Oligodendrocytes are relatively ubiquitously distributed in the adult CNS. While the highest density of oligodendrocytes and myelin is in white matter axon tracts, there is substantial myelination in gray matter and a large population of morphologically identified and phenotypically characterized oligodendrocytes. Classical studies suggested that in any specific region of the CNS, the three major cell types of the CNS—neurons, astrocyte, and oligodendrocytes—were derived from cells of neural tube as a result of periods of coordinated cell induction, proliferation, and differentiation. In general, birth-dating studies indicate that neuronal populations arise first followed by astrocytes and then oligodendrocytes (Jacobson 1978). Remarkably, the timing of the relative appearance of different CNS cell types is largely recapitulated in cultures of neonatal neural cells (Abney et al. 1981) and even neural stem cells. Using an approach pioneered by Reynolds and Weiss (1992), culturing dissociated neural cells on a nonadherent substrate results in the death of the vast majority of cells. Some of the residual cells, however, form neurospheres that can be passaged repeatedly and when replated give rise to neurons, astrocytes, and oligodendrocytes. Such preparations have been extensively studied, and in general, neurospheres derived from virtually all regions of the developing and adult CNS have the capability to give rise to cells of the oligodendrocyte lineage, although the number of OPCs is relatively small compared to the numbers of neurons or astrocytes and they are often the last cell type to arise. It seems likely that this reproducible timing of cell development is a reflection of the interactions occurring between cells in the neurosphere culture rather than a cell intrinsic property of neural stem cells. While useful for defining the properties of neural stem cells and potential molecular regulators of early oligodendrocyte development, neurosphere studies provide few insights into the localization of initial appearance of OPCs during normal development.

The application of specific markers and growth factor receptors expressed by OPCs to in vivo preparation combined with appropriate in vitro assays allowed a detailed description of the timing and localization of OPC’s initial appearance in different regions of the CNS. In the spinal cord, for example, separation of dorsal and ventral regions along the sulcus limitans early in embryogenesis suggested that OPCs originated selectively from ventral tissues (Warf et al. 1991), an unexpected result given the wide spread distribution of oligodendrocytes in the adult CNS. Subsequent studies identified an initial source of OPCs at the ventral midline directly above the floor plate. In mouse, the localized expression of PDGF alpha receptor (PDGFαR)-expressing cells revealed the earliest OPCs around embryonic day 12 (Pringle and Richardson 1993), while in chick embryos, OPCs first appear in the ventral ventricular zone around stage 25 (Ono et al. 1995; Orentas and Miller 1996). In both species, there is a close correlation between the appearance of OPCs with the development of motor neurons from the same embryonic domain that reflects the potential shared ancestry of the two cell types (Rowitch 2004). This localized ventral origin of OPCs is a consequence of signaling between neighboring tissues. For example, in chick embryos, transplantation of an additional notochord adjacent to lateral or dorsal regions of the neural tube results in the local induction of an additional floor plate, motor neuron, and OPCs (Jessell 2000; Ono et al. 1995; Orentas and Miller 1996). The accurate spatial and temporal characterization of the initial appearance of OPCs in the spinal cord facilitated the identification of major regulators of OPC induction, including environmental cues such as Sonic hedgehog (Shh) and intracellular cues such as the expression of distinct cohorts of transcription factors including olig1 and olig2 that orchestrate oligodendrocyte development (Rowitch 2004). The ventral midline is not the only source of OPCs in the developing spinal cord. A smaller source of OPCs arises from a discrete location in the dorsal spinal cord later in development (Cai et al. 2005), while intermediate regions of the spinal cord germinal zone do not appear to give rise to oligodendrocytes although they generate both neurons and astrocytes.

In more rostral regions of the CNS, multiple discrete sources of OPCs have been identified (Spassky et al. 1998). These include the floor of the third ventricle, which generates a subset of OPCs that migrate down the optic nerve (Gao and Miller 2006) and the medial ganglion eminence (Perez Villages et al. 1999; Spassky et al. 1998; Timsit et al. 1995) that contributes oligodendrocytes to specific domains of the CNS. Not all these regions are adjacent to an obvious source of an inductive signal such as the floor plate or notochord, and at least in the optic system, the source of inductive signals appears to be retinal ganglion cell axons (Gao and Miller 2006) that express Shh (Dakubo et al. 2008) and are the ultimate target of these OPCs. Our current understanding of the origins of OPCs in the developing CNS remains incomplete, and it seems likely based on cell tracing studies that many regions of the neural tube generate OPCs (Kessaris et al. 2006) but do so in response to specific local cues in a temporal and spatially regulated manner.

The localized appearance of OPCs at defined stages in the developing CNS provided insights into environmental cues mediating their initial appearance. The ability to induce oligodendrocytes in dorsal and lateral regions of the developing spinal cord through notochord transplantation (Orentas and Miller 1996, 1998; Pringle et al. 1996) demonstrated the importance of the cellular environment and implicated the morphogen Sonic hedgehog in their initial appearance.

Sonic hedgehog is a member of the hedgehog family that contains three different members termed Sonic, Indian, and Desert hedgehogs (Ingham and McMahon 2001). The hedgehog pathway is critical in a wide range of morphogenetic events including patterning of the digits, lungs, and neural axis. Each member of the hedgehog ­family is functionally similar but has a distinct distribution in the embryo (Ingham and McMahon 2001). In the nervous system, the most highly expressed member of the hedgehog family is Sonic hedgehog (Shh) that is responsible for contributing to the patterning of the neural tube in both rostral–caudal and dorsoventral axis (Ericson et al. 1997; Jessell 2000). In caudal regions of the embryo during early development, Shh is expressed in the notochord where it contributes to the induction of ventral midline derived neural cells including the floor plate and motor neurons (Jessell 2000). In many cases, there is a connection between the types of cells induced by Shh and their subsequent expression of Shh. For example, the floor plate is induced by high concentrations of Shh generated by the adjacent notochord but then contributes to the further patterning of adjacent ventral spinal cord cell populations through the release of Shh (Jessell 2000). At high concentrations, Shh released by the notochord and floor plate induces adjacent ventral midline cells of the neural tube to develop into motor neurons and subsequently OPCs (Jessell 2000; Rowitch 2004). It has been proposed that the graded signaling of Shh from the notochord and overlying floor plate establishes a series of ventral to dorsal domains in the ventricular zone in which precursor cells become specified to distinct cell fates depending on the specific repertoire of transcription factors that are induced (Jessell 2000; Rowitch 2004). The downstream signaling of the hedgehog pathway has been extensively studied (Jiang and Hui 2008). In general, in the absence of hedgehog, ligands signaling through the smoothened receptor are inhibited by a surface molecule termed patched. The binding of hedgehog to patched releases this inhibition, and smoothened signaling is initiated. Downstream of this interaction, in the absence of hedgehog, the Gli proteins are cleaved to generate a Gli repressor that blocks the transcription of hedgehog-responsive genes, while in the presence of hedgehog, the activation of smoothened leads to a switch from a repressor to an activator (Jiang and Hui 2008). The activation of the Gli signaling pathway has been described in neural stem cells and OPCs, and although the details of the hedgehog signaling pathway in the oligodendrocyte lineage have yet to be fully resolved (Hu et al. 2009), it seems likely that they are similar to those described in other mammalian cell systems.

The inductive effects of hedgehog signaling are countered by inhibitory cues that suppress the initial appearance of OPCs. Several factors have been implicated in inhibiting OPC appearance including members of the bone morphogenetic protein family (BMPs) and the Wnts. The appearance and development of OPCs in dorsal spinal cord are significantly delayed compared to ventral regions. One mechanism that inhibits OPC development in dorsal regions is the localized expression of BMPs (Gomes et al. 2003; Grinspan et al. 2000). Bone morphogenetic proteins are members of the transforming growth factor beta superfamily and signal through SMAD intracellular pathways (Massague 1998). In the spinal cord, BMPs are primarily expressed in dorsal midline regions of the spinal cord until mid-embryogenesis and in vitro studies have shown that exposure of developing neural cells to BMP4 or 7 inhibits the appearance of oligodendrocytes and enhances the appearance of astrocytes (Grinspan et al. 2000; Miller et al. 2004). Signaling through the BMP pathway may not simply inhibit the induction of oligodendrocytes. Since in vitro OPCs are bipotential cells capable of giving rise to type 2 astrocytes or oligodendrocytes, one possibility is that BMPs influence OPCs to give rise to type 2 astrocytes rather than oligodendrocytes and this has been clearly shown in a series of in vitro studies (Grinspan et al. 2000; Miller et al. 2004). Consistent with this hypothesis, perturbation of the BMP receptor 1 in OPCs in vivo results in a transient reduction in the number of immature oligodendrocytes and an enhancement in the number of astrocytes (Gomes et al. 2003). Remarkably, this cellular imbalance is rapidly corrected during subsequent development (Samanta et al. 2007). Direct evidence for an antagonism between BMP and Shh signaling comes from in vitro studies in which the induction of oligodendrocytes by Shh can be directly inhibited by BMPs (Miller et al. 2004). It seems likely that in normal development the inhibition of oligodendrocyte differentiation by BMPs effectively outcompetes the inductive effects of Shh, and in regions where Shh is operative, additional local inhibitors of BMP signaling such as noggin are also present. The induction of OPCs in dorsal regions of the spinal cord may also reflect localized blockade of BMP signaling. The generation of OPCs in dorsal spinal cord is independent of Shh but stimulated by fibroblast growth factor (FGF) (Cai et al. 2005). In vitro studies suggest that in neural precursor cells treatment with FGF blocks BMP-stimulated nuclear translocation of phosphorylated Smads and subsequent development of OPCs (Bilican et al. 2008). Not only are BMPs important during development for oligodendrogenesis, it is becoming clear they regulate neural cell responses and fate after demyelination or spinal cord injury (Cheng et al. 2007; Fuller et al. 2007).

Additional signaling pathways may regulate OPC development. One such pathway that has received considerable attention is the Wnt signaling system that may contribute to the initial patterning of the spinal cord. Multiple studies implicated Wnts in the control of dorsal cell proliferation and more recently in neural tube patterning (Ulloa and Marti 2010). Explant studies using the chick spinal cord demonstrated expression of Wnt in dorsal regions of the developing spinal cord at the time that OPCs were being generated ventrally and that inhibition of Wnt signaling in the ex vivo model resulted in a dramatic increase in the number of OPC cells (Shimizu et al. 2005) suggesting that Wnt had an inhibitory effect on OPC development. Subsequent studies suggest that Wnt may have a more profound effect on OPC maturation than induction. Cell culture studies implicated Wnt signaling through β-catenin in the regulation of OPC maturation that was independent of the induction of OPCs or the control of their proliferation or cell fate determination (Feigenson et al. 2009). These functional characteristics distinguish Wnt signaling from other inhibitors of OPC development such as BMPs. Genetic manipulations resulting in constitutive activation of Wnt/β-catenin signaling specifically in cells of the oligodendrocyte lineage resulted in similar numbers of OPC compared to wild-type littermate controls but delayed the development of mature oligodendrocytes and myelination. Somewhat unexpectedly, these effects were transient and the histology of the adult animals was relatively normal (Feigenson et al. 2009). Confirmation of a critical role for Wnt in the timing of OPC development and identification of a persistent role for Wnt/β-catenin in the adult followed the identification of the transcription factor Tcf4 as a major mediator of oligodendrocyte Wnt signaling (Fancy et al. 2009). Conditional activation of the β-catenin pathway or perturbation of APC, an inhibitor of Wnt, resulted in a significant delay in the timing of oligodendrocyte maturation during development and a delay in remyelination in the adult CNS (Fancy et al. 2009). Thus far all the data suggest Wnt signaling in oligodendrocytes or OPCs is mediated through the canonical pathway, and whether there is a role for the noncanonical pathway in mediating the effects of Wnt on OPCs and oligodendrocytes remains to be determined.

Multiple factors may initiate the induction of OPCs in the developing CNS. While Shh is clearly important in early OPC development, OPCs also develop in cultures derived from Shh null animals indicating the existence of alternative pathways for their induction. The best defined of these is through FGF signaling pathway. As discussed above, FGF signaling may induce a blockade of BMP signaling (Bilican et al. 2008) and promote OPC induction. Later in development, OPCs express multiple FGF receptors that initiate a variety of cellular responses (Fortin et al. 2005). The relative roles of FGF in OPC induction are likely to be more pronounced in rostral regions of the CNS that contain multiple anatomically distinct areas of OPC appearance (Kessaris et al. 2006; Spassky et al. 1998). Axonal signals such as the neuregulins have been implicated in OPC induction; however, their role remains somewhat controversial. While in vitro the absence of NRG-1 significantly inhibited the appearance of OPCs (Vartanian et al. 1999) and suggested involvement of distinct ErbB receptors (Park et al. 2001; Sussman et al. 2005), subsequent studies using genetic manipulations in vivo have failed to provide any evidence of a role for neuregulins or their receptors in the development of CNS myelination (Brinkmann et al. 2008). It seems likely that while not essential in vivo for OPC development and CNS myelination, distinct isoforms of NRG and their receptors contribute significantly to normal OPC development and further studies are needed to clarify their precise roles.