The cellular diversity of the vertebrate central nervous system (CNS) relies upon the generation of distinct neuronal subclasses at defined positions and times from a relatively small pool of proliferating progenitors. As neural progenitors proliferate, they are exposed to secreted inductive signals that initiate cell fate decisions by regulating expression of transcription factors. These transcription factors, in turn, impose developmental restrictions on multipotent progenitor cells before ultimately effecting their final differentiation [1–3]. Understanding how extracellular and cell-intrinsic mechanisms are coordinated during CNS development is important not only for understanding embryonic patterning but also for gaining insight into the developmental potential of neuronal stem cells and progenitors isolated from different regions of the CNS .
Neural progenitors from different CNS regions exhibit varying degrees of restriction during their development. Heterochronic transplantation studies have revealed that in the cortex and retina, where neurons are born in a temporal order, progenitors acquire critical aspects of their phenotype during their final cell division [5–7]. Young cortical progenitors, which normally generate deep layer neurons in their normal environment, can respond to signals from an older host environment and generate superficial layer neurons [8, 9]. Similarly, young retinal progenitors are multipotent and can adopt cell fates characteristic of the host environment, whereas older progenitors are somewhat limited in their developmental potential [10, 11]. However, time-lapse lineage analysis in vitro has revealed that even young cortical or retinal progenitors have the intrinsic potential to recapitulate the correct sequence of laminar identities when grown as single progenitors in culture [12–14]. Although clonal analysis has also revealed the importance of cell-intrinsic mechanisms for regulating progenitor cell fate decisions, the degree to which the intrinsic program can be modified by extrinsic cues has not been rigorously tested because the identities of the inductive signals as well as molecular markers for distinct progenitors are poorly defined.
In the vertebrate spinal cord, the signals involved in the conversion of progenitor cells to distinct neuronal subtypes have been defined [1, 15, 16], making it an excellent system to address if and when motor neuron (MN)-restricted progenitors arise during development and whether they can be isolated in culture. Moreover, the purification of MN-restricted cells has important implications for therapeutic efforts to treat neurodegenerative diseases that affect MNs. During spinal cord development, Sonic Hedgehog (Shh), secreted from the notochord and floor plate, induces the expression of several class II homeodomain (HD) proteins primarily belonging to the Nkx gene family (e.g. Nkx6.1/6.2/2.2) in a concentration-dependent manner within ventral neural progenitors [17–19]. The class II HD factors repress other class I HD proteins (for example, Pax7, Pax6, Dbx2 and Irx3) in ventral neural progenitors to establish, refine, and stabilize distinct progenitor domains . Nkx6.1 protein is expressed throughout the ventral third of the neural tube, spanning three ventral progenitor domains: p3, pMN and p2. From these domains arise the V3 interneurons, MNs and V2 interneurons, respectively. Genetic studies have revealed an essential role for Nkx6.1 in MN and V2 interneuron fates, through repression of Dbx2 and establishment of a ventral region of the neural tube [17, 21]. Establishment of the Nkx6.1+ region is followed by the expression of 'subtype determinants' that define specific progenitor domains and coordinate neuronal specification and differentiation . These subtype determinants include two closely related Nkx repressor proteins (Nkx2.2/2.9) that are expressed by p3 progenitors and specify V3 neurons , as well as the basic helix-loop-helix (bHLH) protein Olig2, which is restricted to pMN progenitors and coordinates MN fate [23, 24]. Therefore, labeling progenitors using the Nkx6.1 regulatory elements would allow the isolation and purification of MN and ventral interneuron progenitors.
Despite their uniform generation from the same progenitor domain within the spinal cord, MNs differ along the rostrocaudal axis by expression of Hox proteins that govern their acquisition of distinct motor columnar identities and ultimately their innervation of a variety of targets, such as limbs, intercostals muscles or sympathetic ganglia [25–27]. The expression of HoxC6 by brachial MNs, Hoxc9 by thoracic MNs and Hoxd10 by lumbar MNs depends on graded Fibroblast Growth Factor (FGF) signaling from Hensen's node [25–27]. These transcription factors ensure that distinct MNs acquire lateral motor column identities at limb levels and a preganglionic, fate at the thoracic level, respectively [25–29]. Therefore, graded activities of Shh along the dorsoventral axis and FGFs along the rostrocaudal axis initiate expression of distinct transcriptional programs that are required for generation of generic and columnar identities of MNs.
Despite our understanding of the molecular mechanisms that contribute to the establishment of ventral progenitor and neuronal fates, several issues pertaining to progenitor fate assignment remain unresolved. First, it is unclear whether all cells that express similar levels of transcription factors within a given progenitor domain are able to maintain their identities in a cell-autonomous manner when deprived of endogenous environmental cues. Second, it has not been established whether all pre-patterned MN progenitors exhibit restrictions in their developmental potential, and what the range of neuronal fates are that they can acquire when exposed to defined signals. Neural tissue explants are heterogeneous, which precludes an analysis of progenitor specification at the single-cell level. Third, if MN-restricted progenitors are present in the spinal cord, are they able to proliferate and generate MNs for a prolonged period in culture? Finally, what is the rostrocaudal identity of MNs born from a restricted progenitor in culture? Motor neurons that have been derived in vitro from embryonic stem cells share several features with those developing in vivo. However, embryonic stem cell-derived MNs have a cervical identity and do not form functional synapses with limb muscles . Isolating lineage-restricted progenitors that can generate MNs in culture may provide an alternative way to produce MNs that maintain their regional identity.
To address these questions, we have generated mice expressing enhanced green fluorescent protein (eGFP) under the control of the Nkx6.1 promoter to genetically label ventral progenitors. We prospectively isolated ventral progenitor cells from this strain by fluorescence-activated cell sorting (FACS), cultured individual Nkx6.1+ progenitors at clonal density, and analyzed the molecular identities of their progeny. We demonstrate that ventral Nkx6.1+ cells isolated after progenitor domains are established are heterogeneous in their ability to maintain their identity in culture. The majority of cultured progenitors lose their ventral identity after successive cell divisions without acquiring dorsal fates. A small subset of progenitors, including pMN progenitors, is able to maintain its identity, suggesting that subtype-restricted progenitors from the Nkx6.1+ region are present in the ventral spinal cord, although at a lower frequency than expected from the apparently uniform expression of transcription factors within these progenitors. The pMN-restricted clones have a Hox profile characteristic of thoracic MNs, despite their origin from either forelimb or thoracic levels. The fraction of subtype-restricted progenitors increases over time, suggesting that neural progenitors become progressively more independent of patterning signals. However, exposure of subtype-restricted cells to signals that specify dorsal fates leads to acquisition of some dorsal characteristics, revealing that lineage-restricted progenitor subtypes are not committed to their fate. These findings support a model whereby continuous Shh signaling is required to stably maintain progenitor domain identity after isolation and in vitro expansion.