Current Research

Our research aims to elucidate the molecular mechanisms underlying the development of the peripheral nervous system (PNS) and synaptic transmission. Recently, key players that affect both processes have been shown to play a critical role in neurological disease and have therefore sparked our interest.

Molecular mechanisms of PNS development

Many key processes underlying neurogenesis were discovered in flies, including the proneural paradigm (1,2), Notch signaling (3,4,5), and asymmetrically localized fate determinants (6). The Notch signaling pathway controls numerous developmental decisions in the fly and vertebrates and has been implicated in neurodegeneration and memory formation (7). Through F3 lethal and F1 mosaic genetic screens, we have identified seven new proteins or protein complexes that affect Notch signaling (8,9,10,11,12,13).

senseless, a key player in PNS and eye development

Selection of a single sensory organ precursor (SOP) from a proneural cluster is in part mediated by the Senseless (Sens) protein. We isolated sens in an F3 genetic screen (14). Sens is a zinc-finger transcription factor that is required in all SOPs (15). The interplay between Notch signaling and proneural proteins leads to Sens expression in a few cells of the proneural cluster (Figure 1). Sens is necessary and sufficient for SOP specification and PNS organ differentiation (9,15,16).

Figure 1
Figure 1. Two ectodermal cells of a proneural cluster: the orange cell will become an SOP, expresses high levels of Sens and proneural proteins, and low levels of Enhancer of split E(spl). Sens expression is directly controlled by proneural genes. Sens in turn upregulates transcription of the proneural proteins Achaete (Ac) and Scute (Sc). The balance of E(spl) and Sens protein levels determines which cells will become SOP. Low levels of E(spl) coincide with high levels of Sens (orange cell). The elevated production of Delta in the presumptive SOP leads to the upregulation of Notch signaling in the blue cell where E(spl) antagonizes Sens and proneural gene expression. This Notch mediated feedback loop ensures the selection of the SOP. (Da = Daughterless).

We have shown that sens is induced by proneural gene activation (9). At low levels, Sens binds DNA and represses transcription of proneural genes (9,17). Lowering Notch signaling in the SOP lowers E(spl) (Figure 1), a Sens antagonist, thereby elevating levels of sens transcription in the presumptive SOP. Sens then interacts with proneural proteins to activate proneural gene transcription, a prerequisite for proper SOP specification (9,17). In some SOPs of the wing margin, Sens itself functions as a proneural gene (16), whereas in other cells it suppresses apoptosis and promotes differentiation (15,18). Sens is also critical in the visual system, where it directs R8 specification and axonal targeting (19,20,21), Bolwig organ differentiation (22), and Rhodopsin expression (23). The vertebrate homologue of Sens, Gfi-1, performs similar functions in neural development and hematopoiesis (24,25,26,27,28). In collaboration with Huda Zoghbi (HHMI, BCM) we have also shown that Gfi-1 binds Ataxin-1 and plays a role in mouse models for Spinocerebellar ataxia type 1 (29). We are currently characterizing the control of sens transcription and translation by microRNAs (30) collaboratively with Pascal Heitzler (Strasbourg).

Postranslational modifications of Notch affect signaling in the signal receiving cell

Notch signaling is pivotal in development, immune function and neurological disease, and memory (5,7). Because hundreds of sugar moieties are added to Notch's 36 EGF repeats (Figure 2; grey ovals), defining the role of post-translational modifications in Notch function is a challenge. We and others have shown that some modifications are crucial for Notch signaling, (4,31,32). We identified the first enzyme, named Rumi, that glycosylates any protein and showed that it mainly modifies Notch. This modification is critical for γ-secretase mediated cleavage of Notch (10,33). We have also discovered an unexpected role for a conserved thiol oxidase, Ero1L (Endoplasmic reticulum oxidoreductin 1-like) in Notch folding (11). In yeast, Ero1L plays a widespread role in forming cysteine disulfide bridges (34). Surprisingly, in flies, Ero1L mutant clones show Notch-specific defects in lateral inhibition and inductive signaling. We found that the fly Ero1L mediates disulfide bonding in Lin12-Notch repeats, which are unique to Notch proteins (LNR, orange ovals in Figure 2).

Figure 2
Figure 2. Notch signaling. Notch is modified in the ER by Ero1L (green oval in yellow cell), O-fucosyltransferase 1 (O-Fut1, blue), and O-glucosyl-transferase (Rumi, yellow). Fringe (turquoise) further modifies fucose residues in the Golgi. The Notch protein is integrated into the cell membrane where it binds activated Delta (displayed by green cell), leading to cleavage of Notch by ADAM proteases (purple) and gamma-secretases (orange). The Nicd domain is then internalized and enters the nucleus where it binds Su(H) to activate transcription. In the signaling cell, Delta is endocytosed via Neuralized and Epsin and trafficked back to the membrane via Sec15 and Rab11 as an active ligand.

Sec15 and Arp2/3/WASp affect trafficking of Delta in cell signaling cells

Delta, the ligand of Notch, must be endocytosed to signal properly (Figure 2). Two models have been proposed to rationalize this requirement: 1) Delta endocytosis stretches Notch to allow cleavage by ADAM proteases (35,36,37); or 2) endocytosis is required to convert Delta from an inactive to an active form, which is then exocytosed for signaling (38). Consistent with the second model, we isolated two novel complementation groups, sec15 and Arp3, that reveal the importance of Delta endocytosis and trafficking back to the membrane (Figure 3). Loss of sec15, an exocyst complex component implicated in intracellular vesicle transport (39), leads to extra neurons at the expense of support cells, consistent with disrupted Notch signaling (8). Delta aberrantly accumulates in basal areas of sec15 mutant cells. Moreover, Rab11, a recycling endosome protein that binds Sec15, is mislocalized in sec15 clones (Figure 3) (40). Together, our results suggest that sec15 mediates intracellular trafficking of Delta via recycling endosomes (8,40,41). We proposed that a defect in Delta recycling triggers the fate transformation in sec15 mutant sensory lineages, indicating that Delta endocytosis and subcellular trafficking is essential for Notch signaling (8).

Figure 3
Figure 3. Delta (purple rod) is inserted in the apical membrane as an inactive ligand, and is then endocytosed through the action of Neuralized and Epsin. Via Rab11 and Sec15 positive compartments, Delta is transported (orange rod) to the actin network (gray) and apically via WASp/Arp2/3.

This model is bolstered by our finding that Actin related protein-3 (Arp3), Arp1C, and WASp (Wiscott Aldrich Syndrome protein) mutations similarly disrupt Notch signaling (12,42). They are part of the WASp/Arp2/3 complex. During SOP specification, endocytosed vesicles containing Delta traffic to a previously undocumented prominent apical and lateral actin-rich structure (ARS) rich in microvilli (Figure 3). In Arp2/3 and WASp mutants, Delta endocytosis occurs properly, but the basal to apical transport of Delta-rich vesicles along actin bundles to microvilli is severely compromised, leading to a loss of Notch signaling. This work also suggests that defective Notch signaling may underlie immunodeficiencies associated with loss of WASp in humans (12,43).

Future goals

The discovery of new players that affect Notch signaling is high on our priority list. This information improves our mechanistic understanding of how Notch signaling is achieved and provides new insights in the molecular complexity of interacting networks, and allows us to develop a comprehensive picture of this pathway. We are currently characterizing three new players from our screens: Epsin homology domain binding protein 1 (CG15609), which controls trafficking of key regulators of Notch signaling, including Sanpodo and Scabrous; Transportin (CG7398), which may affect the nuclear protein distribution of RNA binding proteins; and tempura, a prenyl transferase. Note that none of the genes that we are studying were identified in a genome-wide analysis of Notch signaling in Drosophila by transgenic RNAi (44).

Molecular mechanisms of synaptic vesicle trafficking

Our goal is to define the role of specific proteins in exo- and endocytosis of synaptic vesicles (SV). These include proteins previously implicated on the basis of biochemical experiments (45) as well as new proteins isolated through genetic screens in my lab. Through forward and reverse genetic screens, we have identified mutations in numerous genes that affect neurotransmitter release and have defined their function in vivo (Table 1). We performed four F1 autosomal and one F3 X-chromosome (section 3) chemical mutagenesis screens using the eyeless-FLP/FRT system (46,47,48,49,50,51). This approach generates heterozygous flies (m/+) whose eyes are homozygous for randomly induced mutations (m/m). This method circumvents the lethality caused by disrupting synaptic transmission and allows us to screen on the basis of electroretinograms (ERGs). These recordings reveal electrical activity induced by light pulses and permit identification of mutant flies that are blind. We then perform secondary assays using immunohistochemistry and transmission electron microscopy (TEM) on photoreceptor terminals to determine if synapses are structurally altered. By combining genetic analyses, protein localization studies, electrophysiological recordings, FM 1-43 dye uptake experiments and TEM at the neuromuscular junction (NMJ), we have provided valuable insights into the function of essential synaptic proteins (Table 1).

Genes affecting exocytosis

All exocytic mutants that we have characterized typically display decreased evoked responses at low frequency stimulation at the fly NMJ, whereas miniature excitatory junctional potential (EJP) size is either unaffected or absent (Table 1). Ultrastructurally, exocytic mutants display subtle defects, such as increased SV numbers and aberrant SV distribution profiles. These exocytic mutants include the Vha100-1a subunit of the V0-ATPase (52) that binds Syntaxin and SNAP25 (53), and our data indicate that Vha100-1a function is required for SV fusion (49).

We have also identified a role for a palmitoyltransferase, Huntingtin-interacting protein 14 (HIP14) in SV exocytosis. hip14 mutants show exocytic defects and a nearly complete loss of synaptic transmission at high temperature (48). Our data indicate that HIP14 controls neurotransmitter release by regulating synaptic protein trafficking. Two exocytic components, Cysteine String Protein (CSP) and SNAP25, are mislocalized in hip14 mutants. CSP is a chaperone required for SV exocytosis and neuronal survival (54,55). The data indicate that palmitoylation of CSP by HIP14 is required for CSP's presynaptic localization. Notably, artificially targeting CSP to SV partially suppresses the exocytic defects in hip14 mutants (48). We are currently assessing whether Huntingtin, another protein that is palmitoylated by HIP14 (56), is likewise altered in hip14 mutants. These studies are aimed at elucidating the Ca2+ homeostasis defects that we observed in a fly model for Huntington's disease in collaboration with Juan Botas (BCM) (57,58).

Table 1. Exocytic and endocytic proteins studied in the lab.
Gene/Protein Function Evoked response Mini EJP Amplitude Ultrastructure References
  1Hz 10Hz   Normal SVs Large SVs  
Exocytosis
Hip14*§ transport of CSP Low Normal Normal 48
Straightjacket*§ targeting of Ca2+ channel Low Normal Normal 59
Synaptotagmin Ca2+ sensor Low Normal Normal 13,60
Rop/Munc18 docking and fusion of SV Low Normal Not tested 21,61
Syntaxin fusion of SV None None Normal 61,62
Complexin* clamp/fusion of SV Low Normal Normal 63
Vha100-1a§ fusion of SV Very low Normal/None Normal 49
Endocytosis
Flower* Ca2+ channel couples exocytosis to endocytosis Normal Low High Large 50
Tweek* PIP2 delivery Normal Low High Large 48
AP180 early endo. adaptor Normal Low High Large 64
Dap160§ scaffolding protein endo Normal Low High Large 65
Eps15§ scaffolding protein endo Normal Low Normal Large 66
Endophilin recruitment of Synj for uncoating Normal Very Low High Normal 67
Synaptojanin clathrin uncoating Normal Very Low Not tested Normal 47

The endocytic machinery

The phenotype associated with the loss of proteins that are required for endocytosis sharply contrast with the phenotypes associated with the loss of exocytic proteins (Table 1). Endocytic mutants typically show additional satellite NMJ boutons (68), fewer SV, normal EJPs at low frequency and a rundown at high frequency (10 Hz) stimulation. Furthermore, mutations in early endocytic genes (Table 1 and Figure 4: Flower, Tweek, AP180, Eps15 and Dap160) cause enlarged SV and an increase in mEJP amplitude. These data help us to pinpoint where and how the synaptic vesicle cycle is affected.

We recently identified two new genes, tweek and flower, not previously implicated in synaptic transmission and endocytosis. tweek encodes a novel giant (~500 kDa) protein that is evolutionarily conserved but contains no known protein domains (51). Our data show that PI(4,5)P2 and endocytic adaptors known to bind PI(4,5)P2 are mislocalized. Tweek may thus regulate endocytosis by regulating PI(4,5)P2 availability at synapses. Our current hypothesis is that Tweek is a scaffolding protein for PIP kinases.

Figure 4
Figure 4. SV endocytosis. Proteins studied in my lab (highlighted in yellow) are indicated at the position in the endocytic cycle at which they function. PIP2 (red dots), Clathrin (blue rectangles), AP180, AP2 (brown and green) and Dynamin (purple).

Endocytosis, like exocytosis, is dependent on influx of extracellular Ca2+ (69). Influx of Ca2+ through voltage-gated calcium channels is responsible for vesicle fusion, which requires Ca2+ in the tens of micromolar range (70,71). By contrast, less than 1 µM is required for endocytosis (72) and a subtle local Ca2+ elevation may trigger endocytosis. The source of this Ca2+ is however unknown. We recently identified a completely novel type of Ca2+ channel encoded by flower (flow). Flower is an evolutionarily conserved SV-associated 25 kDa tetraspanin protein (50). Our data show that loss of Flower disrupts endocytosis. Upon SV exocytosis, the protein localizes to endocytic zones. Flower contains a critical negatively charged residue (E) in the middle of the second transmembrane domain, and overexpression of Flower in salivary gland cells enhances Ca2+ influx. In addition, the purified protein forms a homotetramer that allows Ca2+ entry into reconstituted proteoliposomes. We are currently trying to determine the biophysical properties and 3-D structure of this new Ca2+ channel using X-ray crystallography (in collaboration with Youxing Jiang at UT Southwestern). Our data suggest that Flower plays a critical role in coupling exocytosis with endocytosis (50).

In summary, we propose that upon exocytosis, the SV membrane-associated protein Flower promotes Ca2+ influx, thereby triggering adaptor-protein recruitment and clathrin assembly (Figure 5). The enrichment of adaptor proteins is dependent on PI(4,5)P2 accumulation (73), which requires a novel protein, Tweek. Clathrin assembly is mediated by several adaptors including AP180 and AP2. Upon clathrin cage formation, Eps15 and Dap160 promote Dynamin function (66), which pinches off the vesicle (74). Endophilin then recruits Synaptojanin (47,75), which hydrolyzes PI(4,5)P2 (76) leading to disassembly of AP proteins and clathrin.

Figure 5
Figure 5. Protein interactions in the endocytic pathway. Proteins that are investigated in my lab are highlighted with yellow boxes and discussed in the text.

Upon endocytosis and uncoating of SV, they need to travel back to active zones for release. The loss of Dynamin related protein 1 (Drp1), a protein that affects the fission of mitochondria, indicates that the loss of mitochondria has surprisingly subtle effects on synaptic transmission, and mainly impairs SV transport through loss of ATP (77).

Novel genes in the Notch pathway, synaptic transmission and neurodegeneration

Given the efficiency of FLP/FRT screens in dissecting Notch signaling, synaptic transmission and neurodegeneration (78,79), we decided to tackle a similar, but more ambitious screen on the X-chromosome. We mutagenized flies with low EMS concentrations and screened 35,000 stocks for lethality (F2), thereby obtaining 5,900 homozygous lethal FRT X-chromosome stocks. We screened the F3 progeny with ey-FLP for ERG defects to identify synaptic transmission and neurodegenerative defects, and Ubx-FLP for wing notching and bristle loss to identify PNS development mutants. However, we also retained many mutations that cause gross morphological defects. After this analysis we retained ~2,200 mutants and have integrated more than 50% into complementation groups. We have already identified the corresponding gene for 100 complementation groups, which include mutations in Rab3 GEFs and AP1, and expect to map another 100 complementation groups in the next few months using duplication mapping and whole genome sequencing (80). Based on preliminary data, we anticipate that at least 50 groups will correspond to previously uncharacterized genes. Integrating these genes/proteins in Notch signaling, the SV cycle, and neurodegenerative pathways will significantly further our understanding of these processes.

Expanding the toolkit for Drosophila biology

We have had the privilege to advance Drosophila as a model system through large-scale development of new tools and reagents, which we make freely available to the Drosophila community. Together with Allan Spradling (Carnegie) and Roger Hoskins (LBNL), my lab has generated more than 12,000 publicly available stocks carrying single transposable element insertions that can be imprecisely excised to create mutations (81). This is the most commonly used method to create mutations in fly genes using reverse genetics (82). Currently, insertions in ~65% of all fly genes are available from the Bloomington Drosophila Stock Center and the GDP Project Database. We are now expanding the size and usefulness of this collection by creating strains carrying a new transposable element, MIMIC (Minos Mediated Integration Cassette). MIMIC inserts preferentially in introns and allows integration of any DNA in a gene of interest based on Recombination Mediated Cassette Exchange (83,84), enhancing our ability to manipulate flies.

We have also created a new transgenesis platform for flies. The P[acman] (ΦC31 artificial chromosome for manipulation) vector allows integration of large DNA fragments (85). Based on this technology, we constructed two highly versatile, publicly available whole-genomic libraries that allow manipulation of virtually all fly genes (see http://www.pacmanfly.org) (86). They provide direct access to recombineering- and transformation-ready genomic clones that can be integrated at precise locations in the Drosophila genome. Tagged clones allow one to assess gene expression and protein distribution and to efficiently rescue mutations and deletions (86).

References