Clonal multipotency and effect of long-term in vitro expansion on differentiation potential of human hair follicle derived mesenchymal stem cells

hHF-MSCs maintain their proliferative capacity over long culture period

hHF-MSCs were isolated from human scalp hair follicles and expanded in DMEM supplemented with 10% FBS and 1 ng/mL bFGF. The cells exhibited small, elongated, fibroblast-like morphology (Fig. 1A), had normal karyotype (Fig. 1B) and expressed surface markers characteristic of mesenchymal stem cells (Fig. S1). To examine their proliferation capacity, hHF-MSCs were seeded at 3,000 cells/cm² and passaged every fourth day. In each passage the total cells were counted and the number of divisions was calculated. The number of population doublings remained relatively constant for 11–12 passages (approximately 2.5 to 3.5 divisions per passage) and declined sharply thereafter, reaching less than one doubling over a 4-day period by passage 14 (Fig. 1C). At the same time the cells appeared larger and flatter possibly as a result of cellular senescence. Indeed, staining for the senescence associated β-galactosidase (SA β-Gal) revealed that the fraction of SAβ-gal positive cells increased over time in culture (Fig. 1D, E). While, only a small fraction of cells (<5%) appeared SA β-Gal positive on passages 5 and 8, this number increased to 91±7% by passage 14 (n=3) (Fig 1E).

Kinetics of gene expression during differentiation of hHF-MSCs

Next we evaluated the multi-lineage differentiation potential of hHF-MSCs toward myogenic, osteogenic, adipogenic and chondrogenic lineages using quantitative real time PCR. Myogenic differentiation was induced by treatment with 2 ng/mL TGF-β1 [17, 19, 20] and total mRNA was isolated daily for a period of 6 days. The expression level of three smooth muscle specific genes, smooth muscle α-actin (αSMA), SM22 and myocardin [21, 22], was measured as a function of time. As illustrated in Fig. 2A all three genes were significantly up-regulated upon TGF-β1 treatment as compared to non-treated cells (p<0.05). More specifically, SMA exhibited the highest expression on the first two days of treatment (day 2: 10±1.8 fold, n=3, p<0.05 as compared to day 0) and decreased thereafter to 4.4±0.78 fold on day 6 (n=3, p<0.05 as compared to day 2). Similar to αSMA the level of SM22 increased on day 2 (day2: 11.2±3.7 fold, n=3, p<0.01 as compared to day 0) and decreased significantly on day 6 (day 6: 5.9 ±0.5 fold, n=3, p<0.05 as compared to day 2). Interestingly, expression of myocardin increased dramatically and showed a periodic kinetic pattern, demonstrating maxima on days 3 and 5 (day 3: 731±172 fold, n=3; and day 5: 932±204, n=3) and minima on days 2, 4 and 6 (day 2: 10.4±3.8 fold, n=3; day 4: 220±11, n=3; and day 6: 495±181, n=3). It is worth noting that medium was changed every 2 days, suggesting a transient increase in myocardin expression concomitant with TGF-β1 stimulation.

During adipogenic induction hHF-MSCs rounded up and at later stages, started accumulating oil droplets signifying terminal adipogenic differentiation. To monitor the transcriptional profile of adipogenesis, RNA was isolated at days 2, 4, 8, 14, 16 and 21 and the relative expression levels of PPARγ, lipoprotein lipase and leptin were measured by RT-PCR (Fig. 2B). Statistical analysis demonstrated significant upregulation of lineage specific markers during adipogenesis (p<0.001). More specifically, the regulatory transcription factor PPARγ increased continuously reaching expression levels of 7.2±0.8 fold on day 21 (n=3, p<0.001 as compared to day 0). Early marker, lipoprotein lipase [23] and late marker, leptin [24] are secreted by adipocytes and their expression is believed to be highly up-regulated during adipogenesis [25]. Both markers showed a sharp increase after day 8 reaching the highest levels of expression on day 21. In particular, the level of leptin increased to 231±26 fold on day 21 and lipoprotein lipase increased dramatically to 45,600±2,114 fold as compared to non-differentiated cells (n=3).

hHF-MSCs cultured under osteogenic conditions increased the expression of osteogenic genes such as RUNX-2, alkaline phosphatase, and osteonectin (p<0.001) (Fig. 2C). RUNX-2, a key transcription factor for osteogenesis [26], showed a steady expression throughout the differentiation period reaching 3.9±0.8 fold on day 21 (n=3, p<0.05 as compared to day 0). Osteonectin increased to a similar level reaching maximum on day 8 (3.09 ±0.18 fold, n=3, p<0.05 as compared to day 21). On the other hand, alkaline phosphatase was highly induced throughout the differentiation period, reaching maximum level on day 21 (301±70 fold, n=3, p<0.05 as compared to day 0).

Chondrogenic differentiation conditions induced expression of three lineage specific genes, namely SOX-9, aggrecan and collagen type X (p<0.001) (Fig. 2D). In 2D cultures, mesenchymal condensation occurred rapidly and became apparent under the microscope by day 4. The transcription factor SOX-9 that regulates chondrogenic differentiation [27] was up-regulated early on and was expressed at 159±31fold on day 21 (n=3, p<0.05 as compared to day 0). Aggrecan, a mature marker of chrondrogenesis [28], was expressed at later stages of differentiation reaching maximum levels on day 16 (day 16: 10,171±2,166 fold, n=6, p<0.05 as compared to day 14) and remained constant thereafter (day 21: 8,314±3,454 fold, n=3, p>0.05 as compared to day 16). Another chondrogenic marker, collagen type X also increased upon chondrogenic induction (day 4: 23.4±4 fold, n=3, p<0.05 as compared to day 0) and continued to increase at later times (day 21: 60.1 ±20.3 fold, n=3, p<0.05 as compared to day 4).

Long-term in vitro expansion reduces the differentiation potential of hHF-MSCs

Next we examined the effect of prolonged in vitro expansion on the differentiation potential of hHF-MSCs. To this end, hHF-MSCs were differentiated towards the osteogenic, adipogenic, chondrogenic or myogenic lineage at different passages (passage 5, 8 and 11). The degree of differentiation was assessed by lineage specific functional assays, namely Oil Red O stain for adipogenesis, Von Kossa stain for osteogenesis, Alcian Blue stain for chondrogenesis and α-SMA expression and fibrin gel compaction for myogenesis.

As shown in Fig. 3A hHF-MSCs that were differentiated to smooth muscle cells (SMC) showed increased expression of α SMA, calponin and myosin heavy chain (MHC). hHF-MSCs that were cultured in proliferation medium (PM) showed low levels and diffuse staining of all three proteins. On the other hand in differentiation medium (DM), all three proteins showed robust expression and filamentous organization that is characteristic of SMC phenotype. To determine the effect of in vitro expansion on myogenesis we performed two independent experiments. First, we measured the fraction of cells expressing α SMA at different passages using flow cytometry. As shown in Fig. 3B the fraction of α SMA positive cells was 8–15 times higher in myogenic DM as compared to PM for all passages. Interestingly, the fraction of α SMA+ cells was higher at passage 8 (P5: 67.5 ±2% vs. P8: 83.7 ±3%; n=3, p<0.001) and decreased at passage 11 (P11: 51.4 ±3.2; n=3, p<0.001 as compared to P8). Second, since the quintessential attribute of SMC is their ability to generate force, we also measured fibrin gel compaction by hHF-MSCs as a function of time in culture. As shown in Fig. 3C, prolonged in vitro culture had a small but statistically significant effect on gel compaction. After 96hr, P5 hHF-MSCs decreased the area of fibrin hydrogels to 7.7 ±1% of the initial area compared to P11 hHF-MSCs which compacted the hydrogels to 19±2% of the initial area (n=3; p<0.05) suggesting that hHF-MSCs maintain their ability to contract even at late passages (Fig 3C). Taken together these results suggest that myogenic differentiation potential and contraction of hHF-MSCs appear to be largely unaffected by in-vitro expansion up to passage 11.

Osteogenesis showed similar trend with myogenesis as evidenced by Von Kossa staining, which is the standard method of identifying matrix mineralization. Mineralization was found to be the highest for P8 cells and then decreased significantly by P11 (Fig 4A). Von Kossa positive area (quantified using ImageJ) showed that 85±7% area was mineralized in P8 hHF-MSCs compared to 36±7% (p=1.8×10⁻⁷) and 17±4% (p=1.3×10⁻⁸) in P5 and P11 hHF-MSCs, respectively (Fig 4A). Chondrogenesis was examined through alcian blue staining that positively stains sulfated glycosaminoglycans (GAG) (blue color). Measurements of the area that was stained blue revealed that chondrogenic potential decreased gradually over time, reaching approximately half of its initial value by P11 (42±6% blue area) (Fig 4B). GAG deposition was highest in P5 hHF-MSCs (74±7% blue area) compared to P8 hHF-MSCs (61±11% blue area, p<0.05) (Fig 4B). In contrast, adipogenesis showed dramatic reduction between P5 and P11 as demonstrated by Oil Red O staining and oil droplet quantitation (15±3% oil droplet area in P5 compared to 1±0.6% in P11, p=1.6×10⁻⁶) (Fig. 4C). These results demonstrate that the adipogenic and osteogenic differentiation potential of hHF-MSCs were more sensitive to prolonged in vitro culture compared to myogenic and chondrogenic potential.

Clonal populations of hHF-MSCs exhibited multipotent differentiation potential

Although these results demonstrated the multi-lineage differentiation potential of hHF-MSCs, it was not clear whether hHF-MSCs were true multipotent stem cells or a collection of various progenitors each with unipotent differentiation ability. To address this issue, we plated hHF-MSCs as single cells in each well of five 96-well plates. Wells containing more than one cell were discarded, whereas single cells were allowed to expand. Thirty seven clones were selected and evaluated for adipogenic, osteogenic and chondrogenic differentiation potential. (Since the vast majority of hHF-MSCs could differentiate to SMCs (>80% of P8 cells were α SMA+) the clones were not examined for myogenic potential. These results are summarized in Table 1 and representative images for tri-potent, bi-potent and uni-potent clones are shown in Fig 5.

Interestingly, 13 out of 37 clones (~35%) demonstrated tri-lineage differentiation potential and 14 out of 37 clones (~38%) exhibited bi-potent differentiation to chondrogenic and osteogenic lineages. Seven clones (~19%) showed unipotent characteristics, whereas only three clones (~8%) failed to differentiate to any of the three lineages. It is worth noting that these experiments were carried out with passage 4 cells that were expanded from single cells, and therefore, underwent approximately 20 population doublings (corresponding to passage 10–11) before the differentiation potential was evaluated. Therefore, our results demonstrate that a significant fraction of hHF-MSCs retained multipotency even after prolonged culture times.

Isolation and cultivation of human Hair Follicle Mesenchymal Stem Cells (hHF-MSCs)

Isolation of hHF-MSCs was performed as described previously [16, 17]. Briefly human scalp skin tissues were obtained from the Cooperative Human Tissue Network (CHTN, Philadelphia, PA). Small (2mm × 4mm) pieces of skin were digested with 0.5% Collagenase Type I (Invitrogen, Carlsbad, CA) in DMEM (Dulbecco’s Modified Eagle Medium, GIBCO, Grand Island, NY) at 37ºC with occasional agitation for 3–4 hr. Single hair follicles were unplugged, washed extensively and seeded each in a well of a 24-well plate (BD Biosciences, San Jose, CA) with 500 μL of DMEM supplemented with 10%FBS (Fetal Bovine Serum; GIBCO). Over the next 7–10 days, cells migrated out onto the culture plate and proliferated. Only wells containing cells originating from the dermal sheath or papilla (fibroblast-like cells) were used in this study, whereas wells containing epithelial cells (keratinocytes) were discarded. Wells containing cells far from the hair follicle were also discarded to ensure that cells originated from the follicle and not from remnants of dermal tissue that might have been carried over accidentally. Expanded cells were cultured in proliferation media (DMEM supplemented with 10% MSC FBS (GIBCO) and 1 ng/mL bFGF, (BD Biosciences)) and the medium was replenished every other day unless otherwise indicated.

For determining the proliferation potential of hHF-MSCs, 3000 cells/cm² were seeded in triplicates at the indicated passage number. On the fourth day, cells were counted and the number of population doublings was calculated assuming geometric growth. To identify senescent cells in culture, we performed senescence associated β-Galactosidase (SA-β-Gal) staining (Cell Signaling, Danvers, MA) according to manufacturer’s instructions and the ratio of total peri-nuclear blue cells to total number cells was determined in 10 images per passage (each image contains approximately 50 cells).

Adipogenic, Chondrogenic, Myogenic and Osteogenic differentiation of hHF-MSCs

To determine the differentiation capacity of hHF-MSCs towards the myogenic, osteogenic, chondrogenic or adipogenic lineage, 6×10³ cells/cm² were seeded in tissue culture plates and cultured in proliferation medium until they reached confluence. The next day, the proliferation medium was replaced with the appropriate differentiation medium as indicated. Thereafter, the differentiation medium was replenished every two days for myogenic lineage and every 3 days for adipogenic, chondrogenic and osteogenic lineage.

For adipogenesis, hHF-MSCs were cultured in DMEM supplemented with 10% MSC FBS (Invitrogen), 0.5 mM of isobutyl-methylxanthine (Sigma, St Louis, MO), 1 μM of dexamethasone (Sigma), 10 μM of insulin (Sigma) and 200 μM of indomethacin (MP Biomedicals, Solon, OH) for 14 days. For chondrogenesis, differentiation was performed on cell culture dishes as previously described [44]. Briefly, hHF-MSCs were cultured in DMEM supplemented with 10% MSC FBS, 6.25 μg/ml insulin, 10 ng/ml TGF-β 1 (US Biological, Swampscott, MA) and 50 nM of ascorbate-2-phosphate (Sigma) for 21 days. For myogenic induction, hHF-MSCs were cultured in DMEM supplemented with 10% MSC FBS and 2 ng/mL TGF-β1(US biological) for 6 days. For osteogenesis, hHF-MSCs were cultured in DMEM supplemented with 10% MSC FBS, 0.1 μM dexamethasone, 50 μM ascorbate-2-phosphate and 10mM β-glycerolphosphate (Alfa Aesar, Ward Hill, MA) for 28 days.

Oil Red O staining

After two weeks in culture in adipogenic medium cells were fixed with 10% formalin at room temperature for 60 min and then incubated with 60% (v/v) isopropanol for 5 min at room temperature. Then hHF-MSCs were dried and incubated with 0.6% (w/v) filtered Oil red O in 60% isopropanol/40% water (Alfa Aesar) at room temperature for 10 min and washed four times with water to remove any unbound dye. The cells were imaged using a Zeiss Axio Observer (Zeiss, Thornwood, NY) inverted microscope and photographed with an ORCA-ER CCD camera (Hamamatsu, Japan). Quantification of images containing oil droplets was performed using ImageJ software [45].

Alcian Blue staining

After 3 weeks of culture in chondrogenic differentiation medium, the cells were fixed with 4% para-formaldehyde (RT; 10 min), stained with 0.5% Alcian Blue (Sigma) in 0.1N HCl (PH=1.0) overnight and washed with water. Mosaic images were taken using a Zeiss Axio Observer inverted microscope. Alcian blue stained area was quantified using ImageJ software.

Von Kossa staining

The degree of osteogenesis of hHF-MSCs was evaluated by Von Kossa staining (Diagnostic Biosystems, Pleasanton, CA). Briefly, after 28 days in osteogenic differentiation medium cells were fixed in 4% paraformaldehyde (RT; 60min) and incubated with 1% (w/v) silver nitrate solution in dark (RT;30 min). After several washes cells were exposed to UV light for 60 min followed by incubation with 5% (w/v) of sodium thiosulfate (RT; 5min) to remove any un-reacted silver nitrate. Mosaic images were taken using a Zeiss Axio Observer inverted microscope. Calcium phosphate deposition area was quantified using Image J.

Gel compaction assay

To compare the ability of differentiated smooth muscle cells to generate force at different passages we performed fibrin gel compaction assay as described previously [17]. Briefly, 1×10⁶ cells were mixed with 800μL of fibrinogen, which was induced to polymerize with addition of 200μL thrombin in a 24-well plate at 37°C. The concentrations of fibrinogen and thrombin in the resulting fibrin gel were 2.5mg/mL and 2.5 U/mL respectively. After one hour incubation, the gels were released from the surface of the plate and the initial area of the gel (t=0 hr) was captured by an imaging system (UVP, Upland, CA) and measured using Image J. Fibrin gels were cultured in myogenic medium and at the indicated times the area (A) of each gel was measured and normalized to the initial area (A₀).

Immunocytochemistry

Immunocytochemistry was performed as described previously [17]. Briefly, cells were fixed (4% paraformaldehyde), permeabilized (PBS with 0.1% triton X-100), blocked (1%BSA/0.01% triton X-100 ) and incubated with one of the following antibodies: mouse anti-human αSMA (1:50; 4°C overnight; Serotec, Raleigh, NC), mouse anti-human calponin (1:50; 4°C overnight; Dako, Carpinteria, CA) or rabbit anti-human myosin heavy chain (1:100; 4°C overnight; Biomedical Technologies, Stoughton, MA,) diluted in blocking buffer. Subsequently the cells were incubated with Alexa Fluor 488 or 594-conjugated goat anti-mouse (1:100; 1 hr at RT; Invitrogen) and counter-stained with Hoechst nuclear dye (1:400 in PBS; 10 min; Sigma). Cells that were incubated with only secondary antibody served as negative controls.

Flow Cytometry

Cells expressing αSMA were measured by flow cytometry as described previously [17]. Briefly, on the fourth day of myogenic differentiation cells were resuspended in 4% ice cold paraformaldehyde, permeabilized with PBS containing 0.1% w/v saponin (MP Biomedical, Inc., Irvine, CA, USA) and 0.05% w/v NaN₃ (Fisher Scientific, PA, USA) for 10 min and blocked with blocking buffer (0.1% w/v saponin, 0.05% w/v NaN_3, 1% BSA in PBS) for 30 min. Cells were then incubated for 1 hr with mouse anti-human αSMA (1:50 in blocking buffer, Serotec) followed by incubation with Alexa Fluor 488-conjugated goat anti-mouse secondary antibody (1:100 dilution, Invitrogen) for 30 min. As a negative control, cells were incubated with only secondary antibody. After secondary antibody incubation cells were resuspended in PBS and run in a FACS Calibur flow cytometer (Becton Dickinson).

RNA isolation and cDNA synthesis

At the indicated times total RNA was isolated using Promega’s RNA isolation kit (Promega Corporation, Madison, WI) according to manufacturer’s instructions. First strand cDNA was synthesized from total RNA using superscript III reverse transcriptase (GIBCO) and oligo dT primer (GIBCO).

Quantitative real time PCR

To determine the kinetics of gene expression during differentiation, quantitative real time PCR was performed with the SYBR Green Kit (Bio-Rad, Hercules, CA) according to manufacturer instructions (see Supplemental Table 1 for primer sets). The level of each mRNA was quantified using the ΔC_T method and normalized with the expression level of the housekeeping gene, RPL32. The specificity of each product was verified by gel electrophoresis.

Clonal differentiation potential of hHF-MSCs

Passage 4 hHF-MSCs were plated using the limiting dilution method (1 cell/well) in 96-well plates. All wells were examined daily and those containing more than one cell were discarded for further analysis. Single cell clones were cultured for 15–20 days in proliferation medium. Selected clonal populations showing the highest proliferation capacity were further expanded into 48 well plates. Finally, thirty seven clones in total were induced towards osteogenic, adipogenic or chondrogenic lineage and the degree of differentiation was assessed by Von Kossa, Oil Red O and Alcian Blue staining, respectively.

Statistical analysis

Data were expressed as mean ± standard deviation and statistical significance (defined as p<0.05) was determined using Student’s t-test. For the kinetics of gene expression, we performed one way ANOVA (statistical significance defined as p<0.05) followed by post-hoc analysis (Fisher’s method with statistical significance defined as p<0.05).