Abstract
Acquisition of highly efficient neural differentiation based on understanding of initial lineage commitment of human embryonic stem (hES) cells remains a challenge. This study describes a simple three-stage protocol to induce hES cells into neural lineage cells using a 2-week coculture with murine bone marrow stromal cell (BMSC) PA6 followed by a 2-week propagation culture in PA6-conditioned medium and an additional 2-week selection culture in chemically defined neurobasal medium. This protocol generated a relatively high yield of neural lineage cells without mesodermal and endodermal lineage cell contamination. Notably, we demonstrated that PA6 coculture can significantly enhance the expression level of Notch signaling components and promote neural lineage entry of hES cell derivatives. Manipulation of Notch signaling can boost or suppress neural differentiation of hES cell derivatives, suggesting that Notch signaling may underlie the PA6-mediated neural induction. In vivo studies demonstrated that derived neural cells could improve the cognitive function of ischemic stroke rats. Intrastriatal human neural cell grafts were noted to migrate to damaged cerebral regions, enhance basic fibroblast growth factor production in the hippocampus, and restore the pyramidal neuron density and morphology in the hippocampal CA1 region, although only a small number of human donor cells were present in the hippocampus, suggesting that donor cells can boost hippocampal reconstruction by promoting the endogenous regeneration process. These findings demonstrate a pivotal role for Notch in hES cell fate determination and that manipulation of Notch signaling is therefore likely to be a key factor in taking command of hES cell lineage choice. This study suggested the potential of utilizing PA6 coculture to imitate the embryonic niche for hES cell neural induction via Notch signaling and a high application potential of BMSC-involved protocol, which can yield a whole lineage of human neural cells to promote endogenous regeneration in the hippocampus upon transplantation for potential therapy of ischemic stroke.
Introduction
Current stroke treatments are typically limited to supportive care and secondary stroke prevention, with stroke survivors left with debilitating behavioral deficits. Initial work with human embryonic stem (hES) cells has demonstrated that these cells have the ability to renew themselves and differentiate into various somatic cell types. Particularly, the ability of hES cells to differentiate into defined neural lineages may suggest a new cell source for cell replacement therapies for nervous system diseases (9). Many approaches to direct neural differentiation of hES cells have been reported: for instance, directing differentiation of hES cells to form embryonic bodies followed by propagating neural precursors in adherent culture (34,40), inducing hES cell neural differentiation in feeder-free monolayer adherent culture with supplemented cytokines (8,14), or coculture with embryonic astrocytes for neural induction of hES cells (29,34). However, the efficiency of neural differentiation remains low, or a chaotic mixture of differentiated cell types is generated, which confines the perspective application of defined neural cells in cell therapy for neural system diseases. Moreover, the initial lineage commitment process remains obscure and uncontrolled. To address these, by designing a simple protocol, we report that a high efficiency of neural differentiation can be achieved and also analyze the mechanism underlying the neural induction process. Notch signaling is deployed in many different tissues to regulate differentiation decisions by mediating signaling between adjacent cells. In this study, we report that Notch ligands and receptors are expressed in bone marrow stromal cell (BMSC) PA6 cell line and cocultured hES cell derivatives where they play a significant role in the promotion of primary neural fate. Further, hES cell-derived neural cells are transplanted into the caudate putamen of stroke rats to determine their therapeutic potential in cell therapy for ischemic stroke.
Materials and Methods
Stem Cell Maintenance
The hES cell lines H14 [WA14, male; Wicell Research Institute, Madison, WI, USA; passages (P) 37–54] and H9 (WA09, female; Wicell Research Institute; P45–62) were cultured on irradiated mouse embryonic fibroblasts [MEFs; generated in house using WiCell protocols; irradiated by exposure to 8,000 rads (80 Gy) γ-irradiation in a γ-irradi-ator (Gammacell 1000 Elite 214, MDS Nordion, Ottawa, Ontario, Canada)] in Dulbecco's modified Eagle's medium (DMEM)/F12 supplemented with 20% knockout serum replacement (KSR), 1 mM L-glutamine, 1% nonessential amino acids (NEAAs), 0.1 mM β-mercaptoethanol, and 4 ng/ml human basic fibroblast growth factor (bFGF; all from Invitrogen, Carlsbad, CA, USA). The hES cells were passaged about once per week by mechanical dissection with the cell divider (StemPro EZPassage tool, Invitrogen) and transferred onto freshly prepared MEF feeders (33).
Preparation of PA6 Coculture and PA6-Conditioned Medium
Neural differentiation of hES cells was induced by using male murine BMSC PA6 (31). The hES cell induction medium consists of DMEM/F12, 20% KSR, 1% NEAA, 1 mM L-glutamine, 0.1 μM β-mercaptoethanol, 10 ng/ml bFGF, 10 ng/ml human epidermal growth factor (EGF), 10 ng/ml brain-derived neurotrophic factor (BDNF), 10 ng/ml neurotrophin-3 (NT-3), B27 (1:50), 0.5 mM dibu-tyryl cAMP (all from Invitrogen), 1 ng/ml transforming growth factor β3, and 500 ng/ml noggin (both from R&D Systems, Minneapolis, MN, USA). For the preparation of direct PA6 coculture, γ-irradiated PA6 cells at 2 × 105/well were seeded onto six-well tissue culture plates (Greiner Bio-One, Frickenhausen, Germany) 1 day before coculture with hES cells. To prepare noncontact PA6 coculture, γ-irradiated PA6 cells (2 × 105) were seeded onto tissue culture inserts (ThinCerts, Frickenhausen, Germany) housed in six-well plates 1 day before hES cells were replated onto each well of the same plates. To prepare PA6-conditioned medium (PA6-CM), γ-irradiated PA6 cells (1 × 107) in 10 ml of hES cell induction medium were seeded onto 75-cm2 tissue culture flasks (Greiner BIO-One). The spent medium was collected daily for 4 consecutive days. Prior to use, collected PA6-CM was diluted at the ratio of 1:1 (equal volume of fresh induction medium added into PA6-CM) and 2:1 (one-half volume of fresh induction medium added into PA6-CM), respectively. All six-well plates were precoated with Matrigel (BD Biosciences, Bedford, MA, USA) diluted at 1:20 with DMEM/F12. The hES cells were seeded as small pieces composed of around 10–25 cells per piece at a concentration of 1 × 104 pieces/well into six-well tissue culture plates pretreated as above.
Three-Stage Culture for Neural Derivation of hES Cells
Neural commitment of hES cells was initiated by direct coculture on irradiated PA6 feeder cells for 2 weeks in hES cell induction medium (stage 1). Then hES cell derivatives were subjected to a 2-week propagation culture in 1:1 diluted PA6-CM (stage 2) followed by another 2-week selection culture in neurobasal medium supplemented with 1% fetal calf serum (FCS), 20 ng/ml bFGF, 20 ng/ml EGF, B27 (1:50), N2 (1:100) (all from Invitrogen), 10 μM Forskolin (Calbiochem, San Diego, CA, USA), and 0.2 mM ascorbic acid (Sigma, St. Louis, MO, USA) (stage 3). At the end of stage 2, derived cells were enzymatically segregated and 2 × 105 cells in stage 3-specific medium were seeded into each well of poly-L-ornithine- and laminin-coated (both Sigma) six-well plates for the selection culture. The control culture series run in parallel was without PA6 coculture in stage 1, during which hES cells were directly seeded into six-well plates precoated with Matrigel, but with the same stage 2- and stage 3-specific treatment. All the media used were refreshed every other day.
Interference with Notch Signaling
L-685458 (Calbiochem) inhibits Notch signaling by inactivating γ-secretase activity (27). When indicated, L-685458 at 4 μM in 0.4% dimethyl sulfoxide (DMSO, Sigma) was supplemented into the medium for 8 days from the first day when the culture was set up. In parallel, 0.4% DMSO alone in the medium was used as a control.
Construction of Expression Vector
Human Notch intracellular domain (NICD) of Notch1 (corresponding to amino acids 1759–2444 of GenBank NM_017617) was amplified by polymerase chain reaction (PCR) using Notch1-specific primers (forward, 5′-CGC GGA TCC ATG CGC AAG CGC CGG CGG CAG CAT-3′; reverse 5′-ACG TCT AGA CAC GTC TGC CTG GCT CGG-3′; Invitrogen), with human pla-cental tissue cDNA (from a 30-year-old healthy female donor with consent) as the template. The PCR product (2,058 bp) was inserted into the BamHI/XbaI restriction sites of the mammalian expression vector pcDNA3.1 (Invitrogen), generating a vector pcDNA3.1/Notch1, which was termed pcNICD. The intact pcDNA3.1 was used as a negative control. Cells seeded at 2×104/cm2 in six-well plates (0.5 ml medium/cm2) were cultured for 24 h before transfection with 1 μg of pcNICD and 2 μl of Lipofectamine 2000 (Invitrogen) per well according to the manufacturer's instructions. Transfected cells were then selected by the antibiotic G418 (Invitrogen) at 250 μg/ml for 2–4 weeks until individual colonies appeared.
Reverse Transcription (RT)–Polymerase Chain Reaction (RT-PCR)
Total RNA was extracted by using an RNeasy Extraction Kit (Qiagen, Valencia, CA, USA). After RT, the first-strand cDNA was used for amplification of a specific gene sequence (Table 1). RNA integrity was confirmed by RT-PCR of a ubiquitous mRNA glyceraldehyde-3-phosphate dehydro-genase (GAPDH) for mouse cells or pyruvate dehydroge-nase (PDH) for human cells. Quantitative real-time PCR was performed in an ABI Prism 7000 Sequence Detector (Applied Biosystems, Foster City, CA, USA) using SYBR green PCR Master Mix reagent as the detector according to the manufacturer's instructions. Reactions were in 20 μl of a mixture containing 2 μl of cDNA dilution, 0.2 μM each primer, and 10 μl of 2× SYBR master mix composed of reaction buffer, AmpliTaq gold DNA polymerase, dNTP mixture with dUTP, passive reference, and SYBR Green I (all from Applied Biosystems). Gel electrophoresis and melting curve analyses were performed to confirm correct PCR product sizes and absence of nonspecific bands. Human primer sequences were as follows: β-actin: (forward) CGC ACC ACT GGC ATT GTC AT, (reverse) TTC TCC TTG ATG TCA CGC AC; Nanog: (forward) ACC AGA ACT GTG TTC TCT TCC ACC, (reverse) GGT TGC TCC AGG TTG AAT TGT TCC; Reduced Expression Protein 1 (Rex-1): (forward) GGC GGA AAT AGA ACC TGT CA, (reverse) CTT CCA GGA TGG GTT GAG AA; forkhead box D3 (FoxD3): (forward) GTC GTT CAG CAT CGA GAA CA, (reverse) GGA GAG TGG CAC GCT AAG AA; sex-determining region Y box 1 (Sox-1): (forward) GGG AAA ACG GGC AAA ATA AT, (reverse) CCA TCT GGG CTT CAA GTG TT; paired box 6 (Pax-6): (forward) ATG AGG CTC AAA TGC GAC TT, (reverse) CAT TTG GCC CTT CGA TTA GA; neurofilament heavy chain (NFH): (forward) TGA ACA CAG ACG CTA TGC GCT CAG, (reverse) CAC CTT TAT GTG AGT GGA CAC AGA G; bone morphogenetic protein 4 (BMP4): (forward) AAG CGT AGC CCT AAG CAT CA, (reverse) TGG TTG AGT TGA GGT GGT CA; Brachyury: (forward) ACC CAG TTC ATA GCG GTG AC, (reverse) ATG AGG ATT TGC AGG TGG AC; muscle segment homeobox 1 (Msx-1): (forward) CCT TCC CTT TAA CCC TCA CAC, (reverse) CCG ATT TCT CTG CGC TTT TC; pancreatic and duodenal homeo-box 1 (Pdx-1): (forward) TGG ATG AAG TCT ACC AAA GC, (reverse) GGT CAA GTT CAA CAT GAC AG; NK6 homeobox 1 (Nkx6.1): (forward) CTA TTC GTT GGG GAT GAC A, (reverse) AGC TGC GTG ATT TTC T; gua-nine-adenine-thymine-adenine binding protein 4 (GATA4): (forward) TCC AAA CCA GAA AAC GGA AG, (reverse) AAG ACC AGG CTG TTC CAA GA. The expression level of each target gene was normalized to the level of β-actin using the comparative CT method (36). Data were presented as the fold change in expression relative to hES cell derivatives in the control culture (without PA6 or L-685458).
Primer Sequences for RT-PCR and Product Sizes of Targeted genes
GAPDH, glyceraldehyde-3-phosphate dehydrogenase; bFGF, basic fibroblast growth factor; IGF-I and -II, insulin-like growth factor-I and -II; BDNF, brain-derived neurotrophic factor; CNTF, ciliary neurotrophic factor; GDNF, glia-derived neurotrophic factor; NGF, nerve growth factor; NT-3, neurotrophin 3; EPO, erythropoietin; VEGF, vascular endothelial growth factor; CXCL-12, chemokine (C-X-C motif) ligand 12; PDH, pyruvate dehydrogenase; GFAP, glial fibrillary acidic protein; MAP-2, microtubule-associated protein 2; Nurr-1, nuclear receptor related protein-1; OCT-4, octamer binding transcription factor 4; ALP, alkaline phosphatase; c-kit, v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog; Sox-9, sex-determining region Y box 9; PPARγ, peroxisome proliferator-activated receptor gamma; AFP, alpha fetoprotein; GLUT-2, glucose transporter 2.
Electrophysiological Analysis
Cells were superfused with artificial cerebrospinal fluid (containing 140 mM NaCl, 3 mM KCl, 2 mM CaCl2, 1 mM MgCl2, 1.25 mM NaH2PO4, 26 mM NaHCO3, 25 mM D-glucose, and 10 mM HEPES/NaOH; pH 7.35, 310 mOsmol/kg; all Sigma) at 1–2 ml/min. Whole-cell voltage-clamp recordings were conducted at room temperature with an Axopatch-200B amplifier (Axon Instruments Inc., Foster City, CA, USA) interfaced by an A/D converter (Digidata 1320, Axon Instruments Inc.) to a PC running PClamp software (Axon Instruments Inc.). With a tip resistance of 3–5 mΩ, the patch pipette contained 120 mM potassium gluconate, 20 mM KCl, 10 mM NaCl, 10 mM ethylene glycol tetraacetic acid, 1 mM CaCl2, 2 mM Mg-ATP, 0.3 mM Na-GTP, and 10 mM HEPES/KOH (all Sigma; pH 7.2, 280 mOsmol/kg). Signals were filtered at 10 kHz. Holding currents were adjusted to 0 pA at the beginning of recording to identify the membrane potential of a neuron.
Dopamine Release Assay
After three-stage culture, derived cells were rinsed in a low-KCl (4.7 mM) solution and then incubated in 2 ml of a high-KCl solution (60 mM KCl, 85 mM NaCl, 2.5 mM CaCl2, 1.2 mM MgSO4, 1.2 mM KH2PO4, 11 mM D-glucose, and 20 mM HEPES/NaOH; pH 7.4) for 15 min. The concentration of dopamine was determined by high-performance liquid chromatography (HPLC) using a reverse-phase column and an electrochemical detector system (HTEC 500, Eicom Corp., San Diego, CA, USA).
Western Blot Analysis
Total proteins (20 μg) were fractionated by 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and electrotransferred to polyvinylidene difluoride (PVDF) membranes (both Millipore, Billerica, MA, USA). The following primary antibodies were applied to membranes pretreated with 5% fat-free milk (Millipore): rabbit anti-NICD (1:1,000; Cell Signaling Technology, Beverly, MA, USA), anti-hairy and enhancer of split (Hes) 1 (Hes1; 1:500; Chemicon, Temecula, CA, USA), anti-Hes5 (1:500; Chemicon), and mouse anti-α-tubulin (1:2,000; Chemicon) followed by further incubation with horseradish peroxi-dase (HRP)-conjugated goat anti-rabbit or anti-mouse IgG or IgM (1:2,000; Millipore). Band intensities were quantified by using Image Lab Software (Bio-Rad, Hercules, CA, USA), and the protein expression level was normalized to the level of α-tubulin. Data were presented as the fold change in expression relative to hES cell derivatives in the control culture (without PA6 or L-685458).
Stroke Surgery
Animal experiments were carried out in accordance with NIH guidelines and were approved by local governmental authorities. The four-vessel occlusion method was applied to adult male Wistar rats (Shanghai Laboratory Animal Center, Chinese Academy of Sciences, Shanghai, China) (10 weeks of age and weighing 250–300 g) to make stroke models with bilateral hemispheric ischemia (30). Briefly, the alar foramen on both sides of the first cervical vertebra was exposed and a 0.5-mm-diameter electrocautery needle (Bovie Monopolar Electrocautery, Louisville, KY, USA) was inserted through the alar foramen on each side to electrocauterize the vertebral artery. After that, both common carotid arteries were clamped (Spectrum Surgical Instruments Corp., Stow, OH, USA) to produce four-vessel occlusion. Carotid clamps were removed following 15-min occlusion. Control rats were sham operated without artery occlusion.
Cell Transplantation
Three days poststroke surgery, 2 × 105 derived neural cells in 5 μl of 0.9% saline were injected into each caudate putamen (4 × 105 cells/brain; 0.5 mm anterior to bregma, 3 mm lateral to midline, and 5 mm ventral to the dura) at a rate of 1 μl/min. An equal volume of saline was injected into ischemic control rats. As cellular control for the transplant study, human mesenchymal stromal cells (hMSCs) were propagated before transplantation. Briefly, human bone marrow aspirates (30 ml) were taken from the iliac crest of healthy donors (10 males, 25–42 years old). The harvest was approved by the Regional Ethics Review Board, and informed consent was obtained from each donor. Heparinized bone marrow was separated by a 1,073 g/ml Percoll density gradient (GE Healthcare, Piscataway, NJ, USA). Mononuclear cells were collected, washed, and resuspended in low-glucose DMEM (Invitrogen) supplemented with 10% FCS, 1% NEAA, and 1 mM L-glutamine. Cells were plated at 1.6×105 cells/cm2 in culture flasks. After passage once, a single-cell suspension was prepared and transplanted into recipient rats in the same manner as for hES cell-derived cells. The skull holes were then cemented (Kaiser Technology Co., Ltd., Taichung, Taiwan). Cyclosporine A (Novartis, Basel, Switzerland) was applied intraperitoneally to all rats every day at a 10 mg/kg dose.
Behavioral Analysis
The spatial cognition ability of rat models was assessed with the Morris water maze system (Harvard Apparatus, Holliston, MA, USA) as reported previously (23). Before stroke surgery, rats that could reach the hidden platform within the water maze in approximately 2 min were recruited to in vivo studies. Six weeks posttransplant, rats were trained daily by using a set of four trials on each of 12 consecutive days prior to formal assessments on the following 2 days. On the last day of the water maze test, the spatial probe test was also conducted. The platform was removed and the rat was allowed to swim for 60 s. The number of times each rat crossed the area in which the platform had previously been located was counted. The time spent in the target quadrant in which the platform had previously been located was also recorded.
Enzyme Immunoassay
The hippocampus from each excised brain was dissected, homogenized in buffer solution (20 mM Tris-HCl, 137 mM NaCl, 1 mM DTT, 0.5% Triton X-100, and 0.5 mM PMSF; pH 8.0; all from Sigma), and centrifuged at 14,000×g for 30 min at 4°C. Protein concentration in each tube of supernatant was determined by a protein assay kit (Bio-Rad). Enzyme immunoassay for bFGF was conducted using an assay kit (Quantikine HS; R&D Systems) according to the manufacturer's instruction.
Tissue Preparation and Histology
Excised brains were coronally cryosectioned at a thickness of 5 μm. The number of donor cells was quantified by counting immunoreactive cells in every fifth section throughout the graft and surrounding tissue, and then corrected by the Abercrombie formula (26). Additionally, images of every fifth section of thionin (0.2%; Sigma)-stained CA1 region of the hippocampus 2.8–3.0 mm posterior to bregma were captured and overlaid to an image of a counting frame (1 mm × 0.25 mm) derived from the Neubauer hemocytometer (EMS Corporation, Hatfield, PA, USA). The preponderance of pyramidal neurons with large nuclei, discrete nucleoli, and clear cell periphery in the hippocampal CA1 region were counted as reported previously (26). The morphology of the hippocampal CA1 region was graded according to previously reported criteria (17): grade 0, less than 10% of total cells with pyknotic morphology within the counting frame; grade I, 10–40% of total cells with pyknotic morphology; grade II, 40–70% of total cells with pyknotic morphology; grade III: more than 70% of total cells with pyknotic morphology.
Immunocytochemistry and Immunohistochemistry
Cells were originally seeded onto coverslips (Lab-Scientific, Highlands, NJ, USA) or cells (1 × 105) were cytospun onto a cytosmear (Autosmear CF-120, Sakura Seiki Co., Tokyo, Japan). The following primary antibodies were applied to cells: mouse anti-β-tubulin III (TuJ1, 1:500; Sigma), anti-oligodendrocyte marker 4 (O4; 1:50; Chemicon), anti-stage-specific embryonic antigen-4 (SSEA-4, 1:100; Santa Cruz Biotechnology, Santa Cruz, CA, USA); rabbit anti-Musashi-1 (1:100; Millipore), anti-Nestin (1:400; Chemicon), anti-glial fibrillary acidic protein (GFAP, 1:400; Chemicon), anti-Synapsin I (1:500; Chemi-con), anti-tyrosine hydroxylase (TH, 1:100; Chemicon), anti-γ-aminobutyric acid (GABA, 1:200; Sigma), anti-choline acetyltransferase (ChAT, 1:200; Chemicon), anti-serotonin (1:100; Chemicon), whereas sections were incubated with the following primary antibodies: mouse anti-human mitochondria (hMito, 1:40; Chemicon), anti-human nuclei (HuN, 1:100; Chemicon), and anti-human neural cell adhesion molecules (hNCAM, 1:100; Santa Cruz Biotechnology). Cells or sections were further incubated with fluorescein isothiocyanate (FITC)- or tetram-ethyl rhodamine isothiocyanate (TRITC)-conjugated goat anti-mouse or anti-rabbit IgG or IgM (1:100; Millipore). Rinsed cells or sections were counterstained with 10 μg/ml propidium iodide (PI; Sigma) or 4′,6-diamidino-2-phenylindole (DAPI; Sigma). Additionally, biotinylated goat anti-mouse IgG (1:100; Vector, Burlingame, CA, USA) was applied to sections incubated with the anti-hMito or anti-hNCAM. After cells were rinsed, streptavidin-HRP (Vector) was added and immunoreactivity was visualized by using diaminobenzidine (DAB; Vector).
Statistical Analysis
Data were presented as mean ± SEM. Unless stated otherwise, data were based on H14 cells. Multiple groups were compared by ANOVA followed by post hoc analysis (Student–Newman–Keuls test), whereas the t test was used when only two groups were compared. Histological grading analysis for the hippocampal pyramidal layer was performed by Nemenyi test for pairwise comparison of ranked data. The software SPSS 17.0 (SPSS Inc., Chicago, IL, USA) was used and values of p < 0.05 were considered statistically significant.
Results
PA6-Mediated Neural Induction
Previous reports have shown that PA6 cells could support the growth of hematopoietic stem cells (35). This effect of PA6 was confirmed by our observation that PA6 cells could express various transcripts encoding growth-promoting factors, such as bFGF, insulin-like growth factor-I (IGF-I) and IGF-II (Fig. 1a). However, neurotrophic effects of PA6 on hES cells were largely understudied. It was noted that several transcripts of genes encoding neurotrophins including BDNF, NT-3, ciliary neurotrophic factor (CNTF), glia-derived neu-rotrophic factor (GDNF), and nerve growth factor (NGF) could be detected in PA6 cells. Neuroprotective factors, erythropoietin (EPO), vascular endothelial growth factor (VEGF), and the chemokine (C-X-C motif) ligand 12 (CXCL-12) were also detectable in PA6 (Fig. 1a). These data suggested that PA6 cells may have a neurotrophic and neuroprotective potential that appears to involve detected factors. Here, to optimize PA6-mediated induction on hES cells, we compared different PA6-involved induction methods. In H14 cell derivatives at day 8, human cell viability and percent of Musashi-1+ neural precursors after direct contact PA6 coculture were 91.8 ± 1.1% and 48.7 ± 2.7%, significantly higher than those derived from noncontact PA6 coculture, culturing in PA6-CM, and control cultures, respectively (Fig. 1b). Consistent data were also derived from H9 cell derivatives (Fig. 1c), suggesting that PA6 coculture in a direct contact manner had preponderance in the generation of neural precursors over any other PA6-involved methods. We further asked how cell–cell interaction between PA6 and hES cells really made the difference.
PA6 cells could elicit neurotrophism to human embryonic stem (hES) cells. (a) RT-PCR analysis of gene transcripts encoding indicated factors in PA6 cells. (b, c) Cell viability and Musashi-1 positivity in hES cell line H14- (b) or H9-derived cells (c) after 8 days of culture under indicated PA6-involved conditions (CM, conditioned media). *p < 0.05 versus any other culture groups (n = 12). bFGF, basic fibroblast growth factor; IGF-I and -II, insulin-like growth factor-I and -II; BDNF, brain-derived neurotrophic factor; NT-3, neurotrophin 3; CNTF, ciliary neurotrophic factor; GDNF, glia-derived neurotrophic factor; NGF, nerve growth factor; EPO, erythropoietin; VEGF, vascular endothelial growth factor; CXCL-12, chemokine (C-X-C motif) ligand 12; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
Notch Signaling Underlies PA6-Mediated Neural Induction
Notably, Notch ligands Delta-1, Delta-3, Jagged-1, and Jagged-2 could all be readily detected in PA6 cells (Fig. 2a), whereas Notch receptors Notch1, Notch2, and Notch3 were detectable in hES cells and their derivatives (Fig. 2b), suggesting that the candidate for such a signaling system underlying interaction between PA6 and hES cells may be Notch signaling. Generally, upon activation by ligands, Notch receptors can be cleaved by γ-secretase to release the NICD that can regulate the expression of downstream target genes including Hes1 and Hes5 (2). At day 8, protein expression levels of NICD, Hes1, and Hes5 in hES cell derivatives after direct contact PA6 coculture were noted to be significantly higher than those in control cultures, indicating Notch signaling was activated during PA6 coculture (Fig. 2c, d). We also found by quantitative RT-PCR that expression of transcripts representative of neuroectodermal formation, such as Sox-1, Pax-6, and NFH, was significantly enhanced, whereas transcript expression of mesodermal markers (BMP4, Brachyury, and Msx-1), endodermal markers (Pdx-1, Nkx6.1, and GATA4), and pluripotency markers (Nanog, Rex-1, and FoxD3) was obviously decreased in PA6-cocultured hES cell derivatives as compared to those in control cultures (Fig. 2e). These data suggested that PA6 coculture in a direct contact manner may play a significant role in promoting neural lineage entry of hES cells while blocking differentiation into nonneural fates. We went on to test whether Notch signaling was requisite for PA6-mediated neural lineage entry. By blocking Notch signaling with L-685458 for the initial 8 days, expression of NICD, Hes1, and Hes5 was noted to markedly decrease in PA6-cocultured cells. Expression of NICD and Hes5 was reduced to the level that had no significant difference with that in control cultures in the presence of L-685458 (Fig. 2c, d), suggesting that Notch cleavage and downstream activation were fully blocked by L-685458. Interestingly, regulation of Hes1 appeared to be minimally affected by L-685458 (Fig. 2c, d). This observation was in agreement with other reports showing that Hes1 expression does not exclusively depend on Notch signaling (38). Notably, promotive effects of PA6 coculture on neural transcript expression of Sox-1, Pax-6, and NFH were abolished, whereas expression of nonneu-ral markers was derepressed in the presence of L-685458 (Fig. 2e), demonstrating that the effects of PA6 coculture cannot override Notch signaling inhibition. Even in the control cultures, L-685458 was noted to have similar effects on neuroectodermal marker expression (Fig. 2e), suggesting that endogenous Notch signaling within hES cell derivatives may also play a role in neural specification. Consistently, L-685458 obviously decreased the frequency of Musashi-1 expression in both PA6-cocultured cells (without vs. with L-685458: 48.7 ± 2.7% vs. 11.3 ± 1.2%, p < 0.0001) and control cultures (without vs. with L-685458: 12.3 ± 2.1% vs. 7.1 ± 0.6%, p = 0.033), without impact on cell viability of PA6-cocultured cells (without vs. with L-685458: 91.8 ± 1.1% vs. 87.1 ± 2.1%, p = 0.06) or control cultures (without vs. with L-685458: 74.5 ± 2.2% vs. 70.7 ± 1.9%, p = 0.203). For further confirmation, hES cells were transfected with pcNICD initially, and it was noted that hES cells had a propensity to differentiate to neural lineage cells thereafter, with Musashi-1+ cells coming up to 41.4 ± 2.3% after spontaneous commitment for 8 days, in marked contrast to Musashi-1 positivity of 10.5 ± 1.1% derived from control cultures priorly transfected with pcDNA3.1 (vector only). Taken together, these data suggested that PA6 coculture in a direct contact manner could exclusively promote neural lineage entry of hES cells, at least in part, by activating Notch signaling.
Notch signaling underlies PA6-mediated neural induction. (a, b) RT-PCR analysis of Notch ligands in PA6 cells (a) and receptors in hES cell (H9 and H14) derivatives (b). Coc, coculture; Ctr, control culture. (c) Western blotting analysis of Notch signaling components after 8 days of culture of H14 cells under conditions with PA6 or without PA6 coculture (control) in the presence (+) or absence (-) of L-685458. (d) Quantitative analysis (n = 12) for band intensities in (c). (e) Quantitative RT-PCR analysis (n = 6) of germ layer-specific gene expression in hES cell (H14) derivatives after 8 days of culture under different conditions as indicated. Data in (d, e) were presented in logarithmic scale as the fold change in expression relative to control cultures (with neither PA6 nor L-685458). #p < 0.05 versus Control; *p < 0.05 versus Control or PA6 + L-685458; ▲p < 0.05 versus Control + L-685458. PDH, pyruvate dehydrogenase; NICD, Notch intracellular domain; Hes1, hairy and enhancer of split 1; Sox-1, sex-determining region Y box 1; Pax-6, paired box 6; NFH, neurofilament heavy chain; BMP4, bone morphogenetic protein 4; Msx-1, muscle segment homeobox 1; Pdx-1, pancreatic and duodenal homeobox 1; Nkx6.1, NK6 homeobox 1; GATA-4, guanine-adenine-thymine-adenine binding protein 4, Rex-1, Reduced Expression Protein 1; FoxD3, forkhead box D3.
Three-Stage Culture for Multilineage Neural Commitment of hES Cells
In the serial culture, the frequency of Musashi-1+ neural precursors was noted to be on the rise progressively, coming up to 89.1 ± 1.7% upon completion of stage 2, when the frequency of SSEA-4+ undifferentiated cells showed a progressive decrease until undetectable level (Fig. 3a), consistent with RT-PCR analysis showing that pluripotency markers octamer binding transcription factor 4 (Oct-4) and alkaline phosphatase (ALP) were undetectable at the end of stage 2, whereas neural precursor markers Nestin and Musashi-1 were substantially expressed during sequential cultures (Fig. 3b, c). Further, in stage 3, the emergence of more mature neural subtypes, such as TuJ1+ neurons, GFAP+ astrocytes, and O4+ oligodendrocytes, which accounted for 23.3 ± 1.5%, 9.4 ± 0.8%, and 5.3 ± 0.6%, respectively, among H14 derivatives (Fig. 3d), resulted in a decrease in Musashi-1 positivity (62.4 ± 1.7% in stage 3) (Fig. 3a), suggesting the third stage favored specification and maturation of derived neural precursors. Consistently, expression of the astroglial marker GFAP, neuronal subtype markers microtubule-associated protein-2 (MAP-2, for postmitotic neurons), and nuclear receptor related protein-1 (Nurr-1, for dopaminergic neurons) was readily detected in stage 3, whereas markers representing mesodermal [v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog (c-kit), Sox-9, and peroxisome proliferator-activated receptor gamma (PPARγ)] and endodermal development [alpha fetoprotein (AFP), glucose transporter 2 (Glut-2), and Amylase] were no longer detectable finally (Fig. 3b), further suggesting the three-stage culture did not favor nonneural commitment. Morphologically, undif-ferentiated hES cells robustly express SSEA-4 (Fig. 3e). Upon induction, central cells in each colony gradually developed a columnar morphology and formed neural tube-like rosettes consisting of Musashi-1+ progenitors and TuJ1+ neurons (Fig. 3f). In stage 2, stem cell derivatives were spontaneously disrupted into small clusters (Fig. 3g), which kept proliferating, and further differentiated into more mature neural cells, as shown by emergence of numerous cells immunoreactive to TuJ1 in stage 3 (Fig. 3h). A substantial number of Nestin+ neural cells were noted (Fig. 3i), among which Synapsin I expression between neurons was apparently present (Fig. 3j). Various subpopulations of dopaminergic neurons (TH+) (Fig. 3k), GABAergic neurons (GABA+) (Fig. 3k), cho-linergic neurons (ChAT+) (Fig. 3l), and serotonergic neurons (Serotonin+) (Fig. 3l) accounted for 35.3 ± 2.0% (vs. control 12.2 ± 0.9%, p < 0.001), 21.4 ± 2.3% (vs. control 8.5 ± 0.9%, p < 0.001), 10.2 ± 1.1% (vs. control 3.1 ± 0.4%, p < 0.001), and 3.4 ± 0.5% (vs. control 0.5 ± 0.1%, p < 0.001), respectively, within TuJ1+ neurons ultimately, significantly higher than those in control cultures. Additionally, at each culture stage, the total number of hES cell derivatives was significantly higher than that in control cultures run in parallel (Fig. 3m), indicating that PA6-involved culture methods also had growth- enhancing effects, consistent with the data showing several transcripts of growth factors could be detected in PA6. Furthermore, after the three-stage culture, derived neurons were able to fire repetitive action potentials (Fig. 3n) and release 61.3 ± 5.8 pmol of dopamine per 106 cells in response to high K+ depolarizing stimuli as assayed by HPLC (n = 8) (Fig. 3o), suggesting a proportion of functional neurons (say, dop-aminergic neurons) were generated after the three-stage culture. RT-PCR showed a progressive loss of expression of pluripotency markers, mesodermal markers, and endo-dermal markers upon completion of three-stage cultures, in contrast to the persistent expression of c-kit, Sox-9, and AFP in control cultures (Fig. 3c). Figure 3d further showed the preponderance in generation of TuJ1+ neurons, GFAP+ astrocytes, and O4+ oligodendrocytes over control cultures. It was suggestive that the three-stage culture generated a hierarchical population of neural lineage cells, but selected against nonneural derivatives.
The three-stage culture favors neural commitment of hES cells (H14). (a) Positivity of Musashi-1 and stage-specific embryonic antigen-4 (SSEA-4) upon completion of each culture stage (n = 12). (b, c) RT-PCR analysis of lineage-specific gene expression profile was performed during hES cell (H14) differentiation. Progressive loss of nonneural marker expression during three-stage culture was obvious (b), in contrast to persistent expression of nonneural markers [v-kit Hardy-Zuckerman 4 feline sarcoma viral onco-gene homolog (c-kit), sex-determining region Y box 9 (Sox-9), and alpha fetoprotein (AFP)] in control cultures run in parallel (c). Coc, coculture; Ctr, control culture. (d) Preponderance over control culture (n = 12) in generation of neural subtypes after the three-stage culture. *p < 0.05 versus Control. (e–h) Immunocytochemistry for indicated markers before induction culture (e) and upon completion of culture stage 1 (f), stage 2 (g), and stage 3 (h–l). Nuclei were counterstained with propidium iodide (PI; e) or 4′,6-diamidino-2-phenylindole (DAPI; f–i, k–l). Scale bars: 5 μm (e), 20 μm (f, l), 10 μm (g–i, k), and 30 μm (j). (m) Cell counting (n = 12) upon completion of each culture stage. Cell number in stage 3 was derived from 2 × 105 discrete cells that were enzymatically dispersed after stage 2 and reseeded onto newly coated 35-mm wells. *p < 0.05 versus Control. (n) Whole-cell voltage-clamp recordings were performed on a newly generated neuron after three-stage culture, which can fire a train of action potentials in response to depolarization. (o) Dopamine can be released by neural cell derivatives (of H14) after three-stage culture, as assayed by high-performance liquid chromatography (HPLC; n = 8), in response to high K+ depolarizing stimuli (top), with dopamine standard as a reference (bottom). TuJ1, antibody for beta III tubulin; GFAP, glial fibrillary acidic protein; O4, oligodendrocyte marker 4; MAP-2, microtubule-associated protein 2; Nurr-1, nuclear receptor related protein-1; Oct-4, octamer binding transcription factor 4; ALP, alkaline phosphatase; PPARγ, peroxisome proliferator-activated receptor γ; Glut-2, glucose transporter 2; PDH, pyruvate dehydrogenase; TH, tyrosine hydroxylase; GABA, γ-aminobutyric acid; ChAT, choline acetyltransferase.
Derived Neural Cell Grafts Improved Cognitive Function of Stroke Rats
It has been documented that the hippocampus is one of the most vulnerable parts of the brain to ischemic injury. Thionin staining demonstrated a remarkable number and extent of pyknotic cells and necrosis in the hippocampal CA1 region at 12 h shortly after ischemic attack (Fig. 4a). Over time, the number of morphologically intact pyramidal neurons with large nuclei, discrete nucleoli, and clear cell periphery in the CA1 region decreased more severely on day 3 poststroke surgery, in contrast to that derived from rats at 12 h and 1 day poststroke surgery. No significant difference in the number of intact pyramidal neurons was found between days 3 and 5 (Fig. 4a). Upon completion of the three-stage culture, derived neural cells at 2 × 105 cells in saline were transplanted bilaterally into the caudate putamen of ischemic rats 3 days poststroke surgery, at which time an extensive loss of 81.1 ± 0.8% of hippocampal CA1 pyramidal neurons was evident. Rats that were able to reach the escape platform in 140 ± 3 s in the Morris water maze were recruited to in vivo studies. After the stroke surgery, 90.9% of the rats survived and displayed signs of drowsiness, paucity of movement, and coma. Six weeks posttransplant, no death was observed, and the rats did not show abnormal behavior or develop dyskinesia, which may suggest no obvious adverse reaction of intrastriatal transplantation of hES cell-derived neural cells. Neural cell-transplanted rats exhibited a progressive reduction of retention time in the water maze over 14 days of behavioral assessment compared to a relatively constant time latency derived from hMSC-transplanted rats and saline-injected ischemic controls (Fig. 4b). On day 14 of behavioral assessment (i.e., day 56 posttransplant), neural cell-transplanted rats took significantly less time (39.5 ± 2.6 s) to reach the escape platform in the maze, in contrast to saline-injected ischemic controls (148.6 ± 5.9 s, p < 0.0001) and even hMSC-transplanted rats (91.5 ± 6.2 s, p < 0.0001). There was no difference in the time latency between cell-transplanted rats (39.5 ± 2.6 s) and sham- operated normal controls (31.8 ± 2.4 s, p = 0.243). However, the swimming speed was comparable between ischemic rats with and without cell grafts, as evident by an experiment conducted in a 150-cm straight path within the water maze on day 14 of testing (Fig. 4c), suggesting the spatial cognitive ability played a critical role in determination of time duration rats spent in the water maze. Moreover, in the spatial probe test, the number of platform location crossings derived from neural cell-transplanted rats was greater than that from saline-injected ischemic controls and hMSC- transplanted rats, respectively (Fig. 4d). A similar scenario was also noted when detecting the amount of time spent in the target quadrant (Fig. 4e). Taken together, these data suggested that hES cell-derived neural cell grafts helped improve the spatial learning and memory ability of ische mic stroke rats, and notably, the therapeutic effects were superior to those derived from hMSC transplantation.
Behavioral assessment. (a) Enumeration of morphologically intact pyramidal neurons within the counting frame of 1 mm × 0.25 mm dimension in every fifth thionin-stained coronal section of the CA1 region in the hippocampus 2.8–3.0 mm posterior to bregma of sham-operated control rats (0 h) and rats at 12 h to 5 days poststroke surgery (n = 12). *p < 0.05. (b) Time latency taken by rats (n = 20) to reach the escape platform in the Morris water maze over 14 days of assessment. (c) Swimming speed was comparable among different groups of rats (n = 20) on day 14 of the Morris water maze test. (d, e) In the spatial probe test, the number of platform location crossings (d; n = 20) and the amount of time spent in the target quadrant (e; n = 20) were recorded, respectively. (b–e) *p < 0.05 versus saline-injected ischemic control and human mesenchymal stromal cell (hMSC)-transplanted rats.
Intrastriatal Cell Grafts Promoted Recovery of Hippocampal Injury
By tracking human neural xenograft in rat brain at 8 weeks posttransplant, histological analysis revealed 52,698 ± 1,995 donor cells, among which 51.2 ± 1.2% was found in the caudate putamen (Fig. 5a), 18.2 ± 0.7% in the corpus callosum, 14.2 ± 0.5% in the cortex, and 8.3 ± 0.3% in the hippocampus (Fig. 5b, c). Donor cells were also evident in the lateral septal nucleus (4.4 ± 0.2%), ventral pallidum (2.3 ± 0.2%), and other regions (1.4 ± 0.2%). In the caudate putamen, donor cells were noted to spread cytoplasmic processes into surrounding striatal parenchyma (Fig. 5a), suggesting donor cells can survive and adapt to the host niche after ischemic injury. Neither intracerebral hemorrhage nor microglial soakage was found, and no teratoma formation was seen. There were significantly greater numbers of intact pyramidal neurons in the hippocampal CA1 regions of stroke rats with (53,383 ± 2,723 CA1 neurons/brain) than without neural cell grafts (saline-injected ischemic controls: 20,648 ± 932 CA1 neurons/brain, p < 0.0001; hMSC-transplanted rats: 41,973 ± 1,417 CA1 neurons/brain, p = 0.0001) (Fig. 5d–h). The densities of CA1 pyramidal neurons were comparable between neural cell-transplanted rats (53,383 ± 2,723 CA1 neurons/brain) and sham-operated normal controls (58,926 ± 2,454 CA1 neurons/brain, p = 0.056) (Fig. 5d–h). Human donor cell migration was obvious (Fig. 5d'–g'), because hNCAM+ cells with neurites spreading into the hippocampal radiatum layer were present in hippocampi of ischemic rats with either hES cell-derived neural cell grafts (Fig. 5g') or hMSC grafts (Fig. 5f'), indicating both types of human donor cells can directly participate in cytoarchitecture reconstruction of injured hippocampus after further differentiation and migration. In hippocampi of neural cell-transplanted rats, these hNCAM+ neurons were found to cluster along the original pyramidal layer and exhibited morphology highly reminiscent of intact pyramidal neurons (Fig. 5g'). Consistently, according to HuN immunostaining, a greater number of donor cells in the hippocampal CA1 region were noted in neural cell-transplanted rats (2,584±158 donor cells in bilateral CA1 regions) than in hMSC-transplanted rats (971 ± 48 donor cells in bilateral CA1 regions, p < 0.0001) (Fig. 5i), implying that hES cell-derived neural cells had a better capacity in promoting hippocampal reconstruction upon cell transplantation therapy, in contrast to hMSC grafts. However, endogenous regeneration in the hippocampus of saline-injected ischemic controls was also detected, although to a lesser extent, because 20,648±932 CA1 neurons/brain were noted, in contrast to 11,553±894 CA1 neurons/brain (p<0.01) 8 weeks previously, suggesting endogenous neurogenesis spontaneously resulted in a slight increase in the number of CA1 neurons by only 0.8-fold over 8 weeks, at which time a severe loss of 65.0±0.3% of hippocampal CA1 neurons was still obvious. On the other hand, as compared to saline-injected ischemic controls, 32,735±2,199 more CA1 neurons were noted in each neural cell- transplanted rat at 8 weeks posttransplant. However, histological analysis revealed that only 2,584±158 cells were immunoreactive to HuN in bilateral CA1 regions (Fig. 5c, i), suggesting exogenous donor cells promoted recovery of cytoarchitecture in the hippocampal CA1 region mainly by enhancing endogenous neurogenesis. In other hip-pocampal regions including CA2, CA3, and dentate gyrus, a total of 1,802±144 hES cell-derived donor cells were detected. Donor cells within the CA1 region accounted for 59.4 ± 1.7% over those migrating into the entire hippocampus. We further speculated whether the progenitor-recruiting effects exerted by donor cells in the hippocampus were correlated with paracrine secretion of bFGF, because intraventricular administration of bFGF has been demonstrated to stimulate not only the proliferation of endogenous progenitors but also their differentiation into neurons (26). To attest this hypothesis, bFGF content was detected in hippocampi isolated from rats at 8 weeks posttransplant. It was noted that the bFGF levels in the hippocampi of neural cell-transplanted rats (171.7±13.0 pg/mg protein) were significantly higher than those derived from saline-injected ischemic controls (114.5±9.3 pg/mg protein, p=0.0004) and hMSC-trans-planted rats (140.6±11.6 pg/mg protein, p=0.041) (Fig. 5j), respectively, suggesting bFGFs may be involved in regulation of endogenous neurogenesis in the hippocampus, which process was promoted by transplantation of hES cell-derived neural cells via paracrine secretion. Consistently, in the histological grading analysis, a lesser degree of lesions of the hippocampal pyramidal layer in neural cell-transplanted rats than saline-injected ischemic controls was also noted (Table 2). Taken together, these data suggested that intrastriatal hES cell-derived neural cell grafts help promote recovery of hippocampal injury after stroke attack mainly by reinforcing endogenous neurogenesis in the hippocampus, which can underlie the improvement in cognitive function of stroke rats after cell transplantation.
Histological grading of the Hippocampal CA1 Regions
hMSC, human mesenchymal stromal cell.
p < 0.05 versus saline-injected ischemic controls and p > 0.05 versus sham-operated normal controls.
Intrastriatal cell grafts participated in the reconstruction of injured hippocampus at 8 weeks posttransplant. (a) Human mitochondria positive (hMito+) cells (brown) on coronal section of caudate putamen at 2.5 mm posterior to injection site of transplant. (b) Thionin-stained hippocampus with the counting frame of 1 mm × 0.25 mm dimension in CA1 region. (c) Human nuclei positive (HuN+) cells (with green nuclei) within the frame of (b). Nuclei were counterstained by DAPI. (d–g) Thionin-stained sections derived from sham-operated normal controls (d), saline-injected ischemic controls (e), hMSC-transplanted rats (f), and neural cell-transplanted rats (g) showing morphology of CA1 pyramidal neurons within the frame of (b). (d'–g') Immunohistochemistry (via DAB method) showing human donor cells [human neural cell adhesion molecule positive (hNCAM+); brown] migrated into hippocampal CA1 regions, corresponding to (d–g), respectively. Scale bars: 50 μm (a, c–g, d'–g'); 500 μm (b). (h, i) Quantification of morphologically intact pyramidal neurons (h; n = 20) and HuN+ human donor cells (i; n = 20) in bilateral CA1 regions of the rat brain according to Abercrombie formula. (j) Enzyme immunoassay for bFGF level in the hippocampus (n = 10). (h–j) *p < 0.05 versus saline-injected ischemic control and hMSC-transplanted rats.
Discussion
If hES cells are to be harnessed effectively in regenerative therapies for neurodegenerative diseases, it will be necessary to develop robust methods for directing neural commitment and simultaneously suppressing nonneural differentiation. In this study, we have presented evidence for a significant role of Notch signaling in promoting primary fate choice in hES cell differentiation. Then we demonstrated the newly designed three-stage culture could exert neurotrophic effects and impose unitary neural lineage commitment on hES cell derivatives, which could effectively improve neurological deficits in ischemic stroke models.
PA6 Directed Neural Lineage Entry of hES Cells
PA6 cells have been shown to support in vitro proliferation of hematopoietic stem cells by expressing stem cell antigen-1 (35). Recent studies have demonstrated that stromal cells derived from murine bone marrow, W-20-17, PA6, and MS5, could not only promote maturation of neuroblastoma cells but also induce neural commitment of mouse embryonic stem cells (12,24,41). In this study it was noted that the neurotrophic effect of PA6 on hES cells was similar to the PA6 stromal cell-derived inducing activity (SDIA) (12,13). PA6-CM was also found to exert the neural induction effect, although to a lesser extent, suggesting the presence of soluble neurotrophic factors secreted by PA6. Besides, PA6 was also noted to enhance the cell number of hES cell derivatives compared to control culture run in parallel. These observations were consistent with results of RT-PCR analysis for transcript expression of growth factors (bFGF, IGF-I, and IGF-II), neurotrophic factors (BDNF, NT-3, CNTF, GDNF, and NGF), and neu-roprotective factors (EPO, VEGF, and CXCL-12) in PA6. In our study, direct coculture with PA6 was believed to be superior in neural induction efficiency to noncontact PA6 coculture or use of PA6-CM. In addition to the effects of neurotrophic and neuroprotective factors secreted by PA6, one important reason for this superiority can be that PA6 also expressed cell surface ligands, namely, Delta and Jagged family members, which resulted in activation of Notch signaling in neighboring hES cells via direct contact. It was reported that Notch signaling could play an important role in developmental conversion of neural precursors to mature neurons (11). By coculturing hES cells directly with PA6, our group indicated that cell–cell interaction through Notch signaling emerged as a key tool for steering hES cells toward the neural lineage and away from nonneural fates. Consistently, Notch inhibition arrested neural differentiation and resulted in commitment toward nonneural lineages irrespective of the presence of PA6 feeders. On the other hand, forced expression of NICD via transfection endowed hES cells with a differentiation propensity to neural lineage cells, suggesting that Notch signaling both mediates and is a limiting factor for neural specification in hES cell cultures. Kawasaki et al. have reported that SDIA via direct contact coculture played a critical role in neural induction (12,13). Our findings suggested that activation of Notch signaling could be one of the key mechanisms underlying SDIA. In addition, Notch signaling has been shown to act in certain contexts to promote neural competence in tissues that would otherwise become nonneural, as reported by an earlier investigation demonstrating that activated Notch promoted generation of sensory patches in the developing inner ear of the chick (6). However, the role of Notch may be of greater significance for cultured hES cells confronted with conflicting autocrine, paracrine, and exogenous differentiation cues, than that in the gastrulating embryo where differentiation signals are tightly restricted, spatially and temporally. Our findings provoked consideration of a possible role for Notch signaling in vertebrate germ layer-specific differentiation. Although neural commitment in vivo is regulated by interaction between family members of bone morphogenetic proteins, fibroblast growth factors, and wingless-related integration (Wnt) signals (39), recent investigations indicated that several yet unidentified signaling events are also requisite for neural specification (20). Our study raised the possibility that Notch may be a key component of this circuit. Taken together, these data suggested the potential of utilizing PA6 coculture to imitate the embryonic niche for hES cell neural induction via Notch signaling.
Induction, Propagation, and Selection of Derived Neural Cells
In normal developmental process, the neural tube of humans was thought to form 3 weeks after fertilization (18). Considering that hES cells were derived from a 5- to 6-day-old blastocysts, in vitro neural induction of hES cells was conducted by coculture with PA6 for 2 weeks to mimic the developmental microenviron-ment and the duration. Subsequent cultures in optimally diluted PA6-CM and supplemented neurobasal medium allowed the propagation and selection of neural lineage cells, facilitating the application of purified neural cells on a large scale (19,32). Because PA6-mediated effects declined gradually with γ-irradiated PA6 cells dying off at the end of stage 1, optimally diluted PA6-CM at 1:1 (which had a better effect on the viability of hES cell derivatives as compared to the ratio of 2:1) was then used to resume the inducing activity during stage 2. In stage 3, the ascorbic acid, Forskolin, and N2 supplement we used in neurobasal medium were believed to favor the survival of postmitotic neurons and promote maturation of neural progenitors (28,42). During the sequential culture, the frequency of Musashi-1+ neural precursors was noted to be on the rise progressively until completion of stage 2, at which time a progressive loss of SSEA-4 positivity until undetectable level was obvious, indicating neuro-trophism was refreshed in stage 2 by PA6-CM and further supported propagation of neural progenitors. These observations were consistent with early studies demonstrating that several soluble growth factors (bFGF, IGF-I, VEGF, and EGF) and neurotrophins (BDNF and NGF) could be secreted by BMSCs (31). More mature neural subtypes such as TuJ1+ neurons (including dopaminer-gic neurons, GABAergic neurons, cholinergic neurons, and serotonergic neurons), GFAP+ astrocytes, and O4+ oligodendrocytes were enriched in stage 3, suggesting culture in stage 3 favored specification and maturation of derived neural precursors. Synapsin I is located only in the brain nerve terminal and is thought to be involved in regulating the number of synaptic vesicles available for release via exocytosis at any one time. Emergence of Synapsin I after the three-stage culture suggested the functionality of newly formed neurons from hES cells. Immunofluorescence study showed a preponderance of the three-stage culture in generation of various neuronal subpopulations and glial cells. Moreover, derived neurons could fire repetitive action potentials and exhibit KCl-evoked synaptic release of the neurotransmitter dop-amine, further confirming the functional characteristics of derived neurons after the three-stage culture. RT-PCR revealed robust expression of transcripts representative of different stages of neural development, but no expression of endoderm- and mesoderm-specific genes and hES cell pluripotency markers upon completion of three-stage culture, in marked contrast to the persistent expression of nonneural markers in control cultures run in parallel. Apparently, the three-stage culture supported neither nonneural differentiation nor maintenance of undiffer-entiated cells. Consistently, in vivo study demonstrated no aberrant development or teratoma formation in brains posttransplant. Compared to previously reported methods (4,8,10), the protocol established in this study evaded the usage of complex and procedural embryoid body-based differentiation methods and offered a simple and efficient procedure to differentiate hES cells into multilineage neural cells without contamination by nonneural derivatives.
Intrastriatal Graft Improved Cognitive Function and Restored Hippocampal Structure
The interaction of neural progenitors, neurons, and glial cells underlines the significance of transplantation of neural lineage cells to the ischemic brain, which is characterized by a severe loss of different neural cell types. Previous studies demonstrated that fully mature neurons survived less efficiently than neural progenitors upon implantation to the brain, which may be related to the susceptibility of mature neurons to mechanical stress and harvesting for transplant (25). Cognitive impairment was found to correlate with the neuronal loss in the hippocampus (37). We conducted cell transplant to rats on day 3 after stroke to explore the therapeutic potential of hES cell-derived neural lineage cells, at which time the cell loss was most severe, accounting for 81.1±0.8% in the hippocampal CA1 region. The neural cell grafts, which were composed of largely neural precursor cells (62.4±1.7%), neurons (23.3±1.5%), and glial cells (9.4±0.8% astrocytes and 5.3±0.6% oligodendrocytes), were shown to improve the cognitive function of stroke rats significantly, as demonstrated by the water maze test and spatial probe test. During normal development of human brain, approximately 6 weeks were required for endogenous neural stem cells to further differentiate into functional neurons following physiological stimuli (3,22,34). We speculated that transplanted hES cell-derived neural precursors may exhibit their plasticity and further differentiate in vivo in response to environmental cues within the rat brain in a similar time duration. Therefore, behavioral assessments of the animals were conducted 6 weeks posttransplant. We further determined whether the reversal of lesion-induced neurological deficits observed in behavioral assessment was attributable to donor cell differentiation or enhanced endogenous neuro-genesis or both.
The hippocampus is the major site of learning and memory functions. To a lesser extent, the caudate putamen is also critical for procedural memory (21). Injured caudate putamen was noted to sustain the interaction between transplanted neural cells and striatal neurons (7,10,15), making itself a target of cell transplantation therapy for cerebral ischemic damage. Histological analysis revealed neither aberrant development nor microglial soakage in host brain, excluding the possibility that the therapeutic effect of transplant had any connection with modulation of inflammation. Tracking of human donor cells in the host brain demonstrated that a majority of hES cell-derived neural cells were localized in the caudate putamen, whereas some were found to migrate into different regions of the brain including the hippocampus. Earlier studies demonstrated that genes of growth factors, neurotrophins, and chemokines, including bFGF, IGF-I, IGF-II, VEGF, EPO, BDNF, CNTF, GDNF, NGF, NT-3, and CXCL-12, were expressed by neural stem cells, microglia, and astro-cytes (1,41). The administration of neurotrophic factors and growth factors was also noted to recruit endogenous progenitors in the periventricular zone near the hippocampus (which is believed to be the reservoir of endogenous neural stem cells) to induce massive regeneration of CA1 pyramidal neurons after ischemic stroke (16,26,43). In this study, we found that hippocampal bFGF level in neural cell-transplanted rats was significantly higher than that in saline-injected ischemic controls or hMSC-transplanted rats, suggesting human neural cell grafts helped generate bFGF during and after migration into the hippocampus of ischemic rats. Consistently, only 20,648±932 CA1 neurons/brain were noted in saline-injected ischemic controls after endogenous neurogenesis for 8 weeks, which only elicited a slight increase in CA1 neuron number by 0.8-fold from initial 11,553 ± 894 CA1 neurons/brain right before saline injection, whereas 53,383 ± 2,723 CA1 neurons/ brain were found in neural cell-transplanted rats, which included 2,584 ± 158 donor cells in bilateral CA1 regions, suggesting that exogenous human neural cell grafts promoted recovery of cytoarchitecture in the hippocam-pal CA1 region mainly by enhancing endogenous neuro-genesis. Thus, data of the pyramidal neuron density and histological grading of the hippocampal CA1 region supported the hypothesis of regeneration by neural stem and progenitor cells present in the periventricular region near the hippocampus subjected to ischemic stress, which can mobilize endogenous neural progenitors (5). Furthermore, a greater number of human donor cells and a higher level of bFGFs were noted in the hippocampus of ischemic rats with hES cell-derived neural cell grafts than with hMSC transplant, which may underlie a better recovery of cognitive functions in ischemic rats with neural cell grafts than with hMSC transplant, as evident by the Morris water maze test and spatial probe test. These data further suggested that donor cell-mediated bFGF production can play a pivotal role in hippocampal reconstruction by recruiting more endogenous progenitors during the regeneration process, which led to a remarkable recovery in CA1 pyramidal neuron density of the hippocampus. On the other hand, hNCAM+ cells with neurites were evident in the hippocampus, indicating that hES cell-derived neural cell grafts can also directly participate in the cytoarchitecture reconstruction of injured hippocampus. Here, the presence of donor cells was demonstrated to have a synergistic effect with ischemic stress on mobilization and recruitment of endogenous neural progenitors. In addition to the CA1 region, a total of 1,802 ± 144 donor cells were detected in other hippocampal regions including CA2, CA3, and dentate gyrus. Donor cells within the CA1 region accounted for 59.4 ± 1.7% over those migrating into the entire hippocampus. A comparable degree of neuronal density in the hippocampal CA1 region was noted between neural cell-transplanted rats and sham-operated normal controls at 8 weeks posttransplant, whereas a significant cell loss of 65.0 ± 0.3% of CA1 neurons was still evident in saline-injected ischemic controls, despite an insufficient neuro-genesis in the ischemic hippocampus. Additionally, it has been documented that even after intraventricular infusion of growth factors (such as bFGF and EGF) from day 2 to day 5 after ischemia attack, CA1 pyramidal neurons were progressively degenerated in the initial week, and most of the cells were subjected to apoptosis (26). The extent of neuronal loss was indistinguishable at day 7 after ischemia attack between growth factor-treated and untreated rats, indicating that intraventricular administration of growth factors exhibited no substantial neuroprotective effects on dying neurons (26). Likewise, in this study, what were transplanted intrastriatally were growth factor-secreting neural lineage cells, which were noted to promote hip-pocampal reconstruction at 8 weeks posttransplant. Thus, it can be ruled out that newly generated CA1 neurons in the reconstructed hippocampus were derived from prevention of initial loss of hippocampal CA1 neurons by cerebral transplant. Taken together, it was suggestive that the small number of hES cell-derived neural cells engrafted to the hippocampus could enhance endogenous regeneration and restore pyramidal neuron density in the hippocampal CA1 region, which may be attributable to generation of donor cell-derived growth factors and neurotrophins (such as bFGF), as well as direct cell replenishment after donor cell differentiation.
In summary, this study suggested that by activating Notch signaling during PA6 coculture, the three-stage culture could exert neurotrophic effects and impose unitary neural lineage commitment on hES cell derivatives, which could effectively improve neurological deficits in ischemic stroke models. After transplantation into the caudate puta-men, neural cell grafts not only enabled migration, homing, and engraftment to the injured brain parenchyma, but also could enhance endogenous regeneration and restore pyramidal neuron density in the hippocampus.
Footnotes
Acknowledgments
The research work is supported by Shanghai Natural Science Foundation of China (grant No. 12ZR1425200) and National Natural Science Foundation of China (grant No. 30970659). The authors declare no conflict of interest.