A New Human Somatic Stem Cell from Placental Cord Blood with Intrinsic Pluripotent Differentiation Potential

Somatic human stem cells which can be propagated in large quantities while retaining their ability to differentiate into different tissue cell types could serve as a highly valuable resource for the development of cellular therapeutics (1). Although embryonic stem cells have the broadest differentiation potential (2), their use for cellular therapeutics is excluded for several reasons: the uncontrollable development of teratomas in a syngeneic transplantation model (3), imprinting-related developmental abnormalities (4), and ethical issues (1). On the other hand, previously published claims that adult tissue–specific stem cells possess an intrinsic differentiation potential to other tissues may have been premature. Over the past years, studies have shown that after hematopoietic allo-transplantation with BM or G-CSF–mobilized blood, a small percentage of donor cells can also be detected as nonhematopoietic tissue (5–9). However, this so-called “plasticity” appears to be an extremely rare event and may not carry a physiologically relevant impact in vivo (10). Moreover, subsequent data have shown that certain results may be the result of cell fusion (11). Recently, a rare cell from BM of rodents, called multipotent adult progenitor cell (MAPC), was identified which differentiated in vitro into cells of all three germ layers and contributed to most somatic tissues when injected into an early murine blastocyst (12). A phenotypically identical cell was isolated from human BM (13). It is unclear, however, whether such MAPCs decline with donor age, a phenomenon that has been observed for the hematopoietic (14) and the mesenchymal stem cell (MSC) compartment from BM (15). In contrast to adult BM, the stem cell compartment in cord blood (CB) is less mature. This has been documented for the hematopoietic stem cells, which in CB are more abundant than in BM and have a higher proliferative potential associated with an extended life span and longer telomeres (16–19). Besides this biological superiority, CB is abundantly available, is routinely harvested without risk to the donor, and infectious agents such as CMV are rare exceptions (20)—a definite advantage for the development of cell therapeutics in regenerative medicine. We identified a rare, CD45 and HLA class II–negative stem cell candidate displaying robust in vitro proliferative capacity without spontaneous differentiation but with intrinsic and directable potential to develop into mesodermal, endodermal, and ectodermal cell fates. Thus, we termed the primary population unrestricted somatic stem cell (USSC).

CB was collected from the umbilical cord vein with informed consent of the mother (21). USSCs were generated from 94 (40.3%) of 233 CB. The mononuclear cell fraction was obtained by Ficoll (Biochrom) gradient separation followed by ammonium chloride lysis of RBCs. Cells were plated out at 5–7 × 10⁶ cells/ml in T25 culture flasks (Costar). Two different media were used to initiate growth of the adherent USSC colonies: myelocult medium (StemCell Technologies) and low glucose DMEM (Cambrex) with 30% FCS, dexamethasone (10⁻⁷ M; Sigma-Aldrich), penicillin (100 U/ml; Grünenthal), streptomycin (0.1 mg/ml; Hefa-pharma), and ultraglutamine (2 mM; Cambrex). Expansion of the cells was performed in the same media but with a lower concentration or in the absence of dexamethasone. Cells were incubated at 37°C in 5% CO₂ in a humidified atmosphere. When cells reached 80% confluency, they were detached with 0.25% trypsin (Cambrex) and replated 1:3 as outlined in Table S1. USSC karyotyping was performed by cytogenetic standard protocols.

FITC-conjugated Ab CD4 (13B8.2), CD8 (B9.11), CD11a (25.3.1), CD14 (RMO52), CD15 (80h5), CD31 (5.6E), CD33 (D3HL60), CD34 (581), CD44 (J-173), CD62L (DREG56), CD80 (MAB104), CD90 (F15.42), CD71 (YDJ122), HLA-ABC (B9.12.1), F(ab′)2 goat anti–mouse IgM/IgG, and isotype controls were from Beckman Coulter; CD45 (2D1), CD49b (AK7), CD86 (FUN-1), and CD40 (5C3) were from BD Biosciences; CD62E/P (1.2B6) were from Serotec; CD105 (SN6) was from Caltag; and CD106 (BBIG-V3) was from R&D Systems. PE-conjugated Ab CD11b (BEAR1), CD16 (3G8), CD25 (B1.49.9), CD29 (K20), CD38 (T16), CD49e (SAM1), CD50 (HP2/19), CD54 (84H10), CD56 (NKH-1), CD117 (95C3), and isotype controls were from Coulter; glycophorin A (GA-R2), CD10 (HI10a), CD13 (L138), CD49d (L25.3), CD49f (GoH3), CD73 (AD2), CD123 (9F5), CD166 (3A6), and HLA -DR (G46-6) were from BD Biosciences; and CD133/1 (AC133) and CD133/2 (AC141) were from Miltenyi Biotech. Unconjugated Ab CD58 (AICD58) and isotype controls from were Immunotech; FLK-1/KDR (A-3), cytokeratin 8 (C51), cytokeratin 18 (DC-10), CK8/18 (C51), and vimentin (V9) were from Santa Cruz; CD44H (2C5) was from R&D Systems; and human endo (P1H12) was from Chemicon. Goat anti–rat Cy2 and normal rat IgG were from Dianova. Analysis was performed on a Beckman Coulter EPICS XL-MCL.

RT reactions were performed at 50°C with Omniscript (QIAGEN) and the 3′ primers following supplier's instruction. The PCR reactions were performed with HotStar Taq Master Mix (QIAGEN) at 93°C for 5 min, 35 × 93°C for 30 s, 55°C for 30 s, 72°C for 1 min, and 72°C for 5 min. Expression of the following molecules was detected by RT-PCR: CD49e (5′-ggcttcaacttagacgcgg; 3′-ccaggttgatcaggtactc; 640 bp), CD105 (5′-cctgccactggacacagg; 3′-atggcagctctgtggtgttg; 411 bp), CHAD (5′-aggaaccagctgtccagct; 3′-agtcaccaggactggctg; 513 bp), PDGFRa (5′-acagtggagattacgaatgtg; 3′-cac-atcagtggtgatctcag; 251 bp), EGFR (5′-tgccacaaccagtgtgct; 3′-gaccagttcatcagattcatc; 205 bp), IGFR (5′-cgagtggagaaatctgcgg; 3′-gaccagggcgtagttgtag; 272 bp), RUNX1 (5′-gcaagctgaggagcggcg; 3′-gaccgacaaacctgaggtc; 296 bp), and GAPDH (5′-ctcaagatcatcagcaatgcc; 3′-gatggtacatgacaaggtgc; 755 bp).

Telomere length measurement was performed with the Telo TAGGG Telomere Length Assay (Roche Diagnostics) according to manufacturer's instructions. DNA was treated with RsaI and Hinfl. Detection was performed with Image Master (Amersham Biosciences). Calculation of mean telomere restriction fragment lengths (mean TRF) was performed according to Harley et al. (22).

USSCs were seeded on glass coverslips coated with 1 mg/ml poly-d-lysine and 13 μg/ml laminin in differentiation medium XXL containing DMEM, 15% heat-inactivated FCS, 100 U/ml penicillin/streptomycin, 50 ng/ml nerve growth factor, 20 ng/ml bFGF, 1 mM dibutyryl cAMP, 0.5 mM isobuthyl methyl xanthine (IBMX), and 10 μM retinoic acid for up to 4 wk. For immunostaining, cells were fixed for 15 min with 4% paraformaldehyde (PFA). For immunostaining against gamma amino butyric acid (GABA) and tyrosine-hydroxylase (TH), cells were fixed in 4% PFA, 0.3% glutaraldehyde at RT for 5 min. After washing with PBS, cells were incubated in 1 M ethanolamine at RT for 20 min, preincubated and permeabilized in 10% normal goat serum with 0.03% Triton X-100 for at least 30 min, followed by incubation with the primary Ab (GABA 1:1000 (Sigma-Aldrich), TH 1:100 (Sigma-Aldrich), neurofilament (NF) cocktail 1:1,000 (BioTrend), DOPA-decarboxylase 1:500 (Sigma-Aldrich), glial fibrillary acidic protein (GFAP) 1:300 (Chemicon), synaptophysin 1:100 (Sigma-Aldrich), and Na⁺ channel 1:50 (Sigma-Aldrich). Secondary Ab were anti–mouse-FITC (1:100; Southern Biotechnology) and anti–rabbit rhodamine X Ab (1:1,000; Molecular Probes). Cell nuclei were labeled with 4-6′diamidino-2-phenylindoline (DAPI; Roche).

USSCs were labeled with pKH26 (Sigma-Aldrich) according to the supplier's protocol 2 d before transplantation. Adult male Wistar rats were anesthetized with Rompun (5 mg/kg) (Bayer) and Ketavet (100 mg/kg) (Pharmacia Upjohn). For transplantation, animals received a single, unilateral stereotactic injection of 1 μl (75,000 USSC/μl) into the hippocampus region using a 22-gauge needle. Coordinates were set according to the atlas of Paxinos and Watson (23). The animals received immune suppression by s.c. injection of cyclosporin A (15 mg/kg Sandimmune; Novartis) for 10 d, starting with a double dose injection 1 d before surgery. At different time points after transplantation the animals were killed. The brains were immediately removed and frozen in methyl butane (Sigma-Aldrich) on dry ice. 20-μm coronal sections were cut through the brain using a freezing microtome (CM 3050; Leica). At intervals of ∼400 μm, sections were stained for human Tau protein (hTau). Sections were fixed with 4% PFA for 10 min followed by 50%, 100%, 50% acetone (Merck), each for 2 min. After a washing step, sections were quenched by 0.3% H₂O₂ (Merck) for 30 min and blocked with 3% goat serum for 30 min. The hTau Ab (1:100; Chemicon) was incubated overnight at 4°C. After washes with PBS, sections were incubated with goat anti–rabbit biotinylated secondary Ab (1:200; Vector) for 1 h and rinsed with PBS. Later, sections were transferred to an avidin–biotin complex Vectastain ABC kit (Vector) and were developed with 3,3-diaminobenzidine (DAB; Sigma-Aldrich) and H₂O₂ for 10 min, dehydrated, and mounted in Entellan (Merck).

For differentiation into osteoblasts, USSCs were plated at 8,000 cells/cm² in 24-well plates and at 70% confluency supplemented with 10⁻⁷ M dexamethasone, 50 μM ascorbic acid-2 phosphate, and 10 mM dexomethasone, asorbic acid, β-glycerol phosphate (DAG) (24). For Alizarin red staining, cells were fixed for 5 min with 70% ethanol at 4°C to determine calcium deposition. Alkaline phosphatase (ALP) activity was determined in cell lysates obtained by treating cell cultures with 1% Triton X-100. The ELISA-based method was performed according to the manufacturer's protocol (Sigma-Aldrich). Results are expressed in nmol of p-nitrophenol produced per min. For quantitative Ca²⁺ determination, cell layers were scraped off the dish in 0.5 N HCl according to manufacturer's instructions (Sigma-Aldrich). For chondrogenic differentiation, a micromass culture system was used (25). 2 × 10⁵ USSCs were cultured in DMEM high glucose supplemented with antibiotic, 100 nM dexamethasone, 35 μg/ml ascorbic acid-2-phosphate, 1 mM sodium pyruvate, ITS + premix (1:100 dilution), and 10 ng/ml TGFβ1. Aliquots of 2 × 10⁵ cells in 0.5 ml of medium were centrifuged at 150 g in 15 ml polypropylene conical tubes. The pelleted cells were incubated at 37°C and 5% CO₂ for 21 d. For Alcian blue staining, cell aggregates were fixed in 4% formalin and cut into 10-μm sections and stained for histology. For immunostaining, frozen sections were fixed with 100% ethanol, incubated in 0.2 U/ml chondrotinase ABC for 40 min at 37°C. Blocking of nonspecific Ab binding sites was performed in 5% BSA/PBS for 1 h. The sections were incubated with the primary Ab diluted in 0.5% BSA/PBS for 1 h. Collagen type II (Chemicon) was detected by fluorescence microscopy after incubation for 30 min with a FITC-labeled secondary Ab diluted in 0.5% BSA/PBS. To induce differentiation into adipocytes, cells were plated at 1,000 cells/cm² in 24-well plates in DMEM with 1 μM dexamethasone (Sigma-Aldrich), 10 μg/ml insulin, 0.5 mM IBMX (Sigma-Aldrich), and 100 μM indomethacin (Sigma-Aldrich) (26). After 2 wk of adipogenic stimulation, cells were fixed in 5% PFA for 30 min and incubated with Oil Red-O to stain lipid vacuoles.

The surgical details of the femoral gap bone repair model were performed according to a method described by Bruder et al. (27), and differentiation into cartilage was shown previously (28).

10⁵ USSCs were expanded for 2 wk with 100 ng/ml Flt3-L (CellGenix), 100 ng/ml SCF (CellGenix), 100 ng/ml IL-3 (Cellsystems), 100 ng/ml IL-6 (CellSystems), 100 ng/ml TPO (CellGenix), and 100 ng/ml G-CSF (Amgen) in Myelocult medium as described previously (29). Human CFU assays were performed on days 0 and 14 (29) with 10⁴ cells in Methocult (Stem Cell Technologies).

Sheep heart was dissected into right and left atria, right and left ventricles, and septum. Several random strips from each area were fixed in 4% PFA, cut into 1 × 1-mm cubes, and embedded in OCT medium. 7–10-μm cryosections were probed with a human-specific anti–heat shock protein 27 (HSP27) Ab (Stressgen) (30) or the antidystrophin Ab NCL DYS2 (Novocastra) and anti–protein gene product 9.5 (PGP 9.5, ubiquitin c terminal hydroxylase) (Biogenesis). The secondary Ab for anti-HSP27 and antidystrophin was goat anti–mouse conjugated to Alexa 488, and the secondary Ab for anti-PGP 9.5 was goat anti–rabbit conjugated to Alexa 647 (Molecular Probes). Livers of the sheep were fixed in buffered formalin and embedded in paraffin. Liver sections (2 μm) were dewaxed and incubated for 10 min at 90°C for target retrieval. After blocking endogenous peroxidase by EnVision blocking reagent (Dako), sections were incubated in serum-free protein block (Dako) for 10 min and for 2 h in TBS containing 0.1% gelatin and the primary Ab anti–human serum albumin (clone HSA-11; 1:100; Sigma-Aldrich) or monoclonal anti–human hepatocyte Ab (clone OCH1E5; Dako). After washing, sections were incubated for 30 min with labeled polymer (EnVision System; Dako), and immunoreactivity was visualized by incubation with diaminobenzidine tetrahydrochloride. For microdissection applying the PALM Micro Beam System, immunohistochemistry was performed as described above with the monoclonal anti–human hepatocyte, clone OCH1E5 Ab except that sections were mounted on foil-laminated glass slides.

Isolation of single cells of human origin from human and sheep chimera liver tissue sections with >20% human cells and single cells of ovine origin from chimera liver tissue sections were performed using the PALM Micro Beam System (P.A.L.M. Microlaser Technologies). After target cell identification (the chimeric sheep liver slides were stained previously with the human hepatocyte specific Ab) and dissection from the surrounding tissue by a nitrogen laser beam (LMM), another strong laser was used to catapult (LPC) the microdissected material directly into the tube cap containing 20 μl of PCR buffer (Promega). Before PCR amplifications, cells were digested by proteinase K as described (31). To analyze individual micromanipulated parenchymal liver cells from human liver and human (stained) and ovine (unstained) cells from chimeric sheep liver tissue, genomic fragments of the human V_H1, ovine V_H7 and human TCRVβ7.2 and ovine TCRCδ genes were amplified with all external primer pairs in a first PCR round. 1-μl aliquots from the first amplification round were then subjected to a second round of fully nested PCR amplification using internal PCR primers. The sequences of the primers used for PCR amplification are given in Table S2. Single-cell PCR amplifications were performed in a 60-μl reaction mix volume containing 2.5 mM MgCl₂ [Promega], 200 μM of each dNTP, 2.5 μM of each primer, 1× PCR buffer [Promega], and 2.5 U platinum Taq-polymerase (Invitrogen). Cycling conditions for the first round included a single 2-min denaturation step at 95°C followed by 34 cycles at 95°C for 1 min, 56°C for 30 s, and 72°C for 1 min, and a 5-min incubation at 72°C. For the second PCR round, 45 cycles of amplification were used. To rule out interspecies cross-reactivity of the primer pairs used, we also amplified DNA from ovine cells with human-specific primers and vice versa. Amplification products were analyzed by agarose gel electrophoresis.

Serum proteins (human 50 ng/lane, ovine 500 ng/lane) were separated on a 12% SDS-polyacrylamide gel. After tank blotting onto a nitrocellulose membrane (Amersham Biosciences), blots were blocked with 5% blocking-grade nonfat dry milk (Amersham Biosciences) in PBS containing 0.05% Tween (PBS-T), incubated at 4°C for ∼12 h with a human albumin-specific Ab (HAS-11, 1:5,000; Sigma-Aldrich). After five washes for 5 min in PBS-T, blots were incubated with horseradish peroxidase–coupled anti–mouse secondary Ab (1:2,000; Amersham Biosciences) at RT for 1 h. Membranes were washed again and developed using the Amersham Biosciences ECL system according to manufacturer's instructions.

Expansion kinetic of USSCs (cell numbers, passages, and population doublings) are shown in Table S1. The sequences of the primers used for PCR amplification in the single cell PCR analysis of fusion/cell hybrids of the chimeric liver tissue are given in Table S2. Tables S1 and S2 are available.

USSCs were generated from 94 CB samples of a mean gestational age of 39.5 ± 1.46 wk (range 34–42), a net volume of 80 ± 21.4 ml including 29 ml citrate-phosphate dextrose (range 43–140 ml) and a total nucleated cell count of 7.6 ± 3.1 × 10⁸ (range 2.4–19.4). No correlation was detected between gestational age (>30 wk), hours elapsed after CB collection, volume, number of mononuclear cells in the CB after gradient separation and the success in generation of USSC. After 6–25 d (mean 15 d), adherently growing cells (USSC colonies) of fibroblastic morphology were detected with a median frequency of four colonies per CB (range 1–11). As shown in Fig. 1, A and B, USSCs are adherent, spindle-shaped cells, and have a size of 20–25 μm. Hematopoietic CD45⁺ cells were no longer detected after three passages. The USSC karyotype was normal 46XX or 46XY as analyzed for six individual USSC specimens for passages 5 (21 population doublings) to 19 (45 population doublings). USSCs were negative for CD14, CD33, CD34, CD45, CD49b, CD49c, CD49d, CD49f, CD50, CD62E, CD62L, CD62P, CD106, CD117, glycophorin A, and HLA-DR and expressed high levels of CD13, CD29, CD44, CD49e, CD90, CD105, vimentin, and cytokeratin 8 and 18, human Endo, low levels of CD10, and FLK1 (KDR), and showed variable but weak expression of HLA-ABC (Fig. 1 C). USSCs can be cultured for >20 passages equivalent to >40 population doublings without any spontaneous differentiation (Fig. 1 D and Table S1). The average telomere length of USSCs obtained after both 21 (passage 4) and 25 population doublings (passage 6) was 8.93 kbp; after 36 population doublings, the average telomere length of USSCs obtained was 8.60 kbp (passage 13). This is significantly longer than the telomere length of MSCs generated from a BM donor (age 30 yr) with 7.27 (passage 4) and 7.11 kbp (passage 9) at 19 and 27 population doublings, respectively (Fig. 1 E). USSCs showed expression of transcripts for epidermal growth factor receptor, platelet-derived growth factor receptor, insulin-like growth factor receptor, runt related transcription factor (Runx1), YB1, CD49e, and CD105 (Fig. 1 F). They were negative for the chondrogenic extracellular protein chondroadherin (Fig. 1 F), the bone-specific markers collagenase X, bone sialoprotein, the liver and pancreas-specific markers Cyp1A1 and PDX-1, and neural markers such as NF, synaptophysin, TH, and glial fibrillary acid protein (not depicted). Preliminary cDNA microarray analysis performed for two individual USSCs and one MSC preparation from human BM suggested differential expression of HAS1, which was only detected in MSCs but not in USSCs. This differential expression of HAS1 was confirmed by RT-PCR (n = 55; unpublished data) and immunochemistry.

Individual USSC preparations (n = 10) from passages 3–16 were analyzed for their neural differentiation potential. In XXL medium, the number of NF-positive cells increased, starting with ∼30% after 1 wk to more than 70% homogeneity after 4 wk (Fig. 2 A). Double immunostaining revealed colocalization of NF and sodium-channel protein in a small proportion of cells (Fig. 2 B). Expression of synaptophysin was detected after 4 wk (Fig. 2 C), indicating a more mature phenotype of neurons. Consistently, ∼90% of the cells stained positive for the inhibitory neurotransmitter GABA (Fig. 2 D). USSC-derived neurons could be identified that stained positive for TH (∼30% of the cells), the key enzyme of the dopaminergic pathway (Fig. 2 E). On the other hand, the number of DDCs (Fig. 2 F), the subsequent enzyme of the same transmitter pathway, was highly variable. Choline acetyltransferase, the enzyme of the cholinergic pathway, was detected in ∼50% of the cells (Fig. 2 G). Since 90% of USSCs in vitro showed positive immunoreactivity for GABA, coexpression with the enzymes of other neurotransmitter pathways was observed. These findings could be explained by the function of GABA during very early embryogenesis, where it is described to act as a signal molecule in neuronal development (32). Extensive analyses by patch-clamp recording of different USSC batches did not thus far reveal a voltage-activated, fast inactivating Na⁺ current, which is typical of terminally differentiated neurons (Copi, A., and K. Gottmann, personal communication). This finding suggests that USSCs differentiate to a precursor-like phenotype in vitro, in which distinct neuron-specific proteins are expressed, but a fully functional neuronal phenotype has not yet developed. Moreover, USSCs differentiated into astrocytes expressing GFAP, an astroglial intermediate filament reaching a transient maximum level of ∼45% at 2–3 wk in XXL medium (Fig. 2 H). During early differentiation, coexpression of GFAP and NF could be observed, indicating a common progenitor cell type (unpublished data). Interestingly, after 2–3-wk GFAP immune reactivity declined resulting in a more enriched neuronal cell population. To analyze the potency of USSCs to migrate, integrate, and differentiate into neuronal-like cells in vivo, USSCs were labeled with pKH26 and transplanted stereotactically into the hippocampus region of an intact adult rat brain. Numerous pKH26-labeled cells could be detected after 3 mo postgrafting (unpublished data). At 3 mo postimplantation, USSCs expressing human Tau protein could be identified widely distributed throughout the brain, indicating a high migratory activity of USSCs in vivo. Human Tau immune staining further revealed the neuronal-like, highly differentiated morphology of implanted USSCs in different ipsi- and contralateral regions of the adult brain including the neocortex (Fig. 2 I). In the brain of a nongrafted control animal, no human Tau immunoreactivity could be detected (unpublished data).

All USSCs tested (n = 30; up to passage 21) were capable of differentiating along the osteogenic and chondrogenic lineage. Adipogenic differentiation was observed for all six USSCs tested. Differentiation into osteoblasts was induced by dexamethasone, ascorbic acid, and DAG. After 5 d, cells showed initial calcium phosphate deposits. Bone-specific ALP activity was detected (Fig. 3 A) and continuous increase in Ca²⁺ release was documented (Fig. 3 B). Osteogenic differentiation was confirmed by expression of ALP, osteocalcin, osteopontin, bone sialo-protein, and collagen type I detected by RT-PCR (unpublished data). A pellet culture technique was employed to trigger USSCs toward the chondrogenic lineage (25). The chondrogenic nature of differentiated cells was assessed by Alcian blue staining (Fig. 3 C) and by expression of the cartilage extracellular protein type II collagen (Fig. 3 D). Chondrogenesis was further confirmed by RT-PCR for the cartilage-specific mRNAs encoding Cart-1, collagen type II, and chondroadherin (unpublished data). For induction of adipogenic differentiation, USSCs were cultured with dexamethasone, insulin, IBMX, and indomethacin (26). Adipogenic differentiation was demonstrated by Oil Red-O staining of intracellular lipid vacuoles (Fig. 3 E).

To determine the in vivo regeneration capacity for bone, the repair of critical size bone defects with USSC-loaded calcium phosphate ceramic cylinders was demonstrated. USSCs (n = 7) were expanded to passage 5 and loaded into porous ceramic cylinders of 5-mm length before implantation into the femur critical size bone defect of athymic Harlan Nude rats. 4 wk after transplantation, human cells were still alive within the defect bone (Fig. 3 F). Images of longitudinal (Fig. 3 G) and cross sections (Fig. 3 H) demonstrated bony healing between the cell-loaded implant and the host bone. Bony integration was established in forms of cancellous bone as detected by Toluidine blue staining. After 12 wk, a clear bony reconstitution was observed (Fig. 3 I). Cell-free implants served as negative controls (Fig. 3 J). However, it is impossible to quantitatively assess the number of human cells in the responding zone, since serial section reconstruction was not performed. Moreover, as shown by Allay et al. (33) the human cells initiate the fabrication of lamellar bone in implanted ceramic vehicles, but since the osteoblasts half-lives are 8–10 d, eventually the momentum of bone formation is joined by host-derived cells. However, the observation shown here indicates that the USSCs initiated the bone formation in the orthotopic site. The in vivo chondrogenic potential was assayed by loading USSCs into gelatin sponges (4 × 4 mm, Gelfoam^®; Upjohn Pharmacia). They were cultured in chondrogenic medium with supplementation of TGF-β (5 ng/ml) for up to 2 wk (Fig. 3, K and L) and then implanted s.c. into nude mice. After an additional 3 wk, the implanted cells demonstrated strong chondrogenic differentiation as shown by Toluidine blue staining (Fig. 3 M).

To evaluate the potential of USSCs to differentiate in vivo into hematopoietic cells, in utero transplantation into fetal sheep was employed (34). USSCs were transplanted intrauterine before immunological maturity. USSCs originating from one expanded colony tested negative for the human hematopoietic antigen CD45 by flow cytometry and PCR and were injected i.p. (1,500 USSC/sheep) into six preimmune (day 57–62) fetal sheep. To determine donor cell engraftment, blood and BM cells were analyzed by flow cytometry for the presence of human cells as described previously (35). Multilineage hematopoietic engraftment was observed, which included cells of erythroid (glycophorin A), myeloid (CD13, CD33), and lymphoid (CD3, CD7, CD10, CD20) lineages. Four out of six animals showed clear evidence of human hematopoietic reconstitution (up to 5% human hematopoietic and lymphopoietic engraftment in the total nucleated cell fraction in the blood after RBC lysis) 4-mo posttransplantation (Table I). The level of these human cells in the BM of these sheep was maintained through 12 mo posttransplant. In an attempt to elucidate these primarily unexpected in vivo results in vitro, 10⁴ and 10⁵ USSCs (n = 14; passages 5–8) were plated directly into a standard CFC assay. However, immediate hematopoietic colony formation was not observed for any of the 14 USSC preparations. When 10⁵ USSCs (n = 14) were cultured for 2 wk in the presence of hematopoietic growth factors, the cell number decreased to 3 × 10⁴ cells (range 2.0–3.4 × 10⁴). When these growth factor pretreated cells were subjected to the same standard CFC assay with 10⁴ total cells, in 2 of 14 USSC preparations hematopoietic CFC were observed, but the frequency of two and six colonies per 10⁴ seeded USSCs was very low. Since the morphology of such CFC was slightly different than that of regular CFC in CB or BM, the USSC-derived CFCs were pooled and analyzed by flow cytometry. The expression of glycophorin A (Fig. 4, 2.7%) and CD33 (Fig. 4, 85%) on these colony-derived cells confirmed the hematopoietic nature.

To explore the in vivo differentiation potential of USSCs into myocardial cells, the preimmune fetal sheep model was applied as described above. Differentiation of USSCs into human cardiomyocytes was analyzed 8 mo after transplantation. We have demonstrated previously the strict human specificity of the anti-HSP27 in sheep heart (30). Positive staining with the human-specific anti-HSP27 Ab was found in both atria, both ventricles, and the septum of the heart of in utero USSC-transplanted sheep, indicating that in this developmental in vivo model USSCs can robustly engraft throughout the heart (Fig. 5, A and B). Within the cells displayed in the longitudinal tissue section (Fig. 5 A), the HSP27 staining shows a striped pattern, which is consistent with previous findings (36). To demonstrate that the cells labeled with the human-specific HSP27 Ab were mature cardiomyocytes, tissue sections were probed with several antibodies that have a distinctive staining in cardiac cells. These antibodies employed had a broad specificity so that they could be used to confirm that this staining pattern was identical for the cells derived from the USSCs and from the surrounding cardiac sheep cells. Antibodies included antiryanodine receptor, anti-MHC, and as shown in Fig. 5, C and D, antidystrophin. In the adult heart, dystrophin is expressed at significant levels and localized at the plasma membrane (37, 38). As can be seen, both engrafted human cells and sheep cells have the mature pattern of distribution of dystrophin and are indistinguishable. The distribution pattern of HSP27 (Fig. 5 A) and other proteins such as dystrophin (Fig. 5, C and D) indicate that the engrafted human cells have a mature cardiac phenotype and the correct localization of proteins to be functional cardiomyocytes. The distribution of the engrafted cells observed was not homogeneous throughout the heart. In certain areas only single HSP-27–positive cardiomyocytes were detected, whereas in other areas groups of positive human cardiomyocytes were interspersed with sheep cardiomyocytes (Fig. 5 E). In some areas the engrafted cells were found in patches, with 10–20% of the cells in these areas being human cardiomyocytes interspersed with sheep cardiomyocytes. These quantities ranged from 0–3% of the sections analyzed (n = 30); however, in the atria and ventricle-positive patches of up to 10–20% USSC-derived cells were found. There was no pattern in the distribution, such as proximity to blood vessels. Using the preimmune fetal sheep model, we had shown previously that human MSCs from both fetal and adult sources engraft predominantly in the Purkinje fiber system and there are very few human cardiomyocytes (30). In this study, engraftment of USSCs in the Purkinje fiber system was detected also. Identification was by the characteristic morphology and staining with the characteristic Purkinje fiber marker PGP 9.5 (30). Similar to our earlier demonstration in preimmune sheep injected with human MSC, the engrafted human cells derived from USSCs were found in aggregates (Fig. 5 F) and not interspersed with sheep Purkinje fiber cells. This result is in contrast to the engrafted human cardiomyocytes, which are interspersed with the sheep cardiomyocytes.

To examine human hepatocyte development, livers of the sheep were taken 14-mo post-USSC transplantation. Fig. 6 A shows the specific staining of the human hepatocyte Ab OCH1E5 with a human liver. A liver from a nontransplanted sheep showed no cross-reactivity with the human-specific Ab (Fig. 6 D). As shown in Fig. 6 C in close association with portal veins, donor parenchymal liver cells generated from the USSCs represent the majority of total hepatic cells. Counting of representative areas of tissue sections of chimeric sheep liver revealed that 21.1 ± 3.2% of total liver cells were stained positive by the Ab that specifically recognizes human hepatocytes in association with portal veins >80%. To use a different human-specific Ab, the mAb HSA-11 recognizing human albumin was applied, which shows a specific staining for the human liver (Fig. 6 E) and no reactivity with the sheep liver (Fig. 6 H). In contrast, the liver of the sheep transplanted with USSCs showed a strong staining with human albumin (Fig. 6 G). Since human albumin is a secreted protein, the majority of cells stained positive, some showed a very strong pattern of albumin distribution, and the ones in the periphery showed a weaker staining. Fig. 6 I reproduces a Western blot showing a specific human albumin band and thus the functional production of this human protein in vivo in serum of the sheep obtained 17 mo after transplantation of the USSCs in utero.

To determine whether the human cells integrated into the sheep liver parenchyma acquired an organ-specific differentiated phenotype through cell fusion with indigenous ovine liver parenchymal cells or by liver cell differentiation, we tested single microdissected liver parenchymal cells from chimeric liver tissue for the coexistence of human and ovine genomes in these cells. To this end, single liver parenchymal cells, in which either expression of human proteins could be detected or were devoid of human proteins, were micromanipulated from chimeric liver sections and separately analyzed by single cell PCR. Each cell was transferred into a PCR reaction tube, digested by proteinase K, and subjected to two rounds of nested PCR amplification using four sets of human and sheep DNA–specific primers. To check for potential interspecies cross-reactivity of these primer sets specific for IGH and TCR loci of either human or sheep origin, we microdissected single liver parenchymal cells from human liver tissue sections as a control. From one third of the microdissected cells, we obtained a human PCR product but not a single PCR product of sheep origin. While analyzing 80 cells from chimeric tissue prestained with the human hepatocyte-specific Ab OCH1E5, human DNA fragments originating from the human IGH or TCRB loci and sheep DNA fragments from ovine IGH and TCRD loci were coamplified. In 24 of these 80 cells derived from paraffin sections, amplification gave rise to a PCR product for human DNA, whereas not a single sheep-specific PCR product was detected (Table II). Conversely, coamplification of human and ovine DNA fragments from cells micromanipulated from chimeric tissue, which did not stain for human proteins, only yielded PCR products of ovine but not human DNA (Table II). Using 16 PCR reactions, to which only PCR buffer but no cells were added as negative controls, neither human nor ovine DNA could be amplified. We conclude that fusion events, if they occur at all under physiological noninjury conditions, at best could account at a very low frequency for the differentiated liver cell–specific phenotype of the USSC-derived hepatocytes integrated into the chimeric liver tissue.

In this study, an adherent CD45-negative unrestricted stem cell population from human placental CB was shown to be pluripotent. In contrast to other data from CB identifying mesenchymal cells that differentiated only into osteoblasts, chondrocytes, adipocytes (39, 40), or neural progenitors in vitro (41), this is the first time that an adherent cell population has been identified, which after ex vivo expansion allows directed differentiation into bone, cartilage, hematopoietic cells, neural, liver, and heart tissue in vivo in various animal models. Previously, even the existence of mesenchymal/mesodermal cells in term CB has been very controversial. Several groups were unable to generate MSCs (42, 43) or generated MSCs only in a limited number of CB specimens (39, 41). Although the USSCs described here have a very low primary frequency in CB, they can be expanded to at least 10¹⁵ cells and maintain a normal karyotype. In contrast to MSCs from BM (26), the USSCs have a wider differentiation potential and differ in immunophenotype (44) and in their mRNA expression profile. Further differences between MSCs and USSCs include absent expression in USSCs of CD50, CD62L, CD106, and HAS1 (45), all of which are present in MSC. In contrast, USSCs are positive for the epithelial markers cytokeratin 8 and 18 and the endothelial marker KDR. In addition, CD44 was expressed in USSCs but not found on human or rodent MAPCs (12, 13). One major biological difference between USSCs and human MAPCs generated from BM (13) is the ease of generation of USSCs in cytokine-free cultures. The potential to generate hematopoietic cells as shown here from USSCs in vitro and in vivo has not been described for human MAPC. The USSC population can differentiate in vitro into osteoblasts, chondroblasts, adipocytes, and neural cell types in a homogeneous fashion. In these in vitro experiments with a clear, directed differentiation in the initial absence of the target tissue, cell fusion cannot account for the observed differentiation events. Osteogeneic, chondrogeneic, and neural differentiation were documented in vivo and in vitro; however, as shown for neuronal in vitro differentiation, only a precursor-like phenotype was generated which might be explained by the lack of a proper microenvironment with special cytokines and/or molecular signals which are not present in the in vitro culture system. Thus, these specific biological niches are able to fully exploit the intrinsic pluripotentiality of USSCs. Also, the in vivo differentiation of hematopoietic cells, which is seen in a similar quantity as that observed for purified CD34⁺ from CB (46), could not be confirmed in all experiments in vitro. However, a low frequency of hematopoietic colonies in 2 out of 14 culture experiments with cytokine-primed USSCs was observed. The reason for this in vitro versus in vivo difference cannot be explained yet. The immature mesodermal nature of these cells could be one possible reason: this means that a hematopoietic precursor cell derived from USSCs requires different growth/differentiation conditions, which are present in the in utero fetal sheep model (19).

One major difference compared to human MSCs is the distribution of USSC-derived cells in the heart. In previous studies using the preimmune fetal sheep model, we have shown that human MSCs engraft predominantly in the Purkinje fiber system and there are very few ventricular or atrial cardiomyocytes of human origin (30). In contrast, USSC-derived cells form both Purkinje fiber cells and cardiomyocytes. This would indicate that the USSCs are of an earlier cell type than multipotent MSC, possibly representing also the precursor cell for MSC. The ability of the USSCs to form more than one cell type suggests that these cells may be a valuable source of cells for the repair of the infarcted heart. Future studies will address whether USSCs engraft in other cell types in the heart. Another interesting observation is that the engrafted USSCs, which differentiated into cardiomyocytes, are interspersed with the sheep cardiomyocytes and are thus likely to participate in myocardial function. However similar to the engraftment of human MSCs (30) into the Purkinje fiber system, USSC-derived human cells form aggregates. This could indicate that during fetal development there are different mechanisms of expansion and/or recruitment of cells in the fetus between ventricular or atrial cardiomyocytes and the Purkinje fiber system. Given the remarkably high frequency of human cells in the noninjury transplant model, it was critical to determine whether any of the further differentiation pathways of USSCs into hematopoietic cells, parenchymal liver cells (>20%), and cardiomyocytes could be a result of cell fusion events. Fused cells have been described in the liver and in models of severe injury where the mouse will die if there is not a significant degree of regeneration (11). Because the human sheep chimeric model is a noninjury model, no driving force for potential fusion would exist apriori. As shown recently by Alvarez-Dolado et al. (47) in a noninjury mouse model using the Cre/lox system after BM transplantation, the frequency of newly generated tissue-specific cells in brain, heart, and liver resulting from fusion with respective tissue-specific cells after 10 mo at maximum was only 2.95 fused cells per section in the liver and 2.84 fused cells per section in the heart, respectively. This low frequency would hardly justify attributing fusion as the major or principal mechanism for the de novo generation of organ-specific cells. In contrast, in analyzing fusion versus nonfusion events the data shown here applying the noninjury in utero sheep model document a substantial degree of USSC-differentiated human parenchymal liver cells/slide in the liver and the heart. In the chimeric sheep liver, no indication of cell fusion was detectable. Therefore, the de novo generation of liver and heart cells from certain immature somatic progenitors appear predominantly engaged in noninjury organ regeneration. Importantly, the application of USSCs into different species thus far has not induced macroscopic or microscopic tumors month or even years (in sheep) after transplantation. In addition, USSCs lack HLA class II and costimulatory molecule expression. Preliminary experiments such as mixed culture inhibition suggest that USSCs, similar to adult and fetal MSCs (48), are also nonimmunogeneic and could even be immunosuppressive. This issue is under further investigation.

Thus, on the basis of their pluripotency and expansion under GMP conditions into large quantities, these USSC, when pretested for infectious agents and matched for the major transplantation antigens, may serve as a universal allogeneic stem cell source for the future development of cellular therapy for tissue repair and tissue regeneration.

We gratefully acknowledge the expert performance of electrophysiological experiments by Prof. K. Gottmann and A. Copi from the Department of Neurophysiology at the University of Düsseldorf, Germany, and are indebted to Anja Mottok and Martin-Leo Hansmann (Department for Pathology, University of Frankfurt, Germany) for sharing their expertise in single cell micromanipulation with us.

The work was supported by grants from the Stem Cell Network North Rhine-Westphalia and Eurocord III, and the National Institutes of Health grants AR45112 and HL077976.