ATM engages the TSC2/mTORC1 signaling node to regulate autophagy

The link between reactive oxygen species (ROS) and induction of autophagy has been well documented, but the molecular mechanisms regulating this phenomenon are only beginning to be elucidated. Autophagy is now being appreciated as an integral part of the cellular response to many diverse types of cellular stresses including nutrient deprivation, hypoxia, oxidative stress, and DNA damage, and likely the mechanism(s) for each type of stress vary considerably. The cellular outcome of inducing autophagy in response to stress is also quite complex, and depends on many factors including cellular context, type and magnitude of stress.

Autophagy induced by low levels of stress may enhance cell survival via recycling of cellular components and macromolecules to allow the cell to engage pathways of repair and checkpoint activation; higher levels of damage over a prolonged period of time may trigger type II programmed cell death (known as autophagic cell death). Elucidating the molecular mechanisms underlying this dichotomy has considerable implications for understanding fundamental aspects of cell biology, as well as treating many human diseases characterized by elevated stress or dysfunctional cell death pathways, such as cancer and neurodegenerative diseases.

Recent work from our group has identified an important cytoplasmic signaling pathway that responds to oxidative stress. Our recent paper demonstrates that cytoplasmic ATM acts as an ROS sensor and activates the TSC2 tumor suppressor by signaling to LKB1 and AMPK to relieve mTORC1 repression of autophagy. We show that both exogenous ROS (direct addition of H₂O₂ or doxorubicin to cells) and endogenous ROS (menadione and phenylethylisothiocyanate) potently and rapidly induce mTORC1 repression via this pathway.

ATM is a well-characterized tumor suppressor protein central to the DNA damage response, which in the nucleus is activated by DNA double-strand breaks, inducing autophosphorylation at a conserved serine residue. This phosphorylation allows the inactive dimers to become catalytically active monomers that regulate a large network of proteins involved in diverse cellular processes. Recently, it has been postulated that ATM could function outside of the nucleus—either as a result of nuclear activation followed by export from the nucleus, or alternatively via activation of a cytoplasmic pool of ATM. Using leptomycin B to block nuclear export and cell fractionation, we found that a substantial cytoplasmic pool of ATM was phosphorylated in the presence of leptomycin B (similar to the control cells) and this ATM could initiate signaling via LKB1, AMPK and TSC2 in the cytoplasm to suppress mTORC1. Our study did not explore the precise biochemical mechanism for ATM activation by ROS. ATM is a large protein containing many cysteine residues, which could be potential targets for direct oxidation by ROS, leading to conformational changes that could activate this kinase. Alternatively, it is possible that lipid peroxide intermediates could mediate signaling to ATM, since ATM has been reported to localize to different membrane-bound compartments in the cytoplasm including peroxisomes, endosomes and the plasma membrane. Further studies will be needed to identify the precise mechanisms for ATM activation in the cytoplasm.

Upon activation by ROS in the cytoplasm, we found that ATM rapidly (<1 h) activates LKB1 by phosphorylation at the same LKB1 site that Sapkota et al found to be phosphorylated by ATM in response to DNA damage. However, in contrast to that study, we found that in MCF7 cells at 1 h after treatment, LKB1 is exclusively cytoplasmic, and that LKB1 activation by ATM results in AMPK activation, which was not examined in this earlier report. Interestingly, when we compare the downstream signaling pathways engaged by DNA damage agents in MCF7 cells, we do not see AMPK activation; likely due to the fact that mTORC1 suppression by DNA damaging agents occurs with much delayed kinetics relative to ROS-induced mTORC1 repression. Following AMPK activation, the tumor suppressor TSC2 becomes activated, resulting in mTORC1 repression. Since mTOR suppresses autophagy, we decided to look at whether autophagy was induced in response to ROS when mTOR was repressed. As expected we could show through multiple molecular and cellular techniques, including LC3-II accumulation, an increase in punctate GFP-LC3, and autophagosome formation by electron microscopy, that autophagic flux was indeed increased by ROS.

Why would cells induce autophagy during oxidative stress? One potential explanation could be that autophagy is important for organelle homeostasis. During periods of high oxidative stress, the proteins and lipids that make up organelle membranes can become damaged leading to organelle dysfunction. Perhaps the removal of mitochondria and possibly other organelles such as the peroxisome, by either selective autophagy (mitophagy/pexophagy) or non-selective autophagy (macroautophagy), could serve to maintain appropriate organelle number and function.

This new cytoplasmic pathway linking ATM and autophagy may have therapeutic implications in cancer. Small molecule ATM inhibitors have been proposed for chemo/radio-sensitization; however, the impact on autophagy of these inhibitors as beneficial or deleterious to cells in these settings remains to be studied. Recently, a number of groups have proposed that ROS-generating agents could be used to kill cancer cells selectively, since cancer cells possess an elevated basal level of ROS, and therefore may easily be elevated above the threshold to induce cell death. Since autophagy may play a cytoprotective role as well as functioning as a mechanism of cell death, the impact on autophagy of increasing endogenous ROS levels must be taken into consideration, especially in settings where ATM, LKB1 and TSC2 tumor suppressor function is lost. It is possible, and perhaps even likely, that tumors which lack components of this signaling pathway would be unable to mount a cytoprotective autophagic response to ROS-inducing agents or would be more prone to undergo cell death. Clearly there are still many questions remaining about targeting cellular pathways to induce autophagy in cancer therapy.