|Home | About | Journals | Submit | Contact Us | Français|
Functionally diversified neuronal populations have now been efficiently generated from human pluripotent stem cells (hPSCs). However, directed differentiation of hPSCs to functional astroglial subtypes remains elusive. In this study, hPSCs were successfully directed to nearly uniform populations of immature astrocytes in large quantities (>90% S100β+ and GFAP+). The immature human astrocytes exhibit similar gene expression patterns as primary astrocytes, display functional properties such as glutamate uptake and promotion of synaptogenesis, and become mature astrocytes by forming connections with blood vessels following transplantation into the mouse brain. Furthermore, hPSC-derived neuroepithelia, patterned to rostral-caudal and dorsal-ventral identities with the same morphogens used for neuronal subtype specification, generate immature astrocytes that carry distinct homeodomain transcription factors and display phenotypic differences. These human astroglial progenitors and immature astrocytes will be instrumental for studying astrocytes in brain development and function, for revealing their roles in disease processes, and for developing novel treatments for neurological disorders.
Astroglial cells are the most abundant cell type in our brain and spinal cord and are now understood to be as important as neurons for brain function1, 2. During development, astroglial progenitors are specified after neurogenesis, although the identity of these progenitors is not well defined due to lack of reliable markers3, 4. These progenitors differentiate to immature astrocytes which are essential for the formation of functional synapses5, 6. In the adult brain, mature astrocytes insulate synapses and remove excessive transmitters, e.g., glutamate, released during neural excitation, thus preventing excitotoxicity7. Astrocytes are crucial for maintaining a homeostatic environment for the healthy brain by supporting neurovascular coupling at the blood brain barrier (BBB), regulating blood flow8, and supplying energy metabolites throughout the brain9. Abnormalities in astroglial cells are implicated in a number of human pathologies including astrocytomas, epilepsy10, Alexander disease11 and neurodegenerative diseases12, 13. Thus, understanding how to regulate generation, maintenance, and functions of human astroglial cells will likely benefit the treatment of a range of neurological injuries and diseases.
Astrocytes are heterogeneous in many aspects including morphology, growth rates14, gene expression profiles15, 16, electrophysiological properties17, gap junction coupling, and calcium wave propagation dynamics18, 19. However, how these various astroglial phenotypes are gained is largely unknown. In the mouse and chick spinal cord, homeodomain and helix-loop-helix (HLH) transcription factors are expressed in progenitors of specific dorsal-ventral domains and genetic alterations of these factors shift the gene expression pattern of astrocytes generated from these domains20–22. This suggests that astroglial diversity may be determined through regional patterning of progenitor cells. However, it remains unknown if astroglial subtypes may be generated by simply patterning neural stem/progenitor cells, especially human stem cells.
In the present study, we found that immature astrocytes appear following neurogenesis in our chemically defined differentiation system of human pluripotent stem cells (hPSCs), including embryonic (hESCs) and induced pluripotent stem cells (iPSCs), and exhibit the typical cellular, molecular, and functional characteristics of cells that are born in the brain. Their progenitors can expand in culture for a long term, thus producing large numbers of immature astrocytes. Importantly, we discovered that regionally and functionally distinct human astroglial subtypes are induced by patterning neuroepithelial cells (NE) at an early stage, and they maintain their identities following transplantation into ectopic mouse brain regions, providing a possible cellular source for regenerative medicine.
The generation of astroglial cells follows neurogenesis during vertebrate development. We hypothesized that hPSC-derived neural progenitors, after temporal expansion, become gliogenic and generate astroglia under conditions that facilitate glial differentiation (Fig. 1a). hPSCs were directed to NE, followed by differentiation to neural progenitors from days 10 to 21 in the presence of either the posterior patterning molecule, retinoic acid (RA, 0.5 μM), or the anterior patterning morphogen, fibroblast growth factor 8 (FGF8, 50 ng/ml)23, 24 in order to examine whether early specification will lead to divergent astroglial subtypes (see below). Differentiation of day-30 RA-treated progenitors from the H9 hESC line showed that a small population of cells (7.7% ± 1.5) were positive for S100β and hardly any cells (<0.1%) were GFAP+, the two common astroglial markers. Most other cells remained columnar epithelial shaped, indicative of progenitor identity, whereas some exhibited neuronal phenotypes and were positive for βIII-tubulin (4.4% ± 0.8), which decreased over time. At increasing time periods, the number of S100β+ cells continued to increase, and displayed the typical stellate morphology of astroglia (Fig. 1b, c). Similarly, GFAP-expressing cells began to appear after 8 weeks, and the percentages increased over time (Fig. 1b, c). The same trend was observed with the H7 hESC line and the (IMR90)-4 iPSC line (Supplementary Fig. 3a, b). The GFAP+ cells always co-labeled with S100β. A similar progression of astroglial marker expression was observed in cells that were specified with FGF8, but significantly slower at certain time points (Fig. 1c). Aldh1L1, a newly identified marker for the astroglial lineage25, but not NG2, a proteoglycan expressed by NG2 cells26, was also detected in GFAP+ cells (Fig. 1d). Western blotting analysis of day-180 immature astrocytes demonstrated the expression of GFAP and the astrocyte specific glutamate transporter GLT-1 (Fig. 1e). These results further confirm the astroglial identity. Leukemia inhibitory factor (LIF) had a similar effect as ciliary neurotrophic factor (CNTF) in increasing the proportion of GFAP+ cells after treatment of the day-180 progenitors for 6 days (Supplementary Fig. 1a>).
To discern progenitors during astroglial differentiation, we examined putative glial progenitor markers, including A2B5, CD4427, and NFIA28. A2B5 was expressed in a subset of S100β+ cells at days 30 and 90 (9.8% ± 3.2 and 8.7% ± 2.1) and the percentage decreased as cells became GFAP+. CD44, however, was not observed until day-60, and by 90 days, 79.5 ± 1.9% of S100β+ cells expressed CD44 (Fig. 1f and Supplementary Fig. 3c). NFIA, which was completely absent in PAX6+ NE at day-11, began to express by day-30 as cells down-regulated PAX6. Day-180 astroglial cells all expressed high levels of NFIA (Fig. 1g). Thus, astroglial progenitors or immature astrocytes can be identified by expression of NFIA-S100β at 4–8 weeks after hPSC differentiation and more mature astrocytes can be marked additionally by CD44-GFAP after 8–12 weeks of differentiation. This expression pattern is remarkably similar to that of in vivo human development29, 30. Quantitative RT-PCR analysis indicated that the hPSC-derived day-210 astroglia expressed high levels of additional astroglial genes, including NF1X, CHL1, GLAST, GLT1, and aquaporin 4 (Supplementary Fig. 1c). Taken together, these results not only confirm the astroglial identity but also suggest functional attributes of the hPSC-derived astroglia.
The hPSC-derived astroglial progenitors were expanded continuously for at least 8 months and survived freeze-thaw cycles. BrdU pulse-labeling for 10 hours in day-60 progenitors revealed a higher percentage of cells undergoing DNA synthesis, a correlate of cellular proliferation, in FGF8-specified progenitors (74.2% ± 2.0, n = 6) compared to RA-specified progenitors (50.6% ± 6.1) (Fig. 1h). At 6 months, the extent of BrdU labeling decreased, the two groups exhibited a similar proliferation rate, and removal of growth factors inhibited DNA synthesis (Fig. 1h). In the presence of EGF and FGF2, the RA-specified astroglial progenitors maximally expanded at a rate of 7.6 ± 1.2 times every 6 days between 4–5 months when seeded at 24,000 cells/cm2. Although hESCs and iPSCs exhibit slightly different efficiency in NE generation31, differentiation of NE and expansion of astroglial progenitors from these various hPSC lines (H9, H7, (IMR90)-4)) showed remarkable similar efficiencies. If one hPSC is differentiated to NE, converted to glial progenitors, and then expanded in suspension, an estimation of at least 2.8 × 1012 immature astrocytes can be generated by 6 months, even after excluding cell loss during the procedure. Therefore, this procedure allows generation of large quantities of astroglia.
Like neurons, astrocytes in different regions of the CNS exhibit differential phenotypes. We hypothesized that regionally distinct astroglia may be specified from hPSCs in the same way as for neuronal types through patterning of NE and subsequent differentiation (Fig. 2a). At day-30 of differentiation from H9 hESCs, nearly all of the cells patterned with RA from days 10–21 expressed the hindbrain-spinal cord specific transcription factor Hoxb4 (98.6% ± 0.7, n = 5) and very few expressed the mid-forebrain marker Otx2 (3.1% ± 0.8). The FGF8 treated cells, similar to those treated without FGF832, expressed Otx2 (95.4% ± 3.0) and none expressed Hoxb4. This expression pattern of homeodomain transcription factors persisted as cells began to express astroglial markers S100β and GFAP (Fig. 2b–d and Supplementary Fig. 2a), but with a slight decrease of Otx2 in GFAP+ astrocytes. A similar expression pattern of homeodomain transcription factors was observed in primary astrocytes isolated from the mouse embryonic brain, spinal cord (Supplementary Fig. 2c), and in astrocytes specified from iPSCs (Supplementary Fig. 3). Quantitative RT-PCR analysis confirmed differential expression of additional homeodomain genes and functional genes in the anterior and posterior immature astrocytes (Supplementary Fig. 2b), signaling potential functional diversities.
To determine whether astroglia with a dorsal-ventral distinction can also be specified, NE were treated with or without the ventralizing factor, sonic hedgehog (SHH, 500 ng/ml). In the absence of morphogens, hESC-derived neural progenitors exhibit a dorsal telencephalic phenotype33; none of the astroglial progenitors expressed the ventral markers Nkx2.1 (Fig. 2e, f). In contrast, the majority of astroglial progenitors (day-30) differentiated from the SHH-treated NE were labeled for Nkx2.1, though Nkx2.1 was noticeably decreased upon S100β expression (Fig. 2f). Regional marker expression was confirmed in subsets of GFAP+ astrocytes in P1 mouse brain and spinal cord sections (Supplementary Fig. 2d). Together, our results indicate that region-specific astroglia can be specified from hPSCs in the same way as for region-specific neuronal types.
In contrast to neurons, astroglia are described as passive cells because they cannot generate action potentials and their voltage gated currents decrease during maturation34. Whole cell patch clamp recording of both RA-specified (n = 18) and FGF8-specified astroglia (n = 10, expanded for 120 days from line H9 and treated with CNTF for 7 days) each displayed a voltage dependent initial rapid outward current that was inactivated within 100 ms, a lower sustained current, and never displayed action potentials in current clamp. This passive electrophysiological property resembles that observed in immature primary mouse astrocytes34. To determine whether neuronal signaling promotes astroglial maturation, 120-day red fluorescent-labeled astroglia were cultured either with or without hESC-derived neurons (day-28) for 2 weeks (Figure 3a). Co-cultured astroglia displayed a change in morphology and a significantly decreased transient outward current (n = 10 with and without neurons), suggesting maturation of astroglial cells in vitro.
One critical function of astrocytes is signaling and buffering of neurotransmitters released during neuronal excitation. To determine if hESC-derived immature astrocytes contain functional glutamate receptors, AMPA or L-glutamate (100 μM) was applied in the absence and presence of CNQX and AP5 (20 μM). An inward current was activated upon addition of AMPA, which was completely blocked in the presence of CNQX and AP5 (RA-specified n = 5, FGF8-specified n = 5) (Fig. 3b). In contrast, the L-glutamate response was partially blocked. L-glutamate induced a large inward current in all cells tested (RA-specified n = 8, FGF8-specified n = 8). 100 μM of D, L-aspartate produced a similar inward current (not shown). When the GLT-1 specific inhibitor dihydrokainate (DHK, 100μM) was applied to the same cells for 1 minute before L-glutamate administration, the inward current was significantly smaller, suggesting that glutamate-induced inward currents depend on the function of glutamate transporters. Addition of the general glutamate transporter blocker L-serine-o-sulfate (SOS, 100μM) further decreased the current (Fig. 3b). No significant differences in induced currents were observed between the two subtypes of immature astrocytes due to the high variability of peak currents between cells. Furthermore, hESC-derived immature anterior astrocytes were competent to uptake glutamate from media over time at a rate significantly increased compared to that of HEK293 cells, but not in the presence of the glutamate transporter inhibitor PDC nor sodium-free media (n = 3) (Fig. 3c). Together, these results indicate that the hESC-differentiated immature astrocytes possess functional glutamate receptors and transporters.
Propagation of calcium waves across astrocytes, activated by various stimuli, plays a critical role in glial and neuron-glial communication35 and calcium wave dynamics are different in regionally distinct astrocytes18, 19. Fluorescent intensity of Fluo-4 loaded hESC(line H9)-derived 6-month immature astrocytes was observed during mechanical stimulation of a small area <20 μm. In all cells tested, stimulation induced an intra- and inter-cellular calcium wave in the vicinity, which traveled outward to adjacent cells (Fig. 3d and Supplementary movie 1). Wave propagation was found to be dependent upon extracellular ATP signaling, because it was reduced by the presence of apyrase, an ATP hydrolytic enzyme, but not the gap junction blocker carbenoxolone (Supplementary Fig. 4). The distance traveled by the calcium waves was significantly different between RA- and FGF8-specified astroglia (Fig. 3d). Application of the InsP3 receptor blocker 2-APB36 reduced the calcium wave distance for both RA-and FGF8-astroglial groups (Fig. 3d). Astroglia generated without FGF8 treatment, which also display anterior characteristics (Supplementary Fig. 2b), propagated waves in similar distances as that of the FGF8-specified astroglia (30 seconds; 106.5 ± 2.3 μm, p = 0.1369) and significantly different from RA-specified astroglia (p < 0.001) (not shown). Thus, similar to astrocytes in vivo, hESC-derived immature astrocytes are competent for network communication.
Another function of astrocytes, especially immature astrocytes, is promotion of synaptogenesis5, 6, 37. To determine if hPSC-derived immature astrocytes possess the same function, we cultured hESC-derived (day-21) neuronal progenitors alone, or in direct contact with immature astrocytes. The density of synapsin 1+ puncta along the neurites was significantly increased in neurons after 3 weeks of direct co-culture on hESC-derived (anterior) immature astrocytes as compared to those without (Fig. 3e), suggesting the ability of hPSC-astrocytes to modulate synaptogenesis.
To determine if hPSC-differentiated astroglia maintain their identity in vivo, dissociated immature astrocytes, treated for one week with CNTF after 6 months of differentiation from line H9, were transplanted into the lateral ventricles of neonatal mice (Fig. 4a). Thirty (n = 2 for RA group, n = 2 for FGF8 group) and one hundred days (n = 3 for RA group, n = 4 for FGF8 group) after transplantation, grafted human cells, identified by human nuclear protein (hNuc), were observed as clusters adjacent to the lateral ventricles and as a stream of migrating cells in the corpus callosum (CC). In every brain section examined, all of the transplanted hNuc+ cells expressed a high level of GFAP (Fig. 4b). Cells that entered the CC exhibited an elongated shape in parallel with axons. Astroglial progenitor clusters (after 6 months without CNTF treatment, H9 line) transplanted directly into the adult mouse hippocampus also survived post-engraftment (2 weeks; n = 4 for RA group and n = 2 for FGF8 group, 6 weeks n = 4 for both groups) and continued to express GFAP (Fig. 4d, e), but not the neuronal marker βIII-tubulin+, even in the neurogenic dentate gyrus (Fig. 4f). This result indicates that the hESC-derived astroglial progenitors or immature astrocytes retain their identity in vivo.
Immunostaining of the ventricle transplanted cells with homeodomain transcription factors showed that all hNuc+ cells in clusters and those migrating through the CC were positive for Hoxb4 (day-30 = 70/70, day-100 = 65/65), but not for Otx2, and no hNuc−/GFAP+ endogenous astrocytes in this region co-labeled with Hoxb4 for the RA-specified astroglial group (Fig. 4c). In the brains transplanted with FGF8-specified immature astrocytes, all hNuc+ cells were positive for Otx2 (day-100 = 52/52), but not Hoxb4, and endogenous Otx2+ astrocytes could be observed (Fig. 4c). Therefore, the regional specificity of astrocytes, which is acquired during early in vitro differentiation, is retained and not altered by the ectopic in vivo brain environment.
To determine if human astroglia may mature and integrate into the mouse brain, we examined the possibility of close apposition of human astrocytes to blood vessels, a sign of contribution or signaling of astrocytes to the BBB38. Since BBB formation requires coordinated development of blood vessels and astrocytes, we transplanted hESC-derived neural progenitors (day-21) to the ventricles of neonatal mouse brain and examined for connections of human GFAP+ fibers directly with vessels. Few GFAP+ cells are generated from grafted hESC-derived NE within 3 months, as previously described39. However, 6 months post transplantation, a number of GFAP+/hNuc+ cells project elongated fibers to blood vessels and end in direct contact at end feet (Fig. 4g, h, and Supplementary movie 2). In contrast, endogenous mouse astrocytes are smaller and the somas directly surround vessels. This distinction is similar to the phenomena observed in adult human and rodent tissues40. These results indicate that the hPSC-derived astroglia can indeed mature and participate in BBB structure formation and, interestingly, the in vitro generated astrocytes exhibit some unique features of human astrocytes even in the mouse brain.
We have developed a chemically defined system for efficiently directing hPSC-derived neural progenitors to a nearly uniform population of astroglial progenitors and immature astrocytes through long term expansion of dissociated progenitors in the presence of FGF2 and EGF. The in vitro produced human immature astrocytes possess functional hallmarks of primary astrocytes, including responses to glutamate, propagation of calcium waves, promotion of synaptogenesis, and participation in BBB formation. Importantly, we provide evidence for the first time that regionally and functionally diversified astroglial subtypes can be efficiently specified through patterning of early NE with the same set of morphogens used for generating neuronal subtypes. Furthermore, both the astroglial identity and regional characteristics are maintained after long-term in vitro expansion and after transplantation into the mouse brain. Thus, the enriched human immature astrocytes, which can be generated in large quantities from an almost unlimited source of stem cells, provide an important platform for basic research, drug discovery, and regenerative biology.
Regional and functional heterogeneity of astroglia has been well recognized41. However, it remains unclear whether this heterogeneity is attributed to intrinsic developmental programs or exclusively from adaptation to environmental cues. We have shown here that the regional identity of NE, specified by a single morphogen, is maintained during subsequent differentiation. Comparisons of astroglial subtypes revealed differential onset of S100β and GFAP, cellular proliferation, gene expression, and calcium wave propagation. Thus, the regional identity of astroglial cells is at least partially determined during neuroepithelial patterning, as has been previously postulated22, 42, 43 and we propose that combinatorial morphogen patterning of NE may lead to generation of increasingly diversified astroglial subtypes (Fig. 5). This hypothesis suggests that regionalized neural progenitors migrate to target brain regions, then give rise to neurons before becoming astroglial cells, which may explain why we do not observe clear differential migration patterns of regionalized astroglia. Nevertheless, the astroglial progenitors and immature astrocytes tend to target to the white matter.
Functional properties of astroglia, particularly those of regional astroglia, are generally considered the outcome of responses to local brain environments44. Given the minimal presence of neurons and absence of immune cells in our culture system, our data suggest that at least some functions, described here, are intrinsically endowed when astroglia are born. It should be pointed out that the astroglial cells in our culture system correspond to those in an early stage of the developing human brain. In the developed brain, astrocytes may well acquire additional mature functions, especially in response to local cues45. Furthermore, both the astroglial identity and regional characteristics are maintained after long-term in vitro expansion of their progenitors and after transplantation into the mouse brain environment. The intimate connections of hPSC-derived astrocytes with blood vessels suggest functional ability in vivo, though further work is needed to determine whether these cells can affect brain signaling or behavior.
Astrocytes in the human nervous system appear more complex than those in lower mammals40. Our present findings indicate that differentiation of human astroglia is substantially slower than that from mouse ESCs (which usually takes about 2 weeks), corresponding to astroglial development in the human brain. This prolonged development explains why human PSC-derived astroglial cells in vitro exhibit characteristics of immature rodent astrocytes in vivo, including the presence of voltage gated currents34 and induction of neuronal maturation37. Nevertheless, these immature astrocytes appear to mature over time when co-cultured with neurons or transplanted into the brain to participate in BBB formation. Our ability to derive and expand an enriched population of astroglial progenitors, as well as differentiating them to immature astrocytes, open up a new avenue for studying the role of human astrocytes in the normal and diseased brain and for the development of transplantation therapy in neurological diseases such as amyotrophic lateral sclerosis (ALS), as suggested previously with mouse primary astrocytes46. In addition, astroglial cells derived from patient specific iPSCs offer yet another novel tool for therapeutic discovery.
hESCs (line H9, passages 20–30; H7, passages 35–40) and iPSCs ((IMR90)-4)47 were cultured as previously described32. Briefly, cells were passaged weekly by dispase (1mg/ml, Gibco) treatment and by plating on a layer of irradiated mouse embryonic fibroblasts. The hPSC medium consisted of Dulbecco's modified Eagle's medium (DMEM)/F12, 20% Knockout serum replacement, 0.1 mM β-mercaptoethanol, 1 mM L-glutamine, nonessential amino acids (Gibco) and 4 ng/ml FGF-2 (R&D Systems).
hPSCs were first differentiated to neuroepithelia for 10 days, as detailed elsewhere48. From days 10–21, cells were treated with either 0.5 μM of retinoic acid (RA, Sigma), 50 ng/ml of FGF8 (Peprotech Inc.), or 500 ng/ml of sonic hedgehog (SHH, R and D Systems). Neural progenitors in a form of rosettes were blown off by a pipette at day-15 and expanded in a suspension culture containing EGF (Sigma) and FGF2 (R and D Systems, 10 ng/ml) starting from day-21. The neural progenitor spheres were disaggregated into small clusters with a Pasteur pipette to reduce cell contact, thus promoting gliogenesis instead of neurogenesis49. For astroglial differentiation, progenitor spheres were dissociated with accutase (Chemicon) to single cells, attached with a laminin substrate in the presence of CNTF (10 ng/ml, R&D System), LIF (10 ng/ml, Millipore), or fetal bovine serum (FBS, 10%, Gibco). Cells were additionally passaged to coverslips for immunocytochemistry.
For immunocytochemistry, fixed cells were stained as previously described32. For quantification of each sample (n), 10 optic fields were chosen randomly under the fluorescent filter for nuclear staining throughout the coverslips in areas which contained a similar density of Hoechst+ cells and the total cells were counted with Metamorph software. The fluorescent filters were shifted during imaging to count the cells labeled by different antibodies in the same field in the same manner. The quantitative data were repeated twice or more in different cultures or those from different cell lines. For Western blotting, 30 μg of cell lysates were resolved with SDS-PAGE and transferred to nitrocellulose membranes. Detection was performed with horseradish peroxidase-conjugated secondary antibodies and the ECL system (Thermo Scientific). Primary antibodies are listed in Supplementary Table 1.
Cells were attached on coverslips for 48 hours and treated with 0.2 μM 5-Bromo-2'-deoxyuridine (BrdU) for 10 hours. Cells were fixed with methanol for 10 minutes, followed by incubation with 2 N HCL for 20 minutes. Cells were immunostained with the BrdU antibody as described above.
cDNA was prepared using Superscript III First-Strand Synthesis System (Invitrogen). qRT-PCR was performed with Power SYBR Green PCR Master Mix (Applied Biosciences) on a StepOnePlus System with standard parameters and values were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH), n = 3 for each. The primer sets used are listed in Supplementary Table 2.
Primary astrocyte cultures were prepared from E13.5 timed pregnant CF-1 mice (Charles River). Brain regions were surgically dissected based on anatomical markers, dissociated with trypsin (Invitrogen), and cultured in DMEM + 10% FBS until experimentation. All experiments were performed with cells between passages 2–5.
Whole cell patch clamp recordings were performed and analyzed as previously described6. During the procedure, cells were bathed in a modified Hank's Buffered Saline Solution (HBSS) that contained (in mM): 140 NaCl, 3 KCl, 2 CaCl2, 1 MgCl2, 15 HEPES and 23 glucose (pH 7.4, 300 mOsm). The following chemicals were applied through a gravity-fed drug barrel system: 4-aminopyridine (4-AP, 1mM), L-glutamate (100 μM), alpha-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA) (100 μM), (2R)-amino-5-phosphonopentanoate/6-cyano-7-nitroquinoxaline-2,3-dione (AP5/CNQX) (20 μM), D,L-aspartate (100 μM), dihydrokainic acid (DHK, 100 μM), and L-serine-o-sulfate (SOS, 100 μM ), all obtained from Sigma. For co-culture experiments, immature astrocytes were infected with lentiviral particles to express transgenic mCherry protein driven by the CMV promoter, and co-cultured without or with day-28 hESC-derived neurons in neural media as previously described6.
The method for measuring the decrease of glutamate over time was modified from Abe et al.50 using the Glutamine/Glutamate Determination Kit (Sigma). Anterior astroglia were differentiated for 7 months, plated at a concentration of 20,000 cells per well in a 48-well plate, and culture for an additional 7 days in the presence of CNTF. Before the assay, cultures were equilibrated in HBSS buffer for 10 minutes. 50 uM L-glutamate solutions were prepared with either HBSS +/− L-trans-pyrrolidine-2,4-dicarboxylic acid (PDC, 1mM, Sigma) or Na+ free HBSS (modified by replacing equimolar KCL with NaCl) and incubated with the cells. After various time periods, the glutamate concentration remaining in the media was measured at 340 nm following the enzymatic reaction. HEK293 cells, which do not significantly uptake glutamate compared to primary astrocytes, were used as controls. After subtraction of the blanks (0 glutamate added), the decrease in the media, or uptake of glutamate by cells, was reported as uM of glutamate per ug of protein after being normalized to the total protein in each well. The protein content was determined by a BCA protein assay (Pierce).
Astroglial cells were incubated at room temperature with HBSS and 1 μl each of Fluor-4 (4 μM, Invitrogen) and Pluronic F-127 (0.01%, Invitrogen) for 30 minutes. Cells were washed with HBSS and imaged with an immersion objective on a confocal microscope (described below). Calcium wave induction was performed by mechanical stimulation with a flame polished pulled glass pipette controlled manually with a micromanipulator (WPI Inc.). Five random fields were chosen under microscopy and averaged for each n. Fluorescent images were taken every 2 seconds with or without 2-aminoethoxydiphenyl borate (2-APB, Tocris, 100 μM), carbenoxolone (Sigma, 100 μM), or apyrase (Sigma, 50 units/ml). Calcium wave distances were quantified using Metamorph software. Post-fixation nuclear counting confirmed similar plating densities of astrocytes (RA = 121.3 ± 7.2, FGF8 = 125.7 ± 26.7 per 428 μm2, p = 0.88).
Human ESC-derived neural progenitors (day-21) were cultured in the neuronal differentiation media alone or directly on a layer of hESC-derived immature astrocytes (10,000 cells/cm2) for 3 weeks, similar as previously described6. The cultures were then fixed with 4% paraformaldehyde and immunostained for βIII-tubulin and synapsin 1. Neurons with elongated neurites were chosen by visualization of βIII-tubulin under confocal microscopy. Fluorescent filters were then switched for Synapsin 1 imaging, and the synapsin 1+ puncta along the βIII-tubulin+ neurites were counted with ImageJ software. The results were expressed as the number of puncta per unit neurite length.
Transplantation studies were conducted following protocols approved by the Animal Care and Use Committees at the University of Wisconsin-Madison. Cells were prepared for transplantation in artificial cerebral spinal fluid (Harvard Apparatus) at a concentration of 50,000 cells/μl. For ventricle transplants, 2 μl of the cell suspension was injected 1 mm from the midline between the Bregma and Lambda and 1 mm deep into the anterior lateral ventricles of both hemispheres of severe combined immunodeficiency (SCID)-beige (Taconic) P1 mice. For transplantation into the adult SCID mouse hippocampus, 2 ul of cells were injected with the sterotaxic coordinates of −2.46 mm for anterior-posterior, ±2 mm for lateral, and −2.25 mm for dorsal-ventral. At various time periods after transplantation, animals were anesthetized, perfused with 4% paraformaldehyde, and processed for immunohistochemistry with antibodies listed in Supplementary Table 2. Sections were imaged with a confocal microscope (Nikon, D-Eclipse C1), and EZ-C1 software (version 3.5).
Results were expressed as mean ± S.E.M. For quantification, each data set (n) was generated from a separate passage of hPSCs. n = 3 unless noted differently. Fields were randomly selected and p values were calculated by unpaired t test. * = p ≤ 0.05.
The authors thank Dr. Albee Messing for critical reading of the manuscript. This study was supported by the ALS Association, National Institute of Neurological Disorders and Stroke (NS045926, NS057778, NS064578), National MS Society (NMSS TR-3761), NYSTEM (C024406), Bleser Family Foundation, Busta Family Foundation, Neuroscience Training Program (T32 GM007507) and partly by a core grant to the Waisman Center from the National Institute of Child Health and Human Development (P30 HD03352).