The decision of whether to allocate resources toward cellular growth or toward quality control is a matter of cellular life and death; disruption of growth pathways is an emerging driving force in diseases ranging from cancer to neurodegeneration. In mammalian cells, the protein kinase activity of mTORC1 promotes cellular anabolism and impedes cellular catabolism, ultimately achieving a balance that dictates the rate of cell growth. In response to nutrient levels, mTORC1 is activated upon recruitment to the lysosome, an organelle whose role as a nutrient sensing integrator has recently come into focus. The Rag GTPases are required for mTORC1 recruitment to the lysosome, but the mechanisms via which the Rags sense nutrients and precisely couple the degree of mTORC1 lysosomal recruitment levels to the level of available nutrients were unknown.
I report new live-imaging and reconstitution approaches that enabled the discovery that when the Rags transition from their inactive nucleotide binding state to their active nucleotide state in response to nutrient stimulation, they also loosen their binding affinity for their lysosomal scaffold, Ragulator. The resulting spatial cycling between the lysosome and the cytoplasm ultimately limits mTORC1 accumulation on its Rag-Ragulator lysosomal scaffold, and promotes rapid responsiveness of mTORC1.
Next, I asked whether the nucleotide states of the two Rag GTPase domains are coordinated. Prior work had established that a complex of Folliculin (FLCN) and FLCN-interacting protein 2 (FLCN:FNIP2) serves as a RagC-specific GTPase-activating protein (GAP) and thus has a positive role in mTORC1 stimulation. However, genetic evidence placed FLCN as a tumor suppressor, suggesting a negative role. I reconstituted the “Lysosomal Folliculin Complex” (LFC), a supercomplex composed of Ragulator, inactive-loaded Rags, and FLCN:FNIP2 that localizes to lysosomes. I discovered that in the LFC, FLCN:FNIP2 clamps Rags in their inactive state (RagAGDP:RagCGTP) by directly inhibiting nucleotide exchange in RagA, concomitant with inhibition of its RagC GAP activity, a conclusion reinforced by a high-resolution (3.6 Å) structure of the LFC. Thus, when nutrients are low, FLCN:FNIP2 is able to maintain the Rag heterodimer in its inactive state, but, in response to a rise in nutrients, FLCN:FNIP2 is converted into a functional GAP.
Finally, by assessing newly available structures of active nucleotide-bound and inactive nucleotide-bound Rag heterodimers, along with recent structural information about Rag interactors, I was able to assemble an integrated structure-guided model of the Rag-mediated cycle of recruitment and activation of mTORC1 at the lysosome. My findings increase our understanding of the molecular logic of nutrient sensing and point to new opportunities for manipulating mTORC1 signaling in disease contexts.