Regulation of Mtorc1-Dependent Growth Signaling by Lysosomal Cholesterol by Ashley M

Home , Nutrient sensing

Regulation of mTORC1-Dependent Growth

Signaling by Lysosomal Cholesterol

Ashley M. Thelen

A dissertation submitted in partial satisfaction

of the requirements for the degree of

Doctor of Philosophy

Molecular and Cell Biology

in the

Graduate Division

of the

University of California, Berkeley

Committee in charge:

Professor Roberto Zoncu, Chair

Professor James Hurley

Professor Jeremy Thorner

Professor Daniel Nomura

Spring 2019

ABSTRACT

Regulation of mTORC1-Dependent Growth Signaling by Lysosomal Cholesterol by Ashley M. Thelen Doctor of Philosophy in Molecular and Cell Biology University of California, Berkeley Professor Roberto Zoncu, Chair

Mechanistic target of rapamycin complex 1 (mTORC1) is an evolutionarily conserved serine/threonine protein kinase that acts as a master regulator of growth and proliferation in eukaryotic cells. mTORC1 integrates a variety of signals including cellular stress, nutrients and growth factors, and, in response, phosphorylates key substrates involved in coordinating cellular metabolism. When nutrients are abundant, mTORC1 fuels the biosynthesis of cellular building blocks including nucleotides and lipids, and enhances mRNA translation and ribosomal biogenesis. In addition, mTORC1 inhibits catabolic pathways such as autophagy, which consume cell mass. A heterodimer of small guanosine triphosphatases, known as the Rag GTPases, communicates cellular nutrient status to mTORC1, such that the active nucleotide state (RagAGTP/RagCGDP) promotes translocation of mTORC1 from the cytosol to the surface of the lysosome. At the lysosome, another small GTPase, Rheb, binds to mTORC1 and unlocks the autoinhibited kinase domain in response to growth factor signals. Thus, growth regulation occurs via an elegant coincidence detection mechanism, whereby intracellular nutrients direct mTORC1 to a physical location, and growth factors provide long-range signals from distant tissues to activate mTORC1 kinase via Rheb. Despite major advances in our understanding of mTORC1 regulation, it remains unclear how mTORC1 senses and integrates chemically diverse molecules. Here, I identify cholesterol, an essential lipid for cellular growth, as a nutrient input that drives mTORC1 recruitment and activation at the lysosomal surface. I found that the lysosomal transmembrane protein SLC38A9 is required for mTORC1 activation by cholesterol through conserved cholesterol-responsive motifs. Moreover, SLC38A9 enables cholesterol-mediated mTORC1 activation independently from its arginine-sensing function. Conversely, the Niemann-Pick C1 (NPC1) protein, which regulates cholesterol export from the lysosome, binds to SLC38A9 and inhibits mTORC1 signaling through its sterol transport function. Thus, lysosomal cholesterol drives mTORC1 activation and growth signaling through the SLC38A9-NPC1 complex. Importantly, this suggests that aberrant mTORC1 activity may contribute to the metabolic derangement observed in Niemann-Pick Type C (NPC) disease in humans. The mechanism of mTORC1 recruitment by the lysosomal scaffolding complex is not fully understood. To address this, I purified the full-length human Ragulator:RagA/C complex in the active nucleotide loading state and, in collaboration with the Hurley Lab, determined the cryo-EM structure. This is the first structure to reveal the nucleotide binding G domains of the Rag GTPases in the context of their lysosomal scaffold, Ragulator. Overall, this work elucidates a new nutrient input to the mTORC1 signaling network and provides structural insights into the regulation of mTORC1 docking at the lysosome, suggesting new strategies for therapeutically targeting metabolic diseases.

ACKNOWLEDGEMENTS

The completion of this dissertation and the experiments within would not have been possible without the generous contributions from so many people throughout my time in graduate school. The vibrant and collaborative environment at Berkeley made it a truly wonderful place to do science. To the entire MCB research community, including faculty, staff, and students, thank you. Two people stand out as particularly influential. First, Roberto, who took me into his lab as a naïve first year and devoted countless hours towards my scientific training and development. He taught me to approach research with a razor-sharp focus and intensity, which helped me push my own limits farther than I knew possible. I admire his dedication to maintaining lab productivity and his eagerness for generating new hypotheses. Second, I will forever be grateful for Brian Castellano, the co-founder of the Zoncu Lab who quickly became my co-conspirator and eventually, co-author. His outgoing and fun-loving personality made the lab a brighter place. I can’t imagine enduring the roller coaster of graduate school without Brian’s friendship, advice, and relentless practical jokes. There is no one else I would rather smuggle a free futon into an emergency spill closet with. I would like to thank my thesis committee, Professors Jim Hurley, Jeremy Thorner, and Dan Nomura for their scientific insights and constructive feedback. Jim has been especially kind and supportive of my research endeavors, offering endless ideas and opportunities for collaborative projects. Jeremy’s wisdom, mentorship, and tremendous enthusiasm for academic research were invaluable during critical moments of my training. I would also like to acknowledge Professors David Raulet and Michel DuPage, both of whom provided an abundance of knowledge and inspiration and strongly influenced my decision to pursue future research in the field of cancer immunotherapy. Furthermore, I was fortunate to have incredible role models and mentors early on in my scientific career, including my high school biology teacher, Julie Olson and my undergraduate research advisors, Professors Joe Barycki and Melanie Simpson. I can’t thank them enough for the lasting impact they have made on my life by sharing their passion for science. I am grateful for all of the past and present members of the Zoncu Lab for maintaining a collegial and intellectually stimulating lab environment. In particular, the graduate students, Brian, Rosalie, Oliver, Justin, and JP were a vital source of support and encouragement. Thank you also to Adam Yokom, a postdoc in the Hurley Lab, and the best collaborator I could’ve asked for. He introduced me to the wonders of electron microscopy and Jolly Pumpkin beer. Last but not least, I feel incredibly fortunate for friends and family who have provided endless love and support over the years. I am particularly appreciative that my parents and siblings didn’t disown me for moving halfway across the country. Thank you to Emily who has kept my life in line as the unofficial “house mom” of the Fulton House and to Natalie who is my favorite friend to talk science with, even after a long day of science. Finally, I am immensely grateful for Kevin, whose contribution to my ability to find balance in life and joy outside of the lab has been immeasurable.

DEDICATION

To my parents, Corey and Lisa Thelen

Mom, you taught me the importance of perseverance and the value of education. Although “bedtime stories” from your Lehninger textbook did not excite me as a child, I’ve finally come around to appreciating the content.

Dad, you taught me something that cannot be found in a book, the ability to see the world though a lens of endless curiosity. You challenged me to think outside the box and to learn by experimenting when the answer to a question was unknown.

TABLE OF CONTENTS

Abstract ...... 1 Acknowledgements ...... i Dedication ...... ii Table of Contents ...... iii List of Figures ...... v

Chapter 1: Introduction to Nutrient Sensing and mTORC1 Signaling 1.1 Overview of mTORC1 signaling ...... 2 1.1.1 Coordination of cellular metabolism by mTORC1 ...... 2 1.1.2 The lysosome as a hub for communicating nutrient status ...... 4 1.1.3 Regulation of mTORC1 kinase activation at the lysosome ...... 5 1.1.4 Mechanisms of nutrient sensing upstream of mTORC1 ...... 6 1.2 Emerging roles for the lysosome in lipid metabolism ...... 7 1.2.1 Lipid homeostasis is key for the function of cells, tissues, and organisms ...... 8 1.2.2 The lysosome as a trafficking station for lipids ...... 9 1.2.3 Mechanisms of lipid transport within the lysosome ...... 10 1.2.4 Regulation of lipid export from the lysosome ...... 12 1.2.5 Feedback inhibition of lipid biogenesis ...... 12 1.2.6 Control of lipid metabolism by mTORC1 ...... 13

Chapter 2: Lysosomal Cholesterol Activates mTORC1 via an SLC38A9-Niemann-Pick C1 Signaling Complex 2.1 Chapter Summary ...... 18 2.2 Introduction ...... 18 2.3 Methods ...... 19 2.3.1 Cholesterol starvation and restimulation in cells ...... 19 2.3.2 Amino acid starvation and restimulation in cells ...... 19 2.3.3 Double starvation of cholesterol and either arginine or leucine ...... 19 2.3.4 mTORC1 kinase assay ...... 19 2.3.5 Immunofluorescence assays ...... 20 2.3.6 Co-immunoprecipitation from cell lysates ...... 20 2.3.7 Lysosome immunoprecipitation for mass spectrometric analysis ...... 20 2.3.8 Lysosomal cholesterol measurements using mass spectrometry ...... 21 2.3.9 Intracellular amino acid measurements ...... 21 2.3.10 siRNA knockdown experiments ...... 21 2.3.11 SLC38A9 peptide cholesterol binding assays ...... 21 2.3.12 SLC38A9 labeling with photoactivatible cholesterol probe ...... 22

2.4 Results ...... 22 2.4.1 Cholesterol regulates mTORC1 activity ...... 22 2.4.2 Cholesterol specifically activates mTORC1 in a pathway distinct from growth factor signaling ...... 24

iii

2.4.3 Cholesterol regulates lysosomal recruitment of mTORC1 via the Ragulator-Rag GTPase complex ...... 25 2.4.4 SLC38A9 mediates mTORC1 activation in response to lysosomal cholesterol ...... 27 2.4.5 Niemann-Pick C1 protein interacts with SLC38A9 and the Ragulator-Rag GTPase complex ...... 33 2.4.6 Cholesterol and arginine signals are integrated via SLC38A9 to coordinately regulate mTORC1 activity ...... 35 2.5 Discussion ...... 37 2.5.1 Role for SLC38A9 and NPC1 in cholesterol-mediated activation of mTORC1 ...... 37 2.5.2 Regulatory feedback loop of mTORC1 activity by other lipid species ...... 38 2.5.3 Disruption of lysosomal lipid homeostasis in disease and aging ...... 39 2.5.4 Outstanding questions and future directions ...... 41

Chapter 3: Structural Characterization of the mTORC1 Lysosomal Scaffolding Complex 3.1 Chapter Summary and Introduction ...... 43 3.2 Methods ...... 44 3.2.1 Expression and purification of Ragulator from Sf9 cells ...... 44 3.2.2 Expression and purification of the RagA/C heterodimer rom Sf9 cells ...... 44 3.2.3 Nucleotide loading of the RagA/C heterodimer ...... 45 3.2.4 Gel filtration of the Ragulator:RagA/CT90N complex ...... 45 3.3 Results ...... 45 3.3.1 Reconstitution of the active Ragulator:RagA/CT90N complex ...... 45 3.3.2 Structural characterization of the active Ragulator:RagA/CT90N complex ... 47 3.4 Discussion ...... 49

References ...... 51

LIST OF FIGURES

Figure 1.1 The mTORC1 signaling network ...... 3 Figure 1.2 Mechanisms of nutrient mediated mTORC1 activation at the lysosome ...... 7 Figure 1.3 Cellular routes for cholesterol trafficking ...... 11 Figure 1.4 mTORC1 activation promotes lipid biogenesis and suppresses lipid catabolism ...... 15 Figure 2.1 Cholesterol regulates mTORC1 activity ...... 23 Figure 2.2 Cholesterol specifically activates mTORC1 in a pathway distinct from growth factor signaling ...... 25 Figure 2.3 Cholesterol regulates lysosomal recruitment of mTORC1 via the Ragulator- Rag GTPase complex ...... 26 Figure 2.4 SLC38A9 is required for mTORC1 activation by lysosomal cholesterol ...... 29 Figure 2.5 SLC38A9 binds LKM-38, a photoactivatable cholesterol probe ...... 31 Figure 2.6 Mutations in the putative cholesterol responsive motifs of SLC38A9 diminish mTORC1 activation and lysosomal translocation in response to cholesterol ...... 32 Figure 2.7 Niemann-Pick C1 protein interacts with SLC38A9 and the Ragulator-Rag GTPase signaling complex ...... 34 Figure 2.8 Coordination of arginine and cholesterol signaling by SLC38A9 ...... 36 Figure 2.9 Model for mTORC1 regulation by LDL-derived cholesterol ...... 38 Figure 3.1 Reconstitution of the active Ragulator:RagA/CT90N complex ...... 46 Figure 3.2 Negative stain EM of the active Ragulator:RagA/CT90N complex ...... 47 Figure 3.3 Low-resolution cryo-EM structure of the active Ragulator:RagA/CT90N complex ...... 48

Chapter 1

Introduction to Nutrient Sensing and mTORC1 Signaling

A portion of the content presented in this chapter has been previously published as part of the following review article: Thelen, A.M. and Zoncu, R. (2017) Emerging Roles for the Lysosome in Lipid Metabolism. Trends in Cell Biol. 27, 833-850.

1 1.1 Overview of mTORC1 signaling

1.1.1 Coordination of cellular metabolism by mTORC1

Cells import nutrients from the extracellular space in the form of small molecules including sugars, lipids, and amino acids, which serve as precursors for the synthesis of macromolecules that make up the cell. Metabolism, the sum of biochemical reactions in the cell, can be divided into anabolic reactions, in which energy in the form of ATP is used to build macromolecules, and catabolic reactions, in which macromolecules are broken down into their constituent parts. Precise regulation of metabolism is required for cellular fitness, as metabolic dysfunction is a hallmark of several diseases including cancer, diabetes, and neurodegeneration. A major breakthrough in our understanding of how metabolism is coordinated at the cellular level occurred in the early 1990’s upon the discovery of a serine/threonine protein kinase called mechanistic target of rapamycin (mTOR). mTOR was initially discovered in yeast via genetic screens for resistance to the immunosuppressive drug rapamycin (Heitman et al. 1991) and the human homolog was identified soon thereafter (Sabatini et al. 1994, Brown et al. 1994). mTOR is evolutionarily conserved amongst eukaryotes and functions as the catalytic subunit of two functionally distinct protein kinase complexes, mTOR complex 1 (mTORC1) and mTORC2 (reviewed in Saxton and Sabatini 2017). mTORC1 is composed of three core subunits: mTOR, mLST8 and the mTORC1-specific component, Raptor (Hara et al. 2002, Kim et al. 2002, Kim et al. 2003). Raptor facilitates substrate recruitment by binding to TOR signaling (TOS) motifs found in many mTORC1 substrates (Nojima et al. 2003). In addition to the three core components, mTORC1 also associates with two inhibitory proteins, PRAS40 and DEPTOR, which bind to Raptor (Sancak et al. 2007, Vander Haar et al. 2007, Peterson et al. 2009). Early structural studies revealed that mTORC1 forms an obligate dimer, corresponding to a roughly 1 mega Dalton complex (Yip et al. 2010, Baretic et al. 2016). Nearly three decades of research has uncovered a critical role for mTORC1 in orchestrating cell growth and proliferation in response to nutrients and growth factors (reviewed in Saxton and Sabatini 2017). mTORC1 acts as a signaling node, integrating diverse inputs and coordinating downstream processes, thereby enabling fine-tuning of the cellular metabolic state (reviewed in Valvezan and Manning 2019). In particular, mTORC1 promotes anabolic pathways including synthesis of proteins, lipids, and nucleic acids, and inhibits catabolic pathways such as autophagy, the process of cellular recycling via “self-eating”. Upstream inputs of mTORC1 can be grouped into three categories: cellular stress, nutrients, and growth factors (Fig 1.1). Cellular stress signals in the form of hypoxia, DNA damage, and low ATP levels prevent cell growth by inhibiting mTORC1 (reviewed in Saxton and Sabatini 2017). Nutrients are detected by a complex network of proteins that directly bind small molecules and transmit information via two small guanosine triphosphatases (GTPases), the Rag GTPases (discussed in detail in later sections). The Rag GTPases control the subcellular localization of mTORC1, enabling its binding to Rheb, a small GTPase that is required for the activation of mTORC1 kinase activity (Long et al. 2005, Tee et al. 2003). The nucleotide state of Rheb is independently regulated by the presence of growth factors and hormones such as insulin, which

Figure 1.1 The mTORC1 signaling network mTORC1 kinase is activated at the surface of the lysosome in the presence of both nutrients and growth factors, and inhibited by cellular stress (including DNA damage, hypoxia, and low energy levels). Phosphorylation of substrates by mTORC1 can result in their activation (orange) or inhibition (grey). The functional outputs of mTORC1 signaling are listed, with processes that are activated shown in green, and processes that are inhibited shown in red. promote GTP-bound, active Rheb. Specifically, binding of insulin to the insulin receptor activates the serine/threonine kinase Akt. In turn, Akt phosphorylates and inhibits Tuberous Sclerosis Complex (TSC), composed of TSC1, TSC2, and TBC1D7, which functions as a negative regulator of Rheb via its GTPase activating protein (GAP) activity (Inoki et al. 2003, Tee et al. 2003). This is an elegant example of coincidence detection, in which the intracellular availability of nutrients directs mTORC1 to a physical location, and long-range signals from distant tissues (i.e. growth factors and insulin) trigger mTORC1 kinase function through Rheb. The nutrient sensing and growth factor signaling pathways act in parallel, and inactivation of either arm blocks the kinase activity of mTORC1 (otherwise known as a molecular “AND-gate”). The complexity of this pathway underlies the importance of metabolic coordination both within cells and across tissues in the evolution of multicellular organisms. Following its activation by the combined action of nutrients and growth factors, mTORC1 phosphorylates a number of proteins that either promote biosynthetic

3 processes or inhibit catabolic ones. The canonical substrates are described below, although this is not an exhaustive list, as the number of mTORC1 substrates is continuously expanding. mTORC1 phosphorylates and activates S6-kinase 1 (S6K1), which drives numerous pro-growth processes including ribosome biogenesis, along with lipid and nucleotide biosynthesis (Ben-Sahra et al. 2013, Ben-Sahra et al. 2016, Duvel et al. 2010). mTORC1 phosphorylates and inhibits eukaryotic translation initiation factor 4E-binding proteins (4E-BP) 1 and 2. 4E-BPs block mRNA translation initiation by binding to eIF4E, a component of a multi-subunit complex that recruits 40S ribosomal subunits to the 5' end of mRNAs; thus, their phosphorylation by mTORC1 triggers translation of a class of mRNAs that contain a 5’ terminal oligopyrimidine (TOP) motif (Ma and Blenis 2009, Thoreen et al. 2012). The serine/threonine kinase ULK1 promotes initiation of autophagy by phosphorylating several proteins required for the initiation and the expansion of the phagophore (Egan et al. 2015, Lin and Hurley 2016). ULK1 is inactivated by mTORC1- mediated phosphorylation, resulting in suppression of autophagy under nutrient-replete conditions. A class of recently identified mTORC1 substrates is the MiT/TFE family, basic-helix loop helix (bHLH) transcription factors including TFEB, MiTF and TFE3. These proteins function as transcriptional activators of lysosomal biogenesis and autophagy via their ability to bind a DNA consensus sequence, known as the Coordinated Lysosomal Expression And Regulation (CLEAR) motif, located upstream of target genes (Sardiello et al. 2009, Settembre et al. 2011). mTORC1 directly phosphorylates TFEB, triggering binding of 14-3-3 proteins to TFEB and leading to its exclusion from the nucleus (Martina et al. 2012, Martina et al. 2014, Roczniak-Ferguson et al. 2012, Settembre et al. 2012). Thus, in high nutrient conditions, mTORC1- mediated phosphorylation of MiT/TFE factors serves to block the transcriptional activation of pro-catabolic gene programs.

1.1.2 The lysosome as a hub for communicating cellular nutrient status Lysosomes are small (100-500 nanometer diameter) organelles characterized by an acidic interior, which enables the activity of up to 60 different hydrolytic enzymes including proteases, lipases and nucleases. Its limiting membrane is spanned by numerous transmembrane proteins, which possess highly glycosylated luminal domains that protect the proteins and lipids of the limiting membrane from hydrolytic digestion (reviewed in Perera and Zoncu 2016, Saftig and Klumperman 2009, Settembre et al. 2013b). Lysosomal transmembrane proteins include an ATP-driven proton pump, known as the vacuolar H+ (v-ATPase), responsible for the acidification of the lysosome interior, as well as numerous transporters that mediate the export of lysosomal digestion products to the cytoplasm (Perera and Zoncu 2016, Saftig and Klumperman 2009, Settembre et al. 2013b). The lysosome is the end-point of numerous vesicle trafficking pathways including the endocytic, phagocytic and autophagic pathways. Its ability to fuse with endosomes, phagosomes and autophagosomes enables the lysosome to break down a wide range of both endogenous and exogenous cargo including macromolecules, certain pathogens and old or damaged organelles (reviewed in He and Klionsky 2009, Hurley and Schulman 2014, Mizushima et al. 2008, Singh and Cuervo 2011).

4 The traditional view of the lysosome as the cellular ‘recycle bin’ has been transformed by a number of recent discoveries, including the identification of a physical and functional association of this organelle with mTORC1 kinase (reviewed in Perera and Zoncu 2016). In particular, we now appreciate the ability of the lysosome to function as a “command-and-control center”, where nutrient supplies are directly monitored and this information is leveraged to influence critical decisions related to cell growth and survival (reviewed in Lim and Zoncu 2016). A deeper understanding of the lysosome and its role in communicating cellular nutrient status has elevated this previously neglected organelle to the center stage in disease-related research, including cancer, neurodegeneration, diabetes and aging (reviewed in Perera and Zoncu 2016, Saxton and Sabatini 2017, Settembre et al. 2013b).

1.1.3 Regulation of mTORC1 kinase activation at the lysosome In response to amino acids, mTORC1 translocates from the cytoplasm to the lysosomal surface, where it becomes competent to phosphorylate its substrates. This amino acid-dependent recruitment is mediated by the Rag GTPases, which exist as heterodimers of RagA or RagB in complex with RagC or RagD (yeast orthologs are known as Gtr1/Gtr2) (Kim et al. 2008, Sancak et al. 2010, Sancak et al. 2008). RagA/B and RagC/D are functionally equivalent, but are differentially expressed in certain tissues. A crucial property of the Rag GTPases is that their nucleotide state changes as a function of the nutrient status of the cell. When amino acid levels are low, it is thought that RagA/B is loaded with GDP, whereas RagC/D is loaded with GTP. The resulting RagAGDP-RagCGTP complex cannot bind to mTORC1, which therefore remains inactive in the cytoplasm. Amino acids trigger the switch to the RagAGTP-RagCGDP state, which is competent for mTORC1 binding and causes rapid translocation of mTORC1 to the lysosomal surface (Kim et al. 2008, Sancak et al. 2010, Sancak et al. 2008). Switching of the Rag GTPases from the inactive to the active state is mediated by the concerted action of guanine nucleotide exchange factors (GEFs) and GTPase activating proteins (GAPs) that act either on RagA/B or on RagC/D. The emerging picture is that the GEFs and GAPs for Rag GTPases are activated by an array of specialized sensor proteins located on the lysosomal limiting membrane and in the cytoplasm. These sensors appear to be strategically placed in order to integrate the availability of two amino acid pools, one in the cytoplasm and the other within the lysosomal lumen (Perera and Zoncu, 2016, Saxton and Sabatini, 2017, Settembre et al. 2013b). The integrated action of these two pools is essential to achieve full recruitment and activation of mTORC1 at the lysosomal surface. Translocation to the lysosome enables mTORC1 to interact with Rheb, the small GTPase that transmits signals from growth factors. The cryo-EM structure of the activated Rheb-mTORC1 complex was recently solved, revealing that apo mTORC1 adopts a conformation in which the kinase domain is autoinhibited due to misalignment of the kinase N- and C- lobes and the corresponding ATP-binding residues required for catalytic activity (Yang et al. 2017). Upon binding to Rheb, mTORC1 undergoes a conformational change that allosterically activates the kinase by promoting proper alignment of the catalytic cleft. Furthermore, cancer-associated mTOR mutants (Grabiner et al. 2014, Wagle et al. 2014, Urano et al. 2007) adopt the catalytically

5 unlocked conformation in the absence of Rheb binding, thus providing a clear explanation for their well-characterized hyperactive kinase phenotype.

1.1.4 Mechanisms of nutrient sensing upstream of mTORC1 A large body of work has shed light on the mechanisms by which amino acids cause the Rag GTPases to switch from the inactive to the active state. The pentameric Ragulator complex (composed of Lamtor1-5, the S. cerevisiae ortholog is known as the Ego 1-3 complex) is tethered to the lysosome through lipidation of the Lamtor1 subunit and acts as an essential platform for the Rag GTPases, enabling their lysosomal localization (Sancak et al. 2010, Binda et al. 2009, Powis et al. 2015). In addition, Ragulator is thought to function as a GEF that mediates the loading of RagA/B with GTP, thus promoting mTORC1 recruitment to the lysosome (Sancak et al. 2010, Bar- Peled et al. 2013). Conversely, the trimeric GATOR1 complex (SEACIT in S. cerevisiae) was shown to function as a RagA/B-specific GAP that enables mTORC1 inactivation when amino acid levels are low (Bar-Peled et al. 2013, Panchaud et al. 2013, Wolfson et al. 2017). It was recently shown that GATOR1 is anchored to the lysosomal surface by the recently described SZT2 protein, which is part of a GATOR1-regulating protein complex known as KICSTOR (Wolfson et al. 2017, Peng et al. 2017). The Folliculin (FLCN)/FNIP complex (Lst4/7 in yeast) is a GAP that converts RagC/D to the GDP- bound form; this step appears critical for the ability of the Rag heterodimer to bind to mTORC1 (Peli-Gulli et al. 2015, Petit et al. 2013, Tsun et al. 2013). A GEF for RagC/D has not been yet identified. An intriguing question is why the Rag GTPases are dimeric, whereas most signaling GTPases such as Ras, Rheb, and Rab GTPases are monomers. Possibly each Rag monomer plays a specialized role, such as binding to mTORC1 or to Ragulator. Also, having two GTPases, each with their own GAP and GEF, may allow fine-tuning of mTORC1 lysosomal recruitment by multiple nutrient inputs (Figure 1.2). Two amino acid pools are thought to cooperate in switching the Rag GTPases to the active state. In the cytoplasm, leucine-bidning protein Sestrin2 and an arginine- binding CASTOR1/2 interact with the GATOR2 complex and block its ability to inhibit GATOR1 when they are free from their respective amino acid ligands (Chantranupong et al. 2016, Saxton et al. 2016a, Saxton et al. 2016b, Wolfson et al. 2016). By binding to Sestrin2 and CASTOR1/2, respectively, leucine and arginine enable GATOR2 to block the GAP activity of GATOR1, thus causing RagA/B to remain in the GTP-bound state. Thus, mTORC1 binds to the lysosome in response to high micromolar concentrations of leucine and arginine in the cytoplasm (Chantranupong et al. 2016, Saxton et al. 2016a, Saxton et al. 2016b, Wolfson et al. 2016). The lysosomal lumen also contains amino acids, which are generated from proteolysis of both intracellular and extracellular substrates, and may also be imported from the cytoplasm by permeases on the lysosomal limiting membrane. The putative amino acid exporter, SLC38A9, which specifically localizes to the lysosome, has been implicated in mTORC1 activation by lysosomal amino acids (Jung et al. 2015, Rebsamen et al. 2015, Wang et al. 2015). Moreover, v-ATPase is also essential for mTORC1 recruitment to the lysosome in response to amino acids (Zoncu et al. 2011). Both SLC38A9 and the v-ATPase physically interact with the Ragulator complex, and may trigger its GEF activity toward

6 RagA/B in response to high amino acid concentrations within the lysosomal lumen (Bar- Peled et al. 2012, Wang et al. 2015, Zoncu et al. 2011, Shen and Sabatini 2018). Recent evidence indicates that mTORC1 recruitment to the lysosome is also regulated at the transcriptional level. The mTORC1 substrate TFEB strongly drives RagD expression, leading to enhanced lysosomal localization and signaling activity of mTORC1. This feedback mechanism may enable rapid resumption of growth as cells transition from starved to nutrient-replete states (Di Malta et al. 2017).

Figure 1.2 Mechanisms of nutrient mediated mTORC1 activation at the lysosome The Rag GTPases signal nutrient status to mTORC1 through their nucleotide loading state, which is regulated by GAPs and GEFs. In the nutrient replete state (shown on right), lysosomal nutrients including amino acids and cholesterol act upstream of Ragulator to promote the GTP loading of RagA/B. Cytosolic sensors of leucine (Sestrin2) and arginine (CASTOR1/2) block the RagA/B GAP activity of GATOR1 via GATOR2 binding. The Folliculin/FNIP complex acts as a GAP toward RagC/D to promote the GDP bound state. Together, this nucleotide state of the Rag GTPases adopts a conformation that promotes the recruitment of mTORC1 to the surface of the lysosome, leading to its subsequent activation by GTP bound Rheb. Upon nutrient depletion (shown on left), RagA/B is bound to GDP and RagC/D is GTP bound. This state is maintained, in part, via the RagA/B GAP activity of GATOR1 in response to the lack of cytosolic arginine and leucine. Inactivation of mTORC1 can also be triggered in response to growth factor depletion, leading to the inactivation of Rheb via the GAP activity of TSC1/2.

1.2 Emerging roles for the lysosome in lipid metabolism Precise regulation of lipid biosynthesis, transport, and storage is key to the homeostasis of cells and organisms. Cells rely on a sophisticated but poorly understood

7 network of vesicular and non-vesicular transport mechanisms to ensure efficient delivery of lipids to target organelles. The lysosome stands at the crossroads of this network due to its ability to process and sort exogenous and endogenous lipids. The lipid sorting function of the lysosome is intimately connected to its recently discovered role as a metabolic command-and-control center, which relays multiple nutrient cues to the master growth regulator, mTORC1 kinase. In turn, mTORC1 potently drives anabolic processes, including de novo lipid synthesis, while inhibiting lipid catabolism. Ultimately, the lysosome plays a dual role in lipid transport and biogenesis, and integration of these two processes play important roles both in normal physiology and in disease.

1.2.1 Lipid homeostasis is key for the function of cells, tissues, and organisms Lipid balance is intricately coordinated at multiple levels, including cellular uptake, synthesis, storage, metabolism, and export. All lipid classes, including fatty acids, sphingolipids, sterols and phospholipids undergo dynamic regulation, the extent of which depends on the specific tissue type and nutrient status of the cell (reviewed in (van Meer et al. 2008). A prominent and relatively well-studied example is cholesterol, a lipid with unique biophysical properties that support its essential role in regulating membrane structure and function (reviewed in Goldstein and Brown, 2015). Cholesterol modulates the activity of transmembrane proteins such as G protein-coupled receptors, and facilitates signal transduction and trafficking processes. In addition, cholesterol is the precursor for steroid hormones and bile acids, which function in signal transduction and absorption of dietary lipids, respectively. At the systemic level, organs perform specialized roles in maintaining cholesterol homeostasis. Dietary cholesterol is absorbed from the gut by intestinal enterocytes, which express a cholesterol-translocating membrane protein known as Niemann-Pick C1-like 1 (NPC1L1) in their apical membranes (reviewed in Chang et al. 2006, Ikonen 2008). Enterocytes package diet-derived lipids into chylomicrons, which are transported to the liver. There, hepatocytes package cholesterol, phospholipids and triglycerides into blood-soluble protein-containing particles known as very low-density lipoprotein (VLDL). Upon remodeling of their lipid and protein contents, VLDL mature into low- density lipoprotein (LDL), which deliver esterified cholesterol to peripheral tissues. The liver is responsible for the bulk of cholesterol biosynthesis, although the ability to synthesize endogenous cholesterol is common to most cell types (Chang et al. 2006, Goldstein and Brown 2015, Ikonen 2008). Endogenous sterol synthesis is critical in the brain, which is isolated from the circulating plasma cholesterol due to the inability of lipoproteins to traverse the blood-brain barrier. Failure to maintain the levels of circulating plasma cholesterol within a tight range promotes the development of atherosclerotic plaques in the arteries, resulting in high blood pressure and increased risk of cardiovascular disease. Plasma cholesterol levels are buffered by the uptake of LDL by extrahepatic tissues. Adipose tissue, in particular, serves as a reservoir for excess cholesterol through its incorporation into intracellular storage compartments known as lipid droplets. Cholesterol accumulation in extrahepatic tissues is counteracted by a process known as reverse cholesterol transport. In this case, cholesterol is transported across the plasma membrane via ABCA1, packaged into high-density lipoprotein (HDL) at the cell surface, and returned

8 through the plasma to the liver for biliary secretion. Oxidation of cholesterol to bile acids and their subsequent excretion is a liver-specific process, allowing for bulk elimination of sterols from the body (reviewed in Goldstein and Brown, 2015). At the cellular level, membrane-bound organelles are distinguished by their characteristic lipid compositions. Cholesterol is distributed heterogeneously on cellular membranes. A majority of cholesterol is found at the plasma membrane, while significant populations also exist at the endocytic recycling compartment (ERC) and the trans-Golgi network (van Meer et al. 2008). The endoplasmic reticulum, mitochondria, and lysosomes contain a relatively minimal amount of cholesterol. Within any given membrane, cholesterol is distributed heterogeneously. Through tight association with sphingolipids, cholesterol becomes enriched in ‘raft-like’ microdomains that are thought to play important roles in signal transduction (Das et al. 2014). Furthermore, organelles also serve diverse functional roles in maintaining cholesterol homeostasis. Endogenous cholesterol synthesis occurs in the endoplasmic reticulum, under the control of the rate- limiting enzyme HMG-CoA reductase (HMG-CoAR) (reviewed in Espenshade and Hughes 2007). Excess cholesterol in the ER is esterified via acetyl-CoA acetyltransferase (ACAT) 1 and stored in lipid droplets (reviewed in Thiam et al. 2013). In steroidogenic tissues such as gonads and adrenal glands, mitochondria import cholesterol into the inner mitochondrial membrane, where it is used as a substrate for steroid hormone synthesis. Lysosomes serve a key role in maintaining cholesterol homeostasis by acting as a processing center for LDL-derived cholesterol. Thus, cholesterol homeostasis is a highly organized process that requires cross-talk among many membranous compartments.

1.2.2 The lysosome as a trafficking station for lipids The convergence of several trafficking pathways at the lysosome makes this organelle a key player in coordinating the sorting and delivery of lipids to various membrane compartments. For instance, exogenous sterols, triglycerides, and phospholipids carried by low-density lipoprotein (LDL) enter the cell via receptor- mediated endocytosis and are processed by the lysosome (Goldstein and Brown 2015). The lysosome also sorts and recycles endogenous lipids. For example, fusion of lysosomes with double-membraned autophagosomes brings the autophagosomal internal membrane, along with engulfed organellar cargo, inside the lysosomal lumen. During endocytosis, a protein machinery known as the ESCRT complex mediates the inward budding of vesicles from the limiting membrane of endosomes (reviewed in Schoneberg et al. 2017). As these structures mature into lysosomes (either via direct conversion or fusion), the lysosomal lumen is progressively filled with vesicles. These vesicles eventually coalesce into membrane sheets or ‘whorls’ as their lipid composition is modified in the internal acidic environment of the lysosome. These exogenous or endogenous lipids must be removed from the lumen and inserted into the lysosomal limiting membrane. From there, the lipids can be further moved to other organelles such as the ER, where they may be esterified and stored in lipid droplets, or the mitochondria, where simple lipids such as cholesterol may be converted into steroid hormones.

9 1.2.3 Mechanisms of lipid transport within the lysosome A well-characterized lipid transport process involves the uptake and processing of LDL by the lysosome (Fig. 1.3) (reviewed in Goldstein and Brown 2015, Mesmin et al. 2013a). LDL particles bind to the LDL Receptor (LDLR) on the plasma membrane of target cells, and are taken up via clathrin-mediated endocytosis (CME). Following clathrin coated vesicle (CCV) uncoating, the LDL-receptor complexes enter the endosomal pathway. As the endosomal lumen progressively acidifies, LDL particles separate from LDLR and remain in the endosomal lumen, whereas LDLR is sorted back to the plasma membrane. As endosomes progressively mature into lysosomes, LDL particles are attacked by the lysosomal acid lipase type A (LipA), a hydrolase that de- esterifies the cholesterol and triglyceride molecules as they are liberated from LDL particles (Chang et al. 2006) The transport of LDL-derived, de-esterified cholesterol out of the lysosomal lumen has been studied in great detail. Free cholesterol is highly insoluble and cannot diffuse within the aqueous environment of the lysosomal lumen. Thus, it requires the concerted action of two proteins known as Niemann-Pick type C (NPC) 1 and 2 (Infante et al. 2008, Mesmin et al. 2013a). NPC2 is the first player in this pathway. It is a luminal protein that plays an obligate role in shuttling cholesterol through the lysosomal lumen. Structural studies have revealed that NPC2 harbors a cholesterol-binding pocket lined with hydrophobic residues. The highly hydrophobic iso-octyl chain is buried within the pocket, whereas the 3b-hydroxyl group on the first tetracyclic ring is exposed to the solvent (Xu et al. 2007). NPC2 is thought to deliver cholesterol to the luminal N-terminal domain of NPC1, although direct deposition of cholesterol into the limiting membrane has also been proposed (Deffieu and Pfeffer 2011, Infante et al. 2008, McCauliff et al. 2015). NPC1 is a polytopic transmembrane protein that harbors an N-terminal cholesterol-binding domain (NTD) protruding into the lumen. The crystal structure of the NPC1-NTD revealed a hydrophobic pocket in which cholesterol is oriented with the 3b- hydroxyl group inside and the iso-octyl chain facing out (Kwon et al. 2009). This orientation is opposite from the one in the NPC2 sterol binding pocket, supporting a model in which cholesterol is transferred in a direct manner from NPC2 to NPC1, via coordination between the respective sterol binding pockets. Recent structural works based on cryo-electron microscopy (cryo-EM) and X-ray crystallography have shed light on the overall structural organization of NPC1. In particular, the models suggest cooperation between the NTD and the middle luminal domain to allow docking of NPC2 and subsequent cholesterol transfer to the NTD (Gong et al. 2016, Li et al. 2016b). This work also provided for the first time a structural view of the transmembrane sterol-sensing domain of NPC1, which is homologous to that of other sterol-regulated proteins such as SCAP, Patched and HMG-CoA reductase (HMG-CoAR) (Brown et al. 2002, Martin et al. 2001, Millard et al. 2005, Sever et al. 2003, Yabe et al. 2002). Whereas the NPC1 SSD has no assigned function so far, an attractive model is that it could be involved in accepting cholesterol from the NTD and promoting its insertion into the limiting membrane. Recently, the lysosomal proteins LAMP1 and LAMP2 were shown to bind to cholesterol with high affinity and specificity via their luminal domain 1. This domain was also capable of interacting with NPC1 and NPC2. Based on these data, a model emerges where LAMP1/2 proteins, which are highly abundant at the lysosomal

10 membrane, may be acceptors for cholesterol from NPC2 and possibly facilitate the flow of cholesterol through the NPC2-NPC1 export pathway (Li and Pfeffer 2016).

Figure 1.3 Cellular routes for cholesterol trafficking LDLRs on the plasma membrane bind LDL particles and are delivered into EEs via clathrin-mediated endocytosis (1). As EE acidify and mature into LE, LDL dissociates from the LDLR and the LDLR is sorted into the ERC and returned to the plasma membrane. LDL continues along the endocytic pathway and ultimately reaches the lysosome (2), where the LDL is broken down (3). LipA hydrolyzes cholesteryl esters to release free cholesterol. Cholesterol is shuttled through the lysosomal lumen by NPC2, and subsequently transferred to the luminal N-terminal domain of the polytopic transmembrane protein NPC1 (4). The mechanisms of cholesterol export from the lysosome are still unknown, although they are likely to occur through several parallel routes that may involve the formation of membrane contact sites (shown by bidirectional dashed lines) between the lysosome and acceptor membranes (5). Endogenous cholesterol synthesis at the ER requires multiple biosynthetic enzymes including the rate-limiting enzyme HMG-CoA reductase. Under conditions of cholesterol abundance,

11 Figure 1.3 Cellular routes for cholesterol trafficking (legend continued from previous page) ER cholesterol can be esterified by ACAT1 and stored in lipid droplets (6). Abbreviations: ACAT1, acetyl-CoA acetyltransferase 1; EE, early endosome; ER, endoplasmic reticulum; ERC, endocytic recycling compartment; HMG-CoA, hydroxymethylglutaryl-CoA; LDLR, low-density lipoprotein receptor; LE, late endosome; LipA, lysosomal acid lipase; NPC, Niemann-Pick type C

1.2.4 Regulation of lipid export from the lysosome Once cholesterol has been deposited on the lysosomal membrane, its transport to extra-lysosomal locations is thought to occur rapidly (Mesmin et al. 2013a, Mobius et al. 2002). Emerging evidence indicates that sterol export relies on cooperation between vesicular and non-vesicular routes. The extensive lysosomal fission and fusion that has been observed in multiple cell types may serve to distribute lipids among lysosomes or deliver them to downstream membranes and organelles. Trafficking routes connect the late endosome/lysosome (LE/LY) to the plasma membrane, the Golgi apparatus and the endoplasmic reticulum. In addition to vesicular traffic, a major development in recent years has been identification of non-vesicular routes for intracellular lipid transport. Non-vesicular trafficking is mediated by several classes of lipid-binding proteins that localize at sites of physical proximity between different organelles, in both mammalian and yeast cells (reviewed in Daniele and Schiaffino 2016, Lahiri et al. 2015, Phillips and Voeltz 2016). A growing number of reports indicate that the lysosome may rely on membrane contacts for lipid transfer with other organelles, particularly the ER (for a detailed discussion, refer to Thelen and Zoncu 2017).

1.2.5 Feedback inhibition of lipid biogenesis Cells have evolved sophisticated mechanisms that sense the availability of specific lipid ligands. Upon binding of lipids to dedicated receptors, signaling responses are triggered which fine tune the rate of lipid synthesis, transport and storage in order to maintain membrane homeostasis. Perhaps the best understood example is the cholesterol sensing system in the ER membrane, which was discovered by Brown and Goldstein two decades ago (Goldstein and Brown 2015). This system is centered on a sterol-sensing transmembrane protein in the ER, known as sterol-responsive element- binding proteins (SREBP) Cleavage Activating Protein (SCAP) (Yang et al. 2002). The function of SCAP is to promote the loading of the membrane-anchored SREBP1/2 transcription factors onto COPII vesicles, which then transport SREBPs to the Golgi. At the Golgi, dedicated proteases cleave the transmembrane anchor of SREBPs (Sakai et al. 1998) and allow them to move into the nucleus, where the SREBPs bind to sterol responsive elements (SREs) on the promoter of genes involved in cholesterol uptake (e.g. LDL Receptor) and for de novo cholesterol synthesis (e.g, HMG-CoAR) (Sudhof et al. 1987). When cholesterol levels in the ER membrane are high, translocation and activation of SREBPs to the Golgi is inhibited (Radhakrishnan et al. 2004, Yang et al. 2002). Conversely, when cholesterol levels drop, SCAP undergoes a conformational change that causes its dissociation from Insig and allows SREBP to be loaded onto COPII vesicles (Radhakrishnan et al. 2008). In addition to cholesterol, SREBP1

12 integrates information about other lipid species, such as phosphatidylcholine (PC), to fine-tune cellular lipid balance (Smulan et al. 2016, Walker et al. 2011).

1.2.6 Control of lipid metabolism by mTORC1 The activation of mTORC1 in response to sufficient nutrients and growth factors has important consequences for cellular lipid homeostasis. mTORC1 promotes de novo lipid synthesis while concomitantly suppressing lipid catabolism (Fig. 1.4). Both actions synergize in promoting membrane biosynthesis during proliferation, as well as building lipid stores for future usage when precursor nutrients are plentiful (reviewed in Caron et al. 2015, Ricoult and Manning 2013). Activation of mTORC1 strongly drives the expression, trafficking and proteolytic processing of sterol-responsive element-binding proteins (SREBPs), SREBP1c and SREBP2, which are master regulators for the synthesis of fatty acids and sterols, respectively (Duvel et al. 2010, Owen et al. 2012, Yecies et al. 2011). In cellular and animal models, hyper-activating mTORC1 (e.g. by deletion of TSC2) resulted in enhanced activation of SREBPs and their target genes (Duvel et al. 2010, Wang et al. 2012). Conversely, ablating mTORC1 activity via rapamycin treatment or liver-specific deletion of its essential subunit Raptor led to suppression of SREBP-dependent lipid biogenesis (Peterson et al. 2011, Porstmann et al. 2008). At least some of the pro- lipogenic activities of mTORC1 are mediated by its substrate S6-Kinase 1 (S6K1). S6K1 was shown to promote SREBP1/2 activation and the downstream biosynthetic machinery for fatty acids and sterols (Duvel et al. 2010, Owen et al. 2012). A further mechanism for SREBP regulation is provided by lipin1, a phosphatidic acid-phosphatase involved in the synthesis of glycerolipids. In mammals lipin1 has been proposed to function as an mTORC1 substrate that regulates the SREBPs. Liver- specific inactivation of mTORC1 protected mice from lipid accumulation in the liver (a process known as hepatic steatosis) caused by high fat diet (HFD), but this protective effect was largely lost upon knock down of lipin1 (Peterson et al. 2011). Thus, lipin1 appears to be a key mediator of mTORC1-dependent lipogenesis. The ability of mTORC1 to promote lipid synthesis is particularly evident during adipogenesis, a developmental process in which precursor cells mature into full adipocytes though the enhanced synthesis and accumulation of triglycerides (Caron et al. 2015, Laplante et al. 2012, Polak et al. 2008, Zhang et al. 2009). Adipogenesis is induced by high circulating levels of insulin and nutrients in the bloodstream. Insulin and nutrients synergistically promote the synthesis and uptake of free fatty acids in adipocytes, and their subsequent esterification to glycerol in the ER of these cells. The resulting triglycerides are stored in lipid droplets (LD), which fill the cytoplasm of mature adipocytes. In in vitro models for adipogenesis, inhibiting mTORC1 by rapamycin treatment completely abolishes insulin-induced adipocyte differentiation by preventing the activation of peroxisome proliferator-activated receptor γ (PPAR-γ), a nuclear receptor that promotes pro-adipogenic and lipogenic transcriptional programs (Caron et al. 2015, Kim and Chen 2004). When glucose and amino acids are scarce, cells turn to fat as a primary energy source. In adipose cells, triglycerides within LDs are hydrolyzed to FFAs, which are released into the bloodstream and are taken up by tissues such as skeletal muscle, heart and liver. In these tissues, a process known as beta-oxidation, which takes place

13 inside mitochondria and peroxisomes, converts FFAs to acetyl-CoA while reducing + NAD and FAD molecules to NADH and FADH2, respectively. In the liver, beta-oxidation is required to trigger gluconeogenesis and maintain glycemia during fasting. This process is also essential to support the production of ketone bodies, which are released into the bloodstream and used by the brain (which is unable to perform beta-oxidation) during prolonged fasting. True to its role as a suppressor of catabolism, mTORC1 potently suppresses LD breakdown and peripheral beta-oxidation. In liver, accumulating evidence suggests that a specialized form of autophagy known as lipophagy participates in lipid store breakdown by targeting LDs for degradation in autolysosomes (Schulze et al. 2013, Singh et al. 2009). There, the lysosomal acidic lipase LIPA de-esterifies triglycerides and releases FFAs, which are then transported to mitochondria for beta-oxidation (Singh and Cuervo 2011). Consistent with this model, treating hepatocytes with rapamycin was sufficient to induce an autophagy-dependent reduction in LD number and a spike in the rate of beta-oxidation (Singh et al. 2009). Although this paradigm of autophagy-mediated LD consumption seems to hold true in liver cells, an alternative model has been put forward whereby autophagy actually promotes LD formation, possibly by recycling fatty acids and sterols from other membrane compartments. In this model, mTORC1 inhibition would be expected to increase LD number rather than causing LD disappearance (Nguyen et al. 2017, Rambold et al. 2015). How autophagy- derived lipids are funneled into LD formation, and how the subsequent transfer of LD- derived FFAs to mitochondria occurs are the subjects of active investigation (Barbosa and Siniossoglou 2017). TFEB has recently emerged as a major regulator of lipid catabolism in whole- organism settings (reviewed in Settembre and Ballabio 2014). Mice carrying a liver- specific deletion of TFEB (TFEB LiKO) failed to degrade LDs upon starvation, and displayed high circulating FFAs and defective peripheral beta-oxidation. Moreover, these animals gained more weight than control littermates when challenged with a HFD. Conversely, mice overexpressing TFEB in the liver displayed increased rates of beta- oxidation relative to wild-type littermates, had lower circulating FFAs and were strongly protected from weight gain induced by HFD (Settembre et al. 2013a). These actions of TFEB were due, at least in part, to its ability to control the expression of a transcriptional regulator of lipid catabolism, peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α), which, together with downstream nuclear receptor peroxisome proliferator- activated receptor α (PPAR-α), promotes the expression of numerous genes for FFA uptake and beta-oxidation in liver. TFEB also contributes to LD breakdown in the liver via its ability to directly induce the expression of autophagic genes. Similar findings in C. elegans show that the TFEB orthologue, HLH-30, promotes lipid catabolism, thereby influencing metabolic homeostasis and lifespan (O'Rourke and Ruvkun 2013). Thus, TFEB (possibly along with other MiT-TFE family members) exerts a powerful influence on lipid catabolic programs in the liver. Due to the role of mTORC1 as a gatekeeper of TFEB activity, the nutrients-mTORC1-TFEB axis emerges as a key mediator of the fasting response at the whole organism level (Perera and Zoncu 2016, Perera et al. 2019). Membrane biogenesis is essential for cellular growth and proliferation. Thus, one might expect that mTORC1 would be able to sense available lipids and facilitate their

14 utilization as well as the production of new lipid species for membrane production. Consistent with this possibility, feeding mice with a HFD boosts mTORC1 signaling in multiple tissues (Dibble and Manning 2013, Khamzina et al. 2005, Korsheninnikova et al. 2006, Um et al. 2004). However, whether this response is an indirect consequence of hyperinsulinemia triggered by lipid-driven insulin resistance, or if it reflects dedicated, cell autonomous mechanisms for lipid sensing remains unclear.

Figure 1.4 mTORC1 activation promotes lipid biogenesis and suppresses lipid catabolism In the presence of nutrients, mTORC1 translocates from the cytoplasm to the surface of the lysosome where it becomes activated and phosphorylates downstream substrates thereby activating (green box) or inhibiting them (red box). S6K1 is an mTORC1 substrate that promotes proteolytic processing of the transcription factors SREBP1/2. Inactive, full-length SREBP is retained at the ER membrane by its association with SCAP and INSIG in the presence of cholesterol. Under low cholesterol conditions in the ER, SCAP undergoes a conformational change releasing SREBP and promoting its COPII vesicle mediated trafficking to the Golgi. At the Golgi, two proteases, S1P and

15 Figure 1.4 mTORC1 activation promotes lipid biogenesis and suppresses lipid catabolism (legend continued from previous page) S2P, cleave SREBP to release the soluble, N-terminal fragment, which translocates to the nucleus and binds to SREs upstream of promoters for genes involved in sterol uptake and de novo lipid biosynthesis. Furthermore, mTORC1 phosphorylation blocks the nuclear entry of lipin1, a phosphatidic acid phosphatase that acts as an inhibitor of nuclear SREBP activity. Similarly, phosphorylation of TFEB prevents its translocation to the nucleus, repressing transcriptional activation of CLEAR genes involved in β- oxidation of fatty acids, lipid catabolism, and lysosomal biogenesis. mTORC1 is thought to promote adipocyte differentiation by releasing a translational block mediated by 4E- BP1. Finally, mTORC1 inhibits autophagic degradation of lipid droplets through phosphorylation of ULK1 kinase. Abbreviations: CLEAR, coordinated lysosomal expression and regulation; 4E-BP1, eukaryotic translation initiation factor 4E-binding protein 1; ER, endoplasmic reticulum; INSIG, insulin-induced gene 1 protein; S6K1, S6- kinase 1; SCAP, SREBP cleavage-activating protein; SRE, sterol response element; SREBP, sterol-responsive element-binding protein; TFEB, transcription factor EB

16 Chapter 2

Lysosomal Cholesterol Activates mTORC1 via an SLC38A9-Niemann-

Pick C1 Signaling Complex

A portion of the content presented in this chapter has been previously published as part of the following research article: Castellano, B.M.*, Thelen, A.M.*, Moldavski, O., Feltes, M., van der Welle, R. E. N., Mydock-McGrane, L., Jiang, X., van Ejkeren, R. J., Davis, O. B., Louie, S. M., Perera, R. M., Covey, D., Nomura, D. K., Ory, D. S., and Zoncu, R. (2017) Lysosomal Cholesterol Activates mTORC1 via an SLC38A9-Niemann Pick C1 Signaling Complex. Science. 355, 1306-1311.

Brian Castellano and Ashley Thelen were designated co-first authors. Brian Castellano, Ashley Thelen, and Roberto Zoncu conceived the study, designed experiments and wrote the manuscript. McKenna Feltes performed photoactivatable cholesterol binding assays and Xuntian Jiang performed mass spectrometry analysis with input from Daniel Ory. Sharon Louie performed mass spectrometry analysis of whole cell amino acids with input from Daniel Nomura. Ashley Thelen and Brian Castellano, executed all experimental work with assistance from Ofer Moldavski. All authors discussed the data and approved the manuscript.

17 2.1 Chapter Summary The mTORC1 protein kinase is a master growth regulator that becomes activated at the lysosome in response to nutrient cues. Here, we identify cholesterol, an essential building block for cellular growth, as a nutrient input that drives mTORC1 recruitment and activation at the lysosomal surface. The lysosomal transmembrane protein, SLC38A9, is required for mTORC1 activation by cholesterol through conserved cholesterol-responsive motifs. Moreover, SLC38A9 enables mTORC1 activation by cholesterol independently from its arginine-sensing function. Conversely, the Niemann- Pick C1 (NPC1) protein, which regulates cholesterol export from the lysosome, binds to SLC38A9 and inhibits mTORC1 signaling through its sterol transport function. Thus, lysosomal cholesterol drives mTORC1 activation and growth signaling through the SLC38A9-NPC1 complex.

2.2 Introduction Cholesterol is an essential building block for membrane biogenesis and rapidly proliferating cells rely on enhanced cholesterol synthesis and uptake to sustain their growth (reviewed in Nohturfft and Zhang 2009). Eukaryotic cells have evolved dedicated protein machinery that senses cholesterol and adjusts the rate of de novo cholesterol synthesis and uptake to meet the metabolic requirements of the cell (Espenshade and Hughes 2007, Goldstein and Brown 2015). Whether direct cholesterol sensing extends beyond the control of its own synthesis and, specifically, whether it integrates with pathways that regulate the control of cellular growth and proliferation is currently uncertain (Dibble and Manning 2013, Kalaany and Mangelsdorf 2006). The lysosome is a major processing station for dietary cholesterol. Low-density lipoproteins (LDL) carrying cholesterol esters and fatty acids enter the cell by receptor- mediated endocytosis and are disassembled by hydrolases in the lysosomal lumen (Goldstein and Brown 2015). A sterol transport system composed of the Niemann-Pick C1 (NPC1) and NPC2 proteins is localized to the late endosome or lysosome, where it mediates the export of free cholesterol to diverse cellular compartments including the plasma membrane and endoplasmic reticulum (ER) via mechanisms that remain poorly understood (Kwon et al. 2009). The sterol transport function of NPC1 is critical for proper cellular function, and its inactivation results in a lipid storage disorder, Niemann- Pick type C, characterized by spleen and liver abnormalities and neurodegeneration (Kobayashi et al. 1999, Ory 2004). The lysosome has also emerged as the cellular site where the master growth regulator, mTORC1 kinase, is activated. Nutrients including glucose and free amino acids act as input signals, which modulate the nucleotide loading state of the Rag guanosine triphosphatases (GTPases). Active Rag GTPases promote the recruitment of mTORC1 to the surface of the lysosome, where interaction with the small GTPase Rheb allosterically activates mTOR kinase. Activation of mTORC1 drives the synthesis of fatty acids and sterols by increasing the translation and proteolytic processing of the master lipogenic transcription factors, SREBP1c and SREBP2 (Owen et al. 2012, Yecies et al. 2011, Duvel et al. 2010). Proliferating cells also actively take up exogenous lipids in the form of LDL (Goldstein and Brown 2015). Considering the central role of mTORC1 in coordinating lipid metabolism, we hypothesized that excess lipids might feed back on mTORC1 activity. In particular, abundance of dietary cholesterol could be

18 a valuable signal given that cholesterol is bioenergetically costly to synthesize de novo and is often the limiting component of membrane synthesis (Lunt and Vander Heiden 2011). Given the dual role of the lysosome as a nutrient gateway and signaling platform for mTORC1 activation, we tested whether the lysosomal membrane contains dedicated machinery that transmits information about cholesterol availability to mTORC1.

2.3 Methods

2.3.1 Cholesterol starvation and restimulation in cells HEK-293T, CHO or MEF cells in culture dishes were rinsed once with serum-free media and incubated in DMEM containing 0.5-1.0% methyl-beta cyclodextrin (MCD) supplemented with 0.5% lipid-depleted serum (LDS) for 2 hours. Cells were then transferred to DMEM supplemented with 0.5% LDS and 0.1% MCD (starved condition), or to DMEM + 0.5% LDS containing 20µg/ml cholesterol pre-complexed with 0.1% MCD (resulting in MCD:cholesterol at 1:1 molar ratio, 50µM), or to DMEM + 0.5% LDS + 0.1% MCD supplemented with 25-50µg/ml low-density lipoprotein (LDL) and incubated for 1-2 hours. MCD:cholesterol complexes were prepared by diluting a 20mg/ml cholesterol stock solution (in EtOH) 1000-fold into a 15-ml falcon tube containing DMEM + 0.1% MCD + 0.5% LDS, resulting in 50µM final concentration of both cholesterol and MCD. The tube was vortexed and incubated in a 37C water bath for 2h. Lipid-depleted serum was prepared as described in (Goldstein et al. 1983).

2.3.2 Amino acid starvation and restimulation in cells HEK-293T cells in culture dishes were rinsed twice with serum-free, amino-acid free RPMI and incubated for 50 minutes, followed by stimulation with a 10x mixture of amino acids for 10 minutes. After stimulation, the final concentration of amino acids in the media was the same as in regular RPMI. The 10x mixture was prepared from individual powders of amino acids. For specific starvation of arginine or leucine, cells were washed twice in arginine- or leucine-free RPMI media and incubated for 50 minutes, followed by stimulation with a 10x mixture of either arginine or leucine for 10 minutes.

2.3.3 Double starvation of cholesterol and either arginine or leucine HEK-293T cells were seeded in 6 well plates to reach 90% confluency at the time of the experiment. Cells were rinsed twice in arginine-free RPMI and incubated in arginine- free RPMI containing 0.5% MCD for 2 hours. For cholesterol re-stimulation, the starvation media was replaced with 50µg/ml LDL in arginine-free RPMI containing 0.1% MCD for 1 hour 50 minutes. Then, a 10x solution of arginine was spiked in for the final 10 minutes followed by immediate harvesting and processing of cell lysates.

2.3.4 mTORC1 kinase assay HEK-293T, CHO, or MEF cells in culture dishes were treated as indicated for the given experiment. Immediately thereafter, media was aspirated and cells were lysed in ice-cold signaling buffer (10mM Na-PPi, 10mM Na-β-glycerophosphate, 40mM HEPES, 4mM EDTA, 1% Triton X-100, pH 7.4 supplemented with Pierce protease inhibitor (Pierce, Thermo Fisher Scientific), mechanically scraped from the plate, and allowed to

19 lyse on a rotator for 30 minutes at 4C. After centrifugation at 20,000 x g for 10 minutes in a 4C centrifuge, supernatant was collected and sample concentrations were normalized using the standard Bradford assay. Normalized samples were denatured in 5x SDS sample buffer, incubated for 10 minutes at 95C, resolved by SDS-PAGE, and analyzed by immunoblotting.

2.3.5 Immunofluorescence assays HEK-293T, CHO or MEF cells were plated on fibronectin-coated glass coverslips in 6-well plates (35mm diameter/well), at 300,000-500,000 cells/well. 12-16 hours later, cells were subjected to sterol or amino acid depletion/restimulation and fixed in 4% paraformaldehyde (in PBS) for 15 min at RT. The coverslips were rinsed twice with PBS and cells were permeabilized with 0.1% (w/v) Saponin in PBS for 10 min. After rinsing twice with PBS, the slides were incubated with primary antibody in 5% normal donkey serum for 1 hr at room temperature, rinsed four times with PBS, and incubated with fluorophore-conjugated secondary antibodies produced in goat or donkey (Life Technologies, diluted 1:1000 in 5% normal donkey serum) for 45 min at room temperature in the dark, washed four times with PBS. Coverslips were mounted on glass slides using Vectashield (Vector Laboratories) and imaged on a spinning disk confocal system (Andor Revolution on a Nikon Eclipse Ti microscope).

2.3.6 Co-immunoprecipitation from cell lysates HEK-293T cells stably expressing FLAG-tagged proteins were rinsed once with ice- cold PBS and lysed in ice cold lysis buffer (150mM NaCl, 20 mM HEPES [pH 7.4], 2 mM EDTA, 0.3% CHAPS or 1% Triton X-100, and one tablet of EDTA-free protease inhibitors per 50 ml). Cell lysates were cleared by centrifugation at 13,000 rpm for 10 minutes in a microfuge. For immunoprecipitations, 30 µl of a 50% slurry of anti-FLAG affinity gel (Sigma) were added to each lysate and incubated with rotation for 2-3 hours at 4°C. Immunoprecipitates were washed three times with lysis buffer. Immunoprecipitated proteins were denatured by the addition of 50 µl of 2x urea sample buffer and heating to 37°C for 15 minutes, resolved by 4%–12% SDS-PAGE, and analyzed by immunoblotting. A similar protocol was employed when preparing samples for mass spectrometry.

2.3.7 Lysosome immunoprecipitation for mass spectrometric analysis Confluent HEK-293T cells stably expressing LAMP1-mRFP-FLAGX2 (LRF), plated at 2 x 15cm dishes per condition in quadruplicate, were sterol-depleted for 2h in 0.5% MCD, or sterol-depleted and restimulated with 50µM cholesterol (in complex with MCD at 1:1 stoichiometry) for 2h. Following these incubations, cells were rinsed once in cold PBS, then scraped, spun down and resuspended in 750ul of fractionation buffer: 140mM KCl, 5mM MgCl2, 50mM Sucrose, 20mM HEPES, pH 7.4, supplemented with protease inhibitors. Cells were mechanically broken by spraying 4-5 times through a 23G needle attached to a 1ml syringe, then spun down at 2000g for 10min, yielding a post nuclear supernatant (PNS). PNS aliquots were equalized based on total protein concentration and subjected to overnight immunoprecipitation with 100µl of a 50% slurry of anti-FLAG affinity gel. The next day, beads were washed 4 times in fractionation buffer. After washing, beads were loaded on a spin column and lipids were eluted from

20 the beads by adding hexane:ethyl acetate (1:1, v/v).

2.3.8 Lysosomal cholesterol measurements using mass spectrometry The samples were extracted by modified Bligh-Dyer method in the presence of internal standard d7-cholesterol (20 µg per sample). The samples were further derivatized to improve the mass spectrometric detection sensitivity of cholesterol. Measurement of cholesterol was performed with a Shimadzu 10A HPLC system and a Shimadzu SIL-20AC HT auto-sampler coupled to a Thermo Scientific TSQ Quantum Ultra triple quadrupole mass spectrometer operated in SRM mode under ESI(+). Data processing was conducted with Xcalibur (Thermo). A quality control (QC) sample was prepared by pooling the aliquots of the study samples and was used to monitor the instrument stability. The QC was injected eight times in the beginning to stabilize the instrument, and was injected between every four study samples. The data was accepted if the coefficient variance (CV) of cholesterol in QC sample was < 15%. The quantification of cholesterol was calculated as the peak area ratio of cholesterol to d7- cholesterol multiplied by the quantity of d7-cholesterol (20 µg).

2.3.9 Intracellular amino acid measurements Metabolites from cell pellets were extracted in 40:40:20 acetonitrile:methanol:water with the inclusion of isotopic d3N15-serine as an internal standard. Samples were vortexed, sonicated, and centrifuged at 10,000 g, and an aliquot of the supernatants were injected onto an 6460 Agilent QQQ-LC-MS/MS.

2.3.10 siRNA knockdown experiments siRNA-based experiments were modified from Wang et al. 2015. HEK-293T cells were plated in a 6-well plate at 250,000 cells/well. 16 hours later, cells were transfected using Dharmafect 1 (Dharmacon) with 250 nM of pooled siRNAs (Dharmacon) targeting SLC38A9 or a non-targeting pool. 48 hours post-transfection, cells were transfected again as described. 24 hours following the second transfection, cells were replated in new 6-well plates at 1,000,000 cells/well for signaling assays . The following siRNAs were used:

Non-targeting: ON-TARGETplus Non-targeting Pool (D-001810-10-05) SLC38A9: SMARTpool: ON-TARGETplus SLC38A9 (L-007337-02-0005)

2.3.11 SLC38A9 peptide cholesterol binding assays The following synthetic peptides, encompassing the CARC-CRAC region of SLC38A9, along with a ‘scrambled’ peptide control, were purchased from Elim Peptides:

CARC-CRACWT: KKKRIFLLFQMMTVWPLLGYLARVKKK CARC-CRACY460I: KKKRIFLLFQMMTVWPLLGILARVKKK CARC-CRACF449I/Y460I: KKKRIFLLIQMMTVWPLLGILARVKKK WT-Scrambled: KKKARVRIFLYGFLLQLLMPVWTMKKK

The N-term and C-term Lys residues increase the solubility of the hydrophobic CARC-CRAC transmembrane segment. Replacing Pro456 located between the CARC

21 and the CRAC with Trp conferred intrinsic fluorescence upon 280nm excitation with maximum emission at 350nm. As previously reported, the conformation of the peptide affects the 350nm emission intensity (Derler et al. 2016). The fluorescent titration assay was performed as previously described (Derler et al. 2016). Stock solutions of 2.5mM peptide in DMSO were diluted to final 2.5µM concentration in assay buffer containing 140mM NaCl, 20mM Tris (pH 8.5), supplemented with 2mM TCEP and 2% (v/v) EtOH. Stock solutions of cholesterol or cholesterol-related ligands in EtOH were diluted to final concentrations ranging between 0.5 and 50 µM in 500 µl volumes of peptide solution, which were incubated overnight at 25C in mild agitation. Fluorescent measurements were carried out in triplicates on a Cary Eclipse fluorescence spectrophotometer (Agilent Technologies) using a 280nm excitation wavelength and collecting the 350nm emission. All measured fluorescence values fell within the linear range of the instrument. Fluorescence intensity values were averaged and expressed as relative increase over the 0µM ligand condition. Data were fitted to saturation binding equations using GraphPad Prism 6. Dotted lines indicate where data could not be fitted.

2.3.12 SLC38A9 labeling with photoactivatable cholesterol probe On day 0, human skin fibroblasts expressing FLAG tagged SLC38A9 were plated at 5x105 in 10cm plates. On day 1, media was changed to 10% lipid-depleted serum (LDS). On Day 3, cells were treated with 2.5µM 7-Azi-27-yne (LKM-38) probe complexed with MCD in the presence or absence of cholesterol-complexed MCD, diluted in DMEM for 1 hour at 37C. Media was aspirated and replaced with cold PBS. Cells were crosslinked for 2.5 minutes on ice in a UV Stratalinker 7000, harvested by scraping, and lysed in 50mM Tris, 150mM NaCl, 1% NP-40 for 1.5 hours with rotation. Protein concentration was determined by BCA assay. Lysate was incubated with anti- FLAG M2 agarose beads (Sigma) for 2hrs at 4C to achieve SLC38A9-FLAG pull down. Beads were washed, then suspended in lysis buffer. Click reagents were added to final concentrations of 1mM CuSO4, 100µM Tris(3-hydroxypropyltriazolylmethyl)amine (THPTA), 100µM ROX azide, and 5mM ascorbic acid. The click reaction proceeded at 37C for 0.5 hours, shaking at 1200 rpm. Beads were washed and then suspended in Laemmli sample buffer before analysis by SDS-PAGE. Fluorescent images were collected at 610nm using a Typhoon Trio imager.

2.4 Results

2.4.1 Cholesterol regulates mTORC1 activity In cultured human embryonic kidney (HEK-293T) cells, bulk depletion of cellular cholesterol using the chemical compound methyl-β-cyclodextrin (MCD) inhibited mTORC1 signaling as indicated by the reduction in phosphorylation of the mTORC1 substrates S6-kinase 1 (S6K1, p-Thr389) and EIF4E-binding protein 1 (4E-BP1, p- Ser65), which could be restored by adding exogenously supplied LDL in a dose- dependent manner (Fig. 2.1A). The kinetics of mTORC1 activation in response to LDL revealed a lag time of approximately 1 hour, which correlated with the time required for bodipy-LDL to traffic to the lysosome as measured by live-cell imaging (Fig. 2.1B). This suggested that breakdown of LDL particles at the lysosome was required for mTORC1

22 activation. Next, we tested whether direct delivery of cholesterol to cells could stimulate mTORC1 signaling, ruling out the effects of other LDL-derived components. Free cholesterol can be directly deposited into cellular membranes by treating cells with pre- complexed MCD:cholesterol (using a 1:1 molar ratio as described in Radhakrishnan et al. 2008). Following cholesterol depletion, re-feeding of cells with MCD:cholesterol for 1 hour restored mTORC1 signaling to a similar extent as LDL. Translocation of the mTORC1 substrate TFEB was measured as an additional readout of mTORC1 activity. Upon mTORC1 inhibition, dephosphorylated TFEB migrates from the cytosol to the nucleus where it assumes its role as a transcription factor. Cholesterol depletion in HEK-293T cells stably expressing TFEB-GFP induced a dramatic redistribution of TFEB to the nucleus, which was reversed upon addition of LDL (Fig. 2.1D,E).

Figure 2.1 Cholesterol regulates mTORC1 activity (A) Dose-dependent activation of mTORC1 by LDL. HEK-293T cells were depleted of sterol with methyl-beta cyclodextrin (MCD, 0.5% w/v) for 2 hours and stimulated for 2 hours with various concentrations (0-100 µg/ml) of LDL. Cell lysates were analyzed for phosphorylation status of S6K (T389) and 4E-BP1 (S65) and for total protein abundance. (B) (top) Time-dependent activation of mTORC1 by LDL. HEK-293T cells were depleted of sterol for 2 hours and re-stimulated with LDL (50 µg/ml) for the indicated times. Cell lysates were analyzed for phosphorylation status of S6K (T389). (bottom) Time-course of LDL delivery to the lysosome. Cells stably expressing LAMP1- mRFP-FLAGX2 (LRF) were treated with BODIPY-LDL for the indicated times. Scale bar, 0.5µm (C) Activation of mTORC1 signaling by cholesterol. HEK-293T cells were

23 Figure 2.1 Cholesterol regulates mTORC1 activity (legend continued from previous page) depleted of sterol for 2 hours and, where indicated, re-stimulated for 2 hours with cholesterol (20µg/ml) complexed with 0.1% MCD or with LDL (100µg/ml) where indicated. Phosphorylation of S6K and 4E-BP1 are shown. (D) Regulation of TFEB nuclear localization by cholesterol. HEK-293T cells stably expressing TFEB-GFP were depleted of sterol and, where indicated, re-stimulated with LDL (50µg/ml). Scale bar, 10µm. (E) Quantification of TFEB-GFP localization from (D). Shown are mean + SD. N=800 cells. ANOVA: p < 0.0001 followed by Tukey’s t-test: ****p < 0.0001 (F) (top) Lysosomes from LRF-expressing HEK-293T cells were immuno-captured onto FLAG affinity beads. (bottom) Mass spectrometry measurement of unesterified cholesterol in immuno-captured lysosomes from LRF-expressing HEK-293T cells subjected to the indicated treatments. Shown are mean + SD. N=4 samples/condition. ANOVA: p < 0.05 followed by Tukey’s t-test: *p <0.05

2.4.2 Cholesterol specifically activates mTORC1 in a pathway distinct from growth factor signaling Altering the lipid composition of membranes can facilitate the creation of membrane ‘raft-like’ microdomains, which have been shown to regulate cellular signaling processes (Das et al. 2014). In cholesterol depleted HEK-293T cells, mTORC1 activation occurred following restimulation with LDL and cholesterol, but not a variety of closely related cholesterol species, including the enantiomer and epimer of cholesterol (Fig. 2.2A). Therefore, activation of mTORC1 by cholesterol appears to be specific, suggesting that cholesterol recognition is protein mediated (i.e. receptor/ligand binding) rather than a consequence of modulating membrane composition. Changes in cholesterol levels can alter the activity of signaling pathways originating at the plasma membrane such as phosphoinositide-3 kinase (PI3K) and the protein kinase Akt, which acts upstream of mTORC1 (Dibble and Manning 2013, Perera and Zoncu 2016). To test whether cholesterol activates mTORC1 through the growth factor signaling arm, we used mouse embryonic fibroblasts (MEFs) in which PI3K-Akt input to mTORC1 is constitutively active due to lack of the Tuberous Sclerosis Complex 2 (TSC2) gene (Yecies et al. 2011, Tee et al. 2003). A postulate of the AND-gate logic of mTORC1 signaling is that shutting off either input to mTORC1 results in inactivation (diagramed in Fig. 2.2C). Thus, if cholesterol acts upstream of Rheb in the growth factor arm, cholesterol depletion should not inhibit the constitutively active mTORC1 signaling in TSC2-/- cells. However, cholesterol depletion in TSC-/- MEFs significantly reduced mTORC1 signaling in a dose-dependent manner (Fig. 2.2B). Thus, cholesterol concentrations affect mTORC1 activity through a mechanism that is, at least in part, distinct from growth factor signaling.

Figure 2.2 Cholesterol specifically activates mTORC1 in a pathway distinct from growth factor signaling (A) HEK-293T cells were sterol depleted for 2 hours and, where indicated, restimulated for 1 hour with the indicated sterol-related compounds. Cell lysates were analyzed for phosphorylation of S6K (T389) and 4E-BP1 (S65) and for total protein abundance. (B) Cholesterol dependence of mTORC1 signaling in TSC2WT and TSC2-/- MEFs. Cells of the indicated genotype were incubated for 3 hours with increasing concentrations of MCD, followed by immunoblotting for the indicated proteins. (C) AND-gate logic of mTORC1 activation by nutrients and growth factors. Bold lines indicate hyperactive signaling. Dashed lines indicate attenuation of signaling.

2.4.3 Cholesterol regulates lysosomal recruitment of mTORC1 via the Ragulator- Rag GTPase complex Given the previous result, we reasoned that cholesterol likely acts through the nutrient sensing arm, which converges on the Rag GTPases and their lysosomal scaffold, Ragulator, to regulate mTORC1 localization. Manipulation of cholesterol levels in HEK-293T cells did not disrupt the integrity of the Ragulator-Rag GTPase scaffold, as RagC and the p18 subunit of Ragulator remained co-localized with LAMP2 positive lysosomes irrespective of cholesterol status (Fig. 2.3A,C). In contrast, cholesterol depletion caused mTOR to leave the lysosome, whereas restimulation with either LDL or MCD:cholesterol restored the lysosomal localization of mTOR (Fig. 2.3B,C). The ability of cholesterol to promote translocation of mTORC1 to the lysosome is consistent with the previous results demonstrating activation of mTORC1 kinase activity. To further test the hypothesis that cholesterol acts upstream of the Rag GTPases, we took advantage of a constitutively active point mutation of RagB, Q99L, which

Figure 2.3 Cholesterol regulates lysosomal recruitment of mTORC1 via the Ragulator-Rag GTPase complex (A) Cholesterol status does not affect Ragulator localization to LAMP2-positive lysosomes. HEK-293T cells were subjected to the indicated treatments, followed by immunofluorescence for endogenous p18 and LAMP2. Scale bar, 10µm. (B) Cholesterol regulates mTORC1 recruitment to LAMP2-positive lysosomes. HEK-293T cells were subjected to the indicated treatments, followed by immunofluorescence for endogenous mTOR and LAMP2. Scale bar, 10µm. (C) Quantification of RagC-LAMP2, p18-LAMP2 and mTOR-LAMP2 co-localization under cholesterol-depleted and cholesterol- stimulated conditions. Shown are mean + SD. N=15 cells/condition. ANOVA: p < 0.0001

26 Figure 2.3 Cholesterol regulates lysosomal recruitment of mTORC1 via the Ragulator-Rag GTPase complex (legend continued from previous page) followed by Tukey’s t-test: ***p < 0.001. (D) Control and RagBQ99L -expressing HEK- 293T cells were sterol-depleted for 2h, or depleted and, where indicated, restimulated for 1h, followed by immunofluorescence staining for endogenous mTOR and LAMP2. Scale bar, 10µm. (E) Quantification of mTOR-LAMP2 co-localization in control HEK- 293T and in HEK-293T expressing RagBQ99L under sterol-depleted and sterol- stimulated conditions. Shown are mean + SD. N=15 cells/condition. ANOVA: p < 0.0001 followed by Tukey’s t-test: ****p < 0.0001. (F) HEK-293T cells stably expressing the constitutively active RagBQ99L mutant, along with control HEK-293T, were sterol- depleted for 2h, or depleted and restimulated for 2h. Cell lysates were immunoblotted for the indicated proteins and phospho-proteins. (G) Cholesterol regulates the interaction between Ragulator and Rag GTPases in cells. HEK-293T cells stably expressing FLAG-p14 or LRF were sterol-depleted for 2h and, where indicated, followed by restimulation for 2h. After lysis, samples were subjected to FLAG immunoprecipitation and immunoblotting for the indicated proteins. causes mTORC1 to be permanently targeted to the lysosomal surface, regardless of nutrient status (Sancak et al. 2008, Kim et al. 2008). In cells stably expressing FLAG- RagBQ99L mTORC1 remained on the lysosome, even upon cholesterol depletion (Fig. 2.3D,E). Moreover, stable expression of RagBQ99L also rescued the phosphorylation of mTORC1 substrates S6K1 and 4E-BP1 in the cholesterol depleted condition (Fig. 2.3F). To independently confirm that cholesterol regulates the Rag GTPases, we used a co-immunoprecipitation assay, which measures the activation state of the Rag GTPases as a function of their affinity for Ragulator. The Ragulator-Rag GTPase interaction is weakened in the presence of amino acids. This effect is thought to reflect GTP loading of Rag A or Rag B by Ragulator, enabling RagA or B to bind to mTORC1 (Zoncu et al. 2011, Bar-Peled, et al. 2012). In HEK-293T cells that were cholesterol depleted and then restimulated with MCD:cholesterol, the FLAG tagged Ragulator subunit p14 bound to endogenous RagA and RagC more weakly than in cholesterol depleted cells (Fig. 2.3G). Thus, cholesterol closely resembles amino acids in its ability to regulate mTORC1 localization and activation via the Rag GTPases.

2.4.4 SLC38A9 mediates mTORC1 activation in response to lysosomal cholesterol Having established that cholesterol acts upstream of the Rag GTPases, we next asked whether cholesterol levels were being directly sensed at the lysosome. Given that the lysosome facilitates the release free cholesterol from LDL and the kinetics of mTORC1 activation paralleled the arrival of LDL to the lysosome, we hypothesized that the lysosomal pool of cholesterol might be particularly important for mTORC1 activation. In order to test whether global cellular cholesterol depletion using MCD impacted lysosomal cholesterol levels, cells stably expressing the lysosomal membrane protein LAMP1 with a C-terminal mRFP-FLAGx2 fusion (LRF) were mechanically lysed and intact lysosomes were immunoprecipitated and subjected to mass spectrometric analysis to quantify lysosomal cholesterol levels (Fig 2.4A, top). Relative to the complete media control, cells treated with MCD showed a greater than 2-fold decrease

27 of total lysosomal cholesterol. Furthermore, refeeding with exogenous cholesterol restored lysosomal cholesterol levels to baseline (Fig. 2.4A, bottom). Using MCD to deplete total cellular cholesterol does not allow for specific manipulation of the lysosomal pool of cholesterol. Thus, further in vitro experiments conducted by the co-author of this work, Brian Castellano, showed that directly altering lysosomal cholesterol levels recapitulated the changes in Ragulator-Rag GTPase affinity measured in live cells. An in vitro assay was performed using light organelle fractions from HEK-293T cells stably expressing FLAG-p14 and treated with recombinant cholesterol oxidase, a bacterial enzyme that converts cholesterol into cholest-4-en-3-one, effectively depleting cholesterol from the limiting membrane of the lysosome. Importantly, cholest-4-en-3-one failed to rescue mTORC1 signaling in cholesterol depleted cells (Fig. 2.2A). Cholesterol oxidase-mediated depletion of lysosomal cholesterol significantly increased the affinity of p14 for Rag A and Rag C, reflecting inactivation of the lysosomal scaffolding complex. Furthermore, the activity of the enzyme was competed by adding increasing amounts of free cholesterol substrate into the light organelle fraction, resulting in a dose-dependent decrease in Ragulator- Rag GTPase interactions. Thus, lysosomal cholesterol modulates protein-protein interactions within the mTORC1 scaffolding complex in a manner consistent with an activating effect toward the Rag GTPases. A search for cholesterol-regulated motifs within the mTORC1 scaffolding complex identified a putative Cholesterol Recognition Amino acid Consensus (CRAC) motif within transmembrane domain 8 (TM8) of the lysosomal Na+ coupled amino acid transporter, SLC38A9 (Fig. 2.4B, top). SLC38A9 has been implicated in the regulation of mTORC1 by amino acids, particularly arginine (Jung et al. 2015, Rebsamen et al. 2015, Wang et al. 2015). The consensus CRAC sequence: (L/V)-X1-5-Y-X1-5-(K/R) of SLC38A9 is conserved across vertebrates (Derler et al. 2016, Fantini and Barrantes 2013). SLC38A9 is the only member of the SLC38 NA+ coupled amino acid transporter family that contains a CRAC motif (Fig. 2.4B, bottom). Moreover, the CRAC motif is immediately preceded by a similar by inverted cholesterol recognition motif known as CARC, such that the tandem motifs together span the lipid bilayer (Fig. 2.4B, top). CRAC and CARC motifs have been described in cholesterol regulated protein such as caveolins and the Ca2+ channel Orai1 (Derler et al. 2016). CRISPR/Cas9 mediated deletion of SLC38A9 in HEK-293T cells largely abolished mTORC1 activation in response to LDL, as well as by arginine as previously reported (Fig. 2.4C) (Wang et al. 2015). Mechanistically, SLC38A9 was required for cholesterol- mediated regulation of Ragulator-Rag GTPase interactions. In cells lacking SLC38A9, FLAG-p14 bound to RagA and RagC more tightly than in control cells, indicative of complex inactivation, and these interactions could not be weakened by addition of cholesterol (Fig. 2.4D). Moreover, in cells lacking SLC38A9, LDL largely failed to induce recruitment of mTORC1 to LAMP2-containing lysosomes (Fig 2.4E,F), further suggesting that the mTORC1 scaffolding complex senses cholesterol, at least in part, via SLC38A9.

Figure 2.4 SLC38A9 is required for mTORC1 activation by lysosomal cholesterol (A) (top) Lysosomes from LRF-expressing HEK-293T cells were immuno-captured onto FLAG affinity beads. (bottom) Mass spectrometry measurement of unesterified cholesterol in immuno-captured lysosomes from LRF-expressing HEK-293T cells subjected to the indicated treatments. Shown are mean + SD. N=4 samples/condition. ANOVA: p < 0.05 followed by Tukey’s t-test: *p <0.05 (B) (top) Conservation analysis of

29 Figure 2.4 SLC38A9 is required for mTORC1 activation by lysosomal cholesterol (legend continued from previous page) transmembrane helix 8 (TM8) of SLC38A9, with the CARC and CRAC motifs indicated by red lines. The essential Phe and Tyr within the CARC and CRAC motif, respectively, are boxed in red. (bottom) Alignment of TM8 of SLC38A9 with other members of the SLC38 transporter family. The red boxes highlight essential Phe 449 and Tyr 460 within the CARC and CRAC domains, respectively. (C) Control or SLC38A9-deleted HEK- 293T cells were double-starved for cholesterol and arginine for 2h and, where indicated, restimulated with LDL, arginine or both for 2h or 30min, respectively. Cell lysates were immunoblotted for the indicated proteins and phospho-proteins. (D) SLC38A9 is required for cholesterol modulation of the Ragulator-Rag GTPase interaction. Control or SLC38A9-deleted HEK-293T stably expressing FLAG-p14 or LRF were subjected to cholesterol starvation for 2h and, where indicated, followed by restimulation for 2h. Cells were lysed and subjected to FLAG immunoprecipitation and immunoblotting for the indicated proteins. (E) SLC38A9-deleted or FLAG-SLC38A9 rescued HEK-293T cells were subjected to the indicated treatments, followed by immunofluorescence for endogenous mTOR and LAMP2. Scale bar, 10µm. (F) Quantification of mTOR-LAMP2 co-localization from the experiment in (D). Shown are mean + SD. N=15 cells/condition. ANOVA: p < 0.0001 followed by Tukey’s t-test: ****p < 0.0001.

To test whether SLC38A9 could directly interact with cholesterol, we synthesized an ultraviolet light (UV)-photoactivatable cholesterol analog, 7-Azi-27-yne (herein called LKM-38), bearing a diazirine group in position 7 and an alkyne handle on the alkyl side chain (Fig. 2.5A). Human fibroblasts stably expressing FLAG-SLC38A9 were incubated with LKM-38:MCD and then exposed to UV light to induce covalent crosslinking. FLAG immunoprecipitation was performed on cell lysates to capture FLAG-SLC38A9 and associated LKM-38 was detected by copper(I)-catalyzed click chemistry, resulting in conjugation of rhodamine-X azide (ROX) dye (Fig. 2.5A). A fluorescent signal was detectable at the expected molecular weight for immunoprecipitated SLC38A9 (Fig. 2.5B) and was competed in a dose-dependent manner when unmodified cholesterol was included in the incubation step (Fig. 2.5C). Importantly, unlike many other sterol- related compounds we tested, LKM-38 stimulated mTORC1 activity with similar potency to cholesterol (Fig. 2.5D) Next, we rescued SLC38A9-depleted cells with wild-type, Y460I mutant or F449I/Y460I double mutant FLAG-SLC38A9 and confirmed their lysosomal localization in HEK-293T cells (Fig. 2.6B). Wild-type SLC38A9 restored mTORC1 activity (Fig. 2.6C) and lysosomal localization (Fig. 2.6E,F) in response to LDL. However, activation and lysosomal localization of mTORC1 in response to LDL was clearly impaired in cells reconstituted with either the Y460I or F449I/Y460I mutants of SLC38A9 (Fig. 2.6C,E,F). Notably, neither mutant disrupted arginine-mediated activation of mTORC1, indicating a specific role for the CARC/CRAC motif in cholesterol-dependent mTORC1 signaling (Fig. 2.6C). Moreover, both SLC38A9Y460I and SLC38A9F449I/Y460I showed increased interactions with Ragulator and the Rag GTPases, a pattern consistent with their failure to activate the mTORC1 scaffolding complex (Fig. 2.6D). Taken together, these results indicate that SLC38A9 is necessary for mTORC1 activation by lysosomal cholesterol and that the tandem CARC and CRAC motifs are important in mediating this process.

Figure 2.5 SLC38A9 binds LKM-38, a photoactivatable cholesterol probe (A) Schematic of the cholesterol photoactivation assay. Human fibroblast cells stably expressing FLAG-SLC38A9 are treated with the photoactivatable cholesterol analog LKM-38, followed by UV-crosslinking, FLAG immunoprecipitation and clicking to rhodamine-X azide (ROX). (B) Crosslinking of the photoactivatable-cholesterol analogue LKM-38 to FLAG-SLC38A9 as described in (A), in the presence of increasing amounts of competing native cholesterol. The FLAG blot shows the immunoprecipitated FLAG-SLC38A9 in each condition. (C) Quantification of 3 independent sterol crosslinking experiments as described in (B). Shown are mean + SD. N=3 samples/condition. ANOVA: p < 0.0001 followed by Tukey’s t-test: **p < 0.01, ****p < 0.0001. (D) HEK- 293T cells were sterol-depleted for 2h and, where indicated, restimulated for 1h with identical concentrations of the indicated sterol-related ligands. Cell lysates were analyzed for phosphorylation status of S6K1.

In order to dissect the specific role of the CARC and CRAC motifs of SLC38A9, in vitro binding assays were performed using a peptide that encompasses the two motifs (amino acids 444-464). As described for other CRAC-containing proteins (Derler et al. 2016), titrating the peptide with increasing concentrations of cholesterol yielded a dose- dependent, saturable increase of the 350nm fluorescence emission of a Trp residue within the peptide (Fig. 2.6A, top), suggesting a conformational change induced by cholesterol. In contrast, 25-hydroxycholesterol and cholest-4-en-3-one, neither of which activates mTORC1 in vivo (Fig. 2.2A), failed to increase the fluorescence emission of the peptide (Fig. 2.6A, top). Mutating the obligate Tyr 460 of the CRAC motif to Ile

31 reduced the cholesterol-induced fluorescence change, whereas a double mutant of the CARC F449I and CRAC Y460I abolished it altogether (Fig. 2.6A, bottom).

Figure 2.6 Mutations in the putative cholesterol responsive motifs of SLC38A9 diminish mTORC1 activation and lysosomal translocation in response to cholesterol (A) Titration curves displaying changes in the intrinsic fluorescence of the SLC38A9

32 Figure 2.6 Mutations in the putative cholesterol responsive motifs of SLC38A9 diminish mTORC1 activation and lysosomal translocation in response to cholesterol (legend continued from previous page) CARC-CRAC peptide (2.5µM) in the presence of increasing concentrations of cholesterol, 25-hydroxycholesterol (25-HC) or cholest-4-en-3-one (3-one). Shown are mean ± SD. N=3 samples/condition. ANOVA: ****p < 0.0001, ***p < 0.001 (B) Titration curves displaying changes in the intrinsic fluorescence of the indicated CARC-CRAC peptides (wild-type and mutants) or a control ‘scrambled’ peptide (Scr, all peptides at 2.5µM) in the presence of increasing concentrations of cholesterol. Shown are mean ± SD. N=3 samples/condition. ANOVA: ****p < 0.0001. (C) SLC38A9-deleted HEK-293T cells stably expressing the indicated wild-type and mutant FLAG-SLC38A9 constructs were starved for the indicated nutrients for 2 hours, or starved and re-stimulated with LDL or arginine for 2h or 30 min, respectively. Cell lysates were immunoblotted for the indicated proteins and phospho-proteins. (D) SLC38A9-deleted HEK-293T cells re- expressing wild-type or Y460I-mutated FLAG-SLC38A9 were subjected to the indicated treatments, followed by immunofluorescence for endogenous mTOR and LAMP2. Scale bar, 10µm. (E) Quantification of mTOR-LAMP2 co-localization from the experiment in (D). Shown are mean + SD. N=15 cells/condition. ANOVA: p < 0.0001 followed by Tukey’s t-test: ****p < 0.0001. (E) SLC38A9-deleted HEK-293T cells re-expressing wild- type or Y460I-mutated FLAG-SLC38A9 were subjected to the indicated treatments, followed by immunofluorescence for endogenous mTOR and LAMP2. Scale bar, 10µm. (F) Quantification of mTOR-LAMP2 co-localization from (E). Shown are mean + SD. N=15 cells/condition. ANOVA: p < 0.0001 followed by Tukey’s t-test ****p < 0.0001.

2.4.5 Niemann-Pick C1 protein interacts with SLC38A9 and the Ragulator-Rag GTPase complex The activating effect of LDL-derived cholesterol on mTORC1 implies that lysosomal proteins that function in cholesterol transport may participate in mTORC1 regulation. FLAG-tagged NPC1, a bona fide cholesterol exporter of the lysosomal limiting membrane, immunoprecipitated multiple v-ATPase and Ragulator subunits including V1B2, p14, p18, RagC, as well as SLC38A9, whereas LRF did not bind to any of these components (Fig. 2.7A,C). Furthermore, FLAG-SLC38A9 associated with endogenous NPC1 along with the Rag GTPases and Ragulator (Fig. 2.7B,D). Binding of NPC1- FLAG to V1B2, Ragulator, and the Rag GTPases was reduced in SLC38A9-null cells, suggesting that NPC1 may associate with the mTORC1 scaffolding complex indirectly through SLC38A9 (Fig. 2.7E). Further experiments conducted by the co-author of this study, Brian Castellano, revealed that NPC1 was essential for mTORC1 regulation by cholesterol. CRISPR/Cas9 mediated deletion of NPC1 in HEK 293T cells (NPC-/-) resulted in a phenotype in which mTORC1 signaling was not reduced following cholesterol depletion with MCD. In the NPC1-/- background, stable expression of wild-type NPC1, but not the transport defective mutant P691S, restored mTORC1 sensitivity to cholesterol depletion. In line with the requirement for SLC38A9 to signal cholesterol sufficiency to mTORC1, siRNA-mediated knockdown of SLC38A9 in the NPC1-/- background led to an overall reduction in mTORC1 signaling relative to the siScramble control, but mTORC1

33 inhibition in response to cholesterol depletion was not fully restored (suggesting that SLC38A9 is not the only cholesterol responsive protein involved).

Figure 2.7 Niemann-Pick C1 protein interacts with SLC38A9 and the Ragulator- Rag GTPase signaling complex (A) Cartoon summarizing mass spectrometry analyses of immunoprecipitates from HEK-293T cells expressing FLAG-NPC1. v-ATPase subunits are color-coded according

34 Figure 2.7 Niemann-Pick C1 protein interacts with SLC38A9 and the Ragulator- Rag GTPase signaling complex (legend continued from previous page) to their peptide representation. Peptide counts for v-ATPase subunits from 5 independent experiments are shown in the table below. (B) HEK-293T cells stably expressing LAMP1-FLAG or NPC1-FLAG were lysed in the presence of Triton X-100 and subjected to FLAG immunoprecipitation followed by immunoblotting for the indicated proteins. (C) Binding of NPC1 to the mTORC1 scaffolding complex. HEK- 293T cells stably expressing LAMP1-FLAG or NPC1-FLAG were lysed and subjected to FLAG immunoprecipitation and immunoblotting for the indicated proteins. (D) HEK-293T cells stably expressing LAMP1-FLAG or FLAG-SLC38A9 were lysed and subjected to FLAG immunoprecipitation (IP) and immunoblotting for the indicated proteins. (E) SLC38A9 is required for the interaction of NPC1 with the mTORC1 scaffolding complex. Control or SLC38A9-deleted HEK-293T cells stably expressing NPC1-FLAG were lysed and subjected to FLAG IP followed by immunoblotting for the indicated proteins. LAMP1-FLAG expressing HEK-293T cells were used as negative control.

2.4.6 Cholesterol and arginine signals are integrated via SLC38A9 to coordinately regulate mTORC1 activity The ability of SLC38A9 to promote mTORC1 activation in response to both arginine and cholesterol raises the question of how these two nutrient inputs are integrated and whether there is a hierarchy of nutrient-derived signals. To rule out the possibility that manipulating cellular cholesterol indirectly alters amino acid levels, we measured total intracellular amino acid levels upon starvation and refeeding with total amino acids or upon MCD-mediated sterol depletion followed by refeeding with LDL. As expected, amino acid levels change dramatically in response to amino acid starvation and restimulation, but cholesterol depletion and refeeding did not result in significant changes in amino acid levels (Fig. 2.8A). Thus, cholesterol sensing can be considered independent from amino acid sensing. To further investigate whether cholesterol depletion and refeeding synergizes with amino acid starvation and restimulation, double starvation experiments for cholesterol and either arginine or leucine were performed in HEK-293T cells. The combined starvation for cholesterol and arginine suppressed signaling more strongly than either of the single nutrient starvations (Fig. 2.8B). Importantly, the cholesterol and arginine signals appear to be additive (Fig. 2.8B), and both are significantly reduced by loss of SLC38A9 (Fig. 2.8D). Conversely, neither LDL nor leucine alone significantly activates mTORC1 following double starvation, suggesting these two nutrients act in separate pathways (Fig. 2.8C). Overall, the results are consistent with our model in which arginine and cholesterol act synergistically and converge on a common effector, SLC38A9, upstream of the Rag GTPases. In the absence of detailed structural information, it is challenging to predict precisely how cholesterol and arginine signals are coordinated via SLC38A9. Previous work has established that overexpression of the soluble N-terminal domain (NTD, amino acids 1- 119) or full-length SLC38A9 is sufficient to render mTORC1 insensitive to amino acid starvation and that several residues in the NTD mediate interactions with the mTORC1 scaffolding complex (Rebsamen et al. 2015, Wang et al. 2015). We found that overexpression of full-length SLC38A9 or the NTD alone strongly boosted mTORC1

35 signaling upon stimulation with LDL, an effect that was not seen with SLC38A9.2, a shorter isoform that lacks the NTD (Fig. 2.8E). Therefore, we conclude that the NTD of SLC38A9 is critical for signals from both cholesterol and arginine to be transmitted to the Rag GTPases.

Figure 2.8 Coordination of arginine and cholesterol signaling by SLC38A9 (A) Cholesterol starvation/refeeding does not alter intracellular amino acid levels. HEK- 293T cells were subjected to starvation and refeeding with amino acids (red bars) or by MCD-mediated sterol depletion followed by refeeding with LDL (grey bars). Following cell lysis, amino acid levels were measured using single reaction monitoring (SRM)- based LC-MS/MS as in (Louie et al. 2016). Metabolites were quantified by integrating the area under the curve for individual SRMs and normalized to internal standard levels and external standard curves. Shown are fold changes relative to starved condition. Each bar represents the mean + SD from n=5 independent samples per condition. (B) HEK-293T cells were starved for both arginine and cholesterol, followed by restimulation with either nutrient individually or with both combined. Cell lysates were analyzed for phosphorylation status of S6K1 (T389) and 4E-BP1 (S65) and for total protein levels. (C) HEK-293T cells were starved for both leucine and cholesterol, followed by restimulation with either nutrient individually or with both combined. Cell lysates were analyzed for phosphorylation status of S6K1 (T389) and 4E-BP1 (S65) and for total protein levels. (D) Control HEK- 293T cells, or HEK-293T cells deleted for SLC38A9 via CRISPR/Cas9 were subjected to starvation for both arginine and

36 Figure 2.8 Coordination of arginine and cholesterol signaling by SLC38A9 (legend continued from previous page) cholesterol followed by restimulation with either nutrient individually or with both combined. Cell lysates were analyzed for phosphorylation status of S6K1 (T389) and 4E-BP1 (S65) and for total protein levels. (E) Overexpression of SLC38A9.1 (full-length) and its 1-119 AA N-terminal fragment, but not SLC38A9.2 (which lacks the N-terminal domain) enhances mTORC1 response to cholesterol. HEK-293T cells overexpressing the indicated FLAG-tagged SLC38A9 constructs were cholesterol depleted, followed by restimulation with LDL, where indicated. Cell lysates were analyzed for phosphorylation status of S6K (T389) and 4E-BP1 (S65) and for total protein levels.

2.5 Discussion

2.5.1 Role for SLC38A9 and NPC1 in cholesterol-mediated activation of mTORC1 This work establishes a role for lysosomal cholesterol in cell-autonomous activation of mTORC1. In particular, cholesterol delivered to the lysosome by endocytosed LDL particles induces the recruitment of mTORC1 to the lysosomal surface. Combining in vitro and cell-based experiments, we demonstrate that lysosomal cholesterol levels modulate the activation status of the Rag GTPases. SLC38A9, the lysosomal amino acid transporter that is involved in mTORC1 activation by arginine, harbors putative cholesterol-interacting motifs in its transmembrane helix 8, known as the CRAC and CARC motifs (Fantini and Barrantes, 2013). These motifs mediate the specific interaction of SLC38A9 with cholesterol, and were shown to be required for mTORC1 activation by cholesterol, but not arginine. The molecular details of how SLC38A9 coordinates signals from both arginine and cholesterol and transmits this information to the Rag GTPases remains an open question. It was recently shown that a single point mutation of SLC38A9 which blocks its interaction with arginine (T133W) fails to rescue mTORC1 signaling in response to arginine, but does not alter the interaction between SLC38A9 and the mTORC1 scaffolding complex (Wyant et al. 2017). Moreover, a crystal structure of zebrafish SLC38A9 bound to arginine revealed that transmembrane domain 1 mediates arginine binding; however, the relationship between arginine binding and the conformation of the N-terminal domain (NTD) could not be examined, as this study used a truncated construct lacking the NTD (Lei et al. 2018). More detailed studies will be required to understand precisely how the conformation of SLC38A9 relates to its ability to promote the active nucleotide state of the Rag GTPases. Furthermore, we found that SLC38A9 directly interacts with NPC1, the major lysosomal cholesterol exporter. NPC1 negatively regulates mTORC1 by reducing the available pool of cholesterol in the lysosomal limiting membrane. Thus, NPC1 is required for mTORC1 to sense changes in dietary lipid supply, as its deletion results in mTORC1 signaling that is insensitive to cholesterol depletion. Consistent with this notion, NPC1 is a risk allele for early-onset obesity in humans, and NPC1 haploinsufficiency in mice has been associated with weight gain and insulin resistance, which are hallmarks of mTORC1 hyperactivation (Jelinek et al. 2010, Meyre et al. 2009). Importantly, this model implies that dysregulated mTORC1 may directly contribute to the metabolic derangement observed in Niemann-Pick type C disease. How this pathway for cell-autonomous cholesterol sensing plays out in a whole

37 organism context, and whether it may take up tissue-specific and organ-specific roles remains to be determined.

Figure 2.9 Model for mTORC1 regulation by LDL-derived cholesterol SLC38A9 stimulates Rag GTPase activation and subsequent recruitment of mTORC1 to the lysosomal surface in response to cholesterol. NPC1 binds to SLC38A9 and inhibits cholesterol-mediated mTORC1 activation via its sterol export activity.

2.5.2 Regulatory feedback loop of mTORC1 activity by other lipid species In addition to cholesterol, there is evidence for cell-autonomous regulation of mTORC1 by phospholipids, particularly phosphatidic acid (PA) (Fang et al. 2001, Foster et al. 2014, Yoon et al. 2011). PA was shown to activate mTORC1 signaling in multiple cell types, and knocking down the main enzyme responsible for PA synthesis, phospholipase D (PLD) 1, suppressed mitogen-induced mTORC1 activation (Fang et al. 2001, Sun et al. 2008). Mechanistically, PA was proposed to bind to the FKBP-

38 Rapamycin Binding (FRB) domain of mTORC1, which is immediately N-terminal to the kinase domain and undergoes rapamycin-induced heterodimerization with FKBP-12 (Fang et al. 2001). How PA binding to the FRB domain affects mTORC1 kinase activity remains unclear. PLD1 was shown to localize to endosomes and lysosomes in a nutrient-regulated manner (Dall'Armi et al. 2010, Yoon et al. 2011). Thus, it was proposed that local production of PA at the lysosomal surface may synergize with other nutrient inputs, primarily amino acids, in activating mTORC1 locally. However, a caveat to this model is the observation that PLD1-/- mice are viable and do not display a phenotype consistent with reduced mTORC1 activation (Elvers et al. 2010). Signaling phospholipids known as phosphoinositides play established roles in mTORC1 regulation. Binding of insulin to the insulin receptor at the plasma membrane triggers the activation of type-I PI3K, which phosphorylates the 3-OH position in the inositol ring of PI(4,5)P2 to generate PI(3,4,5)P3. Production of PI(3,4,5)P3 on the inner leaflet of the plasma membrane is a key event for activation of Akt, which then phosphorylates and suppresses TSC, resulting in Rheb activation and triggering of mTORC1 kinase function at the lysosome (reviewed in Saxton and Sabatini 2017). Additional phosphoinositides found on endomembranes have been linked to mTORC1 regulation, including phosphatidylinositol 3-phosphate (PI3P) (the product of the type-III PI3-Kinase, the human orthologue of yeast Vps34) and phosphatidylinositol 3,4- bisphosphate [PI3,4)P2], generated from phosphorylation of PI4P by class II PI3K β (PI3KC2β) (Gulati et al. 2008, Marat et al. 2017). Overall, our understanding of how information about cellular lipid availability feeds back on mTORC1 remains incomplete. Further studies are needed to elucidate the biochemical mechanisms as well as the physiological context of lipid sensing by mTORC1. The evidence gathered so far sets the stage for exciting future work spanning reconstituted systems, functional studies in cells and genetic and metabolic analysis in mice, which will shed light on fundamental rules of lipid homeostasis.

2.5.3 Disruption of lysosomal lipid homeostasis in disease and aging Loss of the degradative capacity of the lysosome, which occurs over time or as a consequence of gene mutations, is increasingly viewed as a driving force in metabolic derangement and neurodegeneration that characterize aging and certain inherited disorders. Loss of hydrolytic or lysosomal transport ability causes aberrant accumulation of macromolecules within the lumen, leading to broad functional impairment of trafficking, signaling, and quality control pathways (reviewed in Ballabio and Gieselmann 2009). A majority of lysosomal storage disorders involve the aberrant accumulation of lipid species including sphingolipids (Niemann-Pick types A and B, Gaucher, Sandhoff, Krabbe, Fabry), mucolipids (collectively known as mucolipidoses), and sterols (Niemann-Pick type C, Wolman, and cholesterol ester storage disease). Accumulation of these lipid species impairs the overall degradative capacity of the lysosome and alters its physiology leading to aberrant size, trafficking, and attenuated autophagy (Ballabio and Gieselmann 2009, Cox and Cachon-Gonzalez 2012, Platt et al. 2012). These functional deficiencies impact multiple cell types, particularly neurons resulting in the neurodegenerative phenotypes associated with a majority of storage diseases.

39 Often accumulation of a primary lipid species occurs with concomitant buildup of secondary lipids. For example, in Niemann-Pick type C, cholesterol build up occurs along with oxysterols, sphingomyelin, gangliosides GM3 and GM2, lysobisphosphatidic acid (LBPA), and sphingosine (Kobayashi et al. 1999, Lloyd-Evans et al. 2008). Collectively, these effects lead to further functional impairment, which goes beyond the lysosome. In cells lacking NPC1, mitochondria also show increased cholesterol content, possibly through the sterol transfer activity of MLN64 (Balboa et al. 2017, Kennedy et al. 2014). In turn, mitochondrial cholesterol accumulation led to an increase in glycolytic flux and mitochondrial oxidative stress, linking aberrant lipid composition and transfer to metabolic alterations. More generally, there is evidence that lysosomal storage disorders, particularly those involving lipids, result in severe mitochondrial dysfunction including their altered morphology, calcium signaling, redox imbalance, and altered TCA cycle flux (for a comprehensive review see Plotegher and Duchen 2017). A major consequence of lipid accumulation within the lysosome is the disruption of autophagy (Settembre et al. 2008). For example, increased levels of cholesterol in Niemann-Pick type C is thought to disrupt the function of Rab GTPases that control delivery of hydrolases to the lysosome, potentially reducing its degradative capacity (Ganley and Pfeffer 2006). NPC lysosomes display aberrant trafficking, due at least in part to the increased association with the minus end directed microtubule motor dynein- dynactin (Rocha et al. 2009). Moreover, alterations in membrane composition of NPC lysosomes may impair the function of the soluble N-ethylmaleimide-sensitive factor protein (SNAP) receptors (SNAREs), which are key components of vesicle fusion machinery resulting in defective fusion ability of lysosomes with autophagosomes (Fraldi et al. 2010, Sarkar et al. 2013). Loss of lysosomal function is emerging as a driving force in age-related degeneration of cells and tissues (Colacurcio and Nixon 2016, Hughes and Gottschling 2012). A time dependent decrease in autophagic flux progressively impairs cellular homeostasis, as evidenced by accumulation of aggregated proteins and extensive mitochondrial damage observed in brain tissue from healthy aged subjects. Electron micrographs from aged humans as well as mice show that a significant percentage of lysosomes appear to be filled with insoluble lipid precipitates known as lipofuscins. These are the insoluble residues left by multiple rounds of fusion and degradation, and are thought to impair the catabolic capacity of the lysosome they reside in. The accumulation of lipofuscin filled lysosomes is especially evident in non-dividing cells such as neurons, which are critically dependent on the autophagic-lysosomal system for their function and survival (Hara et al. 2006). Interestingly, a class of lipid storage disorders known as neuronal ceroid lipofuscinoses (NCL) display massive accumulation of lipofuscin-like material inside of lysosomes at an early age and are characterized by severe neurodegeneration (Carcel- Trullols et al. 2015). Therefore, some of the pathophysiological mechanisms that underlie lysosomal storage disorders might occur during normal aging albeit at a slower rate. Multiple lines of evidence from organisms spanning from yeast to higher mammals strongly suggests that maintenance of lysosomal function is key for lifespan and healthy aging, or healthspan. In addition to maintenance of lipid homeostasis, there is evidence that lipid processing and trafficking within the lysosome may generate signals that control lifespan in the roundworm C. elegans (Folick et al. 2015).

40 Although many storage diseases are linked to impairment of lysosomal function, enhanced lysosomal activity may also promote disease progression. For example, in the case of KRAS driven cancers, generation of nutrients through lysosomal processing may confer a selective advantage under conditions of nutrient scarcity (Perera and Zoncu 2016). Recently, it has been shown that this enhanced lysosomal activity relies on specific transcriptional programs. For example, in pancreatic ductal adenocarcinoma (PDA), the MiT/TFE transcription factors evade regulation by mTORC1 and become constitutively localized to the nucleus where they drive enhanced lysosomal biogenesis (Perera et al. 2015). This adaptation enables scavenging of energy rich nutrients, including lipids, which are delivered to the lysosome through macropinocytosis and autophagy, driving cellular growth and survival (Commisso et al. 2013, Mancias and Kimmelman 2011). Interestingly, concomitant upregulation of ACAT1 in pancreatic cancer may enable storage of LDL-derived cholesterol in lipid droplets and promote tumor growth and metastasis (Li et al. 2016a). This mechanism may be important not only for cancer growth, but also in the context of resistance to therapy. In mouse models where mutant KRAS can be inducibly ablated the appearance of resistant cells relies on enhanced lysosomal function, lipid droplet accumulation, and lipid catabolism (Viale et al. 2014). Thus, a deeper understanding of lysosomal function, both in normal and disease states, may inform new approaches for preventing and treating cancer and age- related conditions.

2.5.4 Outstanding questions and future directions Lysosomes promote lipid catabolism and transport, both of which are critical in maintaining cellular lipid homeostasis. Recent work has established that the lysosome is a hub for nutrient sensing and signaling, and it is likely that the lipid sensing and trafficking functions of the lysosome are intimately linked. In particular, organelle contact sites may mediate the transfer of information between subcellular compartments, each with specialized roles such as storage of lipids in lipid droplets or beta-oxidation of lipids in mitochondria and peroxisomes. Organelle crosstalk via membrane contact sites is a rapidly progressing area of research with many outstanding questions, and future work may shed light on the role of various lipid species in supporting the function of the organelle in which they reside. In the case of lysosomes, mTORC1 acts as a sensor of lysosomal lipids, although it remains to be seen whether other species beside cholesterol and phosphatidic acid can influence mTORC1 signaling. Interestingly, mTORC2, a distinct protein complex that shares the same core kinase subunit with mTORC1, senses sphingolipid levels at the plasma membrane, and plays a key role in maintaining sphingolipid homeostasis in response to a variety of stressors (Berchtold et al. 2012, Muir et al. 2014). Importantly, aberrant mTORC1 signaling may contribute to the pathogenesis of lysosomal disorders characterized by abnormal lipid storage, and thus targeting mTORC1 may prove beneficial in these disease settings. Clarifying the role of mTORC1 in maintaining lipid homeostasis may increase our understanding of processes where lipids are key building blocks, such as cell growth, proliferation and differentiation processes that require expansion of membranes, as well as energy storage. Furthermore, it will be interesting to compare how these lipid-sensing and lipid-trafficking networks acquire specialized functions in different tissues, according to specific metabolic roles.

Chapter 3

Structural Characterization of the mTORC1 Lysosomal Scaffolding Complex

Ashley Thelen and Adam Yokom conceived the study and designed experiments with input from Roberto Zoncu and James Hurley. Ashley Thelen performed protein expression and purification. Adam Yokom performed electron microscopy, data processing, and model building.

42 3.1 Chapter Summary and Introduction

The AND-gate model of mTORC1 activation in the presence of nutrients and growth factors critically depends on the ability of the lysosomal scaffolding complex (composed of Ragulator and the Rag GTPases) to recruit mTORC1. Despite extensive efforts, we still lack a detailed mechanistic understanding of how this process is orchestrated. A surge of recent structural studies have begun to shed light on the question, including a hybrid crystallography/electron microscopy (EM) structural model of the Ragulator:Rag GTPase complex as well as a high-resolution cryo-EM structure of the GATOR1:Rag GTPase complex (Su et al. 2017, de Araujo et al. 2017, Yonehara et al. 2017, Zhang et al. 2017, Shen et al. 2018). Importantly, this work revealed that the orientation of the Rag GTPase nucleotide binding G domains project away from the Ragulator scaffold, a surprising finding given the current model that Ragulator acts as a guanine nucleotide exchange factor (GEF) for RagA/B (Su et al. 2017, de Araujo et al. 2017, Yonehara et al. 2017, Bar-Peled et al. 2012). The accessibility of the G domains supports their ability to interact with the corresponding GTPase activating proteins (GAPs) for RagA/B and RagC/D, GATOR1 and Folliculin/FNIP2, respectively (Bar-Peled et al. 2013, Panchaud et al. 2013, Peli-Gulli et al. 2015, Petit et al. 2013, Tsun et al. 2013, Shen et al. 2018). Although the spatial and temporal regulation of the GEF/GAP cycles for the Rag GTPases is not yet fully understood, the discovery of KICSTOR, a lysosomal scaffolding complex for GATOR1, suggests that GATOR1 likely promotes RagA/B GTP hydrolysis at the lysosomal surface (Peng et al. 2017, Wolfson et al. 2017). In addition to the implications for GEF and GAP proteins, the accessibility of the Rag GTPase G domains suggests that nucleotide-associated conformational changes may provide a direct structural readout for mTORC1 docking on the lysosome. A series of cryo-EM studies have provided a wealth of structural information about mTORC1 and its activation by the small GTPase Rheb; however, we currently do not know how mTORC1 associates with the lysosomal scaffolding complex and whether this process is gated by the coordinated efforts of other protein factors (Yip et al. 2010, Yang et al. 2013, Aylett et al. 2016, Baretic et al. 2016, Yang et al. 2016, Yang et al. 2017). Moreover, we know even less about the mechanism of mTORC1 disengagement from the lysosome, a program that must be faithfully executed in order to avoid mTORC1 hyperactivation. For example, recurrent mutations in RagC have been reported in follicular lymphoma, resulting in amino acid insensitive increases in Raptor binding to the Rag GTPases and constitutively lysosomal mTORC1 localization (Okosun et al. 2016, Ying et al. 2016, Lawrence et al. 2018). Thus, the ability to selectively regulate mTORC1 association with the lysosome may provide a powerful therapeutic tool for controlling aberrant mTORC1 activation associated with disease. Here, we show purification of the Ragulator:RagA/C complex, loaded in vitro with nucleotides to reflect the active state, RagAGTP/RagCXDP. A point mutation in RagC, T90N, was incorporated in order to stabilize interactions between Ragulator and the Rag GTPases. In collaboration with Adam Yokom, a postdoctoral researcher in the Hurley Lab, the purified complex was used for structural determination via cryo-electron microscopy (cryo-EM). We present an ~8 Å cryo-EM structure of the full-length, active state Ragulator:RagA/CT90N complex, which improves upon the earlier Ragulator:Rag

43 GTPase by providing detailed structural information for the G domains of the Rag GTPases.

3.2 Methods

3.2.1 Expression and purification of Ragulator from Sf9 cells Expression and purification of human Ragulator from Spodoptera frugiperda (Sf9) cells was performed similarly to previously published methods in Su et al. 2017. The pFastBac-GST-TEV-Lamtor1/His6-Lamtor2/Lamtor3/Lamtor4/Lamtor5 construct used for purification was obtained from the Hurley Lab. Lamtor1-5 were co-expressed in Spodoptera frugiperda (Sf9) insect cells using ESF 921 Protein Free Insect Cell Culture Medium (Expression Systems). Baculoviruses were generated in Sf9 cells with the bac-to-bac system (Life Technologies). Sf9 cells were harvested after 48 to 72 hours based on cell viability. Cells were pelleted by centrifugation at 2000 x g for 20 minutes at 4C. Cells were lysed for 30 min at 4C in pH 7.4 Triton Lysis Buffer containing the following: 130mM NaCl, 2.5mM MgCl2, 1% (v/v) Triton X-100, 25mM HEPES, 2mM EGTA, supplemented with 0.5mM TCEP-HCl and protease inhibitors (Pierce, Thermo Fisher Scientific). Lysate was centrifuged at 15,000 x g for 1 hour at 4C. Cleared lysate was added to pre-washed Ni-NTA resin and incubated for 1.5 hours at 4C. Protein was washed 4 times with 3 column volumes pH 8.0 Wash Buffer (50mM NaH2PO4, 300mM NaCl, 20mM imidazole, 0.5mM TCEP-HCl). Bound protein was eluted with 1 column volume Elution Buffer (Wash Buffer supplemented with imidazole to a final concentration of 250mM) after a 10 minute incubation at room temperature. Elution step was repeated 3 times and eluate was pooled and applied to pre-washed glutathione sepharose 4B (GE healthcare) in a gravity-flow column and incubated for 2 hours at 4C. Protein was washed 3 times with 3 column volumes Triton Lysis Buffer supplemented with 0.5mM TCEP-HCl. The last wash was supplemented with NaCl to a final concentration of 330mM and incubated at 4C for 30 minutes before eluting. Protein was eluted from the resin overnight at 4C in Wash Buffer and 0.25mg Tobacco Etch Virus (TEV) protease. Flow-through was collected and concentrated to 500uL using a 30kDa MWCO Amicon centrifugal filter (Millipore) before loading onto a Superdex 200 10/300 GL column (GE Healthcare) equilibrated in Wash Buffer. Eluted fractions containing Ragulator were pooled and flash frozen in liquid N2 for storage.

3.2.2 Expression and purification of the RagA/C heterodimer from Sf9 cells Expression and purification of human RagA/C GTPases from Sf9 cells was performed similarly to previously published methods in Su et al. 2017. The pFastBac- MBP-TEV-RagCD181N/RagAWT and pFastBac-MBP-TEV-RagCT90N/D181N/RagAWT constructs used for purification were obtained from the Hurley Lab. RagA and RagC constructs were co-expressed in Sf9 insect cells using the identical conditions as described for Ragulator purification. Pelleted Sf9 cells were lysed for 30 min at 4C in pH 7.4 Triton Lysis Buffer (same as above) supplemented with 0.5mM TCEP-HCl and protease inhibitors. Lysate was cleared by centrifugation at 15,000 x g for 1 hour at 4C. Cleared lysate was added to pre-washed amylose resin (New England Biosciences) in a gravity flow column and incubated for 2 hours at 4C. Protein was

44 washed 4 times with 3 column volumes pH 7.4 Wash Buffer containing the following: 130mM NaCl, 2.5mM MgCl2, 25mM HEPES, 2mM EGTA supplemented with 0.5mM TCEP-HCl. For the final wash, Wash Buffer was supplemented with NaCl to a final concentration of 330mM, and incubated for 30 minutes at 4C. Protein was eluted by incubation with 0.25mg TEV protease overnight at 4C. Flow-through was collected and concentrated to 500uL using a 30kDa MWCO Amicon centrifugal filter before loading onto a Superdex 200 10/300 GL column equilibrated in Wash Buffer. Eluted fractions containing RagA/C were pooled and flash frozen in liquid N2 for storage.

3.2.3 Nucleotide loading of the RagA/C heterodimer Purified RagAWT/RagCD181N heterodimeric complex was stripped of nucleotides by incubating with 10mM EDTA and 0.5mM TCEP in DPBS (without calcium or magnesium, Gibco) for 20 minutes at 25C with gentle rotating. The desired nucleotides (in this case, GTP and XTP) were each added in 10-fold molar excess relative to the Rag GTPase complex and incubated for 30 minutes at 25C. MgCl2 was added to a final concentration of 20mM and incubated for 30 minutes at 25C. Excess nucleotide was removed by buffer exchange into Wash Buffer (25mM HEPES pH7.4, 130mM NaCl, 2.5mM MgCl2, 2mM EGTA, 0.5mM TCEP-HCl) using a 1.5mL 30kDa MWCO Amicon centrifugal filter (Millipore). In order to convert XTP to XDP for a final heterodimeric complex containing RagAGTP/RagCXDP, the Rag GTPase complex was incubated with purified recombinant RagC GAP complex (FLCN:FNIP2, generously provided by Rosalie Lawrence) at a 100:1 molar ratio of RagA/C:FLCN/FNIP2 for 1 hour at 37C.

3.2.4 Gel filtration of Ragulator:RagA/RagCT90N complex Purified RagAWT/RagCT90N/D181N was re-loaded with nucleotides to form the active state of the heterodimeric complex (RagAGTP/RagCXDP). Purified Ragulator was thawed and combined in an equimolar ratio with RagA/C and incubated for 2 hours at 4C. The complex was loaded onto a Superdex 200 10/300 GL column pre-equilibrated with Gel Filtration Buffer (50mM HEPES, pH 7.4, 150mM NaCl, 0.5mM TCEP-HCl, 0.1% CHAPS). Fractions containing the Ragulator:RagA/C complex were pooled and concentrated using a 1.5mL 30kDa MWCO Amicon centrifugal filter and used immediately without freezing.

3.3 Results

3.3.1 Reconstitution of the active Ragulator:RagA/CT90N complex To further understand the molecular mechanism by which the Ragulator:Rag GTPase complex binds and recruits mTORC1, we sought to purify the 150kDa Ragulator:RagA/C complex in the active nucleotide state and use cryo-EM to elucidate the structure. Given the conformational rearrangements associated with various nucleotide states, it was important to ensure the Rag GTPase complexes were homogenously loaded with the desired diphosphate or triphosphate nucleotides. To achieve this, a point mutation was introduced into RagC, D181N, which alters the nucleotide specificity such that RagC preferentially binds xanthosine nucleotides (Bar- Peled et al. 2013a). Moreover, previous work from our lab has shown that the active nucleotide state of the Rag GTPases promotes dissociation from the Ragulator scaffold

45 and release into the cytoplasm; thus acting as a limiting factor for mTORC1 recruitment to the lysosome (Lawrence et al. 2018). In order to stabilize Ragulator:Rag GTPase association, we incorporated a second point mutation in RagC, T90N, which was initially identified as a recurring mutation in follicular lymphoma and has been shown to dramatically reduce Rag GTPase cycling off of the lysosome in nutrient replete conditions (Okosun et al. 2016, Ying et al. 2016, Lawrence et al. 2018). Full-length human RagAWT/RagCT90N/181N (hereafter referred to simply as RagA/RagCT90N) was co-expressed and purified from Spodoptera frugiperda (Sf9) cells by amylose affinity and size exclusion chromatography (SEC) (Fig. 3.1A, C blue trace). Following purification, the Rag GTPase heterodimer was stripped of bound nucleotides, and subsequently reloaded with GTP and XTP in vitro resulting in a RagAGTP/RagCXTP complex. Next, purified Folliculin/FNIP2 complex, a RagC-specific GAP, was used to convert XTP to XDP, thus arriving at the fully active RagAGTP/RagCXDP complex. Correct nucleotide loading was validated by comparing to nucleotide standards on HPLC as demonstrated in Su et al. 2017 (data not shown).

Figure 3.1 Reconstitution of the active Ragulator:RagA/CT90N complex (A) SDS-PAGE of purified RagAWT/RagCT90N fractions from size exclusion column. Asterisks indicate unknown protein contaminants. (B) SDS-PAGE of purified Ragulator fractions from size exclusion column. (C) Purification of the active Ragulator:RagA/CT90N complex. Superdex 200 10/300 size exclusion column showing absorbance at 280nm for given elution volumes. RagA/CT90N purification (blue trace) is shown as overlay with Ragulator:RagA/CT90N (orange trace).

46 Full-length Ragulator was co-expressed and purified from Sf9 cells (Fig. 3.1B) and incubated with the active RagA/CT90N heterodimer in an equimolar ratio. The active Ragulator:RagA/CT90N complex was purified by SEC (Fig. 3.1C, orange trace). The final Ragulator:RagA/C complex yield was more than doubled by incorporation of the RagCT90N mutation (relative to RagCWT, SEC data not shown), in agreement with increased RagA/C occupancy on Ragulator scaffolds (Lawrence et al. 2018).

3.3.2 Structural characterization of the active Ragulator:RagA/CT90N complex Previously reported models of the Ragulator:RagA/C complex were either high- resolution crystal structures lacking the G domains of the Rag GTPases or low- resolution negative stain EM data modeled using the yeast homologs Gtr1/2 in place of the Rag GTPases (Su et al. 2017, de Araujo et al. 2017, Yonehara et al. 2017, Zhang et al. 2017). Thus, we sought to determine the structure of the full-length human Ragulator:RagA/CT90N complex to elucidate the conformation of the active Rag GTPase G domains. Diluted Ragulator:RagA/CT90N was incubated on glow discharged continuous carbon grids and stained with 1% uranyl formate. Negative stain EM images were collected, representing approximately 16,000 discrete particles with views from several different 2D orientations (Fig. 3.2A,B). The sample contained a relatively small population of unbound RagA/CT90N heterodimer, suggesting a homogenous sample amenable to providing high-quality structural data.

Figure 3.2 Negative stain EM of the active Ragulator:RagA/CT90N complex (A) Raw negative stain EM image of active Ragulator:RagA/CT90N complex. Discrete particles were clearly visualized under the microscope. (B) 2D-clustering of active Ragulator:RagA/CT90N complex particles. Red arrows indicate gap between Rag GTPases.

Next, we analyzed the active Ragulator:RagA/CT90N complex by cryo-EM. The sample was well behaved in vitreous ice and maintained a monodisperse particle spread (Fig. 3.3A). Well-defined particles were clearly visualized after 2D classification revealing distinct structural features and multiple different orientations (Fig. 3.3B).

47 Density corresponding to Rag GTPase G domains show a distinct gap between the two domains. Furthermore, 2D class averages resemble the negative stain EM 2D classes suggesting vitrification did not disrupt the structure. After iterative 2D and 3D classification, the initial 533,000 were pruned to a final 76,000 single particle data set. 3D refinement was performed and the final reconstruction reached a resolution of 8.1 Å (Fig. 3.3C). Structures of Ragulator and the RagA/RagCS75N heterodimer were rigid body docked into the electron density using UCSF Chimera ‘fit in map’ function (Fig. 3.3C) (Yonehara et al. 2017, Shen et al. 2018). At this resolution, no additional modeling was performed. Notably, our structure is missing electron density for the Lamtor1 α2 helix (amino acid residues 77-96), but contains extra density in RagC, representing the missing N-terminal residues from the fitted structure (amino acid residues 1-60). This extra density is distinctly flexible and can be seen as ‘fuzzy’ density in the 2D class averages.

Figure 3.3 Low-resolution cryo-EM structure of the active Ragulator:RagA/CT90N complex (A) Raw cryo-EM image of active Ragulator:RagA/CT90N complex. (B) 2D classification of active Ragulator:RagA/CT90N complex particles. (C) Electron density envelopes from the final 3D reconstruction. (D) Ragulator (5X6V) and RagA/C (6CES) fit to 3D reconstructed volume. Single asterisk indicates missing density for Lamtor1 α2 helix. Double asterisk indicates density for missing RagC N-terminal residues 1-60.

48 3.4 Discussion

In this study, we determined the cryo-EM structure of the full-length human Ragulator:RagA/CT90N complex in the active nucleotide state. This is the first structure to reveal the nucleotide binding G domains of the Rag GTPases in the context of their lysosomal scaffold, Ragulator. Prior to the completion of this work, the first high- resolution cryo-EM structure of the Rag GTPases was reported in complex with the RagA/B GAP, GATOR1 (Shen et al. 2018). Interestingly, we did not detect differences in orientation of the G domains upon comparison of our active structure with their pseudo-active structure, containing GTP bound RagA and nucleotide empty RagCS75N. This suggests that binding of Ragulator to the Rag GTPases through the C-terminal roadblock dimerization domains does not influence the conformation adopted by the G domains in the active state. It remains unclear whether the mTORC1 binding interface includes Ragulator, which could influence the ability of the free Rag GTPase heterodimer to bind mTORC1 in the cytosol following Rag GTPase release from Ragulator (Lawrence et al. 2018). Structural determination of the reconstituted Ragulator:RagA/C:mTORC1 supercomplex is clearly the next step in understanding the molecular details of mTORC1 recruitment to the lysosome. There are several mechanistic questions regarding the spatial and temporal dynamics of Rag GTPase activation that remain unanswered. Although we now have a snapshot of the active state, the accompanying structure of the inactive state would provide a more sophisticated understanding of the conformational changes associated with the nucleotide state of the Rag GTPases. Notably, in the S. cerevisiae homologs Gtr1 and Gtr2, the G domains undergo significant reorientation relative to the C-terminal roadblock domains upon GTP hydrolysis, impacting Raptor binding (Gong et al. 2011, Jeong et al. 2012). Thus, it is likely that the specific orientation of the G domains mediate the switch-like ability of the Rag GTPases to bind mTORC1. The Rag GTPases are highly unusual in that they exist as a heterodimer and their nucleotide states are independently controlled through the coordinated action of specific GEFs and GAPs. This complicated system may have evolved to enable fine-tuning of mTORC1 activity based on the integrated sensing of several nutrient inputs. We do not yet fully understand whether there is a nutrient hierarchy or how these signals are influenced by the tissue type of any given cell. However, we do know that some nutrients act upstream of the Rag GTPases by synergistically influencing multiple regulatory nodes. For example, the presence of the amino acid arginine in the cytosol can promote RagAGTP through CASTOR1-dependent inhibition of the RagA/B-specific GAP GATOR1. In conjunction, arginine inside the lysosome can also promote RagAGTP through SLC38A9 via a mechanism that is unclear, but has been proposed to enhance GEF activity towards RagA (Shen and Sabatini 2018). Some nutrients such as cholesterol (discussed in detail in Chapter 2) have been shown to act upstream of the Rag GTPases through one route (in this case via SLC38A9), but this does not preclude the possibility that they could be acting through multiple upstream effectors. Moreover, certain nutrients such as glucose are thought to act upstream of the Rag GTPases, but a direct sensing mechanism has not yet been identified (Efeyan et al. 2013). Overall, significant progress has been made since the early 1990’s in our understanding of how mTORC1 regulates cell growth and proliferation in response to

49 nutrients and growth factors. However, major conceptual gaps still remain, which may require the development of new biochemical tools or more physiologically relevant animal models in order to gain deeper insights. Ultimately, leveraging our understanding of this critical pathway may translate to the development of therapeutic strategies for targeting metabolic dysfunction in diseases including cancer, diabetes, and neurodegeneration.

50 REFERENCES

Aylett CH, Sauer E, Imseng S, Boehringer D, Hall MN, Ban N, Maier T. 2016. Architecture of human mTOR complex 1. Science 351:48–52. Balboa E, Castro J, Pinochet M-J, Cancino GI, Matías N, Sáez PJ, Martínez A, Álvarez AR, Garcia-Ruiz C, Fernandez-Checa JC. 2017. MLN64 induces mitochondrial dysfunction associated with increased mitochondrial cholesterol content. Redox biology 12:274– 284. Ballabio A, Gieselmann V. 2009. Lysosomal disorders: from storage to cellular damage. Biochimica et Biophysica Acta (BBA)-Molecular Cell Research 1793:684–696. Barbosa AD, Siniossoglou S. 2017. Function of lipid droplet-organelle interactions in lipid homeostasis. Biochimica et Biophysica Acta (BBA)-Molecular Cell Research 1864:1459–1468. Baretić D, Berndt A, Ohashi Y, Johnson CM, Williams RL. 2016. Tor forms a dimer through an N-terminal helical solenoid with a complex topology. Nature communications 7:11016. Bar-Peled L, Chantranupong L, Cherniack AD, Chen WW, Ottina KA, Grabiner BC, Spear ED, Carter SL, Meyerson M, Sabatini DM. 2013. A Tumor suppressor complex with GAP activity for the Rag GTPases that signal amino acid sufficiency to mTORC1. Science 340:1100–1106. Bar-Peled L, Schweitzer LD, Zoncu R, Sabatini DM. 2012. Ragulator is a GEF for the rag GTPases that signal amino acid levels to mTORC1. Cell 150:1196–1208. Ben-Sahra I, Howell JJ, Asara JM, Manning BD. 2013. Stimulation of de novo pyrimidine synthesis by growth signaling through mTOR and S6K1. Science 339:1323–1328. Ben-Sahra I, Hoxhaj G, Ricoult SJ, Asara JM, Manning BD. 2016. mTORC1 induces purine synthesis through control of the mitochondrial tetrahydrofolate cycle. Science 351:728– 733. Berchtold D, Piccolis M, Chiaruttini N, Riezman I, Riezman H, Roux A, Walther TC, Loewith R. 2012. Plasma membrane stress induces relocalization of Slm proteins and activation of TORC2 to promote sphingolipid synthesis. Nature cell biology 14:542. Binda M, Péli-Gulli M-P, Bonfils G, Panchaud N, Urban J, Sturgill TW, Loewith R, De Virgilio C. 2009. The Vam6 GEF controls TORC1 by activating the EGO complex. Molecular cell 35:563–573. Brown AJ, Sun L, Feramisco JD, Brown MS, Goldstein JL. 2002. Cholesterol addition to ER membranes alters conformation of SCAP, the SREBP escort protein that regulates cholesterol metabolism. Molecular cell 10:237–245. Brown EJ, Albers MW, Shin TB, Keith CT, Lane WS, Schreiber SL. 1994. A mammalian protein targeted by G1-arresting rapamycin–receptor complex. Nature 369:756. Cárcel-Trullols J, Kovács AD, Pearce DA. 2015. Cell biology of the NCL proteins: What they do and don’t do. Biochimica et Biophysica Acta (BBA)-Molecular Basis of Disease 1852:2242–2255. Caron A, Richard D, Laplante M. 2015. The roles of mTOR complexes in lipid metabolism. Annual Review of Nutrition 35:321–348. Chang T-Y, Chang CC, Ohgami N, Yamauchi Y. 2006. Cholesterol sensing, trafficking, and esterification. Annu.Rev.Cell Dev.Biol. 22:129–157.

51 Chantranupong L, Scaria SM, Saxton RA, Gygi MP, Shen K, Wyant GA, Wang T, Harper JW, Gygi SP, Sabatini DM. 2016. The CASTOR proteins are arginine sensors for the mTORC1 pathway. Cell 165:153–164. Colacurcio DJ, Nixon RA. 2016. Disorders of lysosomal acidification—The emerging role of v-ATPase in aging and neurodegenerative disease. Ageing research reviews 32:75–88. Commisso C, Davidson SM, Soydaner-Azeloglu RG, Parker SJ, Kamphorst JJ, Hackett S, Grabocka E, Nofal M, Drebin JA, Thompson CB. 2013. Macropinocytosis of protein is an amino acid supply route in Ras-transformed cells. Nature 497:633. Cox TM, Cachón‐González MB. 2012. The cellular pathology of lysosomal diseases. The Journal of pathology 226:241–254. Dall’Armi C, Hurtado-Lorenzo A, Tian H, Morel E, Nezu A, Chan RB, Yu WH, Robinson KS, Yeku O, Small SA. 2010. The phospholipase D1 pathway modulates macroautophagy. Nature communications 1:142. Daniele T, Schiaffino MV. 2016. Lipid transfer and metabolism across the endolysosomal– mitochondrial boundary. Biochimica et Biophysica Acta (BBA)-Molecular and Cell Biology of Lipids 1861:880–894. Das A, Brown MS, Anderson DD, Goldstein JL, Radhakrishnan A. 2014. Three pools of plasma membrane cholesterol and their relation to cholesterol homeostasis. Elife 3:e02882. de Araujo ME, Naschberger A, Fürnrohr BG, Stasyk T, Dunzendorfer-Matt T, Lechner S, Welti S, Kremser L, Shivalingaiah G, Offterdinger M. 2017. Crystal structure of the human lysosomal mTORC1 scaffold complex and its impact on signaling. Science 358:377–381. Deffieu MS, Pfeffer SR. 2011. Niemann–Pick type C 1 function requires lumenal domain residues that mediate cholesterol-dependent NPC2 binding. Proceedings of the National Academy of Sciences 108:18932–18936. Derler I, Jardin I, Stathopulos PB, Muik M, Fahrner M, Zayats V, Pandey SK, Poteser M, Lackner B, Absolonova M. 2016. Cholesterol modulates Orai1 channel function. Sci.Signal. 9:ra10. Di Malta C, Siciliano D, Calcagni A, Monfregola J, Punzi S, Pastore N, Eastes AN, Davis O, De Cegli R, Zampelli A. 2017. Transcriptional activation of RagD GTPase controls mTORC1 and promotes cancer growth. Science 356:1188–1192. Dibble CC, Manning BD. 2013. Signal integration by mTORC1 coordinates nutrient input with biosynthetic output. Nature cell biology 15:555. Düvel K, Yecies JL, Menon S, Raman P, Lipovsky AI, Souza AL, Triantafellow E, Ma Q, Gorski R, Cleaver S. 2010. Activation of a metabolic gene regulatory network downstream of mTOR complex 1. Molecular cell 39:171–183. Efeyan A, Zoncu R, Chang S, Gumper I, Snitkin H, Wolfson RL, Kirak O, Sabatini DD, Sabatini DM. 2013. Regulation of mTORC1 by the Rag GTPases is necessary for neonatal autophagy and survival. Nature 493:679. Egan DF, Chun MG, Vamos M, Zou H, Rong J, Miller CJ, Lou HJ, Raveendra-Panickar D, Yang C-C, Sheffler DJ. 2015. Small molecule inhibition of the autophagy kinase ULK1 and identification of ULK1 substrates. Molecular cell 59:285–297. Elvers M, Stegner D, Hagedorn I, Kleinschnitz C, Braun A, Kuijpers ME, Boesl M, Chen Q, Heemskerk JW, Stoll G. 2010. Impaired αIIbβ3 integrin activation and shear-dependent thrombus formation in mice lacking phospholipase D1. Sci.Signal. 3:ra1.

52 Espenshade PJ, Hughes AL. 2007. Regulation of sterol synthesis in eukaryotes. Annu.Rev.Genet. 41:401–427. Fang Y, Vilella-Bach M, Bachmann R, Flanigan A, Chen J. 2001. Phosphatidic acid- mediated mitogenic activation of mTOR signaling. Science 294:1942–1945. Fantini J, Barrantes FJ. 2013. How cholesterol interacts with membrane proteins: an exploration of cholesterol-binding sites including CRAC, CARC, and tilted domains. Frontiers in physiology 4:31. Folick A, Oakley HD, Yu Y, Armstrong EH, Kumari M, Sanor L, Moore DD, Ortlund EA, Zechner R, Wang MC. 2015. Lysosomal signaling molecules regulate longevity in Caenorhabditis elegans. Science 347:83–86. Foster DA, Salloum D, Menon D, Frias MA. 2014. Phospholipase D and the maintenance of phosphatidic acid levels for regulation of mammalian target of rapamycin (mTOR). Journal of Biological Chemistry 289:22583–22588. Fraldi A, Annunziata F, Lombardi A, Kaiser H, Medina DL, Spampanato C, Fedele AO, Polishchuk R, Sorrentino NC, Simons K. 2010. Lysosomal fusion and SNARE function are impaired by cholesterol accumulation in lysosomal storage disorders. The EMBO journal 29:3607–3620. Ganley IG, Pfeffer SR. 2006. Cholesterol accumulation sequesters Rab9 and disrupts late endosome function in NPC1-deficient cells. Journal of Biological Chemistry 281:17890– 17899. Goldstein JL, Brown MS. 2015. A century of cholesterol and coronaries: from plaques to genes to statins. Cell 161:161–172. Gong R, Li L, Liu Y, Wang P, Yang H, Wang L, Cheng J, Guan K-L, Xu Y. 2011. Crystal structure of the Gtr1p–Gtr2p complex reveals new insights into the amino acid-induced TORC1 activation. Genes & development 25:1668–1673. Gong X, Qian H, Zhou X, Wu J, Wan T, Cao P, Huang W, Zhao X, Wang X, Wang P. 2016. Structural insights into the Niemann-Pick C1 (NPC1)-mediated cholesterol transfer and Ebola infection. Cell 165:1467–1478. Grabiner BC, Nardi V, Birsoy K, Possemato R, Shen K, Sinha S, Jordan A, Beck AH, Sabatini DM. 2014. A diverse array of cancer-associated MTOR mutations are hyperactivating and can predict rapamycin sensitivity. Cancer discovery 4:554–563. Gulati P, Gaspers LD, Dann SG, Joaquin M, Nobukuni T, Natt F, Kozma SC, Thomas AP, Thomas G. 2008. Amino acids activate mTOR complex 1 via Ca2 /CaM signaling to hVps34. Cell metabolism 7:456–465. Hara K, Maruki Y, Long X, Yoshino K, Oshiro N, Hidayat S, Tokunaga C, Avruch J, Yonezawa K. 2002. Raptor, a binding partner of target of rapamycin (TOR), mediates TOR action. Cell 110:177–189. Hara T, Nakamura K, Matsui M, Yamamoto A, Nakahara Y, Suzuki-Migishima R, Yokoyama M, Mishima K, Saito I, Okano H. 2006. Suppression of basal autophagy in neural cells causes neurodegenerative disease in mice. Nature 441:885. He C, Klionsky DJ. 2009. Regulation mechanisms and signaling pathways of autophagy. Annual Review of Genetics 43:67–93. Heitman J, Movva NR, Hall MN. 1991. Targets for cell cycle arrest by the immunosuppressant rapamycin in yeast. Science 253:905–909. Hughes AL, Gottschling DE. 2012. An early age increase in vacuolar pH limits mitochondrial function and lifespan in yeast. Nature 492:261.

53 Hurley JH, Schulman BA. 2014. Atomistic autophagy: the structures of cellular self- digestion. Cell 157:300–311. Ikonen E. 2008. Cellular cholesterol trafficking and compartmentalization. Nature reviews Molecular cell biology 9:125. Infante RE, Wang ML, Radhakrishnan A, Kwon HJ, Brown MS, Goldstein JL. 2008. NPC2 facilitates bidirectional transfer of cholesterol between NPC1 and lipid bilayers, a step in cholesterol egress from lysosomes. Proceedings of the National Academy of Sciences 105:15287–15292. Inoki K, Li Y, Xu T, Guan K-L. 2003. Rheb GTPase is a direct target of TSC2 GAP activity and regulates mTOR signaling. Genes & development 17:1829–1834. Jelinek D, Millward V, Birdi A, Trouard TP, Heidenreich RA, Garver WS. 2010. Npc1 haploinsufficiency promotes weight gain and metabolic features associated with insulin resistance. Human molecular genetics 20:312–321. Jeong J-H, Lee K-H, Kim Y-M, Kim D-H, Oh B-H, Kim Y-G. 2012. Crystal structure of the Gtr1pGTP-Gtr2pGDP protein complex reveals large structural rearrangements triggered by GTP-to-GDP conversion. Journal of Biological Chemistry 287:29648–29653. Jung J, Genau HM, Behrends C. 2015. Amino acid-dependent mTORC1 regulation by the lysosomal membrane protein SLC38A9. Molecular and cellular biology 35:2479–2494. Kalaany NY, Mangelsdorf DJ. 2006. LXRS and FXR: the yin and yang of cholesterol and fat metabolism. Annu.Rev.Physiol. 68:159–191. Kennedy BE, Madreiter CT, Vishnu N, Malli R, Graier WF, Karten B. 2014. Adaptations of energy metabolism associated with increased levels of mitochondrial cholesterol in Niemann-Pick type C1-deficient cells. Journal of Biological Chemistry 289:16278– 16289. Khamzina L, Veilleux A, Bergeron S, Marette A. 2005. Increased activation of the mammalian target of rapamycin pathway in liver and skeletal muscle of obese rats: possible involvement in obesity-linked insulin resistance. Endocrinology 146:1473– 1481. Kim D-H, Sarbassov DD, Ali SM, King JE, Latek RR, Erdjument-Bromage H, Tempst P, Sabatini DM. 2002. mTOR interacts with raptor to form a nutrient-sensitive complex that signals to the cell growth machinery. Cell 110:163–175. Kim D-H, Sarbassov DD, Ali SM, Latek RR, Guntur KV, Erdjument-Bromage H, Tempst P, Sabatini DM. 2003. GβL, a positive regulator of the rapamycin-sensitive pathway required for the nutrient-sensitive interaction between raptor and mTOR. Molecular cell 11:895–904. Kim E, Goraksha-Hicks P, Li L, Neufeld TP, Guan K-L. 2008. Regulation of TORC1 by Rag GTPases in nutrient response. Nature cell biology 10:935. Kim JE, Chen J. 2004. Regulation of peroxisome proliferator–activated receptor-γ activity by mammalian target of rapamycin and amino acids in adipogenesis. Diabetes 53:2748– 2756. Kobayashi T, Beuchat M-H, Lindsay M, Frias S, Palmiter RD, Sakuraba H, Parton RG, Gruenberg J. 1999. Late endosomal membranes rich in lysobisphosphatidic acid regulate cholesterol transport. Nature cell biology 1:113. Korsheninnikova E, Van Der Zon G, Voshol PJ, Janssen GM, Havekes LM, Grefhorst A, Kuipers F, Reijngoud D-J, Romijn JA, Ouwens DM. 2006. Sustained activation of the

54 mammalian target of rapamycin nutrient sensing pathway is associated with hepatic insulin resistance, but not with steatosis, in mice. Diabetologia 49:3049–3057. Kwon HJ, Abi-Mosleh L, Wang ML, Deisenhofer J, Goldstein JL, Brown MS, Infante RE. 2009. Structure of N-terminal domain of NPC1 reveals distinct subdomains for binding and transfer of cholesterol. Cell 137:1213–1224. Lahiri S, Toulmay A, Prinz WA. 2015. Membrane contact sites, gateways for lipid homeostasis. Current opinion in cell biology 33:82–87. Laplante M, Horvat S, Festuccia WT, Birsoy K, Prevorsek Z, Efeyan A, Sabatini DM. 2012. DEPTOR cell-autonomously promotes adipogenesis, and its expression is associated with obesity. Cell metabolism 16:202–212. Lawrence RE, Cho KF, Rappold R, Thrun A, Tofaute M, Moldavski O, Hurley JH, Zoncu R. 2018. A nutrient-induced affinity switch controls mTORC1 activation by its Rag GTPase–Ragulator lysosomal scaffold. Nature cell biology 20:1052. Lei H-T, Ma J, Martinez SS, Gonen T. 2018. Crystal structure of arginine-bound lysosomal transporter SLC38A9 in the cytosol-open state. Nat Struct Mol Biol 25:522–527. Li J, Pfeffer SR. 2016. Lysosomal membrane glycoproteins bind cholesterol and contribute to lysosomal cholesterol export. Elife 5:e21635. Li X, Saha P, Li J, Blobel G, Pfeffer SR. 2016. Clues to the mechanism of cholesterol transfer from the structure of NPC1 middle lumenal domain bound to NPC2. Proceedings of the National Academy of Sciences 113:10079–10084. Lim C-Y, Zoncu R. 2016. The lysosome as a command-and-control center for cellular metabolism. J Cell Biol 214:653–664. Lin MG, Hurley JH. 2016. Structure and function of the ULK1 complex in autophagy. Current opinion in cell biology 39:61–68. Lloyd-Evans E, Morgan AJ, He X, Smith DA, Elliot-Smith E, Sillence DJ, Churchill GC, Schuchman EH, Galione A, Platt FM. 2008. Niemann-Pick disease type C1 is a sphingosine storage disease that causes deregulation of lysosomal calcium. Nature medicine 14:1247. Long X, Lin Y, Ortiz-Vega S, Yonezawa K, Avruch J. 2005. Rheb binds and regulates the mTOR kinase. Current biology 15:702–713. Lunt SY, Vander Heiden MG. 2011. Aerobic glycolysis: meeting the metabolic requirements of cell proliferation. Annual Review of Cell and Developmental Biology 27:441–464. Ma XM, Blenis J. 2009. Molecular mechanisms of mTOR-mediated translational control. Nature reviews Molecular cell biology 10:307. Mancias JD, Kimmelman AC. 2011. Targeting autophagy addiction in cancer. Oncotarget 2:1302. Marat AL, Wallroth A, Lo W-T, Müller R, Norata GD, Falasca M, Schultz C, Haucke V. 2017. mTORC1 activity repression by late endosomal phosphatidylinositol 3, 4-bisphosphate. Science 356:968–972. Martina JA, Chen Y, Gucek M, Puertollano R. 2012. MTORC1 functions as a transcriptional regulator of autophagy by preventing nuclear transport of TFEB. Autophagy 8:903–914. Martina JA, Diab HI, Lishu L, Jeong-A L, Patange S, Raben N, Puertollano R. 2014. The nutrient-responsive transcription factor TFE3 promotes autophagy, lysosomal biogenesis, and clearance of cellular debris. Sci.Signal. 7:ra9.

55 Martın V, Carrillo G, Torroja C, Guerrero I. 2001. The sterol-sensing domain of Patched protein seems to control Smoothened activity through Patched vesicular trafficking. Current Biology 11:601–607. McCauliff LA, Xu Z, Li R, Kodukula S, Ko DC, Scott MP, Kahn PC, Storch J. 2015. Multiple surface regions on the Niemann-Pick C2 protein facilitate intracellular cholesterol transport. Journal of Biological Chemistry 290:27321–27331. Mesmin B, Antonny B, Drin G. 2013a. Insights into the mechanisms of sterol transport between organelles. Cellular and molecular life sciences 70:3405–3421. Mesmin B, Bigay J, von Filseck JM, Lacas-Gervais S, Drin G, Antonny B. 2013b. A four-step cycle driven by PI (4) P hydrolysis directs sterol/PI (4) P exchange by the ER-Golgi tether OSBP. Cell 155:830–843. Meyre D, Delplanque J, Chèvre J-C, Lecoeur C, Lobbens S, Gallina S, Durand E, Vatin V, Degraeve F, Proença C. 2009. Genome-wide association study for early-onset and morbid adult obesity identifies three new risk loci in European populations. Nature genetics 41:157. Millard EE, Gale SE, Dudley N, Zhang J, Schaffer JE, Ory DS. 2005. The sterol-sensing domain of the Niemann-Pick C1 (NPC1) protein regulates trafficking of low density lipoprotein cholesterol. Journal of Biological Chemistry 280:28581–28590. Mizushima N, Levine B, Cuervo AM, Klionsky DJ. 2008. Autophagy fights disease through cellular self-digestion. Nature 451:1069. Möbius W, Ohno-Iwashita Y, Donselaar EG van, Oorschot VM, Shimada Y, Fujimoto T, Heijnen HF, Geuze HJ, Slot JW. 2002. Immunoelectron microscopic localization of cholesterol using biotinylated and non-cytolytic perfringolysin O. Journal of Histochemistry & Cytochemistry 50:43–55. Muir A, Ramachandran S, Roelants FM, Timmons G, Thorner J. 2014. TORC2-dependent protein kinase Ypk1 phosphorylates ceramide synthase to stimulate synthesis of complex sphingolipids. Elife 3:e03779. Nguyen TB, Louie SM, Daniele JR, Tran Q, Dillin A, Zoncu R, Nomura DK, Olzmann JA. 2017. DGAT1-dependent lipid droplet biogenesis protects mitochondrial function during starvation-induced autophagy. Developmental cell 42:9-21. e5. Nohturfft A, Zhang SC. 2009. Coordination of lipid metabolism in membrane biogenesis. Annual Review of Cell and Developmental 25:539–566. Nojima H, Tokunaga C, Eguchi S, Oshiro N, Hidayat S, Yoshino K, Hara K, Tanaka N, Avruch J, Yonezawa K. 2003. The mammalian target of rapamycin (mTOR) partner, raptor, binds the mTOR substrates p70 S6 kinase and 4E-BP1 through their TOR signaling (TOS) motif. Journal of Biological Chemistry 278:15461–15464. Okosun J, Wolfson RL, Wang J, Araf S, Wilkins L, Castellano BM, Escudero-Ibarz L, Al Seraihi AF, Richter J, Bernhart SH. 2016. Recurrent mTORC1-activating RRAGC mutations in follicular lymphoma. Nature genetics 48:183. O’Rourke EJ, Ruvkun G. 2013. MXL-3 and HLH-30 transcriptionally link lipolysis and autophagy to nutrient availability. Nature cell biology 15:668. Ory DS. 2004. The Niemann-Pick disease genes: regulators of cellular cholesterol homeostasis. Trends in cardiovascular medicine 14:66–72. Owen JL, Zhang Y, Bae S-H, Farooqi MS, Liang G, Hammer RE, Goldstein JL, Brown MS. 2012. Insulin stimulation of SREBP-1c processing in transgenic rat hepatocytes requires p70 S6-kinase. Proceedings of the National Academy of Sciences 109:16184–16189.

56 Panchaud N, Peli-Gulli M-P, De Virgilio C. 2013. Amino acid deprivation inhibits TORC1 through a GTPase-activating protein complex for the Rag family GTPase Gtr1. Sci.Signal. 6:ra42. Péli-Gulli M-P, Sardu A, Panchaud N, Raucci S, De Virgilio C. 2015. Amino acids stimulate TORC1 through Lst4-Lst7, a GTPase-activating protein complex for the Rag family GTPase Gtr2. Cell reports 13:1–7. Peng M, Yin N, Li MO. 2017. SZT2 dictates GATOR control of mTORC1 signalling. Nature 543:433. Perera RM, Di Malta C, Ballabio A. 2019. MiT/TFE Family of Transcription Factors, Lysosomes, and Cancer. Annual Review of Cancer Biology 3:203–222. Perera RM, Stoykova S, Nicolay BN, Ross KN, Fitamant J, Boukhali M, Lengrand J, Deshpande V, Selig MK, Ferrone CR. 2015. Transcriptional control of autophagy– lysosome function drives pancreatic cancer metabolism. Nature 524:361. Perera RM, Zoncu R. 2016. The lysosome as a regulatory hub. Annual Review of Cell and Developmental Biology 32:223–253. Peterson TR, Laplante M, Thoreen CC, Sancak Y, Kang SA, Kuehl WM, Gray NS, Sabatini DM. 2009. DEPTOR is an mTOR inhibitor frequently overexpressed in multiple myeloma cells and required for their survival. Cell 137:873–886. Peterson TR, Sengupta SS, Harris TE, Carmack AE, Kang SA, Balderas E, Guertin DA, Madden KL, Carpenter AE, Finck BN. 2011. mTOR complex 1 regulates lipin 1 localization to control the SREBP pathway. Cell 146:408–420. Petit CS, Roczniak-Ferguson A, Ferguson SM. 2013. Recruitment of folliculin to lysosomes supports the amino acid–dependent activation of Rag GTPases. J Cell Biol 202:1107– 1122. Phillips MJ, Voeltz GK. 2016. Structure and function of ER membrane contact sites with other organelles. Nature reviews Molecular cell biology 17:69. Platt FM, Boland B, van der Spoel AC. 2012. Lysosomal storage disorders: The cellular impact of lysosomal dysfunction. J Cell Biol 199:723–734. Plotegher N, Duchen MR. 2017. Mitochondrial dysfunction and neurodegeneration in lysosomal storage disorders. Trends in molecular medicine 23:116–134. Polak P, Cybulski N, Feige JN, Auwerx J, Rüegg MA, Hall MN. 2008. Adipose-specific knockout of raptor results in lean mice with enhanced mitochondrial respiration. Cell metabolism 8:399–410. Porstmann T, Santos CR, Griffiths B, Cully M, Wu M, Leevers S, Griffiths JR, Chung Y-L, Schulze A. 2008. SREBP activity is regulated by mTORC1 and contributes to Akt- dependent cell growth. Cell metabolism 8:224–236. Powis K, Zhang T, Panchaud N, Wang R, De Virgilio C, Ding J. 2015. Crystal structure of the Ego1-Ego2-Ego3 complex and its role in promoting Rag GTPase-dependent TORC1 signaling. Cell research 25:1043. Radhakrishnan A, Goldstein JL, McDonald JG, Brown MS. 2008. Switch-like control of SREBP-2 transport triggered by small changes in ER cholesterol: a delicate balance. Cell metabolism 8:512–521. Radhakrishnan A, Sun L-P, Kwon HJ, Brown MS, Goldstein JL. 2004. Direct binding of cholesterol to the purified membrane region of SCAP: mechanism for a sterol-sensing domain. Molecular cell 15:259–268.

57 Rambold AS, Cohen S, Lippincott-Schwartz J. 2015. Fatty acid trafficking in starved cells: regulation by lipid droplet lipolysis, autophagy, and mitochondrial fusion dynamics. Developmental cell 32:678–692. Rebsamen M, Pochini L, Stasyk T, de Araújo ME, Galluccio M, Kandasamy RK, Snijder B, Fauster A, Rudashevskaya EL, Bruckner M. 2015. SLC38A9 is a component of the lysosomal amino acid sensing machinery that controls mTORC1. Nature 519:477. Ricoult SJ, Manning BD. 2013. The multifaceted role of mTORC1 in the control of lipid metabolism. EMBO reports 14:242–251. Rocha N, Kuijl C, van der Kant R, Janssen L, Houben D, Janssen H, Zwart W, Neefjes J. 2009. Cholesterol sensor ORP1L contacts the ER protein VAP to control Rab7–RILP– p150Glued and late endosome positioning. The Journal of cell biology 185:1209–1225. Roczniak-Ferguson A, Petit CS, Froehlich F, Qian S, Ky J, Angarola B, Walther TC, Ferguson SM. 2012. The transcription factor TFEB links mTORC1 signaling to transcriptional control of lysosome homeostasis. Sci.Signal. 5:ra42. Sabatini DM, Erdjument-Bromage H, Lui M, Tempst P, Snyder SH. 1994. RAFT1: a mammalian protein that binds to FKBP12 in a rapamycin-dependent fashion and is homologous to yeast TORs. Cell 78:35–43. Saftig P, Klumperman J. 2009. Lysosome biogenesis and lysosomal membrane proteins: trafficking meets function. Nature reviews Molecular cell biology 10:623. Sakai J, Rawson RB, Espenshade PJ, Cheng D, Seegmiller AC, Goldstein JL, Brown MS. 1998. Molecular identification of the sterol-regulated luminal protease that cleaves SREBPs and controls lipid composition of animal cells. Molecular cell 2:505–514. Sancak Y, Bar-Peled L, Zoncu R, Markhard AL, Nada S, Sabatini DM. 2010. Ragulator-Rag complex targets mTORC1 to the lysosomal surface and is necessary for its activation by amino acids. Cell 141:290–303. Sancak Y, Peterson TR, Shaul YD, Lindquist RA, Thoreen CC, Bar-Peled L, Sabatini DM. 2008. The Rag GTPases bind raptor and mediate amino acid signaling to mTORC1. Science 320:1496–1501. Sancak Y, Thoreen CC, Peterson TR, Lindquist RA, Kang SA, Spooner E, Carr SA, Sabatini DM. 2007. PRAS40 is an insulin-regulated inhibitor of the mTORC1 protein kinase. Molecular cell 25:903–915. Sardiello M, Palmieri M, di Ronza A, Medina DL, Valenza M, Gennarino VA, Di Malta C, Donaudy F, Embrione V, Polishchuk RS. 2009. A gene network regulating lysosomal biogenesis and function. Science 325:473–477. Sarkar S, Carroll B, Buganim Y, Maetzel D, Ng AH, Cassady JP, Cohen MA, Chakraborty S, Wang H, Spooner E. 2013. Impaired autophagy in the lipid-storage disorder Niemann- Pick type C1 disease. Cell reports 5:1302–1315. Saxton RA, Chantranupong L, Knockenhauer KE, Schwartz TU, Sabatini DM. 2016a. Mechanism of arginine sensing by CASTOR1 upstream of mTORC1. Nature 536:229. Saxton RA, Knockenhauer KE, Wolfson RL, Chantranupong L, Pacold ME, Wang T, Schwartz TU, Sabatini DM. 2016b. Structural basis for leucine sensing by the Sestrin2- mTORC1 pathway. Science 351:53–58. Schöneberg J, Lee I-H, Iwasa JH, Hurley JH. 2017. Reverse-topology membrane scission by the ESCRT proteins. Nature reviews Molecular cell biology 18:5.

58 Schulze RJ, Weller SG, Schroeder B, Krueger EW, Chi S, Casey CA, McNiven MA. 2013. Lipid droplet breakdown requires dynamin 2 for vesiculation of autolysosomal tubules in hepatocytes. J Cell Biol 203:315–326. Settembre C, Ballabio A. 2014. Lysosome: regulator of lipid degradation pathways. Trends in cell biology 24:743–750. Settembre C, De Cegli R, Mansueto G, Saha PK, Vetrini F, Visvikis O, Huynh T, Carissimo A, Palmer D, Klisch TJ. 2013a. TFEB controls cellular lipid metabolism through a starvation-induced autoregulatory loop. Nature cell biology 15:647. Settembre C, Di Malta C, Polito VA, Arencibia MG, Vetrini F, Erdin S, Erdin SU, Huynh T, Medina D, Colella P. 2011. TFEB links autophagy to lysosomal biogenesis. Science 332:1429–1433. Settembre C, Fraldi A, Jahreiss L, Spampanato C, Venturi C, Medina D, de Pablo R, Tacchetti C, Rubinsztein DC, Ballabio A. 2007. A block of autophagy in lysosomal storage disorders. Human molecular genetics 17:119–129. Settembre C, Fraldi A, Medina DL, Ballabio A. 2013b. Signals from the lysosome: a control centre for cellular clearance and energy metabolism. Nature reviews Molecular cell biology 14:283. Settembre C, Zoncu R, Medina DL, Vetrini F, Erdin Serkan, Erdin SerpilUckac, Huynh T, Ferron M, Karsenty G, Vellard MC. 2012. A lysosome‐to‐nucleus signalling mechanism senses and regulates the lysosome via mTOR and TFEB. The EMBO journal 31:1095– 1108. Sever N, Song B-L, Yabe D, Goldstein JL, Brown MS, DeBose-Boyd RA. 2003. Insig- dependent ubiquitination and degradation of mammalian 3-hydroxy-3-methylglutaryl- CoA reductase stimulated by sterols and geranylgeraniol. Journal of Biological Chemistry 278:52479–52490. Shen K, Huang RK, Brignole EJ, Condon KJ, Valenstein ML, Chantranupong L, Bomaliyamu A, Choe A, Hong C, Yu Z. 2018. Architecture of the human GATOR1 and GATOR1– Rag GTPases complexes. Nature 556:64. Shen K, Sabatini DM. 2018a. Ragulator and SLC38A9 activate the Rag GTPases through noncanonical GEF mechanisms. Proceedings of the National Academy of Sciences 115:9545–9550. Shen K, Sabatini DM. 2018b. Ragulator and SLC38A9 activate the Rag GTPases through noncanonical GEF mechanisms. Proceedings of the National Academy of Sciences 115:9545–9550. Singh R, Cuervo AM. 2011. Autophagy in the cellular energetic balance. Cell metabolism 13:495–504. Singh R, Kaushik S, Wang Y, Xiang Y, Novak I, Komatsu M, Tanaka K, Cuervo AM, Czaja MJ. 2009. Autophagy regulates lipid metabolism. Nature 458:1131. Smulan LJ, Ding W, Freinkman E, Gujja S, Edwards YJ, Walker AK. 2016. Cholesterol- independent SREBP-1 maturation is linked to ARF1 inactivation. Cell reports 16:9–18. Su M-Y, Morris KL, Fu Y, Lawrence R, Stjepanovic G, Zoncu R, Hurley JH. 2017. Hybrid structure of the RagA/C-Ragulator mTORC1 activation complex. Molecular cell 68:835- 846. e3. Südhof TC, Russell DW, Brown MS, Goldstein JL. 1987. 42 bp element from LDL receptor gene confers end-product repression by sterols when inserted into viral TK promoter. Cell 48:1061–1069.

59 Sun Y, Fang Y, Yoon M-S, Zhang C, Roccio M, Zwartkruis FJ, Armstrong M, Brown HA, Chen J. 2008. Phospholipase D1 is an effector of Rheb in the mTOR pathway. Proceedings of the National Academy of Sciences 105:8286–8291. Tee AR, Manning BD, Roux PP, Cantley LC, Blenis J. 2003. Tuberous sclerosis complex gene products, Tuberin and Hamartin, control mTOR signaling by acting as a GTPase- activating protein complex toward Rheb. Current biology 13:1259–1268. Thelen AM, Zoncu R. 2017. Emerging roles for the lysosome in lipid metabolism. Trends in cell biology 27:833–850. Thiam AR, Farese Jr RV, Walther TC. 2013. The biophysics and cell biology of lipid droplets. Nature reviews Molecular cell biology 14:775. Thoreen CC, Chantranupong L, Keys HR, Wang T, Gray NS, Sabatini DM. 2012. A unifying model for mTORC1-mediated regulation of mRNA translation. Nature 485:109. Tsun Z-Y, Bar-Peled L, Chantranupong L, Zoncu R, Wang T, Kim C, Spooner E, Sabatini DM. 2013. The folliculin tumor suppressor is a GAP for the RagC/D GTPases that signal amino acid levels to mTORC1. Molecular cell 52:495–505. Um SH, Frigerio F, Watanabe M, Picard F, Joaquin M, Sticker M, Fumagalli S, Allegrini PR, Kozma SC, Auwerx J. 2004. Absence of S6K1 protects against age-and diet-induced obesity while enhancing insulin sensitivity. Nature 431:200. Urano J, Sato T, Matsuo T, Otsubo Y, Yamamoto M, Tamanoi F. 2007. Point mutations in TOR confer Rheb-independent growth in fission yeast and nutrient-independent mammalian TOR signaling in mammalian cells. Proceedings of the National Academy of Sciences 104:3514–3519. Valvezan AJ, Manning BD. 2019. Molecular logic of mTORC1 signaling as a metabolic rheostat. Nature Metabolism 1: 321–333. Van Meer G, Voelker DR, Feigenson GW. 2008. Membrane lipids: where they are and how they behave. Nature reviews Molecular cell biology 9:112. Vander Haar E, Lee S, Bandhakavi S, Griffin TJ, Kim D-H. 2007. Insulin signalling to mTOR mediated by the Akt/PKB substrate PRAS40. Nature cell biology 9:316. Viale A, Pettazzoni P, Lyssiotis CA, Ying H, Sánchez N, Marchesini M, Carugo A, Green T, Seth S, Giuliani V. 2014. Oncogene ablation-resistant pancreatic cancer cells depend on mitochondrial function. Nature 514:628. Wagle N, Grabiner BC, Van Allen EM, Hodis E, Jacobus S, Supko JG, Stewart M, Choueiri TK, Gandhi L, Cleary JM. 2014. Activating mTOR mutations in a patient with an extraordinary response on a phase I trial of everolimus and pazopanib. Cancer discovery 4:546–553. Walker AK, Jacobs RL, Watts JL, Rottiers V, Jiang K, Finnegan DM, Shioda T, Hansen M, Yang F, Niebergall LJ. 2011. A conserved SREBP-1/phosphatidylcholine feedback circuit regulates lipogenesis in metazoans. Cell 147:840–852. Wang C-Y, Stapleton DS, Schueler KL, Rabaglia ME, Oler AT, Keller MP, Kendziorski CM, Broman KW, Yandell BS, Schadt EE. 2012. Tsc2, a positional candidate gene underlying a quantitative trait locus for hepatic steatosis. Journal of lipid research 53:1493–1501. Wang S, Tsun Z-Y, Wolfson RL, Shen K, Wyant GA, Plovanich ME, Yuan ED, Jones TD, Chantranupong L, Comb W. 2015. Lysosomal amino acid transporter SLC38A9 signals arginine sufficiency to mTORC1. Science 347:188–194.

60 Wolfson RL, Chantranupong L, Saxton RA, Shen K, Scaria SM, Cantor JR, Sabatini DM. 2016. Sestrin2 is a leucine sensor for the mTORC1 pathway. Science 351:43–48. Wolfson RL, Chantranupong L, Wyant GA, Gu X, Orozco JM, Shen K, Condon KJ, Petri S, Kedir J, Scaria SM. 2017. KICSTOR recruits GATOR1 to the lysosome and is necessary for nutrients to regulate mTORC1. Nature 543:438. Wyant GA, Abu-Remaileh M, Wolfson RL, Chen WW, Freinkman E, Danai LV, Vander Heiden MG, Sabatini DM. 2017. mTORC1 activator SLC38A9 is required to efflux essential amino acids from lysosomes and use protein as a nutrient. Cell 171:642-654. e12. Xu S, Benoff B, Liou H-L, Lobel P, Stock AM. 2007. Structural basis of sterol binding by NPC2, a lysosomal protein deficient in Niemann-Pick type C2 disease. Journal of Biological Chemistry 282:23525–23531. Yabe D, Xia Z-P, Adams CM, Rawson RB. 2002. Three mutations in sterol-sensing domain of SCAP block interaction with insig and render SREBP cleavage insensitive to sterols. Proceedings of the National Academy of Sciences 99:16672–16677. Yang H, Jiang X, Li B, Yang HJ, Miller M, Yang A, Dhar A, Pavletich NP. 2017. Mechanisms of mTORC1 activation by RHEB and inhibition by PRAS40. Nature 552:368. Yang H, Wang Jia, Liu M, Chen X, Huang M, Tan D, Dong M-Q, Wong CC, Wang Jiawei, Xu Y. 2016. 4.4 Å resolution cryo-EM structure of human mTOR complex 1. Protein & cell 7:878–887. Yang T, Espenshade PJ, Wright ME, Yabe D, Gong Y, Aebersold R, Goldstein JL, Brown MS. 2002a. Crucial step in cholesterol homeostasis: sterols promote binding of SCAP to INSIG-1, a membrane protein that facilitates retention of SREBPs in ER. Cell 110:489– 500. Yang T, Espenshade PJ, Wright ME, Yabe D, Gong Y, Aebersold R, Goldstein JL, Brown MS. 2002b. Crucial step in cholesterol homeostasis: sterols promote binding of SCAP to INSIG-1, a membrane protein that facilitates retention of SREBPs in ER. Cell 110:489– 500. Yecies JL, Zhang HH, Menon S, Liu S, Yecies D, Lipovsky AI, Gorgun C, Kwiatkowski DJ, Hotamisligil GS, Lee C-H. 2011. Akt stimulates hepatic SREBP1c and lipogenesis through parallel mTORC1-dependent and independent pathways. Cell metabolism 14:21–32. Ying ZX, Jin M, Peterson LF, Bernard D, Saiya-Cork K, Yildiz M, Wang S, Kaminski MS, Chang AE, Klionsky DJ. 2016. Recurrent mutations in the MTOR regulator RRAGC in follicular lymphoma. Clinical Cancer Research 22:5383–5393. Yip CK, Murata K, Walz T, Sabatini DM, Kang SA. 2010. Structure of the human mTOR complex I and its implications for rapamycin inhibition. Molecular cell 38:768–774. Yonehara R, Nada S, Nakai T, Nakai M, Kitamura A, Ogawa A, Nakatsumi H, Nakayama KI, Li S, Standley DM. 2017. Structural basis for the assembly of the Ragulator-Rag GTPase complex. Nature communications 8:1625. Yoon M-S, Du G, Backer JM, Frohman MA, Chen J. 2011. Class III PI-3-kinase activates phospholipase D in an amino acid–sensing mTORC1 pathway. J Cell Biol 195:435–447. Zhang HH, Huang J, Düvel K, Boback B, Wu S, Squillace RM, Wu C-L, Manning BD. 2009. Insulin stimulates adipogenesis through the Akt-TSC2-mTORC1 pathway. PloS one 4:e6189.

61 Zhang T, Wang R, Wang Z, Wang X, Wang F, Ding J. 2017. Structural basis for Ragulator functioning as a scaffold in membrane-anchoring of Rag GTPases and mTORC1. Nature communications 8:1394. Zoncu R, Bar-Peled L, Efeyan A, Wang S, Sancak Y, Sabatini DM. 2011. mTORC1 senses lysosomal amino acids through an inside-out mechanism that requires the vacuolar H - ATPase. Science 334:678–683.