Autoimmune diseases such as rheumatoid arthritis (RA) are associated with debilitating chronic inflammation, autoantibody production, articular bone erosions and systemic bone loss. The underlying mechanisms and cell types that initiate these diseases are not fully understood, and current therapies mainly address downstream mechanisms and do not fully halt disease progression in all patients. Moreover, previous studies have largely focused on the role of adaptive immunity in driving these diseases, and less attention has been given to the contribution of innate immune pathways such as DNA sensor signaling pathways in initiating and/or perpetuating autoimmunity and erosive inflammatory arthritis. Detection of microbial nucleic acids by DNA sensors such as endosomal toll-like receptors (TLRs) and cytosolic sensors is an early form of antiviral defense. Upon detection of nucleic acid, TLRs dependent on Unc93B and cytosolic sensors dependent on the adaptor stimulator of interferon genes (STING) orchestrate production of type 1 interferons and pro-inflammatory cytokines to resolve infection. Additionally, the cytosolic DNA sensor absent in melanoma 2 (AIM2), which is not dependent on STING, also recognizes microbial DNA and coordinates the cleavage of pro-IL-1β. Previous studies have largely focused on the role of these DNA sensors in macrophages and dendritic cells in the context of antiviral immunity. In recent years, however, the inappropriate recognition of host nucleic acids by these sensors has been associated with several autoimmune diseases including RA. This dissertation aims to delineate the mechanisms by which DNA sensors contribute to inflammatory arthritis and bone remodeling in the context of a murine model of autoimmunity. In DNase II deficient mice, excessive accrual of undegraded, endogenous DNA leads to robust production of type 1 interferons (IFNs) and proinflammatory cytokines. The high levels of type 1 IFNs result in anemia and embryonic lethality; therefore, the gene for the type 1 IFN receptor (IFNaR) has also been deleted so that the mice survive. DNase II-/- IFNaR-/- double knockout (DKO) mice develop erosive inflammatory arthritis, anti-nuclear antibodies, and splenomegaly not seen in the DNase II+/- IFNaR-/- (Het) control group. To evaluate whether cytosolic or endosomal DNA sensors contribute to the clinical manifestations of DKO mice, genes involved in TLR or cytosolic sensor signaling were deleted on the DKO background. Genetically altered mice include STING/DNaseII/IFNaR TKO (STING TKO), AIM2/DNase II/IFNaR TKO (AIM2 TKO), and Unc93b/DNase II/IFNaR TKO (Unc93 TKO) mice. Our hypothesis was that the STING, AIM2, and/or Unc93 pathways would contribute to the autoimmune manifestations in DNase II deficient mice. Rigorous examination of inflammation in these lines revealed important roles for both the STING and AIM2 pathways in arthritis. Despite the substantial effects of the STING and AIM2 pathways on arthritis, STING TKO and AIM2 TKO mice still exhibited prominent autoantibody production. Interestingly, inflammation persisted in Unc93 TKO mice while autoantibody production to nucleic acids was abrogated. Collectively, these data indicate that innate immune pathways contribute to the initiation/perpetuation of inflammatory arthritis and demonstrate that cytosolic and endosomal pathways play distinct roles in the manifestations of autoimmunity. Moreover, they reveal a previously undescribed role for AIM2 as a sensor of endogenous nucleic acids in inflammatory arthritis. Thus, therapeutics that target the STING and AIM2 pathways may be beneficial for the treatment of inflammatory joint diseases. While the role of hematopoietic cells in driving autoimmunity has been well established, the contribution of stromal elements to disease pathogenesis is less well understood. Therefore, we generated bone marrow chimeras to delineate the contribution of hematopoietic and non-hematopoietic cells to the various autoimmune manifestations in DKO mice. These studies revealed that both donor hematopoietic and host radioresistant cells are required for inflammation in the joint as well as for other features of autoimmunity in DKO mice, including splenomegaly, extramedullary hematopoiesis, and autoantibody production. This data demonstrates that stromal host cells play a major role in DNA-driven autoimmunity. Moreover, these results suggest that targeting not only hematopoietic but also stromal elements may be advantageous in the setting of inflammatory arthritis. In the final chapter of this thesis, a role for innate immune sensor pathways in bone is described. The majority of inflammatory arthritides have been shown to lead to systemic loss of bone. Surprisingly, however, we found that DKO mice accumulate trabecular bone in the long bones over time as well as ectopic bone in the spleens, both sites of robust DNA accrual. Moreover, deficiency of the STING pathway abrogated this bone accumulation. Collectively, these data demonstrate that DNA accrual promotes dysregulated bone remodeling through innate immune sensing pathways. These findings are the first to reveal a role for the STING pathway in bone and may unveil novel targets for the treatment of diseases associated with bone disorders.

Investigators have been attempting to identify the receptor for ribosomes on the rough endoplasmic reticulum (RER) for almost 20 years, yet the ribosome receptor has remained elusive. Rough microsomal membranes contain endogenous ribosomes bound in at least two types of interactions. Loosely associated ribosomes can be removed by extraction with a high ionic strength solution, but ribosomes that were actively engaged in translocation when the membranes were isolated remain tethered to the membrane by a nascent polypeptide (Adelman et al., 1973). The original assay for the ribosome receptor involved stripping all of the endogenous ribosomes off of intact membranes before adding back a quantitated amount of ribosomes. More recent assays have employed detergent solubilization of the membrane and then reconstitution of the membrane proteins into lipid vesicles before adding back ribosomes. In both cases ribosome binding to its receptor is measured in an assay that does not involve translation or translocation. We utilized a crosslinking assay to attempt to identify membrane proteins that function as a binding site for ribosomes engaged in protein translocation across the endoplasmic reticulum. In vivo bound ribosomes that remain associated with the membrane after extraction with a high ionic strength solution are likely to be bound to a functional translocation site. The water soluble, membrane impermeable, thiol-cleavable crosslinker 3,3'-dithiobis (sulfosuccinimidylpropionate) was selected to limit reaction to protein domains located on the cytoplasmic face of salt extracted microsomal membrane vesicles. A specific subset of RER proteins was reproducibly crosslinked to the endogenous ribosomes. Immunoblot analysis of the crosslinked products with antibodies raised against signal recognition particle receptor, ribophorin I, and the 35 kD subunit of the signal sequence receptor demonstrated that these translocation components had been crosslinked to the ribosome, but each to a different extent. The most prominent polypeptide among the crosslinked products was a 180 kD protein that had recently been proposed to be a ribosome receptor (Savitz and Meyer, 1990). RER membrane proteins were reconstituted into liposomes and assayed with radiolabeled ribosomes in an in vitro binding assay to determine whether ribosome binding activity could be ascribed to the 180 kD protein. Differential detergent extraction was used to prepare soluble extracts of microsomal membrane vesicles that either contained or lacked the 180 kD protein, as determined by Coomassie blue staining of a polyacrylamide gel. Liposomes reconstituted from both extracts bound ribosomes with essentially identical affinity. Additional fractionation experiments and functional assays with proteoliposomes demonstrated that the bulk of the ribosome binding activity present in detergent extracts of microsomal membranes could be readily resolved from the 180 kD protein by chromatography. Taken together, the evidence indicates that the 180 kD protein is in the vicinity of membrane bound ribosomes, yet does not correspond to the ribosome receptor. To continue the investigation of ribosome binding, an assay was designed to characterize ribosome-nascent chain complexes bound to the microsomal membrane during translocation. A series of translocation intermediates consisting of discrete sized nascent chains was prepared by including microsomal membranes in cell-free translations of mRNAs lacking termination codons. Proteinase K was then used as a probe to detect cytoplasmically and lumenally exposed segments of nascent polypeptides undergoing transport. Only those partially translocated nascent chains of 100 amino acids or less were insensitive to protease digestion by externally added protease. It was concluded that the increased protease sensitivity of larger nascent chains is due to the exposure of a segment of the nascent polypeptide on the cytoplasmic face of the membrane. In contrast, shorter nascent polypeptides appear not to have lumenally exposed segments. Ultimately, a functional assay for the ribosome receptor should include binding studies conducted under physiological conditions. For this purpose, an assay was developed that allowed translation, translocation, and termination of a secretory protein to be monitored with probes designed to independently quantitate translating and non-translating ribosomes. A synchronized wheat-germ translation system was programmed with bovine preprolactin mRNA and aliquots were taken at various time points before and after adding membranes. The samples were then separated into membrane bound and soluble species by centrifugation. RNA was isolated from each supernatant and pellet sample and blotted onto nylon sheets. By probing the dot blots with probes that hybridize with either the 5S RNA of wheat germ ribosomes or the preprolactin transcript, the translating ribosomes could be monitored without the interference of the endogenous canine ribosomes on the membrane. By comparing the total amount of preprolactin transcript that bound to the membrane versus the total amount of wheat germ ribosomes bound to the membrane, it was discovered that the vast majority (>99%) of wheat germ ribosomes that bound to the microsomal membrane were non-translating ribosomes. In later experiments it was found that the non-translating ribosomes did not compete with the translating ribosomes; under all conditions tested, the translating ribosomes had access to translocation sites on the microsomal membrane. One interpretation of this data is that all ribosome binding sites are not identical. It may be that functional sites for translocation are a distinct subclass of total ribosome binding sites. Another interpretation is that a ribosome in a nascent chain-SRP complex has a much higher affinity for the ribosome receptor than nontranslating ribosomes or 60S subunits. Perhaps the non-translating ribosome can not compete with ribosomes engaged in translocation. As stated earlier, ribosomes do make at least two kinds of interactions with the microsomal membrane surface. This data may be indicative of those types of interactions.

In an attempt to more fully understand the generation of antibody diversity to carbohydrate antigens, we produced and characterized a panel of hybridoma cell lines specific for group A streptococcal carbohydrate from mice injected with the intact bacteria (minus the hyaluronic acid capsule and cell wall protein antigens). We have analyzed the use of heavy and light chain variable region genes in the early (day 7) and late response (hyperimmune) and have determined the nucleotide sequence of the dominant VH gene used in several of our hybridomas. Our data allowed us to assess the extent to which the recombination of various V, D, and J gene segments and somatic mutation contribute to antibody diversification in this system. In this report we confirm that a minimum of two VH and four VK gene segments are used to encode this response. We extend this analysis to show that multiple D and J gene segments are used and that a significant amount of junctional variability is tolerated in CDR 3. Our results also suggest that there is a positive selection for somatic mutation in CDR 1 during the hyperimmune response to group A streptococcal carbohydrate.

The purpose of this work was to determine whether the previously identified fatty acid activation of large conductance Ca2+-activated K+ (BK) channels from rabbit pulmonary artery smooth muscle cells was due to the direct interaction of the fatty acid with a site on the channel protein. If this was found to be the case, this study would also attempt to identify the site of fatty acid-protein interaction. Fatty acids released from membrane phospholipids by cellular phospholipases or available to the cell from the extracellular environment are important signaling molecules. Fatty acids can modulate the activity of a large number of molecules including protein kinases, phospholipases, adenylate and guanylate cyclases, G-proteins and ion channels. Fatty acids have also been shown to activate transcription of genes belonging to the steroid/thyroid superfamily of receptors. The actions of fatty acids on signal transduction pathways can be direct, whereby the fatty acid molecule itself is responsible for changes in the activity of enzymes, ion channels and other proteins. Alternatively, the effects of fatty acids may be indirect. In this case, biologically active lipids, produced from the metabolism of arachidonic acid are responsible for changes in cellular signaling. A previous study on the fatty acid modulation of rabbit pulmonary artery smooth muscle BK channel activity concluded that channel activation by fatty acids did not involve cycloxygenase, lip oxygenase and P450 metabolites (122), eliminating this indirect action of fatty acids as a possible mechanism. When dealing with the effects of fatty acids on membrane bound ion channel proteins, other mechanisms of action are also possible. For example, fatty acids are capable of entering the cell membrane and can thus affect properties of the lipid bilayer, such as membrane fluidity or membrane surface charge, that may consequently alter the activity of ion channel proteins. In addition, fatty acid mediated alterations of ion channel activity could result from the effect of fatty acids on ion channel associated proteins. To determine the mechanism of action of fatty acids on the activity of BK channels from rabbit pulmonary artery smooth muscle cells, all of the above mentioned mechanisms were considered. Most of the experiments described here were carried out using the patch-clamp technique and current recordings were performed in cell free, excised inside-out or outside-out membrane patches, in the absence of any added nucleotides and calcium. As a first step towards understanding how fatty acids modulate BK channel activity, as well as the type of protein site with which fatty acids may be interacting, we determined the structural features of the fatty acid molecule that are required for channel modulation. To do this the effects of a range of fatty acids and other lipids on BK channel activity were examined. The features required for BK channel activation were found to be the negatively charged head group and a carbon chain of greater than eight carbons. We also found that positively charged lipids produced the opposite effect of negatively charged lipids, a decrease in BK channel activity. A similar chain length requirement was also necessary for channel inhibition by positively charged lipids; short chain compounds did not alter activity while those with fourteen carbons or greater decreased activity. The identification of these required structural features suggested that a specific interaction between the charge on the lipid head group is required for channel modulation by these lipids. The requirement for a chain length of greater than eight carbons also suggests that a hydrophobic interaction is necessary for these lipids to be effective modulators of this channel. In addition, the identification of these required structural features makes it unlikely that modulation of BK channel activity by these lipid compounds is a consequence of a perturbation of the lipid environment in which the channel resides. Experiments were then carried out to determine whether modulation of BK channel activity by fatty acids and other charged lipids involved any of the following indirect mechanisms of action: 1) alterations in the concentration of calcium in the vicinity of the channel due to changes in membrane surface charge, or due to calcium stores attached to excised membrane patches, 2) alterations in the membrane electric field that the channel perceives due to changes in membrane surface charge and 3) changes in the activity of membrane bound protein kinases or protein phosphatases. In experiments where high ionic strength solutions were used to shield membrane surface charge, fatty acids and other charged lipids were still able to modulate BK channel activity suggesting that fatty acids do not act through a mechanism involving surface charge. Experiments carried out in high concentrations of EGTA (20 mM) make it unlikely that calcium is involved in the modulation of BK channels by fatty acids and other lipids. The involvement of membrane bound kinases or phosphatases is also unlikely as fatty acids effectively modulated BK channel activity in the presence of staurosporin, a kinase inhibitor, and okadaic acid, a phosphatase inhibitor. The elimination of these indirect and non-specific suggests that fatty acids and charged lipids modulate BK channel activity by directly interacting with, either the channel protein itself, or some other channel associated protein. To obtain further evidence that this indeed is the mechanism by which these lipids modulate BK channel activity; experiments were carried out to identify the site of action (i.e. side of the membrane) of both negatively and positively charged lipids. The negatively charged palmitoyl coenzyme A (PCoA) and a myristoylated positively charged peptide (myr-KPRPK), two compounds that are incapable of flipping across the bilayer, were used to identify the site of action of negatively and positively charged lipids. PCoA and myr-KPRPK produced their predicted effects of BK channel activation and suppression, respectively, only when they were applied to outside-out membrane patches. These experiments, therefore, support the contention that fatty acids and other charged lipids modulate BK channel activity by interacting with a site on the channel protein or a channel associated protein and that this site is found on the external membrane surface. If the site responsible for channel modulation by fatty acids and other charged lipids is contained within the BK channel protein itself, other members of this family may also possess this site, and thus be modulated by fatty acids. Experiments were performed, therefore, to determine whether the BK cloned channels, mslo, hslo and bslo could also be modulated by fatty acids. These cloned channels were expressed in the Xenopus oocytes, and whole-cell currents were recorded using the two-electrode voltage clamp technique. The fatty acids myristic and arachidonic acid were able to increase whole-cell current of oocytes expressing all clone types. The modulation of these cloned channels by fatty acids did not appear to involve calcium, the BK β-subunit or a bioactive metabolite of arachidonic acid. Although all possible mechanisms of action were not addressed in this study, the results support the idea that the site of fatty acid interaction resides in the channel protein itself. Taken together, therefore, these studies suggest that it is very likely that fatty acids and charged lipids modulate the activity of BK channels from smooth muscle cells of the rabbit pulmonary artery by directly interacting with an externally located site on the channel protein itself. The BK clones, mslo, hslo and bslo, are also modulated by fatty acids and it is likely that they share the same mechanism of action seen for BK channels from rabbit pulmonary artery smooth muscle cells.

Signal transduction by G-proteins (a heterotrimer membrane protein composed of an α, β, and γ subunit) requires that the α-subunit undergoes a transition from a GDP-bound inactive state to an activated GTP-bound state. The exchange of GDP for GTP leads to a conformational change in the α-subunit that results in the loss of affinity for the βγ subunits. We predicted that appropriate genetic manipulation of key regions of the α-subunit could result in the induction of the active conformation that would mimic at least in part the activated GTP-bound state. We have demonstrated that the substitution of the 38 amino acid residue carboxyl termimus of Gαs with the last 36 amino acid residues of Gαi2 resulted in a chimeric Gα-subunit (C4) that exhibits a constitutively active Gαs-like activity. Similarly, the substitution of the amino terminal 61 amino acid residues of Gαs with the first 54 residues of Gαi2 also resulted in a chimeric Gα-subunit that is persistently active (Gs like). We have also generated point mutations in the Gαs subunit that are comparable to the activating mutations in the ras protein. Our results suggest that point mutations in the signature sequence of the A (Val 49) and C (Thr 225) homologous regions that are implicated in regulating the GTPase activity of the molecule also resulted in the activation of the subunit. The present study has identified four key regions of the α-subunit that are critical for the activity and regulation of the Gs protein.

Obesity results from expansion of white adipose tissue. The inability of white adipose tissue to adequately store lipids leads to ectopic deposition of lipids in non-adipose tissue that can lead to systemic insulin resistance. It is well known that insulin resistance correlates with inflammation of adipose tissue in obese animals and humans. Decreasing inflammation in the adipose tissue has been proven as a therapeutic strategy for improvement of insulin sensitivity in vivo. Numerous factors secreted by immune cells, including macrophages, have been suggested as regulating adipose tissue insulin sensitivity. In the first part of my thesis, I describe the role of one such factor, CD40 in adipose tissue inflammation. The CD40-CD40L dyad acts as co-stimulation in the interaction of antigen-presenting cells, such as macrophages and dendritic cells, with effector cells, such as T cells, in adaptive immunity. We found that CD40 knockout mice were smaller but surprisingly more insulin resistant and glucose intolerant compared to wild-type mice when fed a high fat diet. Consistent with their metabolic phenotype, knockout mice displayed increased adipose tissue inflammation with infiltration of immune cells including macrophages and T cells. Consistent with increased inflammation, CD40 knockout adipose tissue displayed decreased lipid storage. Deficiency of CD40 also led to increased lipid deposition in liver, which may be due to increased lipid release into circulation from the adipose tissue as well as increased lipid synthesis in the liver. CD40 knockout mice had increased hepatic insulin resistance and increased gluconeogensis despite decreased hepatic inflammation. These findings suggest that CD40 is a novel regulator of adipose tissue inflammation in diet-induced obesity. In the second part of this thesis we examined perivascular adipose tissue and brown adipose tissue for the presence of inflammation. In contrast to visceral adipose tissue, macrophage infiltration was absent in perivascular and brown adipose tissue as defined by reduced F480+ cells by flow cytometry and immunohistochemistry. We also found that perivascular adipose tissue was similar to brown adipose tissue as shown by gross morphology and gene expression pattern. Inflammatory gene expression was not increased in brown or perivascular adipose tissue in obese mice as determined by microarray gene expression analysis. These findings suggest that perivascular adipose tissue is more similar to brown adipose tissue than white adipose tissue and that both perivascular and brown adipose tissue are resistant to inflammation. We conclude that, (1) CD40 protects against adipose tissue inflammation in diet-induced obesity, (2) the CD40 knockout mouse is an interesting model of hepatic steatosis with decreased inflammation and (3) perivascular adipose tissue is almost identical to brown adipose tissue in obese mice and that both are resistant to inflammation.

Activity plays diverse roles in shaping neuronal development and function. These roles range from aiding in synaptic refinement to triggering cell death during traumatic brain injury. Though the importance of activity-dependent mechanisms is widely recognized, the genetic underpinnings of these processes have not been fully described. In this thesis, I use the motor circuit of Caenorhabditis elegans as a model system to explore the functional and morphological consequences of modulating neuronal activity. First, I used a gain-of-function ionotropic receptor to hyperactivate motor neurons and asked how increased excitation affects neuronal function. Through this work, I identified a cell death pathway triggered by excess activation of motor neurons. I also showed that suppression of cell body death failed to block motor axon destabilization, providing evidence that death of the cell body and of motor axons can be genetically separated. Secondly, I removed excitatory drive from a simple neural circuit and asked how loss of excitatory activity alters circuit development and function. I identified excitatory motor neurons as master regulators of inhibitory synaptic connectivity. Additionally, I was able to identify previously undescribed activity-dependent mechanisms for regulating inhibitory synapses in both developing and mature neural circuits. Finally, I show data to implicate the highly conserved genes neurexin and neuroligin in determining inhibitory synapse connectivity. Collectively this work has lent insight into activity-dependent mechanisms in place to regulate neuronal development and function, a core function of neurobiology that is relevant to the study of a wide range of neurological disorders.

Early events in folding can determine if a protein is going to fold, misfold, or aggregate. Understanding these deterministic events is paramount for de novo protein engineering, the enhancement of biopharmaceutical stabilities, and understanding neurodegenerative diseases including amyotrophic lateral sclerosis and Alzheimer's disease. However, the physicochemical and structural biases within high energy states of protein biopolymers are poorly understood. A combined experimental and computational study was conducted on the small β/α-repeat protein CheY to determine the structural basis of its submillisecond misfolding reaction to an off-pathway intermediate. Using permutations, we were able to discriminate between the roles of two proposed mechanisms of folding; a nucleation condensation model, and a hydrophobic collapse model driven by the formation of clusters of isoleucine, leucine, and valine (ILV) residues. We found that by altering the ILV cluster connectivity we could bias the early folding events to either favor on or off-pathway intermediates. Structural biases were also experimentally observed in the unfolded state of a de novo designed synthetic β/α-repeat protein, Di-III_14. Although thermodynamically and kinetically 2-state, Di-III_14 has a well structured unfolded state that is only observable under native-favoring conditions. This unfolded state appears to retain native-like structure, consisting of a hydrophobic 7 core (69% ILV) stabilized by solvent exposed polar groups and long range electrostatic interactions. Together, these results suggest that early folding events are largely deterministic in these two systems. Generally, low contact order ILV clusters favor local compaction and, in specific cases, long range electrostatic interactions may have stabilizing effects in higher energy states.

Astrocytes densely infiltrate the brain and intimately associate with synaptic structures. In the past 20 years, they have emerged as critical regulators of both synapse assembly and synapse function. During development, astrocytes modulate the formation of new synapses, and later, control refinement of synaptic connections in response to activity dependent cues. In a mature nervous system, astrocytes modulate synapse function through a variety of mechanisms. These include ion buffering, neurotransmitter uptake and the release of molecules that activate synaptic receptors. Through such roles, astrocytes shape the structure and function of neuronal circuits. However, how astrocytes and synapses reciprocally communicate during circuit assembly remains an unanswered question in the field. The vast majority of our understanding of astrocyte biology has come from studies conducted in mammals, where it is challenging to dissect molecular mechanisms with cell type specificity. Drosophila melanogaster is a less established model system for studying astrocyteneuron interactions, but its vast array of genetic tools and rapid life cycle promises great potential for precisely targeted manipulations. My thesis work has utilized Drosophila melanogaster to investigate the reciprocal nature of astrocyte-synapse communication. First, I characterized Drosophila late metamorphosis as a developmental stage in which astrocyte-synapse associations can be studied. My work demonstrates that during this time, when the adult Drosophila nervous system is being assembled, synapse formation relies on the coordinated infiltration of astrocyte membranes into the neuropil. Next, I show that in a reciprocal manner, neural activity can shape astrocyte biology during this time as well and impart long lasting effects on neuronal circuit function. In particular expression of the astrocyte GABA transporter (GAT) is modulated in an activity-dependent manner via astrocytic GABABR1/2 receptor signaling. Inhibiting astrocytic GABABR1/2 signaling strongly suppresses hyperexcitability in a Drosophila seizure model, vii arguing this pathway is important for modulating excitatory/inhibitory balance in vivo. Finally, utilizing the ease of the Drosophila system, I performed a reverse genetic screen to identify additional astrocyte factors involved in modulating excitatory-inhibitory neuronal balance.

A key process underlying synapse development and plasticity is stimulus-dependent translation of localized mRNAs. This process entails RNA packaging into translationally silent granules and exporting them over long distances from the nucleus to the synapse. Little is know about (a) where ribonucleoprotein (RNP) complexes are assembled, and if in the nucleus, how do they exit the nucleus; (b) how RNPs are transported to specific synaptic sites. At the Drosophila neuromuscular junction (NMJ), we uncovered a novel RNA export pathway for large RNP (megaRNP) granules assembled in the nucleus, which exit the nucleus by budding through the nuclear envelope. In this process, megaRNPs are enveloped by the inner nuclear membrane (INM), travel through the perinuclear space as membrane-bound granules, and are deenveloped at the outer nuclear membrane. We identified Torsin (an AAA-ATPase that in humans is linked to dystonia), as mediator of INM scission. In torsin mutants, megaRNPs accumulate within the perinuclear space, and the mRNAs fail to localize to postsynaptic sites leading to abnormal NMJ development. We also found that nuclear envelope budding is additionally used for RNP export during Drosophila oogenesis. Our studies also suggested that the nuclear envelope-associated protein, Nesprin1, forms striated F-actin-based filaments or ‘‘railroad tracks,’’ that span from muscle nuclei to postsynaptic sites at the NMJ. Nesprin1 railroad tracks wrap aoround the postsynaptic regions of immature synaptic boutons, and serve to direct RNPs to sites of new synaptic bouton formation. These studies elucidate novel cell biological mechanisms for nuclear RNP export and trafficking during synapse development.