue, both cells and ECM, within the scaffolds have been shown to profoundly influence in vivo bone formation [24,25]. Currently, there are two major categories of techniques to investigate the neo-tissue structure in cellular constructs [49]. Firstly, microscopy based techniques such as PCLM and confocal imaging are able to image cellular and tissue structure under high resolution. However, they are generally semi-quantitative and are limited to imaging the superficial surface layer. Another commonly utilized technique is based on assaying cellular lysates for enzymes, proteins or DNA, such as the ALP and Picogreen dsDNA assays used in this study. These assays are quantitative and highly sensitivity, but does not yield any information related to the spatial distribution of neo-tissue. In this study, we have explored the use of micro CT imaging technique to investigate the formation and spatial distribution of neo-tissue in a non-invasive manner. Micro CT has been widely used for bone tissue analysis, ever since the pioneering work by Feldkamp’s group [50,51]. It has attracted increasing interest for its ability to image soft tissues at depth at high spatial resolution quantitatively non-invasively. The development of special staining techniques to increase the X-ray absorption of soft tissues using Xray contrast agents such as organically bound iodine, and osmium tetroxide has widened it utility considerably [52,53]. More recently, Dorsey et al. demonstrated the imaging of cells cultured in scaffolds after osmium tetroxide staining, although cells are generally transparent to X-ray [49]. In our present study, we successfully visualized and analyzed neo-tissue consisting mainly composed of hfMSC and ECM without the use of toxic X-ray contrast agents, such as osmium tetroxide. This will allow the timely assessment of such TEBG prior to their transplantation into animal models or for clinical applications. It is likely that the compact cellular structure and mineralized ECM deposited within the scaffolds increase their X-ray absorption, resulting in the visualization of neo-tissue structure at a relatively low thresholding. This non-invasive and non-destructive approach will allow the analysis of large bioreactor cultured scaffolds for evidence of homogenous cellular distribution and if necessary, osteogenic mineralization prior to their transplantation in both animal and clinical scenarios. Aside from the BXR bioreactor cultured scaffolds, we have shown that the bottom regions of larger scaffolds tend to generate more neo-tissues compared to the other regions of the scaffolds. This is likely to be due to the manner in which the hfMSC had been seeded, through static loading and incubation over 3 h. In such a technique, the majority of hfMSC would have settled towards the bottom region of scaffolds under the gravity. Notwithstanding this, following dynamic culture in BXR bioreactor, homogenous redistribution of the cells was achieved. However, scaffolds cultured in the other three bioreactors did not allow homogenous distribution of hfMSC, leading to non-homogenous cellular distributions in the eventual TEBG. This disparity between the bioreactors was not observed in the first set of experiments where hfMSC achieved confluence in all bioreactors tested when seeded in smaller macroporous scaffolds. 5. Conclusion The generation of osteogenic TEBG with BXR bioreactor allows homogenous seeding, rapid proliferation with high cellular viabilities, and potent osteogenic differentiation of hfMSC to be achieved more efficiently than with other commonly utilized bioreactor systems for BTE. In particular, the BXR bioreactor was the only bioreactor capable of achieving a homogenous cellular distribution within larger (785 mm3) macroporous scaffolds of clinical relevance. In addition, we report a micro CT technique for nondestructive imaging of neo-tissue distribution within large cellular tissue engineered grafts which will be of use for assessment prior to clinical transplantation. Acknowledgements The authors would like to thank these people for their help in this project: Tan Lay Geok from department of Obstetrics & Gynaecology; Teh Xiang Sheng, Heng Weiliang Dominic, Ang Xiang Yi Daryl, Foo Hui Juan Kenneth, Chng Yhee Cheng from Singapore Polytechnics. This work is supported by grants from the National Medical Research Council (NMRC1179/2008), the Ministry of Education in Singapore (T208B3114) and Tote Board Student Project Fund of Singapore Polytechnic (11-27801-2714). JC received salary support from the Clinician Scientist Award, NMRC (CSA/012/2009). Appendix Figures with essential color discrimination. All figures in this article are difficult to interpret in black and white. The full color images can be found in the on-line version, at doi:10.1016/j. biomaterials.2010.07.097. Z.-Y. Zhang et al. / Biomaterials 31 (2010) 8684e8695 References [1] Marsh D. Concepts of fracture union, delayed union, and nonunion. Clin Orthop Relat Res; 1998:22e30. [2] Salgado AJ, Coutinho OP, Reis RL. Bone tissue engineering: state of the art and future trends. Macromol Biosci 2004;4:743e65. [3] Banwart JC, Asher MA, Hassanein RS. Iliac crest bone graft harvest donor site morbidity. A statistical evaluation. Spine 1995;20:1055e60. [4] Goulet JA, Senunas LE, DeSilva GL, Greenfield ML. Autogenous iliac crest bone graft. Complications and functional assessment. Clin Orthop Relat Res; 1997:76e81. [5] Parikh SN. Bone graft substitutes: past, present, future. J Postgrad Med 2002;48:142e8. [6] Datta N, Pham QP, Sharma U, Sikavitsas VI, Jansen JA, Mikos AG. In vitro generated extracellular matrix and fluid shear stress synergistically enhance 3D osteoblastic differentiation. Proc Natl Acad Sci U S A 2006;103:2488e93. [7] Service RF. Tissue engineers build new bone. Science 2000;289:1498e500. [8] Ballas CB, Zielske SP, Gerson SL. Adult bone marrow stem cells for cell and gene therapies: implications for greater use. J Cell Biochem Suppl 2002;38:20e8. [9] Mauney JR, Volloch V, Kaplan DL. Role of adult mesenchymal stem cells in bone tissue engineering applications: current status and future prospects. Tissue Eng 2005;11:787e802. [10] Koc ON, Day J, Nieder M, Gerson SL, Lazarus HM, Krivit W. Allogeneic mesenchymal stem cell infusion for treatment of metachromatic leukodystrophy (MLD) and Hurler syndrome (MPS-IH). Bone Marrow Transplant 2002;30:215e22. [11] Di Nicola M, Carlo-Stella C, Magni M, Milanesi M, Longoni PD, Matteucci P, et al. Human bone marrow stromal cells suppress T-lymphocyte proliferation induced by cellular or nonspecific mitogenic stimuli. Blood 2002;99:3838e43. [12] Bruder SP, Jaiswal N, Haynesworth SE. Growth kinetics, self-renewal, and the osteogenic potential of purified human mesenchymal stem cells during extensive subcultivation and following cryopreservation. J Cell Biochem 1997;64:278e94. [13] Campagnoli C, Roberts IA, Kumar S, Bennett PR, Bellantuono I, Fisk NM. Identification of mesenchymal stem/progenitor cells in human first-trimester fetal blood, liver, and bone marrow. Blood 2001;98:2396e402. [14] Guillot PV, Gotherstrom C, Chan J, Kurata H, Fisk NM. Human first-trimester fetal MSC express pluripotency markers and grow faster and have longer telomeres than adult MSC. Stem Cells 2007;25:646e54. [15] Zhang ZY, Teoh SH, Chong MS, Schantz JT, Fisk NM, Choolani MA, et al. Superior osteogenic capacity for bone tissue engineering of fetal compared with perinatal and adult mesenchymal stem cells. Stem Cells 2009;27:126e37. [16] Ishaug SL, Crane GM, Miller MJ, Yasko AW, Yaszemski MJ, Mikos AG. Bone formation by three-dimensional stromal osteoblast culture in biodegradable polymer scaffolds. J Biomed Mater Res 1997;36:17e28. [17] Martin I, Obradovic B, Freed LE, Vunjak-Novakovic G. Method for quantitative analysis of glycosaminoglycan distribution in cultured natural and engineered cartilage. Ann Biomed Eng 1999;27:656e62. [18] Gomes ME, Sikavitsas VI, Behravesh E, Reis RL, Mikos AG. Effect of flow perfusion on the osteogenic differentiation of bone marrow stromal cells cultured on starch-based three-dimensional scaffolds. J Biomed Mater Res A 2003;67:87e95. [19] Chen HC, Hu YC. Bioreactors for tissue engineering. Biotechnol Lett 2006;28:1415e23. [20] Martin I, Wendt D, Heberer M. The role of bioreactors in tissue engineering. Trends Biotechnol 2004;22:80e6. [21] Marolt D, Augst A, Freed LE, Vepari C, Fajardo R, Patel N, et al. Bone and cartilage tissue constructs grown using human bone marrow stromal cells, silk scaffolds and rotating bioreactors. Biomaterials 2006;27:6138e49. [22] Stiehler M, Bunger C, Baatrup A, Lind M, Kassem M, Mygind T. Effect of dynamic 3-D culture on proliferation, distribution, and osteogenic differentiation of human mesenchymal stem cells. J Biomed Mater Res A; 2008. [23] Singh H, Teoh SH, Low HT, Hutmacher DW. Flow modelling within a scaffold under the influence of uni-axial and bi-axial bioreactor rotation. J Biotechnol 2005;119:181e96. [24] Zhang ZY, Teoh SH, Chong WS, Foo TT, Chng YC, Choolani M, et al. A biaxial rotating bioreactor for the culture of fetal mesenchymal stem cells for bone tissue engineering. Biomaterials 2009;30:2694e704. [25] Zhang ZY, Teoh SH, Chong MS, Lee ES, Tan LG, Mattar CN, et al. Neo-vascularization and bone formation mediated by fetal mesenchymal stem cell tissue-engineered bone grafts in critical-size femoral defects. Biomaterials 2010;31:608e20. [26] Polkinghorne J. Review of the guidance on the research use of fetuses and fetal material. London: HMSO; 1989. CM 762. [27] Guldberg RE, Lin AS, Coleman R, Robertson G, Duvall C. Microcomputed tomography imaging of skeletal development and growth. Birth Defects Res C Embryo Today 2004;72:250e9. [28] Lee ES, Chan J, Shuter B, Tan LG, Chong MS, Ramachandra DL, et al. Microgel iron oxide nanoparticles for tracking human fetal mesenchymal stem cells through magnetic resonance imaging. Stem Cells 2009;27:1921e31. 8695 [29] Bilodeau K, Mantovani D. Bioreactors for tissue engineering: focus on mechanical constraints. A comparative review. Tissue Eng 2006;12: 2367e83. [30] Bancroft GN, Sikavitsas VI, Mikos AG. Design of a flow perfusion bioreactor system for bone tissue-engineering applications. Tissue Eng 2003;9:549e54. [31] Sikavitsas VI, Bancroft GN, Holtorf HL, Jansen JA, Mikos AG. Mineralized matrix deposition by marrow stromal osteoblasts in 3D perfusion culture increases with increasing fluid shear forces. Proc Natl Acad Sci U S A 2003;100: 14683e8. [32] Song K, Liu T, Cui Z, Li X, Ma X. Three-dimensional fabrication of engineered bone with human bio-derived bone scaffolds in a rotating wall vessel bioreactor. J Biomed Mater Res A 2008;86:323e32. [33] Sikavitsas VI, Bancroft GN, Mikos AG. Formation of three-dimensional cell/ polymer constructs for bone tissue engineering in a spinner flask and a rotating wall vessel bioreactor. J Biomed Mater Res 2002;62:136e48. [34] Goldstein AS, Juarez TM, Helmke CD, Gustin MC, Mikos AG. Effect of convection on osteoblastic cell growth and function in biodegradable polymer foam scaffolds. Biomaterials 2001;22:1279e88. [35] Meinel L, Karageorgiou V, Fajardo R, Snyder B, Shinde-Patil V, Zichner L, et al. Bone tissue engineering using human mesenchymal stem cells: effects of scaffold material and medium flow. Ann Biomed Eng 2004;32:112e22. [36] Pound JC, Green DW, Chaudhuri JB, Mann S, Roach HI, Oreffo RO. Strategies to promote chondrogenesis and osteogenesis from human bone marrow cells and articular chondrocytes encapsulated in polysaccharide templates. Tissue Eng 2006;12:2789e99. [37] Kim HJ, Kim UJ, Leisk GG, Bayan C, Georgakoudi I, Kaplan DL. Bone regeneration on macroporous aqueous-derived silk 3-D scaffolds. Macromol Biosci 2007;7:643e55. [38] Mygind T, Stiehler M, Baatrup A, Li H, Zou X, Flyvbjerg A, et al. Mesenchymal stem cell ingrowth and differentiation on coralline hydroxyapatite scaffolds. Biomaterials 2007;28:1036e47. [39] Granet C, Laroche N, Vico L, Alexandre C, Lafage-Proust MH. Rotating-wall vessels, promising bioreactors for osteoblastic cell culture: comparison with other 3D conditions. Med Biol Eng Comput 1998;36:513e9. [40] Qiu QQ, Ducheyne P, Ayyaswamy PS. Fabrication, characterization and evaluation of bioceramic hollow microspheres used as microcarriers for 3-D bone tissue formation in rotating bioreactors. Biomaterials 1999;20:989e1001. [41] Yu X, Botchwey EA, Levine EM, Pollack SR, Laurencin CT. Bioreactor-based bone tissue engineering: the influence of dynamic flow on osteoblast phenotypic expression and matrix mineralization. Proc Natl Acad Sci U S A 2004;101:11203e8. [42] Ontiveros C, McCabe LR. Simulated microgravity suppresses osteoblast phenotype, Run2 levels and AP-1 transactivation. J Cell Biochem 2003;88:427e37. [43] Mukundakrishnan K, Ayyaswamy PS, Risbud M, Hu HH, Shapiro IM. Modeling of phosphate ion transfer to the surface of osteoblasts under normal gravity and simulated microgravity conditions. Ann N Y Acad Sci 2004;1027:85e98. [44] Holtorf HL, Sheffield TL, Ambrose CG, Jansen JA, Mikos AG. Flow perfusion culture of marrow stromal cells seeded on porous biphasic calcium phosphate ceramics. Ann Biomed Eng 2005;33:1238e48. [45] Sikavitsas VI, Bancroft GN, Lemoine JJ, Liebschner MA, Dauner M, Mikos AG. Flow perfusion enhances the calcified matrix deposition of marrow stromal cells in biodegradable nonwoven fiber mesh scaffolds. Ann Biomed Eng 2005;33:63e70. [46] Gomes ME, Bossano CM, Johnston CM, Reis RL, Mikos AG. In vitro localization of bone growth factors in constructs of biodegradable scaffolds seeded with marrow stromal cells and cultured in a flow perfusion bioreactor. Tissue Eng 2006;12:177e88. [47] Singh H, Ang ES, Lim TT, Hutmacher DW. Flow modeling in a novel nonperfusion conical bioreactor. Biotechnol Bioeng 2007;97:1291e9. [48] Ontiveros C, Irwin R, Wiseman RW, McCabe LR. Hypoxia suppresses run2 independent of modeled microgravity. J Cell Physiol 2004;200:169e76. [49] Dorsey SM, Lin-Gibson S, Simon CG. X-ray microcomputed tomography for the measurement of cell adhesionand proliferation in polymer scaffolds. Biomaterials 2009;30:2967e74. [50] Feldkamp LA, Goldstein SA, Parfitt AM, Jesion G, Kleerekoper M. The direct examination of three-dimensional bone architecture in vitro by computed tomography. J Bone Miner Res 1989;4:3e11. [51] Thomsen JS, Laib A, Koller B, Prohaska S, Mosekilde L, Gowin W. Stereological measures of trabecular bone structure: comparison of 3D micro computed tomography with 2D histological sections in human proximal tibial bone biopsies. J Microsc 2005;218:171e9. [52] Metscher BD. MicroCT for comparative morphology: simple staining methods allow high-contrast 3D imaging of diverse non-mineralized animal tissues. BMC Physiol 2009;9:11. [53] Faraj KA, Cuijpers VM, Wismans RG, Walboomers XF, Jansen JA, van KT, et al. Micro-computed tomographical imaging of soft biological materials using contrast techniques. Tissue Eng Part C Methods 2009;15:493e9.