SEARCH

SEARCH BY CITATION

Keywords:

  • Gene expression profile;
  • Mesenchymal stem cells;
  • Adipose tissue;
  • Bone marrow;
  • FKBP5

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Mesenchymal stem cells derived from human bone marrow (hBMSCs) and human adipose tissue (hAMSCs) represent a useful source of progenitor cells for cell therapy and tissue engineering. However, it is not clear what the similarities and differences between them are. Like hBMSCs, hAMSCs can differentiate into osteogenic, adipogenic, and chondrogenic cells. Whether MSCs derived from different tissue sources represent fundamentally similar or different cell types is not clear. Given the possible different sources of MSCs for cell therapy, a comprehensive comparison of the different MSCs would be very useful. Here, we compared the transcriptome profile of hAMCS and hBMSCs during directed differentiation into bone, cartilage, and fat. Our data revealed considerable similarities between bone marrow-derived MSCs (BMSCs) and adipose tissue-derived MSCs (AMSCs). We uncovered an interesting bifurcation of pathways in both BMSCs and AMSCs, in which osteogenesis and adipogenesis appear to be linked in a differentiation branch separate from chondrogenesis. Our data suggest that although a set of common genes may be needed for early differentiation into all three lineages, a different set of signature genes is associated with maturation into fully differentiated cells. The recruitment of different late differentiation factors explains and supports our conclusion that BMSCs differentiate more efficiently into bone and cartilage, whereas AMSCs differentiate better into adipocytes. This study not only generated a rich database for continuing molecular characterization of various MSCs but also provided a rational basis for assessing qualities of MSCs from different sources for the purpose of cell-based therapy and tissue engineering.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Mesenchymal stem cells derived from human bone marrow (hBMSCs) represent a source of pluripotent cells that are already in various phases of clinical application [1, [2]3]. Their most immediate use is in the orthopedic context because of the clear demonstration of their ability to differentiate into bone and cartilage [45]. Mesenchymal stem cells derived from human adipose tissue (hAMSCs) manifest multilineage differentiation capacity, including osteogenesis, chondrogenesis, and adipogenesis. Similar to hBMSCs, hAMSCs exhibit stable growth and proliferation kinetics in vitro [6, 7]. Adipose tissue can be obtained by less invasive methods and in larger quantities than bone marrow cells, making the use of hAMSCs as a source of stem cells very appealing. Adipose tissue-derived MSCs (AMSCs) could also potentially be an alternative source to bone marrow-derived MSCs (BMSCs) for use in allogeneic transplants, exploiting the apparent immunosuppressive properties of hBMSCs [8]. Both types of cells have very similar cell surface markers, including CD13, CD29, CD44, CD90, CD105, SH3, and STRO-1 [9]. Thus, MSCs isolated and cultured from various tissues appear to be morphologically similar and do not differ significantly, at least by the expression of commonly used cell surface markers [10]. Although MSCs from various tissues, including bone marrow, periosteum, skeletal muscle, and adipose tissue, have similar epitope profiles, significant differences have been observed in MSC properties according to tissue source, beyond donor and experimental variation. For example, it has been shown that synovium-derived MSCs were superior to MSCs from other sources for clinical applications [11]. Little is known about similarities and differences between BMSCs and AMSCs at the genetic level and during their differentiation into the three major mesenchymal lineages.

To make better use of MSCs for cell-based therapy and tissue engineering, it is useful to understand the process that governs initial commitment and further differentiation into various mesenchymal lineages. In this study, we compared the gene expression profile of human mesenchymal stem cells (hMSCs) derived from adipose tissue and bone marrow during differentiations toward three common mesenchymal lineages. Our data suggest that a set of genes that are upregulated during differentiation is needed for differentiation into all three lineages, whereas late-differentiation genes are essential for terminal differentiation. One of the genes that appear to have a positive role in early differentiation is FKBP5, an immunophilin-binding protein involved in modulating hormone receptor response and transcription regulation.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

MSC Culture and Osteogenic, Chondrogenic, and Adipogenic Differentiation

After informed consent was obtained and following institutional review board guidelines from the National University Hospital of Singapore, hAMSCs were cultured as described [7] from tissue obtained from the knee fat pad or subcutaneous fat. Adipose tissue from the knee fat pad was washed extensively with sterile phosphate-buffered saline (PBS) and treated with 0.075% collagenase (type I; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) in PBS for 30 minutes at 37°C with gentle agitation. The collagenase was inactivated with an equal volume of Dulbecco's modified Eagle's medium (DMEM)/10% fetal bovine serum (FBS), and the infranatant was centrifuged for 10 minutes at low speed. The cellular pellet was resuspended in DMEM/10% FBS and filtered through a 100 μm mesh filter to remove debris. The filtrate was centrifuged as detailed above and plated onto a conventional tissue culture flask in complete medium (DMEM [Gibco-BRL, Auckland, NZ, http://www.gibcobrl.com]; 10% FBS, 100 units/ml penicillin,100 μg /ml streptomycin, and 2 mM l-glutamine [Gibco-BRL]). hBMSCs were cultured as described [12] from bone marrow aspirated from the iliac crest. To prevent spontaneous differentiation, cells were maintained at subconfluent levels. Briefly, 2–10-ml bone marrow aspirates were taken from iliac crest after informed consent was obtained. Nucleated cells were isolated with a density gradient (Ficoll-Paque; Pharmacia, Piscataway, NJ, http://www.amershambiosciences.com) and resuspended in complete culture medium. All of the nucleated cells were plated in 25 ml of medium in a culture dish and incubated at 37°C with 5% CO2. After 24 hours, nonadherent cells were discarded, and adherent cells were thoroughly washed twice with PBS. The cells were incubated for 5–7 days, harvested with 0.25% trypsin and 1 mM EDTA for 5 minutes at 37°C, and replated at 6 cells per cm2 in culture flasks. As described [7], approximately 90% confluent MSCs at passage 2 in a T75 flask were induced to differentiate into adipocytes and osteocytes for 3 and 14 days in adipogenic and osteogenic media, respectively. Adipogenic medium contained 0.5 mM isobutyl methylxanthine, 1 μM dexamethasone (Sigma-Aldrich), 10 μM insulin, 200 μM indomethacin, and 1% antibiotic/antimycotic. Osteogenic medium contained 0.1 μM dexamethasone, 50 μM ascorbate-2-phosphate, 10 mM β-glycerophosphate, and 1% antibiotic/antimycotic. Approximately 90% confluent MSCs at passage 2 were induced into chondrocytes for 3 and 14 days for comparison of gene expression profiles between BMSCs and AMSCs as described [13] without subculture in chondrogenic medium containing 10 ng/ml transforming growth factor (TGF)-β1 (R&D Systems Inc., Minneapolis, http://www.rndsystems.com), 6.25 μg/ml insulin, 50 nM ascorbate-2-phosphate, and 1% antibiotic/antimycotic. Pellet culture as described [12] was used for comparison in chondrogenesis between BMSCs and AMSCs and functional study of FKBP5 in chondrogenesis. Briefly, 2 × 105 MSCs were placed in a 15-ml polypropylene tube (Falcon; Becton, Dickinson and Company, Franklin Lakes, NJ, http://www.bd.com) and centrifuged to pellet. The pellet was cultured at 37°C with 5% CO2 in 500 μl of chondrogenic media that contained 10 ng/ml TGF-β3, 10−7 M dexamethasone, 50 μg/ml ascorbate-2-phosphate, 40 μg/ml proline, 100 μg/ml pyruvate, and 50 mg/ml ITS+Premix (Becton, Dickinson and Company; 6.25 μg/ml insulin, 6.25 μg/ml transferrin, 6.25 μg/ml selenious acid, 1.25 mg/ml bovine serum albumin, and 5.35 mg/ml linoleic acid). The medium was replaced every 3–4 days for 21 days. Oil red stain for adipogenesis and Alizarin Red S stain for calcium deposition in osteogenesis were examined by area of positive stain with BioQuant software (BioQuant, Nashville, TN, http://www.bio-quant.com), COLII for chondrogenesis was examined by quantitative polymerase chain reaction.

cDNA Microarray Analysis

Total RNA was isolated from MSCs or MSCs induced differentiation to the osteocytes, chondrocytes, and adipocytes using the RNeasy mini-kit (Qiagen, Chatsworth, CA, http://www1.qiagen.com) per the manufacturer's protocol. In brief, 1.5 μg of total RNA was used to synthesize double-strand DNA using one-cycle cDNA synthesis kit. cDNA was purified by using a Sample Cleanup Module (Qiagen). In vitro transcription was performed to produce biotin-labeled cRNA using a GeneChip IVT Labeling Kit. Biotinylated cRNA was cleaned and fragmented to 50–200 nucleotides with the Sample Cleanup Module and hybridized for 16 hours at 45°C to Affymetrix HG U133 Plus 2.0 (Santa Clara, CA, http://www.affymetrix.com), containing more than 54,675 human genes. After being washed, the array was stained with streptavidin-phycoerythrin (Molecular Probes Inc., Eugene, OR, http://probes.invitrogen.com). The staining signal was amplified by biotinylated anti-streptavidin (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com), followed by streptavidin-phycoerythrin stain, and then scanned using GCOS 3000 (Affymetrix). The data were analyzed using GeneSpring software V7.2. A t test on normalized intensity with p ≤ .05 followed by ratio change (ratio of normalized intensity, ≥2.0 or ≤0.5) was used to generate the list of genes with significant change in expression profile during three differentiations. In this study, BMSCs from three patients and AMSCs from six patients were used.

Quantitative Real-Time Polymerase Chain Reaction

To confirm the microarray data, real-time polymerase chain reaction (PCR) was performed with the TaqMan expression assay according to the manufacturer's instructions and an ABI 7700 Prism (Applied BioSystems, Foster City, CA, http://www.appliedbiosystems.com); 0.75 μg of total RNA was converted to cDNA using a high-capacity cDNA archive kit and then diluted to 750 μl. Quantitative real time-PCR was done as follows: initial denaturation for 2 minutes at 50°C and 10 minutes at 95°C, followed by 40 cycles of PCR (95°C for 15 seconds, 60°C for 1 minute) by using 10 μl of 2× Master mix, 1 μl of TaqMan probe, and 9 μl of cDNA. All probes were designed with a 5′ fluorogenic, 6-carboxylfluorescein, and a 3′ quencher, tetramethyl-6-carboxyrhodamine. The expression of human glyceraldehyde-3-phosphate dehydrogenase was used to normalize gene expression level. Primers used for real-time PCR included CCAAT-enhancer-binding protein-α (C/EBPα), NOX4, osteomodulin (OMD), and FKBP5.

Immunoblotting Analysis

Cells were collected by centrifugation, and the cell pellet was resuspended in lysis buffer (25 mM Tris, pH 7.5, 150 mM NaCl, 1% Nonidet P40, 0.5% sodium deoxycholate, 0.1% SDS) containing proteinase inhibitors and incubated at 4°C for 30 minutes. Following centrifugation at 16,000g for 15 minutes at 4°C, the supernatant containing total cell extract was collected and kept at −80°C. Protein from cell extract in the gel was electrophoretically transferred onto a Hybond polyvinylidene difluoride membrane (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com). The membrane was incubated for 1 hour at room temperature in Tris-buffered saline and 0.1% Tween 20 (TBS-T) containing 5% skim milk to block nonspecific protein binding and incubated at room temperature for 1 hour with the primary antibody against FKBP5 (Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com) diluted (1:300) in blocking buffer. Following four washes with TBS-T, the membrane was incubated for 1 hour with the horseradish peroxidase-conjugated secondary antibody diluted (1:3000) in blocking buffer for 1 hour. Antibody binding was visualized with an enhanced chemiluminescence Western blotting detection system (Amersham Biosciences).

RNA Interference

Small interfering RNA (siRNA) duplexes (Ambion, Austin, TX, http://www.ambion.com) used in this study consisted of a 21-nucleotide sense strand and a 21-nucleotide antisense strand with a 2-nucleotide T overhang at the 3′ end. The sequences were as follows: FKBP5 siRNA sense, GGAGCAACAGUAGAAAUCCTT; antisense, GGAUUUCUACUGUUGCUCCTT. siRNA (FKBP5 100 nM) was introduced into hMSCs using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). MSCs were transfected with cy3-labeled Silence negative control siRNA (Ambion) as an experimental control. At 48 hours post-transfection, specific-siRNA-treated cells and control siRNA cells were analyzed with real-time PCR. To study the long-term effect of FKBP5 knockdown on differentiation of MSCs, lentiviral vector for knockdown was created by cloning short hairpin FKBP5 RNA into pLL3.7. Lentivirus was generated by cotransfecting pLentiviral vector for FKBP5 knockdown and packaging mix (Invitrogen) into 293FT cells, and supernatant was harvested after 48 hours. MSCs were infected with viral supernatant to achieve FKBP5 knockdown, and infected MSCs were induced to undergo differentiation for 14 days to evaluate adipogenesis and osteogenesis and for 21 days to evaluate chondrogenesis. The empty pLL3.7 vector with no insert was used as a control.

Lentivirus Production and Generation of MSCs Overexpressing Stably Integrated Genes

FKBP5 was amplified from cDNA of BMSCs differentiated into osteogenic differentiation for 14 days, digested with BamHI and EcoRI, and then ligated into pEntry3C (Invitrogen). Via LR (attL and attR) recombination between pEntry3C and pDest6/V5 (Invitrogen), pLentiviral vector for overexpression of FKBP5 was created. Lentivirus was generated by cotransfecting pLentiviral vector for overexpression of FKBP5 and packaging mix (Invitrogen) into 293FT cells, and then MSCs were infected with viral supernatant and were selected with 5 μg/ml blastidin for 7 days. FKBP5-overexpressed MSCs were induced differentiation into three lineages for 2 days and then analyzed with real-time PCR compared with no-insert control.

Data and Statistical Analysis

Data were analyzed using GeneSpring software and using one-way analysis of variance.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

MSCs Differentiated into Adipogenic, Osteogenic, and Chondrogenic Lineages

hAMSCs and hBMSCs were isolated from six and three healthy donors, respectively. Their race, age, and sex are described in supplemental online Table 1. MSCs were expanded by subculture every 2–3 days at a 1:3 dilution. MSCs at passage 2 cultured in a T75 flask were induced into three mesenchymal lineages as described [7, 13].

First, we showed that both mesenchymal stem cells were capable of differentiating into adipogenic, osteogenic, and chondrogenic cells (Fig. 1A), using histochemical staining for lineage-specific markers. Comparison of the degree of tissue-specific staining and expression of lineage-specific markers indicated that AMSCs differentiated less well into chondrocytes by a 170-fold difference in expression of COLII at day 21 and into osteocytes by a 7-fold difference in area of positive Alizarin Red stain for calcium deposition at day 14 (Fig. 1B).

thumbnail image

Figure Figure 1.. Comparison of cellular and transcript changes during late adipogenesis, chondrogenesis, and osteogenesis in BMSCs and AMSCs. (A): Histochemical staining of adipocytes (oil red O), chondrocytes (Alcian Blue), and osteocytes (Alizarin Red). BMSCs and AMSCs were induced into adipogenesis for 14 days, chondrogenesis under pellet culture for 21 days, and osteogenesis for 14 days. Note the quantitative differences observable between BMSCs and AMSCs. (B): Comparisons in differentiation between BMSCs and AMSCs were examined by oil red stain for adipogenesis and Alizarin Red S stain for calcium deposition in osteogenesis by area of positive stain with BioQuant software, and COLII for chondrogenesis was examined by quantitative polymerase chain reaction. The probability associated with Student's test was performed. (C): Upregulated genes involved in lipid metabolism during adipogenesis for 14 days. t test p value was generated with GeneSpring software V7.2. Note the higher fold change in all markers in AMSC adipogenesis. (D): Upregulated extracellular matrix genes during chondrogenesis for 14 days. t test p value was generated with GeneSpring software V7.2. Note the higher fold change in the major markers (DPT and Col10A1) of chondrogenesis in BMSCs. (E): Upregulated extracellular matrix genes during osteogenesis for 14 days. t test p value was generated with GeneSpring software V7.2. Abbreviations: AMSC, adipose tissue-derived MSC; BMSC, bone marrow-derived MSC; COL10A1, collagen, type 10, α1; DPT, dermatopontin; FABP4, fatty acid-binding protein 4; FBN2, fibrillin 2; hAMSC, MSC derived from human adipose tissue; hBMSC, MSC derived from human bone marrow; LPL, lipoprotein lipase; OMD, osteomodulin; PDK4, pyruvate dehydrogenase kinase 4, isoenzyme, adipocyte; PLIN, perilipin, LOC283445, acetyl-coenzyme A carboxylase β; POSTN, periostin; SPON1, spondin 1.

Download figure to PowerPoint

To identify genes involved in the commitment of MSCs to the three mesenchymal lineages, total RNA was isolated from undifferentiated MSCs and the MSCs at days 3 and 14 after initiating induction of differentiation into bone, cartilage, and fat cells. cDNAs were labeled and hybridized to a microarray chip (Affymetrix) that contains probes for 54,695 human genes, mostly with known functions. The gene expression profiles were compared over time between AMSCs and BMSCs during differentiation. The results revealed considerable similarity in gene expression profiles during differentiation, especially the top upregulated genes between the two types of cells (supplemental online Table 2).

Reproducibility of Data Generated from Microarrays

Because of the genetic variability between different individuals, two methods have been used to filter out this variation. Either RNA samples from different donors can be pooled before use in cDNA array analysis [14], or the average changes in gene expression or changes in gene expression found in the majority of samples can be computed [15]. In our analysis, we applied the latter and computed for mean signal intensity between individuals. To determine the reliability of microarray, we determined the global correlation coefficients between transcriptomes of BMSCs and AMSCs from different patients and from same patients (supplemental online Table 3). We found that the average correlation coefficient between BMSCs was 0.64, and the correlation coefficients were not improved by comparing BMSCs of the same sex. We found that the average correlation coefficient between AMSCs was 0.71, and again, the correlation coefficients were not improved by comparing the same sex. The lowest average correlation coefficient, 0.52, was observed when we compared AMSCs with BMSCs. These data indicated that the microarray data generated was reliable and reproducible enough to detect biological differences between AMSCs and BMSCs and that the differences between AMSCs and BMSCs were not due to culture or technical differences or to variations between individuals. This indicated that data generated from microarrays were reproducible and that culture technique might not underlie differences seen in gene expression.

Confirmation by Real-Time PCR

To confirm the data generated from microarray studies, we performed quantitative RT-PCR using TaqMan on the same total RNA samples used in the microarray studies. The average fold change by PCR was compared with average fold change by microarray detection. We selected genes indicative of different lineages, as shown (supplemental online Table 4): C/EBPα for adipogenic marker, NOX4 for chondrogenic upregulated gene, OMD for osteogenesis, and FKBP5 for upregulated common gene in all three lineages. The genes found to be differentially expressed in the microarray analysis were confirmed to be differentially expressed by quantitative RT-PCR (supplemental online Table 4). However, the degree of increased or decreased expression differed for some genes, likely as a result of the difference in the sensitivity of the two assays. Nevertheless, the result of this comparison gave us the confidence to use our microarray data to deduce biological meaning.

Genes Differentially Regulated During Adipogenic Differentiation In Vitro

Induced differentiation of MSCs toward adipogenesis resulted in a larger cell morphology and a time-dependent increase in intracellular lipid-filled droplets stained by oil red O (Fig. 1A). Our microarray data showed that approximately 230 common genes were upregulated more than twofold at both day 3 and day 14 after induction of BMSCs and AMSCs. Analyzing the data in a different way, we found that approximately 105 and 320 common genes were upregulated more than twofold at days 3 and 14, respectively. A substantial number of genes have been identified as regulated in a differentiation-dependent manner, including constitutively activated gene peroxisome proliferator adipogenic transcription factors (PPARγ), C/EBPα, and the entire spectrum of genes associated with lipid metabolism that contribute to the function and phenotype of the mature adipocyte [16]. As expected, C/EBPα and PPARγ were upregulated during adipogenesis in both BMSCs and AMSCs. Genes involved in lipid metabolism (including lipoprotein lipase [LPL]; fatty acid-binding protein 4 [FABP4]; pyruvate dehydrogenase kinase 4, isoenzyme, adipocyte [PDK4]; perilipin [PLIN]; and acetyl-coenzyme A carboxylase β [LOC283445]) were all upregulated (Fig. 1C; supplemental online Table 2A).

Among upregulated genes for lipid metabolism, genes related to energy reserve metabolism and cholesterol metabolism, such as LEP, LPL, PLIN, SAA1, APOD, and ACDC, were expressed at higher levels in AMSCs than BMSCs, suggesting that AMSCs were superior to BMSCs in adipogenesis. Among upregulated genes for cell cycle, growth, and proliferation, ZNF145, RASD1, and INHBB were expressed more highly in AMSCs, whereas expression of G0S2 and FOX1A was higher in BMSCs. Among signal transduction genes, expression of PDK4, RASD1, and LEP was higher in AMSCs than in BMSCs (supplemental online Table 5). These results together illustrated differences in gene expression between BMSCs and AMSCs during adipogenesis.

Genes Differentially Regulated During Chondrogenic Differentiation

BMSCs cultured in condensate culture under chondrogenic medium, including pellet [12] and high-density monolayer [13, 17], undergo chondrogenic differentiation. AMSCs showed a much weaker differentiation into chondrocytes (Fig. 1A). Our microarray data showed that approximately 77 common genes were upregulated more than twofold at both day 3 and day 14 between BMSCs and AMSCs, and approximately 117 and 73 common genes were upregulated more than twofold at days 3 and 14, respectively. Treatment of MSCs on high-density monolayer with chondrogenic medium resulted in the expression of genes consistent with chondrogenesis. This included an increase in expression of mRNA for cartilage matrix proteins such as collagen type XI, cartilage linking protein 1, dermatopontin, and COL10A1 during chondrogenesis. COL11 triple helices bundle together with type II and type IX collagen triple helices to form collagen fibrils [18]. Cartilage linking protein 1, an extracellular matrix protein in cartilage, gives cartilage its tensile strength and elasticity [19, 20]. Mice lacking cartilage linking protein 1 developed dwarfism and craniofacial abnormalities [21]. Dermatopontin (DPT), a low-molecular-mass component of the extracellular matrix, interacts with decorin and TGF-β and inhibits the formation of the decorin-TGF-βcomplex during chondrogenesis. The expression of DPT significantly increased from 8.199-fold at day 3 to 76.87-fold at day 14 for BMSCs. But DPT expression for AMSCs peaked at day 3 (38.09-fold) and then decreased a little (20.17-fold). COL10A1 is the only known hypertrophic chondrocyte-specific molecular marker [22]. COL10A1 significantly increased from 16.79-fold at day 3 to 56.34-fold at day 14 for BMSCs compared with an increase from 5.855-fold at day 3 to 10.07-fold at day 14 for AMSCs, suggesting a difference in gene expression during chondrogenesis between BMSCs and AMSCs (supplemental online Tables 2B, 6).

Among upregulated extracellular matrix genes, in addition to the genes mentioned above, osteoblast-specific factor 2 (fasciclin I-like, POSTN) progressively increased to day 14. Among cell adhesion genes, DPT and POSTN increased, whereas OMD and WISP1 decreased. Among genes for cell cycle, growth, and proliferation, there was no significant change in expression of MAD2L1 and BUB1 (supplemental online Table 6). Among signal transduction genes, WISPI and INPP4B were more highly expressed in AMSCs than in BMSCs.

Genes Differentially Regulated During Osteogenic Differentiation

Calcium deposition was seen 14 days after induced differentiation toward osteogenesis, as shown by Alizarin red staining. Compared with AMSCs, BMSCs accumulated more calcium during osteogenesis (Fig. 1A). Our microarray data showed that approximately 134 common genes were upregulated more than twofold at both day 3 and day 14 between BMSCs and AMSCs, and approximately 214 and 97 common genes were upregulated more than twofold at days 3 and 14, respectively.

Genes known to be expressed in osteoblasts were consistently upregulated in osteogenic differentiation cultures compared with undifferentiated MSCs. Expression of osteomodulin, implicated in biomineralization processes [23], was much higher in BMSC differentiation than AMSCs at both day 3 and day 14. The promyelotic leukemia zinc finger (PLZF; i.e., ZNF145) was shown to play an important role in early osteoblastic differentiation as an upstream regulator of CBFA1 [24]. As expected, we found that the ZNF145 was upregulated (ratios, 34.49 and 86.09 at days 3 and 14, respectively, for BMSCs; 37.19 and 73.57 at days 3 and 14, respectively, for AMSCs) (supplemental online Table 2C). These results confirmed that MSCs underwent differentiation toward the osteoblastic lineage. In addition, growth repressor delta sleep-inducing peptide immunoreactor (DSIPI) was upregulated more than twofold during osteoblastic differentiation. This was in agreement with the results of Qi et al. [25], suggesting that growth inhibition occurred. We also found that this growth repressor was upregulated more than twofold during adipogenesis and chondrogenesis (Table 1).

Table Table 1.. Common upregulated genes during differentiation between BMSCs and AMSCs at early and late stages
Thumbnail image of

Among extracellular matrix genes, OMD and tissue inhibitor of metalloproteinase 4 (TIMP4) progressively increased with BMSCs, whereas in AMSCs, they were decreased or did not change. Tissue factor pathway inhibitor 2 (TFPI2) was also upregulated during osteogenesis. Notably, expression of IGFBP2, WISP1, TACSTD2, NEDD, and RGC32 was higher in AMSCs than in BMSCs, whereas expression of FOXO1A and CDD20 was lower in AMSCs (supplemental online Table 7). The expression of FOXO1A and NR2F1 continuously increased in BMSCs but did not change in AMSCs (supplemental online Table 7), further indicating differences in osteogenesis between BMSCs and AMSCs.

Comparison of BMSCs and AMSCs in Lineage-Related Genes

To obtain a more quantitative comparison of the difference between differentiation of BMSCs and AMSCs into the three lineages, lineage-related genes were chosen for comparison. During adipogenesis (Fig. 1C), lipid metabolism-related genes were upregulated more highly in AMSCs than in BMSCs, including LPL (ratio, 2.04), FABP4 (ratio, 1.47), PDK4 (ratio, 3.41), PLIN (ratio, 1.82), and LOC283445 (ratio, 1.42).

Treatment of MSCs with chondrogenic medium resulted in the expression of genes related to chondrogenesis (Fig. 1D). Compared with AMSCs, differentiated BMSCs showed much higher in expression of extracellular matrix genes DPT (ratio, 3.81) and COL10A1 (ratio, 5.59). Expressions of COL2 and AGC were too low to be detected at days 3 and 14 by RT-PCR; this is consistent with the results of Zuk et al. [7] and probably resulted from different expression levels at various time points. In osteogenesis (Fig. 1E), differentiated BMSCs expressed much more highly than AMSC levels of extracellular matrix genes, including OMD (ratio, 11.13), DPT (ratio, 2.09), SPON1 (ratio, 3.01), and FBN2 (ratio, 4.68). These results suggested that BMSCs were superior to AMSCs in osteogenesis and chondrogenesis but inferior in their potential for adipogenesis compared with hAMSCs.

Common Upregulated Genes During Three Differentiations

We next wanted to look for genes that were upregulated common to both BMSCs and AMSCs. During differentiation of MSCs into three lineages, 11 and 12 common genes between BMSCs and AMSCs were upregulated more than twofold at the early (3 days) and late (14 days) stages, respectively (Table 1). Among these genes, there were six common upregulated genes at both early and late stages, including FKBP5, ZNF145, SAA1, PCDH9, CPM, and DSIPI. FKBP5 has been shown to inhibit the serine/threonine phosphatase activity of calcineurin in the presence of calcium and calmodulin. Zinc finger protein 145 (ZNF145; PLZF) has been shown to be highly expressed during osteoblastic differentiation, playing an important role in early osteoblastic differentiation as an upstream regulator of CBFA1 [24]. Carboxypeptidase M (CPM) is an extracellular glycosylphosphatidylinositol-anchored membrane glycoprotein. It plays an important role in the control of peptide hormones, growth factor activity at the cell surface, and the membrane-localized degradation of extracellular proteins [26].

Our microarray data showed that FKBP5, ZNF145, and CPM expression was upregulated much more highly in adipogenesis and osteogenesis than in chondrogenesis (Table 1), suggesting a differential role of these genes in mesenchymal differentiation and, most interestingly, a linkage between osteogenesis and adipogenesis (Fig. 2). Among common genes at day 3 differentiations to the three mesenchymal lineages, OMD was expressed most highly in osteogenesis. This is consistent with its preferential high expression in osteoblastic lineages [27]. OMD is also essential in cartilage and bone physiology [28]. Among genes expressed at day 14, DPT in chondrogenesis was expressed at much higher levels than in osteogenesis and adipogenesis. DPT has been shown to interact with other ECM components, especially decorin, and to regulate ECM formation [29]. Expression of APOD was much higher in osteogenesis than in adipogenesis or chondrogenesis, suggesting a more important role for APOD in osteogenesis (Table 1A and B). Furthermore, the expression of APOD was higher in BMSCs than in AMSCs, suggesting different capacities for osteogenesis in BMSCs and AMSCs. APOD is primarily associated with high-density lipoproteins in human plasma and has been reported to participate in maintenance and repair within the central and peripheral nervous systems [30]. It was also reported that APOD was involved in cellular differentiation and growth arrest.

thumbnail image

Figure Figure 2.. Molecular signatures predominantly marking similarities and differences between BMSCs and AMSCs during progression of osteogenesis, adipogenesis, and chondrogenesis. Note the linkage between osteogenesis and adipogenesis and the more robust differentiation of BMSCs and the poorer differentiation of AMSCs to osteocytes and chondrocytes. Abbreviations: ADP, immature adipocytes; AMSC, adipose tissue-derived MSC; BMSC, bone marrow-derived MSC; CHOND, immature chondrocytes; CPM, carboxypeptidase M; DPT, dermatopontin; LPL, lipoprotein lipase; OMD, osteomodulin; OSTEO, immature osteocytes.

Download figure to PowerPoint

Effect of FKBP5 Knockdown on the Differentiation of MSCs

As discussed above, there appears to be a common set of genes upregulated during differentiation to all three lineages. Included in our list is ZNF145, which has been shown to be important for osteogenesis. To further test whether the upregulation of these genes is the cause or effect of differentiation, we chose to investigate FKBP5 by manipulating its expression in MSCs. First, we tested the effect of suppressing the upregulation of FKBP5 by RNA interference-mediated knockdown of the transcript. siRNAs targeting FKBP5 (Ambion) were transfected into hMSCs with Lipofectamine 2000 (Invitrogen). The transfection efficiency of MSCs with cy3-labeled negative control siRNA showed very high transfection efficiency (data not shown). Introduction of siRNA targeting FKBP5 resulted in a downregulation of the lineage markers. Compared with negative control siRNA, FKBP5 siRNA knocked down FKBP5 transcripts by 0.5, 0.35, and 0.3 in mRNA level during adipogenesis, chondrogenesis, and osteogenesis, respectively (Fig. 3A); this was consistent with knockdown in protein level during differentiation into three lineages at day 7 (Fig. 3B). During adipogenesis, C/EBPα and PPARγ were reduced to 63.7% and 56.7%, respectively, of control cells. For chondrogenesis, expression of Col2A1 and COMP was decreased to 38.2% and 40%, respectively, compared with control cells. For osteogenesis, osteocalcin was decreased by more than osteopontin and alkaline phosphatase (45.7% vs. 70.5% and 69.3%) (Fig. 3A). Consistent with the molecular reduction, FKBP5-knockdown MSCs showed decreased oil red stain, proteoglycan, and calcium deposition at cellular level (Fig. 3C, 3D). These results showed that a knockdown of FKBP5 resulted in suppression of expression of differentiation lineage markers to various degrees, indicating that FKBP5 was involved in differentiation.

thumbnail image

Figure Figure 3.. Effects of small interfering RNA (siRNA)-mediated gene silencing of FKBP5 on the differentiation of MSCs. (A): After 48 hours of transfection with FKBP5-targeted siRNA (100 nM) and differentiation, efficiencies of reduction of FKBP5 and lineage markers siRNA were measured by real-time polymerase chain reaction (PCR) compared with negative control. The probability associated with Student's test was performed. (B): FKBP5 was efficiently knocked down by FKBP5-targeted siRNA (100 nM) in protein level after 7 days (d) of transfection and differentiation. Bone marrow-derived MSCs (BMSCs) were induced into osteogenesis for 7 d. Lane 1, BMSCs transfected with negative control siRNA and induced into adipogenesis for 7 d. Lane 2, BMSCs transfected with FKBP5 siRNA and induced into adipogenesis for 7 d. Lane 3, BMSCs transfected with negative control siRNA and induced into chondrogenesis for 7 d. Lane 4, BMSCs transfected with FKBP5 siRNA and induced into chondrogenesis for 7 d. Lane 5, BMSCs transfected with negative control siRNA and induced into osteogenesis for 7 d. Lane 6, BMSCs transfected with FKBP5 siRNA and induced into osteogenesis for 7 d. Lane 7, Undifferentiated BMSCs. (C): FKBP5-knockdown and control MSCs infected with lentivirus were induced into adipogenesis, chondrogenesis, and osteogenesis for 14, 21, and 14 d, respectively. Effects of FKBP5 knockdown on differentiation of MSCs were assessed by oil red stain for intracellular lipid-filled droplets (adipogenesis), Alcian Blue stain for sulfated proteoglycan matrix (chondrogenesis), and Alizarin Red stain for calcium deposits (osteogenesis). Compared with negative controls, FKBP5-knockdown MSCs showed a notable lower staining in all three differentiation pathways. (D): Effects of FKBP5 knockdown with lentivirus on differentiation were examined by oil red stain for adipogenesis and Alizarin Red S stain for calcium deposition in osteogenesis by area of positive stain with BioQuant software, and COLII for chondrogenesis by quantitative PCR. The probability associated with Student's test was performed. Abbreviations: CEBPa, CCAAT-enhancer-binding protein-α; PPARγ, peroxisome proliferator adipogenic transcription factors.

Download figure to PowerPoint

Effect of FKBP5 Overexpression on the Differentiation of MSCs

Next, we studied the effect of enforced FKBP5 overexpression on undifferentiated MSCs. Lentiviral vector for FKBP5 overexpression was constructed as described and introduced into MSCs. In our experiments, we used AMSCs for infection. Compared with transfection of vectors, lentiviral vectors allowed stable expression of foreign genes with high efficiency in MSCs (data not shown). Western blotting showed that FKBP5 was overexpressed in undifferentiated lentiviralinfected AMSCs (Fig. 4A). Enforced overexpression of FKBP5 alone did not appear to induce differentiation at the morphological level. However, when FKBP5-overexpressed MSCs were induced to differentiate into the three mesenchymal lineages for 2 days, the levels of expression of lineage-related genes were consistently higher in the FKBP5-overexpressing AMSCs compared with regular AMSCs similarly induced to differentiate. For example, PPARγ was higher by 1.3-fold in adipogenesis, aggrecan was higher by 3.3-fold in chondrogenesis, and osteocalcin and ALP were higher by 1.6 and 1.2-fold, respectively, in osteogenesis (Fig. 4B). Compared with control, FKBP5-ovexpressed MSCs enhanced three mesenchymal lineage differentiations by stain (Fig. 4C). FKBP5 overexpression increased the area of oil red positive stain for adipogenesis by 1.19-fold, COLII for chondrogenesis by 3.55-fold, and alkaline phosphatase expression for osteogenesis by 2.15-fold (Fig. 4D). These results together indicated that FKBP5 was involved in enhancing and supporting differentiation along all three mesenchymal lineages.

thumbnail image

Figure Figure 4.. Effects of FKBP5 overexpression on the differentiation of MSCs. (A): Detection of FKBP5 overexpression in undifferentiated MSCs after infection with lentiviral virus using Western blotting. Lane 1, MSCs induced into the osteoblasts for 14 days (d) (positive control). A strong FKBP5 protein band was observed. Lane 2, MSCs infected with control lentiviral vector. No FKBP5 protein was detected. Lane 3, FKBP5-infected MSCs. An FKBP5 protein level comparable to that observed in differentiated MSC was achieved by lentiviral vector. (B): After FKPB5-overexpressing MSCs were induced into adipogenesis, chondrogenesis, and osteogenesis for 2 d, fold changes of FKBP5 and lineage markers compared with control-infected MSCs were measured by real-time polymerase chain reaction (PCR). In each instance, the level of markers in FKBP5-overexpressing MSCs was higher than that in control MSC differentiation. The probability associated with Student's test was performed. (C): FKBP5-overexpressed and control MSCs were induced into adipogenesis, chondrogenesis, and osteogenesis for 14, 21, and 14 d, respectively. Effects of FKBP5 overexpression on differentiation of MSCs were assessed by oil red stain for intracellular lipid-filled droplets (adipogenesis), Alcian Blue stain for sulfated proteoglycan matrix (chondrogenesis), and AP stain for alkaline phosphatase activity (osteogenesis). Compared with no-insert controls, FKBP5-overexpressed MSCs showed enhanced staining. (D): Effects of FKBP5 overexpression on differentiation were examined by area of positive oil red stain for adipogenesis with BioQuant software, COLII for chondrogenesis by quantitative PCR, and alkaline phosphatase for osteogenesis compared with control MSCs. The probability associated with Student's test was performed. Abbreviation: PPARγ, peroxisome proliferator adipogenic transcription factors.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Although the bone marrow is the major source of multipotent stromal cells, adipose tissue has been proven to be an alternative source [6, 7]. Although MSCs from bone marrow and adipose tissue appear to be morphologically similar and did not differ by the expression of the main marker genes [9, 10, 31], toluidine blue metachromasia and collagen type immunohistochemical staining were more extensive in the aggregates formed by BMSCs during chondrogenesis compared with AMSCs, suggesting that BMSCs might represent a better choice for progenitor cell-based strategies for cartilage repair [32]. So far, no detailed comparison in gene expression profile between BMSCs and AMSCs during three lineage differentiations has been reported.

In the present study, comparison of the gene expression profile during differentiation of MSCs into three mesenchymal lineages between BMSCs and AMSCs revealed considerable similarity in gene expression profiles during differentiation, especially the top upregulated genes between the two types of cells. Both types of MSCs were capable of multiple mesodermal lineage differentiation, as shown by the expression of several lineage-specific genes and stains.

First of all, we found a common set of genes that were upregulated during differentiation toward all three mesenchymal lineages and in both BMSCs and AMSCs. These included ZNF145 and FABP5. By comparing the differentially expressed genes in three mesenchymal lineages, 11 and 12 common upregulated genes during three differentiations between BMSCs and AMSCs were identified. Among the top candidates we uncovered were ZNF145 and FKBP5.

ZNF145 has been previously reported to play an important role in early osteoblastic differentiation as an upstream regulator of CBFA1 [24]. Our finding indicates that this transcriptional factor may be important for the general initiation of MSC differentiation. Identification of the targets of this factor in MSCs would therefore be very interesting.

The other gene we identified was FKBP5. FKBP5 has been most extensively studied in the context of its role in modulating signals from glucocorticoid receptor steroid and hormone receptors. For example, overexpression of FKBP5 inhibits steroid response [33]. FKBP5 is ubiquitously expressed in adult animals in many tissues, especially muscles, liver, and T lymphocytes [34]. FKBP5 has also been implicated in roles unrelated to steroid receptor function. The FKBP5 homolog, PAS-1, in Arabidopsis plays a critical role in growth and development [35]. In higher organisms, FKBP5 has been shown to act as a transcriptional repressor [36, [37]38]. The expression of FKBP5 in osteogenic tissues, cartilage, and adipocytes has not been described before. Therefore, induction of FKBP5 may have a significant functional role in these lineages not related to differentiation. Here, we showed that knockdown of FKBP5 retarded differentiation to all three lineages, whereas in contrast, overexpression of FKBP5 accelerated the differentiation of both hBMSCs and hAMSCs to all three lineages. This was a surprising finding, but these data together suggested that FKBP5 might in fact play an important role in the early differentiation of these three mesenchymal lineages. How it functions in modulating differentiation remains to be further studied.

By comparing the levels of upregulation of genes at early and late stages, a relationship between the lineages emerged. First of all, from the data shown in Table 1 and supplemental online Table 2, it is clear that ZNF145 and CPM are highly increased in adipogenesis and osteogenesis, in both AMSCs and BMSCs. This is not seen with chondrogenesis. In contrast, dermatopontin and collagen 10A1 are increased during early chondrogenesis, but a marked increase in dermatopontin and collagen 10A1 was only observed in late chondrogenesis in BMSCs. In early adipogenesis, whether in BMSCs or AMSCs, there was an increase in expression of LPL, FABP4, PDK4, and ACDC, and with late differentiation, these markers continue to increase even more in AMSC adipogenesis. In early osteogenesis, OMD was increased in both BMSCs and AMSCs, but in late differentiation, OMD continued to be increased further in BMSC osteogenesis, with a marked increase of APOD, also, that was not observed in AMSC osteogenesis.

We therefore propose the model shown in Figure 2. We suggest that the common list of genes upregulated during differentiation signifies a core set of signature genes involved in initiating differentiation into all three lineages. FKBP5, for example, is a member of this core set. For differentiation along each lineage, other genes are needed. For example, ZNF145 and CPM are needed for both adipogenesis and osteogenesis, OMD for osteogenesis, and DPT for chondrogenesis. Then, for maturation of differentiation, additional genes are recruited: OMD, APOD, and ZNF145 for osteogenesis; DPT and COL10A1 for chondrogenesis; and ACDC, FABP5, and LPL for adipogenesis.

As MSCs differentiate along a specific lineage, there are clearly different sets of genes that are increased. During adipogenesis, BMSCs and AMSCs expressed several genes and proteins involved in lipid biosynthesis and storage. We observed adipo-induced expression of PPARγ2, a fat-specific transcription factor that functions in preadipocyte commitment [39]. There is an upregulation of C/EBPα, a transcription factor expressed in adipose tissues that not only modulates the expression of leptin but also affects the cell cycle. C/EBPA-null mice die shortly after birth [40] from hypoglycemia. There is an increased expression of αP2 (FABP4), a protein associated with lipid accumulation within mature adipocytes [41], Finally, we observed an upregulation of LPL, a lipid exchange enzyme that is increased during adipogenesis [42].

Collagens form the major extracellular matrix protein component of cartilage. Our microarray data showed that COL11A1 was upregulated during chondrogenesis. Cartilage linking protein 1, which is indispensable for stable formation of cartilage proteoglycan aggregates [21], also increased its expression during chondrogenesis. Dermatopontin is an extracellular matrix protein with proteoglycan and cell-binding properties and is assumed to play an important role in cell-matrix interactions and matrix assembly [43]. Osteoblast-specific factor (POSTN) is essential in cartilage [28]. OMD, a small, leucine-rich proteoglycan, is important for collagen fibrillogenesis [44]. The presence of these genes was, in general, consistent with gene expression during chondrogenesis.

Osteomodulin is expressed strongly in osteogenesis [27]. Here, we observed that osteomodulin was upregulated much more highly during osteogenesis compared with adipogenesis and chondrogenesis. PLZF (ZNF145), a transcriptional factor that has been shown to play important roles in early osteoblastic differentiation [24] and regulating limb and axial skeletal pattern [45], was also upregulated. Other genes that showed significant differences in expression during osteogenesis between BMSCs and AMSCs included TIMP4, RGC32, FOXO1A, and NR2F1.

It has been shown that BMSCs and AMSCs are not a homogeneous population of multilineage progenitors. Instead, they are made up of a heterogeneous population of both pluripotent stem cells and tripotent, bipotent, and unipotent progenitors [4, 7, 46, 47].

Therefore, the differences between BMSCs and AMSCs observed here may not be due to the inherent difference between a multipotent BMSC and a multipotent AMSC. Rather, it could be due to the fact that BMSC cultures may be dominated by osteogenic and chondrogenic progenitors, whereas AMSCs have mainly adipogenic progenitors. A direct comparison of the distribution and frequencies of different progenitors between BMSCs and AMSCs has not been made in formal clonal assays, and such experiments may reveal additional information about the cellular basis for the differences that we observed between BMSCs and AMSCs. However, such clonal comparisons may not be easy because the frequencies of progenitor types (adipogenic, chondrogenic, and osteogenic), even in same MSCs, such as AMSCs, can be quite different between different laboratories [7, 47]. One of the reasons may be that the frequencies of progenitor clones appear to be sensitive to culture conditions and passage number [47]. Nevertheless, a carefully controlled experiment may still be able to reveal additional useful information.

In conclusion, our study revealed both similarities and differences between BMSCs and AMSCs. A set of common upregulated genes during differentiation identified at early and late stages of differentiation provided useful leads to further investigation of the signaling pathways that initiate MSC differentiation. Our functional study of one of the genes identified, FKBP5, demonstrates the value of a genomic approach to identifying key genetic elements and pathways involved in growth and differentiation of complex tissues and progenitor populations. This work also generated a useful database for comparison and addition to other genetic data being generated about mesenchymal tissues. Finally, our data support the conclusion that bone marrow-derived MSCs would be a better source of progenitors for cartilage and bone repair.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

We thank Yue Fei Lou, Amber Annette Sawyer, Ying Nan Wu, Zheng Yang, and Tam Wai Leong for their assistance in this study. This research was funded by Biomedical Research Council Grant R175-000-055-305. BL is partially supported by Grants (DK04763 and AI54973) from the NIH.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
Supp_Table_1.pdf21KSupplemental Table 1
Supp_Table_2.pdf41KSupplemental Table 2
Supp_Table_3.pdf17KSupplemental Table 3
Supp_Table_4.pdf14KSupplemental Table 4
Supp_Table_5.pdf77KSupplemental Table 5
Supp_Table_6.pdf69KSupplemental Table 6
Supp_Table_7.pdf73KSupplemental Table 7

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.