In Vitro Expansion of Human Mesenchymal Stem Cells: Choice of Serum Is a Determinant of Cell Proliferation, Differentiation, Gene Expression, and Transcriptome Stability

Abstract

Human bone marrow mesenchymal stem cells (hMSCs) represent an appealing source of adult stem cells for cell therapy and tissue engineering, as they are easily obtained and expanded while maintaining their multilineage differentiation potential. All current protocols for in vitro culture of hMSCs include fetal bovine serum (FBS) as nutritional supplement. FBS is an undesirable additive to cells that are expanded for therapeutic purposes in humans because the use of FBS carries the risk of transmitting viral and prion diseases and proteins that may initiate xenogeneic immune responses. In the present study, we have therefore investigated if autologous serum (AS) or allogeneic human serum (alloHS) could replace FBS for the expansion of hMSCs in vitro. We discovered that the choice of serum affected hMSCs at several different levels. First, hMSCs in AS proliferated markedly faster than hMSCs in FBS, whereas use of alloHS resulted in hMSC growth arrest and death. Second, hMSCs in FBS differentiated more rapidly toward mesenchymal lineages compared with hMSCs in AS. Interestingly, genome-wide microarray analysis identified several transcripts involved in cell cycle and differentiation that were differentially regulated between hMSCs in FBS and AS. Finally, several transcripts, including some involved in cell cycle inhibition, were upregulated in hMSCs in FBS at a late passage, whereas the hMSC transcriptome in AS was remarkably stable. Thus, hMSCs may be expanded rapidly and with stable gene expression in AS in the absence of growth factors, whereas FBS induces a more differentiated and less stable transcriptional profile.

Introduction

Human bone marrow mesenchymal stem cells (hMSCs) represent an appealing source of adult stem cells for cell therapy and tissue engineering. Because hMSCs are present at low frequency in the bone marrow, expansion is necessary before performing clinical studies. Recently, techniques for isolation and extensive subcultivation of hMSCs have been developed. In vitro, culture-expanded hMSCs are capable of differentiation along osteogenic, chondrogenic, and adipogenic lineages [1, 2]. In vivo, several studies in a variety of animal models have shown that hMSCs may be useful in the repair or regeneration of cartilage [3], damaged bone [4, 5], tendon [6], and meniscus [7]. In vitro and in vivo, studies have indicated that hMSCs are capable of differentiation also into cardiomyocytes [8, 9], skeletal muscle [10], and neural precursors [11, 12]. hMSCs have been used in clinical trials in children with osteogenesis imperfecta [13, 14] and to promote engraftment and prevent or treat severe graft-versus-host disease (GVHD) in allogeneic stem cell transplantation [15–19].

In practically all studies using in vitro–expanded hMSCs, the cell culture medium has been supplemented with fetal bovine serum (FBS). This is also true for human clinical trials approved by the U.S. Food and Drug Administration [13]. The risk of transmission of prion diseases and zoonoses from the use of FBS is considered to be small [20]. A greater risk associated with the use of hMSCs expanded in FBS seems to be the immunogenicity of the xenogeneic FBS proteins. Recently, it was shown that a single preparation of 10⁸ hMSCs grown under standard condition in FBS would carry with it approximately 7–30 mg of FBS proteins [21]. The full clinical impact of this observation remains to be investigated, but the use of autologous serum (AS) instead of FBS was recently shown to prevent life-threatening arrhythmias after cellular cardiomyoplasty [22]. Thus, it seems likely that in the future both clinical and regulatory issues will motivate the use of serum supplements other than FBS. We have therefore examined the possibility of using AS or allogeneic human serum (alloHS) rather than FBS for in vitro expansion of hMSCs. In this study, we present the results of experiments comparing proliferation, phenotype, differentiation capability, and gene expression of hMSCs expanded in different serum preparations.

Materials and Methods

Chemicals

All chemicals were purchased from Sigma-Aldrich (St. Louis, http://www.sigmaaldrich.com) unless otherwise stated.

Isolation and Culture of hMSCs

Bone marrow (100 ml) was obtained from the iliac crest of healthy voluntary donors after informed consent. The aspirate was diluted 1:3 in Dulbecco's modified Eagle's medium (DMEM)/F12 (Gibco, Paisley, U.K., http://www.invitrogen.com). After density-gradient centrifugation at 750g for 20 minutes, the mononuclear cell layer was removed from the interface, washed twice, and suspended in DMEM/F12 at 10⁷ cells per ml. To reduce contamination by other adherent cells, CD14⁺ monocytes were removed using magnetic beads coupled to mouse anti-human CD14 monoclonal antibody (Mab), a superMACS magnet, and LS columns (Milteny Biotech, Bergisch Gladbach, Germany, http://www.miltenyibiotec.com) according to the manufacturer's recommendations. The CD14⁻ cells were washed and allowed to adhere overnight at 37°C with 5% humidified CO₂ in five parallel 175-cm² flasks (Nunc, Roskilde, Denmark, http://www.nuncbrand.com). Each flask contained DMEM/F12 medium supplemented with 100 U/ml penicillin, 100 μg/ml streptomycin, and 2.5 μg/ml amphotericin B and 20% serum from one of the following sources: FBS from Biochrom AG (Berlin, http://www.biochrom.de; lot no. 098B), FBS from Biochrom AG (lot no. 074EE), FBS from Gibco (lot no. 3954132S), a human off the clot pooled allogeneic serum (lot no. B03123-028; PAA Laboratories, Linz, Austria, http://www.paa.at), and autologous serum (prepared as described below). On day 1, nonadherent cells were discarded and adherent cells were washed with phosphate-buffered saline (PBS) (Gibco) and then cultured in DMEM/F12 medium with antibiotics and 20% of the same serum. Subsequently, medium containing 20% of the serum used exclusively for that flask was replaced every 3 or 4 days. At approximately 50% confluence, the cells were suspended using trypsin-EDTA and replated at approximately 5,000 cells per cm². After the first passage, amphotericin B was removed and 10% instead of 20% serum was used for further cell cultures. Viable cells were counted at each passage.

Preparation of Human Serum

From each bone marrow donor, 400–500 ml of whole blood was drained into blood bags (Baxter, Deerfield, IL, http://www.baxter.com), quickly transferred to 10-ml vacutainer tubes without anticoagulants (BD, Plymouth, U.K., http://www.bd.com), and allowed to clot for 4 hours at 4°C to 8°C. Subsequently, the blood was centrifuged at 1800g at 4°C for 15 minutes. Serum was collected and filtered through a 0.2-μm membrane (Sarstedt, Nümbrecht, Germany, http://www.sarstedt.com). Aliquots of the sterile AS were stored at −20°C.

Flow Cytometric Analysis

For flow cytometric analyses of surface molecule expression, the following Mabs directly conjugated to fluorochromes were used: CD34–peridinin-chlorophyll-protein complex (PerCP), CD36- PerCP, CD49a–fluorescein isothiocyanate (FITC) or –phycoerythrin (PE), CD49d-FITC, CD56-FITC, CD58-FITC, CD62L-PE, CD71-FITC, CD117-PE, CD152-CY, HLA ABC-CY, HLA DR-PerCP, CD45-CY (BD Biosciences, San Diego, http://www.bdbiosciences.com), CD13-PE, CD14-FITC (Diatec, Oslo, Norway, http://www.diatec.com), CD44-PE, CD90-PE, CD106-FITC (Serotec, Oxford, U.K., http://www.serotec.com), CD133-PE (Miltenyi Biotec), NGFR-FITC, or PE (Chromaprobe, Maryland Heights, MO). Irrelevant control Mabs were included for all fluorochromes. Cells were coated with directly conjugated Mab at room temperature for 15 minutes, washed, and fixed in 1% paraformaldehyde. The supernatant of the CD105 (SH2, Endoglin) cell line hybridoma culture (American Type Culture Collection, Manassas, VA, http://www.atcc.org) was used for unconjugated SH2 staining. Staining with SH2 supernatant was performed as follows: cells were incubated with unconjugated SH2 supernatant at room temperature for 15 minutes, washed, incubated with PE-conjugated goat anti-mouse IgM + IgG + IgA (H + L) (Southern Biotech, Birmingham, AL, http://www.southernbiotech.com) for 15 minutes at room temperature, washed, and fixed. Cells were analyzed using a FACSCalibur flow cytometer (Becton, Dickinson and Company, San Jose, CA, http://www.bd.com). Gates were set based on staining with combinations of relevant Mab and irrelevant Mab so that no more than 1% of the cells were positive using irrelevant Mab.

Mesodermal Lineage Differentiation

Studies on the capability of hMSCs to differentiate along adipogenic, osteogenic, and chondrogenic lineages were performed at passage 4 on cells cultured in AS and on cells cultured in the FBS preparation that supported the most active proliferation (Gibco). For adipogenic differentiation, confluent cultures were incubated in DMEM/F12 containing 10% AS or FBS, 0.5 μM 1-methyl-3 isobutylxanthine, 1 μM dexamethasone, 10 μg/ml insulin (Novo Nordisk, Copenhagen, Denmark, http://www.novonordisk.com), and 100 μM indomethacin (Dumex-Alpharma, Copenhagen, Denmark, http://www.alpharma.com/pages/splash.aspx). Fresh induction medium was replaced every 3 days. After 3 weeks, differentiated cells were fixed with 4% formalin, washed in 50% isopropanol, and subsequently incubated for 10 minutes with Oil-Red O to visualize lipid droplets. Cells were then washed in isopropanol and subjected to nuclear staining with hematoxylin. For osteogenic differentiation, cells were incubated at 3,000 cells per cm² in DMEM/F12 containing 10% AS or FBS, 100 nM dexamethasone, 10 mM β-glycerophosphate, and 0.05 mM L-ascorbic acid-2-phosphate. Fresh induction medium was replaced every 3 days. After 3 weeks, differentiated cells were fixed for 1 hour with 4% formalin and rinsed with PBS without Ca²⁺ and Mg²⁺ (Gibco). Mineralization of the extracellular matrix was visualized by staining with 40 mM Alizarin Red S, pH 4.2, for 5 minutes. For chondrogenic differentiation, 1.5 × 10⁵ cells were pelleted in conical tubes. Subsequently, 500 μl chondrogenic induction medium containing high-glucose DMEM (4.5 g/ml) supplemented with 500 ng/ml bone morphogenic protein- 6 (BMP-6) (R&D Systems, Abingdon, U.K., http://www.rndsystems.com), 10 ng/ml recombinant human transforming growth factor-β1 (R&D Systems), 1 mM sodium pyruvate, 0.1 mM ascorbic acid-2-phosphate, 10⁻⁷ M dexamethasone, 1% ITS (insulin 25 μg/ml, transferrin 25 μg/ml, and sodium selenite 25 ng/ml), and 1.25 mg/ml bovine serum albumin was added. Fresh induction medium was replaced every 3 days. Tissue spheres were collected after 4 weeks and fixed overnight in a 0.1-M cacodylate buffered mixture of 2% glutaraldehyde and 0.5% paraformaldehyde. The samples were embedded in an epoxy resin, and 2-μm-thick sections were cut on a microtom (Leica RM2165, Waldkreiburg, Germany, http://www.leica.com). Sections were then stained with a drop of 0.4% acidic toluidine blue solution for 1 minute, rinsed in distillated water, mounted with Eukitt (O. Kindler GmbH & Co., Freiburg, Germany), and immediately micrographed.

Real-Time Polymerase Chain Reaction

Total RNA was extracted from differentiated cells using Trizol (Invitrogen, Carlsbad, CA, http://www.invitrogen.com). After treatment with DNase I (Ambion, Huntingdon, U.K., http://www.ambion.com), reverse transcription (RT) was performed according to the manufacturer's protocol (Applied Biosystems, Abingdon, U.K., http://www.appliedbiosystems.com) with 100 ng total RNA per 50 μl RT reaction. Real-time quantitative RT–polymerase chain reaction (PCR) assays for peroxisome proliferator-activated receptor (PPAR)–γ 2 [forward primer: 5′-TCCATGCTGTTATGGGTGAAACT-3′, reverse primer: 5′-GTGTCAACCATGGTCATTTCTTGT-3′, TaqMan probe: FAM (5′) -AAGCGATTCCTTCACTGATACACTGTCTG-Darquencher (3′)] were designed using the Primer Express software version 1.5 (Applied Biosystems). Assays for collagen I, collagen II, and aggrecan were performed according to Martin et al. [23]. The assay for osteomodulin was purchased from Applied Biosystems. All assays were designed to overlay a junction between two exons to avoid hybridization to genomic DNA. 18S (Applied Biosystems) was included as an endogenous normalization control to adjust for unequal amounts of RNA. Quantification of mRNA was performed using the ABI Prism 7700 (Applied Biosystems). Each sample (each reaction, 2.5 μl cDNA; total volume, 25 μl) was run in triplicate. Cycling parameters were 95°C for 10 minutes followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. Gene expression was calculated using the relative standard curve method (ABI Prism 7700 Sequence Detection System, User Bulletin 2, PE Applied Systems).

Microarray Analysis

RNA sample preparation and microarray assay were performed according to the Affymetrix GeneChip Expression Analysis Technical Manual (Affymetrix, Santa Clara, CA, http://www.affymetrix.com). In brief, hMSCs were cultured either in FBS (Gibco) or AS. At passage 4, hMSCs from each of three donors (donors 2, 3, and 4) were pelleted and snap-frozen in liquid nitrogen. At passage 10, hMSCs from donors 3 and 4 were treated in the same way. Total RNA was extracted from the cells with Trizol (Invitrogen) following the manufacturer's protocol. For all samples, 10 μg of cRNA was hybridized to the HG-U133A array (Affymetrix) with 22,284 probes representing approximately 14,500 genes. Arrays were scanned at 3 μm using the Agilent Gene Array Scanner (Affymetrix). Gene expression data were analyzed using the Affymetrix Microarray Suite (MAS) 5.0, Affymetrix MicroDB 3.0, and Affymetrix Data Mining Tool (DMT) 3.0 programs. Briefly, a target value of 100 was set for scaling signal intensities for all probe sets. For each comparison, differentially expressed genes were obtained as follows: genes with a present (P) call in one or both populations were selected. Only genes that showed increased (I) or decreased (D) calls were kept for further analysis. Within these genes, only those with a log₂ ratio >1 or <−1 were selected and published using MicroDB 3.0 into DMT 3.0 to obtain gene names and descriptions. Raw data of the microarray analyses are available at http://www.ebi.ac.uk/arrayexpress/under the accession numbers E-MEXP-214 and E-MEXP-215.

Results

hMSCs Are Efficiently Expanded In Vitro in AS and FBS but Not in alloHS

To compare the effect of different serum preparations on the proliferative capability of hMSCs, 10⁸ bone marrow mononuclear cells (BMMCs) depleted of CD14⁺ cells were established in parallel cultures supplemented with one of three different preparations of FBS, AS, or alloHS. Of the three FBS preparations, FBS (Gibco) consistently yielded the highest cell counts (data not shown). Thus, FBS (Gibco) was selected for all further analyses. A comparison of calculated cumulative cell counts between cells expanded in FBS and AS for four donors is shown in Figure 1. Extrapolation of growth curves suggests that there were approximately 10⁵ hMSCs within the 10⁸ CD14⁻ BMMCs on day 0, giving a precursor frequency of one hMSC in 10³ CD14⁻ BMMCs. For each of the donors, the population doubling time was always shorter in hMSCs in AS (median for the four donors, 53 hours; range, 41–54 hours) compared with hMSCs in FBS (median, 84 hours; range, 76–89 hours). In cultures supplemented with alloHS, the number of adherent cells on day 1 was always lower than for cultures with other serum supplements (data not shown). The adherent cells in cultures supplemented with alloHS spread and formed small colonies, but proliferation soon ceased, and as cells cultured in alloHS did not survive (beyond passage 1), cell counts are not entered into Figure 1.

Figure Figure 1.

Proliferation of human marrow mesenchymal stem cells (hMSCs). Calculated cumulative cell counts in cultures of hMSCs from four different donors are plotted with filled symbols for hMSCs expanded in autologous serum and open symbols for hMSCs expanded in fetal bovine serum. Identically shaped symbols (e.g., open and filled squares) represent results from the same donor.

hMSCs Cultured in AS Are Morphologically and Phenotypically Similar to hMSCs Cultured in FBS

From initial colony formation until senescence, hMSCs showed a fibroblastoid appearance, with no discernible morphological differences between cells expanded in AS and FBS at any time (supplemental online Fig. 1). However, we observed that hMSCs grown in FBS required prolonged exposure to trypsin to detach from the plastic surface (data not shown). This could not be explained, as determined by flow cytometry, by differential expression of the adhesion molecules tested. Indeed, these were almost equally expressed on the cells regardless of the serum supplement (supplemental online Table 1). As expected, hMSCs expressed HLA class I, CD13, CD44, CD90, and CD105 (SH2) [1].

Table Table 1.. Relative expression of molecules characteristic for different lineages

	Control		Osteogenic		Chondrogenic		Adipogenic
	FBS	AS	FBS	AS	FBS	AS	FBS	AS
Osteomodulin	1	ND	30.7	22.2	333	1,237	5.3	12.2
Collagen I	9.4	11.3	1.1	1.2	18.3	21	1	4.3
Collagen II	ND	ND	ND	ND	8.6	1	ND	ND
Aggrecan	348	42.6	9.3	9.1	8,470	3,190	8.9	1
PPAR-γ2	ND	ND	ND	ND	ND	62.5	245	1

• ^a mRNA levels were scaled according to the sample with the lowest detectable expression. This value was set to 1. Adjusted for the concentration of expression of 18S. Values are median of three experiments.
• ^b Abbreviations: AS, autologous serum; FBS, fetal bovine serum; ND, not detected; PPAR, peroxisome proliferator-activated receptor.

Choice of Serum Impacts on the Rate of Differentiation of hMSCs

The multilineage differentiation capability of hMSCs expanded in FBS or AS was examined by culturing cells collected at passage 4 under conditions favorable for chondrogenic, osteogenic, and adipogenic differentiation [1]. Examples of staining assays obtained after 3 or 4 (chondrogenic differentiation) weeks of culture in induction medium are shown in Figure 2. At this time, staining for fat globules (adipogenic differentiation) and mineralization (osteogenic differentiation) was observed in cells from all donors and both serum preparations. However, for cells in adipogenic induction medium, fat globules could be observed under the light microscope already after 3–4 days in hMSCs in FBS, whereas fat globules appeared 8–12 days later in hMSCs in AS (data not shown). The same tendency was observed in chondrogenic-induced cultures. Here, cells were cultured in pellets that lost attachment and appeared as spheres. For hMSCs in FBS, the spheres increased in size, indicating production of extracellular matrix (ECM), from approximately 2 weeks of culture. A similar increase in size occurred later (or not at all) for hMSCs in AS. After toluidine blue staining, all spheres of hMSCs in FBS stained positive, whereas spheres from only one out of three of the donor's hMSCs cultured in AS stained positive (Fig. 2A).

Figure Figure 2.

Differentiation of hMSCs. Results of staining assays on hMSCs induced to differentiate in chondrogenic direction (A, stained with toluidine blue), osteogenic direction (B, stained with alzarin red), and adipogenic direction (C, stained with oil-red O) are shown. Left, cells expanded in FBS before differentiation; right, cells expanded in AS. Within each of these categories, differentiated cells are shown in the left panels and undifferentiated cells are shown in the right panels. All images magnified ×200. Abbreviations: AS, autologous serum; FBS, fetal bovine serum; hMSCs, human bone marrow mesenchymal stem cells.

Observations from real-time PCR analyses were in line with the results from the staining assays and our observations of the cells in differentiation cultures (Table 1). Collagen type II mRNA was present only in cells undergoing chondrogenic differentiation and at higher levels in cells grown in FBS than in AS. Aggrecan mRNA was detected at much higher levels after chondrogenic differentiation than in other cultures and to a higher extent in cells expanded in FBS than in AS. Collagen type I mRNA was present at relatively high levels in all cultures, upregulated upon chondrogenic differentiation relative to undifferentiated cells and downregulated in osteogenic and adipogenic differentiation. Osteomodulin mRNA appeared most strongly in cultures after chondrogenic or osteogenic induction. PPAR-γ2 mRNA was detected in cells induced toward adipogenic differentiation at much higher levels in cells expanded in FBS than in AS. Surprisingly, PPAR-γ2 mRNA was observed also in AS cells induced toward chondrogenic differentiation.

Serum Supplement Is a Determinant of Gene Expression

To assess whether the serum supplement used for hMSC culture affected their gene expression, we performed a series of microarray analyses on cells at passage 4. Out of 22,284 probes expressed on the chips, only a few hundred probes represented genes that were differentially expressed depending on the serum supplement. Although some of these differences in expression were donor-specific, many were shared between the three donors (supplemental online Fig. 2). Using twofold upregulation as cut-off, 79 probes representing 59 genes were upregulated in hMSCs in FBS compared with hMSCs in AS. Some of the most highly overexpressed genes, together with genes found to be functionally related to other observations in this study, are presented in Table 2. A detailed list of all the genes overexpressed in hMSCs cultured in FBS is presented in supplemental online Data 1.

Table Table 2.. Selected genes upregulated in human marrow mesenchymal stem cells maintained in fetal bovine serum versus autologous serum at passage 4

Gene name	Fold upregulation
Cell cycle–related and proliferation-related genes
Growth arrest-specific 1^a (GAS1)	5
Antiproliferative protein 2, B-cell translocation gene (BTG1)	3
Genes associated with extracellular matrix and cytoskeleton
Keratin 14^a (KRT14)	21
Cartilage glycoprotein-39, Chitinase 3-like 1 (CHI3L1)	11
Decorin^a (DCN)	10
Collagen, type XIV, alpha 1, undulin^a (COL14A1)	7
Collagen I inducible protein, phosphatidic acid phosphatase type 2B (PPAP2B)	5
Collagen, type XI, alpha 1^a (COL11A1)	4
Growth factor–related genes
Cytokine receptor-like factor 1^a (CRLF1)	16
Leptin receptor^a (LEPR)	13
Expressed sequence tag consensus, including insulin-like growth factor 2 (IGF2)	12
Fibroblast growth factor 7 (keratinocyte growth factor) (FGF7)	6
Fibroblast growth factor receptor 2^a (FGFR2)	6
Interleukin 7 (IL7)	4
Genes associated with organ-specific cells
Ectonucleotide pyrophosphatase/phosphodiesterase 2, autotaxin^a (ENPP2)	8
cDNA FLJ36690, highly similar C1r component precursor^a (C1R)	5
Glycoprotein (transmembrane) nmb^a (GPNMB)	4
Complement component 1, s subcomponent^a (C1S)	4
Complement component 1, r subcomponent-like (C1RL)	3
Cathepsin K (CTSK)	3
Genes associated with signaling pathways and transcription
Inhibitor of DNA binding 4, dominant-negative helix-loop-helix proteina (ID4)	6
MAD, mothers against decapentaplegic homologue 6 (SMAD6)	6
Inhibitor of DNA binding 2, dominant-negative helix-loop-helix proteina (ID2)	4
Frizzled homologue 1 (FZD1)	3
OLF-1/EBF-associated zinc finger gene, zinc finger protein 423 (ZNF423)	3
Metabolism-related genes
Prostaglandin D2 synthase 21 kDaa (PTGDS)	9
Prostaglandin F synthase, aldoketo reductase family 1, member C3 (AKR1C3)	6
Heme oxygenase (decycling) 1 (HMOX1)	4
Prostaglandin E synthasea (PTGES)	3

• a ^aDifferentially expressed at passage 10.

Consistent with their slower rate of proliferation, hMSCs in FBS overexpressed genes associated with prolongation of the cell cycle, that is, growth arrest–specific 1 and antiproliferative protein 1. Several genes coding for constituents of the ECM were also upregulated in hMSCs in FBS. Chondrocyte differentiation– related genes, including ECM genes, cytokine receptor-like factor 1 [24], leptin receptor [25], and ectonucleotide pyrophosphatase/phosphodiesterase 2 [26], were overexpressed as well. Genes connected to the chondrocyte-inducting factors transforming growth factor-β and BMP-6 signaling pathways, such as SMAD6 and OLF-1/EBF-associated zinc finger gene, were also upregulated. Moreover, genes related to differentiation into osteoblasts were expressed at higher levels in hMSCs grown in FBS such as cytokine receptor–like factor 1 [24] and glycoprotein (transmembrane) nmb [27]. Other upregulated genes were associated with adipogenesis, such as the leptin receptor [28], inhibitor of DNA binding 4 [29], and members of the complement system [30]. Finally, genes related to the synthesis of prostaglandins were also upregulated in hMSCs expanded in FBS. Importantly, many of these genes were overexpressed in FBS also at passage 10.

Considerably fewer genes were consistently upregulated in hMSCs expanded in AS compared with hMSCs in FBS. The entire list is presented in Table 3 (for details, see supplemental online Data 2). Among these genes, angiopoietin-like 4 was previously shown to inhibit apoptosis in endothelial cells and thereby to contribute to the increased cell number observed in these cell cultures [31]. In addition, angiopoietin-like 4 is a target gene for PPARs and thus likely to be involved in adipogenesis. Another gene, ectonucleotide pyrophosphatase/phosphodiesterase 1, has been shown to act as an antagonist of bone mineralization [32]. As for hMSCs in FBS, many of the differentially regulated genes in AS at passage 4 were differentially expressed also at passage 10.

Table Table 3.. Genes upregulated in human marrow mesenchymal stem cells maintained in autologous serum versus fetal bovine serum at passage 4

Gene name	Fold upregulation
Genes associated with signaling pathways and transcription
Regulator of G-protein signaling 4^a (RGS4)	6
PTPL-1 associated RhoGAP 1^a (PARG1)	3
Signal sequence receptor gammaa (SSR3)	3
Calcium/calmodulin-dependent protein kinase II (CaMKIINalpha)	3
Metabolism-related genes
Angiopoietin-like 4a (ANGPTL4)	12
Ectonucleotide pyrophosphatase/phosphodiesterase 1, PC-1 (ENPP1)	5
DnaJ (Hsp40) homologue, subfamily B, member 4a (DNAJB4)	5
Transglutaminase 2 (TGM2)	4
Tumor necrosis factor superfamily, member 10d, TRAIL-R4-B (TNFRSF10D)	3
Spastic ataxia of Charlevoix-Saguenay, sacsin (SACS)	2
Ser (or cys) proteinase inhibitor, cladeE (SERPINE1)	2
Other genes and expressed sequence tags
Leucine-rich repeat containing 17a (LRRC17)	4
EST from clone containing 2 transglutaminase 2 isoforms (TGM2)	3
Downregulated in ovarian cancer 1 (DOC1)	2

• a ^aDifferentially expressed at passage 10.

Serum Supplement Is a Determinant of Transcriptome Stability

To be safe for use in therapeutic protocols, in vitro–expanded hMSCs should not change in the course of cell culture. To determine the effect of serum supplement on transcriptome stability in ex vivo–expanded hMSCs, we compared transcript expression of hMSCs derived from two donors expanded in AS or FBS at passage 4 versus passage 10. Between these time points, the cells underwent approximately 13.5 population doublings, corresponding to approximately 10⁴-fold expansion. Scatter plot analyses of the results are shown in online Figure 3. Although the numbers of differentially expressed genes varied between donors, they were consistently much higher for hMSCs expanded in FBS. For hMSCs in FBS, 46 genes, listed in Table 4 and described in supplemental online Data 3, were differentially expressed over time in both donors. The most striking changes were observed for genes involved in the cell cycle. In passage 10 cells, which were close to proliferative senescence, several genes with essential functions for the normal progression through cell cycle were dramatically downregulated. These genes included topoisomerase (DNA) II alpha (−66-fold), cell division cycle 2 (−30-fold), ribonucleotide reductase M2 polypeptide (−30-fold), Asp (abnormal spindle)-like, microcephaly associated (−24-fold), and others. The effect of FBS on the production of constituents of the ECM and organization of the cytoskeleton was underscored by the fact that a large number of genes important for the cytoskeleton and ECM were upregulated. Moreover, several genes associated with organ specification, including neuronal tissues (neurotrypsin, neurotrophic tyrosine kinase receptor, myelin basic protein, nerve growth factor beta, neurophilin, Eph5A, and tetraspan 3), vessels (prostaglandin 12, tumor endothelial marker 8, neurotrypsin, and Nur77), and bone (osteoprotegrin and oxytocin receptor) were also differentially expressed.

Table Table 4.. Changes in gene expression in human marrow mesenchymal stem cells maintained in fetal bovine serum at passage 10 versus passage 4

	Fold change
Gene name	Up	Down
Cell cycle–related and proliferation-related genes
Topoisomerase (DNA) II alpha 170 kDa (TOP2A)		−66
Cell division cycle 2, G1 to S and G2 to M, Cdk1 (CDC2)		−30
Ribonucleotide reductase M2 polypeptide (RRM2)		−30
Asp (abnormal spindle)-like, microcephaly associated (ASPM)		−24
Nucleolar protein ANKT (NUSAP1)		−8
Cyclin D2 (CCND2)	6
Cyclin B2 (CCNB2)		−6
Cyclin A2 (CCNA2)		−4
Aurora kinase B (AURKB)		−4
Serine/threonine kinase 6 (STK6)		−4
Genes associated with adhesion, extracellular matrix, and cytoskeleton
Integrin, alpha 4, antigen CD49D (ITGA4)	5
Keratin 18 (KRT18)	4
Collagen, type XIV, alpha 1, undulin (COL14A1)	4
Keratin 14 (KRT14)	3
Bone marrow stromal cell antigen 1, CD157 (BST1)	3
Genes associated with signaling pathways and transcription
Phosphodiesterase 5A, cGMP-specific (PDE5A)	3
PTPL1-associated RhoGAP 1 (PARG1)	2
Growth factor–related genes
Hyaluronan-mediated motility receptor, CD168, RHAMM (HMMR)		−6
Oxytocin receptor (OXTR)	5
Leptin receptor (LEPR)	5
Secreted frizzled-related protein 4 (SFRP4)	4
Transforming growth factor, beta 2 (TGFB2)	4
Nerve growth factor, beta polypeptide (NGFB)	3
Organ-specific genes
Neurexin 3 (NRX3)	23
Osteoprotegerin, tumor necrosis factor receptor superfam memb 11b (TNFRSF11B)	13
Myosin ID (MYO1D)	10
Neurotrypsin, motopsin, strong similarity protease serine 12 (PRSS12)	7
Nuclear receptor subfamily 4, group A, member 2, NGFI-B/Nur77 (NR4A2)		−7
Neurotrophic tyrosine kinase receptor (NTRK2)	6
Myelin basic protein (MBP)	4
Tumor endothelial marker 8, anthrax toxin receptor 1 (ANTXR1)	4
Prostaglandin I2 (prostacyclin) synthase (PTGIS)	4
Neuropilin (NRP) and tolloid (TLL)-like 2 (NETO2)		−2
Ephrin type A receptor 5 (EPHA5)	2
Tetraspan 3 (tetraspanin 3, transmembrane 4 superfamily member 8) (TM4SF8)	2
Metabolism-related genes
Fatty acid binding protein 5 (FABP5)		−6
Ubiquitin-specific protease 53 (USP53)	4
Solute carrier family 4, sodium bicarbonate cotransporter member 4 (SLC4A4)	4
Cytochrome P450, family 1, subfamily B, polypeptide 1 (CYP1B1)	3
Thioredoxin-interacting protein (TXNIP)	3
Other genes and expressed sequence tags
Chromosome 10 open reading frame 3 (C10orf3)		−21
KIAA0101 gene product (KIAA0101)		−7
Clusterin, apolipoprotein J (CLU)		−4
H2B histone family, member S (HIST1H2BK)	3
Hypothetical protein FLJ12428, DEP domain containing 6 (DEPDC6)	3
SRY (sex-determining region Y)-box 11 (SOX11)		−3

In contrast, only six genes, listed in Table 5 and described in supplemental online Data 4, were transcriptionally altered in hMSCs in AS in both donors after 12 population doublings. These were all downregulated at passage 10 and were predominantly genes associated with the ECM and cytoskeleton.

Table Table 5.. Changes in gene expression in human marrow mesenchymal stem cells maintained in autologous serum at passage 10 versus passage 4

	Fold change
Gene name	Up	Down
Genes associated with extracellular matrix and cytoskeleton
Collagen, type XI, alpha 1 (COL11A1)		−8
Calponin 1, basic (smooth muscle) (CNN1)		−3
Actin, gamma 2 (smooth muscle, enteric) (ACTG2)		−2
Kinesin family member 18A (KIF18A)		−2
Genes associated with signaling pathways and transcription
G protein–coupled receptor (chemokine orphan receptor 1) (CMKOR1)		−4
Stromal cell–derived factor 1 (SDF-1), Chemokine (C-X-C motif) ligand 12 (CXCL12)		−2

Discussion

The increasing number of clinical protocols using hMSCs underscores the need for serum supplements other than FBS. Serum-free media have been investigated, but none has been published so far that supports the proliferation of hMSCs in the absence of growth factors. Recently, the persistence of xenogeneic proteins in hMSCs expanded in FBS was examined extensively [21]. Results from this study showed that, after intravenous administration of autologous rat MSCs expanded in FBS, humoral immune responses against FBS proteins were observed in the recipient. To reduce the immunogenic effect of FBS carried within hMSCs, the authors changed to AS supplemented with growth factors after an initial period of expansion in FBS [21]. Another study has described the comparison between hMSCs in FBS versus AS, but only in short-term cultures and with limited phenotypic characterization [33]. In the present study, we show that expansion of hMSCs in AS in the absence of FBS and without any cytokines is as effective as supplementing the culture medium with FBS. To the best of our knowledge, this is the first study that provides extensive phenotypic and gene expression analysis of hMSCs expanded to large numbers entirely without FBS or cytokine supplements.

In our cultures of CD14⁺-depleted BMMCs, we calculated the incidence of hMSCs to be 1 in 10³. In previous studies, the hMSC precursor frequency has been calculated to be 2 to 5 per 10⁶ BMMCs [34], 1 in 10⁴ BMMCs based on selection by the D7-FIB Mab and colony-forming units fibroblastic (CFU-F) assays, and 3 in 10⁴ BMMCs based on isolation of STRO-1^bright/vascular cell adhesion molecule–positive (VCAM⁺) cells and CFU-F assays in medium supplemented with growth factors [35]. Polyclonal cultures may provide a growth advantage to all clonally competent hMSC precursors in the culture by the secretion into the culture medium of dickkopf-1 by the earliest hMSC precursors entering into cell cycle. In this setting, high levels of dickkopf-1 stimulates the entry into cell cycle by other hMSC precursors and also provides a proliferation stimulatory signal [36]. Thus, our estimate of the incidence of cells in BMMCs that are able to plate and expand to form cultures of cells with all the characteristics of hMSCs in the absence of growth factors may closely reflect the true precursor frequency of hMSCs in BMMCs. However, the hMSC cultures supplemented with alloHS were dramatically different from the others. Here, fewer cells attached and formed colonies. The attached cells survived through a few cell divisions but then died. hMSCs supplemented with alloHS never reached 60% subconfluence in the first flask. Similar results have recently been briefly presented in another study [21]. Clearly, allogeneic differences in serum composition affect hMSC survival and proliferation in vitro to a much greater extent than xenogeneic differences. The allogeneic proteins responsible for this growth inhibition, however, remain to be identified.

We observed that hMSCs expanded in AS proliferated faster but differentiated more slowly than hMSCs expanded in FBS. Changes in gene expression reflected these differences. For example, genes associated with the cell cycle were differentially expressed. These were overexpressed in hMSCs cultured in FBS and were all associated with prolongation of the cell cycle. On the other hand, angiopoietin-like 4, which was upregulated in hMSCs expanded in AS, has been shown to inhibit apoptosis [31]. This may further contribute to the increased cumulative cell numbers in hMSC cultures supplemented with AS. Other genes overexpressed in hMSCs in FBS were genes associated with differentiation into osteoblasts, adipocytes, and chondrocytes. Together, this pattern of genes overexpressed in hMSCs grown in FBS suggests that these cells may have been taken through some steps along differentiation pathways that hMSCs in AS have yet to pass through. The overexpression of collagen genes may have contributed to another phenomenon observed in the cell cultures, namely the tendency to increased adherence to plastic by hMSCs in FBS compared with those in AS. Finally, some of the genes upregulated in hMSCs expanded in FBS compared with hMSCs in AS may be of particular significance to clinicians involved in protocols in which the immunosuppressive properties are being exploited. The immunosuppressive effect exerted by hMSCs has been mechanistically linked to production of prostaglandins [16]. Therefore, although AS may be preferable for regulatory reasons, the fact that several prostaglandin synthase genes were highly upregulated in FBS-supplemented hMSCs may tilt the balance in favor of FBS as serum supplement for these particular protocols.

To determine the stability of the transcriptome of hMSCs expanded with different serum supplements, we examined gene expression at passage 4 versus passage 10. Passage 4 coincided with a cumulative cell number of approximately 10⁸, whereas passage 10 occurred at approximately 10¹² cells. At this point, hMSCs in FBS were closer to replicative senescence than hMSCs in AS. This was clearly apparent from the list of genes differentially expressed in FBS-supplemented hMSCs between passages 4 and 10. Many genes encoding proteins with crucial roles in cell-cycle progression were strongly downregulated at passage 10. Topoisomerase II, for instance, is obligatory in the remodeling of chromatin during mitosis [37]. Induction of proliferative quiescence has been shown to be associated with a dramatic downregulation of this enzyme [38]. Cell division cycle 2 also codes for an enzyme that is central in the regulation of G2/M transition [39]. The combined observations of a reduced calculated cumulative cell count and differential expression of genes at passages 4 and 10 suggest that FBS induces premature replicative senescence by downregulation of genes involved in cell-cycle progression.

Many of the other differentially expressed genes at passage 10 were genes upregulated as a result of prolonged exposure to FBS. These were genes known to be associated with cells of the central nervous tissue (neurexin 3, neurotrypsin, neurotrophic tyrosine kinase receptor, myelin basic protein, nerve growth factor beta, ephrin type A receptor 5, and tetraspan 3), bone remodeling (osteoprotegrin), muscle cytoskeleton (myosin 1D), and vasculo-genesis (tumor endothelial marker 8). In addition, several growth factor–related genes and genes related to ECM were upregulated. Many of these same genes were upregulated or belong to a functional group that was also upregulated in hMSCs expanded in FBS versus AS at passage 4. Altogether, these observations suggest that FBS may be conducive toward differentiation.

In contrast, gene expression in hMSCs supplemented with AS was remarkably stable over many cell doublings. No genes were upregulated in any of the donors examined, and only a handful of genes, mostly associated with the cytoskeleton and ECM, were downregulated. Hence, should hMSC be needed in large quantities for cell therapeutic purposes, they can be expanded with AS as serum supplement without a risk of transcriptome instability.

In conclusion, we have shown that hMSCs may be expanded rapidly to very high cumulative cell counts in the presence of AS without growth factors. Compared with cells expanded in FBS, hMSCs in AS seemed less differentiated and remained transcriptionally more stable over time in culture. As AS should be universally acceptable by regulatory bodies, this expansion protocol may well be preferable over hMSCs in FBS for many protocols of cellular therapy.

Acknowledgements

We would like to acknowledge the skilled technical assistance of Siv Haugen Tunheim and Aileen Murdoch Larsen. This work was supported through the Norwegian Center for Stem Cell Research by the Research Council of Norway and Gidske og Peter Jacob Sørensens Foundation for the Promotion of Science.

Disclosures

The authors indicate no potential conflicts of interest.