They were initially identified in the fruit fly Drosophila, following the discovery of mutant flies lacking specific subsets of external sensory organs. Analysis of this mutant led to the identification of the achete scute (as-c) gene complex, which encodes for proteins that share sequence similarity with each other and with other transcription factors containing the bHLH domain. Subsequently, another proneural gene in Drosophila, Atonal (ato) was identified. Atonal mutant flies lack a complementary set of sensory organs, namely internal mechanosensory transduction organs, suggesting that proneural genes not only provide neural competence to ectodermal cells but also encode neuronal subtype information (Guillemot 1999).
These works paved the way for the identification of many vertebrate genes that are related to the Drosophila asc and ato and account for the specification of most of the neuronal cell types present in the mammalian nervous system. The first Neurogenin member (Neurog1) was cloned in 1996 by PCR screening using degenerate oligonucleotide primers based on sequence conservation with other bHLH proteins (Ma et al. 1996). A similar PCR screening subsequently identified two additional Neurog members, Neurog2 and Neurog3, which together with Neurog1, define a novel subfamily of Atonal-related mouse genes (therefore also called Mammalian atonal homologs, Math4A, B, and C) (Sommer et al. 1996). Independently, Neurog2 was also identified in a yeast two-hybrid screen through its ability to interact with another mammalian bHLH factor (Guillemot 1999).
By analogy with the proneural genes in Drosophila, vertebrate Neurogs were predicted to have proneural activities during mammalian neurogenesis since the time of their discovery in the late 1990s. However, it is only in the subsequent 20 years that researchers have extensively characterized the requirements of Neurog activity for the generation of various neuronal subtypes and have disclosed novel functions in the specification of nonneuronal cells.
Biochemical Properties of Neurogenins
Neurogenins are characterized by the presence of specific residues in their bHLH domain, which distinguish them from other bHLH factors. The amminoacidic sequence in the bHLH of the three Neurog members is highly conserved, sharing 82% identity (Fig. 1c), in comparison, for example, to the 67% identity shared between the bHLH domain of Neurogs and another Atonal-related factor NeuroD (Bertrand et al. 2002). The high level of conservation in the bHLH domain reflects the similarities in biochemical properties, including the dimerization with cofactors, the binding to DNA, and the regulation of transcription.
Dimerization and DNA-Binding Properties
Like other bHLH factors, Neurogs bind DNA as heterodimeric complexes (Fig. 1b). The HLH domain mediates Neurog binding to the dimerization partners, such as the ubiquitously expressed bHLH proteins, or E proteins, which are encoded by the Drosophila daughterless (Da) gene or by the vertebrate E2A (with the two alternative products E12 and E47), HEB, and E2-2 genes. Since dimerization is a requirement for DNA binding, factors that interfere with Neurog-E protein interaction can impair Neurog functions. The ID (Inhibitor of differentiation) proteins, which have an HLH domain but lack the basic DNA-binding region, can bind E proteins and sequester them away from Neurogs. Therefore, by competition for E-protein binding, ID proteins inhibit Neurog activity (Bertrand et al. 2002).
Upon dimerization with E proteins, Neurogs use the basic domain to bind DNA sequences; the majority of the residues that directly contact the DNA lie within the basic region, with three additional residues present in the HLH domain (Fig. 1c). The conserved DNA sequence bound by Neurogs contains a core hexanucleotide motif, originally found in the immunoglobulin k-chain enhancer (kE2 motif) and mainly known as E-box (Bertrand et al. 2002).
Regulation of Transcription
Given their role as transcriptional regulators, Neurogs promote differentiation by ultimately activating a series of target genes, whose expression is upregulated by Neurog activity through direct Neurog binding to promoter or enhancer regions. Identification of target genes is crucial to understand the mechanism through which Neurogs operate during neurogenesis and although progresses have been made, little is known about the transcriptional program activated by Neurogs on a genome-wide scale. Several genes have been proposed to be direct Neurog targets, because their expression was deregulated in loss-of-function and gain-of-function experiments; however, it is not clear whether they were directly bound by Neurog or activated further downstream in the transcriptional cascade (Gohlke et al. 2008; Serafimidis et al. 2008). Chromatin immunoprecipitation (ChIP) experiments can directly test whether a genomic region is bound by a transcription factor and when analyzed in combination with gene expression studies, they help identify bona fide target genes. In this way, a few Neurog targets have been characterized, including Neurod1 and Insm1, which are of particular interest since they are involved in neuronal subtype determination in the brain and also in endocrine cell differentiation in the pancreas (Masserdotti et al. 2015; Mellitzer et al. 2006). In more recent years, the identification of additional direct targets also provided crucial insights into novel functions of Neurog2 that extend beyond neuronal specification activities (Heng and Guillemot 2013), such as neuronal migration. For example, the small GTP-binding protein Rnd2 is a Neurog2 direct target that is activated in newborn neurons to control cytoskeletal remodeling and neuronal migration in the cerebral cortex (see below for details).
As well as acting as transcriptional activators at specific promoters and enhancers, Neurogs can also modify the chromatin around their target genes by recruiting epigenetic modifiers. There is indeed evidence that a subunit of SWI-SNF chromatin-remodeling complex, Brg1, interacts with and mediates the transcriptional activity of Neurog1 in Xenopus embryos and teratocarcinoma P19 cells (Heng and Guillemot 2013). Moreover, Neurog1 and Neurog2 have been shown to recruit the transcriptional coactivators p300/CBP and PCAF within multiprotein complexes that exert histone acetyltransferase activity and promote activation of specific genes for example in motoneuron precursors.
In addition to activating neuronal genes, Neurogs promote neuronal differentiation by inhibiting alternative fates. However, there is no clear evidence that Neurogs can act as transcriptional repressors, but the inhibitory activity is likely to be an indirect effect. For example, Neurog1 inhibits astrocyte fate in the cerebral cortex by sequestering CBP/p300/Smad1 transcriptional complex away from STAT, thus preventing the activation of glial genes (Heng and Guillemot 2013).
Neurogenin Role in Lineage Specification
Neurog1 and Neurog2 in Neurogenesis
Neurogs are expressed in several regions of the embryonic and adult brain, where they control the generation of specific neuronal subtypes. In particular, the developing cerebral cortex represents the site in which Neurog activities have been predominantly investigated. Neurog1 and 2 are both expressed in the dorsal cortical epithelium, where they instruct both pan-neuronal and cortical-specific identity to dorsally located neural progenitors and promote the generation of cortical pyramidal neurons (Azzarelli et al. 2015; Heng and Guillemot 2013). While Neurog1 knockouts do not exhibit overt phenotype in the cerebral cortex, Neurog2 knockouts show reduced number of early born neurons that occupy deep cortical layers, a phenotype that is exacerbated in Neurog1; Neurog2 double knockouts. Moreover, the neurons that are generated in the absence of Neurogs are not properly specified, because they express some ventral markers, at the expenses of dorsal markers typical of excitatory glutamatergic neurons (Fode et al. 2000; Heng and Guillemot 2013).
Neurog2 role in instructing a cortical glutamatergic neuron phenotype has also been demonstrated by in vitro experiments, in which overexpression of Neurog2 in neural stem cell cultures or in glia cells drives the generation of neurons exhibiting characteristics typical of this lineage (Heinrich et al. 2010). Furthermore, Neurog2 exhibits a similar subtype specification activity in the dentate gyrus of the hippocampus and in the subependymal zone of the adult rodent brain, where it promotes the generation of glutamatergic granule neurons and olfactory neurons. However, the specificity of the neuronal sublineage properties induced by Neurog2 strongly depends on the cellular context, because ectopic Neurog2 expression in the ventral division of the embryonic brain is not sufficient to impart full dorsal identity to ventral progenitors.
In keeping with Neurog role in the regulation of multiple aspects of cortical neuron development, Neurog also controls the pattern of axonal projections. Neurog2 knockout has abnormal corpus callosum and Neurog2 silencing decreased contralateral projections of upper layer neurons, while redirecting some of the projections toward the lateral cortex or subcortical regions of the same hemisphere, which are typical of lower layer neuronal subtypes. Therefore, through the coordinated regulation of the distinct aspects of cortical neuron development, Neurogs contribute to the early steps of circuit formation and to the overall cortical architecture.
Neurog1 and 2 are also expressed in neural progenitor populations in other regions, including the spinal cord and the peripheral nervous system (PNS), where they play important function in the specification of both pan-neuronal and subtype-specific properties. For example, Neurog2 expression in the ventral half of the spinal cord contributes to the specification of motoneurons, whereas during PNS development Neurogs specifically activate sensory markers and promote the generation of several cranial and spinal sensory ganglia (Bertrand et al. 2002).
Neurog3 in Neuroendocrine Differentiation
Neurog3 is generally expressed in regions of the brain where Neurog1 and Neurog2 are absent or less abundant. Particular attention has been dedicated to Neurog3 exclusive expression in specific subregions of the hypothalamus, which is a portion of the ventral brain composed by several small nuclei with important neuroendocrine and metabolic functions. In this area, Neurog3 plays a fundamental and nonredundant role in the specification of pro-opiomelanocortin (POMC)+ neurons, which control feeding behavior. Indeed, reduction in the number of POMC+ neurons is observed in defined nuclei of the Neurog3 knockout hypothalamus and leads to obesity due to increased food intake and decreased energy expenditure (Anthwal et al. 2013).
Interestingly, Neurog3 is also the only Neurog member present in regions outside the nervous system, such as the gut, stomach, spermatogonia, and embryonic pancreas, thus playing unique functions in these territories (Fig. 2b). In the developing pancreas, for example, Neurog3 specifies the distinct endocrine cell types that will form pancreatic islets. These cells secrete hormones like insulin and glucagon directly into the bloodstream and control glucose homeostasis. Due to their important role in glucose metabolism, loss or dysfunction of these cells and, in particular, deficiency in insulin-producing β-cells leads to life-threatening diseases, such as diabetes. As expected from the important role of Neurog3 in the specification of endocrine cells, Neurog3 knockout animals are deprived of endocrine pancreatic cells and die perinatally of diabetes (Gradwohl et al. 2000). In addition to the role in the endocrine pancreas, Neurog3 also specifies cells with endocrine properties in the stomach and gut, thus suggesting that the genetic program activated downstream of Neurog3 may exhibit some degree of conservation in different cell types.
Neurogenin Role in Cell Reprogramming
While Neurog2 drives neuronal differentiation from various sources, Neurog3 induces the generation of endocrine pancreatic cells, including insulin-producing β-cells, which are lost in diabetes. Ectopic expression of Neurog3 in cells of the exocrine pancreas, such as acinar or ductal cells, is able to reprogram them into the endocrine lineage. However, the specific endocrine subtypes obtained strongly depends on the presence of additional transcription factors, with the combination of Neurog3, Pdx1, and MafA being specifically required for the generation of insulin-positive cells (Zhou et al. 2008) (Fig. 3b). Therefore, controlled expression of Neurogs in human stem cells or somatic cells might be sufficient to generate in vitro cell types of interest to either study human diseases or to replace lost or damaged tissues.
Regulation of Neurogenin Activity
More recently, work from the laboratory of Ryoichiro Kageyama has shown that this process is more dynamic than previously thought. Indeed, the expression of Neurog2 in neural progenitors has been found to oscillate out of phase with the expression of the Notch signaling effector Hes1 (Shimojo et al. 2008), with a period of 2–3 h (Fig. 4b). The fast dynamics of this process indicate that Neurog-Hes oscillation may occur several times during cell cycle, which in neural progenitor range from 18–24 h. Therefore, Neurog transient and oscillatory expression may still be compatible with cycling progenitors, whereas sustained Neurog expression and Hes1 downregulation mark the transition from proliferating progenitor to differentiating neurons (Fig. 4c).
Neurog2 is also highly phosphorylated on multiple sites, although alteration in protein stability due to phosphorylation has not been reported yet. Phosphorylation of Neurog2 has been shown to regulate the ability of Neurog to interact with additional transcription factors (see next section) or to bind target gene promoters (Fig. 5b, c). For example, in the developing cerebral cortex, Neurog2 is phosphorylated on multiple Serine-Proline (SP) sites by cyclin-dependent kinases (CDKs), which are highly expressed in cycling progenitors (Ali et al. 2011). CDK-mediated phosphorylation on 9 SP sites has been shown to diminish the ability of Neurog2 to specifically bind regulatory regions of differentiation genes, such as Neurod1 and Neurod4 (Fig. 5c). Therefore, as development progresses and neurogenesis declines, the rise in Neurog phosphorylation correlates with a decrease in Neurog proneural activity.
Interaction with Cofactors
The ability of Neurogs to initiate distinct genetic programs in different cellular contexts raises the question of how they achieve specificity for target genes. Studies on Neuorg2, which specifies distinct neuronal sublineages in different regions of the nervous system, provide crucial insights into the mechanisms of target gene diversification.
Context-dependent functions of Neurog2 result, at least in part, from the differential expression of additional transcription factors, which confer target specificity by binding to specific DNA sequences in the vicinity of the E-box. In ventral spinal cord progenitors, Neurog2 interacts with LIM-homeodomain transcriptional complexes via the adaptor protein NLI to promote the expression of the motoneuron determination gene Hb9 (Heng and Guillemot 2013). Likewise, Neurog2 binding to the same bridging factor NL1 seems to be required in the cerebral cortex for Neurog2 interaction with a different homeodomain transcription factor, thus resulting in coactivation of cortical neuron genes (Heng and Guillemot 2013).
Interestingly, the interaction with cofactors is highly dynamic and often requires the presence of posttranslational modifications on Neurog proteins. For example, Neurog2 interaction with NL1 in the spinal cord requires the phosphorylation of Neurog2 on Serine 231 and 234 by GSK3β. Whether this site-specific phosphorylation is required for Neurog2-NL1 interaction also in the cerebral cortex is not known, but, in this context, Neurog2 phosphorylation has been shown to facilitate the formation of Neurog2-E protein heterodimers at the expenses of Neurog2-Neurog2 homodimers, which exhibit a different potency in activating cortical neuronal genes (Li et al. 2012). Therefore, understating the context-dependent activity of Neurogs is a fundamental question that may provide crucial insights into how Neurogs specify distinct cell types.
The discovery of Neurogs strongly contributed to understanding how different cell types originate during embryonic development, especially in the nervous system. Although the role of Neurogs in cell fate specification has been extensively studied, important questions remain unanswered. For example, the exact function of Neurogs in instructing subtype-specific properties is still in its infancy, as is the identification of direct transcriptional target genes on a genome-wide scale. Recent advances in genomic and transcriptomic techniques will definitely contribute to the identification of the transcriptional programs activated by Neurogs. Such data will also help to understand how the same Neurog specifies distinct cell types in different tissues. Deciphering how Neurogs achieve target specificity is also crucial in view of the expanding field of cell reprogramming, which aims at changing cell fate by manipulating Neurog expression or activity. Indeed, it is important to control that Neurog overexpression selectively activates genes for the intended lineage and generates only the specific cell type of interest. Furthermore, the recent discovery that Neurogs undergo several posttranslational modifications provides a more amenable way to alter the activity of endogenous Neurog proteins. As biological and genomic research continues to reveal novel Neurog functions and novel regulatory mechanisms of Neurog activity, the impact of Neurog-mediated reprogramming in regenerative medicine will also grow.
- Anthwal N, Pelling M, Claxton S, Mellitzer G, Collin C, Kessaris N, Richardson WD, Gradwohl G, Ang SL. Conditional deletion of neurogenin-3 using Nkx2.1iCre results in a mouse model for the central control of feeding, activity and obesity. Dis Model Mech. 2013;6:1133–45.PubMedPubMedCentralCrossRefGoogle Scholar
- Heng J, Guillemot F. Proneural proteins and the development of the cerebral cortex. In: Cortical development: neural diversity and neocortical organization. 2013; Tokyo: Springer. p 54419–41. doi: 10.1007/978-4-431-54496-8.
- Masserdotti G, Gillotin S, Sutor B, Drechsel D, Irmler M, Jorgensen HF, Sass S, Theis FJ, Beckers J, Berninger B, et al. Transcriptional mechanisms of proneural factors and REST in regulating neuronal reprogramming of astrocytes. Cell Stem Cell. 2015;17:74–88.PubMedPubMedCentralCrossRefGoogle Scholar