Generation of human induced pluripotent stem cells from osteoarthritis patient–derived synovial cells

Authors


Abstract

Objective

This study was undertaken to generate and characterize human induced pluripotent stem cells (PSCs) from patients with osteoarthritis (OA) and to examine whether these cells can be developed into disease-relevant cell types for use in disease modeling and drug discovery.

Methods

Human synovial cells isolated from two 71-year-old women with advanced OA were characterized and reprogrammed into induced PSCs by ectopic expression of 4 transcription factors (Oct-4, SOX2, Klf4, and c-Myc). The pluripotency status of each induced PSC line was validated by comparison with human embryonic stem cells (ESCs).

Results

We found that OA patient–derived human synovial cells had human mesenchymal stem cell (MSC)–like characteristics, as indicated by the expression of specific markers, including CD14−, CD19−, CD34−, CD45−, CD44+, CD51+, CD90+, CD105+, and CD147+. Microarray analysis of human MSCs and human synovial cells further determined their unique and overlapping gene expression patterns. The pluripotency of established human induced PSCs was confirmed by their human ESC–like morphology, expression of pluripotency markers, gene expression profiles, epigenetic status, normal karyotype, and in vitro and in vivo differentiation potential. The potential of human induced PSCs to differentiate into distinct mesenchymal cell lineages, such as osteoblasts, adipocytes, and chondrocytes, was further confirmed by positive expression of markers for respective cell types and positive staining with alizarin red S (osteoblasts), oil red O (adipocytes), or Alcian blue (chondrocytes). Functional chondrocyte differentiation of induced PSCs in pellet culture and 3-dimensional polycaprolactone scaffold culture was assessed by chondrocyte self-assembly and histology.

Conclusion

Our findings indicate that patient-derived synovial cells are an attractive source of MSCs as well as induced PSCs and have the potential to advance cartilage tissue engineering and cell-based models of cartilage defects.

Osteoarthritis (OA), also known as degenerative arthritis, is a chronic and progressive disorder characterized by the breakdown of joint cartilage, which causes severe pain and stiffness in the joints. OA can be caused by aging, heredity, and injury from trauma or other disease (1), but details of the biologic etiopathogenesis of OA in humans have remained elusive. There is no proven disease-modifying treatment for OA. Current treatments focus mainly on controlling pain and improving joint function (2). Irreversible joint damage in advanced OA usually requires surgical management. Numerous surgical procedures for repairing articular cartilage defects have been developed, but these procedures are still considered challenging (3).

Animal models of OA have been used extensively for understanding disease progression and testing potential antiarthritis drugs for clinical use or evaluating the disease-modifying effects of agents currently used to treat patients (4, 5). The relevance of animal models to human disease is not based on a proven track record of predictability of drug-induced changes in disease progression, but rather on the clinical and histologic similarities to human disease. Clinical efficacy data in humans are still largely lacking due to the difficulty of assessing and monitoring disease progression and the long duration of clinical trials.

Recent research on cartilage and disc tissue engineering has focused on grafting heterologous or autologous cartilage or on the transplantation of chondrocytes (6–8). Much effort has been focused on engineering cartilage with mesenchymal stem cells (MSCs) recovered from various adult tissue types as a promising alternative to chondrocytes (9–11). The heterogeneity of MSC populations isolated from different tissue types or exposed to different environmental factors, such as inflammatory conditions, can influence MSC properties and generate discrepancies in the differentiation and expansion capabilities of undifferentiated MSCs. However, the underlying mechanisms of action and possible roles of the interaction between MSCs and other specialized cells remain unknown. Therefore, there are still many questions about the most appropriate tissue source for MSCs.

Recent advances in cellular reprogramming technology have provided entirely new approaches to the development of human disease models and therapeutic strategies. Induced pluripotent stem cells (PSCs) derived from patients' somatic cells and differentiated cells have made it possible to develop patient-specific disease models that can be tested for the initiation and progression of disease, and are a human therapeutic cell population that can be used in cell-based medical products.

Herein, we show that human MSC-like synovial cells from patients with OA can be efficiently reprogrammed into a pluripotent state and demonstrate that these OA patient–derived human induced PSCs can develop into specialized cell types, allowing them to be used for drug discovery and regenerative medicine.

MATERIALS AND METHODS

Culture of human synovial cells and human MSCs.

The study was approved by the local ethics committee, and informed consent was obtained from all patients. Human synovial tissue was obtained aseptically from 2 patients with OA (two 71-year-old women) who were undergoing total hip arthroplasty. Tissue was digested using 0.05% collagenase (Invitrogen) in α-minimum essential medium (α-MEM) supplemented with 100 μg/ml penicillin, 100 μg/ml streptomycin, and 0.25 μg/ml amphotericin B for 2–3 hours and centrifuged at 1,500 rpm (380g) for 5 minutes. The washed pellet was resuspended in synovial cell culture medium (α-MEM containing 10% fetal bovine serum [Invitrogen]) in a 100-mm culture dish and allowed to attach for 4 days. Nonadherent cells were removed by changing the media, and the cell layer was washed 2 or 3 times with Hanks' balanced salt solution. Established human synovial cells were maintained in synovial cell medium or in MSC growth medium (catalog no. PT-3001; Lonza/Cambrex). Human MSCs (catalog no. PT-2501; Lonza/Cambrex) were maintained in MSC growth medium at 37°C in an atmosphere of 5% CO2. Differentiation of human synovial cells and human MSCs into mesenchymal lineage cells was performed according to the recommendations of the manufacturer (Lonza/Cambrex).

Retrovirus production, infection, and human induced PSC generation.

For retrovirus production, GP2-293 cells were transfected with pMXs-Oct-4, SOX2, Klf4, and c-Myc (Addgene) by Lipofectamine according to the recommendations of the manufacturer (Invitrogen). Forty-eight hours and 72 hours after transfection, the supernatants of the transfectant were collected and concentrated in an Ultracentrifuge (Beckman). For induced PSC generation, passage-4 human synovial cells (1.5 × 105 cells per well) were transduced with 1–5 multiplicities of infection of retroviruses encoding human Oct-4, SOX2, Klf4, and c-Myc at a 1:1:1:1 ratio in 6-well culture dishes (defined as day 0). This transduction procedure was repeated a total of 2 more times, once on day 2 and once on day 4, and cells were maintained in synovial cell medium. On day 6, the cells were passaged onto γ-irradiated mouse embryonic fibroblasts and cultured in human embryonic stem cell (ESC) culture medium (80% Dulbecco's modified Eagle's medium [DMEM]–F-12, 20% knockout serum replacement [Invitrogen], 1% nonessential amino acids [Invitrogen], 1 mM L-glutamine [Invitrogen], 100 units penicillin, 100 μg/ml streptomycin [Invitrogen], 0.1 mM β-mercaptoethanol [Sigma], and 4–6 ng/ml basic fibroblast growth factor [R&D Systems]). The medium was changed every other day. Human ESC–like colonies were manually picked between days 23 and 25 and transferred onto 12- or 6-well culture dishes preplated with γ-irradiated mouse embryonic fibroblasts. Isolated human induced PSC colonies were subsequently maintained and expanded under standard human ESC culture conditions.

Culture of human ESCs and human induced PSCs.

Undifferentiated human ESCs (H9; WiCell) and established human induced PSCs were maintained with human ESC medium and passaged once a week using mechanical passaging, as described previously (12). Autologous feeder cells (H9 embryonic body–derived fibroblasts [ebF]) were differentiated from H9 human ESCs through embryoid body outgrowth using a previously published procedure (13, 14). After mitotic inactivation of H9 ebF, passages 3–10 were used for maintenance of undifferentiated human ESCs and human induced PSCs.

Polymerase chain reaction (PCR) analysis of genomic integration.

Genomic DNA samples were isolated using the DNeasy kit (Qiagen). Each PCR amplification reaction was performed with 300 ng of genomic DNA extracted from human induced PSCs and human synovial cells. The primers used to amplify the transgene are presented in Supplementary Table 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

Bisulfite pyrosequencing.

One microgram of genomic DNA (per sample) isolated from human synovial cells, human ESCs, and human induced PSCs was used for bisulfite conversion and subsequent sequencing. Bisulfite conversion was performed using the EZ DNA Methylation kit according to the recommendations of the manufacturer (Zymo Research). Promoter regions of Oct-4 and Nanog were amplified by PCR and cloned into the pCR2.1-TOPO vector (Invitrogen). Eight to ten random clones were sequenced with the M13 forward and M13 reverse primers. Primer sequences used in the PCR amplification are shown in Supplementary Table 1.

Microarray analysis.

Total RNA from human ESCs, human synovial cells, human MSCs, and the established induced PSC lines was extracted using the RNeasy Mini Kit (Qiagen), labeled with Cy3, and hybridized to Agilent Human Whole Genome 4x44K Microarrays (one-color platform) according to the recommendations of the manufacturer (Agilent Technologies). The gene expression results were analyzed using GeneSpring microarray analysis software.

Karyotype analysis.

Expanded human induced PSCs cultured in human ESC culture medium for 20 passages were processed for chromosomal G-band analysis by GenDix. Images were captured by ChIPS-Karyo (Chromosome Image Processing System; GenDix).

In vitro differentiation of human ESCs and human induced PSCs.

For embryoid body formation, human ESCs and human induced PSCs were transferred onto nonadherent plates and maintained in suspension with embryoid body culture medium containing DMEM–F-12, 10% knockout serum replacement, 1% nonessential amino acids, 100 units penicillin, 100 μg/ml streptomycin, and 0.1 mM β-mercaptoethanol. After 5 days of growth in suspension, the cell aggregates were seeded onto Matrigel-coated dishes and cultured for 15 additional days. The medium was changed every 2 days. Detailed information on the conditions for mesenchymal differentiation of human ESCs and human induced PSCs into osteoblasts, chondrocytes, and adipocytes is available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

Teratoma formation.

For teratoma formation, one million human induced PSCs were harvested and injected subcutaneously into the dorsal flank of 6-week-old SPF/VAF-immunodeficient mice (Orientbio). Eight to twelve weeks after injection, visible tumors were dissected and fixed overnight with 4% paraformaldehyde/phosphate buffered saline solution. Paraffin-embedded tissue was sliced, stained with hematoxylin and eosin, and examined for the presence of tissue representative of all 3 germ layers. The antibodies used for immunofluorescence staining are listed in Supplementary Table 2, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

Semiquantitative reverse transcriptase PCR (RT-PCR), real-time quantitative PCR (qPCR), cytochemistry, and flow cytometric analysis.

A detailed list of the primers and PCR conditions and details regarding cytochemistry and flow cytometry analysis conditions is available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

RESULTS

Isolation and characterization of human synovial cells.

We isolated human synovial cells from 2 patients (71-year-old women) with advanced OA and determined whether the cells had MSC–like properties by examining the expression of MSC-specific markers and the differentiation potential of the cells. Commercially available human MSCs derived from bone marrow were used for comparison. Phase-contrast images revealed that human synovial cells were relatively uniformly spindle-like and had fibroblast-like shapes. (See Supplementary Figure 1A, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.) Under the culture conditions used in this study, 2 human synovial cell lines (hSC52 and hSC65) were expanded over at least 15 passages with a similar growth rate, and <0.1% of the cells stained for SA-β-gal activity as a marker of cellular senescence at passage 9 (Supplementary Figure 1). Semiquantitative RT-PCR and real-time qPCR showed that human synovial cells (passages 3–5) that were expanded in monolayer culture expressed the key MSC marker genes CD44, CD51, CD90, CD105, and CD147, and a fibroblast marker, vimentin, but did not express hematopoietic markers (CD14, CD19, CD34, and CD45) or the endothelial marker fetal liver kinase 1 (Figures 1A and B).

Figure 1.

Characterization of human synovial cells (hSC) from osteoarthritis patients. A and B, Semiquantitative reverse transcriptase–polymerase chain reaction (RT-PCR) (A) and real-time quantitative PCR (B) analyses of the expression of fibroblast markers (vimentin and CD90), hematopoietic markers (CD14, CD19, CD34, and CD45), an endothelial marker (fetal liver kinase 1 [FLK-1]), and human mesenchymal stem cell (hMSC) markers (CD44, CD51, CD90, CD105, and CD147) in human MSCs and in synovial cell lines hSC52 and hSC65. Real-time PCR results were normalized to the expression of GAPDH and displayed as ratios of the indicated marker gene versus GAPDH using the formula 2inline image (× 100). Bars show the mean ± SEM (n = 3 samples per group). C, Immunohistochemical staining for CD45, CD44, CD105, and STRO-1 in human MSCs and in the synovial cell lines hSC52 and hSC65. D, Fluorescence-activated cell sorter analysis for CD45, CD44, and CD90 in human MSCs and the synovial cell lines hSC52 and hSC65. Representative results obtained from passage-5 cells are shown.

Immunohistochemical analysis revealed that >90% of human synovial cells showed positive staining for essential MSC marker proteins, including CD44, CD105, and STRO-1 (Figure 1C), and a fibroblast marker but were negative for the pluripotency markers Oct-4 and Nanog as well as a macrophage marker. (See Supplementary Figure 2A, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.) Using fluorescence-activated cell sorting, we determined that >90% of human synovial cells were positive for the MSC makers CD44 and CD90, and the fibroblast marker, for both the hSC52 and hSC65 lines (Figure 1D and Supplementary Figure 2B). However, CD45 was barely detected in either hSC52 or hSC65. Similar results were obtained using commercially available human MSCs (Figures 1A–D). In addition, although heterogeneity in the induction of various lineage markers was detected, all isolated human synovial cell lines showed differentiation potential to all 3 mesenchymal lineages, including bone, cartilage, and fat (data not shown).

Transcription profiles of human synovial cells and human MSCs.

We next explored the complementary DNA expression profiles of human synovial cells and human MSCs using Agilent Human Whole Genome 4x44K microarrays to further elucidate the differences or similarities between human synovial cells and human MSCs. Although human synovial cells and human MSCs displayed similar characteristics, the transcription profiles showed that human synovial cells and human MSCs had highly distinct and uniform gene expression patterns. (See Supplementary Figures 3A–C, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.)

As confirmed by semiquantitative RT-PCR and real-time qPCR, the microarray analysis revealed that expression levels of human MSC–specific genes, including CD13, CD44, CD51, CD59, CD73, CD90, CD105, and CD147 were similar in human MSCs and human synovial cells, as shown in Supplementary Table 3, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131. In addition, the microarray identified 17,187 genes (39%) that showed ≥2-fold differences in expression level in the 2 human synovial cell lines (hSC52 and hSC65). The microarray analysis showed that 8,261 genes (18.8%) were up-regulated by >2-fold and 8,745 genes (19.9%) were down-regulated by >2-fold in both in hSC52 and hSC65 cells compared to human MSCs (Supplementary Figure 3B). The 20 genes with the greatest degree of up- or down-regulation in both hSC52 and hSC65 were ranked in order and are listed in Supplementary Table 4, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.

Generation and characterization of induced PSCs from human synovial cells.

The method of reprogramming human synovial cells is described in detail in Materials and Methods. The scheme for induced PSC induction is shown in Figure 2A. Human ESC–like colonies with typical human ESC–like morphology appeared following 2–3 weeks of culture. (See Supplementary Figure 4A, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.) Based on human ESC–like morphology, alkaline phosphatase staining, and TRA-1-60 immunostaining, the calculated reprogramming efficiency was ∼0.007–0.01%.

Figure 2.

Generation and characterization of human induced pluripotent stem cell (iPSC) lines. A, Scheme for human induced PSC induction using retroviral transduction (TD) of genes encoding the 4 transcription factors Oct-4 (O), SOX2 (S), Klf4 (K), and c-Myc (M). Cells were maintained in synovial cell (SC) medium until day 6, when they were transferred to human embryonic stem cell (hESC) medium. B, Morphologic analysis and immunohistochemical staining of established human induced PSC lines for pluripotency markers. Original magnification × 200. C and D, Semiquantitative reverse transcriptase–polymerase chain reaction (RT-PCR) (C) and real-time quantitative PCR (D) analyses of H9 human ESCs (hES) and established human induced PSC lines for the pluripotency markers Oct-4, SOX2, Klf4, c-Myc, human telomerase reverse transcriptase (hTERT), Rex1, Lin28, Nanog, and TDGF. Semiquantitative RT-PCR was performed for total, endogenous (Endo), and exogenous (Exo) transcription factors. Real-time PCR results were normalized to the expression of GAPDH. Bars show the mean ± SEM (n = 3 samples per group). Representative results obtained from human induced PSC lines (iPS-SC52-1, iPS-SC52-2, iPS-SC52-3, iPS-SC65-2, iPS-SC65-3, and iPS-SC65-4) at passage 5 are shown. ALP = alkaline phosphatase.

Selected human synovial cell–derived induced PSCs were all positive for alkaline phosphatase staining and uniformly expressed the typical human ESC–specific markers Oct-4, Nanog, TRA-1-60, and SSEA-3 as determined by immunocytochemistry (Figure 2B and Supplementary Figure 4B). Semiquantitative RT-PCR and real-time qPCR showed that levels of messenger RNA (mRNA) for key pluripotency marker genes, including Oct-4, SOX2, human telomerase reverse transcriptase (hTERT), Rex1, Lin28, Nanog, and TDGF were markedly increased in all human synovial cell–derived induced PSCs compared to the parental human synovial cells and were expressed at levels comparable to those in undifferentiated control H9 human ESCs (Figures 2C and D). Semiquantitative RT-PCR also confirmed that retroviral exogenous genes became largely silenced, whereas strong reactivation of endogenous reprogramming transcription factors was assumed (Figure 2C).

Genomic PCR revealed integration of all 4 transgenes (Oct-4, SOX2, Klf4, and c-Myc) in all human induced PSC lines tested (Figure 3A). Bisulfite genomic sequencing revealed that the Oct-4 and Nanog promoter regions were demethylated in human synovial cell–derived induced PSCs to a similar extent as in control human ESCs, relative to their hypermethylated state in human synovial cells (Figure 3B). Global gene expression profiling using Agilent Human Whole Genome 4x44K Microarrays demonstrated that 2 human synovial cell–derived induced PSC lines (iPS-SC52-1 and iPS-SC65-2) from different donors displayed a high degree of similarity to H9 human ESCs and a low degree of similarity to parental cells (Figure 3C). The results of short tandem repeat analysis (shown in Supplementary Table 5, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131) and karyotype analysis (Figure 3D) confirmed that OA patient–specific human synovial cell–derived induced PSCs were generated from the parental synovial cells and maintained a normal diploid karyotype.

Figure 3.

Additional characterization of human induced PSCs. A, Semiquantitative RT-PCR results, showing the genomic integration of the exogenous factors in selected human synovial cell–derived induced PSC colonies. Donor cells (hSC52 and hSC65) were included as negative controls. GAPDH was used as a control. B, DNA methylation (Me) profile of the Oct-4 and Nanog proximal promoters in selected induced PSC colonies. Donor cells were included as negative controls, and human ESCs were included as the positive control. Open circles indicate unmethylated CpG; solid circles indicate methylated CpG. C, Microarray data comparing global gene expression profiles of H9 human ESCs, human synovial cell–derived induced PSCs, and human synovial cells. Heat map and hierarchical clustering analysis by Pearson's correlation are shown at the top. Ratios are color coded, with green indicating minimum (min) and red indicating maximum (max) values. Scatterplots comparing global expression patterns between human ESCs, human synovial cell–derived induced PSCs, and human synovial cells are shown at the bottom. The positions of the pluripotency genes Oct-4, SOX2, and Nanog are indicated. D, Karyotype analysis of human synovial cell–derived induced PSCs. Representative results obtained from human induced PSC lines (iPS-SC52-1, iPS-SC52-2, iPS-SC52-3, iPS-SC65-2, iPS-SC65-3, and iPS-SC65-4) at passages 15–20 are shown. See Figure 2 for other definitions.

In vitro differentiation of human synovial cell–derived human induced PSCs.

To confirm the developmental potential of induced PSCs derived from human synovial cells, 4 individual human induced PSC lines from 2 patients were tested in both in vitro and in vivo differentiation assays through the formation of embryoid bodies and teratomas. Semiquantitative RT-PCR showed that human synovial cell–derived induced PSC expression of the pluripotency marker genes Oct-4 and Nanog was markedly down-regulated, while the expression of makers for the ectodermal lineage (neural cell adhesion molecule, Pax6, and Nestin), mesodermal lineage (Brachyury and GATA-2), and endodermal lineage (GATA-6 and α-fetoprotein) were up-regulated upon embryoid body differentiation of human synovial cell–derived induced PSCs. The results were similar to those for H9 human ESCs cultured under the same conditions (Figures 4A and B). Consistent with the results of RT-PCR, immunohistochemical analysis for ectodermal (Tuj1, Nestin), mesodermal (desmin, α-smooth muscle actin), and endodermal (SOX17, FoxA2) markers confirmed successful in vitro differentiation of human synovial cell–derived induced PSCs at the protein level (Figure 4C).

Figure 4.

In vitro differentiation of human induced PSCs into all 3 germ layers. A, Undifferentiated and differentiated human synovial cell–derived induced PSCs. Undifferentiated cells can form embryonic bodies and differentiate into cells of ectodermal, endodermal, and mesodermal lineages. B and C, Semiquantitative RT-PCR (B) and immunohistochemical (C) analyses of the expression of ectoderm (neural cell adhesion molecule [NCAM], Pax6, Nestin, and Tuj1), mesoderm (Brachyury, GATA-2, desmin, and α-smooth muscle actin [α-SMA]), and endoderm (GATA-6, α-fetoprotein [AFP], SOX17, and FoxA2) lineage markers in the cell types indicated. Results are representative of at least 3 independent experiments. Representative results obtained from human induced PSC lines (iPS-SC52-1, iPS-SC52-2, iPS-SC65-2, and iPS-SC65-3) at passages 15–20 are shown. See Figure 2 for other definitions.

Mesenchymal differentiation of human synovial cell–derived induced PSCs.

We further determined the potential of human synovial cell–derived induced PSCs to differentiate into mesenchymal lineages, such as osteoblasts, chondrocytes, and adipocytes, after incubation with lineage-specific differentiation medium for 3–4 weeks, via embryoid body formation. After 3 weeks, osteogenic differentiation was confirmed by alizarin red staining for calcium deposition in the matrix (Figure 5A) and increased expression of mRNA for type I collagen, osteoprotegerin, and runt-related transcription factor 2, as determined by real-time qPCR (Figure 5B). Approximately 60–70% of all cells showed positive alizarin red staining. After 3 weeks of chondrogenic differentiation, >50% of all cells stained positive with Alcian blue, a specific stain for extracellular matrix proteoglycans (Figure 5A). Chondrogenic differentiation of human induced PSCs was further confirmed by the increased expression of mRNA for type IIA collagen, aggrecan, and type X collagen, as determined by real-time qPCR (Figure 5B). After 3 weeks of adipogenic differentiation, >70% of all cells stained positive with oil red O (Figure 5A). Real-time qPCR revealed up-regulation of the adipocyte-specific genes AFABP2/aP2 and peroxisome proliferator–activated receptor γ (Figure 5B). These lineage-specific markers were not detected in undifferentiated human ESCs or human synovial cell–derived induced PSCs and were significantly induced only after stimulation of cells into defined lineages by culture in differentiation medium.

Figure 5.

Directed differentiation of human induced PSCs into mesenchymal lineages. Embryonic bodies predifferentiated from induced PSCs were further differentiated into mesenchymal lineage cells for an additional 3 weeks as described in Materials and Methods. A, Staining of differentiated cells for osteoblasts (alizarin red S), chondrocytes (Alcian blue), and adipocytes (oil red O). B, Real-time quantitative PCR analysis of specific markers for osteoblasts (type I collagen, osteoprotegerin, and runt-related transcription factor 2 [RUNX-2]), chondrocytes (type IIA collagen, type X collagen, and aggrecan), and adipocytes (aP2 and peroxisome proliferator–activated receptor γ [PPARγ]). Real-time PCR results were normalized to the expression of GAPDH. Bars show the mean ± SEM (n = 3 samples per group). Results are representative of at least 3 independent experiments. C, Chondrogenic differentiation of human induced PSCs in pellet culture (i), agarose substratum culture (ii), and 3-dimensional (3-D) scaffold culture (iii). Cryosections of chondrocyte pellets were stained with antibodies against SOX9, type I collagen, type II collagen, and type X collagen, and Alcian blue–Safranin O after 3 weeks of chondrogenic differentiation (i and ii). Cells in scaffolds were stained with antibodies against SOX9, type I collagen, type II collagen, and type X collagen, and Alcian blue after 2 months of chondrogenic differentiation (iii). Representative results obtained from human induced PSC lines at passages 15–25 are shown. See Figure 2 for other definitions.

To further evaluate the differentiation potential of human induced PSCs into functional chondrocytes, we differentiated and maintained human induced PSCs in pellet culture (Figure 5C part i), agarose culture (Figure 5C part ii), and implantable scaffolds (Figure 5C part iii). After 3 weeks of chondrogenic differentiation, dense cartilage-like aggregates were formed both in suspension (Figure 5C part i) and in gels such as agarose (Figure 5C part ii) in vitro and stained positive with either Alcian blue or Safranin O (Figure 5C). Immunohistochemical analysis showed that the cartilaginous constructs positively expressed chondrogenic transcription factor SOX9 and cartilage-specific matrix molecules (type I, type II, and type X collagen) (Figure 5C). In addition, the cells from human induced PSCs that underwent chondrogenic differentiation were seeded at a density of 1 × 106 cells/polycaprolactone polymer scaffold (thickness 400 μm, pore size 800 μm) and allowed to grow for up to 2 months in vitro. Cells cultured in 3-dimensional scaffolds displayed the spherical morphology of chondrocytes, and produced intense and well-defined positive staining with antibodies to SOX9 and type I, type II, and type X collagen and Alcian blue stain (Figure 5C). These findings demonstrate that established human induced PSCs have a high potential for differentiating into functional chondrocytes, and that differentiation of PSCs using conventional differentiation protocols is feasible.

Teratoma formation of human synovial cell–derived human induced PSCs.

After transplantation into nude mice, human induced PSCs formed teratomas consisting of representative derivatives of all 3 germ layers, including neural tissue, neural rosette, and epidermis (ectoderm); gut-like epithelium (endoderm); and adipose tissue, smooth muscle, bone, cartilage, and myxoid tissue (mesoderm) (Figure 6A). (See Supplementary Figure 5A, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.) In addition, sections from human induced PSC–derived teratomas stained positive for antibodies recognizing SOX9, type I, type II, and type X collagen, and aggrecan and Alcian blue stain (Figure 6B and Supplementary Figure 5B).

Figure 6.

Histologic analysis of a teratoma derived from human induced pluripotent stem cells (PSCs). A, Hematoxylin and eosin–stained sections from human induced PSC–induced teratomas. Passage-15 iPS-SC65-2 cells were used. Differentiation into multiple derivatives of the 3 germ layers is shown. Original magnification × 200 in left panels. Images at the right are higher-magnification views of the boxed areas at the left. B, Immunohistochemical analysis of SOX9, type I collagen, type II collagen, type X collagen, and aggrecan and Alcian blue staining in cartilage formed within the teratomas.

Comparative transcriptome analysis of MSCs and induced PSCs.

Using genome-wide microarray analysis, we compared the gene expression profiles of human MSCs and human induced PSCs to better understand their similarities and differences. (See Supplementary Figure 6, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.) Global transcription profiles of human MSCs and human induced PSCs were highly distinct, as confirmed by hierarchical clustering (Supplementary Figure 6A) and scatterplots (Supplementary Figure 6B).

The microarray identified 19,771 genes (44.96%) that showed ≥2-fold differences in expression level in human induced PSCs and human MSCs. Analysis showed that 10,066 genes (22.88%) were up-regulated by >2-fold and 9,705 genes (22.06%) were down-regulated by >2-fold in human induced PSCs compared to human MSCs (Supplementary Figure 6C). Human ESC–specific genes (TDGF3, CLDN6, DPPA4, LEFTY1, SOX2, ZFP42, CKMT1B, POU5F1, and Nanog) were significantly up-regulated by ≥180 times in human induced PSCs compared to MSCs, while lineage-specific genes were down-regulated in human induced PSCs (Supplementary Figure 6D). The 50 genes with the greatest up- or down-regulation in human induced PSCs compared to human MSCs are listed in Supplementary Table 6, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131. The human MSCs displayed more or less undetectable expression of the genes important for pluripotency of human induced PSCs, while lineage-specific genes, especially mesoderm markers, were relatively up-regulated in human MSCs compared to induced PSCs. (See Supplementary Figure 7D, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.) These results suggest that in an undifferentiated state, there are significant differences in gene expression between induced PSCs and MSCs.

Maintenance of human synovial cell–derived human induced PSCs with human fibroblast feeder cells.

We further evaluated whether induced PSCs derived from patient synovial cells could be maintained on human feeders. Fibroblast-like cells were differentiated from the outgrowth of H9 human embryoid bodies (H9 ebF) based on the findings of previous studies (13, 14) and passaged periodically for >15 passages. After 2 or 3 passages, fibroblast-like cells displayed homogeneous populations that stained positive for a human fibroblast marker (Supplementary Figure 7A). The fibroblast-like characteristics of H9 ebF were further confirmed using semiquantitative RT-PCR, which revealed increased expression of the fibroblast-specific markers vimentin and prolyl 4-hydroxylase β as well as decreased expression of Oct-4 and Nanog, which are human ESC–specific markers (Supplementary Figure 7B). Undifferentiated human induced PSCs and H9 human ESCs were grown on H9 ebF and passaged every 5–7 days. As with mouse embryonic fibroblast feeder cells, both human induced PSCs and H9 human ESCs were successfully maintained on H9 ebF in an undifferentiated state for >15 passages, which was confirmed by morphology and by the elevated expression of the human ESC–specific markers alkaline phosphatase, Oct-4, Nanog, TRA-1-60, and SSEA-4, as determined by semiquantitative RT-PCR and immunohistochemistry (Supplementary Figure 7C and D), and by teratoma formation (Supplementary Figure 7E).

DISCUSSION

Tissue engineering with MSCs is one of the most promising approaches for the treatment of rheumatic diseases, including OA, rheumatoid arthritis (RA), and genetic bone and cartilage disorders, as well as bone metastasis, because of the immunosuppressive characteristics and trilineage differentiation potential of MSCs. A newly identified source of MSCs for cartilage regeneration, fibrous synovium-derived MSCs, which possess high chondrogenic potential, have been successfully isolated from the human knee joint by arthroscopy (15–17). Comparative studies showed that MSCs isolated from synovial tissue have superior chondrogenic potential compared to MSCs from other tissue sources, including bone marrow (18, 19). In this study, we successfully isolated adherent synovial cells from 2 OA patients undergoing hip arthroplasty in the clinic. Consistent with the results of previous studies (15–17, 19–21), we found that isolated synovial cells display a phenotype and differentiation capacity similar to that of bone marrow–derived MSCs (Figure 1) while having a different gene expression signature. (See Supplementary Figure 3, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.)

Obstacles to the clinical application of MSCs still exist. There have been conflicting results as to whether or not functionally normal MSCs can be isolated from patients with OA and RA. Dudics and colleagues (22) showed that MSCs from patients with OA and RA possess chondrogenic potential similar to that of MSCs from healthy donors. Similarly, Scharstuhl and colleagues (23) demonstrated that the chondrogenic potential of MSCs is independent of age or OA etiology. In contrast, Murphy and colleagues (24) showed that MSCs from patients with advanced OA displayed reduced proliferative and chondrogenic activity, while their osteogenic activity was unchanged. Some studies revealed that human MSCs from patients with OA showed rapid induction of the hypertrophic marker type X collagen (COL10A1) (25, 26), which is associated with endochondral ossification (27, 28). The expansion and differentiation potential of MSCs is considered to be linked to several factors, such as chronic inflammation and age, but the underlying mechanisms and possible roles of interaction between MSCs and other specialized cells remain undefined.

Against this backdrop, the generation and differentiation of human induced PSCs from various different cell types represents a new strategy for human disease modeling and drug discovery. Previous studies have provided evidence of the therapeutic efficacy of using human induced PSCs for tissue repair (29–33). Previous studies have shown that human induced PSCs can differentiate into a large number of multipotent MSCs, and that MSCs derived from human induced PSCs are easily expandable to higher passages without changes in multipotent differentiation potential and show no clear signs of replicative senescence, compared to bone marrow–derived MSCs (30, 34). No major differences between human induced PSC–derived MSCs and human ESC–derived MSCs were demonstrated with regard to differentiation and proliferation potential (30, 35).

In an attempt to reprogram primary human MSC-like synovial cells to a pluripotent state, in this study human synovial cells were transduced with a subset of core reprogramming factors (Oct-4, SOX2, Klf4, c-Myc, Nanog, Lin28, and TERT). Human synovial cells acquired pluripotency when cotransduced with 4 transcription factors (Oct-4, SOX2, Klf4, and c-Myc), and this pluripotency was confirmed by pluripotency marker expression, global gene expression profile, CpG methylation profile, and in vitro and in vivo differentiation potential (Figures 2–6). Previous studies have shown that the cell numbers and expansion and differentiation potential of stem cells are closely linked with aging (36–38). Under the conditions used in this study, exogenous expression of the TERT gene was not necessary to reprogram human synovial cells isolated from elderly patients with OA (age 71) and did not noticeably increase the reprogramming efficiency (data not shown).

Established human induced PSCs can successfully differentiate into multiple mesenchymal lineages, such as osteoblasts, chondrocytes, and adipocytes. Our results confirm that conventional methods for differentiating mesenchymal lineages that are effective for human ESCs can also be used for human synovial cell–derived induced PSCs. We did not observe any significant differences between individual human induced PSC lines from different patients in our tests, which were potentially due to their similar age and sex. As expected, human induced PSCs and human MSCs displayed distinctive gene expression profiles. (See Supplementary Figure 6 and Supplementary Table 6, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131.) The identification of differentially expressed genes in this study may contribute to an increased understanding of the molecular characteristics of stem cells and their capacity for self-renewal and differentiation.

To our knowledge, this is the first study to identify MSC-like synovial cells from a hip joint and to show reprogramming of OA patient–derived synovial cells to human induced PSCs. We also provide evidence that established human induced PSCs possess high chondrogenic differentiation potential both in vitro and in vivo. Chondrogenic differentiation of human induced PSCs was highly efficient, with >80% of differentiating cells producing proteoglycans in extracellular matrix, as confirmed by Alcian blue and Safranin O staining, and >50% of cells being positive for key regulators in chondrogenic differentiation (SOX9, type I, type II, and type X collagen) in both pellet and scaffold culture (Figure 5C). Cartilage formation within human induced PSC–induced teratomas was confirmed by histologic analysis (Figure 6A and Supplementary Figure 5A) and positive immunohistochemical staining for chondrogenic-specific molecules (SOX9, type I, type II, and type X collagen, and aggrecan) (Figure 6B and Supplementary Figure 5B). We determined that the expression of aggrecan mRNA in chondrogenically differentiated human induced PSCs was markedly higher than that in human ESCs.

Previous studies showed that chondrocytes from OA joints develop hypertrophy (39, 40), while chondrocytes from healthy articular cartilage maintain a stable articular cartilage phenotype without evidence of hypertrophy (41). Interestingly, increased expression of type X collagen, which is characteristic of hypertrophic chondrocytes, was observed in only 1 human induced PSC line of the 3 cell lines examined, which included 2 human induced PSC lines and an H9 human ESC line (Figure 5B). We assumed that the state of the donor cells may influence the differentiation potential of established human induced PSC lines. However, further studies are certainly needed to develop a greater understanding of the relationship between donor cells and established human induced PSCs. Further studies are also needed to examine whether differences in differentiation potential among OA patient–derived induced PSCs can be overcome by modifying reprogramming methods and/or differentiation protocols.

Insight into the pathogenesis of OA has been obtained largely from animal models, but differences between animal models and the human disease, such as differences due to immune and inflammation mediators and long-term efficacy, are still not understood. Information obtained from human induced PSCs generated from OA patients and their differentiation into specific cell types relevant to the disease will provide valuable insight into OA pathogenesis. Our results demonstrate that autologous synovial cells extracted from OA patients could be used for drug discovery and development and cartilage regeneration in the future, and will provide an important tool in the search for clues to the cellular and molecular defects that cause OA.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Cho had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Janghwan Kim, Han, Chang, Cho.

Acquisition of data. Min-Jeong Kim, Myung Jin Son, Seol, Jongjin Park, Jung Hwa Kim, Su A Park, Kang-Sik Lee, Cho.

Analysis and interpretation of data. Min-Jeong Kim, Myung Jin Son, Mi-Young Son, Yong-Hoon Kim, Chul-Ho Lee, Cho.

Acknowledgements

We thank Professor Dae-Sik Lim for kindly providing GP2-293 cells for virus production.

Ancillary