At last, the crystal structure is revealed for the catalytically active mTOR kinase enzyme, a master regulator of cell growth. The structure indicates a gatekeeper mechanism that controls substrate access to the active site. See Article p.217
Whether or not a cell grows is decided by a remarkable protein kinase enzyme called mTOR. As part of two complexes, mTORC1 and mTORC2, mTOR integrates and interprets all sorts of factors that influence cell growth — including nutrients, stressors and the outputs of signal-transduction networks — by targeting a multitude of substrates that drive processes such as protein translation, metabolism and cell division. Research into mTOR-mediated signalling has taken on added urgency since it was discovered that most cancers contain mutations that inappropriately activate this protein1. In an eagerly awaited study published in this issue, Yang et al.2 (page 217) report the crystal structure of the part of mTOR that has enzymatic activity and encompasses many of its crucial domains (the FAT, FRB, catalytic core and FATC domains), in complex with mLST8, a component of both mTORC1 and mTORC2Footnote 1.
The authors' stunning structure, at 3.2 ångströms resolution, comprises some 1,500 amino-acid residues and reveals how mTOR, an atypical member of the phosphoinositide kinase-related kinase family, is evolutionarily related to a group of lipid phosphoinositide 3-kinases and to other phosphoinositide kinase-related kinases (ATR, ATM, DNA-PKcs, SMG1 and TRRAP). It also shows that the 600-residue FAT domain consists mostly of α-helical repeats, which wind like a twisted telephone wire to form a C-shaped lobe around a kinase domain (Fig. 1a, b). Like all other kinases so far described, at its core mTOR possesses the characteristic N and C bi-lobal architecture. However, the kinase domain is roughly 300 residues larger than those of most kinases.
The immunosuppressant drug rapamycin inhibits mTORC1 — and hence signalling downstream of mTOR — by binding to the protein FKBP12. Yang et al. find that the additional residues in the kinase domain of mTOR include a 100-residue FRB region located in the N lobe. And in the C lobe, a 40-residue insertion called LBE interacts with mLST8. This creates a network of stabilizing interactions that probably explains why mLST8 is indispensable for mTOR activity3. The 35-residue FATC domain at the carboxy terminus forms an integral part of the C lobe and is packed against the activation loop.
Most kinases can readily switch between an inactive 'open' conformation and an active 'closed' conformation in response to various signals. By contrast, the mTOR structure shows that its kinase domain has an intrinsically active conformation, in which the catalytic residues and ATP (the molecule that provides phosphate groups) are poised to phosphorylate substrates. But how is the activity of the always-active mTOR controlled? It seems that the FRB domain operates as a gatekeeper of the active site, perhaps preventing access to all but genuine substrates.
Yang et al. also provide evidence that FRB interacts with specific residues (those close to the threonine residue at position 389) in one of its substrates, S6K1, and that this markedly enhances phosphorylation of S6K1 at this threonine residue. Intriguingly, the substrate-binding region on FRB is also the binding site of the FKBP12–rapamycin complex. Therefore, the mechanism by which the FKBP12–rapamycin complex inhibits mTOR probably involves sequestering the FRB docking site and physically blocking substrate access to the catalytic cleft of mTOR.
The authors' detailed sequence inspection shows that some of the mTOR-related kinases (such as DNA-PKcs, SMG1 kinase and TRRAP) contain an FRB-like insertion. It would be fascinating to learn whether the FRB-like domain in these enzymes also participates in substrate recruitment and whether inhibitors similar to rapamycin can be used to target these domains.
The structure lacks the amino-terminal residues that bind regulatory subunits such as RAPTOR (in mTORC1) and RICTOR (in mTORC2). These subunits might be important for controlling the localization and recruitment of substrates to mTOR complexes1. For example, most mTORC1 substrates have a docking site termed the TOS motif, and RAPTOR recognizes this motif4.
Yang et al. present an attractive model (Fig. 1c) for substrate selectivity by mTORC1: initially, RAPTOR recognizes substrates through their TOS motifs and recruits them, and then FRB functions as a secondary, fine-tuning docking site, recognizing residues that lie close to the substrate phosphorylation site before guiding it into the mTOR active site. The authors suggest that this two-part substrate-recruitment mechanism facilitates substrate selectivity and entry into the otherwise restricted catalytic site. It might also provide an additional layer of regulation to ensure that mTOR does not inappropriately phosphorylate non-physiological substrates that do not possess a TOS motif and an FRB-binding domain.
In cancer cells, several mutations that hyperactivate mTOR have been identified in the FAT domain and its interface with the kinase domain5. Yang and colleagues argue that these mutations cluster to regions that control substrate access and loosen the structural skeleton, including the FRB domain, which otherwise restricts substrate admission to the active site.
There are more than 250 clinical trials in progress or planned to evaluate the efficacy of diverse mTOR inhibitors as anticancer agents6. Yang et al. have crystallized mTOR with a highly specific inhibitor (Torin2), as well as with two other, less selective compounds7 (PP242 and PI-103). These compounds bind and inhibit mTOR in markedly different ways, providing a wealth of information on features of the mTOR active site that will undoubtedly be exploited to develop even more potent and selective inhibitors.
Previous work8 showed that the complete mTORC1 complex is a dimer with a striking empty central cavity, which was proposed to facilitate substrate access. The same paper also found that rapamycin treatment caused disassembly of mTORC1 subunits. The current study provides no insight into how kinase activity and substrate phosphorylation would be affected by mTORC1 dimerization, nor does it suggest how rapamycin could induce complex disassembly.
Further work is also required to uncover why rapamycin does not inhibit mTORC2. Also, much remains to be understood about how RAPTOR and RICTOR sense the diverse upstream signals that continuously bombard mTOR complexes, and how this information is coupled to mTOR activation and substrate access to its active site. Fruitful knowledge is to be gained from research in this area, especially from structural analysis of even larger mTOR fragments in complex with other subunits.
*This article and the paper under discussion were published online on 1 May 2013.
Laplante, M. & Sabatini, D. M. Cell 149, 274–293 (2012).
Yang, H. et al. Nature 497, 217–223 (2013).
Kim, D. H. et al. Mol. Cell 11, 895–904 (2003).
Schalm, S. S. & Blenis, J. Curr. Biol. 12, 632–639 (2002).
Hardt, M., Chantaravisoot, N. & Tamanoi, F. Genes Cells 16, 141–151 (2011).
Wander, S. A., Hennessy, B. T. & Slingerland, J. M. J. Clin. Invest. 121, 1231–1241 (2011).
Liu, Q. et al. J. Biol. Chem. 287, 9742–9752 (2012).
Yip, C. K., Murata, K., Walz, T., Sabatini, D. M. & Kang, S. A. Mol. Cell 38, 768–774 (2010).
About this article
Cite this article
Alessi, D., Kulathu, Y. Security measures of a master regulator. Nature 497, 193–194 (2013). https://doi.org/10.1038/nature12101
Nature Communications (2017)