Peter A. Rubenstein

All cells experience force. Tension, compression, and shear stress are all forces that can arise from normal cellular processes. Imbalances in sensing and responding to force can contribute to multiple diseases such as, cancer, diabetes, and cardiovascular disease. Identifying mechanisms by which cells sense and respond to force under normal conditions can deepen our understating of how these diseases progress and occur. External forces are sensed by cell surface adhesion receptors, such as the cadherins. In epithelia, E-cadherin is the principal adhesion receptor at intercellular junctions and mediates adhesion between neighboring cells. In response to force, E-cadherin triggers robust actin cytoskeletal reinforcement and growth of the adhesion complex. Collectively, known as cell stiffening or reinforcement. These changes allow cells to resist the force and are energetically costly. However, little is known on how energy generation and E-cadherin mediated force transmission are linked. In this thesis I investigate how the application of force on E-cadherin stimulates actomyosin contractility, glucose uptake, and ATP production. More specifically, I focus on determining the force-induced linkage between glucose uptake and the reinforcement of the actin cytoskeleton. I determine that force on E-cadherin localizes and enriches the glucose transporter, GLUT1. Localization of GLUT1 is required for force-induced glucose uptake, ATP production and reinforcement of actin and the cadherin-adhesion complex. Additionally, I identify a novel linkage between GLUT1, E-cadherin, and an adaptor protein, ankyrin G. Ankyrin G is known to meditate the retention of several membrane-bound proteins and bind to E-cadherin directly. I demonstrate that binding of Ankyrin G to E-cadherin is required for the localization and enrichment of GLUT1 to the cadherin-containing junctions in response to force. Furthermore, the reinforcement of actin and the adhesions are dependent on ankyrin G binding to E-cadherin. The work in this thesis provides three significant conceptual advancements: (1) a mechanism by which force induces glucose uptake, (2) the identification of a force-activated glucose transporter, (3) a novel link between cytoskeletal reinforcement and nutrient uptake.

The voltage-gated sodium channel 1.2 (NaV1.2) is a transmembrane protein primarily expressed in the central nervous system where it is responsible for generating the rising phase of an action potential. To do this, NaV1.2 undergoes voltage-dependent conformational changes that transition the pore of the channel from a closed to an open state, which allows Na+ to rapidly enter the neuron. Milliseconds after opening, the flow of Na+ through NaV1.2 is halted through a process known as inactivation. The ability of NaV1.2 to rapidly transition among closed, open and inactivated states is critical to its physiological function. Mutations found throughout the protein sequence of NaV1.2 that disrupt this process have been shown to cause debilitating disorders such as epilepsy and autism. Cycling of NaV1.2 among its functional states is tightly regulated by a network of interactions. These include intramolecular interactions between the cytosolic domains of the channel and intermolecular interactions between the cytosolic N- (NTD) and C-terminal domains (CTD) of the channel and several auxiliary proteins. Among these auxiliary proteins are fibroblast growth factor homologous factor 12 (FGF12) and the ubiquitously expressed Ca2+-sensor calmodulin (CaM). FGF12 and CaM modulate NaV1.2 function through binding to the cytosolic CTD of the channel, with FGF12 binding to an EF-hand like domain (EFL), and CaM to an IQ motif (IQxxx[R,K]Gxxx[R,K]). However, how these two auxiliary proteins regulate NaV1.2 function is poorly understood. My thesis has focused on understanding the thermodynamic and structural mechanism through which CaM functions as a Ca2+ sensor when bound to NaV1.2. I have used spectroscopic methods to investigate the interaction between Ca2+-depleted (apo) and Ca2+-saturated CaM and the human NaV1.2 IQ motif as a peptide (NaV1.2IQp) and in a larger fragment of the NaV1.2 CTD containing both the EFL and IQ motif (NaV1.2CTD), and examined how pathogenic mutations in CaM perturb its interaction with the NaV1.2 IQ motif. My solution structure of apo CaM bound to the NaV1.2IQp and other NMR studies of apo and Ca2+-saturated CaM bound to NaV1.2IQp and NaV1.2CTD have demonstrated that both endstates of CaM bind the NaV1.2IQp and NaV1.2CTD exclusively through the C-domain of CaM (CaMC), with the N-domain of CaM (CaMN) making no persistent contacts. Binding of NaV1.2IQp had markedly different effects on the residues in CaMC of apo and Ca2+-saturated CaM, and were consistent with the binding of Ca2+ inducing rotation of full-length CaM on both the NaV1.2IQp and NaV1.2CTD. This suggests that as the local Ca2+ concentration rises, CaMC may transiently release and re-associate with the NaV1.2 IQ motif. Pathogenic CaM mutants bound the NaV1.2 IQ motif as well as or with greater affinity than WT CaM under apo conditions. However, mutations in CaMC inhibited or ablated Ca2+ binding by CaMC when bound to the NaV1.2 IQ motif. Thus, CaM harboring these mutations would not be expected to support NaV1.2 regulation requiring rotation or transient release of CaM from the IQ motif. CaM and FGF12 can interact simultaneously with the NaV1.2 CTD, placing these two auxiliary proteins near one another in the context of the channel. However, prior to this thesis no direct interaction had been reported between CaM and any FGF12 splice variant. My studies have discovered that both domains of CaM bind to a novel CaMBD in the NTD of the A-type splice variant of FGF12 (FGF12A CaMBD). CaM was found to bind the FGF12A CaMBD with high affinity in a Ca2+-dependent manner. Binding of full-length FGF12A to the NaV1.2CTD was found to increase the Ca2+ affinity of both CaM domains, consistent with Ca2+-saturated CaM interacting favorably with the NTD of full-length FGF12A when bound to NaV1.2CTD. This suggests that at a high cytosolic Ca2+ concentration, CaM may translocate from the NaV1.2 IQ motif to the FGF12A CaMBD to modulate NaV1.2 function. This provides new insights into the structural mechanism by which Ca2+, CaM, and FGF12A may act as regulators of NaV1.2.

The Mitochondrial Pyruvate Carrier (MPC) occupies a central metabolic node by transporting cytosolic pyruvate into the mitochondrial matrix, thereby linking glycolysis with mitochondrial metabolism. Two reported human MPC1 mutations cause developmental abnormalities, neurological problems, metabolic deficits, and for one patient, early death. We aimed to understand biochemical mechanisms by which the human patient c.C289T and c.T236A MPC1 alleles disrupt MPC function. MPC1 c.C289T encodes two protein variants, a mis-spliced, truncation mutant (A58G) and full-length point mutant (R97W). MPC1 c.T236A encodes a full-length point mutant (L79H). Using human patient fibroblasts and complementation of CRISPR-deleted, MPC1 null mouse C2C12 cells, we investigated how MPC1 mutations cause MPC deficiency. Truncated MPC1 A58G protein was intrinsically unstable and failed to form MPC complexes. The MPC1 R97W protein was less stable but when overexpressed formed complexes with MPC2 that retained pyruvate transport activity. Conversely, MPC1 L79H protein formed stable complexes with MPC2, but these complexes failed to transport pyruvate. These findings inform MPC structure-function relationships and delineate three distinct biochemical pathologies resulting from human patient MPC1 mutations and inform fundamental MPC structure-function relationships. These results also demonstrate an efficient molecular genetic system using the mouse C2C12 cell line to mechanistically investigate human inborn errors in pyruvate metabolism.

Metabolic cycles are a fundamental element of cellular and organismal function. Among the most critical in higher organisms is the Cori Cycle, the systemic cycling between lactate and glucose. Here, skeletal muscle-specific Mitochondrial Pyruvate Carrier (MPC) deletion in mice increased muscle glucose uptake but diverted pyruvate into the circulation as lactate, driving increased Cori Cycling and energy expenditure. Loss of muscle MPC activity evoked adaptive glutaminolysis, increased fatty acid oxidation, and resulted in a striking resistance to gains in fat mass with age with perfect sparing of muscle mass and strength. Furthermore, chronic and acute muscle MPC deletion accelerated fat mass loss on a normal diet after high fat diet-induced obesity. Our results illuminate the role of the skeletal muscle MPC as a central node for whole-body carbohydrate, fat, and amino acid metabolism. They highlight the potential utility of decreasing muscle pyruvate oxidation to ameliorate obesity and type 2 diabetes.

Prostaglandins (PGs) are small, lipid signaling molecules produced downstream of cyclooxygenase (COX) enzymes. PG signaling regulates many processes including pain, inflammation, fertility, cardiovascular function and disease, and cancer. One mechanism by which PG signaling exerts its function is by regulating the dynamics of the actin cytoskeleton; however, the exact mechanisms remain largely undefined.Drosophila oogenesis provides an ideal system to determine how PG signaling regulates the actin cytoskeleton. Drosophila follicles, or eggs, pass through 14 well- characterized, morphologically defined stages of development. Each developing follicle is comprised of 16 interconnected germline-derived cells (15 nurse cells and 1 oocyte) that are surrounded by a layer of somatically derived epithelial cells. During Stage 10B (S10B), the nurse cells form a cage-like network of parallel actin filament bundles that extend from the nurse cell membranes inward, toward the nurse cell nuclei. During Stage 11 (S11), the nurse cells rapidly transfer their cytoplasmic contents into the oocyte in an actomysoin-dependent contraction termed nurse cell dumping. Previous work uncovered that the Drosophila COX-like enzyme, Peroxinectin-like (Pxt), and thus PG signaling, is required to promote both actin filament bundle formation during S10B and subsequent nurse cell dumping. This finding established Drosophila oogenesis as a genetically tractable model in which to elucidate the conserved mechanisms underlying PG- dependent actin remodeling.The research presented in this dissertation focused on identifying actin-binding proteins that are regulated by PG signaling during Drosophila oogenesis. To identify these downstream effectors, we performed a dominant modifier screen to uncover factors that could suppress or enhance the ability of COX inhibitors to block nurse cell dumping in vitro. This screen revealed a number of actin-binding proteins that enhance the dumping defects caused by COX-inhibition, including the actin bundling protein, Fascin (Drosophila Singed, Sn); the actin filament elongation factor, Enabled (Ena); and the actin filament capper, Capping protein (Drosophila Capping protein alpha, Cpa, and beta, Cpb). Through a collaborative effort between Christopher Groen and myself, Fascin was shown to mediate PG-dependent cortical actin integrity and actin bundle formation during Drosophila ooogenesis.Ena and Capping protein regulate actin filament elongation through opposing actions: Ena promotes their elongation, while Capping protein binds to, or caps, the growing end of actin filaments to prevent their further elongation. However, genetic reduction of either Ena or Capping protein enhance the nurse cell dumping defects caused by COX inhibition. These findings suggest that Ena activity must be balanced to promote proper actin remodeling during S10B. Ena localization to the growing ends of actin filament bundles is reduced in pxt mutants during S10B, suggesting that PG signaling is required to promote Ena localization at this stage. Together, these data support a model in which PG signaling promotes actin remodeling during S10B, at least in part, by modulating Ena-dependent actin remodeling.While PG signaling promotes parallel actin filament bundle formation during S10B, PGs also restrict actin remodeling during Stage 9 (S9). Loss of Pxt results in early actin remodeling, including the formation of extensive actin filaments and actin aggregate structures within the posterior nurse cells of S9 follicles. Wild-type follicles exhibit similar structures at a low frequency. Ena preferentially localizes to the early actin structures observed in pxt mutants and reduced Ena levels strongly suppress early actin remodeling in pxt mutants. These data indicate that PG signaling temporally restricts actin remodeling during Drosophila oogenesis, at least in part, through negative regulation of Ena localization or activity during S9.The data presented here support a model in which PG signaling coordinates the concerted activity of a number of actin-binding proteins to regulate actin remodeling during Drosophila oogenesis. Specifically, PG signaling temporally restricts actin remodeling during S9 of Drosophila oogenesis, but promotes parallel actin filament bundle formation during S10B. PG signaling achieves this temporal regulation, at least in part, through differential regulation of Ena-dependent actin remodeling. Based on prior pharmacologic studies, we hypothesize that PGE2 is required to restrict Ena-dependent actin remodeling during S9, while PGF2Α; is required to promote Ena-dependent actin remodeling during S10B. Determining how these signaling cascades achieve differential regulation of Ena throughout Drosophila oogenesis is an important area for future investigation. As both the actin cytoskeletal machinery and PG signaling are conserved across species, the data presented here provide new and significant insights into the likely conserved mechanisms by which PG signaling regulates actin remodeling across species.

Photoreceptors are highly compartmentalized neurons in the retina, and they function by detecting light and initiating signaling through the visual network. The photoreceptor contains several compartments including the outer segment (OS) which is a sensory cilium for detecting photons and the inner segment (IS) that carries out important modulatory functions via its resident channels and transporters. Those proteins are membrane proteins that function together to shape electrical properties of the cell membrane during both rest and active states. Therefore it is essential to maintain proper function of the membrane proteins in the IS. One important way to regulate the function of a membrane protein is via controlling its trafficking to ensure a proper amount of the protein in the proper cellular compartment. To date, little is known about how IS membrane protein trafficking is controlled in photoreceptors. In this study, our goal is to understand those mechanisms using cell biology and biochemistry approaches. To achieve the goal, we investigated trafficking of two unrelated IS resident proteins: the hyperpolarization-activated cyclic nucleotide-gated channel 1 (HCN1) that mediates a feedback current in photoreceptors, and the sodium potassium ATPase (NKA) which maintains the basic electrochemical property of the cell. In order to study trafficking of HCN1, we first investigated the dependence of HCN1 trafficking in photoreceptors on TRIP8b, an accessory subunit that influences trafficking of HCN1 in hippocampal neurons. By studying TRIP8b knockout mice we found that TRIP8b is dispensable for HCN1 trafficking in photoreceptors but required for maintaining the maximal expression level of HCN1. Since we revealed that HCN1 trafficking can be regulated in a cell-type specific manner, we subsequently focused on the amino acid sequence of HCN1 to identify novel trafficking signals that function in photoreceptors. By examining localization of a series of HCN1 mutants in transgenic Xenopus photoreceptors, we discovered a di-arginine ER retention motif and a leucine-based ER export motif. These two sequence motifs must function together to maintain equilibrium of HCN1 level between the endomembrane system and the cell surface. The study of HCN1 uncovered a mechanism for the photoreceptor to control membrane protein trafficking via the early secretory pathways. To reveal additional trafficking machineries in photoreceptors, we investigated trafficking of NKA. We first tested for an interaction with ankyrin, an adaptor protein that regulates NKA trafficking in epithelial cells, and found these proteins do not co-localize in photoreceptors. We then aimed to identify novel trafficking signals by studying the trafficking behavior of two NKA isozymes: NKA-α 3 and NKA-α 4. When expressed in transgenic Xenopus photoreceptors, these two proteins localize to the IS and the OS respectively. By studying localization of multiple chimeras and truncation mutants, we found that the distinct localization pattern is due to a VxP OS/ciliary targeting motif present in NKA-α 4. Since NKA-α 4 is naturally expressed in the ciliary compartment of the sperm, our finding in the photoreceptor suggests a mechanism for NKA-alpha 4 trafficking in its native environment. Overall, our studies of HCN1 and NKA together provide new insights into controlling membrane protein trafficking in photoreceptors and help establish the basics for future therapeutic intervention targeting trafficking pathways that are linked to about one third of proteins reported in retinal diseases.

Defects in BRCA1 and BRCA2 tumor suppressors predispose one to breast and ovarian cancer. The current treatment for BRCA-deficient cancers is mastectomy. Because both copies of the tumor suppressor need to be defective for cancer to occur, identifying cellular mechanisms that specifically target BRCA-deficient cells is of paramount importance. Luckily, recent experiments have shown that depletion of a protein named RAD52 in BRCA1 or BRCA2 cancer cells causes them to die. Therefore, we can use small molecules to stop the RAD52 protein from functioning. We need, however, to know which of the RAD52 activities to inhibit and how. One function of RAD52 that likely underlies all cellular activities is its ability to bind single-stranded DNA (ssDNA). To identify if small molecules could inhibit the RAD52-ssDNA complex, I screened a small library of compounds and found 13 potential inhibitors. We validated that these small molecules bind to RAD52 and inhibit RAD52 DNA binding and annealing activities. The identification of these small molecules is important because we can use them to dissect the function of RAD52 in normal and malignant cells, which to date remains elusive. In an attempt to further advance our understanding of RAD52 function and regulation we are also investigating how a novel binding partner, DSS1, interacts with RAD52 and modulates its activities. My data show that this protein enhances the way RAD52 finds separate complementary DNA templates and anneals them to make a double-stranded product. At least in part, these studies have identified some residues likely involved in the binding site of DSS1 on RAD52. In aggregate, the outcome of the two projects deepens our understanding of the complex and interconnected cellular pathways that support the integrity of genomes.

The dynamic turnover of the actin cytoskeleton is regulated cooperatively by force and biochemical signaling. We previously demonstrated that actin depolymerization under force is governed by catch-slip bonds mediated by force-induced K113:E195 salt-bridges. Yet, the biochemical regulation as well as the functional significance of actin catch bonds has not been elucidated. Using AFM force-clamp experiments, we show that formin controlled by RhoA switches the actin catch-slip bonds to slip-only bonds. SMD simulations reveal that the force does not induce the K113:E195 interaction when formin binds to actin K118 and E117 residues located at the helical segment extending to K113. Actin catch-slip bonds are suppressed by single residue replacements K113E and E195K that interrupt the force-induced K113:E195 interaction; and this suppression is rescued by a K113E/E195K double mutant (E/K) restoring the interaction in the opposite orientation. These results support the biological significance of actin catch bonds, as they corroborate reported observations that RhoA and formin switch force-induced actin cytoskeleton alignment and that either K113E or E195K induces yeast cell growth defects rescued by E/K. Our study demonstrates how the mechano-regulation of actin dynamics is modulated by biochemical signaling molecules, and suggests that actin catch bonds may be important in cell functions.