State Key Laboratory of Oral Diseases, Sichuan University, Chengdu, People's Republic of China
Department of Orthopaedic Surgery, Massachusetts General Hospital for Children and the Pediatric Orthopaedic Laboratory for Tissue Engineering and Regenerative Medicine, Harvard Medical School, Boston, MA, U.S.A.
Department of Orthopaedic Surgery, Massachusetts General Hospital for Children and the Pediatric Orthopaedic Laboratory for Tissue Engineering and Regenerative Medicine, Harvard Medical School, Boston, MA, U.S.A.
Recent research has shown that adipose tissues contain abundant MSCs (mesenchymal stem cells). The origin and location of the adipose stem cells, however, remain unknown, presenting an obstacle to the further purification and study of these cells. In the present study, we aimed at investigating the origins of adipose stem cells. α-SMA (α-smooth muscle actin) is one of the markers of pericytes. We harvested ASCs (adipose stromal cells) from α-SMA-GFP (green fluorescent protein) transgenic mice and sorted them into GFP-positive and GFP-negative cells by FACS. Multilineage differentiation tests were applied to examine the pluripotent ability of the α-SMA-GFP-positive and -negative cells. Immunofluorescent staining for α-SMA and PDGF-Rβ (platelet-derived growth factor receptor β) were applied to identify the α-SMA-GFP-positive cells. Then α-SMA-GFP-positive cells were loaded on a collagen—fibronectin gel with endothelial cells to test their vascularization ability both in vitro and in vivo. Results show that, in adipose tissue, all of the α-SMA-GFP-positive cells congregate around the blood vessels. Only the α-SMA-GFP-positive cells have multilineage differentiation ability, while the α-SMA-GFP-negative cells can only differentiate in an adipogenic direction. The α-SMA-GFP-positive cells maintained expression of α-SMA during multilineage differentiation. The α-SMA-GFP-positive cells can promote the vascularization of endothelial cells in three-dimensional culture both in vitro and in vivo. We conclude that the adipose stem cells originate from perivascular cells and congregate around blood vessels.
Stem cells are self-renewing, have long-term cell viability, and possess multilineage differentiation potential, all of which make them advantageous as seeding cells for regenerative medicine (Langer and Vacanti, 1993; Peng and Zhou, 2009; Reyes et al., 2001). Although embryonic stem cells have multipotent differentiation ability, their use has been restricted owing to ethical and potential immunorejective problems. BMSCs (bone-marrow-derived stem cells) have excellent capacity to proliferate in culture and differentiate into a wide variety of cell types (Cai et al., 2007; Wu et al., 2010). Unfortunately, there remains concern about the discomfort and pain associated with bone marrow procurement, low numbers of BMSCs and their heterogeneous nature (Li Pira et al., 2006).
Recent work has demonstrated that adipose tissue is not only a reservoir of energy, but also a repository of stem cells. ASCs (adipose stromal cells) can be induced to differentiate into adipocytes, osteoblasts (Lin et al., 2008a), chondrocytes and myocytes under specific culture conditions (Lin et al., 2006; Zuk et al., 2002). Adipose tissue is derived from the embryonic mesenchyme and contains a stroma in which the ASCs can be induced to multiple lineages in specific culture systems similar to BMSCs (Cowan et al., 2004; Qi et al., 2009). Compared with BMSCs, ASCs are easier to obtain, have relatively lower donor-site morbidity and are available in large quantities at harvest (Lin et al., 2005; Miranville et al., 2004). There is more cell homogeneity in ASCs, but apparent heterogeneity in BMSCs (Janssen et al., 1994). For these reasons, ASCs represent a promising alternative to BMSCs for use in cell-based musculoskeletal tissue-engineering strategies (Zuk et al., 2001).
Many questions remain about ASCs, including the locale where they reside within adipose tissue, the existence of specific markers for adipose stem cells and how the adipose stem cells can be purified (Traktuev et al., 2008; Zannettino et al., 2008; Zimmerlin et al., 2010). Previously a subpopulation of human perivascular cells that express both pericyte and MSC (mesenchymal stem cell) markers in situ (Andersen et al., 2008; Crisan et al., 2008) have been reported. The isolated population can expand and is clonally multipotent in culture, hypothesizing that MSCs found throughout fetal and adult tissues are members of the pericyte family of cells (Amos et al., 2008; Caplan, 2008). However, direct proof of the origin and location of the stem cells in tissues was not obtained (Kalajzic et al., 2008; Khan et al., 2008). Our recent work, however, has confirmed that bone marrow derived pluripotent cells arise from pericytes that contribute to vascularization modelled by in vitro culture (Cai et al., 2009).
In the present study, we sought to use α-SMA (α-smooth muscle actin), a promoter, to direct the expression of enhanced GFP (green fluorescent protein), a transgenic model to study the differentiation of the adipose stem cells and determine their origin in order to better direct future research efforts (Kalajzic et al., 2008; Yokota et al., 2006). We set out to obtain purified α-SMA-GFP-positive cells from the adipose tissue of transgenic mice. We then applied multilineage differentiation and immunostaining of pericyte markers to identify the characteristics of the α-SMA-GFP-positive cells. Finally, the purified adipose derived α-SMA-GFP-positive cells were co-cultured with endothelial cells in a three-dimensional collagen—fibronectin gel to act as a source of perivascular cells to generate a vascular network in vitro and in vivo.
Histological analysis of adipose tissue from α-SMA-GFP transgenic mouse
In the Oil Red O staining of the adipose tissue from α-SMA-GFP transgenic mice (Figure 1A), the adipose tissue consisted of mature adipocytes (black arrow) and connective tissue. There were many blood vessels in the connective tissues (white arrow). By fluorescence microscopy of the adipose tissue, the green fluorescence was expressed exclusively along the blood vessels (Figure 1B). No green fluorescent cells were evident in any other region of the adipose tissue. These results strongly suggest that the α-SMA-GFP-positive cells reside exclusively around the blood vessels.
Immunofluorescent staining of CD31 (Figures 1C–1F) and vWF (Figures 1G–1J) were applied to identify the position of the α-SMA-GFP-positive cells. Figure 1(C) demonstrates DAPI (4′,6-diamidino-2-phenylindole) staining of the adipose tissue. In Figure 1(D) of the same field, the α-SMA-GFP-positive cells distributed around the blood vessels. Figure 1(E) showed that the endothelial cells were CD31-positive (red). The merged picture demonstrates that all of the α-SMA-GFP-positive cells reside around the endothelial cells and occupy the abluminal wall of the blood vessels (Figure 1F). vWF is another typical marker of endothelial cells (Figures 1G–1J). Figure 1(G) shows the DAPI staining of adipose tissue. Figure 1(H) of the same field shows the α-SMA-GFP-positive cells located around the blood vessel. Figure 1(I) shows that the endothelial cells are vWF-positive (red). The merged picture demonstrates that α-SMA-GFP-positive cells surround the endothelial cells and form the abluminal wall of the blood vessel (Figure 1J).
Characterization of α-SMA-GFP-positive cells
Different passages of ASCs from α-SMA-GFP transgenic mice were analysed by FACS to determine the percentage of α-SMA-GFP-positive cells. In passage 0 of ASCs, the percentage of α-SMA-GFP-positive cells was 31±5.11%; in passage 1, the percentage was 56±8.15%; in passage 2, the percentage was 79±8.70%; and in passage 3, the percentage was 91±5.45% (Figure 2A).
The number of sorted α-SMA-GFP-positive and -negative cells were counted at different time points to construct a growth curve. The initial number of both GFP-positive and GFP-negative cells were 1×104. At day 3, the α-SMA-GFP-positive cells proliferated into (2.1±0.2)×104 cells and the negative cells were (1.6±0.1)×104. At day 6, there were (4.5±0.4)×104 GFP-positive cells and (2.0±0.3)×104 GFP-negative cells. At day 9, there were (8.1±0.6)×104 GFP-positive cells and (3.1±0.5)×104 GFP-negative cells. At day 12, there were (11.0±0.9)×104 GFP-positive cells and (4.0±0.8)×104 GFP-negative cells. According to the growth curve, α-SMA-GFP-positive cells proliferated much faster than the α-SMA-GFP-negative cells (Figure 2B). After passage 4, FACS results showed that most of the ASCs were GFP-positive (Figures 2C and 2D).
The primary ASCs of the α-SMA-GFP transgenic mouse are polygonal in shape and only some of them were GFP-positive (Figures 3A–3C). After the third passage, most of the ASCs were GFP-positive (Figures 3D–3F). The morphology of cultured DsRed-labelled HUVEC cells is shown for comparison (Figures 3G–3I).
Immunostaining for the pericyte's characteristic proteins, α-SMA and PDGF-Rβ, shows that the α-SMA-GFP-positive cells express both GFP and α-SMA in the same cells (Figures 3J–3L). These GFP-expressing cells are also PDGF-Rβ-positive (Figures 3M–3O).
Multilineage differentiation of GFP-positive and GFP-negative ASCs: adipogenesis
Seven days after being transferred into adipogenic medium, the morphology of induced α-SMA-GFP-positive ASCs modulated from an elongated to a round shape. Two weeks after induction, there were more lipid vesicle-filled cells (Figure 4A). Lipids appear black on the green fluorescence microscopic images (Figure 4B). The α-SMA-GFP expression remained positive during adipogenesis (Figure 4B). The adipogenic differentiation was demonstrated by positive Oil Red O staining of lipid droplets. The lipid droplets were red in fluorescence microscopy (Figure 4C). In contrast, GFP-negative ASCs expressed weak adipogenic differentiation after induction. The expression of adipocyte-specific genes PPARγ (peroxisome-proliferator-activated receptor γ) and LPL (lipoprotein lipase) was also examined using RT—PCR. The level of PPARγ and LPL in the α-SMA-GFP-positive ASCs is much higher than in the GFP-negative cells (Figure 4J).
Both GFP-positive and GFP-negative ASCs were placed in osteogenic medium. Five days after induction, the GFP-positive cells changed from an elongated fibroblast-like shape to a multilateral form. After culture for 14 days, mineralized nodules were observed (Figure 4D). α-SMA-GFP remained positive throughout the process of osteogenesis (Figure 4E). The mineralized nodules were assessed by AR-S staining (Figure 4F). Similar changes were not observed in the GFP-negative ASCs. To further verify osteogenesis, cells were examined by RT—PCR for the expression of the osteoblast-specific transcription factor OSX (osterix) (Figure 4J). OSX was observed in differentiated α-SMA-GFP-positive ASCs but not observed in the negative cells.
To prove chondrogenesis, the α-SMA-GFP-positive and -negative ASCs were placed into chondrogenic medium for 7 days. The cells changed their appearance from fibroblast-like to flat and multi-angled (Figure 4G). The α-SMA-GFP remained positive throughout chondrogenesis (Figure 4H). The cells were then pelleted by centrifugation and cultured as a small mass for another 3 weeks. The pellet was next stained with Toluidine Blue. The staining demonstrated large nuclei with multiple nucleoli similar to chondrocytes. Most of the cells were surrounded by proteoglycan (Figure 4I). To further verify chondrogenesis, cells were examined by RT—PCR for the expression of SOX9 [SRY (sex determining region Y)-box 9] and Col-II genes (Figure 4J). Both SOX9 and Col-II were observed in differentiated α-SMA-GFP-positive ASCs and were not observed in the GFP-negative cells.
Formation of vascular network within the three-dimensional collagen—fibronectin gels in vitro
The ability of α-SMA-GFP-positive cells to assist HUVECs in forming and maintaining a network was evaluated by loading in a three-dimensional collagen—fibronectin gel in vitro (Figure 5). The gels were cultured and observed under fluorescence microscopy to demonstrate the presence of an extensive network. In the experimental group, the HUVECs were cultured with α-SMA-GFP-positive ASCs from adipose tissue. HUVECs began to grow and contact with other cells at day 1 (Figure 5A). The α-SMA-GFP-positive cells were distributed around the HUVECs (Figures 5B and 5C). At day 4, the HUVECs formed tube-like structures and connected to each other (Figure 5E), while the α-SMA-GFP-positive cells were distributed around the tubes (Figures 5F and 5G). At day 7, the networks were primarily formed by HUVECs (Figure 5I) and the α-SMA-GFP-positive cells became denser around the networks (Figures 5J and 5K). At day 14, the HUVECs formed a robust network (Figure 5M) with the α-SMA-GFP-positive cells distributed around the network (Figures 5N and 5O). In the control group, HUVEC cultures without α-SMA-GFP-positive cells proliferated slowly and could not form a vascular network at day 1, day 4, day 7 or day 14 (Figures 5D, 5H, 5L and 5P).
The formation of vascular structure was analysed from images of the different culture groups using ImageJ software. The control HUVECs alone group at day 1 was set as standard. The fold-increase in formation of vascular structure by co-culture of α-SMA-GFP-positive ASCs with HUVECs was 3.010±0.39 (day 1), 6.03±1.64 (day 4), 9.53±1.55 (day 7) and 12.54±2.28 (day 14), with significant differences from the control (*P<0.01) (Figure 6).
Formation of vascular network within the three-dimensional collagen—fibronectin gels in vivo
In order to evaluate the role of α-SMA-GFP-positive cells in blood-vessel formation, cells seeded in a collagen—fibronectin gel were subcutaneously implanted into nude mice. The size of in vivo implants became smaller with time, reducing to 10–20% of their original size after 2 weeks. In the HUVECs alone control group, some discontinuous blood vessels formed by DsRed-labelled HUVECs were observed at 1 week after implantation (Figure 7A). These vessels became continuous and more complex at 2 weeks (Figure 7C). In the experimental group (α-SMA-GFP-positive cells plus HUVECs), continuous red networks surrounded by green pericyte-like cells were observed at 1 week (Figure 7B). After 2 weeks in vivo, more robust networks surrounded by green pericyte-like cells were observed (Figure 7D). The blank control injections of pure collagen—fibronectin gel disappeared and nothing was retrievable after 1 week.
The in vivo formation of vascular structure was analysed by ImageJ in the different groups of gel implants (Figure 7E). The fold-increase in formation of vascular structure in the experimental group (α-SMA-GFP-positive cells plus HUVECs) was 2.27±0.09 (1 week) and 4.00±0.62 (2 weeks), with significant differences from the control (*P<0.05).
Based on previous work, ASCs are thought to reside around small blood vessels (Andersen et al., 2008; Lin et al., 2008b; Tang et al., 2008). It has been proven that MSCs are members of the pericyte family of cells from different fetal and adult tissues (Amos et al., 2008; Caplan, 2008; Crisan et al., 2008; Doherty et al., 1998). Our previous work showed that bone-marrow-derived pluripotent cells are pericyte-like and can contribute to vascularization. In this study, we sought to prove that these ASCs reside within the pericyte population surrounding these blood vessels in adipose tissue as well. We initially identified this population of cells in α-SMA-GFP transgenic mice and localized the perivascular cells. We then subjected the GFP-positive and GFP-negative cells to differentiation culture conditions to assess the pluripotent potential of the suspected ASCs. Only the GFP-positive cells demonstrated pluripotent properties, as evidenced by their commitment to adipogenesis, osteogenesis and chondrogenesis, while the GFP-negative cells did not. We have provided proof that the adipose pluripotent stem cells originate from perivascular cells and express α-SMA during differentiation. The perivascular cells are mainly composed of pericytes and the population of smooth muscle cells derived from pericytes. We determined that ASCs isolated from adipose tissue originated from the GFP-positive perivascular cells and quite probably are a subset of the pericytes located around adipose tissue blood vessels.
While the population of pericytes in adipose tissue is large, not all of them are adipose stem cells (Kahn, 2008; Peng and Zhou, 2009; Planat-Benard et al., 2004). It is likely that only a limited number of pericytes are adipose stem cells (Zannettino et al., 2008; Zimmerlin et al., 2010). An important goal is to identify specific characteristics and markers of adipose stem cells that allow enrichment (Rehman et al., 2004).
Vascularization is a major obstacle for the clinical application of regenerative medicine because cells in the centre of engineered constructs have insufficient oxygen and nutrients (Qi et al., 2009; Tuan et al., 2003). This often results in necrosis or inflammation and subsequent failure of tissue implantation or transplantation (Melero-Martin et al., 2008; Traktuev et al., 2008). Engineered tissues must rapidly generate a vascular network with connections to the host vasculature for adequate nutrients and oxygen, and to eliminate waste. Pericytes play an important role in vascularization which is critical to wound healing and tissue regeneration (Doherty et al., 1998; Kang et al., 2009). The pericytes can stimulate the endothelial cells to proliferate and form networks, then pericytes support the external wall of the blood vessels (Miranville et al., 2004; Traktuev et al., 2009). The interaction between the endothelial cells and pericytes is still unclear (Benedito et al., 2009; Liu et al., 2009). Our study suggests that ASCs originate from pericytes and can contribute to vascularization both in vitro and in vivo. The ASCs not only demonstrate potential for multilineage differentiation, but also can synergize with endothelial cells to form vascular networks. These ASCs may therefore be ideal seeding cells for vascularized tissue engineering.
ASCs originate from perivascular cells and congregate around blood vessels, these cells can promote the vascularization of endothelial cells in three-dimensional culture both in vitro and in vivo. Adipose stem cells may be ideal seeding cells for vascularized tissue engineering.
Materials and methods
Experimental mouse model
To study the origination and differentiation of the adipose stem cells, we utilized α-SMA, a promoter, to direct the expression of enhanced GFP (Kalajzic et al., 2008). The α-SMA-GFP transgenic mice originally developed by Jen-Yue Tsai (NEI/NIH) were obtained through the courtesy of John D. Ash and Sanai Sato (University of Oklahoma Health Science Center). The strain was rederived and characterized at the Bone Biology Research Laboratory, Department of Orthopaedic Surgery, Children's Hospital Boston.
Histological analysis of adipose tissue from α-SMA-GFP mice
Two-week-old α-SMA-GFP transgenic mice were used for this study. The animal use was reviewed and approved by the IACUC at Harvard Medical School and conducted in accordance with NIH and AAALAC guidelines. The inguinal fat-pads were harvested and divided into fresh frozen tissue sections 10 μm thick and mounted on slides. The slides were air-dried for 45 min at 37°C and fixed in ice-cold 10% formalin for 10 min. They were then air-dried for another 45 min and rinsed in distilled water three times.
For histological staining, slides were placed in absolute propylene glycol for 2 min. They were next stained in Oil Red O solution for 10 min at 65°C, differentiated in 85% propylene glycol solution for 2 min, stained in haematoxylin for 30 s and washed thoroughly in running tap water for 3 min.
For immunostaining, the slides were incubated with 3% H2O2 in methanol for 30 min. After washing with PBS, they were blocked in 1% BSA and 1.5% normal goat serum at room temperature (25°C) for 30 min. Slides were then incubated overnight at 4°C with goat anti-mouse polyclonal antibodies against CD31 and vWF (von Willebrand factor; Santa Cruz Biotechnology, Santa Cruz, CA, U.S.A.) for endothelial cells. The slides were then incubated with secondary rabbit anti-goat IgG—Alexa Fluor® 594 (Invitrogen, Eugene, OR, U.S.A.)
Isolation and characterization of ASCs from α-SMA-GFP mice
Mice were killed by CO2 inhalation. The inguinal fat-pads were harvested and extensively washed with sterile PBS. They were then incubated with 0.075% type II collagenase (Sigma—Aldrich, St Louis, MO, U.S.A.) in PBS for 60 min at 37°C with vigorous agitation. After neutralization of the collagenase, ASCs released from adipose specimens were filtered and collected by centrifugation at 1200 g for 10 min. The pellet was then re-suspended, washed three times with medium and sent for FACS. The ASCs were sorted into two groups: GFP-positive and GFP-negative. The GFP-positive and GFP-negative cells were seeded on to plastic tissue-culture dishes separately in control medium containing α-MEM (Minimal Essential Medium), 10% FBS (fetal bovine serum), 100 units/ml penicillin and 100 μg/ml streptomycin.
The percentages of α-SMA-GFP-positive cells of each passage of ASCs were analysed by FACS. The cells were sorted into two groups: the α-SMA-GFP-positive population against the negative group. The sorted α-SMA-GFP-positive cells are expected to be α-SMA-positive because the GFP transgene expression is directed by the α-SMA promoter. Immunofluorescent staining for α-SMA and PDGF-Rβ (platelet-derived growth factor receptor β) were carried out to differentiate α-SMA-GFP-positive cells and -negative cells. α-SMA and PDGF-Rβ are markers of pericytes. If the α-SMA-GFP-positive cells express both α-SMA and PDGF-Rβ, the adipose tissue-derived α-SMA-GFP-positive cells should derive from pericytes.
For immunostaining, the cells were blocked in 1% BSA and 0.1% Triton X-100 at room temperature for 30 min. Slides were then incubated for 1 h at 4°C with mouse anti-mouse monoclonal antibodies against α-SMA and rabbit anti-mouse polyclonal antibodies against PDGF-Rβ (Santa Cruz Biotechnology, CA, U.S.A.). The slides were then incubated with secondary goat anti-mouse and goat anti-rabbit IgG Alexa Fluor® 594 (Invitrogen) for 30 min.
HUVECs (human umbilical vein endothelial cells)
HUVECs were purchased from the A.T.T.C. and labelled with DsRed by lentivirus vector. HUVECs were maintained in 0.1% gelatin-coated plates in EGM-2 (endothelial cell growth medium-2; Lonza).
Multilineage differentiation of α-SMA-GFP-positive and -negative ASCs
After culture and expansion in the control medium, the third passage GFP-positive and GFP-negative ASCs were trypsinized and reseeded on to 24-well plates at a density of 104 cells per well. The cells were incubated in the control medium for 2 days and then changed to specific medium (Table 1) to induce multilineage differentiation of ASCs. We utilized three mice for each experiment, and each trial was performed at least three times.
Table 1. Lineage-specific differentiation induced by medium supplementation
The adipogenic cells were assessed by Oil Red O staining. The slide was rinsed in PBS and fixed in 4% (w/v) paraformaldehyde for 15 min, then stained with 1% Oil Red O for 10 min. The slide was rinsed with 60% propan-2-ol and washed with 70% ethanol. The slides were then observed under fluorescence microscopy for intracellular red lipid droplets (Wu et al., 2010).
The mineralization nodules characteristic of the osteogenic lineage were stained with AR-S (Alizarin Red-S). The sections on each slide were rinsed in PBS and incubated with 40 mM AR-S (in 1% KOH) with rotation for 10 min. They were then rinsed in PBS and observed through the microscope.
After the ASCs were placed into chondrogenic medium for 7 days, the cells were digested, pelleted by centrifugation at 1200 g for 5 min and cultured as a small mass for another 2 weeks. The cells were next stained in Toluidine Blue for 10 s and washed in water and differentiated in 0.2% uranyl nitrate until the background appeared colourless.
RNA isolation and RT—PCR (reverse transcription—PCR)
Total RNA was extracted using RNeasy Mini Kit (Qiagen Sciences, Gaithersburg, MD, U.S.A.), and quality and concentration determined by absorbance. Then 0.5 μg was converted into cDNA using MMLV (Moloney-murine-leukaemia virus) reverse transcriptase (New England Biolabs, Beverly MA, U.S.A.). PCR amplification of target mRNA was performed by TaKaRa PCR kit. PCR oligonucleotide primers are listed in Table 2. All primers were determined through established GenBank® sequences. The products were electrophoresed on 2% agarose gels, stained with ethidium bromide and visualized with Quantity One software (Bio-Rad).
Table 2. Specific primers for PCR amplification
Three-dimensional collagen—fibronectin gel preparation and formation of vascular network in vitro
Collagen/fibronectin gels were prepared as previously described with minor modifications (Khan et al., 2008; Verseijden et al., 2010). HUVECs and α-SMA-GFP-positive cells were mixed at a ratio of 4:1 and suspended in a gel that included 1.5 mg/ml collagen gel (PureCol), 90 μg/ml human plasma fibronectin (Gibco), 25 mM Hepes (Gibco), 38.5% complete EGM-2 (Lonza), with the final cell concentration 2×106 cells/ml. The control group had only HUVECs. The cell suspensions were polymerized in a 48-well plate (100 μl/well) for 30 min at 37°C. Then the gels were covered with 250 μl of complete EGM-2. In order to quantify formation of vascular networks, three random microscopic fields for each group were imaged at ×10 magnification with a Leica DMI 6000B microscope, and the average area of the vascular network (tubular structure formed by red HUVECs) was analysed by NIH Image/ImageJ 1.42.
Formation of vascular network in vivo
Two-month-old female nude mice were anaesthetized with 0.4 mg/g avertin. In the experimental group, the 250 μl mixtures of collagen—fibronectin gels were loaded with HUVECs and α-SMA-GFP-positive cells. In the control group, 250 μl mixtures of collagen—fibronectin gels were loaded only with HUVECs. The gels were injected subcutaneously on the dorsum of each mouse. At 1 week and 2 weeks after implantation, mice were anaesthetized again (n=3 for each time point). Implanted constructs were then carefully separated from the surrounding fibrous capsule, washed in PBS and immediately observed under a laser scanning confocal microscope. The images were analysed by NIH Image/ImageJ 1.42 to evaluate the formation of new blood vessels.
All of the experiments were repeated a minimum of three times and representative data are presented as means±S.D. ANOVA was used to analyse the difference within groups in all assays above. To specify significant difference between experimental groups and the control, Dunnett's t test was conducted. P<0.05 shows significant differences in all t tests.
Yunfeng Lin and Brian Grottkau designed the research; Xiaoxiao Cai and Yunfeng Lin performed the research; Xiaoxiao Cai, Yunfeng Lin, Peter Hauschka and Brian Grottkau analysed the data; and Yunfeng Lin, Peter Hauschka and Brian Grottkau wrote the paper.
This work was supported by the Peabody Foundation, Inc., the Constance and Anthony A. Franchi Fund for Pediatric Orthopaedics at the Massachusetts General Hospital for Children, National Natural Science Foundation of China [grant numbers 30801304, 81071273], New Century Excellent Talents [grant number NCET-08-0373], National Excellent Dissertation [grant number FANEDD 200977], Young Scientists in Sichuan [grant number 2010JQ0066]. The Department of Defense/CDMRP Breast Cancer Research Program under grants managed by the U.S. Army Materiel and Research Command [grant numbers W81XWH-09-1-0292 and W81XWH-03-1-0607] (to P.V.H.), and the Department of Orthopaedic Surgery, Children's Hospital Boston.