imbb logo for mobile

    C. Delidakis Laboratory

    The nervous system contains a large number of cells of diverse specializations; the vast majority of these cells derive from asymmetric cell divisions of neural stem cells (called neuroblasts in Drosophila). bHLH transcription factors play crucial roles in the biology of neural stem cells and in this context we study bHLH proteins belonging to two classes: the proneural activators and the “Orange” (or "Hes/ Hey") repressors. Many of the bHLH-O repressors are transcriptional targets of Notch in a broad range of biological processes. One of our areas of interest is how Notch signalling is sequentially redeployed in various steps of neural development. Notch is a broadly used cell signalling pathway dating back to the earliest metazoans. As both the Notch receptor and its two ligands, Delta and Serrate, are transmembrane proteins, this pathway is cell-contact mediated and serves to allocate cell fates and behaviours in many tissues and at many developmental stages. Consequently, its deregulation is implicated in a variety of diseases in humans, for example cancer and ischemic stroke.


    Although animal nervous systems differ vastly in anatomy and complexity, their development relies on a common battery of transcription factors (TFs) and signalling pathways. From cnidarians to mammals, the bHLH activators of the proneural class (e.g. Scute, Atonal, Neurogenin homologues) endow cells with the potential to produce neurons. Proneural TFs generally function as heterodimers with ubiquitously expressed "E-proteins", also members of the bHLH superfamily. There is only one E-protein gene in the Drosophila genome, daughterless (da). Although the DNA binding and heterodimerization bHLH domains of Da and proneurals were well studied, little was known about other functional domains of these crucial TFs. We have extensively characterized the Scute-Da heterodimer and have identified a phosphorylation motif on Sc, as well as three transactivation domains (TADs), one on Sc and two on Da. These play at least three different roles: (1) they activate transcription of target genes, as expected; (2) they enhance the turnover of Sc, which is a very unstable protein and (3) they negatively regulate Da/Sc function by interacting with E(spl) repressors, which inhibit Da/Sc activity.

    Drosophila embryos exemplifying patterns of different E(spl) enhancer-driven GFP reporters (green). They are distinct from the expression of Hey (red or blue; CycE marks NBs in red in the closeup of the embryo's head).

    E(spl) ("Enhancer of split", a family of 7 paralogous proteins in Drosophila) are also bHLH TFs, but, unlike the proneurals, they contain an additional Orange domain (bHLH-O) and act as repressors. In the early stages of neural precursor birth in both CNS and PNS primordia, E(spl) expression is induced by Notch signalling and acts antagonistically to proneural TFs: E(spl) inhibit neural precursor commitment, whereas proneural TFs promote it. The antagonistic interplay between proneural and E(spl) TFs ensures the proper number of neurons are born. Protein-protein interactions between Da/Sc and E(spl) contribute to their antagonistic interactions and mutual destabilization. It is certain that enhancer architecture of their target genes also plays a crucial role. We have therefore embarked on genomic scale studies to identify the cohort of target genes and interactors for each of these key TFs and so to get a glimpse of the TF networks that orchestrate the adoption of neural fate.

    Proneural bHLH factors (Da/Sc heterodimer) and E(spl) homodimers activate and repress (respectively) common target genes. AD: activation domains. OR: orange domain. The corepressor Groucho is recruited by the C-terminal WRPW motif of E(spl). Although proneural and E(spl) factors have different binding site preferences (green and blue boxes on DNA), we have shown that they can form higher order complexes by interacting with each other (blue arrows). We have also shown that a phosphosrylation site on Sc (green P) is needed for its turnover and the consequent downregulation of its activity.

    The dynamic interplay between proneural and E(spl) TFs necessitates that all of these proteins respond rapidly to changes in cell-cell interactions and other inputs that dictate their expression. We are characterizing the transcriptional and translational regulation of two E(spl) genes, m7 and m8. These are adjacent on the chromosome and are co-regulated by two cis-acting Notch-responsive enhancers. Although these enhancers are very potent, E(spl)m7 and m8 are expressed at very low levels. This is largely due to several micro-RNA binding sites contained in their 3' UTRs. Inactivation of micro-RNA translational repression of E(spl) m7 and m8 leads to a substantial increase in their protein levels accompanied by visible phenotypic consequences.

    RNA-seq data from the E(spl) locus. In embryos with over-activated Notch signaling (UAS-Nicd) most transcripts are upregulated.


    After birth, NBs divide asymmetrically over more than 5 days (half of the life-cycle of Drosophila) to generate the entirety of the central nervous system of the fly. We were surprised to discover that soon after their specification, NBs express E(spl)mγ and m8, despite the fact that E(spl) expression earlier inhibits the NB fate. We have been studying the role of bHLH-O TFs in NB self-renewal. Besides E(spl)mγ and m8, we also study Deadpan (Dpn), which is another bHLH-O protein strongly expressed in NBs. We have shown that E(spl)mγ and Dpn can form heterodimers and act redundantly to prevent premature differentiation of NBs; we therefore view them as "stemness" factors. Both E(spl) genes and dpn respond to Notch signaling in the NBs and we are studying how Notch interfaces with the proliferation and differentiation programme of CNS lineages. Excessive activation of Notch or overexpression of bHLH-O stemness factors generates a hyperplastic CNS with supernumerary stem cells at the expense of differentiated neurons. We are using genomic-scale approaches to characterize transcriptional events mediated by Notch and the bHLH-O TFs during embryonic and post-embryonic neurogenesis.

    Normal brain and an example of a hyperplastic brain due to bHLH-O overexpression. Neuroblasts grow to outnumber differentiating neurons. Microarray analysis can help us sort the genes differentially regulated in five different hyperplastic conditions.

    The bHLH-O family consists of two subfamilies, the Hairy/ Enhancer-of-split (or Hes) proteins (which include E(spl) and Dpn) and the Hey proteins. Hes and Hey have diverged from each other at the base of the metazoan lineage. The three mammalian Hey members have been shown to be targets of Notch and to act redundantly with Hes proteins in several contexts, including neurogenesis. When we raised antisera against the sole Hey homologue of Drosophila, we were surprised to find that it is expressed in a non-overlapping pattern with that of E(spl) proteins. The majority of Hey positive cells were found to be early-born neurons or glia. In most of these cells, Hey expression was Notch-dependent, with the notable exception of the four mushroom body lineages.

    NBs generate differentiated cells by producing a mitotic daughter cell called "ganglion mother cell" (GMC), which divides once to yield two different neurons. The two different neuronal fates, termed A and B, are specified by Notch signalling, such that A neurons need a Notch input, whereas B neurons can only form in the absence of this signal. All identifiable Hey positive neurons (except those of the mushroom body) were shown to be of the A type, suggesting that Hey is expressed as a Notch target immediately after the asymmetric cell division of the GMC. In collaboration with Maria Monastirioti we are characterizing  transcriptional enhancers from the Hey locus to gain insight into the factors that contribute to the cell specificity in the response to Notch.

    Larval brain marked for NBs and GMCs, nascent neurons and the Hey protein. Hey is expressed in nascent neurons.


    The main actors in the Notch signalling pathway are the Notch (N) transmembrane receptor and its two alternative transmembrane ligands, Delta (Dl) and Serrate (Ser), belonging to the class of DSL proteins. Notch is activated by an intracellular juxtamembrane proteolytic cleavage mediated by γ-secretase. The released N intracellular fragment translocates to the nucleus and acts as a transcriptional co-activator of target genes bound by the Suppressor-of-Hairless [Su(H)], or CSL, transcription factor. DSL binding triggers N intracellular proteolysis and activation by first inducing another extracellular proteolytic event, followed by N ectodomain shedding. Our work revolves around the mechanisms used by Dl and Ser to activate Notch.

    Distribution of the two Notch ligands, Delta and Serrate, in the wing disk epithelium. Proneural clusters are areas from which sensory organs will arise.

    We have been studying two cytoplasmic E3 ubiquitin ligases, which from genetic studies had been associated with the Notch pathway. They both belong to the RING family of Ub ligases and are named Neuralized and Mindbomb1. We have shown that Neur promotes Dl internalization and at the same time enhances its activity. In fact Mib1 and Neur are redundantly needed for the activity of both Dl and Ser. Why would endocytosis enhance signalling? One model suggests that the movement generated by the DSL ligand’s endocytosis exerts a force on the Notch receptor in the adjacent cell, which deforms it and stimulates the proteolytic steps needed for its activation. Alternatively, DSL ligands may require a cycle of secretion, uptake and recycling in order to be properly presented to the N receptor.

    Optical cross-section of epithelial cells indicating apical and basal accumulation of Delta and Notch.

    In an effort to better characterize the biology of DSL ligands, we have shown that Neur and Mib1 interact with the cytoplasmic tails of both Dl and Ser and catalyse high MW ubiquitylation of Dl and probably also of Ser. We have mapped the motifs on Dl and Ser that interact with these E3 ligases and have shown that they are indispensable for the activity and endocytosis of these proteins. Interestingly, although both E3 ligases interact via the same Ser motif, they dock to two distinct sites on Dl. Furthermore, Neur prefers a specific lysine of Dl (K742) as a Ub attachment site, whereas Mib shows no apparent preference. These data suggest that the two E3 ubiquitin ligases, impart different types of ubiquitylation on Delta, which may ‘encode’ two different signal transduction modes. We are currently using genetic and biochemical approaches, to examine this possibility.


    Biochemical assays reveal ubiquitylation of Delta by Neur or Mib1.