Individuals with germline inactivating mutations in either the TSC1 or TSC2 gene develop tuberous sclerosis complex, a devastating disease with three distinct categories of manifestations: neurological disease (seizures, intellectual impairment, autism), tumors, which can occur in the brain, heart, skin and kidney, and pulmonary lymphangioleiomyomatosis, which can occur in women with TSC. The TSC1-TSC2 protein complex inhibits mTORC1 via the small GTPase Rheb, which is the direct target of the GTPase activating domain of the TSC2 protein, tuberin. mTORC1 is a key inhibitor of autophagy, via direct phosphorylation of the ULK1 and ULK2 kinases.
TSC provides a unique opportunity to address the implications of autophagy dysregulation in human disease with only three “degrees of separation” between the TSC proteins and autophagy regulation: TSC-Rheb-mTORC1-ULK1. Autophagy dysregulation has been implicated in a variety of human tumors, including tumors with mTORC1 activation. In many human and mouse tumors in which autophagy has been studied, mTORC1 is likely to be activated through a series of upstream signals, and the tumors are likely to be genetically complex. TSC may be the single human disease in which tumorigenesis is mostly closely linked with mTORC1 and autophagy.
This clear biochemical link to mTORC1 activation and autophagy inhibition led us to address the consequences of autophagy inhibition in the pathogenesis of TSC. As predicted, we found that autophagy levels are low in Tsc2-deficient cells at baseline. Autophagy can be induced by stimuli such as hypoxia in Tsc2-deficient cells, but never to the same level as control Tsc2-expressing cells. The first question we addressed was: Do these low levels of autophagy serve to promote or inhibit tumor cell growth in TSC? In models of Tsc2 deficiency in which Atg5 is downregulated, we observed extensive central necrosis of xenograft tumors, and Tsc2+/−Beclin 1+/− mice exhibit fewer renal tumors compared with Tsc2+/− mice, indicating that further autophagy inhibition decreases the growth of Tsc2-deficient tumors.
Interestingly, mTORC1 inhibitors, which activate autophagy, have partial efficacy in the treatment of certain manifestations of TSC, including angiomyolipomas, subependymal giant cell astrocytomas and LAM. Therefore, our second question was: How does autophagy activation by mTORC1 inhibitors affect the survival and growth of Tsc2-deficient cells? To address this, we used the mTORC1 inhibitor sirolimus (rapamycin) and/or the autophagy inhibitor chloroquine (CQ). In vitro, the combination of both drugs more significantly inhibits ATP levels and the survival of Tsc2-deficient cells, compared with either agent alone. In vivo, CQ shows single-agent efficacy in a xenograft model, and the combination of rapamycin and CQ inhibits the growth of xenograft tumors and the development of renal tumors in Tsc2+/− mice more effectively than either agent alone.
What are the clinical implications of these results? Our data utilizing both genetic and pharmacological inhibition of autophagy indicate that Tsc2-deficient cells are highly dependent on autophagy for survival—an “Achilles' heel.” mTORC1 inhibitors potently induce autophagy. Our data suggest that this provides a survival advantage to tumor cells in TSC. This may lead to a type of “dormancy” in which proliferation is blocked because of mTORC1-mediated inhibition of protein translation, but long-term survival is possible because of mTORC1-mediated activation of autophagy. The net result of these opposing influences (growth arrest and autophagy induction) may underlie the partial response observed in human TSC tumors upon treatment with mTORC1 inhibitors. Based on our data, we propose that autophagy inhibitors will have efficacy as single agents in TSC, and that the combination of autophagy inhibition with mTORC1 inhibition will have enhanced efficacy vs. either treatment alone.
What are the next steps toward the potential translation of these results toward better therapeutic strategies for TSC and LAM? There is currently no genetically engineered mouse model that faithfully recapitulates the tumor phenotypes that are responsible for the highest levels of morbidity and mortality in human TSC (angiomyolipomas, subependymal giant cell astrocytomas, and LAM). It is impossible to know how precisely the results from xenograft mouse models of Tsc2-deficient cells or the epithelial renal cystadenomas that develop in the Tsc2+/− mice will translate to human TSC. Therefore, the next steps could include determining the safety and tolerability of autophagy inhibition and the combination of mTORC1 inhibition and autophagy inhibition in individuals with TSC and LAM, with the ultimate objective of a clinical trial to determine whether autophagy inhibitors have efficacy as single agents and/or enhance the response to mTORC1 inhibitors, using tumor size and/or lung function endpoints. The autophagy inhibitors chloroquine and hydroxychloroquine are FDA approved for the treatment of malaria, rheumatoid arthritis and other diseases, enhancing the feasibility of trials in LAM and TSC.
A major area of uncertainty is the contribution of autophagy dysregulation to the neurological manifestations of TSC, which include seizures, intellectual disability and autism. Ongoing studies in humans are addressing the potential role of mTORC1 inhibition in the neurocognitive deficits in TSC, based on promising data from murine studies. Whether and how autophagy induction by mTORC1 inhibitors affects the neurocognitive phenotypes is unknown. Another key area of uncertainty is the role of p62/sequestosome 1 in the pathogenesis and treatment of TSC and LAM. Our data indicate that p62 is required for tumor formation by tuberin-deficient cells, but the mechanism is unknown.
In conclusion, as a genetic disease, TSC provides a window into the roles of autophagy in disease pathogenesis. Our data indicate that tumor cells in TSC are dependent on autophagy for survival. Furthermore, our data indicate that autophagy inhibition may have therapeutic potential in TSC and LAM, either as single-agent therapy or in combination with mTORC1 inhibitors.