Molecular Mechanisms of Peripheral Nervous System Development
To better understand the molecular mechanisms that underlie neural development, we try to identify, through genetic screens, novel genes that affect peripheral nervous system (PNS) development. These genes are then placed in a molecular network of other players. In the past we isolated a gene that we named senseless because in its absence almost all the cells of the PNS are lacking. It encodes a zinc finger transcription factor expressed in every sensory organ precursor (SOP) of the PNS. senseless is necessary and sufficient to specify SOPs and initiate full differentiation of most PNS organs and functions as a binary switch in proneural clusters. Prepatterning genes activate proneural genes in clusters of cells that become competent to initiate neuronal development. The proneural genes then activate senseless transcription in a portion of the proneural cluster.
Expression of senseless can then be subdivided into two domains: a domain with a low level of expression, and a single cell in which a high level of expression is observed. At low levels, Senseless binds a specific DNA sequence to repress proneural genes transcriptionally, thereby quenching the ability of these cells of the proneural cluster to become neural precursors. In the future SOP, however, senseless and the proneural genes synergize, leading to up-regulation of proneural gene expression. This synergism is required for the ectodermal cell to become a neural precursor. Interestingly, this synergism does not depend on DNA binding but rather on the ability of the zinc fingers to interact with proneural and possibly other proteins. We are exploring the mechanism by which Senseless synergizes with proneurals to understand how it blocks Notch signaling in the SOP.
More recently, we have carried out large-scale genetic mosaic screens to identify new players that control PNS development. We have identified numerous new genes, including a number of genes that affect Notch signaling. These include rumi, which encodes a sugar-modifying enzyme that alters the sugar composition of Notch EGF (epidermal growth factor) repeats and affects all the known functions of Notch. Rumi is the first enzyme known to add glucose to proteins and is a highly conserved protein in eukaryotes. This modification is required for the proper folding of Notch, and loss of Rumi impairs cleavage of Notch at the membrane. We also discovered that another protein, Ero1L, is implicated in cysteine bridge formation and is required rather selectively for proper folding of a specific domain of the Notch protein. In the absence of Ero1L, Notch is retained in the endoplasmic reticulum and Notch signaling is temporally and spatially abolished.
By focusing on proteins that cause phenotypes that are similar to the loss of Notch, we discovered that three proteins that control actin polymerization affect the formation of an actin structure in the SOPs. Loss of these proteins also impairs the formation and density of microvilli on SOPs. Although these microvilli have not been described previously, they are abundant and long and seem to be the site where Delta is presented to the cell receiving the signal. The actin-polymerizing proteins are also involved in the transport of Delta within cells. Hence, their loss impairs Notch signaling in many cells.
We have recently identified three novel genes that affect Notch signaling and have initiated detailed phenotypic studies to define which portion of the Notch signaling pathway is affected.
The Function of Presynaptic Proteins in Synaptic Vesicle TraffickingThe endocytic molecular scaffold. Numerous proteins have been implicated in vesicle trafficking during endocytosis, but their precise function remains to be determined. We have isolated mutations in seven key players that affect endocytosis: AP180, synaptojanin (synj), endophilin (endo), dynamin-associated protein/intersectin (Dap160), eps15, tweek, and flower. Most of these genes encode proteins that are dramatically enriched in the nervous system and localized to nerve terminals, including neuromuscular junctions (NMJs). The presence of some of these proteins in the nervous system is essential, as neural expression of endo, eps15, Dap160, and flower rescues the lethality and phenotypes associated with the loss of the respective genes. This implies that they do not play a major role in tissues other than the nervous system. Tweek and Flower have not previously been identified. Both proteins are conserved from C. elegans to human and affect endocytosis at different stages.
The picture that is emerging is that there are three major forms of endocytosis at fly NMJs: a fast kiss-and-run mode of release at active zones, a slower clathrin-mediated form in periactive zones, and bulk endocytosis of membrane. The presence of at least two modes of retrieval has also been documented in hippocampal neurons but may not occur in all synapses. Our data indicate that synaptojanin (a polyphospoinositide phosphatase) and endophilin (a lysophosphatidic acid transferase) are required at precisely the same step during clathrin uncoating. Mutations in both genes display virtually identical phenotypes in FM1-43 dye uptake, transmission electron microscopy (TEM) at NMJs and photoreceptor terminals, and electrophysiological properties of NMJs and photoreceptors. In addition, double mutants (synj; endo) display phenotypes virtually identical to single mutants in flies and worms. Since most synaptic vesicles (SVs) are lost in endo and synj mutants but numerous SVs remain at active zones even after prolonged and intense synaptic activity, and since SVs fail to uptake FM1-43, we proposed that vesicles at active zones mainly use a kiss-and-run mode of release in these mutants. Our data also indicate that synj and endo are not required for kiss-and-run release but are mainly required for clathrin uncoating.
Another picture is emerging for the role of the Dap160 and Eps15 proteins, which have been shown to bind to each other as well as to numerous other endocytic proteins. Our data suggest that Dap160 and Eps15 are required as scaffolding proteins to localize Dynamin properly and to control actin dynamics. In their absence, endocytosis is only mildly impaired. A severe defect in endocytosis is observed, however, when the mutants are shifted to the restrictive temperature (34°C). In the absence of Dap160, numerous collared pits are observed at active zones and periactive zones (Figure 1). We also observe numerous irregularly sized vesicles in NMJ synapses and highly variable miniature excitatory junction potentials, similar to what we observed in AP180 mutants, suggesting a defect in clathrin caging. Dap160 and Eps15 appear to act at the same step in endocytosis, and double eps15;dap160 mutants exhibit many of the same features as eps15 or dap160 single mutants. Their presence seems important to ensure proper localization of Dynamin.
Novel players in presynaptic vesicle trafficking. To identify mutations in novel genes as well as genes that have previously been implicated in SV trafficking, we used the eyeless-FLP/FRT system to carry out four chemical mutagenesis screens. This system allows us to generate flies that have eyes that are homozygous mutant in an otherwise heterozygous fly. We opted to carry out five sequential assays to isolate mutations that affect synaptic transmission or synapse development. First, we only tested flies with apparent normal eye morphology (300,000 flies) for their ability to phototax (retained 10,000). Second, we screened for loss of ON and OFF responses in electroretinograms (ERGs) (retained 650). Third, flies with aberrant ERGs were used to establish stocks and retested (retained 450). Fourth, brains of adult flies were immunohistochemically stained to determine if photoreceptor synapses appear normal (450) (Figure 2). We then carried out TEM in the lamina to determine ultrastructural defects of photoreceptor synapses. We identified 62 genes with two or more alleles that affect neurotransmitter release or synapse formation. To map these genes, we developed a novel high-throughput mapping strategy. This strategy is based on the availability of P-element insertions every 16 kb in the genome. Using this approach, we identified the molecular lesions in numerous genes, a vast improvement to any other mapping strategy, including deficiency mapping. We identified sec15, a homolog of the yeast exocyst component SEC15, which was originally isolated in a yeast secretion screen. We have shown that Sec15 plays a highly specific role delivering cell adhesion molecules as well as the Delta protein in some cells. Loss of sec15 causes a specific targeting defect during synaptic partner selection, as well as cell fate specification defects in the PNS.
More recently, we have characterized a set of other mutants that play a role in synaptic vesicle fusion, including the V0 ATPase a1 100-kDa unit. Some mutations in this gene severely reduce SV fusion but do not affect vesicle acidification, implying that SNAREs (soluble NFS-attachment protein receptors) are not sufficient to promote fusion and that other proteins are required in the process (Figure 3). The phenotypes associated with the partial loss of the V0 ATPase are most similar to the loss of v-SNAREs (such as Synaptobrevin), arguing that it plays a role late in SV fusion. Most recently, we have identified mutations in hip14 (Huntingtin-interacting protein 14). Hip14 is a palmitoyl transferase for cysteine string proteins and SNAP25, proteins that play a pivotal rolein synaptic transmission.
Genome-wide Disruption/Tagging of Drosophila Genes
To permit rapid generation of targeted mutations for functional analysis of most Drosophila genes, we are attempting to create a transposable element insertion in every gene. In collaboration with the laboratories of Allan Spradling (HHMI, Carnegie Institution of Washington), Roger Hoskins (Lawrence Berkeley National Laboratory), and Gerald Rubin (HHMI, Janelia Farm Research Campus), we have created insertions in more than 60 percent of all Drosophila genes. We have now switched from P elements to Minos transposable elements and have developed new Minos-based transposable elements that greatly enhance the usefulness of the insertion stocks. All the stocks that we create are deposited in the Bloomington Stock Center and are available to the fly community (see http://flypush.imgen.bcm.tmc.edu/pscreen/).
We have also developed a new transgenesis platform for Drosophila, named P[acman]. This vector allows us to clone and manipulate very large pieces of DNA and to integrate them at specific genomic sites in the fly genome via ΦC31-mediated integration (http://flypush.imgen.bcm.tmc.edu/lab/pacman.html). We have recently generated two genomic libraries encompassing DNA fragments that average 20 or 80 kb. These libraries will allow unprecedented manipulations of more than 95 percent of the fly genes.