Shared Mechanisms between Drosophila Peripheral Nervous System Development and Human Neurodegenerative Diseases

Abstract

Signaling pathways and cellular processes that regulate neural development are used post-developmentally for proper function and maintenance of the nervous system. Genes that have been studied in the context of the development of Drosophila peripheral nervous system (PNS) and neuromuscular junction (NMJ) have been identified as players in the pathogenesis of human neurodegenerative diseases, including spinocerebellar ataxia, amyotrophic lateral sclerosis, and spinal muscular atrophy. Hence, by unraveling the molecular mechanisms that underlie proneural induction, cell fate determination, axonal targeting, dendritic branching, and synapse formation in Drosophila, novel features related to these disorders have been revealed. In this review, we summarize and discuss how studies of Drosophila PNS and NMJ development have provided guidance in experimental approaches for these diseases.

Introduction

Drosophila is a powerful model organism to study neural development and neuronal maintenance [1]. The development of a neuron typically includes proneural induction, cell fate determination, axonal targeting, dendritic branching, and synapse formation (Figure 1). Based on their anatomical features and accessibility, different types of neurons are better suited to study different steps. For example, proneural induction and cell fate specification is more easily studied in external sensory organs (ESOs) [2], axonal targeting in photoreceptor cells [3], dendrite branching and pruning in multidendritic neurons [4], and synapse formation and function in the neuromuscular junction (NMJ) [5,6]. Although invertebrate biologists typically consider NMJs as part of the CNS, considering their peripheral localization and significant contribution to our understanding of synaptogenesis [5], we also included NMJ studies in this review. By making use of the unique features of different cell types of the fly PNS, and by applying sophisticated genetic manipulations [7], many evolutionarily conserved genes and pathways that regulate these processes have been discovered [1,8].

Figure 1. Molecular links between Drosophila PNS/NMJ development and human neurodegenerative diseases.

Left: Schematic diagrams of different steps of fly PNS/NMJ development. (1) Proneural induction, (2) lateral inhibition, (3) cell fate specification in the ESO (left) and in the eye (right), (4) dendrite development, (5) axon growth and targeting, (6) synapse development and maintenance. Middle: Selected signaling pathways and components that are commonly involved in different steps of fly PNS/NMJ development and human neurodegenerative diseases. Right: Selected human neurological diseases and selected causative genes. See main text for abbreviations.

Studies of genes involved in the development of Drosophila PNS and NMJ have provided important insights in the pathogenesis of several neurodegenerative diseases, including spinocerebellar ataxias (SCA), amyotrophic lateral sclerosis (ALS), and spinal muscular atrophy (SMA). For example, the proneural gene senseless (sens) was shown to regulate PNS development and later was implicated in the pathogenesis of SCA1 [9]. The Eph receptor was found to bind to major sperm protein (MSP) at NMJs and was later shown to be a key modifier of ALS in human [10–12]. Moreover, Fibroblast Growth Factor (FGF) signaling at the NMJ provided a link between this pathway and SMA pathogenesis [13,14]. This review will focus on the contribution and the interplay between basic research in the Drosophila PNS and NMJ fields and neurodegenerative diseases.

Links between molecular mechanisms of neurodegenerative diseases and PNS/NMJ development

Spinocerebellar ataxia (SCA)

SCAs are neurodegenerative diseases that primarily affect the spinal cord and the cerebellum [15]. A subset of SCA types (SCA1–3, 6–7, and 17) are caused by CAG trinucleotide repeat expansion which encode polyglutamine (polyQ) tracts in a variety of proteins. While long polyQ tracts by themselves can be cell toxic, recent studies indicate that modulations of the normal functions of the disease-causing proteins are also important in the pathogenesis [15].

SCA1 is caused by a polyQ expansion in ATAXIN-1 (ATXN1) [16]. The first link between ATXN1 and Drosophila PNS development came from a yeast two-hybrid screen which used Sens as bait to identify ATXN1 as an interactor [9]. Sens, a zinc finger transcription factor, regulates the early development of PNS organs [17]. Absence of sens leads to embryonic PNS cell death in Drosophila [17] and loss of the mouse homolog, Growth Factor Independent 1 (Gfi1), causes inner ear hair cell death [18]. During proneural induction in PNS development, the expression of proneural transcription factors is initiated in a group of ectodermal cells (proneural cluster) [19]. Subsequently, proneural proteins and Sens form a positive feedback loop to designate a single sensory organ precursor cell [20] through lateral inhibition, a process mediated by Notch signaling [17,21]. The SOP then undergoes several rounds of asymmetric cell division to generate a PNS organ. Differential Notch signaling activation and a combinatorial transcription factor code determines the fate of each cell type [2,22,23]. Overexpression of human ATXN1 can impact fly PNS development by promoting degradation of Sens and inhibiting SOP formation [9].

Similar to the fly, human ATXN1 interacts with Gfi1 [9]. Gfi1 is expressed in vertebrate PNS neurons [18] and cerebellar Purkinje cells (PC) [9], the most affected neurons in SCA1. Interestingly, overexpression of the polyQ-containing ATXN1 in mice (SCA1 mice) decreases the levels of Gfi1 prior to PC death, a phenotype that is further enhanced by loss of one copy of endogenous Gfi1. In summary, these data indicate that loss of Gfi1 and the consequent transcriptional dysregulation contributes to PC death in SCA1 pathology, and hence striking parallels can be drawn between PNS development in Drosophila and PC maintenance in vertebrates.

Another study links SCA1 and fly PNS development through Notch signaling [24]. In vertebrates, ATXN1 has another homolog, named ATAXIN-1 like (ATXN1L). To probe its function, Tong et al. (2011) overexpressed ATXN1L in Drosophila and observed a Notch signaling defect [24]. This phenotype is suppressed by mutations in Suppressor of Hairless [Su(H)], a transcriptional effector of Notch signaling [25]. Similarly, ATXN1L and ATXN1 exhibit repressive effects on Notch signaling through binding to the mammalian homolog of Su(H), CBF1/RBP-J_in cultured cells. Hence, ATXN1 and ATXN1L may form a repressor complex with Su(H) to inhibit Notch signaling, an idea that should be tested in vivo in mice. Interestingly, the protein level of ATXN1L is reduced prior to PC death in SCA1 mice [24], although Notch signaling activity in these dying cells has not been documented. Nevertheless, given that Sens/Gfi1 and Notch signaling form a feedback loop in the fly PNS [21] as well as in mammalian hematopoietic stem cell lineages [26], it is tempting to speculate that the disruption of this network in adult PC contributes to SCA1 pathogenesis.

The study of signaling pathways in Drosophila PNS development can also drive the discovery of potential therapeutic treatments. Phosphorylation of ATXN1 at Serine 776 (S776) is critical for its stability, and the polyQ-containing ATXN1 with S776A mutation exhibits reduced toxicity [27]. Therefore, suppressing the kinase activity for S776 phosphorylation is likely to reduce ATXN1 toxicity. A recent RNAi-based screen for such kinases in mammalian cells and fly retina identified the Ras-MAPK pathway as a regulator of ATXN1 phosphorylation [28]. Inhibition of numerous components in this pathway suppresses the neurodegenerative phenotype in SCA1 flies and mice, providing a potential therapeutic target. Ras-MAPK pathway is the downstream effector of receptor tyrosine kinase (RTK) signaling and has been extensively studied in Drosophila photoreceptor cell fate determination [8]. Many important components in this pathway, including downstream of receptor kinases (Drk) and Son of Sevenless (SOS), were identified in Drosophila forward genetic screens [20]. Therefore, identification and characterization of signaling pathway components in the Drosophila PNS facilitates in vivo studies in flies and mice.

Amyotrophic lateral sclerosis (ALS)

ALS is characterized by progressive loss of both upper and lower motor neurons, leading to muscle atrophy [29]. About 15% of all ALS are familial and are inherited dominantly. ALS causing genes are functionally diverse and mutations in these genes often interfere with normal protein functions [29,30]. Most of the genes that cause ALS are evolutionarily conserved and numerous Drosophila ALS related studies have revealed interesting features of pathogenesis [31]. Here, we focus on the studies of VAMP-associated membrane protein B (VAPB), Profilin1 (PFN1), and TAR DNA-binding protein 43 (TDP-43).

VAPB is the causative gene of ALS8 and SMA [32]. Its function was studied at the Drosophila NMJ prior to its identification as an ALS causing gene [33]. Drosophila VAPB (dVAP) mutants exhibit a severe reduction in bouton number and an increase in bouton size, while motor neuron specific dVAP overexpression causes the opposite NMJ phenotype. Based on experiments at the NMJ in flies and worms, the following model has emerged (Figure 2): VAPB is associated with the Endoplasmic Reticulum (ER) [9,34] where it is implicated in protein folding and ceramide transport together with ceramide transfer protein (CERT) to the Golgi complex [35,36]. Loss of VAPB in vertebrate cells leads to defects in ceramide transport from the ER to the Golgi [35]. Consistently, ceramide is enriched in the spinal cord in ALS patients [37]. In addition to its role in lipid homeostasis, VAPB also serves as a signaling molecule. Its MSP domain is cleaved and secreted from motor neurons into circulation, and the MSP moiety is also present in mammalian blood [11,38]. The disease causing mutation (P56S) in VAPB disrupts this cleavage and secretion, leading to accumulation of ubiquitinated aggregates in the ER. This induces an ER-associated unfolded protein response (UPR) [10,11,39], which is also observed in ALS model mice and ALS patients [30]. Secreted VAPB binds to Eph receptor (an RTK) on muscles and leads to un-clustering of glutamate receptors, whereas loss of VAPB causes hyper-clustering [10,11]. Glutamate receptor hyper-clustering promotes Ca²⁺ influx, which may mediate excitotoxicity. Interestingly, EPHA4, one of the human Eph homologs, has recently been identified as a modifier of ALS related phenotypes in mice and human [12]. Secreted VAPB also binds to Robo (a growth cone guidance molecule associated with kinases) and Lar (a receptor tyrosine phosphatase) to control mitochondria structure/function in muscles through Arp2/3-dependent actin cytoskeleton remodeling [38]. Loss of VAPB, Lar, or Robo impairs muscle function, resulting in severe mitochondrial defects and subsequent neurodegeneration. Similar muscular defects have also been reported in ALS model mice [40]. These data suggest that loss of VAPB is an initiator of disease progression, and that reintroduction of exogenous secreted VAPB may be a beneficial therapeutic intervention.

Figure 2. A schematic model of the VAPB signaling pathway.

VAPB (blue) is associated with the ER and regulates protein folding and ceramide transport from the ER to the Golgi together with CERT. The full length VAP protein is cleaved, releasing the MSP domain from the neuron. MSP binds the growth cone guidance receptors in the muscle, including Eph, Lar, and Robo. MSP binding to Eph leads to unclustering of Glutamate receptors, and binding to Lar and Robo is required for proper mitochondrial morphology, localization, and function through regulation of the actin and tubulin cytoskeleton. See main text for abbreviations.

Profilin is a downstream effector of Robo and Lar that regulates the formation of filamentous-actin. Drosophila Profilin is required for axon outgrowth during motor neuron development [41,42]. Recently, PFN1 was found to cause ALS18 and be responsible for 1–2% of familial ALS [43]. Overexpression of mutant PFN1 leads to formation of ubiquitinated protein aggregates and causes cytoskeletal defects including axonal outgrowth defects in mammalian cells, consistent with earlier fly studies. The studies on Profilin and VAPB indicate that the dynamics of actin polymerization in muscles is important in ALS pathogenesis. Actin and microtubules are key coordinators of many cellular processes [44] and in many cases, alterations in actin dynamics affect microtubule structure and vice versa. Moreover, Drosophila dVAP mutants exhibit disorganized microtubular structures [33], and mutations in Dynactin 1 (DCTN1; a microtubule and dynein binding protein) are also associated with ALS [45]. These studies indicate that cytoskeleton alterations play key roles in the pathogenesis of ALS. The main question is to determine if they all impinge on similar or different cellular processes.

TDP-43 is the causative gene of ALS10 [46] and encodes a DNA/RNA-binding protein that regulates transcription, alternative splicing, and stability of hundreds of RNAs [29]. Among these RNA targets, mutations in TDP-43 may affect certain RNAs which play a predominant role in the pathogenesis of ALS. A recent Drosophila study revealed one such potential RNA, microRNA-9a (miR-9a) [47]. miR-9a, a conserved miRNA that regulates SOP development by fine-tuning the expression of Sens [48,49]. Loss of miR-9a leads to Sens upregulation, causing formation of additional sensory organs. In Drosophila TDP-43 mutants, the level of miR-9a is reduced and hence the number of SOPs is increased. While it remains to be established whether miR-9a is reduced and whether Gfi1 is upregulated in TDP-43 mutant mice, fly data provide an interesting inroad to investigate this avenue in vertebrates. Moreover, miR-9a minimizes the phenotypic variation resulting from individual genetic variants in Drosophila: i.e. miR-9a sets a threshold for the amount of Sens, and the consequences of genomic variations that affect the levels of Sens are filtered, providing phenotypic stability in SOP formation [50]. When the level of miR-9a decreases, SOP formation becomes sensitive to small genetic variances. In ALS10, if the level of miR-9a or other TDP-43 target miRNAs are affected, loss of buffering in genetic variation may promote the pathogenesis caused by minor mutations in the patient’s genome.

Spinal muscular atrophy (SMA)

SMA is an autosomal recessive disorder with progressive loss of lower motor neurons and muscle atrophy [51]. Mutations in Survival of Motor Neuron 1 (SMN1) contribute to the majority of SMA [52]. SMN1 is ubiquitously expressed and regulates pre-mRNA splicing. Drosophila smn mutants exhibit alterations in NMJ morphology, neurotransmission, and locomotion, similar to SMA patients [51,53–56]. Novel insights into the molecular mechanisms of SMA originated from a modifier screen.

Components of the BMP/TGFβ and FGF (an RTK) signaling pathways were identified as enhancers of the smn mutant phenotypes [14,54]. BMP/TGFβ signaling is an important retrograde signal for synapse development at the NMJ, while the role of FGF signaling in this context is not well established [6,57]. Loss of BMP/TGFβ signaling at the NMJ leads to defects in pre-synaptic structure and neurotransmitter release [6,57]. When the level of SMN decreases, the level of phosphorylated Mothers Against Decapentaplegic (pMAD), an effector of BMP/TGFβ signaling, is reduced [54]. In addition, elevated BMP/TGFβ signaling can suppress the NMJ morphology defect in smn mutant, indicating that defective BMP/TGFβ signaling contributes to mutant phenotypes [54]. The same modifier screen also isolated FGF signaling components as modifiers, and activation of FGF signaling specifically in the muscle is sufficient to reverse the NMJ morphology defects in smn mutants [14]. Subsequent studies have documented that FGF signaling is also misregulated in an SMA mouse model [13]. Therefore, pharmacological manipulation of BMP/TGFβ and FGF signaling can be considered as possible therapeutic strategies against SMA.

Studies of the PNS of smn mutant flies also shed light on the neuronal circuitry critical for SMA. Reintroduction of SMN protein in a subpopulation of cholinergic proprioceptive and interneurons of the motor circuit, but not muscles or motor neurons, can rescue NMJ phenotypes in smn mutants [55]. Conversely, specific knockdown of smn in these cholinergic neurons causes SMA-related phenotypes. Therefore, the defects in the sensory-motor circuit may be the primary cause for the motor system defects in smn mutant flies. Interestingly, SMN1 overexpression in cholinergic motor neurons largely rescues motor symptoms in SMA model mice [58,59]. These data indicate that cholinergic neurons of motor circuits are the major site of action for SMN in flies (peripheral/interneurons) and in mammals (lower motor neurons) [60].

Since SMN is ubiquitously expressed, the question remains as to which neurons are selectively vulnerable in SMA patients. One recent study found that splicing of several U12 intron-containing pre-mRNAs are impaired in smn mutant flies [56]. Among these, selective knockdown of stasimon (Stas, a putative vesicular trafficking protein) in cholinergic neurons exhibited SMA-related defects, while many of the SMA-related defects in smn mutant flies can be repaired by overexpression of stas in the motor circuit. These data suggest that mRNA splicing defects in the motor circuit, rather than in motor neurons per se, are responsible for SMA-related motor defects in smn mutant flies. The same splicing defect in Tmem41b (mouse homolog of stas) is also observed in SMA mice, indicating that this is an evolutionarily conserved target [56]. It will be interesting to investigate whether overexpression of Tmem41b can correct the SMA-related defects in SMA mice.

Conclusions

Many evolutionarily conserved genes involved in Drosophila PNS and NMJ development have not yet been linked to neurological disorders in humans. On the other hand, sequencing of human exomes is revealing numerous variants in patients with rare genetic diseases, and many homologs of these genes have been or can be studied in flies. Although mutant flies often do not display the exact phenotypes as humans, the information obtained from studying the endogenous function of a conserved gene is very valuable to identify and characterize the molecular mechanisms underlying diverse human diseases. For example, mutations in flies that cause PNS organ defects may affect Notch signaling and help identify genes that cause neurological or other diseases related to Notch signaling in human [61]. Another examples are Atonal and Sens, fly proneural transcription factors which are required for PNS and eye development [62,63]. In vertebrate, loss of Atoh1/Math1, a homolog of Atonal, leads to cerebellar defects [64], and loss of Gfi1 (Sens homolog) leads to PC loss [9]. Hence, similar molecular mechanisms are probably used in both development and maintenance of the nervous system. Moreover, a method to identify non-obvious equivalences between mutant phenotypes in various species has been proposed. Such “phenologs” can now be used to predict genes associated with human diseases [65]. By systematically identifying and characterizing additional genes involved in PNS and NMJ development, we foresee that one will be able to predict and discover novel human disease genes using a similar approach. In sum, the Drosophila PNS and NMJ have been and will continue to be valuable model systems to reveal the molecular mechanisms underlying neurodegenerative disorders.

Highlights.

• VAPB signaling at the NMJ is involved in the pathogenesis of ALS

• PNS studies implicate a link between microRNA mediated genomic stability and ALS

• A major mRNA splicing target of SMN in cholinergic neurons contributes to SMA