Abstract
Embryonic stem (ES) cells are a uniquely self-renewing, pluripotent population of cells that must be differentiated before being useful for cell therapy. Since most studies utilize subcutaneous implantation to test the in vivo functionality of ES cell-derived cells, the objective of the current study was to develop an appropriate and clinically relevant in vivo implantation system in which the behavior and tumorigenicity of ES cell-derived cells could be effectively tested in a tissue-specific (orthotopic) site. Male ES cells were differentiated either into osteoblasts or chondrocytes using protocols that were previously developed and published by our laboratory. The differentiated cells were implanted into a burr-hole fracture created in the proximal tibiae of immunocompetent female mice, strain matched to the ES cell line. The ability of the differentiated ES cell-derived cells (bearing the Y chromosome) to incorporate into the newly formed bone was assessed by micro- computed tomography imaging and histochemistry. ES cells differentiated with either osteogenic or chondrogenic medium supplementation formed a soft tissue mass that disrupted the normal bone architecture by 4 weeks after implantation in some mice. In contrast, mice receiving osteoblastic cells that were differentiated in a three-dimensional type 1 collagen gel showed evidence of new bone formation at the defect site without evidence of tumor formation for up to 8 weeks after implantation. In this injury model, type 1 collagen is more effective than medium supplementation at driving more complete differentiation of ES cells, as evidenced by reducing their tumorigenicity. Overall, the current study emphasizes the importance of using an appropriate orthotopic implantation system to effectively test the behavior and tumorigenicity of the cells in vivo.
Introduction
Embryonic stem (ES) cells derived from the early blastocyst are pluripotent and possess the capacity for long-term self-renewal (7,12,21). This versatility makes ES cells an ideal source of cells for treating diseases and injuries in which insufficient cell numbers or activity limit the natural regeneration and repair of tissues (e.g., Parkinson's disease, spinal cord injury, diabetes, etc.). To successfully implement ES cell therapies in the clinic, it is necessary to develop in vitro techniques to expand, differentiate, and purify these cells. Before they can be used therapeutically, it must be shown that ES cells have not been adversely affected in vitro, which typically is assessed by cell viability and proliferation, by protein and gene expression of lineage-specific or pluripotent markers, and by karyotypic stability. While these attributes can be assayed to a certain extent in vitro, a more definitive functional assay of the quality of ES cell-derived cells is in vivo implantation, which is typically studied by subcutaneous injection into immunocompromised animals (5,8,10,25,26). Whereas subcutaneous implantation can help determine if ES cell-derived cells form the desired tissue type in vivo, it is unclear whether implantation at an ectopic site accurately predicts the fate and tumorigenicity of ES cell-derived cells. Unlike the ectopic subcutaneous injection of ES cell-derived cells into an immunocompromised host, implanting the cells into an appropriate, tissue-specific orthotopic microenvironment in an immunocompetent host provides a direct test of the capacity of ES cell-derived cells to restore the health of diseased or injured tissue in vivo, which is the ultimate goal of stem cell therapies.
For the current study, we used differentiation protocols that were previously developed by our group to drive ES cell differentiation into skeletal cell types. The efficacy of these protocols was reported previously (11,25,26). For example, when exposed to a collagen I matrix, murine ES cells differentiated into osteoblasts and by day 15, 75% of the cells were positive for osteopontin, osteocalcin, and osterix protein markers (11). Furthermore, these cell-loaded collagen matrices produced radio-opaque bony nodules and not tumors or teratomas, following subcutaneous implantation into immunecompromised mice (11). In a separate study, ES cells exposed to chondrogenic medium supplements (CHON medium) displayed a significant increase in glycosaminoglycan production, positive mRNA, and protein expression of collagen II and aggrecan and formed cartilaginous nodules following subcutaneous implantation of the cells into immunecompromised mice (25). By contrast, exposure of ES cells to osteogenic medium supplements (OB medium) lead to an increased expression in the osteoblast transcription factor core-binding factor subunit α-1/Runt-related transcription factor 2 (Cbfa1/runx2), osteocalcin, and calcium production was enhanced (26).
The objective of the current study was to compare the in vivo fate and functionality of these three ES cell-derived skeletal cells when implanted in an orthotopic (clinically relevant) skeletal injury model. The orthotopic site used in this study consisted of a burr-hole fracture (23) in the proximal tibiae of immunocompetent, strain-matched recipient mice. The burr-hole fracture model is not a critical-size defect; however, this injury model has a known geometry, is highly reproducibly, and is a stable fracture—requiring no external fixation—and thus effectively permits the assessment of cell functionality and tumorigenicity at an orthotopic site in vivo.
Materials and Methods
All supplies were obtained through Invitrogen (Carlsbad, CA, USA) unless otherwise noted.
ES Cell Culture
The pluripotent murine D3 ES cell line from male Sv129 mice (American Type Culture Collection, Rockville, MD, USA) was cultured on mitomycin C-treated (10 μg/ml, 37°C for 2 h; Sigma, Oakville, Ontario, Canada) mouse embryonic fibroblast (MEF) feeder layers on gelatin-coated tissue culture dishes. For regular maintenance, the cells were cultured in high-glucose Dulbecco's modified Eagle's medium (DMEM) supplemented with 15% fetal bovine serum (FBS), 1% nonessential amino acids (NEAA), 50 U/ml penicillin and 50 μg/ml streptomycin, 0.1 mM β-mercaptoethanol, and 1,000 U/ml leukemia inhibitory factor (LIF, Chemicon International/Millipore, Billerica, MA, USA). Undifferentiated D3 cultures were maintained in static tissue culture flasks and passaged every second day on MEFs.
Differentiation of ES Cells Using Media Supplementation
The differentiation of ES cells into osteoblasts and chondrocytes was carried out in micromass cultures using media supplements, as described previously (25,26). Briefly, ES cells were dissociated using 0.1% trypsin-EDTA (Invi trogen) and then cultured for 2 h at a high density, 1.0×105 cells per 10 ml×9 spots in 6-cm culture dish. After incubation, medium was added to each dish without dissociating the cell drops. For differentiation, we used osteogenic (OB) or chondrogenic (CHON) media containing the same base: DMEM, 1% NEAA, 50 U/ml penicillin and 50 μg/ml streptomycin, and 0.1 mM β-mercaptoethanol. The following factors were added to the base medium: (i) osteogenic media: 15% knockout serum replacement (KSR, Invitrogen), 50 (βg/ml ascorbic acid (Sigma), 10 mM β-glycerophosphate (β-GP), 50 nM 1,25-OH2 vitamin D3 (VD3; Calbiochem/EMD Chemicals, Gibbstown, NJ, USA); (ii) chondrogenic media: 1% ITS (Invitrogen; ITS solution consists of 0.00067 g/L sodium selenite, 1.9 g/L insulin, 0.55 g/L transferrin), 1% FBS (Invitrogen), 10 ng/ml transforming growth factor-b1 (TGF-β1; PeproTech, Rocky Hill, NJ, USA), 10 ng/ml bone morphogenetic protein 2 (BMP-2; PeproTech), and 50 mg/ml ascorbic acid (Sigma). Medium was changed every 2 days. During the micromass differentiation, the ES cells formed floating aggregates. On day 5 of differentiation, the floating osteo- and chondrogenic aggregates were transferred to static nonadherent culture flasks containing either osteogenic or chondrogenic differentiation medium. Medium was changed every 3-5 days. Following 15 days differentiation (osteogenic) and 20 days differentiation (chondrogenic), aggregates were separately collected at random for implantation.
Differentiation of ES Cells on Collagen I Matrix
A single cell suspension of ES cells was prepared in an 80% v/v 3 mg/ml bovine collagen I solution (Advanced Biomatrix, San Diego, CA, USA) at a concentration of 106 cells/ml (11). The collagen cell solution was then polymerized as a three-dimensional gel using 20% v/v of 10 mM β-GP in 5× concentrated DMEM. The 5× concentrated DMEM was generated by dissolving powered DMEM (Invitrogen) in 1/5 the recommended amount of water and was supplemented with 15% FBS (Invitrogen), 1% nonessential amino acids, 50 U/ml penicillin and 50 μg/ml streptomycin, 0.1 mM β-mercaptoethanol. The cell gel constructs (MTXOB) were then incubated at 37°C for 15 days.
Burr-Hole Fracture Model and ES Cell Implantation
All surgical and animal care procedures were approved by the University of Calgary Animal Care Committee. Immunocompetent, 8-week-old mice (Jackson Laboratory, Bar Harbor, ME, USA) strain-matched to the ES cells (D3/Sv129) were used as hosts for the burr-hole surgery. Male cells were implanted into female recipients to identify donor cells in host tissues (by Y chromosome in situ hybridization histochemistry, see below). The burr-hole defects were created in the proximal tibiae of the mice as described by Taiani et al. (19), as a modification of a previously published method (23). Briefly, mice were anesthetized with isoflurane, a small incision was made on the medial side of the left proximal tibia and a 0.7-mm hole was drilled though the medial cortex and though the medullary cavity of the metaphysis using a high-speed microdrill (Fine Science Tools, Vancouver, BC, Canada). Fractures were created unilaterally for each animal. The animals were divided into six groups, each consisting of nine mice: (i) noninjured (NI), (ii) control nontreated fractures (CTL); fractures treated with medium-induced (iii) osteoblasts (OB) or (iv) chondrocytes (CHON); (v) collagen matrix-induced osteoblasts (MTXOB) or (vi) acellular collagen matrix (MTX) (Table 1). ES cells differentiated using medium induction were implanted by lifting 8-10 cell aggregates (each approximately 300 μm in diameter) from a culture dish and transferring them to the fracture site using the tip of a scalpel blade. The aggregates were then gently inserted into the hole using a 27-gauge needle, immediately after drilling. ES cells differentiated in collagen gels were removed from their 1-ml well plates and approximately one fifth of the cell gel composite was placed over the fracture using a scalpel blade. A 27-gauge needle was used to gently push the gel into the hole. Each 1-ml cell gel composite had a cell density of 1.0 × 106 cells/ml, and therefore, approximately 2.0 × 105 cells were transplanted into the fracture. The same technique was used to transplant collagen gel without cells (MTX). The skin wound was closed using cynoacrylate tissue adhesive (Vetbond 3M, London, Ontario, Canada). The mice were returned to their cages and were allowed to move freely immediately after surgery. Three mice were euthanized by cervical dislocation from each group at weeks 2, 4, and 8 postimplantation. The tibiae were excised without removal of the soft tissue and were fixed in 10% neutral-buffered formalin (NBF) for a minimum of 5 days.
Experimental Groups
NI, noninjured; CTL, control; OB, media-induced osteoblasts; CHON, media-induced chondrocytes; MTXOB. Matrix-induced OBs; MTX acellular collagen matrix; NA, not applicable.
Micro-Computed Tomography (microCT) Imaging and Analysis
MicroCT scanning was conducted on the tibiae after fixation. The proximal third of the fixed tibiae were scanned at an isotropic resolution of 10 μm (μCT 40, Scanco Medical AG, Basserdorf, Switzerland) using a tube voltage of 55 kV, an integration time of 500 ms, and a tube current of 145 μA. A standardized region of interest (ROI) was selected from each image stack and consisted of 120 slices (1.2 mm) centered at the fracture (hole), at a mean distance of 3 mm from the proximal end of the tibia. For comparison, an ROI of 120 slices starting from 3 mm below the proximal tibia was selected in the noninjured group. Periosteal surface, cortical, and trabecular bone regions were identified using an automated segmentation technique (6) and Gaussian filtering (sigma = 1.2, support=5) followed by global thresholding (threshold=27% of maximum intensity) to obtain binary images of the mineralized phase of the tibia in the ROI. The mean gray scale value of all voxels within the periosteal surface was converted to mg hydroxyapatite (HA)/cm3 and was reported as apparent density. Based on the binary images, morphological parameters including trabecular bone volume ratio (BV/TV), trabecular thickness (Tb.Th), trabecular separation (Tb.Sp), trabecular number (Tb.N), and cortical thickness (Ct.Th) were calculated.
Histology and In Situ Hybridization Histochemistry
Fixed tibiae were decalcified using CalEx (Fischer Scientific, Ottawa, Ontario, Canada) for 5 days, embedded in paraffin, sectioned, and stained with hematoxylin and eosin (H&E). Alternatively, the tibiae used for in situ hybridization were decalcified in 10% aqueous EDTA pH 7.4 for 3 weeks using a method previously described (2), with daily changes for the first week and weekly thereafter. The decalcified samples were then embedded in paraffin. Male cells implanted in the female recipients were identified using Y chromosome in situ hybridization histochemistry. Briefly, the paraffin-embedded specimens were sectioned at a thickness of 5 μm. The sections were deparaffinized with three washes in xylene (5 min each), dehydrated through graded alcohols (3 min each), and then washed twice with phosphate-buffered saline (PBS). The sections were incubated in a preheated 2× SSC (saline-sodium citrate buffer) at 70°C for 30 min and then allowed to cool for 5 min, after which they were rinsed twice in double distilled water (ddH2O) for 3 min and then digested in 0.4% pepsin in 0.2 M HCl at 37°C for 10 min. After rinsing twice with ddH2O, sections were then denatured for 5 min at 60°C in 70% formamide/30% 2× SSC then dehydrated through a graded series of ethanol solutions. The samples were left to air dry at room temperature for 20 min. Labeling of the sections was performed according to the manufacturer's protocol. Briefly, a biotinylated mouse Y chromosome probe (1187-YMB-01; Cambio, Cambridge, UK) was denatured for 10 min at 65°C and applied to each section. A glass coverslip was used to seal the sections, and they were incubated overnight at 37°C. The next day, coverslips were removed in 2× SSC at room temperature. Posthybridization washes were performed by rinsing twice with 50% formamide/50% 2× SSC – 1× SSC – 0.4× SSC/0.3% NP-40 buffer for 5 min at 37°C. Signal detection was achieved using streptavidin-labeled peroxidase (CISH Centromeric Detection Kit, Invitrogen, Burlington, Ontario, Canada) according to the manufacturer's protocol. Nuclei were lightly counterstained with hematoxylin. Labeled cells were visualized as dark brownish dots with the light purple nuclei using a Zeiss Orthoplan microscope (Toronto, Ontario, Canada) equipped with differential interference contrast (DIC) optics and a 63× oil immersion objective (NA 1.4). The sensitivity and specificity of the in situ hybridization protocol for Y chromosome detection was verified using three controls: (i) female mouse bone sections with no transplanted cells (to test for nonspecific binding of the Y chromosome probe), (ii) male mouse bone sections with exclusion of the Y chromosome probe [to test for nonspecific binding of the secondary antibody and the horse radish peroxidase (HRP) detection system], and (iii) male mouse bone sections (a positive control for Y chromosome hybridization in bone).
Statistical Analysis
The microCT data (cortical bone thickness, bone volume ratio, trabecular number, trabecular thickness, trabecular separation, apparent density) of the various treatment groups were compared using an ANOVA with a Scheffé post hoc test at each time point. The tests were run with the STATA (v.9.0, College Station, TX, USA) software package.
Results
Burr-Hole Fractures
Burr-hole fractures were successfully created in the proximal tibia of each mouse without surgical complications. MicroCT and histopathology revealed that all burr holes were located distal to the growth plate and had disrupted one side of the cortical bone as well as the trabecular bone in the medullary cavity (Fig. 1A–D). The reproducibility of the burr-hole creation was evaluated by microCT immediately after surgery (n = 10 mice; not used for cell implantations). The mean burr-hole diameter was 0.71 ± 0.04 mm, the mean distance from the center of the hole to the top of the epiphysis was 2.49 ± 0.26 mm, and the mean depth of the hole was 1.07 ± 0.21 mm (Fig. 1A, B). The tibiae were scanned at 2, 4, and 8 weeks using microCT, and the scans of each tibia were qualitatively examined. As the location of the fracture relative to the growth plate varied somewhat among mice, histomorphometry was done on a region of interest measuring 1.2 mm (120×10 μm sections) in height centered about the burr hole. The selected region encompassed the majority of bony changes that occurred to the cortical and trabecular bone in all of the tibiae over the 8-week period (Fig. 1C). Hence, this volume was used to qualitatively and quantitatively compare the bone quality and histomorphometric changes among the treatment groups (Figs. 2–4).
Burr-hole defect in the proximal mouse tibia. (A) Micro-computed tomography (MicroCT) image rendered to show the medial aspect of the tibia (oriented longitudinally) just after surgery. The red line drawn on the tibia measures the distance from the proximal end of the tibia to the centroid of the burr hole. (B) MicroCT image of a section through the long axis of the burr hole looking proximally. The red line drawn measures the depth of the burr hole. (C) MicroCT image of a 1.2-mm-thick section showing the region of interest (ROI) centered around the burr hole. The cortical bone is semitransparent, and the trabecular bone is shown in white. (D) A histological section along the frontal plane of the fractured tibia immediately after surgery, stained with H&E. Scale bars: 1 mm.
Panel of representative microCT images (one mouse from each treatment group) at 2, 4, and 8 weeks after implantation. (A-C) Control group (CTL) shows that trabecular bone forms rapidly to span the cortical defect and inside the medullary cavity, which is subsequently remodeled to achieve nearly normal cortical and trabecular bone morphology by week 8. (D-F) Burr-hole fractures implanted with collagen matrix alone (MTX) and (G-I) burr-hole fractures implanted with collagen-induced embryonic stem (ES) cell-derived osteoblasts (MTXOB). Trabecular bone bridges the fracture site by week 2; exuberant trabecular bone is still present by week 8. No appreciable difference was observed between the MTX and MTXOB groups for any time points. (J-L) Burr-hole fractures implanted with osteogenic media-differentiated ES cells (OB). Note the cortical bone defect present at 8 weeks (arrow). (M-O) Burr-hole fractures implanted with chondrogenic media-differentiated ES cells (CHON). Note the cortical bone defect present by week 4 (arrow) and the large deformation of the bone at week 8 (arrows). Scale bars: 1 mm.
In the nontreated fracture control group (CTL), trabecular bone bridged the cortical bone defect and filled the medullary cavity at the injury site by 2 weeks. By 8 weeks, the trabecular bone volume in the medullary cavity was similar to that observed in the noninjured (NI) bones, and the cortical defect site was bridged with compact bone (Figs. 2 and 3A-C). Quantitative microCT data revealed that the trabecular number and separation in the CTL fracture group had returned to normal (no surgery, NI) values by week 8 after surgery (Fig. 4).
Histopathology of the burr-hole region. (A–C) CTL group at 2, 4, and 8 weeks postsurgery. Scale bars: 200 μm. (D–F) MTX group and (G–I) MTXOB group at 2, 4, and 8 weeks postimplantation. Scale bars: 200 μm. (J) OB group at 8 weeks. Note the soft tissue mass surrounded by bone (compare with Fig. 2L). Scale bar: 100 μm. (K) CHON group at 8 weeks postimplantation. Note the presence of striated muscle and cystic pockets within the cortex (inset shows magnified view of region within dashed box, compared with Fig. 2O). Scale bar: 100 μm, inset: 25 μm. The sections were stained with H&E.
Bone histomorphometry from microCT images of the fracture site at 2, 4, and 8 weeks postimplantation for the NI, CTL, MTX, and MTXOB groups. *p < 0.05, **p < 0.01. HA, hydroxyapatite.
Orthotopic Implantation of ES Cell-Derived Cells In Vivo
By 8 weeks postimplantation, two of the three mice that had received medium-induced osteoblasts (OB) had abnormal bone architecture. Specifically, these tibiae had cortical defects at the site of the burr hole (Fig. 2L). Histopathology of these regions confirmed the presence of an unorganized soft tissue mass (Fig. 3J). The third OB mouse showed a bone architecture that was characteristic of normal burr-hole fracture repair. By 4 weeks postimplantation, two of the three mice that had received medium-induced chondrogenic cells (CHON) also had abnormal healing of the cortex at the burr-hole site (Fig. 2N). Specifically, a large cavity was found in the medial cortex. By 8 weeks postimplantation, one of three CHON mice had severe disruption of the bone tissue architecture throughout the proximal tibia (Fig. 2O). The presence of a soft tissue mass containing cysts surrounded by striated muscle tissue was observed at the fracture site using histopathology (Fig. 3K). Furthermore, by 8 weeks postimplantation in the CHON-treated bone, two to three layers of lamellar bone had formed along borders where the host bone met with the tumor, indicative of reactive bone formation. Normal bone healing was observed in the other CHON-treated mice. There was no evidence of acute or chronic inflammation indicative of infection in any mice. Careful inspection by microCT and histopathology throughout the experimental period revealed no evidence of soft tissue mass or cyst formation in or around the burr-hole defect sites of the mice receiving either collagen matrix alone (MTX) or collagen matrix loaded with ES cells (MTXOB) (Figs. 2 and 3D–I). Through histological observation, bone formation in the subperiosteal region appeared to be enhanced in the MTXOB group compared to the MTX group by 4 weeks (Fig. 3E, H). No other appreciable differences were observed in the fracture repair process between the MTX-and MTXOB-treated groups, through either histological analysis or microCT imaging.
Bone Histomorphometry
The dramatic changes in the bone architecture in one or two of the three mice in the OB and CHON groups contributed to high variability in the histomorphometry data; hence, these groups were not included in the statistical analysis.
By 2 weeks after implantation, the MTX and MTXOB groups had less trabecular bone bridging the defect and in the medullary cavity than the CTL group (Fig. 4B). The trabecular number in the MTX and MTXOB groups was approximately 25% less than the CTL group at this time point (Fig. 4C). By 4 weeks after implantation, the trabecular bone volume ratio and trabecular thickness, number, and separation were similar among the MTX, MTXOB, and CTL groups; however, the MTX and MTXOB groups had significantly lower cortical bone thickness than the CTL group (p < 0.05) (Fig. 4A). By 8 weeks after implantation, trabecular number and thickness, trabecular bone volume ratio, and cortical bone volume were elevated, although not statistically different, in the MTX and MTXOB groups compared to the CTL group, while trabecular separation was reduced. These histomorphometry data appear consistent with the morphological quality of the bones in microCT images at 8 weeks (Fig. 2). Apparent density increased to a normal (noninjured) level by week 2 in the CTL group; however, these values were significantly less in the MTX (p < 0.05) and MTXOB (p < 0.01) at week 2 (Fig. 4F). The apparent density in the MTX and MTXOB groups increased steadily to reach normal (noninjured) levels by week 8. There were no significant differences in any of the bone parameters examined between the MTX group and the MTXOB group.
In Situ Hybridization for Y Chromosome
Cells bearing a Y chromosome were identified as any cell with brown–black spots located inside the nucleus. In the MTXOB group, Y chromosome-bearing cells were identified in the cortical and trabecular bone near the site of injury (Fig. 5A). In the OB group, no cells bearing a Y chromosome were identified in or around the soft tissue mass that was embedded in the cortical bone or in the surrounding bone tissue (see corresponding H&E stain in Fig. 3J) (Fig. 5B). In the CHON group, the only positive signal was observed along the periphery of the cystic pockets in the soft tissue mass (see corresponding H&E stain in Fig. 3K) (Fig. 5C); however, due to the dispersed appearance of the stain, it did not appear this was indicative of positive cells. Positive and negative controls verified that our methods were both sensitive and specific (Fig. 5D–F).
In situ hybridization histochemistry for Y chromosome horse radish peroxidase/3,3′-diaminobenzidine (HRP/DAB) detection counterstained lightly with hematoxylin show up as small brown–black-positive signals (dots) in the nuclei. (A) Male D3 cells of MTXOB group at 8 weeks postimplantation. Scale bars: 50 μm. Inset shows region within dashed box. Several Y chromosome bearing cells (brown–black dots in nuclei) in the lamellar bone at the burr-hole fracture site in the female host are visible. Scale bars: 25 μm. (B) OB and (C) CHON groups, 8 weeks after implantation (see corresponding H&E stains, Fig. 3J, K). No Y chromosome-positive cells were observed in the OB group. Dispersed DAB staining was observed along the periphery of the cystic pockets in the CHON group. Scale bars: 50 μm. Histological controls: (D) female bone with probe and detection system, (E) male bone without probe and with detection system, (F) male bone with probe and with detection system. Scale bars: 50 μm.
Discussion
The differentiation protocols used in the current study had been previously developed by our group and were reported to generate cells that expressed osteogenic and chondrogenic gene and protein markers and that formed skeletal tissues following subcutaneous implantation into a severe combined immunodeficient (SCID) mouse (11,25,26). No tumors or teratomas were reported to have formed following subcutaneous implantation in any of these studies. In the current study, we sought to examine the fate and tumorigenicity of these ES cell-derived skeletal cells following implantation into an appropriate orthotopic site in vivo.
Although stem cells differentiated with medium supplementation alone were associated with the formation of soft tissue masses and cysts in some recipients, ES cells differentiated into osteoblastic cells in a three-dimensional type 1 collagen gel were able to incorporate into the newly formed bone at the fracture site without any detectable tissue pathology. It is important to note that the ES-derived cells were not purified prior to implantation. These data further support the feasibility of using exogenous stem cells for skeletal repair and regeneration (e.g., 9,15), though they also suggest a careful reexamination of the strategy of using medium supplementation alone to drive the in vitro differentiation of ES cells.
It was previously reported that medium supplementation induces chondrogenic and osteogenic differentiation of the stem cells in micromass culture in vitro, as determined by various molecular and extracellular matrix markers (26). Moreover, ES cells differentiated by medium supplementation reportedly form small tissue masses containing nodules of bone and cartilage and not tumors or teratomas, following subcutaneous implantation in immunocompromised mice (25,26). Hence, in the current study, it was surprising to see the formation of tumors and cysts in some recipients following implantation of the medium-induced OB and CHON cells in the burr-hole defects of immunocompetent mice. We speculate that those animals not forming tumors or cysts after receiving OB- and CHON-treated cells likely had fewer (or no) tumor-forming or tumor-promoting cells implanted or that tumors were undetected at 8 weeks (i.e., tumors may have been detected at a later time point). It is noteworthy that no Y chromosome-positive (implanted) cells were observed in the soft tissue masses that formed at the burr-hole site. The only positive signal that was observed was located along the edges of the cystic pockets in the soft tissue masses arising after implantation of the CHON cells. However, the dispersed appearance of the stain could not be specifically ascribed to any intact cells. It seems possible that the residual staining might represent the remnants of male cells that underwent cell death. The absence of any intact cells bearing a Y chromosome suggests either that a small number of implanted cells went undetected using our screening method, or that the implanted cells migrated away from the fracture site, or that the implanted cells had short longevity. Nevertheless, lacking any cells bearing a Y chromosome, we conclude that the tissue masses generated at the implant site were host-derived and speculate that paracrine signaling between donor (implanted) cells may have overstimulated the proliferation of the host cells near the fracture site. Preliminary studies in our laboratory reveal that the majority of differentiated CHON and OB aggregates (in vitro) stain positively for the oncogene embryonic stem cell expressed Ras (ERas) (data not shown), which is expressed in pluripotent ES cells and is associated with the proliferation and tumorigenicity of these cells (20). Hence, we speculate that the medium supplementation used in the OB and CHON cultures was insufficient to drive the complete differentiation of ES cells and that these ERas-positive cells may have had tumor-promoting effects, through paracrine signaling, on the host cells following their implantation in vivo. This supposition receives support from several studies (e.g., 4,17,18) suggesting that paracrine signaling from ES cells promotes tumorigensis by promoting tissue overgrowth and cell survival. Clearly, the influence of implanted stem cells on host tissues needs to be better elucidated in order to avoid tumor formation and assure the safety of the tissue constructs in vivo. It is possible that purification of the medium-induced differentiated cultures, possibly using ERas expression, could be used to reduce the risk of tumor formation in the recipient, although we believe that cell sorting is impractical in a clinical setting and that research efforts are better spent on the development of efficient differentiation techniques.
In a previous report, exposure of undifferentiated ES cells to a collagen I gel for 15 days led to the osteogenic differentiation of the cells, as determined by the presence of lineage-specific protein and gene markers (11). Moreover, when these cell gel constructs were implanted subcutaneously in immunocompromised mice, they reportedly formed small masses of mineralized tissue. The results from the present study suggest that collagen gel differentiation reduces the risk of tumor formation in vivo compared to the use of medium supplements alone and that the donor cells can incorporate into the new bone tissue in the host skeleton at the injury site. These data are consistent with previous rat studies in which undifferentiated ES cells loaded on type 1 collagen gels participated in healing an osteochondral defect (14) but formed teratomas when injected into a knee joint cavity (24). Since no appreciable difference was observed between the MTX and MTXOB groups through histochemical and microCT analyses, the effect of the ES cell-derived osteoblasts on fracture repair could not be discerned. A similar observation was noted in a separate study by Undale et al. (22). This group examined bone augmentation strategies in a rat nonunion fracture model and found no appreciable difference between fractures treated with human ES cell-derived osteoblasts loaded on an atelocollagen matrix and those treated with the matrix alone. They did however note a significant enhancement of bone formation in fractures treated with osteoblasts generated from bone marrow-derived mesenchymal stem cells (MSCs), loaded on the atelocollagen matrix. It would be interesting to examine whether mouse MSCs would similarly enhance bone formation in the burr-hole fractures. Future studies using larger animal models with critical size defects and larger sample sizes will be required to effectively ascertain the ability of the ES cell-loaded collagen constructs to restore the biological and mechanical integrity of the bone and to determine if the cell-loaded matrices perform more effectively than the collagen matrix alone.
We suspect that, compared to medium supplementation alone, the biochemical environment of the type 1 collagen gel more completely drives the osteoblastic differentiation of ES cells which facilitates their incorporation into the host bone as well as limits the number or activity of undifferentiated cells that might otherwise go on to induce tumor formation. Extracellular matrix (ECM) plays a critical role in the regulation of cell function and differentiation [see reviews (1,3)]. There are at least three mechanisms through which ECM controls cell behavior: (i) composition and structural arrangement of the matrix, (ii) synergistic interactions between growth factors and matrix molecules, and (iii) activation of signaling pathways through cell surface adhesion molecules (1). In our study, each of these mechanisms may have contributed to the enhancement of osteoblastic differentiation of the ES cells. For example, the microarchitecture of the collagen I matrix likely influenced cell shape, and this has previously been found to affect lineage-specific differentiation of stem cells (13). Furthermore, Salasznyk et al. showed that MSCs could be differentiated into osteoblasts using collagen I and vitronectin ECM and that integrin signaling (through cell-matrix attachment) was at least partially responsible for this process (16). In particular, this group discovered that the binding of cell surface integrins to collagen I and vitronectin ECM caused activation of the extracellular signal-regulated kinase (ERK) signaling pathway, which subsequently resulted in the upregulation of Runx2/Cbfa1—a transcription factor critical to osteoblastic differentiation and bone formation. In our study, integrin signaling likely played a critical role in the enhanced osteogenic differentiation of ES cells when collagen I matrix was used for induction instead of medium supplementation. Although these studies provide some insight, it is clear that the mechanisms through which ECM regulate stem cell differentiation are complex and further investigation will be required to identify these processes in ES cells.
The clinical use of ES cells will rely on the development of robust differentiation protocols that not only promote the generation of desired cell types but also eliminate tumorigenicity of the cells. This study clearly demonstrates that not all differentiation protocols will induce ES cell differentiation and loss of tumorigenicity to the same degree, and more importantly, that the use of subcutaneous implantation is not an adequate test of in vivo behavior and tumorigenicity. Future studies should include more robust in vivo assessment methods, such as the use of an orthotopic implantation system.
Footnotes
Acknowledgments
This study was supported by the Alberta Heritage Foundation for Medical Research, the Natural Sciences and Engineering Research Council of Canada, the National Institutes of Health (R21 AR053738-01), and the Canadian Stem Cell Network. The authors declare no conflicts of interest.