Stem Cell Institute, University of Minnesota, Minneapolis, Minnesota, USA
Department of Medicine, Medical School, University of Minnesota, Minneapolis, Minnesota, USA
Professor of Medicine, Director, Stem Cell Institute, University of Minnesota, McGuire Translational Research Facility, 2001 6th St. SE, Mail Code 2873, Minneapolis, Minnesota 55455, USA. Telephone: 612-625-0602; Fax: 612-624-2436
We show that multipotent adult progenitor cells (MAPCs) can be derived from both postnatal and fetal swine bone marrow (BM). Although swine MAPC (swMAPC) cultures are initially mixed, cultures are phenotypically homogenous by 50 population doublings (PDs) and can be maintained as such for more than 100 PDs. swMAPCs are negative for CD44, CD45, and major histocompatibility complex (MHC) classes I and II; express octamer binding transcription factor 3a (Oct3a) mRNA and protein at levels close to those seen in human ESCs (hESCs); and have telomerase activity preventing telomere shortening even after 100 PDs. Using quantitative-reverse transcription-polymerase chain reaction (Q-RT-PCR), immunofluorescence, and functional assays, we demonstrate that swMAPCs differentiate into chondrocytes, adipocytes, osteoblasts, smooth muscle cells, endothelium, hepatocyte-like cells, and neuron-like cells. Consistent with what we have shown for human and rodent MAPCs, Q-RT-PCR demonstrated a significant upregulation of transcription factors and other lineage-specific transcripts in a time-dependent fashion similar to development. When swMAPCs were passaged for 3–6 passages at high density (2,000–8,000 cells per cm2), Oct3a mRNA levels were no longer detectable, cells acquired the phenotype of mesenchymal stem cells (CD44+, MHC class I+), and could differentiate into typical mesenchymal lineages (adipocytes, osteoblasts, and chondroblasts), but not endothelium, hepatocyte-like cells, or neuron-like cells. Even if cultures were subsequently replated at low density (approximately 100–500 cells per cm2) for >20 PDs, Oct3a was not re-expressed, nor were cells capable of differentiating to cells other than mesenchymal-type cells. This suggests that the phenotype and functional characteristics of swMAPCs may not be an in vitro culture phenomenon.
The main function of stem cells in postnatal life is to repair and regenerate the tissue in which they reside. Stem cells have the ability to self-renew and to differentiate into at least one mature cell type. Mesenchymal stem cells (MSCs) [1, –3], which can be isolated from bone marrow, adhere to plastic in vitro and expand in tissue culture, with a finite lifespan of 15–50 population doublings (PDs) [3, –5]. Under proper inductive stimuli, MSCs differentiate in vitro and in vivo into adipocytes [2, , , , , , , , –11], chondrocytes [3, 6, 8, , –11], osteoblasts [2, 6, 9, 12], and smooth and skeletal myoblasts [4, 8, 9].
Recently, we demonstrated that human, mouse, and rat postnatal bone marrow (BM) contains primitive progenitors termed multipotent adult progenitor cells (MAPCs) [9, 13, , , , , –19]. In contrast to another class of adherent BM-derived stem cells, MSCs, MAPCs are CD44- and major histocompatibility complex class I (MHC I)-negative. MAPCs can be expanded under defined low-serum conditions for more than 100 (human) or 400 (rat) PDs without telomere shortening or karyotypic abnormalities. MAPCs not only differentiate into mesenchymal cell types (osteoblasts, chondrocytes, adipocytes, and smooth and skeletal myoblasts) [9, 14, 15] but also into cells with phenotypic and functional characteristics of endothelial cells [9, 17], hepatocytes , and neural cells [13, , –16].
Although many studies have tested the effect of different stem cell populations in rodent models, results seen in rodents can often not be duplicated when applied to human cells, or in human clinical trials. For instance, there are extensive data from the hematopoietic stem cell literature showing that significant differences exist when comparing human and murine hematopoietic stem cells . Likewise, more recent studies demonstrating that grafting of BM-derived cells in a cardiac infarct model in mouse results in significant levels of engraftment and functional improvement, have not been replicated in human clinical trials [21, 22]. Such differences may be due to the remarkable anatomical and physiological differences between the mouse and human heart. In addition, the small size of the mouse heart makes it difficult to evaluate the functional consequences of cell transplantation with good spatial differentiation. These studies demonstrate therefore that a number of observations made in rodent models will need to be confirmed in larger animal models prior to applying to human clinical trials. Hence, the rationale for isolating MAPCs from swine, which would allow, at least in some disease models, results from rodent models to be confirmed and extended preclinically to perhaps more relevant models.
In this study, we demonstrate that MAPCs can be isolated from swine BM by methods similar to those used for mouse, rat, and human MAPCs isolation with minor modifications. These swine MAPCs (swMAPCs) have the same phenotype as human, rat, and mouse MAPCs; express the ESC-specific transcription factor Oct3a [23, –25]; and differentiate in vitro into cells with phenotypic and functional characteristics of tissues of all three germ layers.
Chemicals and Cytokines
MCDB-201, retinoic acid (RA), dimethyl sulfoxide, human epidermal growth factor (EGF), nonessential amino acids, β-glycerophosphate, ascorbic acid, bovine serum albumin (BSA), linoleic acid (LA)-BSA, and insulin-transferrin-selenium (ITS) were obtained from Sigma-Aldrich (St. Louis, http://www.sigmaaldrich.com). Low-glucose Dulbecco's modified Eagle's medium (DMEM) and α-minimal essential medium (α-MEM) were obtained from Invitrogen (Grand Island, NY, http://www.invitrogen.com). Penicillin and streptomycin were obtained from Mediatech Inc. (Herndon, VA, http://www.cellgro.com). Human platelet-derived growth factor (PDGF)-BB, human vascular endothelial growth factor (VEGF)-165, human hepatocyte growth factor (HGF), human basic fibroblast growth factor (bFGF), human fibroblast growth factor-4 (FGF-4), human bone morphogenetic protein 4 (BMP4), human transforming growth factor β1 (TGF-β1), human Noggin, human brain-derived neurotrophic factor (BDNF) and human glial cell-derived neurotrophic factor (GDNF) were obtained from R&D Systems Inc. (Minneapolis, http://www.rndsystems.com). Fetal bovine serum (FBS) was obtained from HyClone (Logan, UT, http://www.hyclone.com).
Monoclonal antibodies (mAbs) against swine neurofilament (NF)-200 (N5389; 1:400); swine glial fibrillary acidic protein (GFAP) (G3893; 1:400); and mAbs against Von Willebrand factor (VWF) (F3520; 1:200), α-smooth muscle (SM)-actin conjugated with Cy3 (C6198; 1:200), calponin (C2687; 1:200), and caldesmon (C4562; 1:200), which are cross-reactive with swine, were from Sigma-Aldrich. Polyclonal goat anti-Oct-3a (sc8628; N-19; 1:300), rabbit anti-hepatocyte nuclear factor (HNF)-1α (sc8986; 1:300), goat anti-tau (sc1995; 1:200), goat anti-vascular endothelial (VE)-cadherin (sc6458; 1:200) and rabbit anti-microtubule-associated protein 2 (anti-MAP2) (sc20172; 1:200), which are cross-reactive with swine, were from Santa Cruz Biotechnology (Santa Cruz, CA, http://www.scbt.com). mAbs against swine CD31 (RDI-RTCD31-3A; 1:200), swine MHC class I (PT85A; 1:200), swine MHC Class II (TH81A5; 1:200), swine CD45 (RDI-PIGCD45-E4; 1:200), and swine CD44 (RDI-PIGCD44abrt; 1:200) were from Research Diagnostics Inc (Flanders, NJ, http://www.scbt.com). Polyclonal rabbit anti-VWF (A0082; 1:200) was from DAKO (Carpinteria, CA, http://www.dako.com). mAbs against swine vascular cell adhesion molecule-1 (VCAM-1) (APG106; 1:200) were from Antigenix America (Huntington Station, NY, http://www.antigenix.com). Polyclonal goat anti-swine albumin (A100; 1:200), polyclonal goat anti-SM22 (ab10135; 1:200), and mAb against swine cytokeratin (CK) 18 (CA1138; 1:300) were from Bethyl Laboratories Inc. (Montgomery, TX, http://www.bethyl.com), Abcam (Cambridge, U.K., http://www.abcam.com) and USBiomax (Rockville, MD, http://www.usbiomax.com) respectively. Secondary anti-mouse/goat/rabbit Alexa Fluor 488 or 594 were from Molecular Probes (Grand Island, NY, http://probes.invitrogen.com)
Control Cell Populations
As no swine ESCs exist, the human ESC line H9 was used as a reference control for Oct3a expression in swine cells. Cell lysate was donated by Dr. Dan Kaufman (University of Minnesota Stem Cell Institute). Swine MSCs were isolated and characterized as described .
swMAPC Isolation, Purification, and Expansion
In initial cultures, we attempted to isolate MAPCs from BM aspirates immediately following harvesting, as we have described for human BM . However, we were unsuccessful (described in Results). Successful swMAPCs isolations were obtained using the following method. Limbs from fetal pigs (∼10 week gestational age) and postnatal pigs (age ∼40 days) were excised and incubated in a 1× phosphate-buffered saline (PBS) (Mediatech) solution containing 5% BSA for 48 hours at 4°C. BM was flushed from the limbs using a 5% BSA 1× PBS solution, and bone marrow mononuclear cells (BMMNCs) were isolated by Ficoll-HistoPaque density gradient centrifugation (Histopaq-1077; Sigma-Aldrich) . BMMNCs were plated on T150 flasks, coated with 10 ml of 10 ng/ml fibronectin (FN) (Sigma-Aldrich), at approximately 2 × 105 cells per cm2 in MAPC medium (60% DMEM-LG/40% MCDB-201 supplemented with 1× ITS, 0.5× LA-BSA, 0.1 mM ascorbic acid 2-phosphate, 100 U of penicillin, 1,000 U of streptomycin, and 2% prescreened FBS (screened on the basis of the ability to isolate and maintain high Oct-4 mouse, rat, and swine MAPCs) with 10 ng/ml each human PDGF-BB and EGF). Once initial colonies formed, cells were recovered by short-term trypsinization (1 minute; 0.25% Trypsin-EDTA; Mediatech) and replated at 100 cells per cm2 until small colonies could again be detected. After the second passage, cells were subcloned at ∼1 cell per well in 48-well plates in MAPC medium. Once clones began to proliferate, wells with only one clone were selected to be expanded, by detaching cells with 0.25% trypsin-EDTA at room temperature for 1 minute and replating at 500 cells per cm2. Cells were maintained for 3–4 days to reach approximately 4,000 cells per cm2 before they were replated again at 500 cells per cm2. All cultures were maintained at 5.5% CO2, normoxia, and 37°C. (More details are given in the supplemental online protocol.)
swMAPCs (5 × 105) were cultured in 1 ml of MAPC basal medium (MAPC expansion medium without serum, EGF, or PDGF) with 10 ng/ml TGF-β1 and 100 ng/ml BMP4 in the tip of a 15-ml conical tube and briefly spun to allow aggregation of the cells in micromass suspension culture. Medium was changed every 4 days. After 21 days, cultures were evaluated by quantitative reverse-transcription-polymerase chain reaction (Q-RT-PCR) for collagen type II and aggrecan transcripts and stained with Alcian Blue to demonstrate cartilage matrix production. A total of eight differentiations were observed on swMAPCs from three donors at 60–100 PDs.
swMAPCs were plated into six-well plates at 3 × 103 cells per cm2 in MAPC medium overnight. On the following day, the medium was replaced with fresh α-MEM containing 10% heat-inactivated FBS, 1% nonessential amino acids, 1% penicillin and streptomycin, 10 mM β-glycerophosphate, and 50 μM ascorbic acid 2-phosphate, with medium changes every 3–4 days. Differentiated swMAPCs were assayed for alkaline phophatase activity and mineral deposition by histochemical staining with the Sigma Kit 85 and Alizarin red methods, respectively, at day 14 . A total of five differentiations were done from swMAPCs from three donors at 60–100 PDs.
Adipocyte differentiation was induced using adipocyte differentiation kit (SCR020; Chemicon, Temecula, CA, http://www.chemicon.com) per the manufacturer's recommendations. Differentiated swMAPCs were evaluated by oil red O stain. A total of six differentiations were done from swMAPCs from three donors at 60–100 PDs.
Smooth Muscle Cell Differentiation.
swMAPCs were plated at 3 × 103 cells per cm2 in MAPC basal medium supplemented with 10 ng/ml PDGF and 5 ng/ml TGF-β1. During the differentiation course, medium was changed every 3–4 days. Smooth muscle cell (SMC) differentiation was evaluated by RT-PCR for calponin, α-SM actin, smoothelin, gata-6, and myocardin and immunofluorescence (IF) staining for calponin, α-SM actin, sm22, and caldesmon. A total of 24 differentiations were done from swMAPCs from three donors at 60–100 PDs.
Endothelial Cell Differentiation.
swMAPCs were plated at 5 × 104 cells per cm2 in MAPC basal medium with 100 ng/ml of VEGF-165 for 10 days. During the differentiation course, medium was changed every 3–4 days. Differentiation cultures were evaluated by Q-RT-PCR for VWF, CD31/Pecam, fms-like tyrosine kinase-1 (Flt-1), fetal liver kinase-1 (Flk-1), VE-cadherin, tyrosine kinase with Ig, and EGF homology domains 1 (Tie-1) and tyrosine kinase endothelial (Tek), every 3 days until day 10. Differentiated endothelial cells were stained for CD31, VWF, VE-cadherin, and VCAM-1 and evaluated for their ability to form tubes on ECMatrix and uptake acetylated low density lipoprotein (a-LDL). Briefly, tube formation was induced by plating MAPC-endothelial cells (MAPC-ECs) using the ECM625 angiogenesis assay (Chemicon) per the manufacturer's recommendations, and a-LDL uptake was performed by using Dil-Ac-LDL staining kit (Biomedical Technologies, Stoughton, MA, http://www.btiinc.com) per the manufacturer's recommendations. A total of 21 differentiations were done from swMAPCs from three donors at 60–100 PDs.
Hepatocyte differentiation was achieved by plating 5 × 104 cells per cm2 swMAPCs on 2% Matrigel-coated (BD354234; BD Biosciences, San Diego, http://www.bdbiosciences.com) plastic chamber slides in MAPC basal medium with 100 ng/ml FGF-4 and HGF for 12 days. During the differentiation course, medium was changed every 3 days. Differentiation cultures were evaluated by Q-RT-PCR for HNF-3β, HNF-1α, CK18 and CK19 albumin, and CYB2B6 every 3 days until day 12. Differentiated cells were evaluated by immunofluorescence microscopy for albumin, CK18 and HNF-1α protein expression. To assess the function of hepatocyte-like cells three assays were done (described below). A total of 25 differentiations were done from swMAPCs from three donors at 60–100 PDs.
swMAPCs were plated at 3 × 103 cells per cm2 on 10 ng/ml FN-coated coverslip placed inside six-well plates in MAPC medium overnight. The medium was then switched to MAPC basal medium with 100 ng/ml bFGF, 10 ng/ml Noggin, 20 μM retinoic acid for 28 days. After 14 days, 10 ng/ml BDNF and GDNF were also added. Half-changes of medium occurred every 7 days until day 28. Neural differentiation was evaluated via Q-RT-PCR for early neural transcript factors, Islet-1 transcription factor, orthodenticle homolog 2 (Otx-2), and paired box gene 6 (Pax-6), as well as neural cell adhesion molecule and the more mature neuronal marker MAP2, NF200, tau, and myelin basic protein (MBP) every 7 days. Cultures were also analyzed via immunofluorescence for NF200, MAP2, τ, and GFAP. A total of 34 differentiations were done from swMAPCs from three donors at 60–100 PDs.
Cells, plated at 500 cells per cm2 48 hours prior to harvesting, were subjected to 10 μl/ml colcemid incubation for 2–3 hours and collected with 0.25% Trypsin-EDTA followed by lysis with a hypotonic solution and fixation in alcohol. Metaphases were analyzed after Giemsa staining.
Telomere Length and Telomerase Activity Measurement
For the telomerase assay equal numbers cells were lysed in 1× 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid (CHAPS) buffer for 10 minutes on ice. Debris was pelleted at 13,000g for 10 minutes. Protein was quantified by the method of Bradford. One to two μg of protein was used in the telomere repeat amplification protocol (TRAP). The TRAP protocol was done according to the manufacturer's instructions (Chemicon). This protocol uses an enzyme-linked immunosorbent assay (ELISA)-based detection system to determine telomerase activity. Positive activity is defined as OD 450–690 reading >0.2 of test samples after subtracting heat-inactivated controls.
RNA Isolation and Q-RT-PCR
RNA was extracted from swMAPCs or swMAPC-derived differentiated progeny using the RNeasy RNA isolation kit (Qiagen, Valencia, CA, http://www1.qiagen.com). mRNA was purified from genomic DNA via Turbo DNase Set (Ambion, Austin, TX, http://www.ambion.com) and reverse-transcribed, and cDNA underwent 40 rounds of amplification (ABI PRISM 7700; PerkinElmer Life and Analytical Sciences, Boston, http://www.perkinelmer.com) under the following reaction conditions: 40 cycles of a two-step polymerase chain reaction (PCR) (95°C for 15 seconds, 60°C for 60 seconds) after initial denaturation (95°C for 10 minutes) with 2 μl of DNA solution, 1× TaqMan SYBR Green Universal Mix PCR reaction buffer. Primers used for amplification are listed in supplemental online Table 1. The mRNA levels were normalized using glyceraldehyde-3-phosphate dehydrogenase as a housekeeping gene and compared with levels in adult or fetal swine tissues. Primer specificity was confirmed by product sequencing.
For staining of cytoskeletal proteins, cells were fixed with −20°C methanol (Sigma-Aldrich) for 2 minutes and treated with 0.1% Triton X-100 (Sigma-Aldrich) for 15 minutes. For other intracellular molecules, cells were fixed with 4% paraformaldehyde (Sigma-Aldrich) at room temperature for 15 minutes and treated with 0.1% Triton X-100 for 15 minutes. For cell surface marker, cells were fixed with 4% paraformaldehyde at room temperature for 15 minutes. For nuclear markers, cells were fixed with 4% paraformaldehyde at room temperature for 15 minutes and treated with 0.5% Triton X-100 for 30 minutes. Fixed cells were blocked by 0.4% fish gelatin (Sigma-Aldrich) in PBS for 30 minutes and incubated sequentially with primary antibodies for 1 hour at room temperature and fluorescence secondary antibodies or secondary biotinylated antibodies for 45 minutes at room temperature. Nuclei were counterstained with 4′,6′-diamino-2-phenylindole (D3571; Molecular Probes). For immunochemistry, samples were incubated for 30 minutes in 0.3% H2O2 (Sigma-Aldrich) after secondary biotinylated antibodies, and then for 30 minutes with Vectastain ABC reagent (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com) and 5 minutes with peroxidase substrate solution (Vector Laboratories) at room temperature. Between each step, fixed cells were washed with PBS three times.
To evaluate the heterogeneity of differentiation approximately 100 cells in four visual fields per slide were randomly selected for approximately six slides per cell line between 60 and 100 PDs; positive cells were counted and are presented as percentage versus total cell number.
Fluorescence-Activated Cell Sorting Analysis
For fluorescence-activated cell sorting, swMAPCs cells were detached by 0.25% trypsin-EDTA and suspended in 100 μl of PBS. Suspended cells were sequentially incubated with primary Abs and fluorescein isothiocyanate secondary Abs for 30 minutes at 4°C and resuspended in 100 μl of PBS and analyzed with a FACSCalibur machine (Becton Dickinson). Between each step, cells were washed with PBS.
Vascular Tube Formation.
Tube formation was induced by plating MAPC-ECs using the ECM625 angiogenesis assay (Chemicon) per the manufacturer's recommendations.
The Dil-Ac-LDL staining kit was purchased from Biomedical Technologies. The assay was performed per the manufacturer's recommendations. Briefly, differentiated swMAPCs were incubated with endothelium differentiation medium containing 10 μg/ml Dil-Ac-LDL for 4 hours at 37°C and rinsed twice by Dil-Ac-LDL free endothelium medium. LDL uptake was visualized via fluorescence microscopy.
Swine albumin concentrations were determined using an ELISA. Concentrations of albumin were determined by generating standard curves from known concentrations of swine albumin. Peroxidase-conjugated and affinity-purified anti-swine albumin and reference swine albumin were from Bethyl Laboratories (E100-110). To verify specificity of results, conditioned medium from endothelial differentiations and unconditioned hepatocyte differentiation medium were used. A total of six experiments were done from swMAPC hepatocytes from two donors at 60–100 PDs.
Urea secretion was assessed by colorimetric assay (DIUR-500 BioAssay Systems) per the manufacturer's instructions. Known concentrations of urea were used to generate standard curves. The assay detects urea directly by using substrates that specifically bind urea. Urea reactions were done in 96-well plates, and urea concentrations were read using a plate reader. Conditioned medium from endothelial differentiations and unconditioned hepatocyte differentiation medium were used as negative controls. A total of six experiments were done from swMAPC hepatocytes from two donors at 60–100 PDs.
Periodic Acid-Schiff Staining.
Slides were oxidized for 5 minutes in 1% periodic acid-Schiff (PAS) (Sigma-Aldrich) and rinsed several times with double-distilled H2O (ddH2O). Samples were incubated with Schiff's reagent for 15 minutes, rinsed several times with ddH2O, immediately counterstained with hematoxylin for 1 minute, and washed several times with ddH2O.
Isolation of swMAPCs
In initial studies (n = 20 postnatal swine BM samples), we attempted to isolate MAPCs from BM aspirates immediately following harvesting, as we have described for human MAPCs . However, we were unsuccessful. After 2–3 days colonies of large flat cells appeared, which were CD44+, CD45−, MHC-I+, MHC-II−, CD90+, consistent with an MSC phenotype . Such cells could only be passaged for ∼30 PDs, and they differentiated as classic MSCs only in mesodermal cell lineages . Initially we thought that this could be due to the age of the donor animals, as rodent and human MAPCs are more frequent in marrow of younger animals and humans  . We therefore tested whether MAPCs could be isolated from the BM of younger, that is, prenatal, animals. From the first attempt, cells with morphological and phenotypic features of MAPCs could be isolated from BM harvested ∼48 hours after the death of a fetal animal. Interestingly, in contrast with cultures initiated with BM from postnatal animals, we did not see cell colonies appear until approximately 7 days after initiation of the culture. These cells were morphological smaller and did not have the typical flat and elongated MSC appearance. Rather, they were small cells, like mouse and rat MAPCs [14, 15]. We hypothesized that a number of differences in the samples used in the initial 20 attempts from postnatal animals and the successful attempt from fetal BM might contribute to the different outcome, including the age of the animal, time of plating after the death of the animal (immediately after harvesting in postnatal samples and 48 hours after the death of the animal for the fetal sample), and method of collecting the sample (aspiration in postnatal animals and flushing from the bone in fetal animals). To discriminate among these possibilities, we next evaluated the ability to isolate MAPCs from fetal swine BM immediately after the death of the animal or 48 hours after death (n = 8 experiments). As we had seen during our initial attempts at isolating MAPCs, when BM obtained immediately after the death of the fetal animal was used to initiate cultures, colonies of relatively large and flat cells appeared by day 3. By day 7, the larger cells grew to near confluence, and we could not detect colonies of the smaller cells. On the other hand, when BM collected 48 hours after the death of the fetal animal was used, smaller cells appeared by day 7. These studies suggested thus that MSC-like cells may proliferate faster than MAPC-like cells and outcompete the latter when both of them are plated together. The fact that BM harvested 48 hours after the death of the animal did not lead to outgrowth of the larger and flatter cells suggests that MSC-like cells may be selectively lost when swine BM is stored at 4°C. As there are no definitive cell surface determinants that can discriminate between the two cell populations, it is not possible to definitively prove this hypothesis. We re-evaluated the ability to isolate MAPCs from postnatal swine BM and found that keeping the bone for 36–48 hours at 4°C also decreased the frequency of MSC-like colonies appearing early after plating and allowed for the outgrowth of cells with MAPC morphology, phenotype, and functional characteristics (described below).
The second difference between the initial experiments performed with postnatal swine BM and the isolation of MAPCs from fetal swine BM was the manner of harvesting (aspiration in postnatal swine and flushing in fetal swine). We also compared in three experiments the two methods of harvesting cells prior to MAPC culture. These studies were done by maintaining the swine bone at 4°C for 36–48 hours, followed by flushing or aspirating the BM. BM obtained by flushing generated more readily MAPCs than when BM was aspirated, possibly indicating that swMAPCs are attached more strongly to bone spiculae, similar to other stem cells within BM [28, 29], and are dislodged more efficiently when the marrow cavity is flushed than when the BM is aspirated.
In subsequent studies, we demonstrated that homogenous swMAPCs populations could be culture-isolated and expanded for more than 100 PDs from three of four fetal pig BM samples and one of three postnatal pig BM samples harvested by flushing from the bone 48 hours after the death of the animal. Hence, aside from the manner in which BM cells are obtained to initiate MAPC cultures, age may also be a factor in the ability to isolate MAPCs.
Characterization of swMAPCs
Culture isolation and subsequent expansion involved initial plating of the total BMMNC for 2–3 passages, as indicated in Materials and Methods, followed by single-cell subcloning. In all cases, cells were plated at ∼1 cell per well in four 48-well plates. Of these, approximately 10 wells contained a single colony of cells with the typical small MAPC morphology (supplemental online Table 2). Approximately 80% of clones could be expanded for >30 PDs. These clones were then evaluated by Q-RT-PCR for Oct3a transcript levels, as MAPCs from other species express Oct3a. We used primers designed against human ESC-specific Oct3a, which also identifies swine Oct3a (determined using testicular swine tissue and confirmed by sequencing). Of these individual clonal populations, 64% had Oct3a transcript levels between 1% and 50% of hESCs, 21% had levels between 0.1% and 1% of hESCs, and 15% had levels <0.1% of hESCs. From each individual BM, one clone with levels of Oct3a mRNA >20% of hESCs was maintained in culture beyond 60 PDs. All other clones were cryopreserved and not evaluated further. For the four clones, one each from three fetal and one postnatal swine BM isolations that were maintained in culture, levels of Oct3a were between 5% and 20% of those identified in hESCs (Fig. 1E) and remained stable for >90 PDs. The expression of Oct3a in swMAPCs was confirmed by immunocytochemistry (Fig. 1D). It should be noted that Oct3a mRNA could not be detected in swine MSCs (Fig. 1E). For all clonal MAPC populations, cell doubling time was 24 hours for the initial 30–40 PDs and 36–48 hours when cultures reached >40 PDs (Fig. 1A). As shown in Figure 1B and 1C, swMAPCs are round or triangular, lightly adherent, less than 10 μm in diameter, and exhibit a very high nucleus-to-cytoplasm ratio. swMAPCs have an instinctive ability to separate from each other following cell divisions, even when cells reach higher density (∼4,000 cells per cm2, just before they are passed). When cultures were allowed to grow to very high densities (>6,000 cells per cm2), proliferation slowed down, demonstrating contact inhibition. Phenotypic analysis after 50 PDs indicated a homogenous population of cells that is negative for CD44, CD45, MHC class I, and MHC class II (Fig. 1J), whereas cells analyzed prior to 50 PDs were negative for CD45 and MHC class II but were mixed for expression of MHC class I and CD44 (Fig. 1I).
Telomere lengths were evaluated at ∼30, ∼60, and ∼90 PDs in the three fetal swMAPC lines and one postnatal swMAPC line. Telomere lengths did not differ between MAPCs isolated from fetal or postnatal swine, and telomeres did not shorten following extensive passaging (Fig. 1G). Significant levels of telomerase activity could be measured in the two swMAPCs populations tested, again irrespective of the age of the donor animal (Fig. 1H).
Cytogenetic analysis of the cultured cells at multiple PDs showed that cells were diploid and contained no cytogenetic abnormalities (based on 20 cells in metaphase analyzed per sample at <90 PDs; Fig. 1F), although some cells evaluated at PDs beyond 90 showed polyclonal karyotypic abnormalities (supplemental online Table 3).
When established populations of swMAPCs were replated at high density (>2,000 cells per cm2), and passaged every 4 days when they reached cell densities of >8,000 cells per cm2 for 3–6 passages, they became morphologically larger (Fig. 2A). This was associated with an acquisition of CD44 and MHC class I antigens on the cell membrane (Fig. 2C), and a loss of Oct3a expression determined by Q-RT-PCR and immunohistochemistry (data not shown). When cultures were subsequently replated at low cell densities (100–500 cells per cm2) for >20 PDs, they did not reacquire the typical small MAPC morphology, and they remained CD44- and MHC class I-positive and Oct3a-low (<0.001% of hESCs). Moreover, we found that telomeres shortened in cells allowed to grow at high density and replated for 20 and 40 PDs at low density (Fig. 2B).
In Vitro Differentiation of swMAPCs
The in vitro differentiation capabilities of swMAPCs to mesoderm, neuroectoderm, and endoderm were evaluated by the addition of cytokines on the basis of previous differentiation studies of ESCs [30, –32] and MAPC differentiation [9, 13, –15, 17, –19]. Studies were done using two fetal MAPC lines and one postnatal MAPC line, and each differentiation was performed multiple times at population doublings 60–100.
Not surprisingly, we demonstrate that swMAPCs can differentiate into osteoblasts, chondroblasts, adipocytes, and SMCs, which typical MSCs can do. When swMAPCs were treated with osteogenesis differentiation medium, a significant increase in alkaline phosphates activity (Fig. 3A) and mineral deposition by Alizarin red staining (Fig. 3B) were seen at day 15. In response to TGF-β1 and BMP4, swMAPCs take on an early chondrocyte phenotype. Differentiating cells expressed aggrecan and collagen mRNA (Fig. 3F) and generated Alcian Blue staining matrix (Fig. 3D, 3E). swMAPCs can also differentiate into adipocytes, as demonstrated by the accumulation of lipid vacuoles that stained with oil red O (Fig. 3C). swMAPCs differentiated in response to PDGF-BB and TGF-β1 to a SMC-like phenotype (Fig. 4). Twelve days after culture, they expressed calponin, α-SM actin, smoothelin, gata-6, and myocardin, as determined by RT-PCR. IF staining confirmed that swMAPCs-derived SMCs expressed calponin, α-SM actin, sm22, and caldesmon in 92% ± 5%, 97% ± 2%, 93% ± 3%, and 93% ± 2% of differentiated swMAPCs, respectively.
In contrast to what is commonly seen for MSCs and consistent with what we have reported for human and rodent MAPCs, swMAPCs differentiated into endothelial cells, hepatocyte-like cells, and neuroectoderm-like cells. When cultured with 100 ng/ml VEGF165, swMAPCs differentiated into ECs expressing VWF, CD31, Flt-1, Flk-1, Tie-1, Tek, and VE-cadherin in a time-dependent fashion, as determined by Q-RT-PCR. This was confirmed at the protein level for CD31 (88% ± 7% of cells staining positive), VWF (95% ± 3% of cells staining positive), VCAM-1 (84% ± 6% of cells staining positive), and VE-cadherin (74% ± 11% of cells staining positive), results similar to those published for human MAPCs  (untreated controls are shown in supplemental online Fig. 1). By day 10, swMAPC-endothelial cells also formed vascular tubes in an in vitro angiogenesis assay and could uptake Dil-Ac-LDL (Fig. 5), whereas undifferentiated MAPCs did not (data not shown).
When swMAPCs were cultured with 100 ng/ml FGF-4 and HGF, they differentiated into cells with hepatocyte phenotype and function. Using Q-RT-PCR, we demonstrated that swMAPCs express in a time-dependent manner hepatocyte-specific transcription factors, as well as transcripts for hepatocyte-specific structural and functional proteins. The expression of HNF-1α, albumin, and CK18 was confirmed at the protein level, with 70% ± 9%, 90% ± 5%, and 89% ± 4% of cells, respectively, staining positive; these results are similar to those published for human MAPCs  (untreated controls are shown in supplemental online Fig. 1). Aside from phenotypic characteristics of hepatocytes, swMAPC-derived cells also had functional characteristics of hepatocytes, including secretion of albumin and urea and storage of glycogen (Fig. 6).
swMAPCs could also be induced down a neuroectodermal pathway. As we showed for endothelium and hepatocyte, we demonstrate sequential activation of early neural commitment transcripts by d14 and more mature transcripts, including NF200 and MBP, by d28 following culture of swMAPCs with noggin, RA, and bFGF, followed by BDNF and GDNF. Again we confirmed neuroectoderm-like differentiation by IF with antibodies against NF200 (91% ± 3%), τ (81% ± 8%), GFAP (5% ± 2%), and MAP2 (86% ± 7%) (Fig. 7) (untreated controls are shown in supplemental online Fig. 1).
When the differentiation ability of swMAPCs allowed to grow at high density for 3–6 passages or swMAPCs grown at high density for 3–6 passages and then replated at 100–500 cells per cm2 for an additional 20 PDs was tested, we demonstrated that although SMC differentiation could be induced, they no longer differentiated to endothelium, hepatocyte-like cells, or neuroectoderm-like cells (data not shown). These observations are consistent with the MSC-like phenotype and the absence of Oct3a (Fig. 2).
When swMAPCs are cultured at low densities (250–500 cells per cm2) they express high levels of Oct3a transcript and protein, maintain telomere lengths as a result of significant telomerase activity, and are capable of trilineage differentiation. We noted stable karyotypes for many population doublings, even though up to 20% of metaphases became abnormal by PD 120. Such late acquisition of cytogenetic abnormalities is also seen for murine ESCs  and human ESCs  and may be inevitable, as cell expansion requires many sequential symmetrical self-renewal cell divisions.
As for human MAPCs and unlike mouse or rat MAPCs, swMAPCs can be isolated without the addition of leukemia inhibitory factor, a cytokine needed for murine ESCs but not hESCs [35, 36]. swMAPCs also differentiate into mesodermal, endodermal, or neuroectoderm cell types, as shown by Q-RT-PCR. Importantly, our studies demonstrated that differentiation of swMAPCs to endothelium, liver, and neural-like cells occurred as would be expected during development. Transcripts for genes typical for commitment to a given lineage can be detected early (such as HNF3-β for hepatocytes and Otx-2 and Islet-1 for neural cells) and are then downregulated, whereas genes expressed in more mature tissues, such as VWF for endothelium, albumin for hepatocytes, and NF200 for neural cells, are expressed 1 or more weeks after induction of differentiation. Double-label IF microscopy verified the findings of Q-RT-PCR of early and late lineage-specific markers. Furthermore, functional tests on the swMAPCs that differentiated into endothelial cells and hepatocyte-like cells demonstrated that cells also acquire functional features of the differentiated cell type.
Although methods to induce differentiation were similar to what we had used for human and rodent MAPCs, some adjustments were needed. For hepatic and endothelial differentiation, we needed to use 10-fold more HGF/FGF4 and VEGF, respectively, to obtain differentiation, which may be due to the fact that human-specific, not swine-specific, cytokines were used. Conditions to induce neural differentiation differed significantly from those described for murine and rat MAPCs . Consistent with the notion that murine and human ESC differentiation to neuroectoderm requires that signaling via BMPs and TGFs is blocked [37, –39] and that retinoic acid induces neural differentiation in these systems [40, 41], we needed to add noggin, which is a TGF-β and BMP4 inhibitor [38, 42], and retinoic acid to commit swMAPCs to a neural fate. This was done in combination with bFGF, a known NCS mitogen . This yielded maximal activation of Islet-1 and Otx-2 expression at 10–14 days. Subsequent maturation was obtained by addition of BDNF and GDNF, which induced maximal expression of NF200 and MBP by day 28.
Consistent with MAPCs from rodents and humans, when swMAPCs were cultured at a higher density (>2,000 cells per cm2) for even a short period of time (3–6 passages), the morphology of the cells began to change. This was associated with the acquisition of CD44 and MHC-I on the cell surface, loss of Oct3a transcript levels and protein, and loss of differentiation ability, with only differentiation to typical mesenchymal cell types, but not endothelium, hepatocyte-like cells, or neural-like cells. In numerous studies, when cells maintained for 3–6 passages at high cell density, were replated at 100–500 cells per cm2 for 20 PDs, Oct3a was not re-expressed, and cells maintained an MSC morphology, cell surface phenotype, and differentiation ability. This strongly suggests, but does not prove, that the MAPC phenotype is not an in vitro phenomenon, since one would expect that the phenotype should be reinduced when cultures conditions favorable to MAPC induction are re-established.
The mechanism underlying the loss of MAPC phenotype when cells are cultured at higher density is currently not yet clear. This phenomenon is true to a greater or lesser extent for rodent, swine, and human MAPCs. Possibilities that are being evaluated are direct cell-cell-mediated interactions; secretion of differentiation-inducing cytokines by MAPCs, which may accumulate faster when cells are maintained at higher densities; or both. The establishment of multiple swMAPCs cell lines may now make it possible to re-evaluate observations made with rodent MAPCs in small animal models in a large animal model such as swine.
The authors indicate no potential conflicts of interest.
This work was supported in part by U.S. Public Health Service Grant HL67828 and a research grant from Athersys Inc. (Cleveland, OH).