Research

We study the relationship between stress, aging and metabolism. Environmental stresses are generally considered to accelerate aging and age-associated tissue dysfunction because they damage important cellular macromolecules such as DNA, protein and lipids. However, cells have several stress adaptation mechanisms that can reduce negative impacts on cell and organismal homeostasis. Interestingly, appropriate activation of such adaptive mechanisms could often lead to not only increased stress resistance but also general extension of health and lifespan. We hope that, in the future, we are able to utilize some of these protective mechanisms to generate therapeutics for age-associated degenerative disorders. We have four specific research projects that cover the stress-aging relationship.

Sestrin was originally discovered as a target of the tumor suppressor p53 in a differential display screening. Follow-up studies have revealed that there are three Sestrin genes (Sestrin1-3) in the mammalian genome and that all three can be transcriptionally upregulated after a variety of environmental stress, including genotoxic, oxidative and hypoxic stresses. In cultured cells, Sestrins exhibit two important biological activities – suppression of reactive oxygen species (ROS) and inhibition of the mammalian target of rapamycin complex 1 (mTORC1) signaling. Because excessive ROS accumulation and mTORC1 activation are well-known facilitators of aging and age-associated diseases, Sestrins are viewed as a potential linker between stress signaling and aging physiology. However, when we first started investigating this protein family, there was very limited information known on its physiological function, due to the absence of any genetic studies in model organisms. Furthermore, the molecular nature of Sestrin and biochemical mechanisms of how Sestrin inhibits ROS and mTORC1 were both unknown.

We were the first to generate Sestrin-deficient and Sestrin-overexpressing animal models using the Drosophila system, and to discover the role of Sestrins as a feedback inhibitor of mTORC1 signaling that attenuates various age- and obesity-associated metabolic pathologies (Lee et al. Science 327, 1213). This work was conducted in Dr. Michael Karin’s lab at UC San Diego, where the PI did his postdoctoral training. After moving to the University of Michigan, we continued following up with these findings using mouse models of obesity and diabetes. As a result, we were able to demonstrate that Sestrins regulate metabolism in the mammalian system. Specifically, we showed that Sestrin2 is induced in the liver upon obesity and subsequently attenuates obesity-induced hepatosteatosis and hepatic insulin resistance through regulation of the AMPK-mTORC1 pathway (Lee et al. Cell Metab, 16, 311).

We further investigated the mechanisms underlying how Sestrins regulate metabolism. We showed that Sestrin2 maintains liver ER homeostasis under conditions of obesity-associated lipotoxic challenges, and thereby attenuates the progression of fatty liver pathologies (Park et al. Nat Commun, 5, 4233). Biochemically, Sestrins regulate mTORC1 through the GATOR1/2 complexes (Kim et al. Sci Rep 5, 9502), two multi-protein complexes recently identified to regulate mTORC1. Sestrin2-GATOR2 interaction was also important for upregulation of mTORC2, which was independent of mTORC1 regulation (Kowalsky et al. J Biol Chem 295, 1769). Sestrin2 also directly regulates autophagy through ULK1, an autophagy-initiating protein kinase (Ro et al. FEBS J 281, 3816). At the structural level, Sestrin2 shows striking internal symmetry with two homologous domains: one domain is responsible for ROS reduction and the other associates with GATOR2 to modulate mTORC1 signaling (Kim et al. Nat Commun 6, 10025); the structural study was done in collaboration with Dr. Uhn-Soo Cho’s lab (UM – BioChem).

The Sestrin functions are not just important in the liver but also in other organs. In brown adipocytes, Sestrin2’s ROS-reducing activity was important for regulating mitochondrial thermogenesis through Ucp1 regulation (Ro et al., PNAS 111, 7849). In the colon, Sestrin2 was important for maintaining epithelial homeostasis and suppressing the growth of colon tumor cells; Sestrin2 was revealed to be one of the important tumor suppressors functioning downstream of p53 (Ro et al. eLife 5, e12204). In T cells, Sestrins suppress immune response upon aging (Lanna et al. Nat Immunol., 18, 354). In heart, Sestrins were also critical for protecting cardiomyocytes from ischemia-reperfusion injury (Morrison et al. FASEB 29, 408) and doxorubicin-induced cardiotoxicity (Li et al. Am J Physiol Heart Circ Physiol., 317, H39). These studies were conducted either in our lab or as a collaboration with Drs. Yatrik Shah (UM – MIP), Arne Akbar (UCL, UK), Ji Li (SUNY Buffalo) and Zunjian Zhang (China Pharmaceutical University).

Most importantly, our recent work, conducted with Dr. Robert J. Wessells' lab (Wayne State University), highlighted the critical role of Sestrins in mediating the benefits of physical movements and exercise. Using both fly and mouse systems, we showed that Sestrin deficiency can nullify exercise benefits in improving endurance performance and metabolic fitness (Kim et al. Nat Commun 11, 190). In contrast, Sestrin overexpression mimicked various effects of exercise. For instance, collaboration with Dr. Pura Muñoz Cánoves (UPF) demonstrated that Sestrin overexpression could be sufficient for preventing muscle loss during disuse and aging (Segalés et al. Nat Commun 11, 189). These results suggest that Sestrin mimetics or activators could be used as a novel way to treat age- or physical inactivity-related muscle wasting.

These series of research findings greatly enhanced our understanding of (1) the physiological functions of Sestrins in a mammalian model, (2) the biochemical basis of Sestrins’ physiological activities, and (3) the potential of Sestrin modulators and mimetics for attenuating or treating various metabolic disorders associated with aging and obesity. This progress, along with the advances made by others in the field, are summarized in recent review articles from our group (Lee et al. Cell Metab 18, 792; Ro et al. Front Endocrinol 6, 114; Ho et al. Trends Biochem Sci 41, 621; Kim et al. Ann. Rev. Physiol 83, 381). 

Autophagy is a physiological process that is responsible for the elimination of toxic protein aggregates, unnecessary nutrient deposits and dysfunctional organelles such as damaged mitochondria. Therefore, autophagy is important for resolving stress-induced damages, and defective autophagy is implicated in many human diseases such as metabolic abnormalities and neurodegeneration. Autophagy is also one of the important targets of the Sestrin-mTORC1 signaling pathway.

Using the Drosophila model system, we devised a novel screening method that can efficiently isolate new autophagy regulators. Using mutant flies for the Fip200 gene, a homolog of the mammalian Fip200/Rb1cc1 gene that controls autophagy, we have confirmed that this scheme is operational for isolating formerly unidentified genes (Kim et al. Autophagy 9, 1201). Among the autophagy-regulating candidate genes, we were intrigued by a gene named Gyf, which is a homologue of a human gene Gigyf2/PARK11, whose mutation was proposed to play a causative role in a familial type of Parkinson’s disease. Through extensive genetic analyses, we established that Gyf is an essential regulator of autophagy that prevents neurodegenerative pathologies in flies (Kim et al., Autophagy 11, 1358). Strikingly, Gigyf2-deficient mice also exhibited neurodegenerative phenotypes, suggesting conservation of gene function across invertebrate and mammalian species.

While conducting these projects, we learned that Dr. Margit Burmeister, in the Department of Human Genetics at the University of Michigan, recently isolated a novel sequence variant (E122D) of the human ATG5 gene, from a family with ataxia patients. Dr. Daniel Klionsky’s lab initially characterized this mutation in yeast and found that the homologous mutation impaired the autophagy-regulating function of the yeast Atg5 gene. We further extended these findings and showed that (1) patient-derived cell lines had a strong impairment in the autophagy process, (2) the mutation interferes with conjugation between Atg5 and Atg12, a critical step in autophagy, and (3) WT, but not the E122D mutant, human ATG5 can rescue ataxia-like neurodegenerative phenotypes of Atg5-null mutant flies (generated by Dr. Gabor Juhasz group in Hungary). This is the first demonstration of a genetically-inherited human mutation in a core autophagy gene with a pathological implication (Kim et al. eLife 5:e12245).

We also investigated the pathogenetic mechanisms of how obesity compromises autophagy. Notably, it has been known that obesity and non-alcoholic steatohepatitis are associated with the accumulation of insoluble protein aggregates that are composed of ubiquitinated proteins and ubiquitin adaptor p62/sequestosome 1 (SQSTM1), which are normally substrates of autophagy. We showed that lipotoxic insults can chronically elevate cytoplasmic calcium levels, which specifically inhibits the fusion between autophagosomes and lysosomes. Correspondingly, inhibition of cytosolic calcium influx through calcium channel blockers restored autophagy and metabolic homeostasis in obese mice. This clarified the mechanism of how obesity inhibits autophagy and how we are able to normalize this dysregulation pharmacologically (Park et al. Nat Commun 5, 4834). In addition to the autophagic mechanism, we also found that lipotoxic activation of TBK1 and subsequent p62/SQSTM1 phosphorylation are critical steps in the NASH pathology of protein inclusion accumulation in hepatocytes (Cho et al. Hepatology 68, 1331). Such dysregulation of protein homeostasis could be similarly observed in a new model of liver failure, which was provoked by unregulated mTORC1 hyperactivation (Cho et al. Cell Discov. 5, 60).

The pathophysiological roles of autophagy in neuronal homeostasis (Kim et al. Int J Mol Sci. 18, E1596; Burmeister et al. Autophagy 12, 1208) and metabolic regulation (Cho et al. Hepatology 66, 700; Park and Lee, Autophagy 10, 2385; Namkoong et al. Mol Cells 41, 3), partially revealed by our studies described above, are summarized in our recent reviews and commentaries.

While studying mechanisms of protein aggregation, we became interested in another insoluble structure, the stress granule (SG), which is a complex of RNA and proteins that accumulates inside the cytoplasm of cells subjected to environmental stress. By focusing on this granular structure, our lab and Dr. Hojoong Kwak’s lab at Cornell University recently developed a novel method, called RG RNA-seq, for characterizing mRNA species trapped in insoluble SGs. Using this new method, we revealed that ER stress, heat shock and arsenic toxicity, which provoke distinct transcriptional responses, induce a conserved response in stress-induced RNA granulation. Interestingly, we found that proto-oncogenic transcripts are selectively recruited to insoluble SGs after all three types of stress. Therefore, SG targeting seems to be a novel layer of stress-dependent gene regulation, which is universally important for stress adaptation but was formerly unappreciated. This story was recently published in Molecular Cell as a cover story article (Namkoong et al. Mol Cell 70, 175).

After the publication of this paper, this new method was soon applied to various research projects. For instance, Dr. Kwak’s lab combined RG RNA-seq with a novel method of measuring the length of poly-A tails (TED-seq) and revealed that SG-targeted mRNAs have a dramatically reduced poly-A tail length. These results suggested that SG targeting is associated with decreased protein expression (Woo et al. Cell Rep 24, 3630). Another collaboration with Dr. Dan Klionsky’s group at the UM Life Science Institute revealed that Dhh1, a stress granule protein, controls protein expression of autophagy-regulating proteins during nutrient starvation (Liu et al. PLoS Biol. 17, e3000219). In collaboration with Dr. Samie Jaffrey’s group at the Weill-Cornell Medical College, we also utilized the RG isolation technique to support that m6A modification of mRNAs is an important driving factor for RNA sequestration in SGs (Ries et al. Nature 571, 424).

Stress insults induce heterogeneous responses in individual cells, which can lead to variations in cell fate determination. However, stress responses were typically measured in bulk RNA or proteins, so cell-specific stress response has not been examined. To overcome this limitation, we collaborated with Dr. Hyun-Min Kang’s lab in the Department of Biostatistics at the University of Michigan, an expert in developing new computational tools for analyzing single cell RNA-seq data. Using the single cell RNA-seq technique, we found that, upon DNA damage, cells undergoing different fates have distinct transcriptomic landscapes that affect their cell fate response. Although DNA damage response genes were thought to be uniformly induced after DNA damage, our data indicated that many of them were actually expressed in a specific subset of cells having a specific cell fate. These findings provide valuable insights into understanding chemotherapy responses of cancer cells, such as fractional killing and chemoresistance development (Park et al. Cell Rep 32, 108077).

The same single cell RNA-seq technique was applied to understand hepatocyte responses to hypernutrition and obesity. Through this, we showed that all hepatocytes, regardless of their metabolic zonation and fat accumulation profiles, undergo substantial transcriptome alteration upon HFD; however, the alteration patterns were highly heterogeneous across the hepatocyte population with zonation-dependent and -independent effects. We also isolated transcriptome features that can explain how different hepatocytes have different rates of lipid droplet accumulation in the liver. This dataset provided a unique and useful resource for understanding hepatocellular alteration during non-alcoholic fatty liver disease (NAFLD) at single cell level (Park et al. AJP-Endo 320, E244).

This is a recently developed area in my research program. While conducting stress transcriptome studies, I came up with an innovative idea of developing new technology, using which microscopic spatial transcriptome could be profiled. With my research team, this technology was successfully implemented with both experimental and computational methods, and named as Seq-Scope. Using Seq-Scope, we were able to visualize spatial transcriptome heterogeneity at multiple histological scales, including tissue zonation according to the portal-central (liver), crypt-surface (colon) and inflammation-fibrosis (stressed/injured liver) axes, cellular components including single cell types and subtypes, and subcellular architectures of nucleus and cytoplasm. Seq-scope is quick, straightforward, precise and easy-to-implement. Therefore, this can make spatial single cell analysis accessible to a wide group of biomedical researchers (Cho et al. Cell 184, 3559).