The TSC Complex — Where the Inputs Meet

Read the diagram left to right and count what arrives at the same box. AKT, downstream of insulin and IGF-1, phosphorylates TSC2 and reduces the brake — the growth-factor input from Lesson 03. ERK and RSK, downstream of mitogen signalling through Ras, phosphorylate TSC2 at different sites with a similar consequence, which is how a second, largely independent growth-signalling system reaches mTORC1. AMPK, active when the energy charge of the cell falls, phosphorylates TSC2 in the opposite direction and strengthens the brake. Additional stresses reported to act on the same complex include hypoxia and DNA damage.

The complex has a third core subunit, TBC1D7, alongside TSC1 and TSC2. It is worth naming for a reason beyond completeness: TSC1 and TSC2 were found as disease genes, and TBC1D7 was found afterwards by asking what else is in the complex. A named component list is a snapshot of the experiments done so far, not a closed inventory.

The right-hand side of the figure is the GTPase cycle itself. TSC pushes Rheb toward Rheb-GDP; something else pushes it back. Notice what is not drawn: a well-characterised, dedicated exchange factor that reloads Rheb with GTP. That asymmetry is real — the deactivating side of this cycle is much better described than the reloading side.

Finally, the complex has an address as well as a substrate. The TSC complex can be recruited to the lysosomal membrane, where Rheb is, and that recruitment is itself regulated by insulin and by nutrient availability — so 'is the brake on?' and 'is the brake where the activator is?' are two different questions with two different answers.

If you model TSC as an on/off switch, two common observations look strange. First, mTORC1 output is graded: cells do not sit at either full activity or none, and the same stimulus can give different amounts of downstream phosphorylation in different conditions. Second, the effect of any one input is context-dependent — removing growth factors from a cell with heavy amino-acid supply is not the same experiment as removing them from a starved cell.

Both fall out of the GAP picture. The level of Rheb-GTP is a steady state between deactivation and reloading, and each input changes one of those rates by some amount rather than setting the answer. Several inputs pushing at once give a level, not a verdict. This is also why the pathway can be described as an integrator without any of the usual hand-waving: the integration is arithmetic on a rate, at one complex.

It has a practical consequence for reading papers. When a study reports that some treatment 'activates mTORC1', ask which rate it moved and by how much, and over what period. A treatment that changes the balance by a small amount for an hour and one that removes the brake permanently are different biology, even when the western blot looks similar.

Tuberous sclerosis complex is caused by loss-of-function variants in TSC1 or TSC2. Losing the complex disconnects all of the inputs above at once, and the result is a disorder with benign tumours in several organs, including the brain lesions called subependymal giant cell astrocytomas.

This is the one place in the lesson where human evidence exists, and it is worth looking at closely. In a phase 3 randomised trial of 117 people with tuberous sclerosis, everolimus — an mTORC1-directed drug — was associated with a tumour-volume reduction of at least 50% in 35% of the treated group, versus none on placebo. As a demonstration that this axis is operating in human tissue, that is about as direct as the corpus gets.

It is also narrower than it first appears, and the narrowness is the lesson. The trial tells you what happens when a drug is given to people whose brake is genetically broken. It does not tell you what mTORC1 inhibition does in someone whose brake works, and it says nothing about aging, metabolism or anything else the pathway is associated with. A strong result in a monogenic disease is strong evidence about that disease.