Author Correction: Tanc2-mediated mTOR inhibition balances mTORC1/2 signaling in the developing mouse brain and human neurons

mTOR signaling, involving mTORC1 and mTORC2 complexes, critically regulates neural development and is implicated in various brain disorders. However, we do not fully understand all of the upstream signaling components that can regulate mTOR signaling, especially in neurons. Here, we show a direct, regulated inhibition of mTOR by Tanc2, an adaptor/scaffolding protein with strong neurodevelopmental and psychiatric implications. While Tanc2-null mice show embryonic lethality, Tanc2-haploinsufficient mice survive but display mTORC1/2 hyperactivity accompanying synaptic and behavioral deficits reversed by mTOR-inhibiting rapamycin. Tanc2 interacts with and inhibits mTOR, which is suppressed by mTOR-activating serum or ketamine, a fast-acting antidepressant. Tanc2 and Deptor, also known to inhibit mTORC1/2 minimally affecting neurodevelopment, distinctly inhibit mTOR in early- and late-stage neurons. Lastly, Tanc2 inhibits mTORC1/2 in human neural progenitor cells and neurons. In summary, our findings show that Tanc2 is a mTORC1/2 inhibitor affecting neurodevelopment. Alterations of the mTOR signalling pathway are associated with neurodevelopmental defects. Regulators of the mTOR kinase activity are not fully described. Here, the authors show that Tanc2, a scaffolding protein, acts as a direct inhibitor of mTOR kinase activity in the developing mouse brain and cultured human neurons.

In this work, we show that Tanc2 interacts with and inhibits mTOR, and that Tanc2 deletion in mice leads to mTOR hyperactivity and synaptic and behavioral abnormalities that are responsive to the mTOR inhibitor rapamycin. Tanc2 and Deptor, a known inhibitor of mTOR, act at early and late postnatal stages, respectively. TANC2 knockdown in human neurons leads to mTOR hyperactivity. These results suggest that Tanc2 is a negative regulator of mTOR with neurodevelopmental impacts.

Results
Abnormal behaviors and synaptic plasticity in Tanc2-mutant mice. To explore in vivo functions of Tanc2, we first characterized mice carrying a heterozygous deletion of the Tanc2 gene (Tanc2 +/− ), encoding the Tanc2 protein. Tanc2 +/− mice, unlike homozygous Tanc2-mutant (Tanc2 -/-) mice that show embryonic lethality 22 , showed better survival. However, Tanc2 +/− mice showed decreased postnatal survival rates (~56% at postnatal day 5 [P5] and~44% at P110, relative to 100% expected values), indicative of substantial early (embryonic or early postnatal) lethality that is followed by moderate lethality during adolescence and adulthood. Body weights of Tanc2 +/− mice were~90% of those in wild-type (WT) mice. These results indicate a dose-dependent impact of Tanc2 deletion on mouse development and survival.
In behavioral tests, adult male Tanc2 +/− mice (2-5 months; male) showed impaired spatial learning and memory in the Morris water maze, but normal novel-object recognition ( Fig. 1a and Supplementary Fig. 1a). These mice also displayed hyperactivity (open-field) and anxiolytic-like behavior (elevated plus-maze), but largely normal social behavior (three-chamber) and moderate anti-depression-like behavior (forced-swim but not tail-suspension), and as neonates, showed suppressed ultrasonic vocalizations upon mother separation (Supplementary Figs. 1b-e and 2). Female adult Tanc2 +/− mice showed largely similar behavioral abnormalities; hyperactivity (open-field) and anxiolytic-like behavior (elevated plus-maze) but normal depression-like behavior (forced-swim and tail-suspension) ( Supplementary Fig. 3). These results indicate that Tanc2 +/− mice are more relevant to human disease conditions compared with Tanc2 -/mice.
The abovementioned decrease in LFS-LTD at 2-3 weeks, which contrasts with the normal LTP at a similar age (4-5 weeks), cannot be explained by the decrease in currents of NMDA receptors (NMDARs), which are known to regulate both LTP and LTD 37,38 . We thus tested whether synaptic signaling downstream of NMDAR activation, also known to control LTP/LTD 37,38 , is altered by immunoblot analysis of neuronal signaling proteins.
Early rapamycin treatment normalizes LTP and behaviors in adult Tanc2 +/− mice. To determine whether the early mTOR   (Fig. 3a). This treatment normalized the suppressed LTP in adult Tanc2 +/− mice (>P56), without affecting WT synapses (Fig. 3b). In addition, it rescued the abnormal behaviors (spatial memory, hyperactivity, and anxiolytic-like behavior) in Tanc2 +/− adults without affecting those in WT mice (Fig. 3c-e). These results suggest that early mTOR hyperactivity leads to late synaptic and behavioral abnormalities and that early correction of mTOR hyperactivity normalizes the mutant phenotypes in adults, highlighting long-lasting effects.
Tanc2 interacts with and inhibits mTOR. To gain mechanistic insight into how Tanc2 deletion induces mTOR hyperactivity, we first tested whether Tanc2 interacts with mTOR using protein-protein binding assays. Purified Tanc2 protein interacted with purified mTOR protein (Fig. 4a). Tanc2 also formed a complex with mTOR in the mouse brain (Fig. 4b, c). This interaction was mediated by multiple regions of Tanc2 protein and the C-terminal region of mTOR containing FRB and kinase domains ( Fig. 4d-f). Here, mTOR was found to additionally interact with Tanc1 (Fig. 4c), a relative of Tanc2 that is strongly expressed in late stages of rat brain development (>P14) and regulates synapse development but is not critical for mouse development 21,22 . In a control experiment, Tanc2 did not interact with Deptor (Fig. 4d), a known inhibitor of mTOR 40 .
The results described thus far suggest that Tanc2 interacts with mTOR, but do not speak to whether Tanc2 inhibits the kinase activity of mTOR. We tested this possibility by overexpressing Tanc2 in HEK293T cells and found that this was sufficient to inhibit endogenous mTOR activity (Fig. 5a). Consistent with this, in vitro assays using purified proteins showed that Tanc2 inhibits mTOR kinase activity, as evidenced by decreased phosphorylation of the mTORC1 (mTOR + Raptor) target S6K in the presence of Tanc2 (Fig. 5b). In addition, Tanc2 decreased phosphorylation of the mTORC2 (mTOR + Rictor) target Akt (Fig. 5c).
Serum and ketamine regulate the Tanc2-mTOR interaction. We next investigated whether Tanc2-mTOR interactions are regulated by extracellular influences, first testing serum, which is known to activate mTOR 2 . Serum starvation promoted the colocalization and biochemical association of Tanc2 with mTOR in HEK293T cells within~4 h. This effect was reversed by serum replenishment for~24 h (Fig. 6a, b), suggesting that mTOR dissociates from Tanc2 upon serum stimulation. Moreover, the Tanc2-mTOR interaction induced by serum starvation was inhibited by rapamycin (Fig. 6c, d), suggesting that Tanc2 and rapamycin compete for binding to the mTOR FRB domain. Tanc1, which also associates with mTOR in the brain, interacted with mTOR in a serum-and rapamycin-dependent manner ( Supplementary Fig. 5).
Tanc2, Deptor, and Tanc1 distinctly inhibit mTORC1/2 in early-and late-stage neurons. Because Deptor, similar to Tanc2, also binds and inhibits mTORC1/2 40 , we tested whether Tanc2 and Deptor show overlapping or distinct spatiotemporal expression patterns. Immunoblot analyses using cultured neurons and mouse brain extracts showed that Tanc2 protein was more strongly expressed in early stages (embryonic and early postnatal) and was less enriched at synapses ( Supplementary Fig. 7). In contrast, Deptor and Tanc1 showed progressive increases in expression across postnatal stages and stronger synaptic enrichment in both cultured neurons and mouse brains, a pattern similar to that reported for rat Tanc1 and Tanc2 21,22 .
Late  [HT-Rapa], *P < 0.05, **P < 0.01, ***P < 0.001, ns not significant, two-way ANOVA with Bonferroni test). c Early, chronic rapamycin treatment improves impaired spatial learning and memory in Tanc2 +/− mice (2-4 months) in the Morris water-maze test, as indicated by escape latency and the number of crossings over the former platform location in the probe test. Data: mean ± SEM (line graphs), minimal, maximal, median, 25  and Deptor double-knockdown did not produce additive effects at early or late stages, except with respect to early-stage (DIV 7-14) mTOR phosphorylation (Fig. 7a-d). These results suggest that Tanc2 and Deptor/Tanc1 distinctly inhibit mTORC1/ 2 signaling at early and late stages of mouse brain development, respectively, in line with the embryonic lethality of Tanc2, but not Deptor or Tanc1, KO mice 14,22 .
To determine whether neurons or glial cells are more important for Tanc2-dependent mTOR inhibition, we selectively knocked down Tanc2 in neuron-or glia-enriched early-stage cultured hippocampal neurons (DIV 7-14). Neuronal, but not glial, Tanc2 knockdown induced mTOR hyperactivity, and, in line with this, Tanc2 expression was much weaker in glial cells ( Supplementary Fig. 8), suggesting that Tanc2 is more important for mTOR inhibition in neurons than in glial cells at early stages.
Within the neuronal populations, Tanc2 mRNAs were detected in both glutamatergic and GABAergic neurons in mouse brains at P7 and P14, as shown by fluorescent in situ hybridization (FISH) (Supplementary Fig. 9).
TANC2 in human neurons inhibits mTORC1 and mTORC2. Finally, we tested whether Tanc2 inhibits mTOR activity in human neurons. To this end, we knocked down TANC2 in human neural progenitor cells (NPCs) developing into neurons for 2 weeks using two independent TANC2 knockdown constructs. Both TANC2 knockdown constructs similarly increased phosphorylation of S6 (S235/236), 4E-BP (T37/46), and GSK3β (S9), although they exerted moderate effects on Akt (S473) phosphorylation (Fig. 8a-c and Supplementary Fig. 10 Tanc1 and Tanc2 form a complex with mTOR in the mouse brain. Whole-brain lysates (P14; mouse) were immunoprecipitated (IP) with pan-Tanc or mTOR antibodies, followed by immunoblotting. Note that mTOR pull-down also coprecipitated PSD-95 through Tanc2. Three independent experiments yielded similar results. d Both N-and C-terminal regions of purified Tanc2 protein directly interact with purified mTOR but not purified Deptor. GST-tagged purified N-and C-terminal regions of human Tanc2 (aa 1-1358 and aa 1359-1990) were coupled to glutathione beads and incubated with purified mTOR and Deptor proteins, followed by GST pull-down and immunoblot analysis. Three independent experiments yielded similar results. e Tanc2 forms a complex with mTOR in HEK293T cells through multiple regions of Tanc2. Lysates of HEK293T cells expressing deletion variants of Flag-Tanc2 (near-full-length, aa 127-1990; N-terminal region, aa 127-835; middle region, aa 836-1358; C-terminal region, aa 1234-1990) and mTOR (endogenous) were immunoprecipitated with Flag antibodies and immunoblotted with anti-Flag (for Tanc2) and mTOR antibodies. Note that all four deletion variants of Tanc2 interacted with mTOR, suggesting that multiple regions of Tanc2 are involved in mTOR binding. We used the near-full-length Tanc2 because the full-length construct was unavailable at the time of the experiment; experiments repeated using the full-length Tanc2 construct yielded the same results. ANK ankyrin repeats, TPR tetratricopeptide repeats, CC coiled-coil domain, PB PDZ-binding motif. Three independent experiments yielded similar results. f The C-terminal region of mTOR containing FRB, kinase, and FATC domains is sufficient for complex formation with Tanc2 (Fig. 7). These results collectively suggest that Tanc2 inhibits mTORC1/2 in both human and mouse neurons.

Discussion
The present study suggests that Tanc2 is a novel and regulated mTOR inhibitor that has strong neurodevelopmental impacts and therapeutic potential. The first important conclusion from our results is that Tanc2 binds to mTOR. In support of this, Tanc2 forms a complex with mTOR in heterologous cells and in the mouse brain. Moreover, purified Tanc2 proteins form a complex with purified mTOR proteins. Tanc2 uses its multiple domains to associate with mTOR, whereas mTOR binds to Tanc2 through its C-terminal region, containing the FRB, kinase, and FATC domains. The latter is further supported by that rapamycin, known to bind to the FRB domain of mTOR, blocks the colocalization and biochemical association between Tanc2 and mTOR. ; human), followed by immunoblot analysis for phosphorylated (Ser-2448) and total mTOR. Cells in which mTOR was suppressed by rapamycin served as controls. Note that the near-full-length variant significantly inhibited mTOR activity, whereas the shorter deletion variants (aa 836-1986, aa 1234-1990) did not. Data: mean ± SD. (n = 3 independent experiments, *P < 0.05, **P < 0.01, ***P < 0.001 [compared with the first bar in each graph], one-way ANOVA with Bonferroni test). b Purified Tanc2 inhibits the kinase activity of purified mTORC1 (containing GFP-mTOR and mTOR-associated proteins such as Raptor), as shown by decreased phosphorylation of purified S6K (an mTOR substrate). In control experiments, EGFP-Tanc2 was replaced with purified EGFP protein (lanes 1-4; not probed). Data: mean ± SD. (n = 3 independent experiments, *P < 0.05 [compared to the absence of Tanc2/the second or fourth bar in the middle and right panels, respectively], ns not significant, one-way ANOVA with Bonferroni test). c Purified Tanc2 inhibits the kinase activity of purified mTORC2 (containing GFP-mTOR-and mTORassociated proteins such as Rictor), as shown by decreased phosphorylation of purified Akt (an mTOR substrate). The baseline phosphorylation in Akt could be attributable to that a small portion of purified Akt proteins is phosphorylated or that the antibody recognizes non-phosphorylated proteins in addition to phosphorylated proteins. Data: mean ± SD. (n = 3 independent experiments, ***P < 0.001 [compared to the absence of Tanc2/the fourth bar], ns not significant, one-way ANOVA with Bonferroni test). See Source Data 1 for raw data values and Source Data 2 for statistical details.
Tanc2 binds to mTOR in a regulated manner. The presence of serum, well known to activate mTOR, weakens the colocalization and biochemical association between Tanc2 and mTOR. In addition, ketamine, a fast-acting antidepressant known to promote excitatory synapse functions and mTOR activity 41 , inhibits the Tanc2-mTOR interaction in the mouse brain. These results suggest that Tanc2 inhibits mTOR in a regulated manner to coordinate mTOR activity under various nutritional states and during brain development and neuronal or synaptic activities. Specific mechanisms that underlie the regulated Tanc2-mTOR interactions remain to be determined, although they could be posttranslational modifications of mTOR or Tanc2 at binding interfaces or regulatory domains. Perhaps the most important conclusion of the present study is that Tanc2 inhibits mTOR. This is supported by multiple lines of in vitro and vivo evidence. Most directly, purified Tanc2 inhibits the kinase activity of mTOR, as shown by the suppression of mTORC1/2-dependent phosphorylation of mTOR substrates (S6K and Akt). In addition, Tanc2 overexpressed in HEK293T cells inhibits mTOR. Moreover, acute knockdown of Tanc2 increases mTOR activity in cultured mouse neurons at around, but not after, the developmental stages of strong Tanc2 expression (P7-14 but not P21-28). Tanc2 +/− mice show increased mTOR activity in both mTORC1 and mTORC2 complexes at P14 but not at P28 or P52. In addition, Cre-dependent acute knockout of Tanc2 in an independent Tanc2-mutant mouse line during P5-14 (but not P19-28) increases mTOR activity in mTORC1/2. In human neurons, Tanc2 knockdown increases mTOR activity in mTORC1/2 in NPCs and neurons. These results collectively suggest that Tanc2 binds to and inhibits mTOR in mouse and human neurons at early stages.
In addition to Tanc2, Tanc1 interacts with and inhibits mTOR in a rapamycin-dependent manner. Tanc1 expression sharply increases during postnatal stages of mouse brain development, whereas Tanc2 expression is stronger at earlier stages. Deptor, a known mTOR inhibitor, also shows strong late-stage expression, similar to Tanc1. It is therefore possible that Tanc2, Tanc1, and Deptor distinctly inhibit mTOR across different developmental stages. Indeed, our results indicate that Tanc2 and Tanc1/Deptor inhibit mTOR more strongly at around postnatal weeks 2 and 4, respectively. These results are in line with the differential impacts of homozygous Tanc2 and Tanc1/Deptor deletions in mice, where the deletion of Tanc2, but not Tanc1 or Deptor, leads to embryonic lethality 14,22 .
Tanc2 and Tanc1 interact with the PSD-95 family of scaffolding proteins, known to mediate the molecular organization of multi-protein complexes at cell-to-cell junctions such as neuronal synapses in order to couple receptor activations with signaling pathways 24,25 . Therefore, Tanc2 and Tanc1 may recruit mTORC1/2 complexes to PSD-95-based multiple protein complexes at excitatory postsynaptic sites. In line with this idea, Tanc2 has been suggested to recruit cargo dense core vesicles driven by the KIF1A motor protein to excitatory synapses 27 . Synaptically localized mTORC1/2 may be inhibited by local Tanc2 until mTOR activity is increased by the activation of synaptic receptors such as TrkB and mGluRs 42 . The four known members of the PSD-95 family (PSD-95, PSD-93, SAP102, and SAP97) display differential spatiotemporal expression patterns; i.e., PSD-95 and PSD-93 are more abundant at later developmental stages whereas SAP102 expression is stronger at earlier stages. It is therefore possible that Tanc2 and Tanc1 may coordinate mTORC1/2 signaling at both synaptic and non-synaptic sites of PSD-95-enriched multi-protein complexes in developing neural and nonneural tissues.
The synaptic and behavioral phenotypes of Tanc2 +/− mice implicate Tanc2 in the regulation of synaptic plasticity and behaviors, including LTP, learning and memory, hyperactivity, and anxiety-like behavior, all of which are reversed by rapamycindependent mTOR inhibition. In humans, TANC2 mutations have been extensively associated with various neurodevelopmental and neuropsychiatric disorders, including intellectual disability, schizophrenia, and ASD 23,28-36 . These results, together with the embryonic lethality in Tanc2 -/mice and strongly increased mTOR activity in Tanc2 +/− mice, suggest that Tanc2 regulates normal bran development and function by coordinating mTOR inhibition and that rapamycin-dependent mTOR inhibition could possibly be used to treat human patients with Tanc2 mutations and resulting mTOR hyperactivity. In addition, modulation of Tanc2 activity, i.e., antisense Tanc2 knockdown, might have therapeutic potential for mTOR-related brain disorders [5][6][7][43][44][45] .
In conclusion, our study reports that Tanc2 is a regulated mTOR inhibitor with strong neurodevelopmental impacts and that mTOR inhibition could be an effective strategy for treating human individuals with TANC2 mutations suffering from neuropsychiatric disorders, including intellectual disability, ASD, developmental delays, and schizophrenia. Moreover, Tanc2 modulations promoting or suppressing mTOR signaling have therapeutic potential for the treatment of various mTOR-related peripheral and brain disorders.
Fluorescent in situ hybridization (FISH). In brief, frozen sections (14-µm thick) were cut coronally through the hippocampal formation. Sections were thawmounted onto Superfrost Plus Microscope Slides (Fisher Scientific #12-550-15). The sections were fixed in 4% paraformaldehyde for 10 min, dehydrated in increasing concentrations of ethanol for 5 min, and finally air-dried. Tissues were then pretreated for protease digestion for 10 min at room temperature. For RNA detection, incubations with the different amplifier solutions were performed in a HybEZ hybridization oven (Advanced Cell Diagnostics, Hayward) at 40°C. The probes used in these studies were three synthetic oligonucleotides complementary to the sequence 289-1198 of Mm-Tanc2, the sequence 62-3113 of Mm-Gad1-C3, the sequence 552-1506 of Mm-Gad2-C3, the sequence 464-1415 of Mm-Slc17a7-C2, and the sequence 1986-2998 of Mm-Slc17a6-C2 (Advanced Cell Diagnostics, Hayward), respectively. The labeled probes were conjugated to Alexa Fluor 488, Atto 550, and Atto 647. The sections were hybridized 2 h at 40°C with labeled probe mixture per slide. Then the nonspecifically hybridized probe was removed by washing the sections, there times for 2 min each in 1X wash buffer at room temperature. Then Amplifier 1-FL (30 min), Amplifier 2-FL (15 min), Amplifier 3-FL (30 min), and Amplifier 4 Alt B-FL (30 min) were sequentially applied for 15 min at 40°C. Each amplifier solution was removed by washing with 1X wash buffer for 2 min at room temperature. Fluorescent images were acquired using TCS SP8 Dichroic/CS (Leica), and the ImageJ program (NIH) was used to analyze the images.
AAV production and injection. AAV1-hSyn-Cre-eGFP (pENN.AAV.hSyn.HI. eGFP-Cre.WPRE.SV40) and AAV1-hSyn-ΔCre-eGFP (pENN.AAV.hSyn.eGFP. WPRE.bGH) were a gift from James M. Wilson (Addgene #105539-AAV1, #105540-AAV1). For the preparation of AAV particles, we used HEK293T cells. For a single virus preparation, three of 150-pi dishes were used. When cells grew up until 90% confluency, 10 µg of target plasmid, 20 µg of PHP.eB plasmid, and 10 µg of pAAV-helper plasmid were co-transfected in a 150-pi dishes using polyetherimide (PEI) (Polysciences, #23966-1) transfection method with the N:P ratio of 25. Twenty-four hours after transfection, the medium was changed with fresh DMEM + 5% fetal bovine serum (FBS). Seventy-two hours after transfection, the medium was collected at 4°C for the next step, and replaced with fresh DMEM + 5% FBS. After 120 h post-transfection, the medium and cells were harvested together and separated by centrifugation. The supernatant was mixed with 1/5 volume of 40% w/v PEG 8000 (Sigma, #89510)/2.5 NaCl solution. After 2 h of incubation on ice, the mixtures were centrifuged at 4000 × g for 30 min. Pellets were suspended in 6-ml SAN digestion buffer with 100-U/ml SAN (HL-SAN , which are less likely to be fully matured neurons at this stage. Pan-NPCs infected with lentivirus particles for TANC2 knockdown were selected and subjected to neuronal maturation for 2 weeks before the analysis of mTORC1/2 signaling by immunoblot analysis. b, c Knockdown of TANC2 in human neurons leads to hyper-phosphorylation of S6 (S235/236), 4E-BP (T37/46), and GSK3β (S9), indicative of increased mTORC1 and mTORC2 hyperactivity. Note that the effects of the two knockdown constructs are not identical, in particular, for Akt (S473) phosphorylation, which was unaltered and increased by the #1 and #2 knockdown constructs, respectively. It could be because of distinct properties of the two constructs, such as differences in the strengths of target sequence binding, time courses of target gene knockdown, or compensatory cellular responses to adjust Akt activity. Human NPCs infected with two independent TANC2 knockdown lentivirus particles (#1 and #2) were differentiated into neurons and analyzed by immunoblot analyses. Data: mean ± SD. (n = 4 independent experiments; *P < 0.05, ***P < 0.001, ns not significant [compared to controls], one-way ANOVA with Bonferroni test). See Source Data 1 for raw data values and Supplementary Table 1 for statistical details. NaCl) and incubated at 37°C for 1 h. AAV virus particles were separated by iodixanol gradient (Optiprep; Sigma D1556) by ultra-centrifugation at 350,000 × g for 2 h. Collected virus particles from the 42/60% iodixanol interface were diluted with Dulbecco's phosphate-buffered saline (DPBS) and dialyzed using Amicon Ultra-15 with 100-kDa cutoff. Injection of AAV1-hSyn1-Cre/ΔCre-GFP into the mouse hippocampus was performed using stereotaxic apparatus (Kopf Instruments) and mice anesthetized with isofluorane (Piramal Healthcare). Virus solutions were injected into the hippocampus dCA3 region in mice at postnatal day or P5 (AP: Lambda + 1.7, ML: ± 1.7, DV: −1.9) at the speed of 100 nL/min. To determine injection sites more precisely with naïve eyes, AAV1-hSyn-ΔCre-eGFP virus with stronger eGFP fluorescence was mixed and coinjected with AAV1-hSyn-Cre-eGFP at 1:3-1:5 ratios. Virus-injected brains were sliced (100-µm thickness) using a vibratome (Leica VT1200) in ice-cold dissection buffer (212-mM sucrose, 25-mM NaHCO 3 , 5-mM KCl, 1.25-mM NaH 2 PO 4 , 0.5-mM CaCl 2 , 3.5-mM MgSO 4 , 10-mM D-glucose, 1.25 L-ascorbic acid, 2-mM Na-pyruvate). At P14, five slices were collected for GFP control and Cre-GFP, which were further dissected for GFP-positive regions. The dissected tissues were fractionated for crude synaptosome preparation and then subjected to immunoblotting.
Tanc2 and mTORC1 protein purification in HEK293T cells. GFP, GFP-Tanc2, GFP-mTOR + HA-Raptor, or GFP-mTOR + Myc-Rictor were overexpressed in HEK293T cells by transfection using Lipofectamine 3000 for 48 h. GFP-or GFP-Tanc2-expressing cells were lysed with RIPA buffer, and the lysates were incubated with anti-GFP antibodies trapped in agarose beads (Chromotech, GFP-Trap_A) for 2 h at 4°C. The immunoprecipitated complexes were washed three times with the RIPA buffer, and the immobilized proteins were eluted by 0.2-mM glycine-HCl, pH 2.5 for the in vitro kinase assay. For mTORC1 and mTORC2 purification, GFP-mTOR + HA-Raptor (or Myc-Rictor) expressed in HEK293T cells were stimulated with 10-µg/ml insulin (Sigma #I9278) for 30 min prior to lysis in mTOR lysis buffer (40- In vitro direct binding test. GST, GST-Tanc2-N, GST-Tanc2-C, or GST-Tanc2 (full-length) proteins immobilized to agarose beads were incubated with the mTOR recombinant protein (0.5 µg; Origene #TP320457), or Deptor protein, in RIPA buffer for 2 h at 4°C. Protein complexes in the agarose beads were washed with RIPA buffer three times and subjected to SDS-PAGE for Coomassie blue staining or immunoblotting.
Cell culture and transfection. HEK293T cells obtained from the American Type Culture Collection were cultured in DMEM medium supplemented with 10% (v/v) FBS (Gibco) and 1% (v/v) penicillin/streptomycin at 37°C in a humidified 10% CO 2 atmosphere. For live imaging, 1.5 × 10 4 cells were plated in each well of µ-Plate 96 plates. Cells were transfected with Lipofectamine (Invitrogen) according to the manufacturer's instructions. After 24 h, adding the transfection mixture into a well, colocalization analysis proceeded. For immunoprecipitation, 1 × 10 6 cells were plated in 60-mm dish and transfected as described above.
Immunoblot analysis. Protein concentrations were determined using the Pierce BCA protein assay kit (Thermo Scientific) according to the manufacturer's protocol. Proteins resolved in SDS-PAGE were transferred to PVDF membrane (Millipore). After blocking with 5% nonfat milk or 5% BSA, the blots were incubated with primary antibodies overnight at 4°C, incubated with HRP-or IRDye-conjugated secondary antibodies for 1 h at room temperature, and detected by using enhanced chemiluminescence (Thermo Scientific) or Odyssey CLx imaging system (LI-COR Biosciences). All of the uncropped full-length images can be found in Source Data 2.
Live imaging and colocalization analysis. Serum starvation and refeeding experiment were performed by plating HEK293T cells on 96-well plates and transfecting the cells as described above. After 24 h of transfection, cells were washed with PBS and maintained with serum-free culture medium at 37°C in a humidified 10% CO 2 air for 4-5 h, then the medium was replaced with normal culture medium supplemented with serum, incubated for following 4 or 24 h. For rapamycin treatment, transfected cells were treated first with 100 nM of rapamycin or vehicle (4% ethanol, 4% PEG400, 4% Tween 80, and sterile water) for 2 h before the serum starvation procedure described above. Live-cell imaging for colocalization was performed using a Nikon A1R confocal microscope (Nikon Instruments) mounted in a Nikon Eclipse Ti body, and equipped with CFI Plan Apochromat VC objectives (60×/1.4-NA oil or 40×/0.95-NA air; Nikon) and a Chamlide TC system (maintained 37°C and 10% or 10% CO 2 ; Live Cell Instrument, Inc). CFP and YFP images (512 × 512 pixels, 72.7 µm 2 ) were taken using 457 and 514-nm laser lines, respectively. Colocalization images were quantified using ImageJ coloc2 plugin.
Mice. Tanc2 +/− mice have been previously described (genetic background: C57BL/ 6J) 22 . All mice used in this study were generated by in vitro fertilization. To generate Tanc2 fl/fl mice, Tanc2 tm2a( KOMP)Wtsi ES cells were obtained from KOMP. The mutant Tanc2 allele was inserted in the intron located between exons 4 and 5 of the Tanc2 gene. Embryonic stem cells were injected into blastocysts to generate chimeric mice. Chimeric mice were further bred to C57BL/6J WT mice to generate F1 heterozygous Tanc2 tm2a (KOMP)Wtsi mice (Tanc2 null/+ ). F1 mice were then bred with Protamine-Flp mice (C57BL/6J0 to remove the Frt-flanked gene trap cassette in the Tanc2 tm2a(KOMP)Wtsi allele, resulting in a conditional allele (Tanc2 fl/+ ). Homozygous Tanc2 fl/fl mice were generated from Tanc2 fl/+ intercrosses, which allow further conditional knockout in the presence of Cre expression. Tanc2 fl/fl mice were identified by PCR with primers (Supplementary Table 2) to amplify a 222-bp fragment from the WT allele and a 440-bp fragment from the flox allele. Mice maintenance and procedures were performed in accordance with the Requirements of Animal Research at KAIST. Experimental procedures were approved by the Committee on Animal Research at KAIST (KA2016-31). Mice were fed ad libitum, and 2-5 animals were housed in a cage with a 12-h dark/12-h light cycle. least 24 h of rest time between tests. All experimental data were analyzed using Ethovision XT 10.1 software (Noldus) unless stated otherwise analysis. All tests were conducted in a blind manner.
Open-field test. Subjects were placed in a white acryl open-field box (40 × 40 × 40 cm), and the movements were recorded for 60 min for adult mice. Illuminations were set at 0 or 100 lux. The distance moved and time spent in the center arena (central area of 20 × 20) were measured.
Laboras test. Various behaviors, including locomotion, immobility, climbing, grooming, and rearing, were recorded automatically using Laboratory Animal Behavior Observation Registration and Analysis System (LABORAS TM ) by Metris 46,47 . Mice were housed in LABORAS recording cages. The cage is directly placed onto the sensing platform with the upper part of the cage, including the top, food hopper, and drinking bottle. The recording was conducted for 72 consecutive hours. Animals were tested at the age of 2-3 months.
Morris water-maze test. Morris water-maze testing was conducted in a round white pool 100 or 120 cm in diameter and 40-cm deep. The maze was placed alone in a room with four cues on its walls. The pool was filled with water until the top of the platform (10 cm in diameter) was submerged 1 cm below the water surface, and white paint was added. The pool temperature was maintained at 22 ± 0.5°C by adding warm water. Each mouse was given 3 trials a day for 5 consecutive days. In each trial, mice were placed in three different quadrants in random order. Irrespective of the trial performance, mice were guided to the platform and allowed to remain there for at least 15 s. In the probe test performed 24 h after the acquisition task, mice were placed in the center of the arena and allowed to swim for 1 min in the absence of the platform. The number of times the mice crossed the learned escape platform location, as well as the time spent in the quadrant, was measured. On the day after the probe test, mice were subjected to a reversal task in which the platform was placed in the quadrant opposite side of the arena and received three trials a day during 3 consecutive days. The procedure remained the same as that of the acquisition task.
Novel-object recognition test. Each mouse was habituated in the open-field box without objects for 30 min a day before the training session. At the training session, the mouse was placed in the open-field arena containing two identical sample objects for 10 min. Twenty-four hours after the training, the mouse was returned to the open-field arena with two objects, one is identical to the sample, and the other is novel. Object recognition was scored by the amount of time with the nose of the mouse pointed and located within 2 cm from the object.
Contextual fear conditioning test. Mice at 4-5 months of age were placed in the conditioning chamber (Coulbourn instruments) and, after a 2-min adaptation period, received three foot shocks (2 s, 0.5 mA) at 2, 3, and 4 min. After the foot shock, mice remained in the chamber for an additional minute and then were returned to their home cage. After 24 h, mice were placed back into the same context for 5 min, and freezing was monitored. Data acquisition and analysis were measured using FreezeFrame (Coulbourn Instruments).
Elevated plus-maze test. The apparatus used for the elevated plus-maze test consisted of two open and closed (30-cm walled) arms (5 × 30 cm each). The maze was elevated to a height 50 cm above the floor. Each mouse was placed in the central zone of the maze facing an enclosed arm and was recorded for 10 min. The frequency of entries and the amount of time spent in each arm were measured.
Light-dark test. The apparatus (40 × 20 × 20 cm) used for light-dark test was divided into two compartments; a smaller dark chamber that occupied roughly 1/3 of the total box, and a larger light chamber (200 lux), which was lit with a bright white light. There was a hole in between that allows mice to move freely between the compartments. The task ran for 10 min, and started after the mice were placed into the dark chamber. The anxiety parameters, time spent in the light compartment, and the number of entries into the light compartment were measured.
Three-chamber social interaction test. The three-chambered social approach test, originally developed for social interaction of rodents 48,49 , was conducted as previously described 50 . In brief, plastic containment cups in the corner of both side chambers were placed. Mice were habituated in the center chamber for 10 min for adaptation, followed by in all three chambers for 10 min. Following habituation, Stranger 1 mouse (129/SvJae strain) was placed in a small container located in the corner of one side chamber, and the Object was placed in another container in the opposite side chamber. Subject interaction was recorded for 10 min. Following the first test, the Object was replaced with Stranger 2 (129/SvJae) and subject interaction was recorded for 10 min.
Forced-swim test. The apparatus, a glass beaker (15-cm diameter, 22-cm high), was filled with water (~24°C) to the height of 15 cm. The time spent floating on the water (immobility time, s) during 6 min was manually scored.
Tail-suspension test. Mice were individually suspended by their tails to a horizontal metal bar using adhesive tape. The distance between the tip of the nose of the mouse and the floor was 20 cm. The time spent immobile was recorded for 6 min and manually scored.
Image acquisition and analysis. Fixed neurons were imaged with a 63X objective oil lens on a Zeiss LSM780 Confocal Microscope driven by ZEN software (Zeiss). Images were obtained as a 1-µm Z-stack with 0.5-µm spacing. For quantification of the spine/shaft fluorescence ratio, line plots of fluorescence intensity were generated across spine heads and the adjacent dendritic shafts using ZEN software. Fluorescence intensity at each compartment was quantified from the maximum intensity corresponding to the spine and the dendrite after background subtraction. Then, ratios were calculated as the ratio between the dendritic spine and shaft. More than ten spines were measured for each neurons and each experiment was repeated at least three times in separate neuronal cultures.
Data acquisition, statistics, and analysis. All experiments were carried out in a blind manner. Statistical analysis was conducted using Prism 7.0 software (GraphPad Software, Inc.). Statistical details are described in Supplementary  Table 1.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All data supporting the findings of this study are provided within the paper and its Supplementary information. A source data file is provided with this paper. All additional information will be made available upon reasonable request to the authors. Source data are provided with this paper.