Introduction: Civilian trauma remains a significant health care problem in North American society. Hemorrhagic shock and resuscitation (S/R) have been shown to prime the immune system for an exaggerated response to subsequent otherwise innocuous inflammatory stimuli such as lipopolysaccharide (LPS), resulting in multiple organ failure or death. Using a rodent model of lung injury, we previously demonstrated that antecedent S/R leads to augmented LPS-induced lung injury by way of heightened NF-κB nuclear translocation, resulting in increased elaboration of pro-inflammatory cytokines in alveolar macrophages. Further studies revealed that oxidative stress generated during S/R is responsible for this priming phenomenon. Our group recently identified two significant alterations to LPS signaling under oxidative stress conditions in macrophages: 1) the rapid recruitment of the LPS receptor Toll-like receptor 4 (TLR4) to membrane lipid rafts, and 2) the reprogramming of LPS signaling to a Src-dependent pathway involving phosphatidylinositol 3-kinase (PI3K). Major Objective and Hypothesis: The objective of this thesis is to elucidate the molecular mechanisms underlying the augmented cellular responsiveness observed in macrophages following oxidative stress. The central hypothesis is that oxidative stress regulates LPS signaling by altering the activation and assembly of TLR4 receptor signaling components through generation of the lipid ceramide. Summary of Findings: In the first paper, we demonstrate that the antioxidant stilbazulenyl nitrone (STAZN), a novel second-generation azulenyl nitrone, is protective in a rodent two-hit model of lung injury involving hemorrhagic S/R and subsequent intra-tracheal LPS injection. Resultant oxidative stress and lung injury in vivo were significantly reduced by STAZN following S/R and LPS. In the second paper, we explore the mechanism underlying oxidant-induced surface up-regulation of TLR4 in macrophages. Using immunofluorescence microscopy and flow cytometry techniques, hydrogen peroxide in vitro and hemorrhagic S/R in vivo are shown to induce TLR4 translocation in macrophages in a ceramide and Src-dependent manner, and the enzyme acid sphingomyelinase (ASM) is shown to mediate ceramide generation. In the third paper, the role of ceramide in oxidant-induced macrophage priming for LPS signaling is investigated. Ceramide generation via ASM is shown to have a prominent upstream role in oxidant activation of the PI3K/Akt pathway via Src kinases in macrophages. Furthermore, oxidative stress is shown to reprogram LPS signaling to a ceramide dependent pathway. Conclusion: Together, these findings highlight the role of oxidative stress in mediating augmented cellular responsiveness following S/R, and describe the role of ceramide as a central upstream mediator of oxidant priming in macrophages. The hierarchy of signaling molecules and interactions described herein represent novel targets for modulating oxidative stress in the treatment of critical illness and organ injury.

Cell migration and extension is essential for development. The ability of a cell or cell extension to reach its target is dependent on spatial cues and receptors that translate positional information into directed plasma membrane extension. For example, the UNC-40/DCC receptor is required to direct circumferential migrations towards the source of the ligand, UNC-6/Netrin, expressed at the ventral midline. To better understand the process of cell extension, I used a specialized process called muscle arms as a model system. In C. elegans, body wall muscles (BWMs) extend membrane projections called muscle arms to the nearest nerve cord at the midline. These muscle arms harbor the postsynaptic element of the neuromuscular junction and extend in a stereotypical and regulated manner. In a screen for muscle arm development defective (Madd) mutants, I isolated madd-2, a novel component of the UNC-40 pathway. MADD-2 is a C-1 TRIM protein and functions cell-autonomously to direct numerous muscle and axon extensions to the ventral midline of worms. In a striking analogy, mutations in a human homologue of MADD-2, MID1, cause numerous ventral midline defects that culminate as Opitz Syndrome. How MID1 regulates midline development is unclear. MADD-2 enhances UNC-40 pathway activity by facilitating the physical interaction between UNC-40 and the downstream Rho-GEF, UNC-73. It is possible that MID1 may mediate the function of a DCC-like pathway at the ventral midline of humans. This work provides the first indication that C1-TRIM proteins may have a conserved biological role of regulating midline-oriented development events and may provide key insights into the role of MID1 in the pathogenesis of Opitz syndrome.

Most human breast tumors have mutations in the growth factor/phosphatidylinositol 3’ kinase (PI3K) pathway. These can occur in genes coding for receptors, adaptor proteins, catalytic and regulatory subunits of PI3K, downstream kinases, or antagonistic tumor suppressors. While each genetic change results in elevated PI3K signaling, and all major breast cancer subtypes show pathway activation, the specific mutations involved in any one tumor may play an important role in defining tumor subtype, prognosis and sensitivity to therapy. Here, I describe mouse models of PI3K-induced breast cancer. First I generated mice that express Pik3ca cDNA under control of the ROSA26 promoter, in a Cre-dependent and therefore tissue specific way. I have generated four strains of knock-in mice: R26-Pik3cawt, R26-Pik3caE545K, R26-Pik3caH1047R, and R26-Pik3caE545K-H1047R, which can be induced to express wild type, helical domain, kinase domain and double mutant forms of mouse p110α, respectively. Mice expressing mutant Pi3kca develop mammary tumors, but the phenotypic spectrum for each mutation is unique. Indeed, many E545K mammary tumors are ii vascularized, whereas H1047R tumors are not. Using these models, I have compared downstream signaling properties of E545K and H1047R. The potential for improved breast cancer treatment lies in combination therapies that target more than one oncogenic pathway. To develop such treatments, we need good mouse models, and an understanding of the oncogenic network. To this end, my Pik3caH1047R model was mated to p53 and PTEN knockout mice, and to mice with active Notch1 signaling. In each case, genetic cooperation was observed and characterized. Oncogenic PI3K cooperated with p53-loss and active Notch1 to decrease survival and alter tumor phenotype in distinct ways. Loss of PTEN cooperated with oncogenic PI3K to alter tumor type, decrease average age at end point, and increase the number of tumors per mouse. Overall, I have shown that Pik3caE545K and Pik3caH1047R are sufficient to induce mammary tumors, and that tumor characteristics differ with these mutations, and with cooperating genetic changes.

The Hippo signaling pathway offers an intrinsic mechanism to control organ sizes, and dysfunction of this pathway can often lead to cancer. Great advancement has been made in recent years into understanding this pathway. Despite all this invaluable knowledge, much remains to be explored. Mass spectrometry offers an unbiased approach to characterize the interactome of any protein of interest and is particularly powerful for identifying potential novel regulators of signalling pathways. I therefore set out to characterize the interactome of all the Hippo pathway main components using mass spectrometry, with the goal of uncovering novel regulatory mechanism(s) of the Hippo pathway. In the end, I was able to identify over 250 novel interactors of the Hippo pathway in total. This study demonstrates the utility of mass spectrometry to identify novel regulators of the Hippo pathway and characterization of one such interactor.

The human saposins are four homologous activator proteins that are essential for the lysosomal degradation of sphingolipids (SLs) with small headgroups. They function in part to increase the solvent accessibility of these SLs to specific acid hydrolases by two proposed modes of action. In the solubilizer model, saposins extract SLs into soluble saposin-lipid complexes, and in the liftase model, the saposins reorganize lipids within membranes. The four saposins, called saposins A, B, C and D, have dual characters and exist as soluble proteins and in membrane-associated states. In this thesis, I first present the crystal structure of saposin D in a lipid-free state. The structure exhibits a closed, monomeric fold as previously described for saposins A and C. Next, I examine the lipid interaction properties of saposin A and determine a crystal structure of SapA in a lipid-bound complex. The complex forms a discoidal lipoprotein particle composed of highly ordered bilayer-like hydrophobic core surrounded by a protein belt consisting of two copies of saposin A in an open conformation. The saposin A-lipid discs are most likely the effective substrate-presenting particles in galactosylceramide hydrolysis. Finally, I compare the lipid interaction properties of the four saposins and address membrane perturbation, liposome binding, lipid solubilization and lipoprotein particle formation for each protein. Each saposin displays a unique behavior in the presence of liposomes under conditions that mimic the lysosomal environment. In particular, saposin D reveals simultaneous formation of different sized protein-lipid complexes, which appear to be primarily dependent on the lipid to protein molar ratio. A comparison of the available structures of saposins A, B and C in the “closed” and “open” conformations reveals structural hinge regions that likely shape the different types of saposin self-association. These states are directly related to the protein-lipid solubilization and/or membrane association properties of the saposins. Collectively, these findings present a more complete understanding of the lipid interaction properties of the saposin proteins and provide new insights into their role as activator proteins.

The human saposins are four homologous activator proteins that are essential for the lysosomal degradation of sphingolipids (SLs) with small headgroups. They function in part to increase the solvent accessibility of these SLs to specific acid hydrolases by two proposed modes of action. In the solubilizer model, saposins extract SLs into soluble saposin-lipid complexes, and in the liftase model, the saposins reorganize lipids within membranes. The four saposins, called saposins A, B, C and D, have dual characters and exist as soluble proteins and in membrane-associated states. In this thesis, I first present the crystal structure of saposin D in a lipid-free state. The structure exhibits a closed, monomeric fold as previously described for saposins A and C. Next, I examine the lipid interaction properties of saposin A and determine a crystal structure of SapA in a lipid-bound complex. The complex forms a discoidal lipoprotein particle composed of highly ordered bilayer-like hydrophobic core surrounded by a protein belt consisting of two copies of saposin A in an open conformation. The saposin A-lipid discs are most likely the effective substrate-presenting particles in galactosylceramide hydrolysis. Finally, I compare the lipid interaction properties of the four saposins and address membrane perturbation, liposome binding, lipid solubilization and lipoprotein particle formation for each protein. Each saposin displays a unique behavior in the presence of liposomes under conditions that mimic the lysosomal environment. In particular, saposin D reveals simultaneous formation of different sized protein-lipid complexes, which appear to be primarily dependent on the lipid to protein molar ratio. A comparison of the available structures of saposins A, B and C in the “closed” and “open” conformations reveals structural hinge regions that likely shape the different types of saposin self-association. These states are directly related to the protein-lipid solubilization and/or membrane association properties of the saposins. Collectively, these findings present a more complete understanding of the lipid interaction properties of the saposin proteins and provide new insights into their role as activator proteins.

Patched-1 (Ptch1) is the primary receptor for Hedgehog (Hh) ligands and mediates both canonical and non-canonical Hh signalling. Previously, our lab identified that mice possessing a Ptch1 C-terminal truncation display blocked mammary gland development at puberty that is overcome by overexpression of activated c-src. Testing the hypothesis that this involves a direct interaction between Ptch1 and c-src, we identified through co-immunoprecipitation that Ptch1 and c-src associate in an Hh-dependent manner, and that the Ptch1 C-terminus regulates activation of c-src in response to Hh ligand. Since the effects of Ptch1 intracellular domain deletions on canonical Hh signalling are ill-defined, we assayed this through luciferase reporter assays and qRT-PCR. Transient assays revealed that the Ptch1 middle intracellular loop is required for response to ligand, while qRT-PCR from primary cells showed that C-terminal truncation impairs canonical Ptch1 function. Together, this indicates that the intracellular domains of Ptch1 mediate distinct canonical and non-canonical functions.

Patched-1 (Ptch1) is the primary receptor for Hedgehog (Hh) ligands and mediates both canonical and non-canonical Hh signalling. Previously, our lab identified that mice possessing a Ptch1 C-terminal truncation display blocked mammary gland development at puberty that is overcome by overexpression of activated c-src. Testing the hypothesis that this involves a direct interaction between Ptch1 and c-src, we identified through co-immunoprecipitation that Ptch1 and c-src associate in an Hh-dependent manner, and that the Ptch1 C-terminus regulates activation of c-src in response to Hh ligand. Since the effects of Ptch1 intracellular domain deletions on canonical Hh signalling are ill-defined, we assayed this through luciferase reporter assays and qRT-PCR. Transient assays revealed that the Ptch1 middle intracellular loop is required for response to ligand, while qRT-PCR from primary cells showed that C-terminal truncation impairs canonical Ptch1 function. Together, this indicates that the intracellular domains of Ptch1 mediate distinct canonical and non-canonical functions.

The Hippo signaling pathway offers an intrinsic mechanism to control organ sizes, and dysfunction of this pathway can often lead to cancer. Great advancement has been made in recent years into understanding this pathway. Despite all this invaluable knowledge, much remains to be explored. Mass spectrometry offers an unbiased approach to characterize the interactome of any protein of interest and is particularly powerful for identifying potential novel regulators of signalling pathways. I therefore set out to characterize the interactome of all the Hippo pathway main components using mass spectrometry, with the goal of uncovering novel regulatory mechanism(s) of the Hippo pathway. In the end, I was able to identify over 250 novel interactors of the Hippo pathway in total. This study demonstrates the utility of mass spectrometry to identify novel regulators of the Hippo pathway and characterization of one such interactor.

Most human breast tumors have mutations in the growth factor/phosphatidylinositol 3’ kinase (PI3K) pathway. These can occur in genes coding for receptors, adaptor proteins, catalytic and regulatory subunits of PI3K, downstream kinases, or antagonistic tumor suppressors. While each genetic change results in elevated PI3K signaling, and all major breast cancer subtypes show pathway activation, the specific mutations involved in any one tumor may play an important role in defining tumor subtype, prognosis and sensitivity to therapy. Here, I describe mouse models of PI3K-induced breast cancer. First I generated mice that express Pik3ca cDNA under control of the ROSA26 promoter, in a Cre-dependent and therefore tissue specific way. I have generated four strains of knock-in mice: R26-Pik3cawt, R26-Pik3caE545K, R26-Pik3caH1047R, and R26-Pik3caE545K-H1047R, which can be induced to express wild type, helical domain, kinase domain and double mutant forms of mouse p110α, respectively. Mice expressing mutant Pi3kca develop mammary tumors, but the phenotypic spectrum for each mutation is unique. Indeed, many E545K mammary tumors are ii vascularized, whereas H1047R tumors are not. Using these models, I have compared downstream signaling properties of E545K and H1047R. The potential for improved breast cancer treatment lies in combination therapies that target more than one oncogenic pathway. To develop such treatments, we need good mouse models, and an understanding of the oncogenic network. To this end, my Pik3caH1047R model was mated to p53 and PTEN knockout mice, and to mice with active Notch1 signaling. In each case, genetic cooperation was observed and characterized. Oncogenic PI3K cooperated with p53-loss and active Notch1 to decrease survival and alter tumor phenotype in distinct ways. Loss of PTEN cooperated with oncogenic PI3K to alter tumor type, decrease average age at end point, and increase the number of tumors per mouse. Overall, I have shown that Pik3caE545K and Pik3caH1047R are sufficient to induce mammary tumors, and that tumor characteristics differ with these mutations, and with cooperating genetic changes.