InteractiveFly: GeneBrief

kon-tiki/perdido: Biological Overview | References

Gene name - kon-tiki/perdido

Synonyms - CG10275

Cytological map position- 36F5-36F5

Function - receptor

Keywords - mesoderm, myotube migration, muscle founder cell, tendon cell

Symbol - kon/perd

FlyBase ID: FBgn0032683

Genetic map position - 2L: 18,487,746..18,496,426 [-]

Classification - Laminin G domain, chondroitin sulfate proteoglycan (CSPG) repeats, and C-terminal PDZ binding domain

Cellular location - transmembrane

NCBI link: EntrezGene
kon/perd orthologs: Biolitmine

Directed cell migration and target recognition are critical for the development of both the nervous and muscular systems. Molecular mechanisms that control these processes in the nervous system have been intensively studied, whereas those that act during muscle development are still largely uncharacterized. This study identified a transmembrane protein, Kon-tiki (Kon, FlyBase term Perdido), that mediates myotube target recognition in the Drosophila embryo. Kon is expressed in a specific subset of myotubes and is required autonomously for these myotubes to recognize their tendon cell targets and to establish a stable connection. Kon is enriched at myotube tips during targeting and signals through the intracellular adaptor Dgrip in a conserved molecular pathway. Forced overexpression of Kon stimulates muscle motility. It is proposed that Kon promotes directed myotube migration and transduces a target-derived signal that initiates the formation of a stable connection (Schnorrer, 2007).

The same gene CG10275 was identified by Estrada (2007) and termed perdido ('lost' in Spanish). perdido is expressed in a subset of embryonic founder myoblasts, and is essential for Drosophila muscles to find their attachment sites. It encodes a conserved single-pass transmembrane cell adhesion protein that contains laminin globular extracellular domains and a small intracellular domain with a PDZ binding consensus sequence at the C-terminus. In vivo visualization of muscle guidance in both perd mutant embryos and dsRNA-injected embryos show that some ventral muscles fail to reach their attachment sites and instead form rounded multinucleated myotubes. Genetic interaction experiments done with a newly developed RNA interference assay combined with in vivo visualization of muscle development suggest that perd may be a ligand for the laminin binding αPS1-βPS integrin heterodimer, which is expressed in the tendon cells. Genetic and biochemical evidence indicates that perd is necessary to localize the muscle guidance PDZ protein DGrip to the plasma membrane. These results suggest that Perd forms an essential protein complex for muscle guidance by engaging the myotube with the tendon cell via an extracellular interaction with the tendon integrin complex and by an intracellular interaction with DGrip. The function of perd in muscle pathfinding resembles the role of its vertebrate orthologs NG2/MCSP in cell migration in the nervous system (Estrada, 2007).

The development of several different organ systems requires migrating precursor cells to locate, recognize, and connect to specific target cells. A striking and intensively studied example of this process occurs during nervous system development, as the axons and dendrites of differentiating neurons seek out their respective synaptic partners. Muscle cells face a similar challenge in finding, identifying, and attaching to their target cells -- the tendon cells that connect to bone (in vertebrates) or epidermis (in invertebrates). Just as the establishment of correct neuronal wiring specificity is critical for neural function, so too is the precise connection specificity between muscles and tendon cells essential for normal muscular function. Yet, whereas great progress has been made over the past decade in defining the molecules and mechanisms that mediate neuronal guidance and target recognition, those that mediate muscle guidance and target recognition are still poorly understood. Indeed, it is not even clear to what extent the superficially similar processes of neuronal and muscle targeting rely on shared or distinct molecular mechanisms (Schnorrer, 2007).

The body muscles of the Drosophila embryo provide an ideal model system for a genetic approach to the problem of muscle targeting. As in vertebrates, the musculature of the Drosophila embryo is highly stereotyped, with uniquely identifiable muscles connecting to specific attachment sites. A total of 30 muscles form in each of the abdominal hemisegments A2-A7, with each muscle having its characteristic size, shape, and epidermal attachment sites. These muscles are single multinucleated cells, and their development can be readily followed by live imaging. Thus, this system offers the opportunity to use genetic and imaging methods to explore muscle targeting in vivo with single-cell resolution (Schnorrer, 2007).

Drosophila muscle cells develop from two types of myoblasts: founder cells and fusion-competent myoblasts (FCMs). Each muscle has a single founder cell, which is thought to determine the characteristic features of the muscle. These founder cells can be defined by the expression of specific combinations of transcription factors, which give each founder cell, and hence each muscle, its unique identity. The FCMs, on the other hand, appear to be more generic in nature, potentially fusing with any myotube. They contribute material rather than identity to the growing myotube (Schnorrer, 2007).

Muscle migration proceeds in three phases. (1) The founder cells migrate relative to each other to assume the approximate position in which the muscle will form. (2) FCMs begin to fuse with the founder cells, forming polarized myotubes with a highly dynamic leading edge at each end. These dynamic tips resemble the growth cones of neurons and appear to have a similar function in directing migration toward specific target cells. (3) Each myotube tip recognizes its specific attachment site and establishes a stable connection. These attachment sites are the tendon cells, specialized epidermal cells located along intersegmental borders and also within segments (Schnorrer, 2007).

How do myotubes locate and recognize their specific tendon cell targets? An attractive model, based on the present understanding of neuronal wiring specificity, is that individual myotubes and tendon cells may express distinct sets of surface or secreted proteins that constitute a molecular recognition system. If this is the case, then what are these molecules, how do they function in myotube targeting, and are they the same as or different than those that operate in the nervous system? To date, only two molecular systems for muscle guidance have been identified, both initially defined by their roles in neuronal wiring. Slit and its Robo family receptors repel both axons and myotubes away from the midline, and may later attract some of the same myotubes to their intersegmental attachment sites. Similarly, the Derailed (Drl) receptor helps some commissural axons to choose the right pathway across the midline and some myotubes to select the right intrasegmental attachment sites. Thus, at least some of the mechanisms involved in axon guidance and targeting are also utilized during muscle development (Schnorrer, 2007).

A less biased approach is to screen directly for genes involved in muscle development, and indeed this is the only way to identify factors that do not also act in neurons. A mutagenesis screen resulted in the identification of the gene kon-tiki (kon), which is essential for promoting directed migration and target recognition of a subset of ventral myotubes. kon mutations do not affect embryonic nervous system development, indicating that muscles and neurons use distinct as well as shared guidance and recognition molecules. The Kon protein is a large transmembrane protein that concentrates at muscle tips. The cytoplasmic domain of Kon contains a PDZ-binding motif and interacts with the PDZ-domain protein Dgrip -- a cytoplasmic adaptor previously shown to function in targeting of these myotubes. Both Kon and Dgrip are highly conserved, and they appear to define an ancient evolutionary molecular pathway that mediates specific muscle targeting (Schnorrer, 2007).

The kon gene was localized, using SNP-on-chip technology, to a position proximal to 36A10 on chromosome 2L. The location of kon was further refined by deficiency mapping: Df(2L)TW137 and Df(2L)M36-S5 delete kon but Df(2L)Exel8083 and Df(2L)Exel6041 do not. This placed kon within the 116 kb interval from 18,451 to 18,567 kb, a region that includes seven annotated genes. Of these genes, CG10275 seemed a strong candidate for kon, based on its size and its predicted product. Indeed, when all of the annotated exons of CG10275 from kon heterozygous adults were sequenced, single-nucleotide substitutions were found in eight of the nine alleles. Five of these alleles are associated with nonsense mutations; the other three are missense mutations in conserved domains (Schnorrer, 2007).

The computational annotation of kon was experimentally refined to include two additional exons which were identified by 5' RACE on embryonic cDNA. The kon mRNA thus includes a total of 12 exons with a total length of 8.3 kb. It is predicted to encode a transmembrane protein of 2381 amino acids, including an N-terminal signal sequence, a large extracellular region, a single membrane-spanning segment, and a cytoplasmic region of 159 amino acids. The predicted extracellular region is composed of 2 lamininG domains followed by 15 chondroitin sulfate proteoglycan (CSPG) repeats, which are structurally related to cadherin domains. The intracellular region lacks any known protein domain, but includes at the C terminus a predicted binding site for PDZ-domain-containing proteins (Schnorrer, 2007).

Kon is closely related to the NG2/CSPG4 and the 'similar to CSPG4' family proteins in vertebrates. These proteins share all of the extracellular domains present in Drosophila Kon, with 21%-25% identity and 39%-46% similarity in the lamininG and CSPG domains. The short cytoplasmic region is less well conserved, with the notable exception of the invariable QYWV sequence of the PDZ-binding motif. The in vivo functions of these Kon relatives in other species are unknown (Schnorrer, 2007).

NG2/CSPG4 is linked to chondroitin sulfate (CS) in human, mouse, and rat. However, the serine-999 residue that carries this modification in the rat homolog (Stallcup, 2001; see

Stallcup and Ng2 for subsequent studies) is not conserved in Drosophila Kon. To test whether Drosophila Kon contains a CS moiety linked to other residues, endogenous or overexpressed Kon protein were immunoprecipitated from embryos, and the electrophoretic mobility of Kon was compared with and without prior treatment with chondroitinase ABC. In both cases, a single sharp band of about 250-300 kD was observed. It is concluded that most Drosophila Kon protein is not modified by CS (Schnorrer, 2007).

kon mRNA is first detected at stage 10, possibly in the longitudinal visceral muscle precursors. These cells arise at the posterior of the embryo and migrate anteriorly to form muscles around the developing gut. More importantly, in the body wall muscles high expression was detected in the ventral-longitudinal myotubes in wild-type embryos at stage 14, the stage at which the first defects become apparent in these myotubes in kon mutant embryos. Taken together, these phenotypic, expression, and molecular data suggest that Kon might be a transmembrane receptor for a guidance or targeting cue provided by specific tendon cells -- a signal that may be transduced intracellularly through Kon's PDZ-binding motif (Schnorrer, 2007).

In each abdominal hemisegment of the Drosophila embryo, 30 muscles connect at their ends to distinct tendon cells in the epidermis. How does each muscle find and recognize its specific tendon cell targets? Myotubes are molecularly distinct before they begin to migrate, they generally migrate directly toward their targets, and they are unperturbed if other muscles are genetically ablated. Thus, it appears that each myotube is endowed with the ability to independently locate and recognize its specific target cells. Kon confers this ability on a specific subclass of myotubes (Schnorrer, 2007).

kon was identified based on the aberrant muscle patterns in kon mutant embryos: the ventral-longitudinal muscles VL1-4 do not connect to their target cells, but most other muscles appear to attach correctly. Time-lapse analysis of these mutants indicates that kon function is required for targeting of the VL myotubes. The directed migration of these myotubes toward the target is also reduced. Nevertheless, filopodia do frequently touch their target cells, but seem indifferent to them. General myotube motility appears normal, and if the VL myotubes are misrouted ventrally (as in slit mutants), then kon function is not required for their migration toward the ventral midline. Hence, kon function is required specifically for these myotubes to recognize their targets. The VL myotubes do not connect to potential alternative targets in kon mutants, such as the intrasegmental attachment sites of the VA muscles. This distinguishes kon from both slit and derailed (drl). Specific myotubes are misrouted in both slit and drl mutants, yet in these mutants the misrouted myotubes are still able to connect to tendon cells (possibly selecting alternate sites according to an unknown hierarchy of preferences). Thus, defective guidance alone does not preclude tendon cell attachment, although it may lead to the selection of the wrong targets. It is therefore concluded that kon function is specifically required for the VL muscles to recognize and attach to their tendon cell targets (Schnorrer, 2007).

Misexpression of Kon in other muscles does not redirect them to other targets. This is perhaps not surprising, since misexpression of many different axon guidance receptors -- such as Frazzled, Robo, and Ptp69d -- similarly does not lead to axonal misrouting. Presumably, in muscles as in neurons, guidance and targeting are controlled by a suite of factors acting in concert. It was however found that overexpression of Kon results in excessive filopodial activity, which persists even at the late stages during which muscles normally cease filopodial activity and establish stable contacts. This gain-of-function phenotype is consistent with the interpretation of the loss-of-function phenotype, namely, that Kon functions to promote migration until the correct connection is established. If too much Kon is present, this putative 'stop' or targeting signal might be overridden (Schnorrer, 2007).

What is the molecular mechanism underlying Kon function in myotube targeting? It was postulate that Kon is a receptor for a ligand produced by specific tendon cells, and that it transduces this signal intracellularly to modulate cytoskeletal dynamics at the myotube tip. This model is based on the following observations. (1) Kon is a single-pass transmembrane protein that localizes to myotube tips during the targeting steps. (2) It functions cell-autonomously. A VL1 myotube that expresses Kon can still target correctly in an embryo that otherwise completely lacks Kon protein. (3) Even if the rescued VL1 fails to attach in one segment, those in adjacent segments can still attach correctly. This argues against alternative models in which Kon might mediate homophilic adhesion across segment boundaries. (4) Kon protein accumulates in juxtaposition to tendon cells, as might be expected if tendon cells express a binding partner for Kon. (5) Full Kon function requires its cytoplasmic domain, including the C-terminal PDZ-binding motif (Schnorrer, 2007).

The identity of the putative Kon ligand is unknown. Potential ligands have however been identified for vertebrate Kon family proteins. For example, NG2 binds to PDGF-AA and bFGF (Goretzki, 1999), to the kringle domains of plasminogen and angiostatin (Goretzki, 1999), to types V and VI collagen (Burg, 1996; Tillet, 1997), and to galectin-3 (Fukushi, 2004; Wen, 2006). The interaction with galectin-3 is thought to lead to integrin activation (Fukushi, 2004), which could potentially modulate a cell's migratory or adhesive properties. Although at least one of the two Drosophila galectins is expressed in migrating myotubes, a similar mechanism is unlikely to apply for Kon because Drosophila integrins are required for stable muscle attachment, not for migration or targeting. Determining the nature, localization, and function of this ligand is a major goal for future genetic and biochemical studies (Schnorrer, 2007).

Such approaches have however already allowed identification of a cytoplasmic partner for Kon: the PDZ protein Dgrip. Biochemically, the C-terminal PDZ-binding motif in Kon interacts in vitro with the seventh PDZ domain of Dgrip. Genetically, truncating Kon before the PDZ-binding domain and completely eliminating Dgrip result in quantitatively indistinguishable muscle defects. Similarly, the seventh PDZ domain of Dgrip was recently shown to be important to mediate Dgrip function, presumably through Kon. This Kon-Dgrip interaction is also conserved in vertebrates, since mouse NG2 binds to the seventh PDZ domain of mouse Grip1 or Grip2 in vitro and in vivo (Stegmuller, 2003). If Dgrip forms homomultimers through PDZ domains 4-6, as its vertebrate counterparts do, then this might facilitate the clustering of Kon proteins. Alternatively, or in addition, Dgrip might function as an adaptor to recruit other signaling components to the Kon-Dgrip complex. Vertebrate Grip1 is thought to perform this function for other transmembrane proteins, recruiting signaling proteins such as the RasGEF Grasp1 or other adaptors such as Liprin-α. Recently the broadly expressed transmembrane protein Echinoid (Ed) was identified as an binding partner for PDZ domains 1, 2, and 7 of Drosophila Dgrip. Although ed mutants do not display major muscle defects, the VL defects observed in Dgrip mutants are enhanced by removing one copy of ed, indicating that ed may modulate Dgrip activity, possibly by binding to PDZ domains 1 and 2 (Schnorrer, 2007).

While Dgrip clearly plays a critical role in Kon signaling, it is also noted that the complete loss of kon function results in a phenotype that is significantly stronger than that which results from the loss of Dgrip, or the C-terminal truncation of Kon. Thus, Kon can exert at least some function that is independent of its interaction with Dgrip. This might reflect the ability of Kon to activate alternative signaling pathways, possibly involving interactions with other transmembrane proteins. It is also possible that Kon might have a dual function as both a receptor and a ligand, with only the former involving Dgrip. Genetic data do not presently offer any insight into the molecular basis for this additional, Dgrip-independent function of Kon (Schnorrer, 2007).

The structures of Kon and Grip proteins, as well as their physical interactions, have been preserved over more than 600 million years of evolution. In Drosophila, this signaling pathway mediates targeting of myotubes during embryonic development. Functions of this pathway in other species are thus far unknown. In vertebrates, there are two subfamilies of Kon proteins, equally related to Drosophila Kon: the NG2/CSPG4 subfamily and the 'similar to CSPG4' subfamily. Most research has been focused on NG2 itself. In a striking parallel to Kon, NG2 is expressed in developing skeletal muscle (Stallcup, 2002). However, NG2 is more broadly expressed than Kon, including in other migrating cells such as developing neurons, glia, mesenchymal cells, and bone (Fukushi, 2003; Niehaus, 1999; Nishiyama, 1991; Stallcup, 2002), as well as a variety of melanomas and cancer cell lines (Stallcup, 2002). NG2 stimulates cell migration in vitro (Fang, 1999; Niehaus, 1999), and smooth muscle cells from NG2−/− mice display a reduced migratory response to PDGF-AA, a putative NG2 ligand (Grako, 1999). Thus, Kon family proteins may have an evolutionary ancient role in the regulation of cell migration. It is anticipated that further functional studies of NG2 or other Kon family proteins might reveal more specific roles for these proteins in muscle migration and targeting in vertebrates. Conversely, further genetic studies of muscle targeting in Drosophila may provide clues as to how this conserved signaling pathway contributes to cancer metastasis (Schnorrer, 2007).

Regenerative neurogenic response from glia requires insulin-driven neuron-glia communication

Understanding how injury to the central nervous system induces de novo neurogenesis in animals would help promote regeneration in humans. Regenerative neurogenesis could originate from glia and glial neuron-glia antigen-2 (NG2) may sense injury-induced neuronal signals, but these are unknown. This study used Drosophila to search for genes functionally related to the NG2 homologue kon-tiki (kon), and identified Islet Antigen-2 (Ia-2), required in neurons for insulin secretion. Both loss and over-expression of ia-2 induced neural stem cell gene expression, injury increased ia-2 expression and induced ectopic neural stem cells. Using genetic analysis and lineage tracing, this study demonstrated that Ia-2 and Kon regulate Drosophila insulin-like peptide 6 (Dilp-6) to induce glial proliferation and neural stem cells from glia. Ectopic neural stem cells can divide, and limited de novo neurogenesis could be traced back to glial cells. Altogether, Ia-2 and Dilp-6 drive a neuron-glia relay that restores glia and reprogrammes glia into neural stem cells for regeneration (Harrison, 2021).

The central nervous system (CNS) can regenerate after injury in some animals, and this involves de novo neurogenesis. Newly formed neurons integrate into functional neural circuits, enabling the recovery of function and behaviour, which is how CNS regeneration is measured. The human CNS does not regenerate after injury. However, in principle it could, as we continue to produce new neurons throughout life that integrate into functional circuits. Through understanding the molecular mechanisms underlying natural regenerative neurogenesis in animals, it might be possible to provoke de novo neurogenesis in the human CNS to promote regeneration after damage or neurodegenerative diseases. Regenerative neurogenesis across animals may reflect an ancestral, evolutionarily conserved genetic mechanism, which manifests itself to various degrees in regenerating and non-regenerating animals. Accordingly, it may be possible to discover molecular mechanisms of injury-induced neurogenesis in the fruit-fly Drosophila, which is a powerful genetic model organism (Harrison, 2021).

Regenerative neurogenesis could occur through activation of quiescent neural stem cells, de-differentiation of neurons or glia, or direct conversion of glia to neurons. Across many regenerating animals, new neurons originate mostly from glial cells. In the mammalian CNS, radial glial cells behave like neural stem cells to produce neurons during development. Remarkably, whereas NG2-glia (also known as oligodendrocyte progenitor cells, OPCs) produce only glia (oligodendrocytes and astrocytes) in development, they can also produce neurons in the adult and upon injury, although this remains controversial. Discovering the molecular mechanisms of a neurogenic response of glia is of paramount urgency (Harrison, 2021).

NG2-glia are progenitor cells in the adult human brain, constituting 5-10% of total CNS cells, and remain proliferative throughout life. In development, NG2-glia are progenitors of astrocytes, OPCs, and oligodendrocytes, but postnatally and upon injury they can also produce neurons. They can also be directly reprogrammed into neurons that integrate into functional circuits. The diversity and functions of NG2-glia are not yet fully understood, but they are particularly close to neurons. They receive and respond to action potentials generating calcium signals, they monitor and modulate the state of neural circuits by regulating channels and secreting chondroitin sulphate proteoglycan perineural nets, and they also induce their own proliferation to generate more NG2-glia, astrocytes that sustain neuronal physiology, and oligodendrocytes that enwrap axons. NG2-glia have key roles in brain plasticity, homeostasis, and repair in close interaction with neurons, but to what extent this depends on the NG2 gene and protein, is not known (Harrison, 2021).

NG2 (also known as chondroitin sulphate proteoglycan 4, CSPG4) is expressed by NG2-glia and pericytes, but not by oligodendrocytes, neurons, or astrocytes. NG2 is a transmembrane protein that can be cleaved upon neuronal stimulation to release a large secreted extracellular domain and an intracellular domain. The intracellular domain (ICD, NG2ICD) is mostly cytoplasmic, and it induces protein translation and cell cycle progression (Nayak, 2018). NG2ICD lacks a DNA binding domain and therefore does not function as a transcription factor, but it has a nuclear WW4 domain and nuclear localisation signals and can regulate gene expression. It is thought that NG2 functions as a receptor, triggering nuclear signalling in response to ligands or partners (Sakry, 2014; Sakry and Trotter, 2016). NG2 protein is abundant in proliferating NG2-glia and glioma. It is also required for OPC proliferation and migration in development and in response to injury. Given the close relationship of NG2-glia with neurons, it is anticipated that key partners of NG2 are produced from neurons, but these remain largely unknown (Harrison, 2021).

The fruit-fly Drosophila is particularly powerful for discovering novel molecular mechanisms. The Drosophila NG2 homologue is called kon-tiki (kon) or perdido. Kon functions in glia, promotes glial proliferation and glial cell fate determination in development and upon injury, and promotes glial regeneration and CNS injury repair. Kon works in concert with the receptor Notch and the transcription factor Prospero (Pros) to drive the glial regenerative response to CNS injury. It is normally found in low levels in the larval CNS, but injury induces a Notch-dependent increase in kon expression in glia. Together, Notch signalling and Kon induce glial proliferation. Kon also initiates neuropile glial differentiation and pros expression, and Pros maintains glial cell differentiation. This glial regenerative response to injury is homeostatic and time-limited, as two negative feedback loops halt it: Kon represses Notch, and Pros represses kon expression, preventing further cell division. The relationship between these genes is also conserved in the mouse, where the homologue of pros, Prox1, is expressed together with Notch1 in NG2-glia. Following cell division, Prox1 represses NG2-glia proliferation and promotes oligodendrocyte differentiation. Together, Notch, Kon, and Pros form a homeostatic gene network that sustains neuropile glial integrity throughout life and drives glial regeneration upon injury. As Kon is upregulated upon injury and provokes glial proliferation and differentiation, it is the key driver of the glial regenerative response to CNS injury (Harrison, 2021).

A critical missing link to understand CNS regeneration was the identification of neuronal partners of glial NG2/Kon that could induce regenerative neurogenesis. Injury to the Drosophila larval CNS also resulted in spontaneous, yet incomplete, repair of the axonal neuropile. This strongly suggested that injury might also induce neuronal events, such as axonal regrowth or generation of new neurons. Thus, this study asked whether Kon may interact with neuronal factors that could contribute to regenerative neurogenesis after injury. Relay of insulin signalling involving neuronal Ia-2 and glial Kon drives in vivo reprogramming of neuropile glia into neural stem cells (Harrison, 2021).

NG2-glia are abundant progenitor cells present throughout life in the adult human brain and can respond to injury. Thus, they are the ideal cell type to manipulate to promote regeneration. However, whether NG2-glia can give rise to neurons is highly debated, and potential mechanisms remained unknown. Using Drosophila in vivo functional genetic analysis this study has identified neuronal Ia-2 as a genetic interactor of the NG2 homologue Kon and shows that it can induce a neurogenic response from glial cells via insulin signalling (Harrison, 2021).

Evidence is provided that Ia-2, Kon, and Dilp-6 induce a regenerative neurogenic response from glia (Ia-2 and Dilp-6 drive a regenerative neurogenic response to central nervous system (CNS) injury). In the un-injured CNS, Kon and Ia-2 are restricted to glia and neurons, respectively (Ia-2 and Dilp-6 drive a regenerative neurogenic response to central nervous system (CNS) injury). Ia-2 is required for neuronal Dilp-6 secretion, Dilp-6 is produced by some neurons and mostly glia, and its production depends mostly on Kon regulated glia. Alterations in Ia-2 levels, increased Dilp-6, and concerted activation of Ras or PI3Kinase downstream of insulin signalling induced ectopic neural stem cells from glia. Both loss and gain of ia-2 function induced ectopic Dpn cells. Ia-2 depends on Pros and in turn negatively regulates Pros. Pros controls the switch from neural stem cell to progenitor state. In this way, cell-cell interactions involving Ia-2 can influence neural progenitor cell fate. ia-2 loss of function would also cause a decrease in Dilp-6 secretion from neurons, but not from glia, as kon mRNA levels were unaffected, and dilp-6 expression depends mostly on glial kon. As neuronal Ia-2 and glial Kon mutually exclude each other, perhaps loss of ia-2 function might increase kon-dependent Dilp-6 production. As Ia-2 is required for Dilp-6 secretion, ia-2 GOF would increase Dilp-6 release triggering the Dilp-6 amplification loop. Conceivably, either way Dilp-6 increased and this induced Dpn. Upon injury, levels of kon and ia-2 expression increased. Ia-2 drives secretion of Dilp-6 from neurons, Dilp-6 is received by glia, and a positive feedback amplification loop drives the further Kon and InR dependent production of Dilp-6 from cortex glia. Dilp-6 can then both promote glial proliferation to generate more glia and induce the neural stem cell marker Dpn in neuropile glia -- the subset known as 'Drosophila astrocytes' and midline glia. Ectopic Dpn+ cells were induced from glia both upon injury and genetic manipulation of Ia-2, Dilp-6, Ras, and PI3Kinase. Importantly, these glial-derived neural stem cells could divide, as revealed by the S-phase marker PCNA-GFP and the mitotic marker pH3, and could generate neurons, albeit to a rather limited extent. Altogether, Dilp-6 is relayed from neurons to cortex and then to neuropile glia. This neuron-glia communication relay could enable concerted glio- and neuro-genesis, matching interacting cell populations for regeneration. Interestingly, Dilp-6 is also involved in non-autonomous relays between distinct CNS cell populations to activate neural stem cells and induce neuronal differentiation in development (Harrison, 2021).

This study has demonstrated that ectopic neural stem cells originate from glia. Regenerative neurogenesis could occur via direct conversion of glia into neurons, glial de-differentiation, or neuronal de-differentiation. Neuronal de-differentiation occurs both in mammals and in Drosophila. However, in most animals, neural stem cells in the adult CNS and upon injury are generally distinct from developmental ones, and can originate from hemocytes, but most often, glial cells. In the mammalian brain, radial glia in the hippocampus respond to environmental challenge by dividing asymmetrically to produce neural progenitors that produce neurons; and astrocytes and NG2-glia can generate neurons, particularly in response to stroke, excitoxic injury, and genetic manipulations. Furthermore, genetic manipulation can lead to the direct conversion of NG2-glia into neurons. The findings that Dilp-6 and InR signalling can induce dpn expression are reminiscent of their functions in the induction of neural stem cells from quiescent progenitors in development. However, the Dpn+ cells induced upon injury and after development are distinct from the developmental neural stem cells normally induced by Dilp-6 in multiple ways. Firstly, in injuries carried out in third instar larvae, the induced neural stem cells were more numerous than normal neural stem cells. Secondly, in injuries carried out late in wandering larvae, Dpn+ cells were found after normal developmental neural stem cells have been eliminated through apoptosis. Thirdly, Dpn+ cells were found in dorsal ectopic locations not normally occupied by developmental neural stem cells. In all injury and genetic manipulation experiments involving overexpression of either ia-2, dilp-6, or PI3K, ectopic Dpn+ cells were located along the midline and surrounding the neuropile, in positions normally occupied by glia. Remarkably, concerted overexpression of ras and dilp-6 induced Dpn in potentially all glial cells and more, consistently with further Dpn+ cell proliferation. Consistent with the current findings, ectopic neuroblasts were also observed upon co-expression of activated rasV12 and knock-down of PTEN in glia, within glioma models in Drosophila. This study has demonstrated that ectopic Dpn+ originated from glia, most particularly neuropile glia (midline glia and 'Drosophila astrocytes'). Firstly, ectopic Dpn+ cells did not have Ia-2YFP, which is expressed in all neurons. Secondly, overexpression of ia-2 or dilp-6, alone or in combination with ras and PI3K, in glia dramatically increased Dpn levels, meaning that insulin signalling induces dpn expression in glia. Thirdly, ectopic Dpn+ cells surrounding the neuropile occupied positions of astrocytes and had the pan-glial marker Repo, and Repo- Dpn+ along the midline had the midline glia marker Wrp. Fourthly, the glial origin of the ectopic Dpn+ cells was demonstrated using two cell-lineage tracing methods (G-TRACE and glial activation of the actin promoter) whereby the expression initiated from the glia repo promoter was turned permanent despite cell state transitions. Consistently with these findings, TRAP-RNA analysis of the normal third instar larva revealed expression of dpn and multiple genes involved in neuroblast polarity, asymmetric cell division, neuroblast proliferation, and neurogenesis in glia. And single cell RNAseq analysis of the larval CNS revealed that in normal larvae some Repo+ glial cells can express dpn, or other neuroblast markers like wor and ase. The current findings show that basal or potential expression of neuroblast genes in glia is switched on and amplified by insulin signalling. It is concluded that Ia-2 and Dilp-6 could reprogramme glial cells in vivo into neural stem cells (Harrison, 2021).

The data showed that the ectopic ia-2 and dilp-6 induced neural stem cells could divide and generate neurons. In fact, concomitant overexpression of dilp-6 and PI3K, and most prominently dilp-6 and ras, dramatically increased Dpn+ cell number. Dilp-6 induced glial-derived Dpn+ cells could express the S-phase marker PCNA-GFP, and Ia-2 induced Wrp+ Dpn+ cells that were pH3+ in mitosis. No mitotic cells surrounding the neuropile were detected, but mitosis is brief, and could have easily been missed. The Dilp-6 induced ectopic Dpn+ cells could generate neurons that could be traced with GFP expression from their glial origin. Thus, ectopic neural stem cells induced by Dilp-6 can divide and produce neuronal progeny cells. However, the clusters of GFP+ cells originating from the in vivo reprogrammed glial cells were rather small, indicating that although neurogenesis was possible in late larvae, it was extremely constrained. This could be due to the fact that in the third instar larva, time is rather limited by pupariation. Injury and genetic manipulation in late larvae may not allow sufficient time for cell lineages to progress, before pupariation starts. Pupariation and metamorphosis bring in a different cellular context, which could interfere with regenerative neuronal differentiation. Alternatively, Ia-2 and Dpn may not be sufficient to carry neurogenesis through either. For instance, gain of ia-2 function resulted only in Dpn+ but not Pros+ or Eve+ cells, suggesting that Ia-2 and Dpn are not sufficient for neuroblasts to progress to GMCs and neurons. In fact, ectopic Dpn+ cells still had Repo. Furthermore, other ectopic neuroblast markers, such as Wor or Ase were not detected in glia. Nevertheless, RNA seq data revealed expression of neuroblast markers, including dpn, wor, and ase in some glia in normal larval CNS, meaning they could potentially be further regulated. Still, to generate neurons, glia may not only require the expression of neural stem cell markers like dpn, but also perhaps receive other yet unknown signals. In mammals, injury creates a distinct cellular environment that prompts glial cells to generate different cell types than in the un-injured CNS. For instance, elevated Sox-2 is sufficient to directly reprogramme NG2-glia into neurons, but only upon injury. Whereas during normal development NG2-glial cells may only produce oligodendrocyte lineage cells, upon injury they can also produce astrocytes and neurons. This suggests that there are injury-induced cues for neuronal differentiation. In the future, it will be compelling to find out what signals could enhance neurogenesis from glial cells reprogrammed in vivo by insulin signalling (Harrison, 2021).

This work has revealed a novel molecular mechanism driving a regenerative neurogenic response from glia, involving Kon/NG2 and insulin signalling. Ia-2 induces an initial secretion of Dilp-6 from neurons, Dilp-6 is received by glia, and a positive feedback loop amplifies the Kon-dependent production of Dilp-6 by cortex glia, Dilp-6 is then relayed to neuropile glia, resulting in the in vivo reprogramming of glial cells into neural stem cells. This mechanism can induce both glial regeneration and neural stem cells from glia, potentially also neurons, matching interacting neuronal and glial cell populations. The incidence of neuropile glia conversion to Dpn+ cells was variable, meaning the process is stochastic. However, all glia converted when activated Ras or PI3K were combined with Dilp-6, meaning levels of insulin signalling matter. Such a mechanism may also operate in mammals. In fact, Ia-2 has universal functions in dense core vesicles to release insulin. Insulin-like growth factor 1 (IGF-1) induces the production of astrocytes, oligodendrocytes, and neurons from progenitor cells in the adult brain, in response to exercise. The transcription factor Sox-2 that can switch astrocytes to neural stem cells and produce neurons is a downstream effector of InR/AKT signalling). NG2 also interacts with downstream components of the InR signalling pathway (e.g., PI3K-Akt-mTOR) to promote cell cycle progression and regulate the expression of its downstream effectors in a positive feedback loop. Together, all of these findings indicate that Ia-2, NG2/Kon, and insulin signalling have a common function across animals in reprogramming glial cells into becoming neural stem cells (Harrison, 2021).

Intriguingly, dpn was mostly induced in neuropile associated glial cells and was only induced in other glial types with overexpression of active RasV12 together with Dilp-6. Thus, perhaps prominently neuropile glia have neurogenic potential. Of the neuropile glia, Drosophila 'astrocytes' and midline glia express Notch, pros, and kon, as well as InR. The cells frequently called 'astrocytes' share features with mammalian NG2-glia. In mammals, the combination of Notch1, Prox1, and NG2 is unique to NG2-glia and is absent from astrocytes. Perhaps Ia-2 and Dilp-6 can only induce neural stem cells from NG2-like glia bearing this combination of factors. Notch activates glial proliferation and kon expression in Drosophila, and in the mammalian CNS, Notch promotes NG2-glia proliferation and maintains the progenitor state. In Drosophila, Notch and Pros also regulate dpn expression: Notch activates dpn expression promoting stemness, and Pros inhibits it, promoting transition to GMC and neuron. Thus, only glial cells with Notch and Pros may be poised to modulate stemness and neuronal differentiation. This study showed that InR is expressed in neuropile glia, which was confirmed by publically available single cell RNAseq data. Insulin signalling represses FoxO, which represses dpn, and thus ultimately activates dpn expression. As Notch and insulin signalling positively regulate dpn expression, and injury induces a Notch-dependent upregulation of Kon, which enables dilp-6 expression, and of Ia-2, which secretes Dilp-6, the data indicate that Notch-Kon/NG2-insulin synergy triggers the activation of dpn expression. Importantly, no evidence was found that Kon functions in neural stem cells. Thus, perhaps induced neural stem cells can generate only glia from daughter cells that inherit Kon, on which Repo and glial cell fate depend, or generate neurons, from daughter cells that lack Kon, but have Pros, on which Ia-2 depends. Thus, upon injury, Notch, Pros, Kon/NG2, Ia-2, and insulin signalling function together to enable the regenerative production of both glial cells and neural stem cells from glia. Intriguingly, developmental neural stem cells are thought to be eliminated through upregulation of Pros, induction of cell cycle exit, and terminal differentiation into glia. The current findings imply that such termination may not be final (Harrison, 2021).

To conclude, a neuron-glia communication relay involving Ia-2, Dilp-6, Kon, and InR is responsible for the induction of neural stem cells from glia, their proliferation, and limited neurogenesis. Neuronal Ia-2 and Dilp-6 trigger two distinct responses in glia: (1) in cortex glial cells, insulin signalling boosts Kon-dependent amplification of Dilp-6, glial proliferation, and glial regeneration. (2) In neuropile-associated NG2-like glial cells, insulin signalling unlocks a neurogenic response, inducing neural stem cell fate. As a result, these genes can drive the production of both glial cells and neurons after injury, enabling the matching of interacting cell populations, which is essential for regeneration (Harrison, 2021).


Burg, M. A., et al. (1996). Binding of the NG2 proteoglycan to type VI collagen and other extracellular matrix molecules. J. Biol. Chem. 271: 26110-26116. PubMed ID: 8824254

Estrada, B., Gisselbrecht, S. S., Michelson, A. M. (2007). Perdido encodes a component of a protein complex required for muscle guidance in Drosophila embryonic muscles. A. Dros. Res. Conf. 48: 15. Flybase abstract: FBrf0199168

Fang, X., et al. (1999). Cytoskeletal reorganization induced by engagement of the NG2 proteoglycan leads to cell spreading and migration. Mol. Biol. Cell 10: 3373-3387. PubMed ID: 10512873

Fukushi, J., et al. (2003). Expression of NG2 proteoglycan during endochondral and intramembranous ossification, Dev. Dyn. 228: 143-148. PubMed ID: 12950088

Fukushi, J., Makagiansar, I. T. and Stallcup, W. B. (2004). NG2 proteoglycan promotes endothelial cell motility and angiogenesis via engagement of galectin-3 and alpha3beta1 integrin. Mol. Biol. Cell 15: 3580-3590. PubMed ID: 15181153

Goretzki, L., et al. (1000). High-affinity binding of basic fibroblast growth factor and platelet-derived growth factor-AA to the core protein of the NG2 proteoglycan. J. Biol. Chem. 274: 16831-16837. PubMed ID: 10358027

Grako, K. A., et al. (1999). PDGF (alpha)-receptor is unresponsive to PDGF-AA in aortic smooth muscle cells from the NG2 knockout mouse. J. Cell Sci. 112: 905-915. PubMed ID: 10036240

Harrison, N. J., Connolly, E., Gascon Gubieda, A., Yang, Z., Altenhein, B., Losada Perez, M., Moreira, M., Sun, J. and Hidalgo, A. (2021). Regenerative neurogenic response from glia requires insulin-driven neuron-glia communication. Elife 10. PubMed ID: 33527895

Niehaus, A., et al. (1999). J. Stegmuller, M. Diers-Fenger and J. Trotter, Cell-surface glycoprotein of oligodendrocyte progenitors involved in migration. J. Neurosci. 19: 4948-4961. PubMed ID: 10366628

Nishiyama, A., Dahlin, K. J. and Stallcup, W. B. (1999). The expression of NG2 proteoglycan in the developing rat limb. Development 111: 933-944. PubMed ID: 1879362

Schnorrer, F., Kalchhauser, I. and Dickson, B. J. (2007). The transmembrane protein Kon-tiki couples to Dgrip to mediate myotube targeting in Drosophila. Dev. Cell 12: 751-766. PubMed ID: 17488626

Stallcup, W. B. and Dahlin-Huppe, K. (2001). Chondroitin sulfate and cytoplasmic domain-dependent membrane targeting of the NG2 proteoglycan promotes retraction fiber formation and cell polarization. J. Cell Sci. 114(Pt 12): 2315-25. PubMed ID: 11493670

Stegmuller, J., Werner, H., Nave, K. A. and Trotter, J. (2003). The proteoglycan NG2 is complexed with alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors by the PDZ glutamate receptor interaction protein (GRIP) in glial progenitor cells. Implications for glial-neuronal signaling. J. Biol. Chem. 278(6): 3590-8. PubMed ID: 12458226

Tillet, E., et al. (1997). The membrane-spanning proteoglycan NG2 binds to collagens V and VI through the central nonglobular domain of its core protein. J. Biol. Chem. 272: 10769-10776. PubMed ID: 9099729

Wen, Y., et al. (2006). Molecular basis of interaction between NG2 proteoglycan and galectin-3. J. Cell. Biochem. 98: 115-127. PubMed ID: 16365873

Biological Overview

date revised: 8 October 2007

Home page: The Interactive Fly © 2006 Thomas Brody, Ph.D.

The Interactive Fly resides on the
Society for Developmental Biology's Web server.