The central moTOR of metabolism

Abstract

The protein kinase mTOR (mechanistic target of rapamycin) functions as a central regulator of metabolism, integrating diverse nutritional and hormonal cues to control anabolic processes, organismal physiology and even aging. This review discusses the current state of knowledge regarding the regulation of mTOR signaling and the metabolic regulation of the four macromolecular building blocks of the cell: carbohydrate, nucleic acid, lipid, and protein by mTOR. We review the role of mTOR in the control of organismal physiology and aging through its action in key tissues and discuss the potential for clinical translation of mTOR inhibition for the treatment and prevention of diseases of aging.

eTOC

mTOR is a central regulator of metabolism, integrating nutritional and hormonal cues to control anabolic processes, organismal physiology and aging. Simcox and Lamming review the current state of knowledge regarding mTOR regulation and functions, and discuss potential clinical applications in the treatment and prevention of diseases of aging.

Introduction

The mechanistic Target of Rapamycin (mTOR) is a serine/threonine protein kinase found in diverse species ranging from yeast to humans. mTOR is so named because it is inhibited by rapamycin, a compound produced by bacteria first isolated from the soil of Easter Island, called “Rapa Nui” by the Indigenous Polynesian inhabitants (Vezina et al., 1975). Rapamycin was originally identified as an antifungal agent but was later realized to have profound effects on cell proliferation. Rapamycin and several derivatives, termed “rapalogs”, were developed as immunosuppressants and trialed as anticancer agents as well as for the treatment of diseases resulting from hyperactivity of mTORC1 such as tuberous sclerosis complex (TSC).

The mTOR kinase is the catalytic subunit of two discrete protein complexes, mTOR complex 1 (mTORC1) and mTORC2. mTORC1 and mTORC2 contain shared as well as unique subunits, are regulated by different nutritional and environmental cues, and phosphorylate distinct substrates. mTORC1 is acutely inhibited by rapamycin, which first binds to the immunophilin FKBP12; the FKBP12-rapamycin complex then binds to mTOR kinase through the FRB domain (Brown et al., 1994; Sabers et al., 1995). The acute sensitivity to rapamycin has enabled many of the cellular and organismal roles of mTORC1 to be explored extensively. Rapamycin sensitive mTORC1 substrates include S6K1 (Ribosomal protein S6 kinase 1) T389 and 4E-BP1 S65 (Kang et al., 2013). In contrast to mTORC1, mTORC2, which phosphorylates substrates including AKT S473, is rapamycin resistant and is only inhibited by prolonged treatment with relatively high doses of rapamycin (Sarbassov et al., 2006). This has led to a delay in understanding the activities of mTORC2, since the exploration required the development and use of mTOR kinase inhibitors as well as genetic tools. Overcoming these challenges has also led to the discovery that the activity of mTORC1 towards some substrates, including 4E-BP1 T37/46, ULK1 S757, and Grb10 S150, is rapamycin-resistant (Kang et al., 2013; Thoreen et al., 2009).

Both mTOR complexes, as well as their core subunits are evolutionally conserved from yeast through humans, underscoring their central role in metabolic regulation. The mTORC1 core complex contains mTOR as well as the scaffold protein Raptor and mLST8/GβL (in yeast TOR1, Kog1 and Lst8, respectively), which is required for complex assembly and stability (Hara et al., 2002; Kim et al., 2002; Kim et al., 2003; Loewith et al., 2002; Sabatini et al., 1994). A host of other mTORC1 interacting proteins have also been reported, which may vary between cell types; these include PRAS40 and DEPTOR, both of which regulate mTORC1 activity (Peterson et al., 2009; Sancak et al., 2007), and Tel2 and Tti1, which are important in the assembly and stability of mTORC1 (Kaizuka et al., 2010). The mTORC2 core complex contains mTOR as well as the scaffold protein Rictor, and other mTORC2-specific protein subunits include mSin1 and Protor1/2 (in yeast, TOR2, Avo3, Avo1, and Bit61/2, respectively) (Frias et al., 2006; Loewith et al., 2002; Pearce et al., 2007; Sarbassov et al., 2004). In addition to other proteins that have been reported to associate with mTORC2 in some cell types, including Sestrin3 and Xpln, mTORC2 associates with a number of proteins also found in mTORC1, including mLST8/GβL, DEPTOR, and Tel2 and Tti1 (Khanna et al., 2013; Tao et al., 2015).

Structural studies in both yeast and humans have determined that both mTORC1 and mTORC2 and their yeast homologues TORC1 and TORC2 are dimers with the dimerization facilitated by the interlocking interactions of mTOR with RAPTOR/KOG1 or RICTOR/AVO3, respectively (Stuttfeld et al., 2018; Yip et al., 2010). This dimerization explains in part the complex specific inhibition by rapamycin, as the FKBP12-rapamycin complex directly blocks the formation of the interaction interface for RAPTOR but not the interface for RICTOR (Gaubitz et al., 2015; Stuttfeld et al., 2018). FKBP12-rapamycin is unable to interact with mTORC2/TORC2 as, RICTOR/AVO3 masks the rapamycin-interacting domain of TOR; deletion of this masking domain of RICTOR/AVO3 sensitizes mTORC2/TORC2 to rapamycin (Gaubitz et al., 2015; Karuppasamy et al., 2017; Scaiola et al., 2020).

Regulation of mTOR

The two mTOR complexes play very different roles in the regulation of metabolism and cellular proliferation. mTORC1 functions as a key integrator of environmental and hormonal cues, activating anabolic processes and suppressing catabolic processes when nutrients including amino acids, glucose, cholesterol, and nucleotides are abundant, cellular energy is plentiful, and hormonal cues such as insulin indicate a permissive environment for growth (Castellano et al., 2017; Hoxhaj et al., 2017). When these signals are all favorable for growth and proliferation (the “and gate” model, (Dibble and Manning, 2013)), mTORC1 localizes to the lysosome, where it can interact with GTP-bound Rheb, which binds to and allosterically activates mTORC1 (Yang et al., 2017). mTORC2 functions principally as an effector of PI3K signaling to regulate cell survival and cytoskeleton formation, responding to the activation of PI3K by insulin, IGF-1 and leptin by potentiating the activation of AKT and other downstream effectors of the insulin signaling pathway.

As discussed in greater detail below, the two mTOR complexes communicate with one another via their substrates. In response to PI3K signaling, mTORC2 signals downstream to mTORC1 via AKT, which promotes mTORC1 activity by phosphorylating the mTORC1-interacting protein PRAS40 (Sancak et al., 2007) as well as mTORC1 regulator TSC2 (Manning et al., 2002). In contrast, mTORC1 negatively regulates mTORC2 via feedback inhibition on insulin receptor substrate mediated by the mTORC1 substrates S6K1 and Grb10 (Harrington et al., 2004; Hsu et al., 2011; Yu et al., 2011). S6K1 may also directly inhibit mTORC2 activity via phosphorylation of RICTOR (Julien et al., 2010).

Regulation of mTORC1 by Rag GTPases

In response to amino acids, mTORC1 is recruited to the lysosome by the Rag family of heterodimeric small GTPases (Figure 1). Under conditions of low amino acids, RagA/B binds GDP and RagC/D binds GTP, a state that does not interact with mTORC1. Upon amino stimulation, Rags flip their nucleotide-bound state, interacting with mTORC1 through binding to Raptor and recruiting mTORC1 to the lysosomal surface (Kim et al., 2008; Sancak et al., 2008). The Rag GTPases bind to the lysosomal surface through interaction with the Ragulator, a multiprotein complex composed of the proteins MP1, p14, HBXIP1, C7ORF59 and p18. In this complex, p18 is responsible for anchoring to the lysosomal membrane through N-terminal myristoylated and palmitoylated sites (Bar-Peled et al., 2012; Sancak et al., 2008; Zhang et al., 2017b).

Figure 1: Regulation of mTORC1 Activity by Amino Acids and Growth Factors.

mTORC1 is activated by interacting with Rheb-GTP at the lysosome. Recruitment of mTORC1 to the lysosome is mediated by the RagGTPases; heterodimers of Rags recruit mTORC1 by binding to the mTORC1 subunit Raptor when RagA/RagB is bound to GTP and RagC/RagD is bound to GDP. The GTP/GDP loading status of the Rags is mediated by Ragulator and GATOR1, as well as other GAPs and GEFs. Specific sensors for the amino acids Methionine (SAMTOR), Leucine (Sestrin2), and Arginine (CASTOR1/2) have been identified; when levels of the sensed amino acids are low, these proteins inhibit GATOR1 or GATOR2 activity to modulate the GAP activity of GATOR1 towards RagA and RagB. Amino acids also stimulate the GEF activity of Ragulator/SLC38A9/v-ATPase complex towards RagA/B and the GAP activity of FLCN towards RagC/D. Glucose levels are sensed indirectly via DHAP levels, which controls GATOR2 activity via an as-yet undescribed mechanism. The GEF activity of Ragulator is modulated by multiple amino acids via the v-ATPase and SLC38A9; cholesterol also activates mTORC1 signaling via SLC38A9. NPC1 (Niemann-Pick C1), the major lysosomal cholesterol transporter, interacts with SLC38A9 and inhibits the activation of mTORC1 by cholesterol. Protein-derived amino acids stimulate lysosomal mTORC1 activity via a HOPS-dependent pathway that has not yet been fully described, but which is antagonized by the RagGTPases. The availability of Rheb-GTP is limited by the activity of the TSC complex, which is a GAP for Rheb; TSC activity is inhibited by insulin/IGF-1 signaling via AKT, which phosphorylates TSC2. Purine nucleotides are sensed via TSC2 as well, inhibiting TSC activity, although the precise sensor has not yet been identified. Figure adapted from (Green et al., 2022a) and used with permission from Springer Nature.

Control of the nucleotide loading state of the Rag GTPases is the critical mechanism by which amino acids and several other stimuli control mTORC1 activity. In addition to serving as a platform for mTORC1 activity, the Ragulator acts as a guanine nucleotide exchange factor (GEF); while originally described as a GEF for RagA and RagB (Bar-Peled et al., 2012), more recent data suggests that Ragulator acts as a GEF for RagC (Shen and Sabatini, 2018). Amino acids control the Ragulator in part via the lysosomal vacuolar ATPase (v-ATPase), which interacts with Ragulator in a manner sensitive to both the availability of lysosomal amino acids and ATP hydrolysis activity of the v-ATPase (Zoncu et al., 2011). A low-affinity lysosomal amino acid transporter, SLC38A9, interacts with both Ragulator and the v-ATPase, and acts as an amino acid sensor upstream of mTORC1 for several amino acids, including asparagine, arginine, glutamine, histidine and lysine (Jung et al., 2015; Rebsamen et al., 2015; Wang et al., 2015). SLC38A9 is also required for mTORC1 activation by cholesterol (Castellano et al., 2017). SLC38A9 has been reported to act as a GEF for RagA (Shen and Sabatini, 2018), but more recently has been shown to act as a GTPase activating protein (GAP) for RagC by acting to destabilize a lysosomal folliculin complex (Fromm et al., 2020).

The nucleotide loading state of RagA and RagB are also controlled by GATOR1, which functions as a GAP for RagA/B (Bar-Peled et al., 2013). GATOR2, a second protein complex, associates with GATOR1 and inhibits its activity. GATOR2 and GATOR1 function as critical regulators of both amino acid and glucose signaling to mTORC1. In the case of leucine and arginine, specific proteins have been identified that in the absence of these specific amino acids bind to and inactivate GATOR2. Leucine is sensed by the Sestrin family of proteins, which bind to and inhibit the activity of GATOR2 (Chantranupong et al., 2014; Parmigiani et al., 2014). Leucine binding to a specific pocket on Sestrin2 relieves the inhibitory action of Sestrin2 on GATOR2, allowing lysosomal recruitment of mTORC1 (Saxton et al., 2016b; Wolfson et al., 2016). This may not be the only mechanism by which leucine activates mTORC1; the major monomethyl branched-chain fatty acid C17ISO, which is synthesized from leucine and acetyl CoA stimulates mTORC1 activity in C. elegans via a Sestrin-independent mechanism, and this activity appears to be conserved in mammalian cells (Zhu et al., 2021).

The CASTOR proteins similarly act as arginine sensors, binding to GATOR2 and inhibiting its activity when arginine is absent (Chantranupong et al., 2016; Saxton et al., 2016a). The SAMTOR protein has been identified as a negative regulator of mTORC1 that binds to GATOR1 directly to modulate its activity (Gu et al., 2017a). Instead of directly sensing methionine, SAMTOR binds to the metabolite S-adenosylmethionine (SAM); SAM binding disrupts the SAMTOR-GATOR1 interaction. Interestingly, mTORC1 activity stimulates SAM synthesis (Cho et al., 2021; Villa et al., 2021), and thus promotes the disassociation of SAMTOR and GATOR1. In addition to being required for amino acid sensing by mTORC1, the Rag family of proteins has also been implicated in the sensing of glucose levels. While glucose itself is not sensed by mTORC1, the glycolytic intermediate dihydroxyacetone phosphate (DHAP) is sensed via a mechanism that is dependent upon both GATOR complexes (Orozco et al., 2020).

There is significant additional complexity in the regulation of mTORC1 by amino acids that remains to be discovered. A recent study suggests that while the RagGTPases are critical for the sensing of exogenous amino acids, lysosomal-derived amino acids can activate mTORC1 via a Rag-independent mechanism that requires the homotypic fusion and vacuole protein sorting (HOPS) tethering complex (Hesketh et al., 2020). In this context, the Rag-GATOR pathway seemingly acts as a negative regulator of mTORC1. Finally, while this review has focused on the lysosome as a central hub for mTORC1 activity, an intracellular reporter of mTORC1 activity has identified amino acid-dependent mTORC1 activity not only at the lysosome but also in the nucleus (Zhou et al., 2015). It is clear that more remains to be discovered about the regulation of mTORC1 activity by amino acids as well as other metabolites.

Regulation of mTORC1 by TSC-Rheb

Recruitment to the lysosome is required for activation of mTORC1, because only at the lysosome can mTORC1 be activated by allosteric binding to the GTP-bound form of the Rheb GTPase. Binding of mTORC1 to Rheb-GTP closes the catalytic cleft of the mTOR kinase domain, aligning the ATP phosphate groups of the “N” lobe into close proximity to the “C” lobe D2338 and H2340 catalytic residues (Yang et al., 2017). Rheb-GTP is regulated principally by the tuberous sclerosis complex (TSC), which is composed of three proteins, TSC1, TSC2 and TBC1D7 (Dibble et al., 2012). TSC acts as a GAP for Rheb; as only Rheb-GTP can activate mTORC1, while Rheb-GDP cannot, active TSC inhibits mTORC1.

Many different kinases, including AMPK and AKT, phosphorylate distinct residues of TSC1 and TSC2, altering the ability of TSC to act as a GAP on Rheb (Corradetti et al., 2004; Inoki et al., 2002). While some of these posttranslational modifications may alter the activity of TSC, TSC localization to the lysosome is regulated by AKT. In response to insulin/IGF-1 PI3K signaling, AKT phosphorylates TSC resulting in its departure from the lysosome and allowing Rheb-GTP to activate mTORC1 (Menon et al., 2014). The lysosomal localization of both Rheb and TSC may also be regulated in part by amino acids (Demetriades et al., 2014; Fawal et al., 2015), and recruitment to the lysosome is dependent on binding to the lysosomal lipid PI3,5P₂ (Fitzian et al., 2021). Interestingly, TSC and Rheb are also required for mTORC1 sensing of purine levels, though the precise metabolite sensed as well as the identity of the sensor is as yet unknown (Hoxhaj et al., 2017).

Regulation of mTORC2

mTORC2 is activated by hormones including insulin, IGF-1, and leptin to regulate cellular processes including metabolism, survival and cytoskeletal organization (Figure 2) (Park and Ahima, 2014). These hormones interact with receptors on the outer surface of the cell to activate signaling cascades that increase the PI3K product, phosphatidylinositol (3,4,5)-trisphosphate (PIP3), which in turn activates mTORC2 through relieving an inhibitory interaction with the mSIN1 subunit (Liu et al., 2015). Because this tyrosine kinase signaling cascade is localized to the plasma membrane, it would make sense that mTORC2 also localizes to the plasma membrane. Consistent with this, intracellular reporters of mTORC2 activity have been used to show that the plasma membrane is indeed a site of mTORC2 activity (Ebner et al., 2017). Further, multiple studies have identified the plasma member as a key site of TORC2 activity in yeast (Berchtold et al., 2012; Berchtold and Walther, 2009; Riggi et al., 2018).

Figure 2: Regulation of mTORC2 activity by Growth Factors.

mTORC2 is activated by insulin, IGF-1, and leptin via activation of PI3K signaling. mTORC2 is activated by binding of mSIN1 to PIP₃; substrates of mTORC2 include the S473 and T450 residues of AKT as well as specific sites on SGK, multiple PKC isoforms, and SIRT6 (reviewed in (Kennedy and Lamming, 2016)).

Interestingly, mTORC2 activity has also been observed at mitochondria and some endosomal vesicles (Ebner et al., 2017). This is consistent with some previous reports that mTORC2 is physically associated with ribosomal subunits and localizes to the mitochondria-associated endoplasmic reticulum, interacting with a specific mitochondrial tethering complex (Betz et al., 2013; Zinzalla et al., 2011). A ribosomal interaction with mTORC2 is also consistent with phosphorylation of AKT (T450), which occurs co-translationally (Oh et al., 2010). Similarly, endosomal localization for mTORC2 is consistent with a recently discovered role for mTORC2 in chaperone-mediated autophagy (Arias et al., 2015).

Metabolic regulation by mTOR

mTORC1 and mTORC2 function to synchronize cellular growth to proliferative environmental cues such as growth factors and nutrient availability. Once activated mTOR regulates the four biological macromolecules that serve as building blocks of the cell: carbohydrates, nucleic acids, proteins, and lipids (Figure 3). The regulation of these macromolecules is the basis of how mTOR functions as a central regulator of cell mass, differentiation, and proliferation. The control of macromolecular metabolism also makes mTOR function as a driver for dysfunction in cancer, immune cell expansion, and aging (Jones and Pearce, 2017; Mossmann et al., 2018).

Figure 3: Control of cellular metabolism and other processes by mTORC1 and mTORC2.

mTORC1 regulates processes required for cell proliferation including the pentose phosphate pathway, protein translation, lipolysis, lipogenesis, and mitochondrial biogenesis. Most of this regulation is mediated downstream of S6K1 and the 4E-BPs. In contrast mTORC2 regulates the pentose phosphate pathway through phosphorylation of transketolase, hexosamine biosynthesis through the phosphorylation of GFAT1 (glutamine:fructose-6-phosphate amidotransferase 1), and controls lipid metabolism through SREBP1, HIF1α, FOXO1 mediated transcriptional regulation. These pathways cumulatively control essential building blocks for cell proliferation, division, and maintenance.

Carbohydrate metabolism

A role for mTOR signaling in carbohydrate metabolism has been the subject of investigation for some time following initial observations that chronic treatment with rapamycin causes fasting hyperglycemia and glucose intolerance in rodents and is associated with an increased prevalence of new-onset diabetes in humans (Cunningham et al., 2007; Houde et al., 2010; Johnston et al., 2008). However, mTORC1 does not play a major role in blood glucose regulation, with mice lacking either hepatic TSC1 or Raptor having no appreciable alterations in glucose homeostasis (Lamming et al., 2012; Sengupta et al., 2010). Inhibition of mTORC1 in other tissues also failed to find substantial impairment of glucose regulation; while mice lacking Raptor in skeletal muscle do have impaired glucose tolerance, this is likely a consequence of the severe muscle dystrophy in these animals (Bentzinger et al., 2008). Deletion of Raptor in adipose tissue alone led to improved insulin sensitivity (Polak et al., 2008). While mTORC1 does have an important role in pancreatic beta cells, this may be a result of a role for Raptor in beta cell identity rather than a direct effect on glucose control (Rachdi et al., 2008; Yin et al., 2020).

Despite this data, there are compelling reasons to believe that mTORC1 does have a role in carbohydrate metabolism. First, mTORC1 activity is sensitive to glucose levels; while this originally was thought to be indirectly via the action of AMPK and TSC (Kim et al., 2013), work has revealed that mTORC1 senses glucose through the glycolytic intermediate DHAP (Orozco et al., 2020). Secondly, neonatal mice expressing a constitutively GTP-bound form of RagA (RagA-GTP) develop normally but are unable to induce autophagy following birth, and as a result die from hypoglycemia if not rescued by rapamycin (Efeyan et al., 2013). RagA-GTP mice that are rescued from this energetic crisis are unable to properly respond to fasting and have metabolic alterations including impaired glucose tolerance (de la Calle Arregui et al., 2021). Finally, mTORC2 has been shown to regulate glucose uptake and glycolysis through HIF1α activation. TSC1^−/− and TSC2^−/− MEFs demonstrated altered expression of HIF1α glycolytic targets including G6PD and GLUT1 (Duvel et al., 2010). HIF1α activation appears to be through mTORC1 phosphorylation of the transcription factor forkhead/winged helix family k1 (FOXK1), and FOXK1 binding sites were identified and validated in the HIF1α promoter (He et al., 2018). The physiological relevance of this role for mTORC1 remains to be determined.

In contrast, mTORC2 has been found to have major effects on carbohydrate metabolism. The induction of glucose intolerance by chronic rapamycin treatment results from hepatic insulin resistance, and genetic disruption of hepatic mTORC2 is sufficient to reproduce this effect (Arriola Apelo et al., 2020; Lamming et al., 2012). Subsequent genetic experiments have revealed that mTORC2 disruption in multiple tissues, including the adipose tissue, brain, and pancreatic islets contribute to the effects of rapamycin on glucose intolerance, and that disruption of mTORC2 in skeletal muscle impairs insulin sensitivity (Chellappa et al., 2019; Cybulski et al., 2009; Gu et al., 2011; Kleinert et al., 2017; Kumar et al., 2008; Kumar et al., 2010; Yu et al., 2019). While the majority of these effects are likely due to the generation of insulin resistance – deletion of the mTORC2 subunit insulin/IGF-1 receptor - mTORC2 also senses (directly or indirectly, via as-yet unknown mechanisms) levels of glucose and α-ketoglutarate (Arriola Apelo and Lamming, 2016; Moloughney et al., 2016). In the face of glutamine or glucose depletion, mTORC2 promotes glutaminolysis to promote anaplerotic filling of the TCA cycle and hexosamine biosynthesis. Interestingly, both mTOR complexes likely cooperate in this process, as mTORC1 helps to sustain anaplerosis by stimulating a-ketoglutarate synthesis from glutamate, the end product of glutaminolysis.

Nucleic acid metabolism

Both mTORC1 and mTORC2 increase nucleic acid production through increasing glucose flux into the pentose phosphate pathway (PPP). After glucose is taken up into the cell it is phosphorylated and can either be broken down further into pyruvate to produce ATP or it can be shunted into the PPP to regulate reduction/oxidation (redox) homeostasis and nucleic acid synthesis. The oxidative branch of the PPP functions primarily to produce ribulose 5-phosphate and NADPH; the NADPH provides the reducing power for synthesis of fatty acid, cholesterol, and amino acids. The non-oxidative PPP branch produces several metabolites including ribose 5-phosphate which serves as the precursor to purines and pyrimidines and can facilitate re-entry of the carbons into glycolysis (Ge et al., 2020).

To regulate both the oxidative and nonoxidative PPP, mTORC1 activity increases the expression of both glucose 6 phosphate dehydrogenase (G6PD) and ribose 5-phosphate isomerase A (RPIA) (Buj et al., 2019; Duvel et al., 2010; Evert et al., 2012). This regulation is mediated through S6K and sterol regulatory-element binding protein 1 (SREBP1), overexpression of Rheb or SREBP1a is sufficient to induce G6PD and RPIA and knockdown of SREBP1 ablates G6PD expression (Duvel et al., 2010).

mTORC1 also regulates pyrimidine synthesis through regulation of the trifunctional multi-domain enzyme CAD (carbamoyl-phosphate synthetase 2, aspartate transcarbamylase, and dihydro-orotase) (Ben-Sahra et al., 2013; Robitaille et al., 2013). CAD phosphorylation on S1859 by S6K1 increases activity. In TSC2-deficient MEFs that have higher mTORC1 activity, there is higher pyrimidine synthesis and rapamycin treatment ablates this increase (Ben-Sahra et al., 2013). Moreover, CAD directly binds mLST8/GβL and this interaction is decreased with depletion of amino acids or serum (Nakashima et al., 2013). Another point of regulation is through mTORC1 control of purine synthesis through ATF4 (activating transcription factor 4), a stress responsive transcription factor that promotes the transcription of stress response genes, including those involved in amino acid uptake and biosynthesis (Pakos-Zebrucka et al., 2016). ATF4 coordinates mTORC1 activity with nutrient ability; while increasing cellular levels of amino acids, it also limits mTORC1 activity by inducing expression of Sestrin2 (Ye et al., 2015). ATF4 similarly limits mTORC1 activity via induction of Sestrin2 in response to mitochondrial distress and reactive oxygen species (Condon et al., 2021; Garaeva et al., 2016). ATF4 activation increases methylenetetrahydrofolate dehydrogenase 2 (MTHFD2) transcripts and regulates metabolites necessary for purine synthesis such as those derived from serine metabolism (Ben-Sahra et al., 2016). While mTORC1 is sufficient to drive purine and pyrimidine synthesis it is not necessary (Ben-Sahra et al., 2013; Kliegman et al., 2013).

mTORC2 regulates the nonoxidative branch of the PPP through AKT mediated phosphorylation of transketolase, facilitating the re-entry of carbons into glycolysis. Treatment of Hela cells with Torin1 or Rictor knockdown led to decreased purine synthesis while rapamycin treatment and knockdown of Raptor had no effect on purine synthesis (Saha et al., 2014). Use of mTORC2 specific inhibitors in yeast demonstrated regulation of PPP enzymes and use of this mTORC2 inhibitor decreased ribose 5-phosphate levels (Kliegman et al., 2013).

Through regulation of carbon flux into the PPP, mTOR controls the production of NADPH for lipid synthesis and the production of purines and pyrimidines which are components of nucleic acids and cofactors such as NAD+. These are necessary components for cellular proliferation and points of dysregulation in cancer initiation and progression, and several cancers such as lung cancer, hepatocellular carcinoma, leukemia, and breast cancer have higher activation of G6PD and transketolase (Vander Heiden and DeBerardinis, 2017). Treatment of 26 different cancer cell lines with an S6K1 inhibitor or with the rapamycin analog everolimus, blocked cell proliferation and decreased PPP metabolites (Pusapati et al., 2016). These results highlight mTORC1 specific inhibition as a potential therapeutic target for cancer proliferation and initiation.

Protein and amino acid metabolism

mTOR regulates cellular protein production through control of translation initiation and ribosomal biogenesis as well as controlling the biogenesis of several amino acids (Park et al., 2009). The majority of this regulation has been shown to be mTORC1 dependent through activation of S6K and the eukaryotic translation initiation factor 4E-binding proteins (4E-BPs). Translation requires the unwinding of secondary structures in the 5’ untranslated region of mRNA, this priming event is performed by the eukaryotic translation initiation factor 4A (eIF4A). eIF4A activity is promoted by complex formation with eIF4G, eIF3, and eIF4E. Insulin activation of mTORC1 leads to phosphorylation of 4E-BP, which then phosphorylates eukaryotic initiation factor 4E (eIF4E) (Haghighat et al., 1995; Pause et al., 1994). mTORC1 also activates eIF4G and eIF4B through S6K-mediated phosphorylation (Holz et al., 2005; Raught et al., 2004).

Another mechanism of translational regulation is through mTORC1 regulation of ribosomal biogenesis through S6K1 phosphorylation of the carboxy terminal activation domain of the rDNA transcription factor upstream binding factor (UBF) (Hannan et al., 2003). Moreover, mTORC1 regulates the nuclear localization of TIF-IA which upregulates ribosomal transcripts (Mayer et al., 2004). Global expression analysis has also implicated a role for mTORC1 in regulating amino acid transport and biogenesis, as rapamycin treatment decreases expression of asparagine synthase and neutral amino acid transporters (Peng et al., 2002). These studies deserve further assessment following the discovery of rapamycin-resistant functions of mTORC1 (Thoreen et al., 2009). Studies using rapamycin and Torin1 have shown that there is rapamycin-resistant regulation of ATF4 by mTOR which controls expression of amino acid transporters CHAC1, SLC7A11, SESN2, SLC7A5, SLC7A1, and SLC3A2. CHAC1, SLC7A11, and SESN2 are regulated by mTORC1 via ATF4, as MEFs lacking both 4E-BP1 and 4E-BP2 have reduced expression of these transporters (Park et al., 2017).

Less is known about how mTORC2 regulates protein synthesis but its function as a regulator of insulin signaling, a known anabolic activator of protein synthesis, suggests that it controls protein production (Haghighat et al., 1995). Moreover, mTORC2 directly interacts with the ribosomal complex when active and this interaction is functionally relevant, with knockdown of ribosomal proteins reducing mTORC2 activity in Hela cells (Zinzalla et al., 2011). While mTORC2 function is dependent on ribosomal interaction, it is unknown how ribosomal function is impacted by mTORC2. In mouse models of breast cancer the mTORC2 interaction with the ribosome has been shown to be a driver of the increased ribosomal biogenesis in cancer progression (Prakash et al., 2019). mTORC2 also regulates amino acid transport through phosphorylation of the cystine-glutamate antiporter xCT on S26. Rictor knockdown prevents this phosphorylation, leading to increased glutamate release and increased cystine uptake (Gu et al., 2017b).

Lipid metabolism

Through regulation of the PPP mTOC1 and mTORC2 provide NADPH necessary for lipid synthesis. Cells require lipids as components of cellular membranes, intra and inter- cellular signaling molecules, and for energy storage. These diverse functions are mediated by a diversity of lipid structural elements including different head groups, backbones, and acyl chains. The steady state lipid abundance is regulated through balance of synthesis and oxidation, mTOR is a key regulator of this balance through both direct and indirect targets. Activation of mTOR is associated with lipid synthesis and increased lipid droplets, while inhibition of mTOR causes higher rates of oxidation. Both mTORC1 and mTORC2 have distinct roles at the level of lipogenesis, lipolysis, and oxidation.

mTORC1 regulates the synthesis of fatty acids through controlling nuclear translocation of SREBP1. Once in the nucleus SREBP1 transcriptionally regulates fatty acid synthase (FASN) and acetyl-CoA carboxylase (ACC), which are required for the synthesis of palmitic acid, as well as stearoyl CoA desaturase 1 (SCD1), required for acyl chain desaturation (Mauvoisin et al., 2007; Porstmann et al., 2008). SREBP1 nuclear translocation is activated by insulin signaling through AKT to increase saturated and unsaturated fatty acids as well as phospholipid synthesis. Inhibition of mTORC1 by rapamycin or Raptor knockdown ablates this AKT induced fatty acid synthesis, while knockdown of Rictor has no impact (Li et al., 2010; Porstmann et al., 2008).

mTORC1 regulates SREBP1 indirectly through phosphorylation of SREBP1 by S6K1 and CREB-regulated transcription coactivator 2 (Dong et al., 2020; Duvel et al., 2010; Han et al., 2015). mTORC1 also indirectly regulates SREBP by phosphorylation of Lipin1, a phosphatidic acid phosphatase. The phosphorylation of Lipin 1 prevents its localization to the nucleus and through unknown mechanisms allows SREBP1 to enter (Peterson et al., 2011). Additionally, mTORC1 regulates lipogenesis is through epigenetic regulation of FASN and ACC. mTORC1 phosphorylates JMJD1c which demethylates H3K9me1 increasing chromatin accessibility and gene expression of FASN and ACC (Viscarra et al., 2020). mTORC1 can also regulate FASN and ACC post-transcriptionally through S6K mediated phosphorylation of SRPK2 which controls alternative splicing of FASN and SCD1 (Lee et al., 2017).

mTORC2 also regulates lipogenesis through tissue specific mechanisms, in adipose tissue this is mediated through controlling expression of carbohydrate response element binding protein β (ChREBPβ) while in the liver mTORC2 regulates SREBP1. Using adipose tissue specific deletion of Rictor driven by Adiponectin-Cre led to decreased FASN, ACC, and ELOVL6 (Tang et al., 2016). This drop in de novo lipogenesis was also associated with decreased body weight in response to high fat diet and decreased fat mass. These results contradict previous observations that adipose tissue specific Rictor ablation driven by ap2-cre led to higher body weight, increased fat mass, and ectopic lipid deposition in the liver (Kumar et al., 2010). These different phenotypes are likely driven by the differences cell type specificity from the promoter driving cre recombinase expression, adiponectin has been shown to be adipocyte specific while ap2 is expressed in adipocytes, macrophages, and neural crest cells (Jeffery et al., 2014). There are also challenges with using Rictor KO to interpret mTORC2 function, since Rictor has been shown to form mTORC2-idependent complexes with PKCζ, Culin1, and Myo1 (Gkountakos et al., 2018). In the liver the mTORC2 regulation is controlled through SREBP1c, liver specific knockout of Rictor (LRicKO) leads to decreased transcriptional expression of SREBP1c and decreased nuclear localization (Yuan et al., 2012). The lower levels of SREBP1c in the L-RicKO mice led to decreased expression of FASN and ACC as well as lower triglyceride levels.

Another point of regulation for lipid homeostasis is in control of lipid breakdown, mTORC1 inhibits both lipolysis and oxidation. mTORC1 regulation of Egr1 inhibits adipose triglyceride lipase (ATGL), the enzyme that hydrolyzes free fatty acid from triglyceride (Singh et al., 2015). mTORC1 also inhibits fatty acid oxidation through regulating the entry of free fatty acids into the mitochondria for β-oxidation. Very long chain fatty acids enter the mitochondria through the carnitine shuttle where carnitine palmitoyltransferase 1 (CPT1) esterifies carnitine to long chain fatty acids, the newly formed acylcarnitine enters the porous outer mitochondrial membrane and is then shuttled into the matrix by carnitine acylcarnitine transferase where CPT2 removes the carnitine for a CoA group. Rapamycin treatment in skeletal muscle cells led to decreased CPT1 and CPT2 activity and subsequent decreases in β-oxidation (Sipula et al., 2006). In the liver, constitutive activation of mTORC1 through deletion of TSC1 leads to decreased oxidation and ketogenesis (Sengupta et al., 2010).

mTORC2 similarly inhibits lipolysis in white and brown adipose tissue. Knockout of Rictor in the adipose tissue leads to increased circulating free fatty acid levels and higher activation of PKA (Kumar et al., 2010). In brown adipose tissue, the regulation of lipolysis has been shown to be FOXO1 mediated. mTORC2 interacts with SIRT6 and inhibits its activity through an unknown mechanism. Brown adipose tissue specific deletion of Rictor led to Foxo1 deacetylation by Sirt6 resulting in increased Foxo1 nuclear localization and increased ATGL expression. Deletion of both Rictor and Foxo1 ablated this induction of ATGL (Jung et al., 2019).

Beyond regulation of cellular lipid metabolism, mTORC1 and mTORC2 have been shown to regulate whole body lipid homeostasis in mouse models through control of adipocyte differentiation and lipolysis. Ablation of mTORC1 through adipocyte specific deletion of Raptor leads to decreased subcutaneous and visceral white adipose tissue as well as brown adipose tissue (Lee et al., 2016). This coincided with hepatic steatosis and increased plasma free fatty acid and triglyceride levels that were ablated in double knockout of Raptor and adipose triglyceride lipase (ATGL) (Paolella et al., 2020). Rapamycin treatment also decreases catecholamine stimulated lipolysis and prevents the differentiation of beige adipocytes via a mTORC1-dependent process (Tran et al., 2016).

mTORC2 also impacts adipogenesis and lipolysis, with a substantial impact on the brown adipocyte program. Adipose tissue specific deletion of Rictor leads to insulin resistance, mild hepatic steatosis, and resistance to high fat diet but no major differences in white adipose tissue morphology or abundance (Tang et al., 2016). However, brown adipose tissue specific deletion of Rictor increased thermogenic transcripts and body temperature in response to a cold challenge and decreased hepatic steatosis (Jung et al., 2019). More recently it has been shown that mice lacking adipose Rictor have decreased thermogenic capacity and uncoupling protein 1 expression, although body temperature is unaffected, likely due to increased shivering (Castro et al., 2021). The differences in the results of these studies could be due to thermoneutral housing, which dramatically alters energy expenditure and metabolic phenotypes (Seeley and MacDougald, 2021; Skop et al., 2020).

Regulation of aging by mTOR

Research in model organisms has demonstrated that genetic inhibition of TOR signaling extends the lifespan of yeast, worms, and flies (Kaeberlein et al., 2005; Kapahi et al., 2004; Vellai et al., 2003). mTORC1 signaling also correlates with longevity, with mTORC1 signaling decreased in long-lived Ames dwarf mice (Sharp and Bartke, 2005). mTORC1 signaling also increases with age in multiple tissues in wild-type mice and rats (Baar et al., 2016; Shavlakadze et al., 2018), suggesting that inhibition of mTORC1 signaling might be a strategy to promote healthspan and longevity in mammals as well.

Over a decade ago, this was shown to be the case, with a landmark study demonstrating that treatment of aged mice with rapamycin can extend lifespan (Harrison et al., 2009). Subsequent work has shown that rapamycin can extend the lifespan of multiple strains of mice as well as mouse models of many different diseases (reviewed in (Dumas and Lamming, 2020)). Importantly, rapamycin works to extend lifespan in both male and female mice, across a broad range of doses and dosing strategies, and when given for varying lengths of time (Arriola Apelo et al., 2016b; Bitto et al., 2016; Miller et al., 2014).

Numerous beneficial effects of rapamycin that may contribute to its positive effects on both lifespan and healthspan have been documented. Rapamycin has anti-cancer effects that likely contribute to its effects on longevity. Rapamycin prevents or delays the onset of age-related changes in multiple rodent tissues including the heart, liver, kidney, muscle and tendons (Dai et al., 2014; Flynn et al., 2013; Joseph et al., 2019; Shavlakadze et al., 2018; Wilkinson et al., 2012; Zaseck et al., 2016). In mice, rapamycin rejuvenates hematopoietic stem cells (Chen et al., 2009), promotes the self-renewal of intestinal stem cells (Yilmaz et al., 2012), delays age-related hearing loss (Altschuler et al., 2021), and reverses age-associated periodontitis, regenerating bone and reducing inflammation (An et al., 2020). Rapamycin may also have beneficial effects on cell senescence and the senescence-associated secretory phenotype (Cao et al., 2011; Wang et al., 2017).

Identifying the molecular mechanisms through which rapamycin promotes healthy aging and lifespan has been a subject of intense study. As discussed above, rapamycin inhibits both mTORC1 and mTORC2 when given chronically, and studies in model organisms, including yeast worms and flies, have shown that inhibition of mTORC1 or signaling pathways downstream of mTORC1 including S6K and translational initiation factors extends lifespan (Hansen et al., 2007; Kaeberlein et al., 2005; Kapahi et al., 2004; Powers et al., 2006; Syntichaki et al., 2007; Vellai et al., 2003). In contrast, reducing mTORC2 signaling in C. elegans modulates lifespan in a way that is dependent upon environmental cues (Mizunuma et al., 2014), and increased mTORC2 signaling extends the lifespan on Drosophila (Chang et al., 2020).

In mice, much the same paradigm has been observed, with inhibition of mTORC1 signaling generally being beneficial. Genetic mouse models of diminished mTORC1 activity, such as mice expressing a hypomorphic allele of mTOR, mice heterozygous for both mTOR and mLST8, and mice overexpressing Tsc1, have longer lifespans (Lamming et al., 2012; Wu et al., 2013; Zhang et al., 2017a). Other modes of reduced mTORC1 or S6K1 signaling, including mice lacking S6K1, mice with constitutive dephosphorylation of the S6K1 substrate glutamyl-prolyl-tRNA synthetase (EPRS), or mice deficient for mitochondrial arginase type II, likewise have increased lifespan (Arif et al., 2017; Selman et al., 2009; Xiong et al., 2017).

Other molecular pathways downstream of mTORC1 may also play a role in the effect of rapamycin or mTORC1 inhibition on lifespan. While the effect of 4E-BP1 activation on mouse lifespan is unknown, increased expression of 4E-BP1 in the whole body of mice or specific activation of 4E-BP1 in skeletal muscle are both associated with favorable metabolic benefits, including reduced adiposity and protection from diet-induced hepatic steatosis and insulin resistance (Tsai et al., 2015; Tsai et al., 2016). Some of these effects may be the result of increased expression of the hormone FGF21 from muscle, expression of which is sufficient to extend mouse lifespan (Zhang et al., 2012).

mTORC1 inhibition also activates autophagy via the release of an inhibitory phosphorylation on the mTORC1 substrates ULK1 and ATG13 (Egan et al., 2011; Puente et al., 2016). While the effect of autophagy activation on the lifespan of mice has not been directly examined, work in C. elegans suggests that activation of autophagy extends lifespan and plays a key role in the response to both dietary restriction and mTORC1 inhibition (Hansen et al., 2008; Seah et al., 2016; Toth et al., 2008). Autophagy activation can also promote healthspan in mammals, restoring hepatic function in aged mice (Zhang and Cuervo, 2008).

Relatively few studies have examined the role of mTORC2 signaling in lifespan and healthy aging. In C. elegans, studies examining the effect of knocking down or deleting the worm homolog of Rictor have found that mTORC2 inhibition can either extend or shorten lifespan depending upon the targeted tissue, the temperature, and the food source (Mizunuma et al., 2014). In flies, overexpression of Rictor extends lifespan (Chang et al., 2020).

In mice, inhibition of mTORC2 in the whole body through deletion of one or both copies of Rictor significantly reduces the lifespan of male mice (Lamming et al., 2014). Inhibition of mTORC2 signaling in the adipose tissue, hypothalamus, or liver likewise impairs healthspan and reduces lifespan (Arriola Apelo et al., 2020; Chellappa et al., 2019; Yu et al., 2019), while long-lived Snell dwarf mice and Ghr^−/− mice have elevated mTORC2 signaling (Dominick et al., 2015; Garratt et al., 2017). Several agents identified as geroprotectors by the National Institute on Aging Interventions Testing Program likewise activate mTORC2 (Garratt et al., 2017).

Recent experiments with mTORC1-specific rapalogs and intermittent treatment regimens have clarified that many of the side effects of chronic treatment with rapamycin – including fasting hyperglycemia, glucose intolerance, hyperlipidemia, and immunosuppression – are due in whole or in part to inhibition of mTORC2 (Arriola Apelo et al., 2016a; Schreiber et al., 2019). Conversely, inhibition of mTORC1 is sufficient to extend lifespan and healthspan in mice (Arriola Apelo et al., 2016b; Lamming et al., 2012; Selman et al., 2009; Zhang et al., 2017a). The side effects of rapamycin have limited enthusiasm and slowed the clinical deployment of rapamycin to prevent or treat age related disease.

However, it remains uncertain if the effects of chronic rapamycin on mTORC2 contribute to the effects of rapamycin on lifespan. Work in model organisms and in mice have shown that mutations that induce insulin resistance can promote longevity, perhaps as a consequence of decreased mTORC1 signaling (reviewed in (Blagosklonny, 2012; Lamming, 2014)). However, rapamycin treatment acutely inhibits mTORC1. While mice with genetically disrupted mTORC2 signaling have reduced lifespan, a few beneficial effects downstream of mTORC2 signaling have been found. Inhibition of mTORC2 helps to slow the progress of certain cancers that are dependent upon elevated PI3K signaling (Guertin et al., 2009), and partial genetic inhibition of Akt1 extends the lifespan of both C. elegans and mice (Nojima et al., 2013). Decreased mTORC2 also partially rescues intrinsic and ultraviolet-induced skin aging (Choi et al., 2016). On balance, inhibition of mTORC2 by rapamycin is very likely deleterious, as rapalogs effectively suppress mTORC1 signaling directly and there appear to be many negative effects of mTORC2 inhibition on metabolism, frailty, and lifespan.

New ways of inhibiting mTOR

There is significant interest in developing new compounds and strategies to selectively inhibit mTORC1 signaling for the treatment of diseases of aging. As mTORC1 activity is potently regulated by dietary components, there has been great interest in understanding how dietary interventions modulate mTORC1 activity. Several studies have shown that protein restriction, which extends lifespan, reduces mTORC1 activity in the liver and other tissues (Lamming et al., 2015; Solon-Biet et al., 2014), and protein restriction appears to have metabolic benefits that apply across a range of genetic backgrounds (Green et al., 2022b). Dietary restriction of specific amino acids, including methionine and the branched-chain amino acids leucine, isoleucine, and valine, extends lifespan and specifically inhibits mTORC1 (Lees et al., 2014; Richardson et al., 2021). A ketogenic diet, which is low in protein and glucose, likewise reduces mTORC1 activity (Roberts et al., 2017). Some early data from randomized clinical trials suggests that while difficult, BCAA restriction in humans is feasible and has metabolic benefits (Karusheva et al., 2019; Lamming et al., 2020). Metformin, a widely-utilized diabetes medication that is actively being studied in humans as a potential geroprotector (Justice et al., 2018), inhibits mTORC1 via AMPK-mediated phosphorylation of TSC2 and Raptor (Howell et al., 2017; Van Nostrand et al., 2020).

It may also be possible to develop drugs that partially block intestinal uptake of specific dietary amino acids. The neutral amino acid transporter B0AT1 (SLC6A19) is the major amino acid transporter for methionine and BCAAs in the intestine, and deletion of this transporter improves metabolic health similarly to restriction of these amino acids (Jiang et al., 2015). While the effect deleting Slc6a19 on mTORC1 signaling has not been examined, chemical inhibition of LAT1, the major transporter for the BCAAs as well as other essential amino acids, reduces mTORC1 activity (Quan et al., 2020).

Despite concerns about side effects, testing of rapamycin in higher organisms including canines and non-human primates is proceeding, and as of yet has not identified any “showstoppers” (Horvath et al., 2021; Urfer et al., 2017). Rapamycin analogs that more specifically target mTORC1, and should therefore display a reduced side-effect profile, are being developed (Schreiber et al., 2019). mTORC1-targeted kinase inhibitors like RapaLink-1 may enable the inhibition of both rapamycin-sensitive and rapamycin-resistant function of mTORC1, although it is not clear if this will be beneficial or lead to new kinds of side effects (Rodrik-Outmezguine et al., 2016). New information regarding the regulation of mTORC1 has opened new doors for the specific pharmacological manipulation of mTORC1, such as a recently reported compound that binds to Rheb and blocks its ability to activate mTORC1 (Mahoney et al., 2018) and a novel small molecule, EN6, that inhibits mTORC1 by targeting the vATPase (Chung et al., 2019). Finally, a novel Sestrin-binding molecules which can activate mTORC1 has been discovered; conceptually, it is easy to imagine that Sestrin-binding molecules may also be able to specifically inhibit mTORC1 (Kato et al., 2019).

Conclusions and Future Directions

The past decade has seen an incredible growth in our understanding of the physiological and metabolic processes controlled by mTOR signaling and the regulation of mTOR signaling by nutrients, metabolites, and hormonal signals. We have gained a more detailed understanding of the rapamycin-dependent and -independent functions of mTORC1 and have a deep molecular understanding of the regulation of mTORC1 at the lysosomal surface. Our understanding of the physiological roles and substrates mediated by mTORC2 has likewise exploded.

Nevertheless, many questions remain about mTOR regulation as well as the physiological roles of mTOR signaling. At the lysosome, while sensors for several specific amino acids and metabolite have been defined, it remains unclear how the majority of the essential amino acids as well as glucose and fatty acids can signal to lysosomal mTORC1. Further, although incredible progress has been made over the last two decades regarding the regulation of mTORC1 at the lysosome by amino acids, much less is known regarding the regulation of mTORC1 activity at non-lysosomal sites. During the next two decade, we expect that the explosion of mechanistic detail regarding the regulation of mTORC1 by nutritional cues will continue, with detailing of the molecular mechanisms by which lysosomal-derived amino acids and DHAP are sensed by mTORC1.

During the last two decades, phosphoproteomic and gene silencing techniques, as well as the availability of mTOR kinase inhibitors, have led to the rapid expansion of known mTOR substrates. While it would now be difficult to list all of the validated mTORC1 and mTORC2 substrates, it is clear that we do not yet know the full signal transduction web downstream of either mTOR complex. Continued definition of these downstream mTOR effectors over the next decade will for the first time allow us to understand how metabolic health, frailty, and longevity are controlled by mTORC1 and mTORC2.

Finally, it is now becoming clear that over the next two decades, the urgent necessity for interventions to promote health in a graying population will lead to the exploration of mTOR inhibitors for the treatment of aging and age-related disease. During the next several years, we expect to learn the results of ongoing studies of rapamycin in dogs and marmosets, and expect to see both private and NIH-funded clinical trials launched to examine the safety and effectiveness of rapamycin (or other approved rapalogs) on human healthspan. These may be informed by a personalized medicine approach that considers genotype, sex, and diet (Green and Lamming, 2021a, b). As we move into the following decade, we expect to see some of the new pharmacological methods for specifically inhibiting mTORC1 move into the clinic, and longer-term trials of rapalogs and other mTOR inhibitors to promote healthy aging.

One overarching certainty for the next twenty years is that our knowledge of the biology of mTOR will continue to advance and shed new insights into development, metabolism, and aging at the cellular and organismal levels. Further, while the precise paths that will be taken remain to be determined, we will see mTOR-based therapies translated to the clinic with greater confidence and fewer side effects as our ability to precisely modulate mTOR activity or downstream processes grows. Twenty years from now, mTOR inhibition will be a standard and important part of the toolkit not only for laboratory-based researchers, but for clinicians treating a broad range of diseases.