Immunocytochemistry revealed four large subsets of cells that expressed IR for nestin (NSCs, NEPs), A2B5 (GRPs and <1% HNPs), GFAP (astrocytes and astrocyte precursors), or E-NCAM and β-III tubulin (HNPs). Acutely passaged, E-NCAM/β-III tubulin+ HNPs coexpressed the early neuronal marker MAP2 (), and BrdU-uptake experiments indicated that many of these cells were mitotic (BrdU+ = 30% at 2 DIC, 5% at 6 DIC). E-NCAM/β-III tubulin+ HNPs matured rapidly in culture (2 weeks) and differentiated into a heterogeneous population of neurons that responded to subsets of applied neurotransmitters and synthesized neurotransmitters and were electrically active. Our data demonstrate that HNPs persist in the developing human cortex until at least 22 weeks of gestation and that E-NCAM IR can be used to identify this neuronal precursor population.
Overall the properties of the HNPs appeared similar to those of the rodent NRPs. As with rodent NRPs, HNPs are mitotic cells that express the early neuronal markers E-NCAM and β-III tubulin; can become post-mitotic in culture; and, upon subsequent differentiation, synthesize mature neuronal markers such as neurofilament, synaptophysin, ChAT, TH, and certain voltage-gated ion channels and ligand-gated neurotransmitter receptors. Also as with the rodent NRPs, HNPs require the presence of the mitogen bFGF to promote division, proliferation, and survival (Kalyani et al., 1998
). After differentiation of HNPs for 14–42 days in vitro, the cells took on characteristic neuronal morphologies, similarly to the rodent NRPs (). Synaptophysin staining suggested that the observed cell–cell contacts formed functional synapses, but additional experiments utilizing dual patch-clamp recording, synaptosomal preparations, or FM-143 vesicle imaging will be required to confirm this hypothesis. Ac-quisition of ligand-gated receptor function followed the arrest of HNP mitosis and coincided with differentiation and expression of mature neuronal markers (RT-97, synaptophysin, ChAT, and TH). Individual neurons responded to one or more neurotransmitters, indicating a heterogeneous pattern of receptor expression. This heterogeneity was more limited than that observed in differentiating rat NRPs (Kalyani et al., 1998
). Differentiated neurons expressed larger whole-cell, voltage-gated, ion currents than acutely passaged HNPs, but, because the neurons were also larger than the acute HNPs, the current densities did not differ significantly. All of the neurons expressed outward K+
currents that included a varying mix of delayed-rectifier and transient A-type currents based on kinetic observations. Some of the HNPs displayed N-shaped, outward current–voltage relationships that suggested the expression of Ca2+
channels. Some also expressed slowly activating and inactivating inward currents that resembled voltage-gated Ca2+
currents. Calcium imaging experiments revealed responses to elevated K+
and supported the notion that HNPs express voltage-gated Ca2+
channels. The majority of differentiated E-NCAM+
cells were capable of firing action potentials. This is similar to the case in rodent NRPs (Kalyani et. al., 1998
) but different from E-NCAM−
cultured HNPs (Piper et al., 2000
), which could not fire action potentials.
Despite these overall similarities, several important differences were noted. Maturation was rapid (2 weeks) compared to that of human NEP cell cultures (>4 weeks; Piper et al., 2000
) but was much slower than that in rodent cultures (5 days; Kalyani et al., 1998
; Mujtaba et al., 1999
) or in immortalized human cell lines (4 days; Li et al., 2000
). Rodent NRPs expressed nestin, and virtually all BrdU-incorporating NRPs were nestin+
(Mayer-Proschel et al., 1997
). In contrast, only occasional HNPs coexpressed nestin, and in triple-labeling experiments we showed that nestin expression was not seen in dividing HNPs. Unlike the case with rodent NRP cultures (Mayer-Proschel et al., 1997
; Kalyani et al., 1998
; Mujtaba et al., 1999
), HNPs have proved difficult to grow at low densities without genetic intervention (Raymon et al., 1999
; Li et al., 2000
) and have remained refractory to strict clonal analyses. HNPs could be isolated by immunopanning and would adhere to laminin/fibronectin but would divide only once or twice in culture. Cells plated in this way survived and differentiated into neurons but did not proliferate. Cells maintained in mass culture continued to divide (incorporate BrdU) for as long as 3 months. Density-dependent proliferation and differentiation have been reported for rodent cultures (Hulspas et al., 1997
; Tsai and McKay, 2000
) and could explain the failure of HNPs to proliferate in the relative absence of neighboring cells.
A possible explanation for the differences between rodent and human cultures may be the time of isolation of human fetal cells compared to that of rodent cells. Rat NRPs are isolated at E13.5 (E10.5 in mice), a time when neurogenesis has just begun. The equivalent age of gestation in humans would be 5 weeks, much earlier than when tissue is available (10–20 weeks). Our analysis of HNPs is therefore confined to later-appearing cells, which may have slower cell cycle times and altered nestin expression. These data raise the possibility that properties of cells change during development and that results obtained in one species may not be readily transferable to another or to a different stage of development.
HNPs have been isolated using a strategy different from the one described here. FACS enrichment of GFP-expressing cells, in which GFP expression was localized to neurons by using the neuron-specific Tα1 tubulin promoter, revealed the existence of a neuronal precursor population (Roy et al., 2000a
; Wang et al., 2000
). Application of the two strategies in an identical population of cells revealed that each identified a separate but partially overlapping set of cells. Only a subset of the Tα1 tubulin: GFP-expressing cells coexpressed either β-III tubulin or E-NCAM. The Tα1 tubulin:GFP+
cells were nestin−
and did not coexpress GFAP, suggesting that they were indeed neurons or neuronal precursors. More detailed comparisons of these overlapping populations are necessary to determine whether these differences are functionally relevant to the phenotypes of cells produced.
Overall, our data support several conclusions. Acute HNPs express several typical neuronal ion channels and neurotransmitter receptors before committing to a specific neuronal fate. Changes in the receptor expression pattern follow the arrest of mitosis and coincide with the expression of morphological and antigenic neuronal phenotypes. Responses to GABA, glycine, DA, ACh, and NE are all up-regulated during differentiation toward a neuronal phenotype. Individual differentiated neurons express heterogeneous complements of the receptors examined. Future experiments will be required to determine the differential roles these receptors play in developmental and neuronal signaling. In summary, our data demonstrate that E-NCAM+/β-III tubulin+/nestin− HNPs, which under differentiating conditions resemble neurons in morphology and certain protein expression patterns, also have the functional capacity to operate as neurons. The availability of renewable sources of human neuronal cells will be an invaluable resource for both therapeutic and research use. These cells will provide a model system in which to examine the effects of intrinsic and extrinsic factors on the process of neuronal development. Should the presumptive synaptic contacts in our cultures prove to exhibit functional properties, then this culture system could be extended to study events underlying synaptogenesis and the characteristics of human neural networks.