Keywords

1 Introduction

It was during the study of the expression and function of the Drosophila runt gene that it was first recognized that members of the runt-related gene family (hereafter collectively referred to as Runx unless otherwise indicated) are involved in the regulation of nervous system development. During Drosophila embryogenesis, runt is expressed in a specific subset of central nervous system (CNS) neurons termed even-skipped-expressing lateral (EL) neurons. In vivo studies showed that runt inactivation causes a selective loss of EL neurons (Duffy et al. 1991). Conversely, ectopic runt expression results in the formation of supernumerary EL neurons that can extend axons along the normal trajectory used by these cells (Dormand and Brand 1998). These observations provided the first in vivo evidence that Runx genes participate in context-restricted mechanisms regulating the specification of defined neuronal subtypes in the developing nervous system and they opened the way to the subsequent investigation of the roles of Runx genes in vertebrate neural development. This chapter will review some of the key functions performed by Runx genes during mammalian nervous system development, focusing mainly on their roles during the formation of neural circuits mediating somatosensory sensation, motor control, and olfaction.

2 Involvement of Runx Genes in Peripheral Nervous System Development

Dorsal root ganglion (DRG) sensory neurons in the peripheral nervous system (PNS) mediate somatosensory stimuli such as the sensations of pain (nociception ), mechanical pressure (mechanoreception ), or relative position (proprioception). Nociceptive neurons (nociceptors) are connected to the dorsal horn of the spinal cord and cutaneous structures and express neurotrophic factor receptors such as TrkA and Ret. Mechanoreceptive neurons (mechanoreceptors) also project their fibers to the spinal cord dorsal horn and the skin, but express other receptors for neurotrophic factors such as TrkB and TrkC. Proprioceptive neurons (proprioceptors) express TrkC and project to the ventral horn and intermediate zone of the spinal cord, as well as to muscle spindles and Golgi tendon organs (Inoue et al. 2002; Kramer et al. 2006; Yoshikawa et al. 2007; Honma et al. 2010; Abdo et al. 2011; Lallemend and Ernfors, 2012).

Both Runx1 and Runx3 are expressed in developing DRG sensory neurons, mostly in non-overlapping patterns. Mouse Runx1 is preferentially expressed in several nociceptive neurons, including pain-related nociceptors and thermoceptors (Levanon et al. 2002; Chen et al. 2006a; Kramer et al. 2006; Marmigere et al. 2006; Lou et al. 2013, 2015). In contrast, Runx3 expression marks for the most part proprioceptive neurons (Inoue et al. 2002; Levanon et al. 2002; Chen et al. 2006b; Kramer et al. 2006; Yoshikawa et al. 2007; Inoue et al. 2008; Lallemend and Ernfors, 2012). The next two sections will discuss evidence that Runx1 and Runx3 act during DRG sensory neuron development to regulate the acquisition of specific neuronal subtype identities, defined gene expression profiles, and formation of precise axonal innervations.

2.1 Runx1 Involvement in Establishment of Cutaneous Sensory Circuits

Most if not all DRG cutaneous sensory neurons mediating pain, itch and thermal sensation initially express TrkA during embryonic development. TrkA+ DRG sensory neurons also express Runx1 in developing embryos (Levanon et al. 2002; Chen et al. 2006a; Kramer et al. 2006; Marmigere et al. 2006; Yoshikawa et al. 2007). At perinatal and postnatal stages, DRG expression of TrkA and Runx1 separates, resulting in the appearance of two main groups of sensory neurons characterized by the expression of TrkA or Runx1 (this latter group also expresses Ret). TrkA+ sensory neurons acquire a ‘peptidergic’ phenotype characterized in part by the expression of the neuropeptide calcitonin-gene-related peptide (CGRP) and specific cell surface proteins. In contrast, Runx1+ cells become ‘non-peptidergic’ sensory neurons defined by specific molecular traits and innervation of skin epidermis and hair follicle targets (Chen et al. 2006a; Kramer et al. 2006; Marmigere et al. 2006; Luo et al. 2007; Yoshikawa et al. 2007; Gascon et al. 2010; Yang et al. 2013; Lou et al. 2015). Runx1 is essential for both the initial separation of TrkA+ and TrkA− sensory neuron lineages and the subsequent generation of cutaneous sensory neuron diversity, a process mediated in part by the transient or persistent nature of Runx1 expression after the separation of TrkA and Runx1 expression (Lou et al. 2015).

During non-peptidergic neuronal fate specification, Runx1 orchestrates transcriptional mechanisms that directly or indirectly regulate the expression of numerous genes defining the non-peptidergic phenotype. These include, to name only a few, genes encoding cold receptors TRPM8 and TRPA1, heat receptors TRPV1 and TRPV2, several Mrgpr class G protein-coupled receptors, and ATP-gated channels. Runx1 is also important to restrict the expression of peptidergic genes, including those encoding TrkA, CGRP, and mu-class opioid receptor (Chen et al. 2006a; Kramer et al. 2006; Yoshikawa et al. 2007; Liu et al. 2008; Ugarte et al. 2013). Runx1 is hypothesized to directly regulate TrkA expression because it can bind in vitro to a TrkA minimal enhancer containing putative Runx binding sites (Marmigere et al. 2006).

As mentioned, Runx1 also plays key roles in the further specification of several cutaneous sensory neuronal subtypes, including polymodal nociceptors, pruciceptors and other neurons associated with pain, as well as specific types of mechanoreceptors (Lou et al. 2013, 2015; Yang et al. 2013). As an example, Runx1 regulates the development of specific unmyelinated low-threshold mechanoreceptors that persistently express vesicular glutamate transporter 3 (VGLUT3) and mediate pleasant touch and/or pain (Lou et al. 2013). The Runx1-dependent transcription factor gene Zfp521 is required to establish molecular features that define VGLUT3+ mechanoreceptors. Runx1 and ZFP521 work in coordination to determine the molecular phenotype of VGLUT3+ mechanoreceptors while suppressing traits typical of other types of sensory neurons such as polymodal nociceptors (Lou et al. 2015). Thus, Runx1 is a key player in the generation of DRG sensory neuron diversity by regulating various developmental processes leading to the separation of peptidergic vs non-peptidergic phenotypes as well as the specification of selected cutaneous sensory neuron subtypes.

Runx1 is also involved in regulating the connectivity of the DRG sensory neurons in which it is expressed. Non-peptidergic and peptidergic DRG neurons normally send their axons to separate targets in the dorsal spinal cord. In Runx1-deficient mice, the majority of nociceptive neurons in which Runx1 would have been expressed had it not been inactivated send their axons to targets more typical of peptidergic neurons (Chen et al. 2006a; Yoshikawa et al. 2007). Conversely, ectopic Runx1 expression in DRG neurons is sufficient to cause axons of TrkA+ neurons to project to layers of the spinal cord normally innervated by non-peptidergic neurons (Kramer et al. 2006). Consistent with these results, in vitro studies suggest that Runx1 might participate in mechanisms promoting axon growth and branching (Marmigere et al. 2006). These results are in agreement with the demonstration that Drosophila runt is involved in the control of photoreceptor neuron axonal targeting choices (Kaminker et al. 2002). Loss of mouse Runx1 function also leads to the selective loss of sensory innervation to the epidermis. In contrast, sensory innervation of ‘deep tissues’, such as muscle and visceral organs, is not affected by Runx1 inactivation, providing further evidence for a role of this gene in genetic programs controlling the differentiation of cutaneous pain pathways (Yang et al. 2013). Given these phenotypic manifestations, it is not surprising that mice lacking Runx1 in DRG sensory neurons display decreased thermal and mechanical pain perception, underscoring further the key role of Runx1 in the formation of functional nociceptive circuits (Chen et al. 2006a, b; Abdel Samad et al. 2010).

In summary, Runx1 acts at multiple levels during the formation of PNS neural circuits mediating pain, thermal, and itch sensations by controlling both sensory neuronal subtype specification and establishment of precise innervations.

2.2 Runx3 Involvement in Formation of Proprioceptive Circuits

During DRG development, TrkC+ proprioceptive neurons are generated from transiently lived TrkB+/TrkC+ cells that also give rise to TrkB+ mechanoreceptive neurons. Runx3 expression becomes detectable in TrkC+ DRG cells at approximately the time when the latter arise from TrkB+/TrkC+ precursors (Inoue et al. 2002; Levanon et al. 2002; Kramer et al. 2006). More importantly, Runx3 inactivation in Runx3-deficient mice results in decreased numbers of TrkC+ cells and also decreased cells expressing Parvalbumin, a protein expressed preferentially in proprioceptive neurons. This phenotype is correlated with a concomitant increase in TrkB+ neurons (Inoue et al. 2002; Levanon et al. 2002; Kramer et al. 2006; Inoue et al. 2007; Nakamura et al. 2008; Lallemend et al. 2012). The converse situation is observed after ectopic expression of Runx3 in all developing DRG neurons (Kramer et al. 2006). These findings show that Runx3 is important for the separation of TrkC+ and TrkB+ sensory neuronal lineages.

Runx3 may be directly involved in the extinction of TrkB expression in proprioceptive neurons because Runx3 binds to, and represses transcription from, a TrkB intronic gene regulatory element containing consensus Runx-binding sites (Inoue et al. 2007). Moreover, analysis of DRG and trigeminal ganglion development in mouse embryos lacking the gene Brn3a revealed that Runx3 fails to be activated in TrkC + neurons in the absence of Brn3a (Runx1 expression is also greatly attenuated in TrkA + nociceptors in Brn3a-deficient mice). These changes are accompanied by expanded expression of TrkB, followed by the loss of TrkC and TrkA expression (Dykes et al. 2010, 2011). Brn3a binds to a conserved upstream enhancer element within the Runx3 locus, suggesting that Runx factors repress TrkB expression downstream of Brn3a (Dykes et al. 2010). It should be noted that separate studies suggest that the negative regulation of TrkB expression by Runx3 in sensory neurons mediating touch sensation is indirect and is mediated by another transcription factor, termed Shox2, which activates TrkB expression and is repressed by Runx3 (Abdo et al. 2011).

Perturbation of Runx3 activity is also correlated with proprioceptive neuron axonal targeting defects. Runx3-deficient embryos display abnormal projections of proprioceptive DRG neurons to both peripheral and central targets. TrkC+ proprioceptive neurons fail to establish proper connections in the ventral spinal cord and instead innervate more dorsal positions (Inoue et al. 2002; Chen et al. 2006b; Nakamura et al. 2008). Conversely, ectopic Runx3 expression in TrkA + DRG neurons results in innervation of more ventral sectors of the spinal cord, similar to the targeting of TrkC + neurons (Kramer et al. 2006). These observations provide evidence that, similar to Runx1 participation in nociceptive circuit formation, Runx3 is important for the connectivity of proprioceptive DRG neurons.

More recent studies have revealed an additional level of involvement of Runx3 during proprioceptive neuron axonal development. Proprioceptive neurons exhibit different rates of axon extension at different axial levels and these differences are determined by a segmental pattern of Runx3 levels at different axial positions. Runx3 is involved in proprioceptive neuron axonal extension at least in part by controlling the level of expression of genes encoding cytoskeletal proteins involved in axon growth (Lallemend et al. 2012). Taken together, these findings provide evidence that Runx3 plays important roles in proprioceptive neuron development by regulating both the specification of proprioceptive sensory neurons and the formation of proprioceptive neural circuits.

In summary, Runx1 and Runx3 mediate analogous developmental functions during the establishment of pathways mediating either cutaneous sensation or propioception, ranging from the regulation of specific Trk family gene expression and acquisition of cell-type specific transcriptional profiles to the control of neuronal connectivity. It should be mentioned that the presence of certain DRG neurons expressing both Runx1 and Runx3 has been observed during embryonic development and after birth. At postnatal stages, some of these Runx1+/Runx3+ cells also express TrkB, Ret, and TrkC, suggesting that these proteins are coexpressed in at least a particular group of mechanoreceptive DRG neurons (Nakamura et al. 2008; Yoshikawa et al. 2013). It remains to be determined whether Runx1 and Runx3 have overlapping or non-redundant roles in the specific sensory neuronal populations in which they are coexpressed at postnatal stages.

3 Roles of Runx Genes in Central Nervous System Development

Motor neurons are specialized cells controlling voluntary and involuntary functions, ranging from the contractile activity of multiple muscle groups to the activity of smooth and cardiac muscle fibers or glands. There are three main classes of motor neurons, termed somatic (innervating muscles that control functions such as speaking, swallowing, breathing, and locomotion), branchial (innervating muscles in the face and upper neck), and visceral (controlling components of the autonomic nervous system such as smooth muscle in the viscera). The next three sections will discuss evidence suggesting that Runx1 acts during the development of selected motor neurons to regulate the establishment of specific motor neuron identities and axonal innervations. Moreover, they will address the expression of Runx1, and other Runx genes, in other types of neural cells in the CNS.

3.1 Runx1 Involvement in the Formation of Motor Circuits

Runx1 is expressed in selected types of post-mitotic motor neurons, but not their undifferentiated mitotic progenitors. Runx1+ neurons include visceral and somatic motor neurons in the murine brainstem and cervical spinal cord (Theriault et al. 2004; Stifani et al. 2008; Guizard et al. 2010; Chen et al. 2015; Yoshikawa et al. 2015). Runx1 also displays a restricted expression in selected motor neuron subtypes in the chick cervical spinal cord (Dasen et al. 2005). The timing of Runx1 expression does not coincide with motor neuron generation and instead roughly correlates with the time when the Runx1+ motor neurons are acquiring their specific molecular identities, cell body positions and/or axonal innervations (Stifani et al. 2008; Chen et al. 2015). Consistent with this finding, lack of Runx1 activity in Runx1-deficient mice does not perturb the generation of the somatic motor neurons in which Runx1 would normally be expressed (Stifani et al. 2008).

The roles of Runx1 during motor neuron development are best characterized in brainstem somatic motor neurons located in the hypoglossal nucleus (12N). Motor neurons in 12N innervate muscles in the tongue controlling vital functions such as chewing, swallowing, and breathing. Specifically, 12N motor neurons innervate two main tongue muscle groups, anatomically defined as intrinsic or extrinsic based on their origin and location. Extrinsic and intrinsic tongue muscles can be further subdivided into two functional categories: muscles controlling either tongue protrusion (‘protrusors’) or tongue retraction (‘retrusors’) (Aldes 1995; Altschuler et al. 1994; McClung and Goldberg 1999, 2000). 12N motor neurons innervating these different tongue muscles are organized into a characteristic ‘somatotopic map ’ in which motor neurons located in the dorsal half of the nucleus generally innervate retrusor muscles, whereas motor neurons located in the ventral part innervate protrusor muscles . Moreover, motor neurons innervating intrinsic muscles are generally located medially, with motor neurons innervating extrinsic muscles found more laterally (summarized in Fig. 8.1) (Aldes 1995; Chibuzo and Cummings 1982; Krammer et al. 1979).

Fig. 8.1
figure 1

Expression and proposed roles of Runx1 during hypoglossal motor neuron development . Schematic representation of the restricted expression of Runx1 in the ventromedial 12N anatomical quadrant at mediocaudal level. Runx1+ 12N motor neurons coexpress the transcription factor SCIP but do not express the transcription factor FoxP1 nor the calcium-binding proteins Parvalbumin and Calbindin, the neurotransmitter CGRP, or the surface protein c-Met. The combinatorial expression of these proteins defines four separate motor neuron groups whose topology roughly corresponds to different 12N quadrants associated with innervation of different tongue muscle groups. Runx1 is thought to act in ventromedial 12N motor neurons to prevent the acquisition of molecular profiles defining dorsomedial and ventrolateral 12N motor neuron phenotypes. Runx1 and FoxP1 are proposed to cross-repress each other’s expression, thereby defining the dorsoventral border of the medial 12N domain. Runx1 is also involved in mechanisms important for ventromedial 12N motor neuron axonal innervation of intrinsic protrusor muscles in the tongue. Abbreviations: CC central canal

Full size image

During 12N development, Runx1 expression is mainly restricted to ventromedial 12N motor neurons characterized by a molecular profile distinct from other 12N motor neurons based on the expression of specific calcium-binding proteins, neurotransmitters, cell surface receptors and transcription factors (Fig. 8.1) (Chen et al. 2015; Yoshikawa et al. 2015). Importantly, the number of motor neurons exhibiting lateral or dorsal 12N motor neuron molecular traits is decreased when Runx1 is ectopically expressed in all developing 12N motor neurons. These results suggest that Runx1 acts to specifically promote a ‘ventromedial’ 12N motor neuron phenotype at the expense of ‘ventrolateral’ and ‘dorsomedial’ 12N motor neuron identities (Chen et al. 2015).

The Runx1+ ventromedial 12N motor neurons send their axons to intrinsic tongue muscles mediating tongue protrusion (Fig. 8.1) (Chen et al. 2015; Yoshikawa et al. 2015). In vivo Runx1 inactivation results in decreased 12N motor neuron axonal projections, and overall reduction of innervation density, to intrinsic protrusor muscles (Yoshikawa et al. 2015). Runx1 inactivation also results in decreased 12N expression of the gene Frizzled3, which contributes to axonal pathfinding of 12N motor neurons (Hua et al. 2013; Yoshikawa et al. 2015). Together, these observations suggest that transcriptional programs involving Runx1 are involved in the establishment of motor circuits controlling tongue protrusion.

One of the consequences of ectopic Runx1 expression in 12N motor neurons is the detectable perturbation of the normal pattern of expression of other transcription factors involved in motor neuron development. A notable example is provided by FoxP1, which is a transcription factor that plays important roles in the regulation of spinal motor neuron subtype development (Dasen et al. 2008; Rousso et al. 2008; Palmesino et al. 2010). During 12N development, FoxP1 is expressed in dorsomedial 12N motor neurons neighboring the ventromedial Runx1+ cells (Fig. 8.1) (Chen et al. 2015). A number of observations suggest that Runx1 and FoxP1 may transcriptionally repress each other’s expression, thereby defining their respective expression domains in the medial part of 12N. In potential agreement with this possibility, forced Runx1 expression in dorsal 12N motor neurons results in a significant decrease in the number of dorsomedial 12N neurons expressing FoxP1 (Chen et al. 2015). These observations raise the possibility that at least one of the functions of Runx1 during 12N development is to prevent FoxP1 expression in the ventral sector of the medial region, thereby contributing to the establishment of separate ventromedial (Runx1+/FoxP1−) and dorsomedial (FoxP1+/Runx1−) 12N motor neuron groups. As mentioned, ventromedial 12N motor neurons project to intrinsic protrusor muscles, whereas dorsomedial 12N motor neurons are believed to project to intrinsic retrusors. Thus, it is conceivable that the activities of Runx1 and FoxP1 are involved in the formation of neural circuits controlling either tongue protrusion or retraction, respectively (Fig. 8.1).

Runx1 is expressed in other selected groups of motor neurons, including spinal motor neurons involved in the control of posture and locomotion. For instance, certain Runx1+ motor neurons at cervical spinal cord level C1–C4 innervate the anterior trapezius muscle. Runx1 expression also marks defined motor neurons at level C4–C5 that innervate the deltoideus muscle (Stifani et al. 2008). It is possible that, similar to its roles in 12N motor neurons, Runx1 may also be important for the acquisition and/or maintenance of specific motor neuron gene expression patterns and axonal connectivity in other somatic motor neuron subtypes.

In summary, the involvement of Runx1 in motor neuron diversity generation and circuit formation shares common themes with the roles of Runx1 during PNS nociceptive neuron development. This situation in turn suggests that Runx1 may perform similar tasks during the formation of additional CNS circuits involving other types of neurons in which Runx1 is expressed.

3.2 Runx1 Involvement in Other Central Neural Circuits

In addition to motor neurons, Runx1 is expressed in at least another neuronal population in the rostral brainstem, which was identified as part of the superior lateral subnucleus of the parabrachial nucleus (LPBS) on the basis of several anatomical and molecular properties (Zagami and Stifani 2010). Runx1 expression in these cells is first observed early during brain development and persists into the postnatal brain, similar to the situation observed in sensory and motor neurons. These observations suggest that Runx1 might be involved in mechanisms controlling the differentiation and/or target connectivity of LPBS neurons, as it does in other central and peripheral neurons.

Although the functional significance of the restricted expression of Runx1 in this selected neuronal population remains to be defined, it is worth mentioning that the LPBS has been implicated in the suppression of food intake in response to pain (Gibbs et al. 1973, 1976; Malick et al. 2001). Moreover, thermal and inflammatory noxious stimuli were shown to activate neurons in the LPBS in which Runx1 is expressed (Hermanson et al. 1998; Bester et al. 1995, 1997; Buritova et al. 1998). Taken together with the important roles of Runx1 in sensory nociceptor development and innervation discussed above, these observations raise the possibility that Runx1 might be involved in the formation of functional networks coordinating nociception and regulation of food intake.

3.3 Runx2 Expression in the Central Nervous System

In contrast to Runx1, little is known about the involvement of Runx2 and Runx3 in the CNS. Previous studies have revealed the presence of Runx2 transcripts in the adult mouse brain (Takarada and Yoneda 2009). Moreover, Runx2 LacZ/+ knock-in mice display restricted expression of βGalactosidase under the control of the Runx2 promoter in the postnatal hippocampus and frontal lobe area (Jeong et al. 2008). More recent work has suggested that mouse Runx2 is expressed under circadian control in specific brain regions including the paraventricular nucleus, olfactory bulb and suprachiasmatic nucleus (Reale et al. 2013). Runx2 expression is presumed to occur in neuronal cells in these brain areas, although it should be noted that separate studies suggest that at least some of the Runx2+ cells in the brain correspond to glial cells (Takarada and Yeoneda 2009). RUNX2 expression was also detected in the adult human hippocampus, and hippocampal RUNX2 expression is decreased in bipolar disorder patients (Benes et al. 2007). It will be important to characterize further the brain cells in which Runx2 is expressed and the functions, and regulation, of Runx2 in these cells to determine whether of not at least some of the themes uncovered by the analysis of Runx1 and Runx3 in central and peripheral neurons also apply to Runx2.

4 Contribution of Runx1 to Olfactory System Development

The previous sections have discussed the important roles performed by Runx1 and Runx3 in specific subtypes of central and peripheral neurons and have addressed how these functions occur mostly, if not entirely, during postmitotic neuronal development. This section will focus on the olfactory system, which offers a compelling example of the importance of Runx proteins in the biology of specific types of undifferentiated mitotic neural cells.

The olfactory system is an evolutionarily ancient sensory system mediating the sense of smell. It has a peripheral component, the olfactory epithelium (OE) , located in the nasal cavity, and central elements, the olfactory bulb and olfactory cortex, in the brain. Both the OE and the olfactory bulb host populations of neurons that are constantly regenerated during adult life and thus the olfactory system is one of the few neural tissues with persistent renewal potential (Leinwand and Chalasani 2011; Takeuchi and Sakano 2014; Suzuki and Osumi 2015).

4.1 Runx1 Involvement in Neural Progenitor Cell Proliferation in the Olfactory Epithelium

The developing OE was first identified as one of the sites of most robust Runx1 expression in the murine nervous system almost two decades ago (Simeone et al. 1995; Levanon et al. 2001). Subsequent work demonstrated that, in contrast to motor and sensory neuronal lineages, Runx1 is mainly expressed in mitotic olfactory sensory neuron (OSN) progenitor cells located on the basal side of the OE (Theriault et al. 2005). During embryonic development and postnatal life, these cells act as the neural stem/progenitor cell population in the OE, comprising the self-propagating and transit-amplifying pools that drive both development and persistent regeneration of OSNs (Kam et al. 2014; Takeuchi and Sakano 2014). In agreement with the ability of Runx proteins to regulate the balance between the undifferentiated and differentiated states in numerous cell lineages, in vivo loss-of-function studies showed that Runx1 acts in OSN progenitor cells to sustain cell proliferation and delay differentiation, thereby contributing to maintenance of the mitotic OSN progenitor pool and regulation of OSN generation. Importantly, forced exogenous Runx1 expression in primary cultures of self-propagating OE progenitor cells revealed that Runx1 is sufficient to enhance proliferation in this cellular context (Theriault et al. 2005). Thus, Runx1 acts to regulate the timing of neuronal differentiation in the OE at least in part by sustaining the proliferation of OSN progenitor cells.

4.2 Runx1 Involvement in Proliferation and Developmental Maturation of Olfactory Ensheathing Cells

OSNs located in the OE send their axons to the olfactory bulb in the rostral part of the brain. The axons of OSNs are enveloped by a particular population of glial cells, termed olfactory ensheathing cells (OECs) . Residing in both the OE and the olfactory bulb, OECs span the PNS and CNS and perform a remarkable array of functions during olfactory development and regeneration. They accompany and ensheath OSN axons, produce growth factors, cell adhesion molecules and extracellular matrix proteins that promote OSN axon growth and targeting, and act as phagocytic cells that engulf and remove apoptotic olfactory nerve debris (Su and He 2010; Chou et al. 2014; Roet and Verhaagen 2014).

Developmentally mature OECs derive from mitotic precursors that originate peripherally and give rise to different subtypes of differentiated OECs populating the olfactory bulb. Runx1 is expressed in at least some OEC precursor cells en route to the olfactory bulb and the proliferative ability of Runx1-expressing OECs is sensitive to Runx1 dosage. Specifically, decreased Runx1 levels are correlated with increased numbers of mitotic OECs , with a parallel decrease in the number of more differentiated OECs. In contrast, Runx1 overexpression results in reduced OEC proliferation (Murthy et al. 2014). Thus, Runx1 contributes to the development, and possibly regeneration, of the olfactory system by acting, at least in part, to either delay or promote the proliferation-to-differentiation transition in OSN or OEC precursors, respectively.

Runx1 expression is not limited to OEC precursors that are migrating to the olfactory bulb, but persists in a defined subgroup(s) of more developmentally mature OECs. These cells are specifically located in the inner portion of the olfactory bulb nerve layer (ONL) and exhibit characteristic molecular features, such as the expression of neuropeptide Y. In contrast, Runx1 expression is not detected in OECs that are located in the outer part of the ONL and express different sets of proteins, including the low-affinity NGF receptor p75NTR (Murthy et al. 2014). The expression of Runx1 in a particular subtype(s) of developmentally mature OECs with a specific topology and molecular profile is akin to the subtype-restricted expression of Runx1 in neuronal cells in the CNS and PNS, as discussed above. This observation raises the possibility that Runx1 may act to regulate specialized glial cell fate acquisition in the olfactory system, similar to its role in the acquisition of specific neuronal identities. Since inner and outer ONL OECs are believed to arise from common precursors, Runx1 may be involved in the specification of inner ONL OECs by promoting the expression of genes associated with this subtype and/or repressing genes associated with outer ONL OEC phenotype(s).

In summary, these findings suggest that Runx1 is involved in olfactory system development and regeneration by participating in multiple mechanisms controlling the transition from an immature proliferating state to more a developmentally mature phenotype in both neuronal and glial cell lineages.

5 Runx Expression in Glial Cells in the Central and Peripheral Nervous Systems

The expression of Runx1 in OECs is not the only example of Runx gene expression in glial cells. Recent studies have provided evidence suggesting that Runx1 is expressed in Schwann cell progenitors in the PNS (Li et al. 2015). Schwann cells share a number of properties with OECs and are involved in many aspects of peripheral nerve biology, including myelination of axons (a property exhibited by many, but not all, Schwann cells), the secretion of growth factors supporting nerve development and regeneration, trophic support to neurons, and antigen presentation (Kidd et al. 2013).

A search for active enhancers in myelinating Schwann cells after peripheral nerve injury has recently identified a Runx2 enhancer bound by c-Jun, a transcription factor required for Schwann cells to support nerve regeneration. These observations suggest that Runx2 might also be involved in Schwann cell biology, possibly by participating in the regulation of genes induced after peripheral nerve injury (Hung et al. 2015). The presence of Runx2 transcripts was also detected in cultured rat forebrain astrocytes and astrocytic C6 glioma cells (Takarada and Yoneda 2009). Moreover, RUNX2 is expressed in human gliomas, brain tumors of astrocytic origin (Vladimirova et al. 2008). Although the functional significance of these observations remains to be determined, they raise the possibility that Runx2 might be involved in mechanisms regulating proliferation and/or differentiation along the astrocyte lineage. In contrast to RUNX2, RUNX3 expression is low or absent in glioma due to promoter hypermethylation (Mueller et al. 2007). RUNX3 mRNA expression was observed in fetal and adult human brain, but whether this expression occurs in astrocytes and/or other glial cells remains to be determined (Mueller et al. 2007).

Together, these observations are suggesting that the involvement of Runx proteins in glial cell biology is not limited to OECs and extends to other glial cells in both the CNS and PNS. In these cells, Runx proteins may act to regulate the expression of genes important for the proliferation-to-differentiation transition and/or the acquisition of defined glial phenotypes. In this regard, it is important to note that Runx1 is upregulated in neurofibroma, a cancer of Schwann cell origin. Impairment of Runx1 activity in Schwann cell progenitor cells delays mouse neurofibroma formation in vivo as a result of decreased cell proliferation and increased cell apoptosis, suggesting that Runx1 may play a role in Schwann cell proliferation (Li et al. 2015). Future studies aimed at clarifying Runx expression and function in glial cells in the developing and postnatal CNS and PNS are expected to offer insight into gliogenesis and glial cell functions.

6 Conclusions and Perspectives

Much progress has been made in understanding the neural functions of Runx genes since the initial observation that runt is important for the differentiation of EL neurons in the Drosophila CNS. Our current knowledge points to essential roles for Runx1 and Runx3 during the development of specific neural circuits in the mammalian CNS and PNS. Moreover, Runx1 is involved in the development, and possibly regeneration, of the olfactory system, one of the few neural tissues with persistent renewal potential throughout life. In the future, it will be important to characterize further the molecular mechanisms underlying the roles of Runx proteins in the regulation of sensory and motor neuron subtype specific gene expression and axonal targeting choices, including the identity of additional Runx-regulated genes during these events. Moreover, little is known about the upstream mechanisms that establish the exquisite temporal and spatial specificity of Runx gene expression in both neuronal and glial cell lineages.

The possible involvement of Runx proteins in the adult CNS and/or PNS also remains to be determined. In the healthy brain, Runx1 is not detectably expressed in neural stem/progenitor cells capable of supporting the genesis of new neural cells. However, endogenous Runx1 expression becomes induced in a subpopulation of putative neural stem/progenitor cells after brain injury in adult mice (Logan et al. 2013, 2015). This effect is thought to be mediated at least in part by mechanisms involving FGF signaling, because culturing forebrain neural progenitor cells in the presence of basic FGF (bFGF) induces Runx1 expression, whereas removal of bFGF from the culture medium decreases Runx1 expression (Theriault et al. 2005; Logan et al. 2015). This possibility is also consistent with the observation that bFGF induces human RUNX1 expression in olfactory neuroblastoma tumors (Nibu et al. 2000).

When expressed in CNS neural stem/progenitor cells as a result of injury or growth factor treatment, Runx1 can participate in mechanisms regulating proliferation and neuronal differentiation (Theriault et al. 2005; Logan et al. 2015), at least in part by enhancing proliferation through repression of cell cycle inhibitory genes such as p21 Cip1 (Theriault et al. 2005). Thus, although Runx1 does not appear to be physiologically expressed in neural stem/progenitor cells in the CNS, it has the potential to participate in neural stem/progenitor cell biology during processes associated with injury and repair of the adult nervous system.

Finally, it is important to mention that Runx1 can contribute to nervous system development and repair in an additional manner, namely by modulating the activity of microglia , the resident immune cells of the nervous system (Zusso et al. 2012). Microglia survey the nervous system for signs of infection , injury or disease and mediate immune responses during many neuropathological conditions (Ransohoff and El Khoury 2015; Shemer et al. 2015). They derive from primitive myeloid precursors that originate from the yolk sac and colonize the nervous system during embryonic development (Ginhoux et al. 2010; Prinz and Mildner 2011). In the developing brain, microglia are initially proliferative, amoeboid in shape and endowed with phagocytic activity. They mediate several important developmental functions including cell debris phagocytosis, guidance of axons in white matter tracts, and synaptic refinement. As development proceeds, microglia lose their amoeboid morphology and gradually progress to a surveillant, non-phagocytic state characterized by a highly ramified morphology (Schlegelmilch et al. 2010; Prinz and Mildner 2011; Wu et al. 2015). Runx1 is expressed in postnatal forebrain amoeboid microglia and its expression is downregulated as microglia progress to the ramified phenotype. Runx1 inhibits amoeboid microglia proliferation and promotes progression to the ramified state (Zusso et al. 2012). Thus, the involvement of Runx1 in microglia biology during brain development provides an additional example of this protein’s ability to regulate proliferation and developmental maturation mechanisms in the developing and postnatal mammalian nervous system. The amoeboid-to-ramified morphological transition of microglia during development is almost recapitulated in reverse during the process of microglia activation in the adult brain, when surveillant microglia undergo a ramified-to-amoeboid transformation and become phagocytic in response to injury or disease (Ransohoff and El Khoury 2015; Shemer et al. 2015). Runx1 expression is upregulated in microglia following nerve injury in the adult nervous system (Zusso et al. 2012), implicating Runx1 in the regulation of at least certain functions of microglia in the injured/diseased adult nervous system. Runx2 expression has also been observed in microglia (Nakazato et al. 2014), but it is unknown whether Runx1 and Runx2 have overlapping or distinct functions in these cells.

In conclusion, increasing evidence implicates Runx proteins in the development, regeneration, and repair of the nervous system through the regulation of several important mechanisms in neurons, macroglia, and microglia. It is therefore anticipated that increased understanding of the roles of Runx proteins in neural cells and microglia will facilitate the study of nervous system development , homeostasis and disease.